code that says this:
nfsm_request(vp, NFSPROC_FSSTAT, p, cred);
if (v3)
nfsm_postop_attr(vp, retattr);
if (!error)
nfsm_dissect(sfp, struct nfs_statfs *, NFSX_STATFS(v3));
The problem here is that if error != 0, nfsm_dissect() will not be
called, which leaves sfp == NULL. But nfs_statfs() does not bail out
at this point: it continues processing until it tries to dereference
sfp, which causes a panic. I was able to generate this crash under
the following conditions:
1) Set up a machine as an NFS server and NFS client, with amd running
(using NIS maps). /usr/local is exported, though any exported fs
can can be used to trigger the bug.
2) Log in as normal user, with home directory mounted from a SunOS 4.1.3
NFS server via amd (along with a few other NFS filesystems from same
machine).
3) Su to root and type the following:
# mount localhost:/usr/local /mnt
# df
To fix the panic, I changed the code to read:
if (!error) {
nfsm_dissect(sfp, struct nfs_statfs *, NFSX_STATFS(v3));
} else
goto nfsmout;
This is a bit kludgy in that nfsmout is a label defined by the nfsm_subs.h
macros, but these macros are themselves more than a little kludgy. This
stops the machine from crashing, but does not fix the overall bug: 'error'
somehow becomes 5 (EIO) when a statfs() is performed on the locally mounted
NFS filesystem. This seems to only happen the first time the filesystem
is accesed: on subsequent accesses, it seems to work fine again.
Now, I know there's no practical use in mounting a local filesystem
via NFS, but doing it shouldn't cause the system to melt down.
Specifically, don't allow a value < 1 for any of them (it doesn't make
sense), and don't let the low water mark be greater than the corresponding
high water mark.
Pre-Approved by: wollman
Obtained from: NetBSD
available to the kernel (VM_KMEM_SIZE). The default (32 MB) is too low
when having 512 MB or more physical memory in a server environment. This is
relevant on systems where "panic: kmem_malloc: kmem_map too small" is a
problem.
value (200) is too low in some environments, causing a fatal
"panic: get_pv_entry: cannot get a pv_entry_t". The same panic might
still occur due to temporary shortage of free physical memory
(cf. PR i386/2431).
Change sd_open, sd_close and sd_ioctl to use this lock to ensure
serialization of some critical operations, thus avoiding some
race conditions. Ideas picked from NetBSD (ccd and sd devices).
This fixes one of the problems noted in PR kern/3688.
Reviewed by: "Justin T. Gibbs" <gibbs@plutotech.com>
nothing good except of opening a can of (potential or real) security
holes. People maintaining a machine with higher security requirements
need to be on the console anyway, so there's no point in not forcing
them to reboot before starting maintenance.
Agreed by: hackers, guido
were returning EFAULT, when it is a completely acceptable thing to do.
Also, at the same time, be a *bit* optimizing and don't allocate any
"stackgrap" memory if we're not going to use it.
This is another Oracle-discovered problem.
Submitted by: Steven Wallace
rather than hard-code it in the message text. Optinally include
the host name in the message if SHOW_HOSTNAME is defined.
The origianl idea and sample code submitted by Angelo Turetta
<ATuretta@stylo.it>.
the XENIX version is packed, and two bytes smaller than ours. So, define
the structure, and have it packed. I used the __attribte__((packed))
modifier for this; I could also have surrounded the struct definition with
#pragma pack(2) -- but that would have meant making ibcs2_timeb's definition
outside the function. This may need to be revisited if we ever want to
compile with a compiler other than gcc. (I also used 'unsigned long'
instead of 'time_t' because I am writing to match an external specification
-- and the definition of time_t could change.)
Reviewed by: Steven Wallace
This eliminates a lot of #ifdef SMP type code. Things like _curproc reside
in a data page that is unique on each cpu, eliminating the expensive macros
like: #define curproc (SMPcurproc[cpunumber()])
There are some unresolved bootstrap and address space sharing issues at
present, but Steve is waiting on this for other work. There is still some
strictly temporary code present that isn't exactly pretty.
