Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
/*-
|
2003-02-20 20:02:32 +00:00
|
|
|
|
* Copyright (c) 1998 - 2003 S<EFBFBD>ren Schmidt <sos@FreeBSD.org>
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
* All rights reserved.
|
|
|
|
|
|
*
|
|
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
|
|
* are met:
|
|
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
|
|
* notice, this list of conditions and the following disclaimer,
|
|
|
|
|
|
* without modification, immediately at the beginning of the file.
|
|
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
|
|
* documentation and/or other materials provided with the distribution.
|
|
|
|
|
|
* 3. The name of the author may not be used to endorse or promote products
|
|
|
|
|
|
* derived from this software without specific prior written permission.
|
|
|
|
|
|
*
|
|
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
|
|
|
|
|
|
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
|
|
|
|
|
|
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
|
|
|
|
|
|
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
|
|
|
|
|
|
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
|
|
|
|
|
|
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|
|
|
|
|
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|
|
|
|
|
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
|
|
|
|
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
|
|
|
|
|
|
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
|
|
|
|
*
|
1999-08-28 01:08:13 +00:00
|
|
|
|
* $FreeBSD$
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
*/
|
|
|
|
|
|
|
1999-10-09 19:57:13 +00:00
|
|
|
|
#include "opt_ata.h"
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
#include <sys/param.h>
|
|
|
|
|
|
#include <sys/systm.h>
|
2001-03-15 15:36:25 +00:00
|
|
|
|
#include <sys/ata.h>
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
#include <sys/kernel.h>
|
2001-03-15 15:36:25 +00:00
|
|
|
|
#include <sys/conf.h>
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
#include <sys/disk.h>
|
1999-04-18 20:48:15 +00:00
|
|
|
|
#include <sys/module.h>
|
|
|
|
|
|
#include <sys/bus.h>
|
2000-05-05 09:59:14 +00:00
|
|
|
|
#include <sys/bio.h>
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
#include <sys/malloc.h>
|
2000-01-27 19:00:51 +00:00
|
|
|
|
#include <sys/sysctl.h>
|
2000-01-24 20:45:24 +00:00
|
|
|
|
#include <machine/stdarg.h>
|
1999-04-18 20:48:15 +00:00
|
|
|
|
#include <machine/resource.h>
|
|
|
|
|
|
#include <machine/bus.h>
|
|
|
|
|
|
#include <sys/rman.h>
|
2000-02-18 20:57:33 +00:00
|
|
|
|
#ifdef __alpha__
|
|
|
|
|
|
#include <machine/md_var.h>
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
#endif
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
#include <dev/ata/ata-all.h>
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
#include <dev/ata/ata-disk.h>
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
#include <dev/ata/ata-raid.h>
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
#include <dev/ata/atapi-all.h>
|
|
|
|
|
|
|
2001-03-15 15:36:25 +00:00
|
|
|
|
/* device structures */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
static d_ioctl_t ataioctl;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
static struct cdevsw ata_cdevsw = {
|
2003-03-03 12:15:54 +00:00
|
|
|
|
.d_open = nullopen,
|
|
|
|
|
|
.d_close = nullclose,
|
|
|
|
|
|
.d_ioctl = ataioctl,
|
|
|
|
|
|
.d_name = "ata",
|
|
|
|
|
|
.d_maj = 159,
|
2001-03-15 15:36:25 +00:00
|
|
|
|
};
|
|
|
|
|
|
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
/* prototypes */
|
2000-02-18 20:57:33 +00:00
|
|
|
|
static void ata_boot_attach(void);
|
|
|
|
|
|
static void ata_intr(void *);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
static int ata_getparam(struct ata_device *, u_int8_t);
|
|
|
|
|
|
static int ata_service(struct ata_channel *);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
static void bswap(int8_t *, int);
|
|
|
|
|
|
static void btrim(int8_t *, int);
|
|
|
|
|
|
static void bpack(int8_t *, int8_t *, int);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
static void ata_change_mode(struct ata_device *, int);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
static u_int8_t ata_enclosure_sensor(struct ata_device *, int, u_int8_t, u_int8_t);
|
|
|
|
|
|
static int ata_enclosure_status(struct ata_device *, int *, int *, int *, int *);
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2001-03-19 08:04:54 +00:00
|
|
|
|
/* sysctl vars */
|
|
|
|
|
|
SYSCTL_NODE(_hw, OID_AUTO, ata, CTLFLAG_RD, 0, "ATA driver parameters");
|
|
|
|
|
|
|
2001-03-06 21:43:46 +00:00
|
|
|
|
/* global vars */
|
2002-12-03 20:20:44 +00:00
|
|
|
|
struct intr_config_hook *ata_delayed_attach = NULL;
|
2001-03-06 21:43:46 +00:00
|
|
|
|
devclass_t ata_devclass;
|
|
|
|
|
|
|
1999-10-09 19:57:13 +00:00
|
|
|
|
/* local vars */
|
2000-12-08 20:09:00 +00:00
|
|
|
|
static MALLOC_DEFINE(M_ATA, "ATA generic", "ATA driver generic layer");
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2002-08-09 20:54:06 +00:00
|
|
|
|
/* misc defines */
|
|
|
|
|
|
#define DEV_ATAPIALL defined(DEV_ATAPICD) || defined(DEV_ATAPIFD) || \
|
|
|
|
|
|
defined(DEV_ATAPIST) || defined(DEV_ATAPICAM)
|
|
|
|
|
|
|
2000-09-19 11:08:39 +00:00
|
|
|
|
int
|
2000-02-18 20:57:33 +00:00
|
|
|
|
ata_probe(device_t dev)
|
|
|
|
|
|
{
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
struct ata_channel *ch;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!dev || !(ch = device_get_softc(dev)))
|
2000-02-18 20:57:33 +00:00
|
|
|
|
return ENXIO;
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
|
2003-03-29 13:37:09 +00:00
|
|
|
|
if (ch->r_irq)
|
2001-03-19 08:04:54 +00:00
|
|
|
|
return EEXIST;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2000-02-18 20:57:33 +00:00
|
|
|
|
/* initialize the softc basics */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->active = ATA_IDLE;
|
|
|
|
|
|
ch->dev = dev;
|
|
|
|
|
|
ch->device[MASTER].channel = ch;
|
|
|
|
|
|
ch->device[MASTER].unit = ATA_MASTER;
|
|
|
|
|
|
ch->device[MASTER].mode = ATA_PIO;
|
|
|
|
|
|
ch->device[SLAVE].channel = ch;
|
|
|
|
|
|
ch->device[SLAVE].unit = ATA_SLAVE;
|
|
|
|
|
|
ch->device[SLAVE].mode = ATA_PIO;
|
|
|
|
|
|
TAILQ_INIT(&ch->ata_queue);
|
|
|
|
|
|
TAILQ_INIT(&ch->atapi_queue);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
|
|
|
|
|
|
/* initialise device(s) on this channel */
|
|
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
|
|
|
|
|
ata_reset(ch);
|
|
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2000-02-18 20:57:33 +00:00
|
|
|
|
return 0;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2001-03-06 21:43:46 +00:00
|
|
|
|
int
|
2000-02-18 20:57:33 +00:00
|
|
|
|
ata_attach(device_t dev)
|
2000-01-18 21:02:59 +00:00
|
|
|
|
{
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
struct ata_channel *ch;
|
2001-02-06 16:44:25 +00:00
|
|
|
|
int error, rid;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!dev || !(ch = device_get_softc(dev)))
|
2000-02-18 20:57:33 +00:00
|
|
|
|
return ENXIO;
|
|
|
|
|
|
|
2001-02-06 16:44:25 +00:00
|
|
|
|
rid = ATA_IRQ_RID;
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->r_irq = bus_alloc_resource(dev, SYS_RES_IRQ, &rid, 0, ~0, 1,
|
|
|
|
|
|
RF_SHAREABLE | RF_ACTIVE);
|
|
|
|
|
|
if (!ch->r_irq) {
|
|
|
|
|
|
ata_printf(ch, -1, "unable to allocate interrupt\n");
|
2000-02-18 20:57:33 +00:00
|
|
|
|
return ENXIO;
|
|
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((error = bus_setup_intr(dev, ch->r_irq, INTR_TYPE_BIO | INTR_ENTROPY,
|
2002-03-03 15:36:21 +00:00
|
|
|
|
ata_intr, ch, &ch->ih))) {
|
|
|
|
|
|
ata_printf(ch, -1, "unable to setup interrupt\n");
|
2000-02-25 09:48:23 +00:00
|
|
|
|
return error;
|
2002-03-03 15:36:21 +00:00
|
|
|
|
}
|
2000-01-18 21:02:59 +00:00
|
|
|
|
|
2003-02-20 20:02:32 +00:00
|
|
|
|
if (ch->dma)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ch->dma->alloc(ch);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
|
2000-01-18 21:02:59 +00:00
|
|
|
|
/*
|
2000-02-18 20:57:33 +00:00
|
|
|
|
* do not attach devices if we are in early boot, this is done later
|
|
|
|
|
|
* when interrupts are enabled by a hook into the boot process.
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
* otherwise attach what the probe has found in ch->devices.
|
2000-01-18 21:02:59 +00:00
|
|
|
|
*/
|
2000-02-18 20:57:33 +00:00
|
|
|
|
if (!ata_delayed_attach) {
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->devices & ATA_ATA_SLAVE)
|
|
|
|
|
|
if (ata_getparam(&ch->device[SLAVE], ATA_C_ATA_IDENTIFY))
|
|
|
|
|
|
ch->devices &= ~ATA_ATA_SLAVE;
|
|
|
|
|
|
if (ch->devices & ATA_ATAPI_SLAVE)
|
|
|
|
|
|
if (ata_getparam(&ch->device[SLAVE], ATA_C_ATAPI_IDENTIFY))
|
|
|
|
|
|
ch->devices &= ~ATA_ATAPI_SLAVE;
|
|
|
|
|
|
if (ch->devices & ATA_ATA_MASTER)
|
|
|
|
|
|
if (ata_getparam(&ch->device[MASTER], ATA_C_ATA_IDENTIFY))
|
|
|
|
|
|
ch->devices &= ~ATA_ATA_MASTER;
|
|
|
|
|
|
if (ch->devices & ATA_ATAPI_MASTER)
|
2002-03-08 11:31:37 +00:00
|
|
|
|
if (ata_getparam(&ch->device[MASTER], ATA_C_ATAPI_IDENTIFY))
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->devices &= ~ATA_ATAPI_MASTER;
|
2001-01-29 10:01:11 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
2003-01-27 09:04:29 +00:00
|
|
|
|
if (ch->devices & ATA_ATA_MASTER)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ad_attach(&ch->device[MASTER]);
|
|
|
|
|
|
if (ch->devices & ATA_ATA_SLAVE)
|
|
|
|
|
|
ad_attach(&ch->device[SLAVE]);
|
2000-02-18 20:57:33 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->devices & ATA_ATAPI_MASTER)
|
|
|
|
|
|
atapi_attach(&ch->device[MASTER]);
|
|
|
|
|
|
if (ch->devices & ATA_ATAPI_SLAVE)
|
|
|
|
|
|
atapi_attach(&ch->device[SLAVE]);
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#endif
|
|
|
|
|
|
#ifdef DEV_ATAPICAM
|
2002-10-22 20:18:51 +00:00
|
|
|
|
atapi_cam_attach_bus(ch);
|
2000-02-18 20:57:33 +00:00
|
|
|
|
#endif
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2000-01-18 21:02:59 +00:00
|
|
|
|
}
|
2000-02-18 20:57:33 +00:00
|
|
|
|
return 0;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2001-03-06 21:43:46 +00:00
|
|
|
|
int
|
2000-02-18 20:57:33 +00:00
|
|
|
|
ata_detach(device_t dev)
|
|
|
|
|
|
{
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
struct ata_channel *ch;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
int s;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
|
2003-03-29 13:37:09 +00:00
|
|
|
|
if (!dev || !(ch = device_get_softc(dev)) || !ch->r_irq)
|
2000-02-18 20:57:33 +00:00
|
|
|
|
return ENXIO;
|
2000-01-18 21:02:59 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
/* make sure channel is not busy */
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
2002-03-11 21:04:32 +00:00
|
|
|
|
