]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoThe PAM module pam_krb5 does not have "session" capabilities.
Mark Murray [Wed, 30 Apr 2003 21:57:54 +0000 (21:57 +0000)]
The PAM module pam_krb5 does not have "session" capabilities.
Don't give examples of such use, this is bogus.

21 years agoAMD64 uses the new-style cpu_switch()/cpu_throw() calling conventions.
Peter Wemm [Wed, 30 Apr 2003 21:45:03 +0000 (21:45 +0000)]
AMD64 uses the new-style cpu_switch()/cpu_throw() calling conventions.

21 years agoCreate a 'legacy' node for AMD64 as well as i386. While we'll never
Peter Wemm [Wed, 30 Apr 2003 21:41:41 +0000 (21:41 +0000)]
Create a 'legacy' node for AMD64 as well as i386.  While we'll never
have to use it since all AMD64 machines are supposed to have acpi etc,
I'm using it during development so I can avoid the acpi code for now.
Yes, this is cheating.

21 years agoEliminate a compiler warning with gcc3.3 on AMD64, where speed is
Peter Wemm [Wed, 30 Apr 2003 21:39:28 +0000 (21:39 +0000)]
Eliminate a compiler warning with gcc3.3 on AMD64, where speed is
a 32 bit int which can never be > ULONG_MAX / 8.  Its an 'always true'
warning.

21 years agoMore factual updates for kerberos 5.
Mark Murray [Wed, 30 Apr 2003 21:15:41 +0000 (21:15 +0000)]
More factual updates for kerberos 5.

21 years agoRemove 80386 bandaids from code repocopied from i386. rtld_start.S still
Peter Wemm [Wed, 30 Apr 2003 21:09:06 +0000 (21:09 +0000)]
Remove 80386 bandaids from code repocopied from i386.  rtld_start.S still
todo.

21 years agoAMD64 support (another IEEEFP platform)
Peter Wemm [Wed, 30 Apr 2003 21:06:30 +0000 (21:06 +0000)]
AMD64 support (another IEEEFP platform)

21 years agoAMD64 support; repocopied from i386
Peter Wemm [Wed, 30 Apr 2003 21:05:33 +0000 (21:05 +0000)]
AMD64 support; repocopied from i386

21 years agoTeach libdisk that AMD64 works just like i386
Peter Wemm [Wed, 30 Apr 2003 21:03:16 +0000 (21:03 +0000)]
Teach libdisk that AMD64 works just like i386

21 years agoMake the defaults for Kerberos 5 a little more up-to-date.
Mark Murray [Wed, 30 Apr 2003 20:58:49 +0000 (20:58 +0000)]
Make the defaults for Kerberos 5 a little more up-to-date.

21 years agoReplace beforeinstall: with plain FILES.
Ruslan Ermilov [Wed, 30 Apr 2003 20:48:03 +0000 (20:48 +0000)]
Replace beforeinstall: with plain FILES.

21 years agoChange the name of the kadmind binary to match reality, now that
Mark Murray [Wed, 30 Apr 2003 20:42:41 +0000 (20:42 +0000)]
Change the name of the kadmind binary to match reality, now that
KerberosIV is no longer an issue.

21 years agoStyle fixes to struct sigacts member comments. Sort function prototypes.
John Baldwin [Wed, 30 Apr 2003 19:57:21 +0000 (19:57 +0000)]
Style fixes to struct sigacts member comments.  Sort function prototypes.

21 years agoForgot to remove Giant around call to kern_sigaction() in
John Baldwin [Wed, 30 Apr 2003 19:45:13 +0000 (19:45 +0000)]
Forgot to remove Giant around call to kern_sigaction() in
freebsd4_sigaction() in revision 1.232.

21 years agoAxe a stale comment.
John Baldwin [Wed, 30 Apr 2003 19:41:04 +0000 (19:41 +0000)]
Axe a stale comment.