This is part of a larger change that has run into some bumps, this part is
standalone so it should be safe. The temporary code goes away when the
full idle cpu support is finished.
Reviewed by: fsmp, dyson
The kernel with USERCONFIG_BOOT and VISUAL_USERCONFIG option presents
the user the kernel configuration menu upon boot.
The user can navigate the menu with cursor keys. I think it would be
nice if the user can navigate and select a menu item with regular keys
as well, so that the user who is using a serial console which is not
so capable of esc sequences still can choose a menu item.
With the following patch we can select an item by typing an item
number, 1, 2, or 3, or mnemonic `s' to skip UserConfig, 'v' to enter
the visual mode, and `c' to start the CLI mode. `p', `u', `n', and `d'
will move cursor up and down.
Submitted by: yokota
adapter during the system boot. It always assumes there is at least a
monochrome adapter.
This is rather strange assumption. If there is no dispaly adapter, the
console driver cannot be any good...
In this patch, scinit() is split into two parts; the first part is
now called scvidprobe() which will detect the presence of video card
at the CGA or MONO buffer address and returns TRUE if found. It is
called during sccnprobe() and scprobe(). Both will fail if no video
card is found.
The second part, whose name stays the same as before, scinit(), is
called from sccninit() and scattach() to complete initialization of
the found video card.
The keyboard probe code is moved from scprobe() to sckbdprobe();
scprobe() now calls scvidprobe() and sckbdprobe() to carry out device
probe. (This is rather a cosmetic change, but it sure makes the code
look better organized.)
The problem pointed out by Joerg.
changes relative to the 2.2 compatable version are include file
related, the new multicast interface (!) and the new PCI interface.
This should work "as-is" but has not been tested (I have not been able
to get a dc21x4x based card for testing).
flag wasn't being respected during vref(), et. al. Note that this
isn't the eventual fix for the locking problem. Fine grained SMP
in the VM and VFS code will require (lots) more work.
out to be a problem with VJ header compression.
davidg spotted this in usr.sbin/ppp/slcompress.c
a while ago, but I believe gave the wrong reasons -
it's too easy to reproduce ! The only scenario that
I've been able to reproduce the problem under is when
m_len is *exactly* 40 ! So go figure !
PR: 3749
Submitted elsewhere by: davidg
Obtained from: usr.sbin/ppp/slcompress.c
since it is impossible to distinguish unpromoted types from small
(struct) types. Renamed __va_promote() to __va_size() since it is
related to sizes of args on the stack and not to promotion.
PR: 3884
Submitted by: mostly by arnej@math.ntnu.no (Arne Henrik Juul)
Obtained from: name of__va_size and some parentheses fixes from NetBSD
by checking whether APM is active in apm_power_off() and returning
if not. (The code was already written with the expectation that
this function would return if it fails.)
like bsd.lib.mk and bsd.prog.mk. It doesn't add it to CXXINCLUDES, I
don't think anybody has written a kernel module with C++. (Not that I
think DavidG will allow it anyway. :)
Reviewed by: bde
Remove previous hack in pppfcs().
This is still not the correct solution. We shouldn't
have any incorrect mbufs. This patch does however make
pppd/natd work (rather than jamming the interface).
NETNATM --> NATM
reported by Bruce Evans.
Bruce also pointed out that NATM is confusing since config(8) defines
NATM as the number of atm pseudo device in "BUILD_DIR/atm.h".
We might change the name in the future but leave it as it is for now.
cause a problem of spiraling death due to buffer resource limitations.
The vfs_bio code in general had little ability to handle buffer resource
management, and now it does. Also, there are a lot more knobs for tuning the
vfs_bio code now. The knobs came free because of the need that there
always be some immediately available buffers (non-delayed or locked) for
use. Note that the buffer cache code is much less likely to get bogged
down with lots of delayed writes, even more so than before.