ATA_SLEEPLOCK_CH(ch, ATA_CONTROL);
|
2001-03-06 21:43:46 +00:00
|
|
|
|
|
2002-03-11 21:04:32 +00:00
|
|
|
|
s = splbio();
|
2001-01-29 10:01:11 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->devices & ATA_ATA_MASTER && ch->device[MASTER].driver)
|
|
|
|
|
|
ad_detach(&ch->device[MASTER], 1);
|
|
|
|
|
|
if (ch->devices & ATA_ATA_SLAVE && ch->device[SLAVE].driver)
|
|
|
|
|
|
ad_detach(&ch->device[SLAVE], 1);
|
2000-01-19 07:25:46 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->devices & ATA_ATAPI_MASTER && ch->device[MASTER].driver)
|
|
|
|
|
|
atapi_detach(&ch->device[MASTER]);
|
|
|
|
|
|
if (ch->devices & ATA_ATAPI_SLAVE && ch->device[SLAVE].driver)
|
|
|
|
|
|
atapi_detach(&ch->device[SLAVE]);
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#endif
|
|
|
|
|
|
#ifdef DEV_ATAPICAM
|
2002-10-22 20:18:51 +00:00
|
|
|
|
atapi_cam_detach_bus(ch);
|
2000-01-19 07:25:46 +00:00
|
|
|
|
#endif
|
2002-03-11 21:04:32 +00:00
|
|
|
|
splx(s);
|
2001-02-06 16:44:25 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->device[MASTER].param) {
|
|
|
|
|
|
free(ch->device[MASTER].param, M_ATA);
|
|
|
|
|
|
ch->device[MASTER].param = NULL;
|
2000-01-18 21:02:59 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->device[SLAVE].param) {
|
|
|
|
|
|
free(ch->device[SLAVE].param, M_ATA);
|
|
|
|
|
|
ch->device[SLAVE].param = NULL;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->device[MASTER].driver = NULL;
|
|
|
|
|
|
ch->device[SLAVE].driver = NULL;
|
|
|
|
|
|
ch->device[MASTER].mode = ATA_PIO;
|
|
|
|
|
|
ch->device[SLAVE].mode = ATA_PIO;
|
|
|
|
|
|
ch->devices = 0;
|
2003-02-20 20:02:32 +00:00
|
|
|
|
if (ch->dma)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ch->dma->free(ch);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
|
|
|
|
|
|
bus_teardown_intr(dev, ch->r_irq, ch->ih);
|
|
|
|
|
|
bus_release_resource(dev, SYS_RES_IRQ, ATA_IRQ_RID, ch->r_irq);
|
|
|
|
|
|
ch->r_irq = NULL;
|
2002-03-11 21:04:32 +00:00
|
|
|
|
ATA_UNLOCK_CH(ch);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2000-02-18 20:57:33 +00:00
|
|
|
|
return 0;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2001-03-06 21:43:46 +00:00
|
|
|
|
int
|
2000-02-18 20:57:33 +00:00
|
|
|
|
ata_resume(device_t dev)
|
|
|
|
|
|
{
|
2002-12-03 20:20:44 +00:00
|
|
|
|
struct ata_channel *ch;
|
|
|
|
|
|
int error;
|
|
|
|
|
|
|
|
|
|
|
|
if (!dev || !(ch = device_get_softc(dev)))
|
|
|
|
|
|
return ENXIO;
|
|
|
|
|
|
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
2002-12-03 20:20:44 +00:00
|
|
|
|
error = ata_reinit(ch);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2002-12-03 20:20:44 +00:00
|
|
|
|
return error;
|
2000-01-18 21:02:59 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2001-03-15 15:36:25 +00:00
|
|
|
|
static int
|
2001-09-12 08:38:13 +00:00
|
|
|
|
ataioctl(dev_t dev, u_long cmd, caddr_t addr, int32_t flag, struct thread *td)
|
2001-03-15 15:36:25 +00:00
|
|
|
|
{
|
2001-05-17 10:29:30 +00:00
|
|
|
|
struct ata_cmd *iocmd = (struct ata_cmd *)addr;
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
struct ata_channel *ch;
|
2002-03-03 15:36:21 +00:00
|
|
|
|
device_t device = devclass_get_device(ata_devclass, iocmd->channel);
|
2002-03-11 21:04:32 +00:00
|
|
|
|
int error;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
2001-05-17 10:29:30 +00:00
|
|
|
|
if (cmd != IOCATA)
|
|
|
|
|
|
return ENOTTY;
|
|
|
|
|
|
|
2003-03-22 12:18:20 +00:00
|
|
|
|
if (iocmd->cmd == ATAGMAXCHANNEL) {
|
|
|
|
|
|
iocmd->u.maxchan = devclass_get_maxunit(ata_devclass);
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2002-03-03 15:36:21 +00:00
|
|
|
|
if (iocmd->channel < -1 || iocmd->device < -1 || iocmd->device > SLAVE)
|
2001-05-17 10:29:30 +00:00
|
|
|
|
return ENXIO;
|
|
|
|
|
|
|
|
|
|
|
|
switch (iocmd->cmd) {
|
2002-03-03 15:36:21 +00:00
|
|
|
|
case ATAATTACH:
|
2001-03-15 15:36:25 +00:00
|
|
|
|
/* should enable channel HW on controller that can SOS XXX */
|
2001-08-21 11:35:47 +00:00
|
|
|
|
error = ata_probe(device);
|
|
|
|
|
|
if (!error)
|
2001-03-15 15:36:25 +00:00
|
|
|
|
error = ata_attach(device);
|
2001-08-21 11:35:47 +00:00
|
|
|
|
return error;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
2002-03-03 15:36:21 +00:00
|
|
|
|
case ATADETACH:
|
2001-05-17 10:29:30 +00:00
|
|
|
|
error = ata_detach(device);
|
2001-03-15 15:36:25 +00:00
|
|
|
|
/* should disable channel HW on controller that can SOS XXX */
|
2001-08-21 11:35:47 +00:00
|
|
|
|
return error;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
2002-03-03 15:36:21 +00:00
|
|
|
|
case ATAREINIT:
|
|
|
|
|
|
if (!device || !(ch = device_get_softc(device)))
|
|
|
|
|
|
return ENXIO;
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
2002-03-11 21:04:32 +00:00
|
|
|
|
ATA_SLEEPLOCK_CH(ch, ATA_ACTIVE);
|
2002-12-03 20:20:44 +00:00
|
|
|
|
error = ata_reinit(ch);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2001-08-21 11:35:47 +00:00
|
|
|
|
return error;
|
2002-03-03 15:36:21 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
case ATAGMODE:
|
2002-03-03 15:36:21 +00:00
|
|
|
|
if (!device || !(ch = device_get_softc(device)))
|
|
|
|
|
|
return ENXIO;
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
|
|
|
|
|
|
if ((iocmd->device == MASTER || iocmd->device == -1) &&
|
|
|
|
|
|
ch->device[MASTER].driver)
|
|
|
|
|
|
iocmd->u.mode.mode[MASTER] = ch->device[MASTER].mode;
|
2001-05-17 10:29:30 +00:00
|
|
|
|
else
|
|
|
|
|
|
iocmd->u.mode.mode[MASTER] = -1;
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
|
|
|
|
|
|
if ((iocmd->device == SLAVE || iocmd->device == -1) &&
|
|
|
|
|
|
ch->device[SLAVE].param)
|
|
|
|
|
|
iocmd->u.mode.mode[SLAVE] = ch->device[SLAVE].mode;
|
2001-05-17 10:29:30 +00:00
|
|
|
|
else
|
|
|
|
|
|
iocmd->u.mode.mode[SLAVE] = -1;
|
2001-08-21 11:35:47 +00:00
|
|
|
|
return 0;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
case ATASMODE:
|
2002-03-03 15:36:21 +00:00
|
|
|
|
if (!device || !(ch = device_get_softc(device)))
|
|
|
|
|
|
return ENXIO;
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((iocmd->device == MASTER || iocmd->device == -1) &&
|
|
|
|
|
|
iocmd->u.mode.mode[MASTER] >= 0 && ch->device[MASTER].param) {
|
|
|
|
|
|
ata_change_mode(&ch->device[MASTER],iocmd->u.mode.mode[MASTER]);
|
|
|
|
|
|
iocmd->u.mode.mode[MASTER] = ch->device[MASTER].mode;
|
2001-03-21 11:44:58 +00:00
|
|
|
|
}
|
2001-05-17 10:29:30 +00:00
|
|
|
|
else
|
|
|
|
|
|
iocmd->u.mode.mode[MASTER] = -1;
|
|
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((iocmd->device == SLAVE || iocmd->device == -1) &&
|
|
|
|
|
|
iocmd->u.mode.mode[SLAVE] >= 0 && ch->device[SLAVE].param) {
|
|
|
|
|
|
ata_change_mode(&ch->device[SLAVE], iocmd->u.mode.mode[SLAVE]);
|
|
|
|
|
|
iocmd->u.mode.mode[SLAVE] = ch->device[SLAVE].mode;
|
2001-03-21 11:44:58 +00:00
|
|
|
|
}
|
2001-05-17 10:29:30 +00:00
|
|
|
|
else
|
|
|
|
|
|
iocmd->u.mode.mode[SLAVE] = -1;
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
return 0;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
case ATAGPARM:
|
2002-03-03 15:36:21 +00:00
|
|
|
|
if (!device || !(ch = device_get_softc(device)))
|
|
|
|
|
|
return ENXIO;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
2001-05-17 10:29:30 +00:00
|
|
|
|
iocmd->u.param.type[MASTER] =
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->devices & (ATA_ATA_MASTER | ATA_ATAPI_MASTER);
|
2001-05-17 10:29:30 +00:00
|
|
|
|
iocmd->u.param.type[SLAVE] =
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->devices & (ATA_ATA_SLAVE | ATA_ATAPI_SLAVE);
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->device[MASTER].name)
|
|
|
|
|
|
strcpy(iocmd->u.param.name[MASTER], ch->device[MASTER].name);
|
|
|
|
|
|
if (ch->device[SLAVE].name)
|
|
|
|
|
|
strcpy(iocmd->u.param.name[SLAVE], ch->device[SLAVE].name);
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->device[MASTER].param)
|
|
|
|
|
|
bcopy(ch->device[MASTER].param, &iocmd->u.param.params[MASTER],
|
2001-03-15 15:36:25 +00:00
|
|
|
|
sizeof(struct ata_params));
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->device[SLAVE].param)
|
|
|
|
|
|
bcopy(ch->device[SLAVE].param, &iocmd->u.param.params[SLAVE],
|
2001-03-15 15:36:25 +00:00
|
|
|
|
sizeof(struct ata_params));
|
2001-08-21 11:35:47 +00:00
|
|
|
|
return 0;
|
2001-08-30 09:47:17 +00:00
|
|
|
|
|
2002-03-30 16:36:41 +00:00
|
|
|
|
case ATAENCSTAT: {
|
|
|
|
|
|
struct ata_device *atadev;
|
|
|
|
|
|
|
|
|
|
|
|
if (!device || !(ch = device_get_softc(device)))
|
|
|
|
|
|
return ENXIO;
|
|
|
|
|
|
|
|
|
|
|
|
if (iocmd->device == SLAVE)
|
|
|
|
|
|
atadev = &ch->device[SLAVE];
|
|
|
|
|
|
else
|
|
|
|
|
|
atadev = &ch->device[MASTER];
|
|
|
|
|
|
|
2003-01-27 09:04:29 +00:00
|
|
|
|
return ata_enclosure_status(atadev,
|
|
|
|
|
|
&iocmd->u.enclosure.fan,
|
|
|
|
|
|
&iocmd->u.enclosure.temp,
|
|
|
|
|
|
&iocmd->u.enclosure.v05,
|
|
|
|
|
|
&iocmd->u.enclosure.v12);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2002-03-27 10:58:59 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
|
|
|
|
|
case ATARAIDREBUILD:
|
|
|
|
|
|
return ata_raid_rebuild(iocmd->channel);
|
|
|
|
|
|
|
|
|
|
|
|
case ATARAIDCREATE:
|
|
|
|
|
|
return ata_raid_create(&iocmd->u.raid_setup);
|
|
|
|
|
|
|
|
|
|
|
|
case ATARAIDDELETE:
|
|
|
|
|
|
return ata_raid_delete(iocmd->channel);
|
2002-04-02 13:48:17 +00:00
|
|
|
|
|
|
|
|
|
|
case ATARAIDSTATUS:
|
|
|
|
|
|
return ata_raid_status(iocmd->channel, &iocmd->u.raid_status);
|
2002-03-27 10:58:59 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
2002-03-10 21:03:04 +00:00
|
|
|
|
case ATAPICMD: {
|
|
|
|
|
|
struct ata_device *atadev;
|
|
|
|
|
|
caddr_t buf;
|
2001-08-30 09:47:17 +00:00
|
|
|
|
|
2002-03-03 15:36:21 +00:00
|
|
|
|
if (!device || !(ch = device_get_softc(device)))
|
|
|
|
|
|
return ENXIO;
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
|
|
|
|
|
|
if (!(atadev = &ch->device[iocmd->device]) ||
|
|
|
|
|
|
!(ch->devices & (iocmd->device == MASTER ?
|
|
|
|
|
|
ATA_ATAPI_MASTER : ATA_ATAPI_SLAVE)))
|
2001-08-30 09:47:17 +00:00
|
|
|
|
return ENODEV;
|
|
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!(buf = malloc(iocmd->u.atapi.count, M_ATA, M_NOWAIT)))
|
2001-08-30 09:47:17 +00:00
|
|
|
|
return ENOMEM;
|
|
|
|
|
|
|
|
|
|
|
|
if (iocmd->u.atapi.flags & ATAPI_CMD_WRITE) {
|
|
|
|
|
|
error = copyin(iocmd->u.atapi.data, buf, iocmd->u.atapi.count);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
if (error) {
|
|
|
|
|
|
free(buf, M_ATA);
|
2001-08-30 09:47:17 +00:00
|
|
|
|
return error;
|
2003-02-20 20:02:32 +00:00
|
|
|
|
}
|
2001-08-30 09:47:17 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
error = atapi_queue_cmd(atadev, iocmd->u.atapi.ccb,
|
2001-08-30 09:47:17 +00:00
|
|
|
|
buf, iocmd->u.atapi.count,
|
|
|
|
|
|
(iocmd->u.atapi.flags == ATAPI_CMD_READ ?
