]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoDon't use the ip_timestamp structure to access timestamp options, as the
Jonathan Lemon [Thu, 25 Oct 2001 06:27:51 +0000 (06:27 +0000)]
Don't use the ip_timestamp structure to access timestamp options, as the
compiler may cause an unaligned access to be generated in some cases.

PR: 30982

22 years agoIf we are bridging, fall back to using any inet address in the system,
Jonathan Lemon [Thu, 25 Oct 2001 06:14:21 +0000 (06:14 +0000)]
If we are bridging, fall back to using any inet address in the system,
irrespective of receive interface, as a last resort.

Submitted by: ru

22 years agoRelocate the KASSERT for a null recvif to a location where it will
Jonathan Lemon [Thu, 25 Oct 2001 05:56:30 +0000 (05:56 +0000)]
Relocate the KASSERT for a null recvif to a location where it will
actually do some good.

Pointed out by: ru

22 years agoMdoc fixes and clarification.
Jonathan Lemon [Thu, 25 Oct 2001 05:53:11 +0000 (05:53 +0000)]
Mdoc fixes and clarification.

Submitted by: sheldonh

22 years agoDocument the link0 option, which downloads the microcode to the board.
Jonathan Lemon [Thu, 25 Oct 2001 05:49:23 +0000 (05:49 +0000)]
Document the link0 option, which downloads the microcode to the board.

22 years agoAdd support for loadable microcode which implements interrupt coalescing
Jonathan Lemon [Thu, 25 Oct 2001 05:32:01 +0000 (05:32 +0000)]
Add support for loadable microcode which implements interrupt coalescing
and packet bundling.  Make the microcode settings controllable via sysctl
and loader tunables.

Submitted by: Marko Zec <zec@tel.fer.hr>
  (with some munging and dynamic sysctl support by me)

Also extend the workaround for Dynamic Standby mode to later '559 chips,
not just the ICH2 variants.

22 years ago . Add structure elements for sysctl.
Jonathan Lemon [Thu, 25 Oct 2001 05:27:25 +0000 (05:27 +0000)]
  . Add structure elements for sysctl.
  . Document default values for microcode

22 years ago . Add structure definition for microcode download.
Jonathan Lemon [Thu, 25 Oct 2001 05:25:58 +0000 (05:25 +0000)]
  . Add structure definition for microcode download.
    Submitted by: Marko Zec <zec@tel.fer.hr>

  . Add some PCI chip revision entries.
  . Make size of txcb dependent on pointer size rather than arch #define

22 years agoAdd Intel's loadable microcode to implement receive side bundling.
Jonathan Lemon [Thu, 25 Oct 2001 05:23:31 +0000 (05:23 +0000)]
Add Intel's loadable microcode to implement receive side bundling.
This is taken verbatim from the Intel's e100-1.6.22 release, with
the addition of their LICENSE file at the top.

Submitted by: Marko Zec <zec@tel.fer.hr>

22 years agocnclose() can potentially race against itself. To avoid vn_close() races,
Jonathan Lemon [Thu, 25 Oct 2001 04:51:37 +0000 (04:51 +0000)]
cnclose() can potentially race against itself.  To avoid vn_close() races,
NULL-out cnd_vp before calling the latter, as it may block.

Submitted by: dillon

22 years agoAdd PCI_ENABLE_IO_MODES option, for BIOSen that neglect this.
Jonathan Lemon [Thu, 25 Oct 2001 04:44:50 +0000 (04:44 +0000)]
Add PCI_ENABLE_IO_MODES option, for BIOSen that neglect this.

Submitted by: Andrew R. Reiter arr@watson.org

22 years agoFix bugs in NgSendMsg() and NgSendAsciiMsg() where the wrong token
Archie Cobbs [Thu, 25 Oct 2001 04:02:25 +0000 (04:02 +0000)]
Fix bugs in NgSendMsg() and NgSendAsciiMsg() where the wrong token
value could be returned when the debug level was non-zero.

Submitted by: Harti Brandt <brandt@fokus.gmd.de>
MFC after: 5 days

22 years agoRe-commit www:www
Andrey A. Chernov [Thu, 25 Oct 2001 03:27:16 +0000 (03:27 +0000)]
Re-commit www:www
If anybody wants to remove them for some reason, please consider "pop"
removing first.