- OVERRIDE_TUNER: allows you to manually choose the tuner type for those
cards that fail to probe properly. See source for legal
values.
- OVERRIDE_DBX: allows you to manually choose DBX or NO DBX for those
cards that fail to probe properly.
0 == no DBX circuit present, 1 == DBX circuit present.
will see it. The name of the file is obtained by appending ".config"
to the name of the kernel file. If the file doesn't exist, then an
annoying message is printed.
Reduced the command line buffer size yet again from 4096 to 1024. It is
oversized to defend against overrun in gets(), but data space is starting
to run out so the buffer size can't be as grossly oversized as before.
It is possible for multiple process to sleep concurrently waiting
for a buffer. When the buffer shortage is a shortage of space but
not a shortage of buffer headers, the processes took turns creating
empty buffers and waking each other to advertise the brelse() of
the empties; progress was never made because tsleep() always found
another high-priority process to run and everything was done at
splbio(), so vfs_update never had a chance to flush delayed writes,
not to mention that i/o never had a chance to complete.
The problem seems to be rare in practice, but it can easily be
reproduced by misusing block devices, at least for sufficently slow
devices on machines with a sufficiently small buffer cache. E.g.,
`tar cvf /dev/fd0 /kernel' on an 8MB system with no disk in fd0
causes the problem quickly; the same command with a disk in fd0
causes the problem not quite as quickly; and people have reported
problems newfs'ing file systems on block devices.
Block devices only cause this problem indirectly. They are pessimized
for time and space, and the space pessimization causes the shortage
(it manifests as internal fragmentation in buffer_map).
This should be fixed in 2.2.
should work with no driver changes, though not all features are currently
used.
Remove code that was conditional on NEW_SCSICONF not being defined. This
was temporary code, that at a time got excluded correctly, until the new
scsiconf became the default, and NEW_SCSICONF was no longer specified.
Add support for quirks defined in scsiconf.c. For now only the HP3724/5
needs an entry, since that drive can't be used with tags.
so don't enforce the MTU as an MRU. Allow bidirectional ppp MTU
negotiation, by checking against a differnt figure for MRU.
Make it large enough for ATM frames at least.
Submitted by: archie@whistle.com (archie cobbs)
- Added support for "dual" internal/serial consoles (-D flag). If -D is set,
then all i/o from the boot blocks is to and from both the internal console
and the serial console. -D has no effect on the kernel (-h decides the
kernel serial console as usual). -D should normally be set in /boot.config.
- Get help messages from /boot.help. You should copy boot.help from the
biosboot directory to the root directory on the boot drive when you
install new boot blocks.
- New, less invasive keyboard probe. Enable keyboard probe dynamically (-P
flag). Should probably never be used (use -h instead).
- Fixed/improved initialization from boot.config. It didn't interact correctly
with the NAMEBLOCK option, and the initialization of the drive/unit/partition
didn't stick when a non-default kernel name was entered.
- Don't reset or forget the default drive/unit/... or kernel name so often.
- Set the default kernel name to something unbootable after `?'.
- Display the defaults better.
- Removed PROBE_KEYBOARD_LOCK option (use -h instead).,
- Removed BOOT_FORCE_COMCONSOLE option (use device flag 0x20 instead).
- Removed -a (RB_HALT) flag. This flag is only used for reboots.
Submitted by: about 2/3 by yokota
then all i/o from the boot blocks is to and from both the internal console
and the serial console. -D has no effect on the kernel (-h decides the
kernel serial console as usual). -D should normally be set in /boot.config.
- Get help messages from /boot.help. You should copy boot.help from the
biosboot directory to the root directory on the boot drive when you
install new boot blocks.
- New, less invasive keyboard probe. Enable keyboard probe dynamically (-P
flag). Should probably never be used (use -h instead).
- Fixed/improved initialization from boot.config. It didn't interact correctly
with the NAMEBLOCK option, and the initialization of the drive/unit/partition
didn't stick when a non-default kernel name was entered.