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ATPR_F_READ : 0) | ATPR_F_QUIET,
|
2001-08-30 09:47:17 +00:00
|
|
|
|
iocmd->u.atapi.timeout, NULL, NULL);
|
|
|
|
|
|
if (error) {
|
|
|
|
|
|
iocmd->u.atapi.error = error;
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
bcopy(&atadev->result, iocmd->u.atapi.sense_data,
|
2001-08-30 09:47:17 +00:00
|
|
|
|
sizeof(struct atapi_reqsense));
|
|
|
|
|
|
error = 0;
|
|
|
|
|
|
}
|
|
|
|
|
|
else if (iocmd->u.atapi.flags & ATAPI_CMD_READ)
|
|
|
|
|
|
error = copyout(buf, iocmd->u.atapi.data, iocmd->u.atapi.count);
|
|
|
|
|
|
|
|
|
|
|
|
free(buf, M_ATA);
|
|
|
|
|
|
return error;
|
2002-03-10 21:03:04 +00:00
|
|
|
|
}
|
2001-09-01 08:47:11 +00:00
|
|
|
|
#endif
|
2002-03-03 15:36:21 +00:00
|
|
|
|
default:
|
2002-03-19 12:14:14 +00:00
|
|
|
|
break;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
}
|
2001-08-21 11:35:47 +00:00
|
|
|
|
return ENOTTY;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2000-09-19 11:08:39 +00:00
|
|
|
|
static int
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_getparam(struct ata_device *atadev, u_int8_t command)
|
2000-01-18 21:02:59 +00:00
|
|
|
|
{
|
|
|
|
|
|
struct ata_params *ata_parm;
|
2000-01-24 20:45:24 +00:00
|
|
|
|
int retry = 0;
|
2000-01-18 21:02:59 +00:00
|
|
|
|
|
2002-03-26 09:31:22 +00:00
|
|
|
|
if (!(ata_parm = malloc(sizeof(struct ata_params), M_ATA, M_NOWAIT))) {
|
|
|
|
|
|
ata_prtdev(atadev, "malloc for identify data failed\n");
|
|
|
|
|
|
return -1;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2000-01-24 20:45:24 +00:00
|
|
|
|
/* apparently some devices needs this repeated */
|
|
|
|
|
|
do {
|
2003-02-14 13:10:40 +00:00
|
|
|
|
if (ata_command(atadev, command, 0, 0, 0,
|
2003-03-29 13:37:09 +00:00
|
|
|
|
dumping ? ATA_WAIT_READY : ATA_WAIT_INTR)) { /* XXX */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_prtdev(atadev, "%s identify failed\n",
|
2001-10-03 08:25:11 +00:00
|
|
|
|
command == ATA_C_ATAPI_IDENTIFY ? "ATAPI" : "ATA");
|
2002-03-26 09:31:22 +00:00
|
|
|
|
free(ata_parm, M_ATA);
|
2000-01-24 20:45:24 +00:00
|
|
|
|
return -1;
|
|
|
|
|
|
}
|
2000-01-25 20:14:51 +00:00
|
|
|
|
if (retry++ > 4) {
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_prtdev(atadev, "%s identify retries exceeded\n",
|
2001-10-03 08:25:11 +00:00
|
|
|
|
command == ATA_C_ATAPI_IDENTIFY ? "ATAPI" : "ATA");
|
2002-03-26 09:31:22 +00:00
|
|
|
|
free(ata_parm, M_ATA);
|
2000-01-24 20:45:24 +00:00
|
|
|
|
return -1;
|
|
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
} while (ata_wait(atadev, ((command == ATA_C_ATAPI_IDENTIFY) ?
|
|
|
|
|
|
ATA_S_DRQ : (ATA_S_READY|ATA_S_DSC|ATA_S_DRQ))));
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_INSW(atadev->channel, ATA_DATA, (int16_t *)ata_parm,
|
2001-09-20 15:25:36 +00:00
|
|
|
|
sizeof(struct ata_params)/sizeof(int16_t));
|
|
|
|
|
|
|
2000-01-18 21:02:59 +00:00
|
|
|
|
if (command == ATA_C_ATA_IDENTIFY ||
|
|
|
|
|
|
!((ata_parm->model[0] == 'N' && ata_parm->model[1] == 'E') ||
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
(ata_parm->model[0] == 'F' && ata_parm->model[1] == 'X') ||
|
|
|
|
|
|
(ata_parm->model[0] == 'P' && ata_parm->model[1] == 'i')))
|
|
|
|
|
|
bswap(ata_parm->model, sizeof(ata_parm->model));
|
2000-01-18 21:02:59 +00:00
|
|
|
|
btrim(ata_parm->model, sizeof(ata_parm->model));
|
|
|
|
|
|
bpack(ata_parm->model, ata_parm->model, sizeof(ata_parm->model));
|
|
|
|
|
|
bswap(ata_parm->revision, sizeof(ata_parm->revision));
|
|
|
|
|
|
btrim(ata_parm->revision, sizeof(ata_parm->revision));
|
|
|
|
|
|
bpack(ata_parm->revision, ata_parm->revision, sizeof(ata_parm->revision));
|
2002-07-31 18:29:34 +00:00
|
|
|
|
bswap(ata_parm->serial, sizeof(ata_parm->serial));
|
|
|
|
|
|
btrim(ata_parm->serial, sizeof(ata_parm->serial));
|
|
|
|
|
|
bpack(ata_parm->serial, ata_parm->serial, sizeof(ata_parm->serial));
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
atadev->param = ata_parm;
|
2000-01-18 21:02:59 +00:00
|
|
|
|
return 0;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2000-02-18 20:57:33 +00:00
|
|
|
|
static void
|
|
|
|
|
|
ata_boot_attach(void)
|
|
|
|
|
|
{
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
struct ata_channel *ch;
|
2000-09-19 11:08:39 +00:00
|
|
|
|
int ctlr;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
|
* run through all ata devices and look for real ATA & ATAPI devices
|
|
|
|
|
|
* using the hints we found in the early probe, this avoids some of
|
|
|
|
|
|
* the delays probing of non-exsistent devices can cause.
|
|
|
|
|
|
*/
|
|
|
|
|
|
for (ctlr=0; ctlr<devclass_get_maxunit(ata_devclass); ctlr++) {
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!(ch = devclass_get_softc(ata_devclass, ctlr)))
|
2000-02-18 20:57:33 +00:00
|
|
|
|
continue;
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->devices & ATA_ATA_SLAVE)
|
|
|
|
|
|
if (ata_getparam(&ch->device[SLAVE], ATA_C_ATA_IDENTIFY))
|
|
|
|
|
|
ch->devices &= ~ATA_ATA_SLAVE;
|
|
|
|
|
|
if (ch->devices & ATA_ATAPI_SLAVE)
|
|
|
|
|
|
if (ata_getparam(&ch->device[SLAVE], ATA_C_ATAPI_IDENTIFY))
|
|
|
|
|
|
ch->devices &= ~ATA_ATAPI_SLAVE;
|
|
|
|
|
|
if (ch->devices & ATA_ATA_MASTER)
|
|
|
|
|
|
if (ata_getparam(&ch->device[MASTER], ATA_C_ATA_IDENTIFY))
|
|
|
|
|
|
ch->devices &= ~ATA_ATA_MASTER;
|
|
|
|
|
|
if (ch->devices & ATA_ATAPI_MASTER)
|
|
|
|
|
|
if (ata_getparam(&ch->device[MASTER], ATA_C_ATAPI_IDENTIFY))
|
|
|
|
|
|
ch->devices &= ~ATA_ATAPI_MASTER;
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2000-02-18 20:57:33 +00:00
|
|
|
|
}
|
2001-01-29 10:01:11 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
2000-02-18 20:57:33 +00:00
|
|
|
|
/* now we know whats there, do the real attach, first the ATA disks */
|
|
|
|
|
|
for (ctlr=0; ctlr<devclass_get_maxunit(ata_devclass); ctlr++) {
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!(ch = devclass_get_softc(ata_devclass, ctlr)))
|
2000-02-18 20:57:33 +00:00
|
|
|
|
continue;
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->devices & ATA_ATA_MASTER)
|
|
|
|
|
|
ad_attach(&ch->device[MASTER]);
|
|
|
|
|
|
if (ch->devices & ATA_ATA_SLAVE)
|
|
|
|
|
|
ad_attach(&ch->device[SLAVE]);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2000-02-18 20:57:33 +00:00
|
|
|
|
}
|
|
|
|
|
|
#endif
|
|
|
|
|
|
/* then the atapi devices */
|
|
|
|
|
|
for (ctlr=0; ctlr<devclass_get_maxunit(ata_devclass); ctlr++) {
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!(ch = devclass_get_softc(ata_devclass, ctlr)))
|
2000-02-18 20:57:33 +00:00
|
|
|
|
continue;
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->devices & ATA_ATAPI_MASTER)
|
|
|
|
|
|
atapi_attach(&ch->device[MASTER]);
|
|
|
|
|
|
if (ch->devices & ATA_ATAPI_SLAVE)
|
|
|
|
|
|
atapi_attach(&ch->device[SLAVE]);
|
2000-02-18 20:57:33 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#ifdef DEV_ATAPICAM
|
2002-10-22 20:18:51 +00:00
|
|
|
|
atapi_cam_attach_bus(ch);
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#endif
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2002-08-09 20:54:06 +00:00
|
|
|
|
}
|
2002-12-03 20:20:44 +00:00
|
|
|
|
if (ata_delayed_attach) {
|
|
|
|
|
|
config_intrhook_disestablish(ata_delayed_attach);
|
|
|
|
|
|
free(ata_delayed_attach, M_TEMP);
|
|
|
|
|
|
ata_delayed_attach = NULL;
|
|
|
|
|
|
}
|
2002-12-06 19:29:53 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
2002-12-03 20:20:44 +00:00
|
|
|
|
ata_raid_attach();
|
2002-12-06 19:29:53 +00:00
|
|
|
|
#endif
|
2000-02-18 20:57:33 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
1999-03-03 21:10:29 +00:00
|
|
|
|
static void
|
2000-02-18 20:57:33 +00:00
|
|
|
|
ata_intr(void *data)
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
{
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
struct ata_channel *ch = (struct ata_channel *)data;
|
2000-01-03 10:26:59 +00:00
|
|
|
|
|
2003-02-20 20:02:32 +00:00
|
|
|
|
/* if device is busy it didn't interrupt */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
if (ATA_IDX_INB(ch, ATA_ALTSTAT) & ATA_S_BUSY) {
|
2001-03-19 08:04:54 +00:00
|
|
|
|
DELAY(100);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
if (!(ATA_IDX_INB(ch, ATA_ALTSTAT) & ATA_S_DRQ))
|
2001-03-19 08:04:54 +00:00
|
|
|
|
return;
|
|
|
|
|
|
}
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2000-09-19 11:08:39 +00:00
|
|
|
|
/* clear interrupt and get status */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ch->status = ATA_IDX_INB(ch, ATA_STATUS);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->status & ATA_S_ERROR)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ch->error = ATA_IDX_INB(ch, ATA_ERROR);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
/* find & call the responsible driver to process this interrupt */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
switch (ch->active) {
|
2001-01-29 10:01:11 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
case ATA_ACTIVE_ATA:
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!ch->running || ad_interrupt(ch->running) == ATA_OP_CONTINUES)
|
1999-11-06 16:50:21 +00:00
|
|
|
|
return;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
break;
|
1999-03-03 21:10:29 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
case ATA_ACTIVE_ATAPI:
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!ch->running || atapi_interrupt(ch->running) == ATA_OP_CONTINUES)
|
1999-11-06 16:50:21 +00:00
|
|
|
|
return;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
break;
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
#endif
|
2000-09-19 11:08:39 +00:00
|
|
|
|
default:
|
2002-07-22 18:35:01 +00:00
|
|
|
|
if (ch->active & ATA_WAIT_INTR)
|
2003-03-02 16:54:40 +00:00
|
|
|
|
wakeup(ch);
|
2002-07-22 18:35:01 +00:00
|
|
|
|
}
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
2002-07-22 18:35:01 +00:00
|
|
|
|
if (ch->active & ATA_CONTROL) {
|
|
|
|
|
|
ATA_FORCELOCK_CH(ch, ATA_CONTROL);
|
|
|
|
|
|
return;
|
2000-09-19 11:08:39 +00:00
|
|
|
|
}
|
2002-07-22 18:35:01 +00:00
|
|
|
|
|
2002-12-03 20:20:44 +00:00
|
|
|
|
if (ch->active & ATA_WAIT_INTR) {
|
|
|
|
|
|
ATA_UNLOCK_CH(ch);
|
|
|
|
|
|
return;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2002-07-22 18:35:01 +00:00
|
|
|
|
if ((ch->flags & ATA_QUEUED) &&
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_INB(ch, ATA_ALTSTAT) & ATA_S_SERVICE) {
|
2002-07-22 18:35:01 +00:00
|
|
|
|
ATA_FORCELOCK_CH(ch, ATA_ACTIVE);
|
|
|
|
|
|
if (ata_service(ch) == ATA_OP_CONTINUES)
|
|
|
|
|
|
return;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->running = NULL;
|
2002-12-03 20:20:44 +00:00
|
|
|
|
ATA_UNLOCK_CH(ch);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_start(ch);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
return;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_start(struct ata_channel *ch)
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
{
|
2001-01-29 10:01:11 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
Nine'th update to the new ATA/ATAPI driver:
The atapi subsystem has gotten better error handeling and timeouts,
it also tries a REQUEST SENSE command when devices returns errors,
to give a little more info as to what went wrong. It might be a
little verbose for now, but I'm interested in as much feedback on
errors as possible, especially timeouts, as I'm a bit in doubt if
I've chosen resonable default values everywhere.