Approved by: arch discussion from Oct 20
MFC after: 3 days

22 years agoResume deorbit burn sequence for <malloc.h>. Inspection of my make logs
Garrett Wollman [Thu, 25 Oct 2001 02:35:29 +0000 (02:35 +0000)]
Resume deorbit burn sequence for <malloc.h>.  Inspection of my make logs
reveals no programs in the standard build which hit the previous warning
message.  This change will NOT be merged to 4-stable.

22 years agoo Remove extern showallsockets, defunct as of the change to
Robert Watson [Thu, 25 Oct 2001 02:03:37 +0000 (02:03 +0000)]
o Remove extern showallsockets, defunct as of the change to
  kern.security.seeotheruids_permitted.  This was missed in the
  commit that made this change elsewhere.

22 years agoBackout 1.61 -- both intrcnt and intrnames are already exported
Luigi Rizzo [Thu, 25 Oct 2001 01:32:24 +0000 (01:32 +0000)]
Backout 1.61 -- both intrcnt and intrnames are already exported
via sysctl under "hw".

22 years agoChange #include "DEFS.h" to <machine/asm.h>.
Peter Wemm [Thu, 25 Oct 2001 01:30:54 +0000 (01:30 +0000)]
Change #include "DEFS.h" to <machine/asm.h>.

22 years agoUse the correct sysctl names for intrcnt and intrnames.
Luigi Rizzo [Thu, 25 Oct 2001 01:29:21 +0000 (01:29 +0000)]
Use the correct sysctl names for intrcnt and intrnames.

22 years agoSplit the per-process Local Descriptor Table out of the PCB and into
John Baldwin [Thu, 25 Oct 2001 00:53:43 +0000 (00:53 +0000)]
Split the per-process Local Descriptor Table out of the PCB and into
struct mdproc.

Submitted by: Andrew R. Reiter <arr@watson.org>
Silence on: -current

22 years agoForce FWRITE on when opening the console, so that the flags passed to
Jonathan Lemon [Thu, 25 Oct 2001 00:14:16 +0000 (00:14 +0000)]
Force FWRITE on when opening the console, so that the flags passed to
vn_close match those from vn_open.  This fixes the panic some people
were seeing about "vrele: missed vn_close".

22 years agoUpdate Labour Day in New Zealand. It's definitely not today. This
Greg Lehey [Thu, 25 Oct 2001 00:01:06 +0000 (00:01 +0000)]
Update Labour Day in New Zealand.  It's definitely not today.  This
year it was on 22 October, which makes me think that it's the 4th
Monday in October.  At least this way it's correct for this year.

22 years agore-undo rev 1.78 now that style(9) is sane in this regard,
Julian Elischer [Wed, 24 Oct 2001 23:40:59 +0000 (23:40 +0000)]
re-undo rev 1.78 now that style(9) is sane in this regard,
(make struct {proc,thread,kse,ksgrp} readable again.)

22 years agoMerge in BSDisms. Slight rewordings in some cases.
Julian Elischer [Wed, 24 Oct 2001 22:57:39 +0000 (22:57 +0000)]
Merge in BSDisms. Slight rewordings in some cases.

Submitted by: bde@freebsd.org
Reviewed by: jhb@freebsd.org

22 years agoDocument the requirements and nature of the logical CPU IDs. It isn't
John Baldwin [Wed, 24 Oct 2001 22:15:38 +0000 (22:15 +0000)]
Document the requirements and nature of the logical CPU IDs.  It isn't
very strict and leaves much up to the platform so that it can define a
convenient mapping.

Requested by: mjacob

22 years agoChange the module name from 'snd_sonicvibes' to 'snd_vibes' as that is
John Baldwin [Wed, 24 Oct 2001 21:44:07 +0000 (21:44 +0000)]
Change the module name from 'snd_sonicvibes' to 'snd_vibes' as that is
the filename (thus the "old" module name) and the name used by
snd_driver.ko.

22 years agoChange module name from 'snd_als' to 'snd_als4000' to match name in
John Baldwin [Wed, 24 Oct 2001 21:42:06 +0000 (21:42 +0000)]
Change module name from 'snd_als' to 'snd_als4000' to match name in
snd_driver.ko.

22 years agoChange the module name from 'via' to 'snd_via82c686' to match the KLD
John Baldwin [Wed, 24 Oct 2001 21:38:08 +0000 (21:38 +0000)]
Change the module name from 'via' to 'snd_via82c686' to match the KLD
filename and in snd_driver.ko.

