1) A new mechanism has been added to prevent pages from being paged
out called "vm_page_hold". Similar to vm_page_wire, but
much lower overhead.
2) Scheduling algorithm has been changed to improve interactive
performance.
3) Paging algorithm improved.
4) Some vnode and swap pager bugs fixed.
Eliminates vm_fault overhead on process startup and
mmap referenced data for in-memory pages.
(process startup time using in-memory segments *much* faster)
2) Even more efficient pmap code. Code partially cleaned up.
More comments yet to follow.
(generally more efficient pte management)
3) Pageout clustering ( in addition to the FreeBSD V1.1 pagein
clustering.)
(much faster paging performance on non-write behind disk
subsystems, slightly faster performance on other systems.)
4) Slightly changed vm_pageout code for more efficiency and
better statistics. Also, resist swapout a little more.
(less likely to pageout a recently used page)
5) Slight improvement to the page table page trap efficiency.
(generally faster system VM fault performance)
6) Defer creation of unnamed anonymous regions pager until needed.
(speeds up shared memory bss creation)
7) Remove possible deadlock from swap_pager initialization.
8) Enhanced procfs to provide "vminfo" about vm objects and user
pmaps.
9) Increased MCLSHIFT/MCLBYTES from 2K to 4K to improve net &
socket performance and to prepare for things to come.
John Dyson
dyson@implode.root.com
David Greenman
davidg@root.com
a pv chain was being traversed while interrupts were
fully enabled in pmap_remove_all ... this is bogus, and
has been fixed in pmap.c. (sorry for adding the splimp)
Description:
The integer overflow instruction (into) and the interrupt instruction with
value 4 (int #4) both give rise to SIGBUS signals rather than SIGFPE. The
problem is that overflow is a trap not a fault (unlike the BOUND instruction).
following is a summary:
1) increased object cache back up to a more reasonable value.
2) removed old & bogus cruft from machdep.c (clearseg, copyseg,
physcopyseg, etc).
3) inlined many functions in pmap.c
4) changed "load_cr3(rcr3())" into tlbflush() and made tlbflush inline
assembly.
5) changed the way that modified pages are tracked - now vm_page struct
is kept updated directly - no more scanning page tables.
6) removed lots of unnecessary spl's
7) removed old unused functions from pmap.c
8) removed all use of page_size, page_shift, page_mask variables - replaced
with PAGE_ constants.
9) moved trunc/round_page, atop, ptoa, out of vm_param.h and into i386/
include/param.h, and optimized them.
10) numerous changes to sys/vm/ swap_pager, vnode_pager, pageout, fault
code to improve performance. LRU algorithm modified to be more
effective, read ahead/behind values tuned for better performance,
etc, etc...
problem with some chipsets (UMC) remapping the 'hole' memory even when
you've got 16MB. People were led to believe that since there was only
16MB of memory in the machine, that they were okay wrt the ISA DMA
limit. This hack simply causes the extra memory to be ignored if it
appears around the 16MB limit.
* Removed pmap_is_wired
* added extra cli/sti protection in idle (swtch.s)
* slight code improvement in trap.c
* added lots of comments
* improved paging and other algorithms in VM system
Changed the output of the isa probe routine, that only devices, that
have an IO address and are smaller than 0x100 to be on the motherboard.
The seagate SCSI adapter is an example of a card, that doesn't have
an IO address and works only memory mapped.
a binary link-kit. Make all non-optional options (pagers, procfs) standard,
and update LINT to reflect new symtab requirements.
NB: -Wtraditional will henceforth be forgotten. This editing pass was
primarily intended to detect any constructions where the old code might
have been relying on traditional C semantics or syntax. These were all
fixed, and the result of fixing some of them means that -Wall is now a
realistic possibility within a few weeks.
John Dyson to make it reliably work under FreeBSD.
2) Added and enabled PROCFS in the GENERICxx and LINT kernels.
3) New execve() from me. Still work to be done here, but this version
works well and is needed before other changes can be made. For
a description of the design behind this, see freebsd-arch or
ask me.