The disk driver has been changed a bit to prepare for tagged queing,
which is next on my list.
The disk driver has grown a dump routine, I got one implementation
from Darrell Anderson <anderson@cs.duke.edu> which also did
partial dumps (usefull on big memory machines) I left out the
partial stuff for now, and changed the rest alot to fit into the new
ad_request framework.
Some minor cleanups and rearrangements as well.
As usual USE AT YOUR OWN RISK!!, this is still pre alpha level code.
Especially the DMA support can hose your disk real bad if anything
goes wrong, again you have been warned :)
Notebook owners should be carefull that their machines dont suspend
as this might cause trouble...
But please tell me how it works for you!
Enjoy!
-Søren
1999-06-25 09:03:07 +00:00
|
|
|
|
struct ad_request *ad_request;
|
2000-02-04 10:20:22 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
struct atapi_request *atapi_request;
|
2000-02-04 10:20:22 +00:00
|
|
|
|
#endif
|
2002-03-11 21:04:32 +00:00
|
|
|
|
int s;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_LOCK);
|
2002-03-11 21:04:32 +00:00
|
|
|
|
if (!ATA_LOCK_CH(ch, ATA_ACTIVE))
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
return;
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
2002-03-11 21:04:32 +00:00
|
|
|
|
s = splbio();
|
2001-01-29 10:01:11 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
1999-10-09 19:57:13 +00:00
|
|
|
|
/* find & call the responsible driver if anything on the ATA queue */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (TAILQ_EMPTY(&ch->ata_queue)) {
|
|
|
|
|
|
if (ch->devices & (ATA_ATA_MASTER) && ch->device[MASTER].driver)
|
|
|
|
|
|
ad_start(&ch->device[MASTER]);
|
|
|
|
|
|
if (ch->devices & (ATA_ATA_SLAVE) && ch->device[SLAVE].driver)
|
|
|
|
|
|
ad_start(&ch->device[SLAVE]);
|
2000-03-13 12:04:58 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((ad_request = TAILQ_FIRST(&ch->ata_queue))) {
|
|
|
|
|
|
TAILQ_REMOVE(&ch->ata_queue, ad_request, chain);
|
|
|
|
|
|
ch->active = ATA_ACTIVE_ATA;
|
|
|
|
|
|
ch->running = ad_request;
|
2002-03-11 21:04:32 +00:00
|
|
|
|
if (ad_transfer(ad_request) == ATA_OP_CONTINUES) {
|
|
|
|
|
|
splx(s);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
return;
|
2002-03-11 21:04:32 +00:00
|
|
|
|
}
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
Fourth update to the new ATA/ATAPI driver:
Well, better late than newer, but things has been hectic
around here, sorry for the long delay.
DMA support has been added to the ATA disk driver.
This only works on Intel PIIX3/4, Acer Aladdin and Promise controllers.
The promise support works without the BIOS on the board,
and timing modes are set to support up to UDMA speed. This
solves the problems with having more than one promise controller
in the same system.
There is support for "generic" DMA, that might work on other
controllers, but now you have been warned :)
More chipset specific code will come soon, I have to find testers
with the approbiate HW, more on that when I have it ready.
The system now uses its own major numbers, please run MAKEDEV
with the devices you need (ad?, acd?, afd?, ast?).
For now the disk driver will also attach to the old wd major
so one can at least boot without this step, but be warned, this
will eventually go away. The bootblocks will have to be changed
before one can boot directly from an "ad" device though.
Fixed problems:
All known hang problems should be solved
The probe code has been sligthly changed, this should solve
the reports I have lying around (I hope).
Hangs when accessing ata & atapi device on the same channel simultaniously.
A real braino in ata_start caused this, fixed.
As usual USE AT YOUR OWN RISK!!, this is still pre alpha level code.
Especially the DMA support can hose your disk real bad if anything
goes wrong, agaiin you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-03-28 18:57:20 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
2000-03-18 22:42:45 +00:00
|
|
|
|
/* find & call the responsible driver if anything on the ATAPI queue */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (TAILQ_EMPTY(&ch->atapi_queue)) {
|
|
|
|
|
|
if (ch->devices & (ATA_ATAPI_MASTER) && ch->device[MASTER].driver)
|
|
|
|
|
|
atapi_start(&ch->device[MASTER]);
|
|
|
|
|
|
if (ch->devices & (ATA_ATAPI_SLAVE) && ch->device[SLAVE].driver)
|
|
|
|
|
|
atapi_start(&ch->device[SLAVE]);
|
2000-03-13 12:04:58 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((atapi_request = TAILQ_FIRST(&ch->atapi_queue))) {
|
|
|
|
|
|
TAILQ_REMOVE(&ch->atapi_queue, atapi_request, chain);
|
|
|
|
|
|
ch->active = ATA_ACTIVE_ATAPI;
|
|
|
|
|
|
ch->running = atapi_request;
|
2002-03-11 21:04:32 +00:00
|
|
|
|
if (atapi_transfer(atapi_request) == ATA_OP_CONTINUES) {
|
|
|
|
|
|
splx(s);
|
2001-04-05 11:15:24 +00:00
|
|
|
|
return;
|
2002-03-11 21:04:32 +00:00
|
|
|
|
}
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
#endif
|
2002-03-14 21:35:55 +00:00
|
|
|
|
ATA_UNLOCK_CH(ch);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->locking(ch, ATA_LF_UNLOCK);
|
2002-07-22 18:35:01 +00:00
|
|
|
|
splx(s);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_reset(struct ata_channel *ch)
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
{
|
2001-03-14 12:05:44 +00:00
|
|
|
|
u_int8_t lsb, msb, ostat0, ostat1;
|
2001-12-25 14:43:08 +00:00
|
|
|
|
u_int8_t stat0 = 0, stat1 = 0;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
int mask = 0, timeout;
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
|
2001-03-14 12:05:44 +00:00
|
|
|
|
/* do we have any signs of ATA/ATAPI HW being present ? */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_DRIVE, ATA_D_IBM | ATA_MASTER);
|
2001-03-14 14:00:09 +00:00
|
|
|
|
DELAY(10);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ostat0 = ATA_IDX_INB(ch, ATA_STATUS);
|
2003-02-25 08:22:01 +00:00
|
|
|
|
if ((ostat0 & 0xf8) != 0xf8 && ostat0 != 0xa5) {
|
2001-12-25 14:43:08 +00:00
|
|
|
|
stat0 = ATA_S_BUSY;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
mask |= 0x01;
|
2001-12-25 14:43:08 +00:00
|
|
|
|
}
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_DRIVE, ATA_D_IBM | ATA_SLAVE);
|
2001-03-14 14:00:09 +00:00
|
|
|
|
DELAY(10);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ostat1 = ATA_IDX_INB(ch, ATA_STATUS);
|
2003-02-25 08:22:01 +00:00
|
|
|
|
if ((ostat1 & 0xf8) != 0xf8 && ostat1 != 0xa5) {
|
2001-12-25 14:43:08 +00:00
|
|
|
|
stat1 = ATA_S_BUSY;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
mask |= 0x02;
|
2001-12-25 14:43:08 +00:00
|
|
|
|
}
|
2001-03-14 12:05:44 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->devices = 0;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (!mask)
|
|
|
|
|
|
return;
|
2000-11-12 20:40:05 +00:00
|
|
|
|
|
|
|
|
|
|
/* in some setups we dont want to test for a slave */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ch->flags & ATA_NO_SLAVE) {
|
2001-12-25 14:43:08 +00:00
|
|
|
|
stat1 = 0x0;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
mask &= ~0x02;
|
2001-12-25 14:43:08 +00:00
|
|
|
|
}
|
2000-11-12 20:40:05 +00:00
|
|
|
|
|
2000-11-16 10:52:00 +00:00
|
|
|
|
if (bootverbose)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ata_printf(ch, -1, "pre reset mask=%02x ostat0=%02x ostat2=%02x\n",
|
2001-03-14 12:05:44 +00:00
|
|
|
|
mask, ostat0, ostat1);
|
2000-11-12 20:40:05 +00:00
|
|
|
|
|
|
|
|
|
|
/* reset channel */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_DRIVE, ATA_D_IBM | ATA_MASTER);
|
2001-10-03 08:25:11 +00:00
|
|
|
|
DELAY(10);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_ALTSTAT, ATA_A_IDS | ATA_A_RESET);
|
2000-11-08 21:25:43 +00:00
|
|
|
|
DELAY(10000);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_ALTSTAT, ATA_A_IDS);
|
2000-11-08 21:25:43 +00:00
|
|
|
|
DELAY(100000);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_INB(ch, ATA_ERROR);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
|
|
|
|
|
|
/* wait for BUSY to go inactive */
|
1999-10-09 19:57:13 +00:00
|
|
|
|
for (timeout = 0; timeout < 310000; timeout++) {
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (stat0 & ATA_S_BUSY) {
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_DRIVE, ATA_D_IBM | ATA_MASTER);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(10);
|
2002-09-12 14:32:33 +00:00
|
|
|
|
|
|
|
|
|
|
/* check for ATAPI signature while its still there */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
lsb = ATA_IDX_INB(ch, ATA_CYL_LSB);
|
|
|
|
|
|
msb = ATA_IDX_INB(ch, ATA_CYL_MSB);
|
|
|
|
|
|
stat0 = ATA_IDX_INB(ch, ATA_STATUS);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!(stat0 & ATA_S_BUSY)) {
|
2000-11-16 10:52:00 +00:00
|
|
|
|
if (bootverbose)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_printf(ch, ATA_MASTER, "ATAPI %02x %02x\n", lsb, msb);
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (lsb == ATAPI_MAGIC_LSB && msb == ATAPI_MAGIC_MSB)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->devices |= ATA_ATAPI_MASTER;
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
if (stat1 & ATA_S_BUSY) {
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_DRIVE, ATA_D_IBM | ATA_SLAVE);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(10);
|
2002-09-12 14:32:33 +00:00
|
|
|
|
|
|
|
|
|
|
/* check for ATAPI signature while its still there */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
lsb = ATA_IDX_INB(ch, ATA_CYL_LSB);
|
|
|
|
|
|
msb = ATA_IDX_INB(ch, ATA_CYL_MSB);
|
|
|
|
|
|
stat1 = ATA_IDX_INB(ch, ATA_STATUS);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!(stat1 & ATA_S_BUSY)) {
|
2000-11-16 10:52:00 +00:00
|
|
|
|
if (bootverbose)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_printf(ch, ATA_SLAVE, "ATAPI %02x %02x\n", lsb, msb);
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (lsb == ATAPI_MAGIC_LSB && msb == ATAPI_MAGIC_MSB)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->devices |= ATA_ATAPI_SLAVE;
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (mask == 0x01) /* wait for master only */
|
|
|
|
|
|
if (!(stat0 & ATA_S_BUSY))
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
break;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (mask == 0x02) /* wait for slave only */
|
|
|
|
|
|
if (!(stat1 & ATA_S_BUSY))
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
break;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (mask == 0x03) /* wait for both master & slave */
|
|
|
|
|
|
if (!(stat0 & ATA_S_BUSY) && !(stat1 & ATA_S_BUSY))
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
break;
|
2000-08-11 10:34:22 +00:00
|
|
|
|
DELAY(100);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
}
|
2000-11-12 20:40:05 +00:00
|
|
|
|
DELAY(10);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_ALTSTAT, ATA_A_4BIT);
|
2000-11-08 19:31:39 +00:00
|
|
|
|
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (stat0 & ATA_S_BUSY)
|
|
|
|
|
|
mask &= ~0x01;
|
|
|
|
|
|