22 years agoChange the module name from 'snd_cmipci' to 'snd_cmi' to match the
John Baldwin [Wed, 24 Oct 2001 21:35:31 +0000 (21:35 +0000)]
Change the module name from 'snd_cmipci' to 'snd_cmi' to match the
filename, module name in the loader, the dependency in snd_driver.ko, and
to be consistent with other sound drivers.

22 years ago* Clear the TLB on boot.
Doug Rabson [Wed, 24 Oct 2001 21:22:08 +0000 (21:22 +0000)]
* Clear the TLB on boot.
* If a pte for a location given to pmap_enter_quick is valid, just give
  up - don't panic, even if the mapping is different.

22 years agoIf we get an unhandled page fault in kernel mode, either panic (if
Doug Rabson [Wed, 24 Oct 2001 21:20:50 +0000 (21:20 +0000)]
If we get an unhandled page fault in kernel mode, either panic (if
pcb_onfault is not set) or arrange to restart at the location in
pcb_onfault.

This ought to help the stability of a system under moderate load. It
certainly stops DDB from hanging the kernel when it tries to access a
non-present page.

22 years agoDe-orbit DEFS.h - the other arches do not use it, and it got replaced
Peter Wemm [Wed, 24 Oct 2001 20:29:14 +0000 (20:29 +0000)]
De-orbit DEFS.h - the other arches do not use it, and it got replaced
with <machine/asm.h>.

Reviewed by: bde

22 years agoTry to get the self-relocator to work with IPLTLSB relocations. Doesn't
Doug Rabson [Wed, 24 Oct 2001 20:14:49 +0000 (20:14 +0000)]
Try to get the self-relocator to work with IPLTLSB relocations. Doesn't
work right though - I can't figure out why.

22 years agoUnbreak alpha builds.
Matt Jacob [Wed, 24 Oct 2001 18:45:35 +0000 (18:45 +0000)]
Unbreak alpha builds.

22 years agounwind v_writecount in fhopen() if we are unable to allocate the
Matthew Dillon [Wed, 24 Oct 2001 18:32:17 +0000 (18:32 +0000)]
unwind v_writecount in fhopen() if we are unable to allocate the
descriptor.

MFC after: 3 days

22 years agocn_tab no longer exists, use cnadd() to add a console device. Note that
Jonathan Lemon [Wed, 24 Oct 2001 18:30:05 +0000 (18:30 +0000)]
cn_tab no longer exists, use cnadd() to add a console device.  Note that
this may result in duplicate console output in some cases.

22 years agoDon't enable write cacheing on devices that don't support tagged queuing
Dag-Erling Smørgrav [Wed, 24 Oct 2001 18:24:09 +0000 (18:24 +0000)]
Don't enable write cacheing on devices that don't support tagged queuing
unless hw.ata.wc is set.

Approved by: sos
MFC after: 1 week

22 years agoDocument C99 conformity.
Jeroen Ruigrok van der Werven [Wed, 24 Oct 2001 18:14:51 +0000 (18:14 +0000)]
Document C99 conformity.
Update text which is printed out to reflect reality.

22 years agoAdd __FBSDID.
Jeroen Ruigrok van der Werven [Wed, 24 Oct 2001 18:12:43 +0000 (18:12 +0000)]
Add __FBSDID.
Change __assert() function to print failing function name.
This makes us C99 conforming.

22 years agoAdd __FBSDID.
Jeroen Ruigrok van der Werven [Wed, 24 Oct 2001 18:12:18 +0000 (18:12 +0000)]
Add __FBSDID.
Change __assert() function to print failing function name.
#if 0 the sccsid block.
This makes us C99 conforming.

22 years agoAdd $FreeBSD$.
Jeroen Ruigrok van der Werven [Wed, 24 Oct 2001 18:10:37 +0000 (18:10 +0000)]
Add $FreeBSD$.
Change assert() macro to print failing function name.
Change K&R function prototype wrapper to ANSI prototype.
This makes us C99 conforming.

22 years agoRemove call to cninit_finish().
Jonathan Lemon [Wed, 24 Oct 2001 17:42:01 +0000 (17:42 +0000)]
Remove call to cninit_finish().