4) Rewrote stack fault code; made user stack VM grow as needed rather
than all up front; improves performance a little and reduces
process memory requirements.
5) Incorporated fix from Gene Stark to fault/wire a user page table
page to fix a problem in copyout. This is a temporary fix and
is not appropriate for pageable page tables. For a description
of the problem, see Gene's post to the freebsd-hackers mailing
list.
6) Tighten up vm_page struct to reduce memory requirements for it. ifdef
pager page lock code as it's not being used currently.
7) Introduced new element to vmspace struct - vm_minsaddr; initial
(minimum) stack address. Compliment to vm_maxsaddr.
8) Added a panic if the allocation for process u-pages fails.
9) Improve performance and accuracy of kernel profiling by putting in
a little inline assembly instead of spl().
10) Made serial console with sio driver work. Still has problems with
serial input, but is almost useable.
11) Added -Bstatic to SYSTEM_LD in Makefile.i386 so that kernels will
build properly with the new ld.
The following patch adds the addr argument to signal handlers.
The kernel with the patch is no more and no less in compliance or in
violation of POSIX and ANSI C than the kernel before the patch.
The added functionality this addr argument provides is quite useful. It
enables an entire class of algorithms which use mprotect to trace memory
references. Beside garbage collectors, I have heard of this technique being
applied to debuggers and profilers. The only benchmarking I've performed is
using akcl to compile maxima: without the kernel patch, it takes 7 hours to
compile maxima, while with stratified garbage collection, it only takes 50
minutes.
Basically, I can't think of a reason not to add the addr argument and there
is a compelling need for it.
If you find the patch acceptable, please let me know so I can send my
FreeBSD akcl config files to wfs for inclusion in the core akcl release.
The old 386BSD config files there won't work on either NetBSD or FreeBSD.
Subject: Page fault in PTE area fails in copyout
Index: sys/i386/i386/trap.c FreeBSD-1.0.2
Description:
Reading files of several megabytes into Emacs, or many small
files all at once, would fail with "IO error - bad address".
Repeat-By:
The bug can be exercised by a test program that malloc()'s
a 5MB chunk of memory, and then, without accessing the memory
first, filling it with data from a file using read().
(I read 64k chunks from /dev/wd0d into successive 64k regions
of the 5MB chunk.) The read() will fail with EFAULT at the first
virtual address boundary that is a multiple of 0x400000.
Fix:
The problem was code in sys/i386/i386/trap.c that tries to
figure out what kind of trap occurred and to handle it appropriately.
It was interpreting any page fault with virtual address
>= vm->vm_maxsaddr as being a user stack segment fault.
In fact, addresses >= USRSTACK are in the user structure/PTE area,
and if they are handled as stack faults, the proper PTE will
not be paged in when it is supposed to be. This situation comes
up in copyout() and copyoutstr(), if PTE's are accessed for the
first time ever. The page fault on accessing the nonexistent PTE
is mishandled as a stack fault, and then the fault that occurs on
the subsequent access to the page itself causes copyout to fail
with EFAULT.
when the machine panics.
i386/i386/locore.s:
1) got rid of most .set directives that were being used like
#define's, and replaced them with appropriate #define's in
the appropriate header files (accessed via genassym).
2) added comments to header inclusions and global definitions,
and global variables
3) replaced some hardcoded constants with cpp defines (such as
PDESIZE and others)
4) aligned all comments to the same column to make them easier to
read
5) moved macro definitions for ENTRY, ALIGN, NOP, etc. to
/sys/i386/include/asmacros.h
6) added #ifdef BDE_DEBUGGER around all of Bruce's debugger code
7) added new global '_KERNend' to store last location+1 of kernel
8) cleaned up zeroing of bss so that only bss is zeroed
9) fix zeroing of page tables so that it really does zero them all
- not just if they follow the bss.
10) rewrote page table initialization code so that 1) works correctly
and 2) write protects the kernel text by default
11) properly initialize the kernel page directory, upages, p0stack PT,
and page tables. The previous scheme was more than a bit
screwy.