if (stat1 & ATA_S_BUSY)
|
|
|
|
|
|
mask &= ~0x02;
|
2000-11-16 10:52:00 +00:00
|
|
|
|
if (bootverbose)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ata_printf(ch, -1, "after reset mask=%02x stat0=%02x stat1=%02x\n",
|
2001-03-14 12:05:44 +00:00
|
|
|
|
mask, stat0, stat1);
|
|
|
|
|
|
if (!mask)
|
2000-11-08 19:31:39 +00:00
|
|
|
|
return;
|
|
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (mask & 0x01 && ostat0 != 0x00 && !(ch->devices & ATA_ATAPI_MASTER)) {
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_DRIVE, ATA_D_IBM | ATA_MASTER);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(10);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_ERROR, 0x58);
|
|
|
|
|
|
ATA_IDX_OUTB(ch, ATA_CYL_LSB, 0xa5);
|
|
|
|
|
|
lsb = ATA_IDX_INB(ch, ATA_ERROR);
|
|
|
|
|
|
msb = ATA_IDX_INB(ch, ATA_CYL_LSB);
|
2000-11-16 10:52:00 +00:00
|
|
|
|
if (bootverbose)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_printf(ch, ATA_MASTER, "ATA %02x %02x\n", lsb, msb);
|
|
|
|
|
|
if (lsb != 0x58 && msb == 0xa5)
|
|
|
|
|
|
ch->devices |= ATA_ATA_MASTER;
|
2000-11-08 19:31:39 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (mask & 0x02 && ostat1 != 0x00 && !(ch->devices & ATA_ATAPI_SLAVE)) {
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_DRIVE, ATA_D_IBM | ATA_SLAVE);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(10);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch, ATA_ERROR, 0x58);
|
|
|
|
|
|
ATA_IDX_OUTB(ch, ATA_CYL_LSB, 0xa5);
|
|
|
|
|
|
lsb = ATA_IDX_INB(ch, ATA_ERROR);
|
|
|
|
|
|
msb = ATA_IDX_INB(ch, ATA_CYL_LSB);
|
2000-11-16 10:52:00 +00:00
|
|
|
|
if (bootverbose)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_printf(ch, ATA_SLAVE, "ATA %02x %02x\n", lsb, msb);
|
|
|
|
|
|
if (lsb != 0x58 && msb == 0xa5)
|
|
|
|
|
|
ch->devices |= ATA_ATA_SLAVE;
|
2000-11-08 19:31:39 +00:00
|
|
|
|
}
|
2000-11-16 10:52:00 +00:00
|
|
|
|
if (bootverbose)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_printf(ch, -1, "devices=%02x\n", ch->devices);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2000-09-19 11:08:39 +00:00
|
|
|
|
int
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_reinit(struct ata_channel *ch)
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
{
|
2001-03-14 12:05:44 +00:00
|
|
|
|
int devices, misdev, newdev;
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
|
2002-12-03 20:20:44 +00:00
|
|
|
|
|
2003-03-29 13:37:09 +00:00
|
|
|
|
if (!ch->r_irq)
|
2001-03-15 15:36:25 +00:00
|
|
|
|
return ENXIO;
|
2002-04-05 18:37:04 +00:00
|
|
|
|
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_FORCELOCK_CH(ch, ATA_CONTROL);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->running = NULL;
|
|
|
|
|
|
devices = ch->devices;
|
2002-07-22 18:35:01 +00:00
|
|
|
|
ata_printf(ch, -1, "resetting devices ..\n");
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_reset(ch);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((misdev = devices & ~ch->devices)) {
|
2001-01-29 10:01:11 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (misdev & ATA_ATA_MASTER && ch->device[MASTER].driver)
|
|
|
|
|
|
ad_detach(&ch->device[MASTER], 0);
|
|
|
|
|
|
if (misdev & ATA_ATA_SLAVE && ch->device[SLAVE].driver)
|
|
|
|
|
|
ad_detach(&ch->device[SLAVE], 0);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (misdev & ATA_ATAPI_MASTER && ch->device[MASTER].driver)
|
|
|
|
|
|
atapi_detach(&ch->device[MASTER]);
|
|
|
|
|
|
if (misdev & ATA_ATAPI_SLAVE && ch->device[SLAVE].driver)
|
|
|
|
|
|
atapi_detach(&ch->device[SLAVE]);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
#endif
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (misdev & ATA_ATA_MASTER || misdev & ATA_ATAPI_MASTER) {
|
2002-03-03 15:36:21 +00:00
|
|
|
|
if (ch->device[MASTER].param)
|
|
|
|
|
|
free(ch->device[MASTER].param, M_ATA);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->device[MASTER].param = NULL;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
}
|
|
|
|
|
|
if (misdev & ATA_ATA_SLAVE || misdev & ATA_ATAPI_SLAVE) {
|
2002-03-03 15:36:21 +00:00
|
|
|
|
if (ch->device[SLAVE].param)
|
|
|
|
|
|
free(ch->device[SLAVE].param, M_ATA);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ch->device[SLAVE].param = NULL;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
}
|
|
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((newdev = ~devices & ch->devices)) {
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (newdev & ATA_ATA_MASTER)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ata_getparam(&ch->device[MASTER], ATA_C_ATA_IDENTIFY))
|
2002-12-01 14:20:08 +00:00
|
|
|
|
ch->devices &= ~ATA_ATA_MASTER;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (newdev & ATA_ATA_SLAVE)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ata_getparam(&ch->device[SLAVE], ATA_C_ATA_IDENTIFY))
|
2002-12-01 14:20:08 +00:00
|
|
|
|
ch->devices &= ~ATA_ATA_SLAVE;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (newdev & ATA_ATAPI_MASTER)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ata_getparam(&ch->device[MASTER], ATA_C_ATAPI_IDENTIFY))
|
2002-12-01 14:20:08 +00:00
|
|
|
|
ch->devices &= ~ATA_ATAPI_MASTER;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
if (newdev & ATA_ATAPI_SLAVE)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ata_getparam(&ch->device[SLAVE], ATA_C_ATAPI_IDENTIFY))
|
2002-12-01 14:20:08 +00:00
|
|
|
|
ch->devices &= ~ATA_ATAPI_SLAVE;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
}
|
2002-12-01 14:20:08 +00:00
|
|
|
|
newdev = ~devices & ch->devices;
|
2001-03-14 12:05:44 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (newdev & ATA_ATA_SLAVE && !ch->device[SLAVE].driver)
|
|
|
|
|
|
ad_attach(&ch->device[SLAVE]);
|
2002-02-12 13:21:51 +00:00
|
|
|
|
else if (ch->devices & (ATA_ATA_SLAVE) && ch->device[SLAVE].driver) {
|
|
|
|
|
|
ata_getparam(&ch->device[SLAVE], ATA_C_ATA_IDENTIFY);
|
|
|
|
|
|
ad_reinit(&ch->device[SLAVE]);
|
|
|
|
|
|
}
|
2003-02-20 20:02:32 +00:00
|
|
|
|
if (newdev & ATA_ATA_MASTER && !ch->device[MASTER].driver)
|
|
|
|
|
|
ad_attach(&ch->device[MASTER]);
|
|
|
|
|
|
else if (ch->devices & ATA_ATA_MASTER && ch->device[MASTER].driver) {
|
|
|
|
|
|
ata_getparam(&ch->device[MASTER], ATA_C_ATA_IDENTIFY);
|
|
|
|
|
|
ad_reinit(&ch->device[MASTER]);
|
|
|
|
|
|
}
|
2001-03-14 12:05:44 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#if DEV_ATAPIALL
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (newdev & ATA_ATAPI_SLAVE && !ch->device[SLAVE].driver)
|
|
|
|
|
|
atapi_attach(&ch->device[SLAVE]);
|
2002-02-12 13:21:51 +00:00
|
|
|
|
else if (ch->devices & (ATA_ATAPI_SLAVE) && ch->device[SLAVE].driver) {
|
|
|
|
|
|
ata_getparam(&ch->device[SLAVE], ATA_C_ATAPI_IDENTIFY);
|
|
|
|
|
|
atapi_reinit(&ch->device[SLAVE]);
|
|
|
|
|
|
}
|
2003-02-20 20:02:32 +00:00
|
|
|
|
if (newdev & ATA_ATAPI_MASTER && !ch->device[MASTER].driver)
|
|
|
|
|
|
atapi_attach(&ch->device[MASTER]);
|
|
|
|
|
|
else if (ch->devices & (ATA_ATAPI_MASTER) && ch->device[MASTER].driver) {
|
|
|
|
|
|
ata_getparam(&ch->device[MASTER], ATA_C_ATAPI_IDENTIFY);
|
|
|
|
|
|
atapi_reinit(&ch->device[MASTER]);
|
|
|
|
|
|
}
|
2001-03-14 12:05:44 +00:00
|
|
|
|
#endif
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#ifdef DEV_ATAPICAM
|
2002-10-22 20:18:51 +00:00
|
|
|
|
atapi_cam_reinit_bus(ch);
|
2002-08-09 20:54:06 +00:00
|
|
|
|
#endif
|
2001-03-14 12:05:44 +00:00
|
|
|
|
printf("done\n");
|
2002-03-11 21:04:32 +00:00
|
|
|
|
ATA_UNLOCK_CH(ch);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_start(ch);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
return 0;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2000-09-19 11:08:39 +00:00
|
|
|
|
static int
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_service(struct ata_channel *ch)
|
2000-09-19 11:08:39 +00:00
|
|
|
|
{
|
|
|
|
|
|
/* do we have a SERVICE request from the drive ? */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((ch->status & (ATA_S_SERVICE|ATA_S_ERROR|ATA_S_DRQ)) == ATA_S_SERVICE) {
|
2003-02-20 20:02:32 +00:00
|
|
|
|
#if 0 /* XXX */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(ch->r_bmio, ATA_BMSTAT_PORT,
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ch->dma->status(ch) | ATA_BMSTAT_INTERRUPT);
|
|
|
|
|
|
#endif
|
2001-01-29 10:01:11 +00:00
|
|
|
|
#ifdef DEV_ATADISK
|
2003-03-29 13:37:09 +00:00
|
|
|
|
if ((ATA_IDX_INB(ch, ATA_DRIVE) & ATA_SLAVE) == ATA_MASTER) {
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((ch->devices & ATA_ATA_MASTER) && ch->device[MASTER].driver)
|
|
|
|
|
|
return ad_service((struct ad_softc *)
|
|
|
|
|
|
ch->device[MASTER].driver, 0);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
}
|
|
|
|
|
|
else {
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((ch->devices & ATA_ATA_SLAVE) && ch->device[SLAVE].driver)
|
|
|
|
|
|
return ad_service((struct ad_softc *)
|
|
|
|
|
|
ch->device[SLAVE].driver, 0);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
}
|
|
|
|
|
|
#endif
|
|
|
|
|
|
}
|
|
|
|
|
|
return ATA_OP_FINISHED;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_wait(struct ata_device *atadev, u_int8_t mask)
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
{
|
2000-09-19 11:08:39 +00:00
|
|
|
|
int timeout = 0;
|
2000-01-17 02:04:19 +00:00
|
|
|
|
|
1999-12-13 10:19:39 +00:00
|
|
|
|
DELAY(1);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
while (timeout < 5000000) { /* timeout 5 secs */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
atadev->channel->status = ATA_IDX_INB(atadev->channel, ATA_STATUS);
|
Fourth update to the new ATA/ATAPI driver:
Well, better late than newer, but things has been hectic
around here, sorry for the long delay.
DMA support has been added to the ATA disk driver.
This only works on Intel PIIX3/4, Acer Aladdin and Promise controllers.
The promise support works without the BIOS on the board,
and timing modes are set to support up to UDMA speed. This
solves the problems with having more than one promise controller
in the same system.
There is support for "generic" DMA, that might work on other
controllers, but now you have been warned :)
More chipset specific code will come soon, I have to find testers
with the approbiate HW, more on that when I have it ready.
The system now uses its own major numbers, please run MAKEDEV
with the devices you need (ad?, acd?, afd?, ast?).
For now the disk driver will also attach to the old wd major
so one can at least boot without this step, but be warned, this
will eventually go away. The bootblocks will have to be changed
before one can boot directly from an "ad" device though.
Fixed problems:
All known hang problems should be solved
The probe code has been sligthly changed, this should solve
the reports I have lying around (I hope).
Hangs when accessing ata & atapi device on the same channel simultaniously.
A real braino in ata_start caused this, fixed.
As usual USE AT YOUR OWN RISK!!, this is still pre alpha level code.