22 years ago- Clean up the comments slightly here to make them more readable.
John Baldwin [Wed, 24 Oct 2001 17:38:40 +0000 (17:38 +0000)]
- Clean up the comments slightly here to make them more readable.
- Set the type and trapframe number for the F00F workaround since type
  can be used later by sv_transtrap().  Debuggers might also want to look
  at the type in the trapframe.

Submitted by: bde (mostly)

22 years agoHelp to recover from bad seek (i.e. negative or too big) happens beyond
Andrey A. Chernov [Wed, 24 Oct 2001 17:25:49 +0000 (17:25 +0000)]
Help to recover from bad seek (i.e. negative or too big) happens beyond
our pre-check control. Do the same way as refill.c does when it set __SERR,
i.e. clear read and ungetc buffers. Clear EOF flag too.

22 years agoMake apm a module.
Hajimu UMEMOTO [Wed, 24 Oct 2001 16:53:54 +0000 (16:53 +0000)]
Make apm a module.
We need further work to be able to specify an equivalence of `flags 0x20'.
Many PCs should work for now.  But, some PCs need `flags 0x20'.

22 years agoRefer more consistently to the different FreeBSD (hardware platform)
Bruce A. Mah [Wed, 24 Oct 2001 16:01:40 +0000 (16:01 +0000)]
Refer more consistently to the different FreeBSD (hardware platform)
ports.  Also use "official" names for platforms where appropriate.

Partly suggested by: phantom

22 years agoAdd some new entities with the "official" names of the platforms
Bruce A. Mah [Wed, 24 Oct 2001 15:37:25 +0000 (15:37 +0000)]
Add some new entities with the "official" names of the platforms
we support (as opposed to the names of ports).

22 years agoAdded the pmc driver which supports power management controller of
Yoshihiro Takahashi [Wed, 24 Oct 2001 14:46:40 +0000 (14:46 +0000)]
Added the pmc driver which supports power management controller of
old NEC PC-98NOTE.

Submitted by: chi@bd.mbn.or.jp (Chiharu Shibata)
MFC after: 1 week

22 years agoFix this to actually compile in the !INVARIANTS case.
John Baldwin [Wed, 24 Oct 2001 14:18:33 +0000 (14:18 +0000)]
Fix this to actually compile in the !INVARIANTS case.

Reported by: Maxime Henrion <mux@qualys.com>

22 years agoUse vm_offset_t instead of caddr_t to fix a warning and remove
Robert Drehmel [Wed, 24 Oct 2001 14:15:28 +0000 (14:15 +0000)]
Use vm_offset_t instead of caddr_t to fix a warning and remove
two casts.

22 years agofix kvprintf panic in attach routine
Hellmuth Michaelis [Wed, 24 Oct 2001 14:12:47 +0000 (14:12 +0000)]
fix kvprintf panic in attach routine

22 years agoCorrect a typo: numtiple -> multiple.
Peter Pentchev [Wed, 24 Oct 2001 12:51:23 +0000 (12:51 +0000)]
Correct a typo: numtiple -> multiple.

PR: 31474
Submitted by: Eric Yu <ericyu@mail2000.com.tw>
MFC after: 1 week

22 years agofix SNDCTL_DSP_GETISPACE behaviour
Cameron Grant [Wed, 24 Oct 2001 12:40:54 +0000 (12:40 +0000)]
fix SNDCTL_DSP_GETISPACE behaviour

22 years agoimprove recording behaviour
Cameron Grant [Wed, 24 Oct 2001 12:33:21 +0000 (12:33 +0000)]
improve recording behaviour

22 years agoFinish the removal of the LABPC driver.
Ruslan Ermilov [Wed, 24 Oct 2001 10:57:30 +0000 (10:57 +0000)]
Finish the removal of the LABPC driver.

Forgotten by: phk

22 years agoIt's safe to assume that ${.OBJDIR} == ${.CURDIR} in the
Ruslan Ermilov [Wed, 24 Oct 2001 09:16:23 +0000 (09:16 +0000)]
It's safe to assume that ${.OBJDIR} == ${.CURDIR} in the
NOOBJ-hinted Makefiles.

22 years agoRemove call to cninit_finish. This is part of the multiple
Marcel Moolenaar [Wed, 24 Oct 2001 08:06:55 +0000 (08:06 +0000)]
Remove call to cninit_finish. This is part of the multiple
low-level console support.

22 years agoBack out my last commit. Ruslan (ru) has a better fix.
Mark Murray [Wed, 24 Oct 2001 07:59:42 +0000 (07:59 +0000)]
Back out my last commit. Ruslan (ru) has a better fix.