- Don't reset or forget the default drive/unit/... or kernel name so often.
- Set the default kernel name to something unbootable after `?'.
- Display the defaults better.
- Removed PROBE_KEYBOARD_LOCK option (use -h instead).,
- Removed BOOT_FORCE_COMCONSOLE option (use device flag 0x20 instead).
- Removed -a (RB_HALT) flag. This flag is only used for reboots.
Submitted by: about 2/3 by yokota
cost since it is only done in cpu_switch(), not for every exception.
The extra state is kept in the pcb, and handled much like the npx state,
with similar deficiencies (the state is not preserved across signal
handlers, and error handling loses state).
that I snuck in to our GDB last year. This allows you to debug headless
machines by sharing the console port between the debugger and the system
console. It's not 100% reliabile, but it works well. It's optional
and disabled by default.
Submitted by: Juniper Networks
speed using the boot blocks, instead of a hardcoded value stuck in the
kernel. This way, you can have systems using the same kernel but different
console speeds.
Add a sysctl entry for changing the system console speed.
Lock the user tty speed to match the system console speed.
Nuke CONSPEED.
Reviewed by: bde
writes sent to the server were synchronous and therefore no commits are
needed. This is the same as the vfs.nfs.async variable on the server but
allows each client to choose whether to work this way.
Also make the vfs.nfs.async variable do the 'right' thing for NFSv3, i.e.
pretend that the write was synchronous.
attached to the vnode, some of them could be re-written synchronously
(if they overflowed the fixed size array nfs_flush had for them). The
fix involves mallocing an array if there are more than its limited
size stack buffer.
Reviewed by: Hidetoshi Shimokawa <simokawa@sat.t.u-tokyo.ac.jp>
These commands are required for the "Disk-At-Once" write process:
WORMIOCREADSESSIONINFO returns the length of the lead-in and lead-out areas
and WORMIOCWRITESESSION is used to send the table of contents of the disk.
device probe of a host to PCI bridge may modify that value, based on
its knowledge of device specific registers. This makes the Intel XXpress
work, as verified by: Terje Marthinussen <terjem@cc.uit.no>.
bit 10 is the old bit for MTRR (presumably this changed, an older P5 I
have has got it, the newer cpus have the new MTRR bit set)
bit 11 is SEP (fast syscalls), bit 23 is MMX
Fill in the other reserved ones with a stub so that we can see them if
they turn up.
Obtained from: Intel AP-485 rev.06
license managers to obtain the host's ethernet address as
a key.
Note that this implementation takes the first hardware address for
the first ethernet interface found, and disregards the interface name
that may be passed in, as linux ethernet devices are all "ethX".
These changes add the ability to specify that a UFS file/directory
cannot be unlinked. This is basically a scaled back version
of the IMMUTABLE flag. The reason is to allow an administrator
to create a directory hierarchy that a group of users
can arbitrarily add/delete files from, but that the hierarchy
itself is safe from removal by them.
If the NOUNLINK definition is set to 0
then this results in no change to what happens normally.
(and results in identical binary (in the kernel)).
It can be proven that if this bit is never set by the admin,
no new behaviour is introduced..
Several "good idea" comments from reviewers plus one grumble
about creeping featurism.
This code is in production in 2.2 based systems
these are quite extensive additions to the ipfw code.
they include a change to the API because the old method was
broken, but the user view is kept the same.
The new code allows a particular match to skip forward to a particular
line number, so that blocks of rules can be
used without checking all the intervening rules.
There are also many more ways of rejecting
connections especially TCP related, and
many many more ...
see the man page for a complete description.
is incompatible with -pg. (We use a different version of mcount for
profiling frame-pointer-less assembler functions, but gcc doesn't know
about this.)
Added a missing dependency.
Cleaned up trailing backslashes.
Added comment about config's limitations/bugs handling dependencies and
backslashe/newlines.
Finished removing support for isdn drivers.
shared function.