Especially the DMA support can hose your disk real bad if anything
goes wrong, agaiin you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-03-28 18:57:20 +00:00
|
|
|
|
|
|
|
|
|
|
/* if drive fails status, reselect the drive just to be sure */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (atadev->channel->status == 0xff) {
|
|
|
|
|
|
ata_prtdev(atadev, "no status, reselecting device\n");
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE, ATA_D_IBM|atadev->unit);
|
2001-03-14 14:00:09 +00:00
|
|
|
|
DELAY(10);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
atadev->channel->status = ATA_IDX_INB(atadev->channel, ATA_STATUS);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (atadev->channel->status == 0xff)
|
2001-03-14 14:00:09 +00:00
|
|
|
|
return -1;
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
}
|
2000-01-17 02:04:19 +00:00
|
|
|
|
|
|
|
|
|
|
/* are we done ? */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (!(atadev->channel->status & ATA_S_BUSY))
|
2000-01-17 02:04:19 +00:00
|
|
|
|
break;
|
|
|
|
|
|
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
if (timeout > 1000) {
|
|
|
|
|
|
timeout += 1000;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
DELAY(1000);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
}
|
|
|
|
|
|
else {
|
|
|
|
|
|
timeout += 10;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
DELAY(10);
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
}
|
2000-01-17 02:04:19 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (atadev->channel->status & ATA_S_ERROR)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
atadev->channel->error = ATA_IDX_INB(atadev->channel, ATA_ERROR);
|
2000-01-17 02:04:19 +00:00
|
|
|
|
if (timeout >= 5000000)
|
|
|
|
|
|
return -1;
|
|
|
|
|
|
if (!mask)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
return (atadev->channel->status & ATA_S_ERROR);
|
2000-01-17 02:04:19 +00:00
|
|
|
|
|
|
|
|
|
|
/* Wait 50 msec for bits wanted. */
|
|
|
|
|
|
timeout = 5000;
|
|
|
|
|
|
while (timeout--) {
|
2003-03-29 13:37:09 +00:00
|
|
|
|
atadev->channel->status = ATA_IDX_INB(atadev->channel, ATA_STATUS);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((atadev->channel->status & mask) == mask) {
|
|
|
|
|
|
if (atadev->channel->status & ATA_S_ERROR)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
atadev->channel->error=ATA_IDX_INB(atadev->channel, ATA_ERROR);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
return (atadev->channel->status & ATA_S_ERROR);
|
2000-01-17 02:04:19 +00:00
|
|
|
|
}
|
|
|
|
|
|
DELAY (10);
|
|
|
|
|
|
}
|
|
|
|
|
|
return -1;
|
|
|
|
|
|
}
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2000-09-19 11:08:39 +00:00
|
|
|
|
int
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_command(struct ata_device *atadev, u_int8_t command,
|
2002-10-01 15:21:57 +00:00
|
|
|
|
u_int64_t lba, u_int16_t count, u_int16_t feature, int flags)
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
{
|
2000-09-19 11:08:39 +00:00
|
|
|
|
int error = 0;
|
Fourth update to the new ATA/ATAPI driver:
Well, better late than newer, but things has been hectic
around here, sorry for the long delay.
DMA support has been added to the ATA disk driver.
This only works on Intel PIIX3/4, Acer Aladdin and Promise controllers.
The promise support works without the BIOS on the board,
and timing modes are set to support up to UDMA speed. This
solves the problems with having more than one promise controller
in the same system.
There is support for "generic" DMA, that might work on other
controllers, but now you have been warned :)
More chipset specific code will come soon, I have to find testers
with the approbiate HW, more on that when I have it ready.
The system now uses its own major numbers, please run MAKEDEV
with the devices you need (ad?, acd?, afd?, ast?).
For now the disk driver will also attach to the old wd major
so one can at least boot without this step, but be warned, this
will eventually go away. The bootblocks will have to be changed
before one can boot directly from an "ad" device though.
Fixed problems:
All known hang problems should be solved
The probe code has been sligthly changed, this should solve
the reports I have lying around (I hope).
Hangs when accessing ata & atapi device on the same channel simultaniously.
A real braino in ata_start caused this, fixed.
As usual USE AT YOUR OWN RISK!!, this is still pre alpha level code.
Especially the DMA support can hose your disk real bad if anything
goes wrong, agaiin you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-03-28 18:57:20 +00:00
|
|
|
|
#ifdef ATA_DEBUG
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_prtdev(atadev, "ata_command: addr=%04lx, cmd=%02x, "
|
2002-11-07 22:23:46 +00:00
|
|
|
|
"lba=%jd, count=%d, feature=%d, flags=%02x\n",
|
2003-03-30 09:27:59 +00:00
|
|
|
|
rman_get_start(atadev->channel->r_io[ATA_DATA].res),
|
2002-11-07 22:23:46 +00:00
|
|
|
|
command, (intmax_t)lba, count, feature, flags);
|
Fourth update to the new ATA/ATAPI driver:
Well, better late than newer, but things has been hectic
around here, sorry for the long delay.
DMA support has been added to the ATA disk driver.
This only works on Intel PIIX3/4, Acer Aladdin and Promise controllers.
The promise support works without the BIOS on the board,
and timing modes are set to support up to UDMA speed. This
solves the problems with having more than one promise controller
in the same system.
There is support for "generic" DMA, that might work on other
controllers, but now you have been warned :)
More chipset specific code will come soon, I have to find testers
with the approbiate HW, more on that when I have it ready.
The system now uses its own major numbers, please run MAKEDEV
with the devices you need (ad?, acd?, afd?, ast?).
For now the disk driver will also attach to the old wd major
so one can at least boot without this step, but be warned, this
will eventually go away. The bootblocks will have to be changed
before one can boot directly from an "ad" device though.
Fixed problems:
All known hang problems should be solved
The probe code has been sligthly changed, this should solve
the reports I have lying around (I hope).
Hangs when accessing ata & atapi device on the same channel simultaniously.
A real braino in ata_start caused this, fixed.
As usual USE AT YOUR OWN RISK!!, this is still pre alpha level code.
Especially the DMA support can hose your disk real bad if anything
goes wrong, agaiin you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-03-28 18:57:20 +00:00
|
|
|
|
#endif
|
|
|
|
|
|
|
2001-10-06 11:07:04 +00:00
|
|
|
|
/* select device */
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE, ATA_D_IBM | atadev->unit);
|
2001-10-06 11:07:04 +00:00
|
|
|
|
|
2000-09-19 11:08:39 +00:00
|
|
|
|
/* disable interrupt from device */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (atadev->channel->flags & ATA_QUEUED)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_ALTSTAT, ATA_A_IDS | ATA_A_4BIT);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
1999-03-07 21:49:14 +00:00
|
|
|
|
/* ready to issue command ? */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ata_wait(atadev, 0) < 0) {
|
|
|
|
|
|
ata_prtdev(atadev, "timeout sending command=%02x s=%02x e=%02x\n",
|
|
|
|
|
|
command, atadev->channel->status, atadev->channel->error);
|
1999-10-09 19:57:13 +00:00
|
|
|
|
return -1;
|
1999-03-05 09:43:30 +00:00
|
|
|
|
}
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
2001-10-06 11:07:04 +00:00
|
|
|
|
/* only use 48bit addressing if needed because of the overhead */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if ((lba > 268435455 || count > 256) && atadev->param &&
|
|
|
|
|
|
atadev->param->support.address48) {
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_FEATURE, (feature>>8) & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_FEATURE, feature);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_COUNT, (count>>8) & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_COUNT, count & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, (lba>>24) & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, lba & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CYL_LSB, (lba>>32) & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CYL_LSB, (lba>>8) & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CYL_MSB, (lba>>40) & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CYL_MSB, (lba>>16) & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE, ATA_D_LBA | atadev->unit);
|
2001-10-06 11:07:04 +00:00
|
|
|
|
|
|
|
|
|
|
/* translate command into 48bit version */
|
|
|
|
|
|
switch (command) {
|
|
|
|
|
|
case ATA_C_READ:
|
|
|
|
|
|
command = ATA_C_READ48; break;
|
|
|
|
|
|
case ATA_C_READ_MUL:
|
|
|
|
|
|
command = ATA_C_READ_MUL48; break;
|
|
|
|
|
|
case ATA_C_READ_DMA:
|
|
|
|
|
|
command = ATA_C_READ_DMA48; break;
|
|
|
|
|
|
case ATA_C_READ_DMA_QUEUED:
|
|
|
|
|
|
command = ATA_C_READ_DMA_QUEUED48; break;
|
|
|
|
|
|
case ATA_C_WRITE:
|
|
|
|
|
|
command = ATA_C_WRITE48; break;
|
|
|
|
|
|
case ATA_C_WRITE_MUL:
|
|
|
|
|
|
command = ATA_C_WRITE_MUL48; break;
|
|
|
|
|
|
case ATA_C_WRITE_DMA:
|
|
|
|
|
|
command = ATA_C_WRITE_DMA48; break;
|
|
|
|
|
|
case ATA_C_WRITE_DMA_QUEUED:
|
|
|
|
|
|
command = ATA_C_WRITE_DMA_QUEUED48; break;
|
|
|
|
|
|
case ATA_C_FLUSHCACHE:
|
|
|
|
|
|
command = ATA_C_FLUSHCACHE48; break;
|
|
|
|
|
|
default:
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_prtdev(atadev, "can't translate cmd to 48bit version\n");
|
2001-10-06 11:07:04 +00:00
|
|
|
|
return -1;
|
|
|
|
|
|
}
|
2003-01-19 11:47:32 +00:00
|
|
|
|
atadev->channel->flags |= ATA_48BIT_ACTIVE;
|
2001-10-06 11:07:04 +00:00
|
|
|
|
}
|
|
|
|
|
|
else {
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_FEATURE, feature);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_COUNT, count);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, lba & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CYL_LSB, (lba>>8) & 0xff);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CYL_MSB, (lba>>16) & 0xff);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (atadev->flags & ATA_D_USE_CHS)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE,
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ATA_D_IBM | atadev->unit | ((lba>>24) & 0xf));
|
2001-10-06 11:07:04 +00:00
|
|
|
|
else
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE,
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ATA_D_IBM | ATA_D_LBA | atadev->unit | ((lba>>24) &0xf));
|
2003-01-19 20:18:07 +00:00
|
|
|
|
atadev->channel->flags &= ~ATA_48BIT_ACTIVE;
|
2001-10-06 11:07:04 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
switch (flags & ATA_WAIT_MASK) {
|
|
|
|
|
|
case ATA_IMMEDIATE:
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CMD, command);
|
2001-10-06 11:07:04 +00:00
|
|
|
|
|
|
|
|
|
|
/* enable interrupt */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (atadev->channel->flags & ATA_QUEUED)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_ALTSTAT, ATA_A_4BIT);
|
2001-10-06 11:07:04 +00:00
|
|
|
|
break;
|
1999-03-05 09:43:30 +00:00
|
|
|
|
|
|
|
|
|
|
case ATA_WAIT_INTR:
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
atadev->channel->active |= ATA_WAIT_INTR;
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CMD, command);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
|
|
|
|
|
/* enable interrupt */
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (atadev->channel->flags & ATA_QUEUED)
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_ALTSTAT, ATA_A_4BIT);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
|
2003-03-02 16:54:40 +00:00
|
|
|
|
if (tsleep(atadev->channel, PRIBIO, "atacmd", 10 * hz)) {
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_prtdev(atadev, "timeout waiting for interrupt\n");
|
|
|
|
|
|
atadev->channel->active &= ~ATA_WAIT_INTR;
|
2000-09-19 11:08:39 +00:00
|
|
|
|
error = -1;
|
1999-03-07 21:49:14 +00:00
|
|
|
|
}
|
1999-03-05 09:43:30 +00:00