22 years agoAdd parse functions for local APIC and I/O APIC entries.
Marcel Moolenaar [Wed, 24 Oct 2001 04:48:05 +0000 (04:48 +0000)]
Add parse functions for local APIC and I/O APIC entries.
Also, show when a local APIC or SAPIC is disabled.

22 years agoo Note an additional potential problem here: LOCKD_MSG directly exports
Robert Watson [Wed, 24 Oct 2001 02:48:38 +0000 (02:48 +0000)]
o Note an additional potential problem here: LOCKD_MSG directly exports
  struct ucred to userland.  In 5.0-CURRENT, it is desirable to instead
  export struct xucred, as ucred contains mutexes, pointers, and other
  kernel evil.  I'll add it to my work queue.

22 years agoFix ktrace enablement/disablement races that can result in a vnode
Matthew Dillon [Wed, 24 Oct 2001 01:05:39 +0000 (01:05 +0000)]
Fix ktrace enablement/disablement races that can result in a vnode
ref count panic.

Bug noticed by: ps
Reviewed by: ps
MFC after: 1 day

22 years agoBack out read buffer invalidating via __SMOD.
Andrey A. Chernov [Tue, 23 Oct 2001 23:52:11 +0000 (23:52 +0000)]
Back out read buffer invalidating via __SMOD.

It was correct, but not needed because internal buffer cleared on each seek
outside of it.

22 years agoTra-La, another QLogic f/w funny- this time with the 2300.
Matt Jacob [Tue, 23 Oct 2001 23:05:20 +0000 (23:05 +0000)]
Tra-La, another QLogic f/w funny- this time with the 2300.
If we get a completion status of RQCS_QUEUE_FULL, it means
that the internal queues are full. Other QLogic boards set
the QFULL SCSI status. But *nooooooooooo*, not the 2300.

MFC after: 1 day

22 years agoChange comment explaining another usage of __SMOD
Andrey A. Chernov [Tue, 23 Oct 2001 23:05:15 +0000 (23:05 +0000)]
Change comment explaining another usage of __SMOD

22 years agoRemove if_wx.
John Baldwin [Tue, 23 Oct 2001 22:56:41 +0000 (22:56 +0000)]
Remove if_wx.

Submitted by: Dave Cornejo <dave@dogwood.com>

22 years agoDocument sx_assert(9).
John Baldwin [Tue, 23 Oct 2001 22:51:59 +0000 (22:51 +0000)]
Document sx_assert(9).

22 years agoDisallow fseek() optimization in internal read buffer, if pointer is moved by
Andrey A. Chernov [Tue, 23 Oct 2001 22:48:00 +0000 (22:48 +0000)]
Disallow fseek() optimization in internal read buffer, if pointer is moved by
seek. It means that beginning of read buffer becomes not the same as current
file position.

22 years agoRegen.
Nick Hibma [Tue, 23 Oct 2001 22:44:36 +0000 (22:44 +0000)]
Regen.

22 years agoAdd Olympus C-700
Nick Hibma [Tue, 23 Oct 2001 22:44:03 +0000 (22:44 +0000)]
Add Olympus C-700

22 years agoChange the sx(9) assertion API to use a sx_assert() function similar to
John Baldwin [Tue, 23 Oct 2001 22:39:11 +0000 (22:39 +0000)]
Change the sx(9) assertion API to use a sx_assert() function similar to
mtx_assert(9) rather than several SX_ASSERT_* macros.

22 years agoAllow hw.acpi.cpu.{economy,performance}_speed to be set from the loader
John Baldwin [Tue, 23 Oct 2001 22:37:36 +0000 (22:37 +0000)]
Allow hw.acpi.cpu.{economy,performance}_speed to be set from the loader
via tunables.

22 years agoChange TUNABLE_*_FETCH to have a return value of 0 if the variable was not
John Baldwin [Tue, 23 Oct 2001 22:35:25 +0000 (22:35 +0000)]
Change TUNABLE_*_FETCH to have a return value of 0 if the variable was not
found or successfully converted and true otherwise.

22 years ago- Change getenv_quad() to return an int instead of a quad_t since it
John Baldwin [Tue, 23 Oct 2001 22:34:36 +0000 (22:34 +0000)]
- Change getenv_quad() to return an int instead of a quad_t since it
  returns an success/failure code rather than the actual value.
- Add getenv_string() which copies a string from the environment to another
  string and returns true on success.