21 years agoAdd __amd64__ to the list of things that use IEEEFP
Peter Wemm [Wed, 30 Apr 2003 19:32:25 +0000 (19:32 +0000)]
Add __amd64__ to the list of things that use IEEEFP

21 years agoTell malloc.c that AMD64 uses the same pagesize as i386.
Peter Wemm [Wed, 30 Apr 2003 19:30:34 +0000 (19:30 +0000)]
Tell malloc.c that AMD64 uses the same pagesize as i386.

21 years agoAdd __amd64__ to the list of things that use HIDENAME() to hide minbrk
Peter Wemm [Wed, 30 Apr 2003 19:29:02 +0000 (19:29 +0000)]
Add __amd64__ to the list of things that use HIDENAME() to hide minbrk

21 years agoUpdate for AMD64. repocopied from i386-elf/crt1.c. Deal with regparm
Peter Wemm [Wed, 30 Apr 2003 19:27:07 +0000 (19:27 +0000)]
Update for AMD64. repocopied from i386-elf/crt1.c.  Deal with regparm
argument passing rather than stack based args.  The kernel passes the
base of the argument/env vector in %rdi (arg1).

21 years agoFix references to non-existing or obsoleted man pages.
Jens Schweikhardt [Wed, 30 Apr 2003 19:18:50 +0000 (19:18 +0000)]
Fix references to non-existing or obsoleted man pages.

PR: docs/51480 (only a small part)
Submitted by: Diomidis D. Spinellis <dds@aueb.gr>

21 years agoIncrease the scope of the vm_object lock in vm_map_delete().
Alan Cox [Wed, 30 Apr 2003 19:18:09 +0000 (19:18 +0000)]
Increase the scope of the vm_object lock in vm_map_delete().

21 years agoRemove redundant strlen checks, do not check the same
Alexander Kabaev [Wed, 30 Apr 2003 19:05:53 +0000 (19:05 +0000)]
Remove redundant strlen checks, do not check the same
symbol twice.

21 years agoTurn off libstand for amd64 for the time being. It is built in i386
Peter Wemm [Wed, 30 Apr 2003 18:42:25 +0000 (18:42 +0000)]
Turn off libstand for amd64 for the time being.  It is built in i386
mode, and we do not need the complications for now.

21 years agoDelete i386_* syscall wrappers and manpages. Rename Ovfork.S to vfork.S.
Peter Wemm [Wed, 30 Apr 2003 18:17:07 +0000 (18:17 +0000)]
Delete i386_* syscall wrappers and manpages.  Rename Ovfork.S to vfork.S.

21 years agoUpdate for AMD64 after repocopy from i386/sys/*. This means:
Peter Wemm [Wed, 30 Apr 2003 18:16:33 +0000 (18:16 +0000)]
Update for AMD64 after repocopy from i386/sys/*.  This means:
- strip out the nasty PIC_PROLOGUE/EPILOGUE stuff, since we dont have
to lose a register in PIC mode anymore (we use %rip-relative addressing).
- update for C register argument passing conventions.
- convert 32 bit to 64 bit register sizes etc.

21 years agoI have no idea why the reboot(2) syscall wrapper ends with iret, but
Peter Wemm [Wed, 30 Apr 2003 18:14:44 +0000 (18:14 +0000)]
I have no idea why the reboot(2) syscall wrapper ends with iret, but
update it to be iretq for completeness.

21 years agoUpdate for AMD64. Depend on %rdi (first syscall argument) being preserved
Peter Wemm [Wed, 30 Apr 2003 18:13:48 +0000 (18:13 +0000)]
Update for AMD64.  Depend on %rdi (first syscall argument) being preserved
across a "syscall"-style syscall

21 years agoUpdate for AMD64 (repocopied from i386/sys/Ovfork.S - why is it O?)
Peter Wemm [Wed, 30 Apr 2003 18:13:13 +0000 (18:13 +0000)]
Update for AMD64 (repocopied from i386/sys/Ovfork.S - why is it O?)
Depend on %rsi being preserved across the "syscall"-style syscall and strip
out the PIC stuff (this cpu has full PC-relative addressing, at last!)