- use p->p_sleepend to try and get more accurate "time remaining" results
when the time has been adjusted.
- verify writeability of return address so that we can fail before sleeping
if the address for the result is bogus.
time for things like nanosleep. These sleep in terms of "ticks" and
calculate the elapsed time relative to the expected wakeup time and do
not return good results when the system time is adjusted.
interrupt mask hackery wasn't happening when being modloaded via the
if_ppp lkm. It seems that the lkm system doesn't particularly like having
two sets of load/unload/etc routines. :-] This really should be fixed
by having a seperate if_ppp and ppp_tty lkm, but that requires that ppp_tty
is loaded after if_ppp, and needs to be able to link with symbols in
if_ppp. This gets messy, it is a better task for the in-kernel linker.
(if_ppp is generic, ppp_tty is a tty-specific bottom end for if_ppp, it's
not _too_ hard to have another "provider" (such as a hdlc sync card)
connected to if_ppp)
the *_imask down in the isa machine dependent layers requires code changes
to all pci drivers, but the interrupt registration mechanism is in flux
at the moment. These can go away when the interface is cleaned and settled.
remove warning about ICU_LEN > 32, it's not likely to happen like this,
and besides, ICU_LEN is not the ideal indicator of "number of interrupts".
Reviewed by: bde, se
that lkm's can use them for fiddling the masks without being dependent on
which mode the kernel is compiled in (SMP or UP). This is particularly
for ppp_tty.c which has some domain crossing between the net and tty
subsystems. The values are not used in the spl code, they are for
reference only (ie: the compiled code uses immediate values rather than
an indirect 32 bit address and 32 bit data fetch).
top of the hardware interrupt handlers. Apparently this is slightly
faster with the bit scanning instruction that looks these up - this set of
changes reverts the original change.
Reviewed by: bde
rather than inlines. These are compiled with -fomit-frame-pointer and
work out pretty close to the original routines, but it might be a fraction
slower. The reason for doing this is to prevent the SWI_* and HWI_* values
from being compiled into drivers and lkms etc which is one of the things
that prevents the same lkm from being used on both SMP and UP kernels.
This gives us a lot more scope for experimenting with the splxxx
implementaton for SMP parallelism etc.
Reviewed by: bde
CPU code-named `M2'.
- Use the result of cpuid instruction instead of DIR to identify
6x86MX cpu. DIR0 and DIR1 are not documented in the data sheet, and
cpuid instruction is enabled at reset time.
- Add a function, init_6x86MX() to initialize 6x86MX cpu. It supports
CPU_SUSP_HLT and CPU_IORT options. It always sets NC1 (640K - 1M is
not cached.), and enables L1 cache in write-back mode.
- Fix typo in the comment in identblue().
Hopefully I've done the proper magic to merge changes between 1.17 and
1.17.2.1 into the main trunk. Description of those changes follows:
Brought in changes sent to me in late 1995 by Rich Murphey.
I cleaned up a few things and am currently running these under
2.2-970205-GAMMA.
The changes deal with software debouncing apparently necessary on
todays faster hardware, and also some problems with the use of the -Select
line for the TW-523 sync. This driver allows use of +PaperEnd as an
alternative.
Changes to pmap.c for lapic_t lapic && ioapic_t ioapic pointers,
currently equal to apic_base && io_apic_base, will stand alone with the
private page mapping.
apic.h has defines like:
#define lapic__id lapic->id
Once private pages and "known virtual addr" mapping of the APICs is
ready all 'lapic__XXX' will be changed to 'lapic.XXX', and the defines
will be removed.
Changes to smp.h for lapic_t lapic && ioapic_t ioapic pointers,
currently equal to apic_base && io_apic_base, will stand alone with the
private page mapping.
Return failure, if the enable bit corresponding to the map type has not
been set in the command register. This feature was requested by Justin
Gibbs, who pointed out that some early PCI to PCI bridges do not correctly
support memory windows (I assume because of the risk of deadlocks that
have been taken care of in the PCI 2.2 spec) and that some BIOS clears
the memory address decode enable bit in the command register of the PCI
device, if it finds them behind such a bridge.