12) change allocation of virtual area of IO hole so that it is
fixed at KERNBASE + 0xa0000. The previous scheme put it
right after the kernel page tables and then later expected
it to be at KERNBASE +0xa0000
13) change multiple bogus settings of user read/write of various
areas of kernel VM - including the IO hole; we should never
be accessing the IO hole in user mode through the kernel
page tables
14) split kernel support routines such as bcopy, bzero, copyin,
copyout, etc. into a seperate file 'support.s'
15) split swtch and related routines into a seperate 'swtch.s'
16) split routines related to traps, syscalls, and interrupts
into a seperate file 'exception.s'
17) remove some unused global variables from locore that got
inserted by Garrett when he pulled them out of some .h
files.
i386/isa/icu.s:
1) clean up global variable declarations
2) move in declaration of astpending and netisr
i386/i386/pmap.c:
1) fix calculation of virtual_avail. It previously was calculated
to be right in the middle of the kernel page tables - not
a good place to start allocating kernel VM.
2) properly allocate kernel page dir/tables etc out of kernel map
- previously only took out 2 pages.
i386/i386/machdep.c:
1) modify boot() to print a warning that the system will reboot in
PANIC_REBOOT_WAIT_TIME amount of seconds, and let the user
abort with a key on the console. The machine will wait for
ever if a key is typed before the reboot. The default is
15 seconds, but can be set to 0 to mean don't wait at all,
-1 to mean wait forever, or any positive value to wait for
that many seconds.
2) print "Rebooting..." just before doing it.
kern/subr_prf.c:
1) remove PANICWAIT as it is deprecated by the change to machdep.c
i386/i386/trap.c:
1) add table of trap type strings and use it to print a real trap/
panic message rather than just a number. Lot's of work to
be done here, but this is the first step. Symbolic traceback
is in the TODO.
i386/i386/Makefile.i386:
1) add support in to build support.s, exception.s and swtch.s
...and various changes to various header files to make all of the
above happen.
the stack area and not memory above VM_MAXUSER_ADDRESS.
That way, copyout and friends now work for pages whose page table entries
have not yet been allocated/been paged out.
machines (now 20% of all memory after the first 3MB). This is necessary
in order for 4MB machine to be able to rebuild the entire source tree
and not run out of physical memory because of fixed memory requirements
of processes and kernel VM.
Date: Tue, 19 Oct 1993 02:22:41 -40962758 (WST)
As the subject line says:
I can;t believe this typo is still here.
Has NOBODY used the isa_dmastart() routine for 16bit DMA?
I know I just hit the dma regs directly for the AHA1542,
and it appears that either everybody else does as well, or
they only use 8bit DMA (e.g. floppy)
Editors Note:
The definition of DMA2_CHN was incorrectly using IO_DMA1!
Remove NKMEMCLUSTERS, it is no longer define or used.
locores.s:
Fix comment on PTDpde and APTDpde to be pde instead of pte
Add new equation for calculating location of Sysmap
Remove Bill's old #ifdef garbage for counting up memory,
that stuff will never be made to work and was just cluttering
up the file.
Add code that places the PTD, page table pages, and kernel
stack below the 640k ISA hole if there is room for it, otherwise
put this stuff all at 1MB. This fixes the 28K bogusity in
the boot blocks, that can now go away!
Fix the caclulation of where first is to be dependent on
NKPDE so that we can skip over the above mentioned areas.
The 28K thing is now 44K in size due to the increase in
kernel virtual memory space, but since we no longer have
to worry about that this is no big deal.
Use if NNPX > 0 instead of ifdef NPX for floating point code.
machdep.c
Change the calculation of for the buffer cache to be
20% of all memory above 2MB and add back the upper limit
of 2/5's of the VM_KMEM_SIZE so that we do not eat ALL
of the kernel memory space on large memory machines, note
that this will not even come into effect unless you have
more than 32MB. The current buffer cache limit is 6.7MB
due to this caclulation.