|
|
|
|
break;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
1999-11-06 16:50:21 +00:00
|
|
|
|
case ATA_WAIT_READY:
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
atadev->channel->active |= ATA_WAIT_READY;
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_CMD, command);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (ata_wait(atadev, ATA_S_READY) < 0) {
|
|
|
|
|
|
ata_prtdev(atadev, "timeout waiting for cmd=%02x s=%02x e=%02x\n",
|
|
|
|
|
|
command, atadev->channel->status,atadev->channel->error);
|
2000-09-19 11:08:39 +00:00
|
|
|
|
error = -1;
|
1999-11-06 16:50:21 +00:00
|
|
|
|
}
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
atadev->channel->active &= ~ATA_WAIT_READY;
|
1999-03-05 09:43:30 +00:00
|
|
|
|
break;
|
|
|
|
|
|
}
|
2000-09-19 11:08:39 +00:00
|
|
|
|
return error;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2002-03-30 16:36:41 +00:00
|
|
|
|
static void
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_start(struct ata_device *atadev)
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
{
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_INB(atadev->channel, ATA_DRIVE);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_INB(atadev->channel, ATA_DRIVE);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_INB(atadev->channel, ATA_CMD);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE, ATA_D_IBM | atadev->unit);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE, ATA_D_IBM | atadev->unit);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE, ATA_D_IBM | atadev->unit);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_INB(atadev->channel, ATA_COUNT);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_INB(atadev->channel, ATA_DRIVE);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(1);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static void
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_end(struct ata_device *atadev)
|
2002-03-30 16:36:41 +00:00
|
|
|
|
{
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_DRIVE, ATA_D_IBM | atadev->unit);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
DELAY(1);
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2002-03-30 16:36:41 +00:00
|
|
|
|
static void
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_start(struct ata_device *atadev)
|
2002-03-30 16:36:41 +00:00
|
|
|
|
{
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x0b);
|
|
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x0a);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(25);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x08);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static void
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_end(struct ata_device *atadev)
|
2002-03-30 16:36:41 +00:00
|
|
|
|
{
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x08);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(64);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x0a);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(25);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x0b);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(64);
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static u_int8_t
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_rdbit(struct ata_device *atadev)
|
2002-03-30 16:36:41 +00:00
|
|
|
|
{
|
|
|
|
|
|
u_int8_t val;
|
|
|
|
|
|
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(64);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x02);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(25);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
val = ATA_IDX_INB(atadev->channel, ATA_SECTOR) & 0x01;
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(38);
|
|
|
|
|
|
return val;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static void
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_wrbit(struct ata_device *atadev, u_int8_t data)
|
2002-03-30 16:36:41 +00:00
|
|
|
|
{
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x08 | (data & 0x01));
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(64);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_SECTOR, 0x08 | 0x02 | (data & 0x01));
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(64);
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static u_int8_t
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_rw(struct ata_device *atadev, int rw, u_int8_t val)
|
2002-03-30 16:36:41 +00:00
|
|
|
|
{
|
|
|
|
|
|
int i;
|
|
|
|
|
|
|
|
|
|
|
|
if (rw) {
|
|
|
|
|
|
for (i = 0; i < 8; i++)
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_wrbit(atadev, (val & (0x80 >> i)) ? 1 : 0);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
}
|
|
|
|
|
|
else {
|
|
|
|
|
|
for (i = 0; i < 8; i++)
|
2003-01-27 09:04:29 +00:00
|
|
|
|
val = (val << 1) | ata_enclosure_chip_rdbit(atadev);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
}
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_wrbit(atadev, 0);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
return val;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static u_int8_t
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_sensor(struct ata_device *atadev,
|
|
|
|
|
|
int rw, u_int8_t idx, u_int8_t data)
|
2002-03-30 16:36:41 +00:00
|
|
|
|
{
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_start(atadev);
|
|
|
|
|
|
ata_enclosure_chip_start(atadev);
|
|
|
|
|
|
ata_enclosure_chip_rw(atadev, 1, 0x5a);
|
|
|
|
|
|
ata_enclosure_chip_rw(atadev, 1, idx);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
if (rw) {
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_rw(atadev, 1, data);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
}
|
|
|
|
|
|
else {
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_end(atadev);
|
|
|
|
|
|
ata_enclosure_chip_start(atadev);
|
|
|
|
|
|
ata_enclosure_chip_rw(atadev, 1, 0x5b);
|
|
|
|
|
|
data = ata_enclosure_chip_rw(atadev, 0, 0);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
}
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_chip_end(atadev);
|
|
|
|
|
|
ata_enclosure_end(atadev);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
return data;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2003-01-27 09:04:29 +00:00
|
|
|
|
static int
|
|
|
|
|
|
ata_enclosure_status(struct ata_device *atadev,
|
|
|
|
|
|
int *fan, int *temp, int *v05, int *v12)
|
|
|
|
|
|
{
|
|
|
|
|
|
u_int8_t id1, id2, cnt, div;
|
2003-03-12 10:59:35 +00:00
|
|
|
|
int error = ENXIO;
|
2003-01-27 09:04:29 +00:00
|
|
|
|
|
|
|
|
|
|
if (atadev->flags & ATA_D_ENC_PRESENT) {
|
2003-02-20 20:02:32 +00:00
|
|
|
|
atadev->channel->locking(atadev->channel, ATA_LF_LOCK);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ATA_SLEEPLOCK_CH(atadev->channel, ATA_CONTROL);
|
|
|
|
|
|
ata_enclosure_sensor(atadev, 1, 0x4e, 0);
|
|
|
|
|
|
id1 = ata_enclosure_sensor(atadev, 0, 0x4f, 0);
|
|
|
|
|
|
ata_enclosure_sensor(atadev, 1, 0x4e, 0x80);
|
|
|
|
|
|
id2 = ata_enclosure_sensor(atadev, 0, 0x4f, 0);
|
2003-03-12 10:59:35 +00:00
|
|
|
|
if (id1 == 0xa3 && id2 == 0x5c) {
|
|
|
|
|
|
div = 1 << (((ata_enclosure_sensor(atadev, 0, 0x5d, 0)&0x20)>>3)+
|
|
|
|
|
|
((ata_enclosure_sensor(atadev, 0, 0x47, 0)&0x30)>>4)+1);
|
|
|
|
|
|
cnt = ata_enclosure_sensor(atadev, 0, 0x28, 0);
|
|
|
|
|
|
if (cnt == 0xff)
|
|
|
|
|
|
*fan = 0;
|
|
|
|
|
|
else
|
|
|
|
|
|
*fan = 1350000 / cnt / div;
|
|
|
|
|
|
ata_enclosure_sensor(atadev, 1, 0x4e, 0x01);
|
|
|
|
|
|
*temp = (ata_enclosure_sensor(atadev, 0, 0x50, 0) * 10) +
|
|
|
|
|
|
(ata_enclosure_sensor(atadev, 0, 0x50, 0) & 0x80 ? 5 : 0);
|
|
|
|
|
|
*v05 = ata_enclosure_sensor(atadev, 0, 0x23, 0) * 27;
|
|
|
|
|
|
*v12 = ata_enclosure_sensor(atadev, 0, 0x24, 0) * 61;
|
|
|
|
|
|
error = 0;
|
|
|
|
|
|
}
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ATA_UNLOCK_CH(atadev->channel);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
atadev->channel->locking(atadev->channel, ATA_LF_UNLOCK);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
}
|
2003-03-12 10:59:35 +00:00
|
|
|
|
return error;
|
2003-01-27 09:04:29 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2002-03-30 16:36:41 +00:00
|
|
|
|
void
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_print(struct ata_device *atadev)
|
2002-03-30 16:36:41 +00:00
|
|
|
|
{
|
2003-01-27 09:04:29 +00:00
|
|
|
|
u_int8_t id, st;
|
|
|
|
|
|
int fan, temp, v05, v12;
|
|
|
|
|
|
|
2003-02-20 20:02:32 +00:00
|
|
|
|
atadev->channel->locking(atadev->channel, ATA_LF_LOCK);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ATA_SLEEPLOCK_CH(atadev->channel, ATA_CONTROL);
|
|
|
|
|
|
ata_enclosure_start(atadev);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
id = ATA_IDX_INB(atadev->channel, ATA_DRIVE);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
st = ATA_IDX_INB(atadev->channel, ATA_COUNT);
|
2002-03-30 16:36:41 +00:00
|
|
|
|
DELAY(1);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
ata_enclosure_end(atadev);
|
|
|
|
|
|
ATA_UNLOCK_CH(atadev->channel);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
atadev->channel->locking(atadev->channel, ATA_LF_UNLOCK);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
|
|
|
|
|
|
switch (id & 0x93) {
|
|
|
|
|
|
case 0x00:
|
|
|
|
|
|
ata_prtdev(atadev, "Universal enclosure");
|
|
|
|
|
|
break;
|
|
|
|
|
|
case 0x01:
|
|
|
|
|
|
ata_prtdev(atadev, "FastSwap enclosure");
|
|
|
|
|
|
break;
|
|
|
|
|
|
case 0x10:
|
|
|
|
|
|
case 0x11:
|
|
|
|
|
|
ata_prtdev(atadev, "SuperSwap enclosure");
|
|
|
|
|
|
break;
|
|
|
|
|
|
default:
|
2003-02-20 20:02:32 +00:00
|
|
|
|
atadev->flags &= ~ATA_D_ENC_PRESENT;
|
2003-01-27 09:04:29 +00:00
|
|
|
|
return;
|
|
|
|
|
|
}
|
|
|
|
|
|
atadev->flags |= ATA_D_ENC_PRESENT;
|
|
|
|
|
|
|
2003-02-20 20:02:32 +00:00
|
|
|
|
ata_enclosure_leds(atadev, ATA_LED_GREEN);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
if (ata_enclosure_status(atadev, &fan, &temp, &v05, &v12))
|
|
|
|
|
|
printf(" detected\n");
|
|
|
|
|
|
else
|
|
|
|
|
|
printf(" [FAN:%drpm TEMP:%d.%01dC %d.%03dV %d.%03dV]\n",
|
|
|
|
|
|
fan, temp/10, temp%10, v05/1000, v05%1000, v12/1000, v12%1000);
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
|
ata_enclosure_leds(struct ata_device *atadev, u_int8_t color)
|
|
|
|
|
|
{
|
|
|
|
|
|
if (atadev->flags & ATA_D_ENC_PRESENT) {
|
|
|
|
|
|
u_int8_t reg;
|
|
|
|
|
|
|
|
|
|
|
|
ata_enclosure_start(atadev);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
reg = ATA_IDX_INB(atadev->channel, ATA_COUNT);
|
2003-01-27 09:04:29 +00:00
|
|
|
|
DELAY(1);
|
2003-03-29 13:37:09 +00:00
|
|
|
|
ATA_IDX_OUTB(atadev->channel, ATA_COUNT,
|
2003-01-27 09:04:29 +00:00
|
|
|
|
(color & ATA_LED_MASK) | (reg & ~ATA_LED_MASK));
|
|
|
|
|
|
DELAY(1);
|
|
|
|
|
|
ata_enclosure_end(atadev);
|
|
|
|
|
|
}
|
2002-03-30 16:36:41 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
static void
|
|
|
|
|
|
ata_change_mode(struct ata_device *atadev, int mode)
|
|
|
|
|
|
{
|
2002-03-11 21:04:32 +00:00
|
|
|
|
ATA_SLEEPLOCK_CH(atadev->channel, ATA_ACTIVE);
|
2003-02-20 20:02:32 +00:00
|
|
|
|
atadev->setmode(atadev, mode);
|
2002-03-11 21:04:32 +00:00
|
|
|
|
ATA_UNLOCK_CH(atadev->channel);
|
2002-12-03 20:20:44 +00:00
|
|
|
|
ata_start(atadev->channel);
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int
|
|
|
|
|
|
ata_printf(struct ata_channel *ch, int device, const char * fmt, ...)
|
|
|
|
|
|
{
|
|
|
|
|
|
va_list ap;
|
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
|
|
if (device == -1)
|
|
|
|
|
|
ret = printf("ata%d: ", device_get_unit(ch->dev));
|
|
|
|
|
|
else {
|
|
|
|
|
|
if (ch->device[ATA_DEV(device)].name)
|
|
|
|
|
|
ret = printf("%s: ", ch->device[ATA_DEV(device)].name);
|
|
|
|
|
|
else
|
|
|
|
|
|
ret = printf("ata%d-%s: ", device_get_unit(ch->dev),
|
|
|
|
|
|
(device == ATA_MASTER) ? "master" : "slave");
|
|
|
|
|
|
}
|
|
|
|
|
|
va_start(ap, fmt);
|
|
|
|
|
|
ret += vprintf(fmt, ap);
|
|
|
|
|
|
va_end(ap);
|
|
|
|
|
|
return ret;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int
|
|
|
|
|
|
ata_prtdev(struct ata_device *atadev, const char * fmt, ...)