21 years agoUpdate for AMD64 contexts. Note that this still has some x87-style FPU
Peter Wemm [Wed, 30 Apr 2003 18:09:33 +0000 (18:09 +0000)]
Update for AMD64 contexts.  Note that this still has some x87-style FPU
instructions here, these may need to be updated for SSE.
(Repocopied from i386/gen)

21 years agoReduce the SRCS to what compiles
Peter Wemm [Wed, 30 Apr 2003 18:08:01 +0000 (18:08 +0000)]
Reduce the SRCS to what compiles

21 years agoUpdate for AMD64 (repocopied from i386/net). We can depend on having
Peter Wemm [Wed, 30 Apr 2003 18:07:23 +0000 (18:07 +0000)]
Update for AMD64 (repocopied from i386/net).  We can depend on having
the bswap instruction (yay!).  Update for register parameter passing
instead of i386 style stack based param passing.

21 years agoUpdate for AMD64-style syscalls. Repocopied from i386/SYS.h.
Peter Wemm [Wed, 30 Apr 2003 18:06:14 +0000 (18:06 +0000)]
Update for AMD64-style syscalls.  Repocopied from i386/SYS.h.
Note that the syscall instruction clobbers %rcx, which is inconvenient
because it is the fourth syscall argument, so we use %r10 (another scratch
register) for the 4th syscall arg instead (I picked %r10 to be the same as
NetBSD).  int 0x80 is still possible though, and it uses %rcx as usual.
Note that the syscall style syscall does *NOT* preserve all the registers,
unlike int 0x80.  We do not preserve the scratch registers except for
%rdi and %rsi.  int 0x80 does preserve everything but the return values.

21 years agoRange check the syscall number before looking it up in the syscallnames[]
John Baldwin [Wed, 30 Apr 2003 17:59:27 +0000 (17:59 +0000)]
Range check the syscall number before looking it up in the syscallnames[]
array.

Submitted by: pho

21 years agoKerberosIV removal. Also crunch down the "krb5" distribution into
Mark Murray [Wed, 30 Apr 2003 17:54:19 +0000 (17:54 +0000)]
KerberosIV removal. Also crunch down the "krb5" distribution into
"secure".

21 years agoWe no longer have KerberosIV. Remove it here also.
Mark Murray [Wed, 30 Apr 2003 17:50:08 +0000 (17:50 +0000)]
We no longer have KerberosIV. Remove it here also.

21 years agoWe no longer have a separate kerberos distribution. Its now just
Mark Murray [Wed, 30 Apr 2003 17:46:24 +0000 (17:46 +0000)]
We no longer have a separate kerberos distribution. Its now just
part of the regular security dist.

21 years agoWe have to hide copyright[] for gcc33. :-(
David E. O'Brien [Wed, 30 Apr 2003 17:40:28 +0000 (17:40 +0000)]
We have to hide copyright[] for gcc33. :-(

21 years agoOnly define platform once -- in a C file.
David E. O'Brien [Wed, 30 Apr 2003 17:14:58 +0000 (17:14 +0000)]
Only define platform once -- in a C file.

21 years agoRemove some KerberosIV references that are now stale.
Mark Murray [Wed, 30 Apr 2003 17:01:43 +0000 (17:01 +0000)]
Remove some KerberosIV references that are now stale.

21 years agoFloating point libc functions traditionally written in ASM.
David E. O'Brien [Wed, 30 Apr 2003 16:21:03 +0000 (16:21 +0000)]
Floating point libc functions traditionally written in ASM.
AMD64 does away with the x87 in 64-bit long mode, so we have to play the
SSE/SSE2 game now.