It seems that we where erroniously allocating bufpages pages
for buffer_map. buffer_map is UNUSED in this implementation
of the buffer cache, but since the map is referenced in
several if statements a quick fix was to simply allocate
1 vm page (but no real memory) to it.
pmap.h
Remove rcsid, don't want them in the kernel files!
Removed some cruft inside an #ifdef DEBUGx that caused
compiler errors if you where compiling this for debug.
Use the #defines for PD_SHIFT and PG_SHIFT in place of
constants.
trap.c:
Remove patch kit header and rcsid, fix $Id$.
Now include "npx.h" and use NNPX for controlling the
floating point code.
Remove a now completly invalid check for a maximum virtual
address, the virtual address now ends at 0xFFFFFFFF so
there is no more MAX!! (Thanks David, I completly missed
that one!)
vm_machdep.c
Remove patch kit header and rcsid, fix $Id$.
Now include "npx.h" and use NNPX for controlling the
floating point code.
Replace several 0xFE00000 constants with KERNBASE
Mark the fact that PGSHIFT and PDRSHIFT are really the same as
PG_SHIFT and PD_SHIFT, these should be collapsed some day soon.
Document that KERNBASE should really be KPTDPTDI << PDRSHIFT, for
now leave it as the constant 0xFE000000 until I make a seperate
common header file for this stuff (vmaddresses.h?)
Remove NKMEMCLUSTERS define, it was only being used to define
VM_KMEM_SIZE, so why have all the indirection. Besides who wants
to work in CLBYTE sizes chuncks.
pmap.h:
Fix $Id$ and some other minor format clean ups.
Remove the XXX comment about NKPDE, since it now has the correct value
of 7.
Remove unused LASTPTDI and move the APTD into the very end of memory to
free up 4MB of kernel virtual address space.
Remove unused RSVDPTDI and free up 12MB of kernel virtual address space.
vmparam.h
Fix $Id$.
Increase SHMMAXPGS to 512 (2MB) now that there is room for it to be
bigger. The XXX comment stays until the kernel moves down in memory
to free up enough space to use the proper default of 4MB.
VM_KMEM_SIZE is now a direct constant stating the size of the kernel
malloc region. Increased the value from 3MB to 16MB.
Bruce Evans had limited the kernel virtual address space to not include the
last 4MB since it was not being used. Other changes are being made that will
reloate the Alternate Page Directory Table (APDT) into this area so the limit
is being fixed to be the last virtual address. (Infact with this patch you
can now do that relocation)
Removed patch kit headers and rcsid strings, add $Id$.
isa.c:
Removed old #ifdef notyet isa_configure code, since it will never be
used, and I have done 90% of what it attempted to.
Add conflict checking code that searchs back through the devtab's looking
for any device that has already been found that may conflict with what
we are about to probe. Checks are mode for I/O address, memory address,
IRQ, and DRQ. This should stop the screwing up of any device that has
alread been found by other device probes.
Print out messages when we are not going to probe a device due to
a conflict so the user knows WHY something was not found. For example:
aha0 not probed due to irq conflict with ahb0 at 11
Now print out a message when a device is not found so the user knows
that it was probed for, but could not be found. For example:
ed1 not found at 0x320
For devices that have I/O address < 0x100 say that they are on the
motherboard, not on isa! The 0x100 magic number is per ISA spec. It
may seem funny that pc0 and sc0 report as being on the motherboard, but
this is due to the fact that the I/O address used is that of the keyboard
controller which IS on the motherboard. We really need to split the
keyboard probe from the display probe. It is completly legal to build
a pc with out one or the other, or even with out both!
npx.c:
Return -1 from the probe routine if we are using the Emulator so
that the i/o addresses are not printed, this is the same trick used
for 486's.
Do not print the ``Errors reported via Exception 16'', and
``Errors reported via IRQ 13'' messages any more, since these just lead
to more user confusion that anything. It still prints the message
``Error reporting broken, using 387 emulator'' so that the person is
aware that there mother board is ill.
to do this as I have changed to using PDTI's as the bases for the vm
system layout.
Eliminate constants SYSPDROFF and SYSPDREND, now use NKPTE to control the size
of the kernel virtual space.