22 years agoSet the code and signal for the F00F hack fault directly instead of
John Baldwin [Tue, 23 Oct 2001 22:29:16 +0000 (22:29 +0000)]
Set the code and signal for the F00F hack fault directly instead of
changing the code in the trapframe and looping back to the top of trap
again.

Tested by: cjc

22 years agoFix RAW dependency violation when compiled with gcc-3
Peter Wemm [Tue, 23 Oct 2001 22:23:22 +0000 (22:23 +0000)]
Fix RAW dependency violation when compiled with gcc-3
Warning: Use of 'br.ret.sptk.many' violates RAW dependency 'PSR.tb' (data)

22 years agoAdd a manual page.
Jonathan Lemon [Tue, 23 Oct 2001 22:13:34 +0000 (22:13 +0000)]
Add a manual page.

22 years agoMake sure the copies: and symlinks: targets exist in the case of the
Mark Murray [Tue, 23 Oct 2001 21:28:39 +0000 (21:28 +0000)]
Make sure the copies: and symlinks: targets exist in the case of the
empty ".for ..." loop.

22 years agoHook up conscontrol to the build.
Jonathan Lemon [Tue, 23 Oct 2001 20:37:07 +0000 (20:37 +0000)]
Hook up conscontrol to the build.

22 years agoAdd conscontrol, a front end interface for manipulating consoles.
Jonathan Lemon [Tue, 23 Oct 2001 20:36:43 +0000 (20:36 +0000)]
Add conscontrol, a front end interface for manipulating consoles.

22 years agoClarification: &release.next; refers to a future FreeBSD release.
Bruce A. Mah [Tue, 23 Oct 2001 20:34:41 +0000 (20:34 +0000)]
Clarification:  &release.next; refers to a future FreeBSD release.

Submitted by: phantom

22 years agoSet RB_MULTIPLE (multiple console support) if the kernel is booted
Jonathan Lemon [Tue, 23 Oct 2001 20:27:48 +0000 (20:27 +0000)]
Set RB_MULTIPLE (multiple console support) if the kernel is booted
with the -D flag.

22 years agoFix typo.
Murray Stokely [Tue, 23 Oct 2001 20:27:06 +0000 (20:27 +0000)]
Fix typo.

PR: docs/31388
Submitted by: Yoshihiko Sarumaru <mistral@imasy.or.jp>

22 years agoAllow the RBX_DUAL flag to appear in bootinfo.
Jonathan Lemon [Tue, 23 Oct 2001 20:27:05 +0000 (20:27 +0000)]
Allow the RBX_DUAL flag to appear in bootinfo.

22 years agoImplement multiple low-level console support.
Jonathan Lemon [Tue, 23 Oct 2001 20:25:50 +0000 (20:25 +0000)]
Implement multiple low-level console support.

22 years agoAdd RB_MULTIPLE flag so the bootloader can turn on mulitiple console
Jonathan Lemon [Tue, 23 Oct 2001 20:22:04 +0000 (20:22 +0000)]
Add RB_MULTIPLE flag so the bootloader can turn on mulitiple console
support.  This flag value is identical to RBX_DUAL from boot2.

22 years agoAdd a siocnterm function for the CONS_DRIVER function vector.
Jonathan Lemon [Tue, 23 Oct 2001 20:16:56 +0000 (20:16 +0000)]
Add a siocnterm function for the CONS_DRIVER function vector.

22 years agoo Add two comments identifying problems with the current nfs_lock.c
Robert Watson [Tue, 23 Oct 2001 19:11:31 +0000 (19:11 +0000)]
o Add two comments identifying problems with the current nfs_lock.c
  implementation, so that the information doesn't get lost.
  (1) /var/run/lock is looked up relative to the current thread's root
      directory, but it's not clear that's desirable.
  (2) A race condition associated with live credential modification on
      a shared credential is present when privilege is granted for
      the purposes of talking to /var/run/lock.

22 years agoo vn_open() fails to call VOP_CLOSE() if vfs_object_create fails. Ideally
Robert Watson [Tue, 23 Oct 2001 19:09:01 +0000 (19:09 +0000)]
o vn_open() fails to call VOP_CLOSE() if vfs_object_create fails.  Ideally
  all successful calls to VOP_OPEN() might be reflected in a call to
  VOP_CLOSE().  For now, simply add a comment reflecting this problem;
  this should be fixed at some point.