Obtained from: NetBSD/x86-64

21 years agoSymlink with -f.
David E. O'Brien [Wed, 30 Apr 2003 15:49:40 +0000 (15:49 +0000)]
Symlink with -f.

21 years agoNo libc_r for AMD64.
David E. O'Brien [Wed, 30 Apr 2003 15:33:50 +0000 (15:33 +0000)]
No libc_r for AMD64.

21 years agoMove the mailbox to the beginning of the thread and align the
Daniel Eischen [Wed, 30 Apr 2003 15:05:17 +0000 (15:05 +0000)]
Move the mailbox to the beginning of the thread and align the
thread so that the context (SSE FPU state) is also aligned.

21 years agoUse %j instead of %q in printf to unbreak sparc64.
Poul-Henning Kamp [Wed, 30 Apr 2003 13:36:51 +0000 (13:36 +0000)]
Use %j instead of %q in printf to unbreak sparc64.

21 years agoFix some easy, global, lint warnings. In most cases, this means
Mark Murray [Wed, 30 Apr 2003 12:57:40 +0000 (12:57 +0000)]
Fix some easy, global, lint warnings. In most cases, this means
making some local variables static. In a couple of cases, this means
removing an unused variable.

21 years agoRestore the ordering of NO's and remove one redundant NOLINT.
Ruslan Ermilov [Wed, 30 Apr 2003 12:38:20 +0000 (12:38 +0000)]
Restore the ordering of NO's and remove one redundant NOLINT.

Reviewed by: markm

21 years agoWarns fixing. Protect against inappropriate linting, and mark
Mark Murray [Wed, 30 Apr 2003 12:23:58 +0000 (12:23 +0000)]
Warns fixing. Protect against inappropriate linting, and mark
GCC-specific assemble code as such (in #ifdefs). Fix an easy
static variable warning while I'm here.

21 years agoHelp out with linting. Print the ${LINT} command line. This makes
Mark Murray [Wed, 30 Apr 2003 12:19:25 +0000 (12:19 +0000)]
Help out with linting. Print the ${LINT} command line. This makes
the all-important -DFOO -IBAR options "hang out".

21 years agoMake bitfields plain old u_int. u_int16_t makes no sense.
Mark Murray [Wed, 30 Apr 2003 12:07:32 +0000 (12:07 +0000)]
Make bitfields plain old u_int. u_int16_t makes no sense.

21 years agoFix "make world" for the WANT_LINT case. The various build tools
Mark Murray [Wed, 30 Apr 2003 11:17:33 +0000 (11:17 +0000)]
Fix "make world" for the WANT_LINT case. The various build tools
must not try to lint(1) stuff at *-tools time; keep the linting
for the actual build (which comes later).

21 years agoGroff after 2002/10/15 has no known backward compatibility issues,
Ruslan Ermilov [Wed, 30 Apr 2003 10:58:52 +0000 (10:58 +0000)]
Groff after 2002/10/15 has no known backward compatibility issues,
but we always want the latest macro files.

21 years agoRelease devq before detach.
Hidetoshi Shimokawa [Wed, 30 Apr 2003 10:32:32 +0000 (10:32 +0000)]
Release devq before detach.

21 years agoBackout my changes in rev. 1.32 and 1.33. There is some code that depends
Mike Makonnen [Wed, 30 Apr 2003 10:25:57 +0000 (10:25 +0000)]
Backout my changes in rev. 1.32 and 1.33. There is some code that depends
on the previous behaviour. This also brings strptime(3) behaviour back in line
with the other BSDs.

Approved by: markm (mentor)

21 years agoSet status to FWDEVINVAL if probe failed.
Hidetoshi Shimokawa [Wed, 30 Apr 2003 10:20:24 +0000 (10:20 +0000)]
Set status to FWDEVINVAL if probe failed.