Eliminate constant PDRPDROFF, now use PDTDTPI to control location of PTD,
PTDmap and PTDpde
Eliminate constant APDRPDROFF, now use APTDPTDI to control location of APTD,
APTDmap and APTDpde.
Still need to fix Sysmap location (it is still a constant).
.globl statements are now consistent with respect to <comma><space>, the
<space> being removed from all .globl statements.
Document the fillkpt macro as to what registers control what.
Fix some comments that went past column 80, and clean/line some others up.
Remove constand for _Crtat, now use KERNBASE+constant, this still needs work.
Replace constants for offsets of sigcode parameters with symbolic names
from assym.s
Mark the sigreturn() call with XXX since we use the hardcoded constant
for the system call number, this is bogus and should be a #define or
something some place!
The kernel before and after this change was verified with cmp, not one
byte changed. These are all cosmetic clean up changes that makes the
code more correct and easier to move the kernels virtual address space
and size.
Cleaned up tabs vs spaces after #define to make file consistent.
Removed now unused definitions of I386_PAGE_SIZE and I386_PDR_SIZE
Note That these two where unused and had the wrong values anyway!
Changed I386_KPDES to NKPDE
Changed I386_UPDES to NUPDE
Redid constant assignments of *PTDI's to be sizeable and relative.
PDRSHIFT.
The SYSTEM constant that was defined in this file has been replaced
with KERNBASE from param.h.
Changed almost all # style comments to /* */ C style comments. Several
comments cleaned up so that they make a little more since.
In the comments that describe C calling conventions to assembler routines
used a comma space sequence to seperate arguments and removed the space
between the function name and the argument list.
Removed useless comments like /* clr eax */.
Changed all comma space sequences on assemble instructions to just be comma.
Removed spaces after $ operators to make the file consistent, this may need
to change again (ie: $KERNBASE should probably be $(KERNBASE), but for now
it all seems to work just fine.) This may become a problem with the C
pre-processor.
Changed several double blank lines to single blank lines that where used
to seperate the I/O routines, these routines are blocked enough that we
don't need double blank lines between them.
Changed sequence of I/O routines to be all input functions, all output
functions instead of just the opposite.
Moved the SHOW_A_LOT debug stuff to near the end of the file.
Changed two occurances of the constant 0xfff to NBPG-1.
the proper I_X86CPU in the config file the following error will occur
while building the kernel: (had to line wrap the error for this message)
../../i386/i386/machdep.c:343: #error This kernel is not configured for one \
of the supported CPUs
Remove patch kit headers, and add $Id$
This is mostly to align some more code with NetBSD.
cpu.h:
Remove the old function vs. include configuration stuff that was
ifdefed out when we went to inline functions.
Remove the define of resettodr that made it a nop, there is
already a function that makes it a nop, no need to #define one.
Remove the #defines of processor types, they are now defined
in cputypes.h, #include that file.
Add struct cpu_nameclass for support of cpu types.
frame.h:
include sys/signal.h, it will be needed in the future.
put the sigframe structure here that was in machdep.c
pcb.h:
Add multiple inclusion protection.
Add pcb_ldt and pcb_ldt_len to pcb structure, this is for the
user mode ldt.
removed patch kit headers and sccsids, add $Id$. This is a general
clean up and reallignment with NetBSD-current where possible.
genassym.c:
removed extranious include of reg.h
removed old FP_* defines that have been ifdefed out since the patch kit
removed PCB_SIGC that is not referenced anywhere
add trapframe and sigframe defines
add KERNBASE define for use in locore.s
locore.s:
include npx.h and use NNPX for turning on and off FPU
include machine/cputypes.h for the types of cpu (used in cpu_identify)
change SYSPDREND to be one higher, this is really the base of the
next area, and will be changing again next time I revise the file
Reverse the NOP defines, you now get slow NOP's by default, this
may be what is casuing us trouble with some systems. If you want
the NOPS to be null you now need to have options DUMMY_NOPS.
Now get esym from the boot blocks which don't pass it yet, and
it is not used, but this will be changing.