22 years agoAssert that Giant is not held in mi_switch() unless the process state
John Baldwin [Tue, 23 Oct 2001 17:52:49 +0000 (17:52 +0000)]
Assert that Giant is not held in mi_switch() unless the process state
is SMTX or SRUN.

22 years agoAlter the suggested way of writing structurtes to make them actuallys
Julian Elischer [Tue, 23 Oct 2001 17:40:37 +0000 (17:40 +0000)]
Alter the suggested way of writing structurtes to make them actuallys
readble when there are compound sub-elements (e.g. other structs).

Reviewed by: {peter,dillon,des,imp,jlemon}@freebsd.org
MFC after: 1 week

22 years agoMove the 'type' variable inside the block where it is actually used.
John Baldwin [Tue, 23 Oct 2001 17:31:11 +0000 (17:31 +0000)]
Move the 'type' variable inside the block where it is actually used.

22 years agoBreak out the bus front ends into their own files. Rewrite
Warner Losh [Tue, 23 Oct 2001 15:17:33 +0000 (15:17 +0000)]
Break out the bus front ends into their own files.  Rewrite
sio_pccard_detach to use new siodetach.  Add an extra arg to sioprobe
to tell driver to probe/not probe the device for IRQs.

This incorporates most of Bruce's review material.  I'm at a good
checkpoint, but there will be more to come based on bde's further
reviews.

Reviewed by: bde

22 years agoCollect multi name handling.
Takanori Watanabe [Tue, 23 Oct 2001 14:54:15 +0000 (14:54 +0000)]
Collect multi name handling.

Reported by :    Stephan Kulow <coolo@caldera.de>

22 years agoDon't avoid setting a 0 second timer in datalink_StartDialTimer() by
Brian Somers [Tue, 23 Oct 2001 13:52:19 +0000 (13:52 +0000)]
Don't avoid setting a 0 second timer in datalink_StartDialTimer() by
not setting any timer.  Instead, set a 1 millisecond timer.

This ensures that ppp will come out of it's select() call after
losing carrier in -ddial mode with a reconnect period of 0 and
going to ST_OPENING, rather than waiting indefinitely for some
other event to wake ppp up.

Bump the ppp version number to indicate the event.

MFC after: 3 days

22 years agoPartially port kvm to ia64 - virtual to physical translation is incomplete.
Doug Rabson [Tue, 23 Oct 2001 11:05:35 +0000 (11:05 +0000)]
Partially port kvm to ia64 - virtual to physical translation is incomplete.

22 years agoTurn off the single-user override. We've been running multi-user
Peter Wemm [Tue, 23 Oct 2001 11:04:43 +0000 (11:04 +0000)]
Turn off the single-user override.  We've been running multi-user
for some time.  Having a machine boot unattended is useful. :-)

22 years agoRemove references to nfsiod(8) and nfs_client_flags
Ruslan Ermilov [Tue, 23 Oct 2001 10:31:52 +0000 (10:31 +0000)]
Remove references to nfsiod(8) and nfs_client_flags
now that they are obsolete.

22 years ago__FBSDID() (second half of src/lib/libcrypt changes)
Peter Wemm [Tue, 23 Oct 2001 10:23:32 +0000 (10:23 +0000)]
__FBSDID()  (second half of src/lib/libcrypt changes)

22 years agoia64 support. Due to the bogus placement of some i4b includes, it is
Peter Wemm [Tue, 23 Oct 2001 10:20:24 +0000 (10:20 +0000)]
ia64 support.  Due to the bogus placement of some i4b includes, it is
i386-only, even though it would work on all the isa-capable platforms.
Change the NOI4B so that instead of excluding alpha, it only includes i386.

22 years agoAdd data serialisations after ptc and mov to rr[] instructions.
Doug Rabson [Tue, 23 Oct 2001 10:20:04 +0000 (10:20 +0000)]
Add data serialisations after ptc and mov to rr[] instructions.

22 years agoia64 support. Hide a.out support solely under i386. While alpha never
Peter Wemm [Tue, 23 Oct 2001 10:18:41 +0000 (10:18 +0000)]
ia64 support.  Hide a.out support solely under i386.  While alpha never
was released in a.out form, it does define the a.out data structures.
The ia64 port does not.