21 years agoThe including makefile's directory is tried first for .include "...".
Ruslan Ermilov [Wed, 30 Apr 2003 07:54:39 +0000 (07:54 +0000)]
The including makefile's directory is tried first for .include "...".

21 years agoMost things depend on !defined(NO_OPENSSL); make it look so.
Ruslan Ermilov [Wed, 30 Apr 2003 07:51:51 +0000 (07:51 +0000)]
Most things depend on !defined(NO_OPENSSL); make it look so.

21 years agoNOSECURE is implied by NOCRYPT, meaning if the latter is defined
Ruslan Ermilov [Wed, 30 Apr 2003 07:34:14 +0000 (07:34 +0000)]
NOSECURE is implied by NOCRYPT, meaning if the latter is defined
we won't be here.

21 years agoDon't clobber Kerberos5 telnet(1) and telnetd(8) with non-crypto versions.
Ruslan Ermilov [Wed, 30 Apr 2003 07:24:35 +0000 (07:24 +0000)]
Don't clobber Kerberos5 telnet(1) and telnetd(8) with non-crypto versions.

21 years agoFixed alignment.
Ruslan Ermilov [Wed, 30 Apr 2003 07:13:50 +0000 (07:13 +0000)]
Fixed alignment.

21 years agoThis script should require mountcritremote, since that is the last
Mike Makonnen [Wed, 30 Apr 2003 07:12:09 +0000 (07:12 +0000)]
This script should require mountcritremote, since that is the last
script at which a filesystem will be mounted in FreeBSD. Without this
the script was being inserted later in the boot than necessary.

Approved by: markm (mentor) (implicit)
Reviewed by: dougb
Noticed by: Andrzej ToboÅa <san@wilk.2a.pl>

21 years agoo Don't announce that the timer is good when in fact it isn't timing
Marcel Moolenaar [Wed, 30 Apr 2003 05:27:01 +0000 (05:27 +0000)]
o  Don't announce that the timer is good when in fact it isn't timing
   at all (ie reads yield constant values). Display the width as the
   difference between max and min so that constant timers have width
   zero.
o  Get the address of the timer from the XPmTmrBlk field instead of
   the V1_PmTmrBlk field. The former is a generic address and can
   specify a memory mapped I/O address. Remove <machine/bus_pio.h>
   to account for this. The timer is now properly configured on
   machines with ACPI v2 tables, whether PIO or MEMIO. Note that
   the acpica code converts v1 tables into v2 tables so the address
   is always present in XPmTmrBlk.
o  Replace the TIMER_READ macro with a call to the read_counter()
   function and add a barrier to make sure that we observe proper
   ordering of the reads.

21 years agoFix tpyo in last commit.
Warner Losh [Wed, 30 Apr 2003 04:14:56 +0000 (04:14 +0000)]
Fix tpyo in last commit.

Noticed by: kuriyama-san

21 years agoSet include path.
Hidetoshi Shimokawa [Wed, 30 Apr 2003 03:39:25 +0000 (03:39 +0000)]
Set include path.

21 years agoFree asyreq.
Hidetoshi Shimokawa [Wed, 30 Apr 2003 03:38:07 +0000 (03:38 +0000)]
Free asyreq.

Submitted by: ryuchi@ryuchi.org

21 years agoEliminate an unused parameter from vm_pageout_object_deactivate_pages().
Alan Cox [Wed, 30 Apr 2003 03:08:16 +0000 (03:08 +0000)]
Eliminate an unused parameter from vm_pageout_object_deactivate_pages().

21 years agoThis particular debugging statement is very verbose, making the rest
Mike Makonnen [Wed, 30 Apr 2003 03:03:47 +0000 (03:03 +0000)]
This particular debugging statement is very verbose, making the rest
of the debug output hard to follow. Comment it out. It's commented out
in NetBSD's version as well.