Move the bit_colors stuff to be in with the rest of Bruces SHOW_A_LOT
things for debugging.
Added NetBSD's CPU type probe code, we now know what type of CPU
we are running on.
Adjust kernel pde calcuation to correct for change in SYSPDREND, no
longer need the +1.
machdep.c
include npx.h and use NNPX for turning on and off FPU
include isa.h, map.h(new file), exec.h in preperation for
changes that are still in process.
Add some of the code for MACHINE_NONCONTIG that will alow us
to better map around the BIOS memory area.
Now print the version, cpu id, real memory and availiable memory
during boot.
Correct the calculation of bufpages, the code was mixing pages
and bytes, it now does the right things. Removed Bill's hack
for limiting the erronous calculation.
add the identifycpu print out code from NetBSD.
remove the definition of the sigframe struct, it belongs in
frame.h
put in printf's about syncing disks on a halt/reboot.
Change the halted message to be a little easier reading.
Clean up of the dump messages, makes the source and the output
much more readable.
Change 0,0 in several places to have spaces after the commas.
it relocates it to be after the BIOS memory hole instead of right below
the 640K limit.
THANK YOU CHRIS!!!
From: <cgd@postgres.Berkeley.EDU>
Date: Wed, 29 Sep 93 18:49:58 -0700
basically, reserve a new 32k space right after firstaddr,
and put the buffer space there...
the diffs are below, and are in ~cgd/sys/i386/i386 (in machdep.c)
on freefall. i obviously can't test them, so if some of you would
look the diffs over and try them out...
Date: Sun, 12 Sep 1993 18:19:05 -0500
This will allow you to compile and run a freebsd kernel with shared
memory support. I haven't tested the shm*() calls yet.
You run out of page table descriptors if you specify 4Mb of sharable
memory (SHMMAXPGS=1024). I don't know what the limit is, but
SHMMAXPGS=64 works. Rich
going to panic shortly after this anyway. Destroys less state, and
keeps the machine from waiting for someone to smash the return key
a few times before it panics!
Added support of DONET({IMP,NS,ISO}) so you can now compile with options
NS and ISO, still missing some IMP code, but since the imp is old and
gone I doubt this will ever be used.
FP and we try to call the emulator when it is not compiled in.
Removed the #if defined(i486) || defined(i387) that use to call the
panic if we did not have a math emulator.
Removed an extranious include of i386/i386/math_emu.h from math_emulate.c.
I posted some patches on the 386bsd_patchkit list to prohibit io access.
Because of a noninitialised filed in the tss, this was possible.
It is included below as the patch to machdep.c
However, when you do this *necessary* fix (security), it will be
impossible form within user space to do io.
therefor, I included another fix: when you open /dev/io, you
get the access. Of course you can rewrite it to use another minor
and thus giving access to the iospace when /dev/mem is opened, e.g.
NOTE: The /dev/io entry has not been added to /dev/MAKEDEV yet.
The patch is in NetBSD.
profiling, and various protection checks that cause security holes
and system crashes.
* Changed min/max/bcmp/ffs/strlen to be static inline functions
- included from cpufunc.h in via systm.h. This change
improves performance in many parts of the kernel - up to 5% in the
networking layer alone. Note that this requires systm.h to be included
in any file that uses these functions otherwise it won't be able to
find them during the load.
* Fixed incorrect call to splx() in if_is.c
* Fixed bogus variable assignment to splx() in if_ed.c
argument definitions outside of the function parameter list. This is
to reduce the copious warning messages that (non-Jolitz) gcc produces.
Also fixed some bogus variable declarations and casts to make the
compiler happy.
way of doing things. There still remain several drivers that need to
be updated. Also added a compile-time option to pccons to switch the
control and caps-lock keys (REVERSE_CAPS_CTRL) - added for my personal
sanity.
Andrey A. Chernov
David Greenman
Garrett Wollman
Jeffrey Hsu
Rodney W. Grimes
Andreas Schulz
Martin Renters
Andrew Moore
Paul Richards
Christoph Robitschko
Nate Williams
Charlie Root