1) Stop at the first map register that contains a zero value.
2) When testing for the map size work up from low values, since
this works around a bug in some BusLogic SCSI card, which has
the 16 upper port base address bits hardwired to zero.
The config register dump printed in the bootverbose case has
been slightly rearranged.
Adjust the data port address by adding the two low order bits of
the register number. The address port takes only a word address
(i.e. ignores the two low order bits written to it).
mode 1. Omission of this bit makes all config register accesses fail in
on recent chip sets ...
(The problem was reported and debug output provided by: Steve Passe)
- vector.s <- stub called by i386/exception.s
- icu_vector.s <- UP
- apic_vector.s <- SMP
Split icu.s into UP and SMP specific files:
- ipl.s <- stub called by i386/exception.s (formerly icu.s)
- icu_ipl.s <- UP
- apic_ipl.s <- SMP
This was done in preparation for massive changes to the SMP INTerrupt
mechanisms. More fine tuning, such as merging ipl.s into exception.s,
may be appropriate.
reality. There will be a new call interface, but for now the file
pci_compat.c (which is to be deleted, after all drivers are converted)
provides an emulation of the old PCI bus driver functions. The only
change that might be visible to drivers is, that the type pcici_t
(which had been meant to be just a handle, whose exact definition
should not be relied on), has been converted into a pcicfgregs* .
The Tekram AMD SCSI driver bogusly relied on the definition of pcici_t
and has been converted to just call the PCI drivers functions to access
configuration space register, instead of inventing its own ...
This code is by no means complete, but assumed to be fully operational,
and brings the official code base more in line with my development code.
A new generic device descriptor data type has to be agreed on. The PCI
code will then use that data type to provide new functionality:
1) userconfig support
2) "wired" PCI devices
3) conflicts checking against ISA/EISA
4) maps will depend on the command register enable bits
5) PCI to Anything bridges can be defined as devices,
and are probed like any "standard" PCI device.
The following features are currently missing, but will be added back,
soon:
1) unknown device probe message
2) suppression of "mirrored" devices caused by ancient, broken chip-sets
This code relies on generic shared interrupt support just commited to
kern_intr.c (plus the modifications of isa.c and isa_device.h).
be (eventually) architecture independent. It provides an emulation
of the ISA interrupt registration function register_intr(), but that
function does no longer manipulated the interrupt controller and
interrupt descriptor table, but calls the architecture dependent
function setup_icu() for that purpose.
After the ISA/EISA bus code has been modified to directly call the new
interrupt registartion functions (intr_create() and intr_connect()),
the emulation of register_intr() should be dropped.
The C level interrupt handler function should take a (void*) argument,
and the function pointer type (inthand2_t) should defined in some other
place than isa_device.h.
This commit is a pre-requisite for the removal of the PCI specific shared
interrupt code.
Reviewed by: dfr,bde
be (eventually) architecture independent. It provides an emulation
of the ISA interrupt registration function register_intr(), but that
function does no longer manipulated the interrupt controller and
interrupt descriptor table, but calls the architecture dependent
function setup_icu() for that purpose.
After theISA/EISA bus code has been modified to directly call the new
interrupt registartion functions (intr_create() and intr_connect()),
the emulation of register_intr() should be dropped.
The C level interrupt handler function should take a (void*) argument,
and the function pointer type (inthand2_t) should defined in some other
place than isa_device.h.
This commit is a pre-requisite for the removal of the PCI specific shared
interrupt code.
Reviewed by: dfr,bde
This is now the default, it delays most of the MP startup to the function
machdep.c:cpu_startup(). It should be possible to move the 2 functions
found there (mp_start() & mp_announce()) even further down the path once
we know exactly where that should be...
Help from: Peter Wemm <peter@spinner.dialix.com.au>