Approved by: markm (mentor)(implicit)

21 years agoo Make the 'Ip-filter module loaded' messages informational
Mike Makonnen [Wed, 30 Apr 2003 02:54:17 +0000 (02:54 +0000)]
o Make the 'Ip-filter module loaded' messages informational
o Make 'No ipnat rules' a warning
o Remove unecessary ' ..'

Approved by: markm (implicit)

21 years agoRunning the script with restart or manually stopping and starting
Mike Makonnen [Wed, 30 Apr 2003 02:19:38 +0000 (02:19 +0000)]
Running the script with restart or manually stopping and starting
it doesn't work because the start_cmd doesn't enable ipfilter if
it is currently disabled.

Approved by: markm (mentor) (implicit)
Submitted by: Michael Lyngbøl <lyngbol@bifrost.lyngbol.dk>
PR: conf/46103

21 years agoGet rid of the redundant 'gone' field, and overload suspend instead.
Warner Losh [Wed, 30 Apr 2003 01:54:38 +0000 (01:54 +0000)]
Get rid of the redundant 'gone' field, and overload suspend instead.
Check for suspend before the device polling, rather than after it.
Check to see if the current thread owns the lock in ioctl and return
EBUSY if it does.

This advances the locking to the point that I can eject my fxp card 10
times in a row, but I agree with Jeff Hsu that we need to get the
network layer locking finished before chasing more of the races here
(actually, he doesn't think this set is worth it even).  There's a
number of races between FXP_LOCK in detach and all other users of
FXP_LOCK, and this gets back to the 'device with sleepers being
forcibly detached' problem as well...

21 years agoIncrease some default values.
David Xu [Wed, 30 Apr 2003 01:18:29 +0000 (01:18 +0000)]
Increase some default values.

21 years agoCall kse_wakeup_mutli() after remove current thread from RUNQ to avoid
David Xu [Wed, 30 Apr 2003 01:15:21 +0000 (01:15 +0000)]
Call kse_wakeup_mutli() after remove current thread from RUNQ to avoid
doing unnecessary idle kse wakeup.

21 years agoCall kse_wakeup_multi() to wakeup idle KSEs when there are threads ready
David Xu [Wed, 30 Apr 2003 01:03:58 +0000 (01:03 +0000)]
Call kse_wakeup_multi() to wakeup idle KSEs when there are threads ready
to run.

21 years agoMake sure rhostip is always initialized.
Dag-Erling Smørgrav [Wed, 30 Apr 2003 00:49:42 +0000 (00:49 +0000)]
Make sure rhostip is always initialized.

PR: bin/51508
Submitted by: Peter Grimshaw <peter@tesseract.demon.co.uk>

21 years agoTreat an empty PAM_RHOST the same as a NULL one.
Dag-Erling Smørgrav [Wed, 30 Apr 2003 00:44:05 +0000 (00:44 +0000)]
Treat an empty PAM_RHOST the same as a NULL one.

PR: bin/51508

21 years agoAdd vm_object locking to vmspace_swap_count().
Alan Cox [Wed, 30 Apr 2003 00:43:17 +0000 (00:43 +0000)]
Add vm_object locking to vmspace_swap_count().

21 years agoSet $HOME to the correct directory (within the chroot tree).
Dag-Erling Smørgrav [Wed, 30 Apr 2003 00:40:24 +0000 (00:40 +0000)]
Set $HOME to the correct directory (within the chroot tree).

21 years agoAdd support to CAM for devices with more than 2^32 blocks. (2TB if you're
Kenneth D. Merry [Wed, 30 Apr 2003 00:35:22 +0000 (00:35 +0000)]
Add support to CAM for devices with more than 2^32 blocks.  (2TB if you're
using 512 byte blocks).

cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to
64 bits and 32 bits respectively, so we can hold larger
device sizes.  cylinders would overflow at about 500GB.
Bump CAM_VERSION for this change.  Note that this will
require a recompile of all applications that talk to the
pass(4) driver.

scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12)
descriptions, add descriptions for SERVICE ACTION IN/OUT.
Add a new function, scsi_read_capacity_16(), that issues
the read capacity service action.  (Necessary for arrays
larger than 2^32 sectors.)  Update scsi_read_write() to use
a 64 bit LBA and issue READ(16) or WRITE(16) if necessary.
NOTE the API change.  This should be largely transparnet
to most userland applications at compile time, but will
break binary compatibility.  The CAM_VERSION bump, above,
also serves the purpose of forcing a recompile for any
applications that talk to CAM.

scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte
READ CAPACITY/SERVICE ACTION IN.  Add scsi_u64to8b() and
scsi_8btou64.

scsi_da.c: The da(4) driver probe now has two stages for devices
larger than 2TB.  If a standard READ CAPACITY(10) returns
0xffffffff, we issue the 16 byte version of read capacity
to determine the true array capacity.  We also do the same
thing in daopen() -- use the 16 byte read capacity if the
device is large enough.

The sysctl/loader code has also been updated to accept
16 bytes as a minimum command size.

21 years agoAdd missing crom_next() in the previous revision.
Hidetoshi Shimokawa [Tue, 29 Apr 2003 23:40:04 +0000 (23:40 +0000)]
Add missing crom_next() in the previous revision.

21 years agoAdd definitions for IN6ADDR_LINKLOCAL_ALLMDNS_INIT and INADDR_ALLMDNS_GROUP.
Matthew N. Dodd [Tue, 29 Apr 2003 22:03:46 +0000 (22:03 +0000)]
Add definitions for IN6ADDR_LINKLOCAL_ALLMDNS_INIT and INADDR_ALLMDNS_GROUP.

21 years agoIP_RECVTTL socket option.
Matthew N. Dodd [Tue, 29 Apr 2003 21:36:18 +0000 (21:36 +0000)]
IP_RECVTTL socket option.

Reviewed by: Stuart Cheshire <cheshire@apple.com>

21 years agoAllow fast instruction and data access mmu miss traps to be handled by
Jake Burkholder [Tue, 29 Apr 2003 21:30:59 +0000 (21:30 +0000)]
Allow fast instruction and data access mmu miss traps to be handled by
user trap handlers.

21 years ago`Hide' strlcpy and strlcat (using the namespace.h / __weak_reference
Jacques Vidrine [Tue, 29 Apr 2003 21:13:50 +0000 (21:13 +0000)]
`Hide' strlcpy and strlcat (using the namespace.h / __weak_reference
technique) so that we don't wind up calling into an application's
version if the application defines them.

Inspired by: qpopper's interfering and buggy version of strlcpy

21 years agoJump to the correct label upon detecting an error.
Daniel Eischen [Tue, 29 Apr 2003 21:05:17 +0000 (21:05 +0000)]
Jump to the correct label upon detecting an error.

21 years agoCreate the thread signal lock as a KSE lock (as opposed to
Daniel Eischen [Tue, 29 Apr 2003 21:03:33 +0000 (21:03 +0000)]
Create the thread signal lock as a KSE lock (as opposed to
a thread lock).

Better protect access to thread state while searching for
threads to handle a signal.

Better protect access to process pending signals while processing
a thread in sigwait().

Submitted by: davidxu

21 years agoUnbreak the build.
Hiroki Sato [Tue, 29 Apr 2003 20:39:38 +0000 (20:39 +0000)]
Unbreak the build.

21 years agoCall exit(0), not exit.
Alexander Kabaev [Tue, 29 Apr 2003 19:53:00 +0000 (19:53 +0000)]
Call exit(0), not exit.

21 years agoFix an obscure fencepost error in GBDE's sector mapping code:
Poul-Henning Kamp [Tue, 29 Apr 2003 19:46:42 +0000 (19:46 +0000)]
Fix an obscure fencepost error in GBDE's sector mapping code:

For certain combinations of sectorsize, mediasize and random numbers
(used to define the mapping), a multisector read or write would ignore
some subset of the sectors past the first sector in the request because
those sectors would be mapped past the end of the parent device, and
normal "end of media" truncation would zap that part of the request.