|
|
|
|
|
|
{
|
|
|
|
|
|
va_list ap;
|
|
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
|
|
if (atadev->name)
|
|
|
|
|
|
ret = printf("%s: ", atadev->name);
|
|
|
|
|
|
else
|
|
|
|
|
|
ret = printf("ata%d-%s: ", device_get_unit(atadev->channel->dev),
|
|
|
|
|
|
(atadev->unit == ATA_MASTER) ? "master" : "slave");
|
|
|
|
|
|
va_start(ap, fmt);
|
|
|
|
|
|
ret += vprintf(fmt, ap);
|
|
|
|
|
|
va_end(ap);
|
|
|
|
|
|
return ret;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
|
ata_set_name(struct ata_device *atadev, char *name, int lun)
|
2001-03-15 15:36:25 +00:00
|
|
|
|
{
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
atadev->name = malloc(strlen(name) + 4, M_ATA, M_NOWAIT);
|
|
|
|
|
|
if (atadev->name)
|
|
|
|
|
|
sprintf(atadev->name, "%s%d", name, lun);
|
2001-03-15 15:36:25 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
ata_free_name(struct ata_device *atadev)
|
2001-03-15 15:36:25 +00:00
|
|
|
|
{
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
if (atadev->name)
|
|
|
|
|
|
free(atadev->name, M_ATA);
|
|
|
|
|
|
atadev->name = NULL;
|
2001-03-15 15:36:25 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2000-02-18 20:57:33 +00:00
|
|
|
|
int
|
2000-09-19 11:08:39 +00:00
|
|
|
|
ata_get_lun(u_int32_t *map)
|
|
|
|
|
|
{
|
|
|
|
|
|
int lun = ffs(~*map) - 1;
|
|
|
|
|
|
|
|
|
|
|
|
*map |= (1 << lun);
|
|
|
|
|
|
return lun;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2001-02-06 12:41:53 +00:00
|
|
|
|
int
|
|
|
|
|
|
ata_test_lun(u_int32_t *map, int lun)
|
|
|
|
|
|
{
|
2001-02-06 12:49:33 +00:00
|
|
|
|
return (*map & (1 << lun));
|
2001-02-06 12:41:53 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2000-09-19 11:08:39 +00:00
|
|
|
|
void
|
|
|
|
|
|
ata_free_lun(u_int32_t *map, int lun)
|
|
|
|
|
|
{
|
|
|
|
|
|
*map &= ~(1 << lun);
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
char *
|
|
|
|
|
|
ata_mode2str(int mode)
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
{
|
|
|
|
|
|
switch (mode) {
|
2000-01-27 19:00:51 +00:00
|
|
|
|
case ATA_PIO: return "BIOSPIO";
|
2000-01-03 10:26:59 +00:00
|
|
|
|
case ATA_PIO0: return "PIO0";
|
|
|
|
|
|
case ATA_PIO1: return "PIO1";
|
|
|
|
|
|
case ATA_PIO2: return "PIO2";
|
|
|
|
|
|
case ATA_PIO3: return "PIO3";
|
|
|
|
|
|
case ATA_PIO4: return "PIO4";
|
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
|
|
|
|
case ATA_DMA: return "BIOSDMA";
|
2003-02-20 20:02:32 +00:00
|
|
|
|
case ATA_WDMA0: return "WDMA0";
|
|
|
|
|
|
case ATA_WDMA1: return "WDMA1";
|
2000-01-03 10:26:59 +00:00
|
|
|
|
case ATA_WDMA2: return "WDMA2";
|
2003-02-20 20:02:32 +00:00
|
|
|
|
case ATA_UDMA0: return "UDMA16";
|
|
|
|
|
|
case ATA_UDMA1: return "UDMA25";
|
2000-01-03 10:26:59 +00:00
|
|
|
|
case ATA_UDMA2: return "UDMA33";
|
2003-02-20 20:02:32 +00:00
|
|
|
|
case ATA_UDMA3: return "UDMA40";
|
2000-01-03 10:26:59 +00:00
|
|
|
|
case ATA_UDMA4: return "UDMA66";
|
2000-08-06 19:51:58 +00:00
|
|
|
|
case ATA_UDMA5: return "UDMA100";
|
2001-10-06 11:07:04 +00:00
|
|
|
|
case ATA_UDMA6: return "UDMA133";
|
2000-01-03 10:26:59 +00:00
|
|
|
|
default: return "???";
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2000-02-18 20:57:33 +00:00
|
|
|
|
int
|
|
|
|
|
|
ata_pmode(struct ata_params *ap)
|
|
|
|
|
|
{
|
|
|
|
|
|
if (ap->atavalid & ATA_FLAG_64_70) {
|
2003-02-20 20:02:32 +00:00
|
|
|
|
if (ap->apiomodes & 0x02)
|
|
|
|
|
|
return ATA_PIO4;
|
|
|
|
|
|
if (ap->apiomodes & 0x01)
|
|
|
|
|
|
return ATA_PIO3;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
}
|
2001-10-06 11:07:04 +00:00
|
|
|
|
if (ap->retired_piomode == 2)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_PIO2;
|
2001-10-06 11:07:04 +00:00
|
|
|
|
if (ap->retired_piomode == 1)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_PIO1;
|
2001-10-06 11:07:04 +00:00
|
|
|
|
if (ap->retired_piomode == 0)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_PIO0;
|
|
|
|
|
|
if (ap->support_dma)
|
|
|
|
|
|
return ATA_PIO4;
|
|
|
|
|
|
return ATA_PIO0;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int
|
|
|
|
|
|
ata_wmode(struct ata_params *ap)
|
|
|
|
|
|
{
|
2001-10-06 11:07:04 +00:00
|
|
|
|
if (ap->mwdmamodes & 0x04)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_WDMA2;
|
2001-10-06 11:07:04 +00:00
|
|
|
|
if (ap->mwdmamodes & 0x02)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_WDMA1;
|
2001-10-06 11:07:04 +00:00
|
|
|
|
if (ap->mwdmamodes & 0x01)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_WDMA0;
|
|
|
|
|
|
if (ap->support_dma)
|
|
|
|
|
|
return ATA_WDMA2;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
return -1;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int
|
|
|
|
|
|
ata_umode(struct ata_params *ap)
|
|
|
|
|
|
{
|
|
|
|
|
|
if (ap->atavalid & ATA_FLAG_88) {
|
2001-10-06 11:07:04 +00:00
|
|
|
|
if (ap->udmamodes & 0x40)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_UDMA6;
|
2000-08-06 19:51:58 +00:00
|
|
|
|
if (ap->udmamodes & 0x20)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_UDMA5;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
if (ap->udmamodes & 0x10)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_UDMA4;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
if (ap->udmamodes & 0x08)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_UDMA3;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
if (ap->udmamodes & 0x04)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_UDMA2;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
if (ap->udmamodes & 0x02)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_UDMA1;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
if (ap->udmamodes & 0x01)
|
2003-02-20 20:02:32 +00:00
|
|
|
|
return ATA_UDMA0;
|
2000-02-18 20:57:33 +00:00
|
|
|
|
}
|
|
|
|
|
|
return -1;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2003-02-20 20:02:32 +00:00
|
|
|
|
int
|
|
|
|
|
|
ata_limit_mode(struct ata_device *atadev, int mode, int maxmode)
|
|
|
|
|
|
{
|
|
|
|
|
|
if (maxmode && mode > maxmode)
|
|
|
|
|
|
mode = maxmode;
|
|
|
|
|
|
|
|
|
|
|
|
if (mode >= ATA_UDMA0 && ata_umode(atadev->param) > 0)
|
|
|
|
|
|
return min(mode, ata_umode(atadev->param));
|
|
|
|
|
|
|
|
|
|
|
|
if (mode >= ATA_WDMA0 && ata_wmode(atadev->param) > 0)
|
|
|
|
|
|
return min(mode, ata_wmode(atadev->param));
|
|
|
|
|
|
|
|
|
|
|
|
if (mode > ata_pmode(atadev->param))
|
|
|
|
|
|
return min(mode, ata_pmode(atadev->param));
|
|
|
|
|
|
|
|
|
|
|
|
return mode;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2000-01-18 21:02:59 +00:00
|
|
|
|
static void
|
2000-09-19 11:08:39 +00:00
|
|
|
|
bswap(int8_t *buf, int len)
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
{
|
2000-02-18 20:57:33 +00:00
|
|
|
|
u_int16_t *ptr = (u_int16_t*)(buf + len);
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2000-02-18 20:57:33 +00:00
|
|
|
|
while (--ptr >= (u_int16_t*)buf)
|
|
|
|
|
|
*ptr = ntohs(*ptr);
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2000-01-18 21:02:59 +00:00
|
|
|
|
static void
|
2000-09-19 11:08:39 +00:00
|
|
|
|
btrim(int8_t *buf, int len)
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
{
|
2000-02-18 20:57:33 +00:00
|
|
|
|
int8_t *ptr;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2000-02-18 20:57:33 +00:00
|
|
|
|
for (ptr = buf; ptr < buf+len; ++ptr)
|
|
|
|
|
|
if (!*ptr)
|
|
|
|
|
|
*ptr = ' ';
|
|
|
|
|
|
for (ptr = buf + len - 1; ptr >= buf && *ptr == ' '; --ptr)
|
|
|
|
|
|
*ptr = 0;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
2000-01-18 21:02:59 +00:00
|
|
|
|
static void
|
2000-09-19 11:08:39 +00:00
|
|
|
|
bpack(int8_t *src, int8_t *dst, int len)
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
{
|
2000-09-19 11:08:39 +00:00
|
|
|
|
int i, j, blank;
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
|
2000-01-24 20:45:24 +00:00
|
|
|
|
for (i = j = blank = 0 ; i < len; i++) {
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
if (blank && src[i] == ' ') continue;
|
|
|
|
|
|
if (blank && src[i] != ' ') {
|
|
|
|
|
|
dst[j++] = src[i];
|
|
|
|
|
|
blank = 0;
|
|
|
|
|
|
continue;
|
|
|
|
|
|
}
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
if (src[i] == ' ') {
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
blank = 1;
|
Ten'th update to the new ATA/ATAPI driver:
It been awhile since the last major update, as a benefit there
are some cool things in this one (and new bugs probably :) )...
The ATA driver has grown "real" timeout support for all devices.
This means that it should be possible to get in contact with
(especially) lost ATAPI devices. It also means that the ATA
driver is now usable on notebooks as it will DTRT on resume.
An experimental hack at utilizing the Promise66's at UDMA66 is
in there, but I cant test it. If someone feels like sending
me one, give me a ping.
The ATAPI DMA enableling scheme has been changed, also better DMA
support for the Aladdin chipset has been implemented for ATAPI
devices. Note that the Aladdin apparently only can do DMA reads
on ATAPI devices, and the Promise cant do ATAPI DMA at all.
I have seen problems on some ATAPI devices that should be able
to run in DMA mode, so if you encounter problems with hanging
atapi devices during the probe, or during access, disable DMA
in atapi-all.c, and let me know. It might be nessesary to do this
via a "white list" for known good devices...
The ATAPI CDROM driver can now use eject/close without hanging and
the bug that caused reading beyond the end of a CD has been fixed.
Media change is also handled proberly. DVD drives are identified
and are usable as CDROM devices at least, I dont have the HW to
test this further, see above :).
The ATAPI tape driver has gotten some support for using the DSC
method for not blocking the IDE channel during read/write when
the device has full buffers. It knows about the OnStream DI-30
device, support is not completed yet, but it can function as a
primitive backup medium, without filemarks, and without bad media
handeling. This is because the OnStream device doesn't handle this
(like everybody else) in HW. It also now supports getting/setting
the record position on devices that supports it.
Some rather major cleanups and rearrangements as well (cvs -b diff
is your freind). I'm closing in on declaring this for beta code,
most of the infrastruture is in place by now.
As usual USE AT YOUR OWN RISK!!, this is still alpha level code.
This driver can hose your disk real bad if anything goes wrong, but
now you have been warned :)
But please tell me how it works for you!
Enjoy!
-Søren
1999-09-21 19:50:40 +00:00
|
|
|
|
if (i == 0)
|
|
|
|
|
|
continue;
|
|
|
|
|
|
}
|
Finally!!
The much roumored replacement for our current IDE/ATA/ATAPI is
materialising in the CVS repositories around the globe.
So what does this bring us:
A new reengineered ATA/ATAPI subsystem, that tries to overcome
most of the deficiencies with the current drivers.
It supports PCI as well as ISA devices without all the hackery
in ide_pci.c to make PCI devices look like ISA counterparts.
It doesn't have the excessive wait problem on probe, in fact you
shouldn't notice any delay when your devices are getting probed.
Probing and attaching of devices are postponed until interrupts
are enabled (well almost, not finished yet for disks), making
things alot cleaner.
Improved performance, although DMA support is still WIP and not
in this pre alpha release, worldstone is faster with the new
driver compared to the old even with DMA.
So what does it take away:
There is NO support for old MFM/RLL/ESDI disks.
There is NO support for bad144, if your disk is bad, ditch it, it has
already outgrown its internal spare sectors, and is dying.
For you to try this out, you will have to modify your kernel config
file to use the "ata" controller instead of all wdc? entries.
example:
# for a PCI only system (most modern machines)
controller ata0
device atadisk0 # ATA disks
device atapicd0 # ATAPI CDROM's
device atapist0 # ATAPI tapes
#You should add the following on ISA systems:
controller ata1 at isa? port "IO_WD1" bio irq 14
controller ata2 at isa? port "IO_WD2" bio irq 15
You can leave it all in there, the system knows how to manage.
For now this driver reuses the device entries from the old system
(that will probably change later), but remember that disks are
now numbered in the sequence they are found (like the SCSI system)
not as absolute positions as the old system.
Although I have tested this on all the systems I can get my hands on,
there might very well be gremlins in there, so use AT YOU OWN RISK!!
This is still WIP, so there are lots of rough edges and unfinished
things in there, and what I have in my lab might look very different
from whats in CVS at any given time. So please have all eventual
changes go through me, or chances are they just dissapears...
I would very much like to hear from you, both good and bad news
are very welcome.
Enjoy!!
-Søren
1999-03-01 21:19:19 +00:00
|
|
|
|
dst[j++] = src[i];
|
|
|
|
|
|
}
|
2000-01-24 20:45:24 +00:00
|
|
|
|
if (j < len)
|
|
|
|
|
|
dst[j] = 0x00;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2000-02-18 20:57:33 +00:00
|
|
|
|
static void
|
|
|
|
|
|
ata_init(void)
|
|
|
|
|
|
{
|
2001-03-15 15:36:25 +00:00
|
|
|
|
/* register controlling device */
|
2001-03-19 11:55:13 +00:00
|
|
|
|
make_dev(&ata_cdevsw, 0, UID_ROOT, GID_OPERATOR, 0600, "ata");
|
2001-03-15 15:36:25 +00:00
|
|
|
|
|
2000-02-18 20:57:33 +00:00
|
|
|
|
/* register boot attach to be run when interrupts are enabled */
|
|
|
|
|
|
if (!(ata_delayed_attach = (struct intr_config_hook *)
|
2000-12-26 12:05:49 +00:00
|
|
|
|
malloc(sizeof(struct intr_config_hook),
|
|
|
|
|
|
M_TEMP, M_NOWAIT | M_ZERO))) {
|
2000-02-18 20:57:33 +00:00
|
|
|
|
printf("ata: malloc of delayed attach hook failed\n");
|
|
|
|
|
|
return;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
ata_delayed_attach->ich_func = (void*)ata_boot_attach;
|
|
|
|
|
|
if (config_intrhook_establish(ata_delayed_attach) != 0) {
|
|
|
|
|
|
printf("ata: config_intrhook_establish failed\n");
|
|
|
|
|
|
free(ata_delayed_attach, M_TEMP);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
SYSINIT(atadev, SI_SUB_DRIVERS, SI_ORDER_SECOND, ata_init, NULL)
|