Rev 1.19+1.20 of g_bde_work.c added the check which should have alerted
me to this happening.  This commit maps the request correctly and
adds KASSERTS to make sure things stay inside the parent device.

This does not change the on-disk layout of GBDE, there is no need to
backup/restore.

21 years agoTypo in last commit: Do not press xZZ to leave vi(1).
Poul-Henning Kamp [Tue, 29 Apr 2003 19:38:55 +0000 (19:38 +0000)]
Typo in last commit:  Do not press xZZ to leave vi(1).

21 years agoWhen a bio comes back from below with a zero error code, check that
Poul-Henning Kamp [Tue, 29 Apr 2003 19:37:36 +0000 (19:37 +0000)]
When a bio comes back from below with a zero error code, check that
it wrote the full length.  The only case where this should be able
to happen is if we try to read/write past the end and the request
is truncated.  We obviously should never try to do that, so this
code should never activate.

21 years agoCatch up with the new acpi files.
Nate Lawson [Tue, 29 Apr 2003 19:19:47 +0000 (19:19 +0000)]
Catch up with the new acpi files.

21 years agoResolve conflicts.
Nate Lawson [Tue, 29 Apr 2003 19:11:00 +0000 (19:11 +0000)]
Resolve conflicts.

21 years agoSupport functions for the new ACPI import.
Nate Lawson [Tue, 29 Apr 2003 18:50:34 +0000 (18:50 +0000)]
Support functions for the new ACPI import.
  * AcpiOsDerivePciId(): finds a bus number, given the slot/func and the
    acpi parse tree.
  * AcpiOsPredefinedOverride(): use the sysctl hw.acpi.os_name to
    override the value for _OS.

Ideas from: takawata, jhb
Reviewed by: takawata, marcel
Tested on: i386, ia64

21 years agoRemove unused declarations and definitions.
Alan Cox [Tue, 29 Apr 2003 18:49:25 +0000 (18:49 +0000)]
Remove unused declarations and definitions.

21 years agoUpdated import-preparation script. It now removes vendor paths from header
Nate Lawson [Tue, 29 Apr 2003 18:48:17 +0000 (18:48 +0000)]
Updated import-preparation script.  It now removes vendor paths from header
info.  With these changes, no manual prep of the dist was necessary.

21 years agoResolve conflicts
Nate Lawson [Tue, 29 Apr 2003 18:46:31 +0000 (18:46 +0000)]
Resolve conflicts

21 years agoThis commit was generated by cvs2svn to compensate for changes in r114239,
Nate Lawson [Tue, 29 Apr 2003 18:39:30 +0000 (18:39 +0000)]
This commit was generated by cvs2svn to compensate for changes in r114239,
which included commits to RCS files with non-trunk default branches.

21 years agoImport of Intel ACPI 20030228 vendor distribution
Nate Lawson [Tue, 29 Apr 2003 18:39:30 +0000 (18:39 +0000)]
Import of Intel ACPI 20030228 vendor distribution

21 years agoThis commit was generated by cvs2svn to compensate for changes in r114237,
Nate Lawson [Tue, 29 Apr 2003 18:39:29 +0000 (18:39 +0000)]
This commit was generated by cvs2svn to compensate for changes in r114237,
which included commits to RCS files with non-trunk default branches.

21 years agoImport of Intel ACPI 20030228 vendor distribution
Nate Lawson [Tue, 29 Apr 2003 18:39:29 +0000 (18:39 +0000)]
Import of Intel ACPI 20030228 vendor distribution

21 years agoUnbreak the build.
Hiroki Sato [Tue, 29 Apr 2003 18:27:38 +0000 (18:27 +0000)]
Unbreak the build.