]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoFix merge conflicts, and because this still has a zillion warnings,
Mark Murray [Thu, 20 Dec 2001 14:25:46 +0000 (14:25 +0000)]
Fix merge conflicts, and because this still has a zillion warnings,
protect the build with a WARNS=2.

Fix the build.

22 years agoBring files back from the dead prior to remote commit to fix import
Mark Murray [Thu, 20 Dec 2001 13:40:48 +0000 (13:40 +0000)]
Bring files back from the dead prior to remote commit to fix import
conflicts.

22 years agoImport of OpenBSD's tip (which was the tip code of a month ago, modulo
Mark Murray [Thu, 20 Dec 2001 13:06:50 +0000 (13:06 +0000)]
Import of OpenBSD's tip (which was the tip code of a month ago, modulo
OpenBSD commits).

22 years agoTemporarily remove tip(1) from the build while I upgrade it.
Mark Murray [Thu, 20 Dec 2001 13:01:47 +0000 (13:01 +0000)]
Temporarily remove tip(1) from the build while I upgrade it.

22 years agoSilence harmless "smbfs_closel: Negative opencount" messages at
Sheldon Hearn [Thu, 20 Dec 2001 11:23:49 +0000 (11:23 +0000)]
Silence harmless "smbfs_closel: Negative opencount" messages at
unmount time.

Thanks to iedowse for the background information.

Submitted by: bp

22 years agoAdd missing snd_ich_load entry.
Jim Mock [Thu, 20 Dec 2001 09:14:45 +0000 (09:14 +0000)]
Add missing snd_ich_load entry.

Noticed by: will
MFC after: 2 days

22 years agoActually make use of the md_version field of 'struct mdio'. In order
Dima Dorfman [Thu, 20 Dec 2001 06:38:21 +0000 (06:38 +0000)]
Actually make use of the md_version field of 'struct mdio'.  In order
not to needlessly break compatibility, decrement MDIOVERSION to 0.

Approved by: phk

22 years agoI am not sure if it is a good idea or not to have my id in the cvs log
Peter Wemm [Thu, 20 Dec 2001 05:34:49 +0000 (05:34 +0000)]
I am not sure if it is a good idea or not to have my id in the cvs log
for this file, but here goes nothing.  This was my first attempt at
tidying up this file.  Unfortunately, it just exposes many more horrors
in the code itself that had been masked by the eyesore that was there
before.  I think this just needs to be put out of its misery.

22 years agoReplace a bunch of:
Peter Wemm [Thu, 20 Dec 2001 05:29:59 +0000 (05:29 +0000)]
Replace a bunch of:
      for (pv = TAILQ_FIRST(&m->md.pv_list);
               pv;
               pv = TAILQ_NEXT(pv, pv_list)) {
with:
      TAILQ_FOREACH(pv, &m->md.pv_list, pv_list) {

22 years agoAdd Apple's filesystem exercising tool.
Jordan K. Hubbard [Thu, 20 Dec 2001 04:15:57 +0000 (04:15 +0000)]
Add Apple's filesystem exercising tool.

Submitted by:   Conrad Minshall <conrad@apple.com>
Obtained from:  Apple Computer

22 years agoFix some whitespace nits, and a minor error that I made in some unused
Peter Wemm [Thu, 20 Dec 2001 03:49:31 +0000 (03:49 +0000)]
Fix some whitespace nits, and a minor error that I made in some unused
#ifdef DEBUG code (VM_MAXUSER_ADDRESS vs UPT_MAX_ADDRESS).

22 years agoFix some whitespace nits, converge with other pmap.c's
Peter Wemm [Thu, 20 Dec 2001 03:44:43 +0000 (03:44 +0000)]
Fix some whitespace nits, converge with other pmap.c's

22 years agoThis commit was generated by cvs2svn to compensate for changes in r88237,
Peter Wemm [Thu, 20 Dec 2001 02:23:44 +0000 (02:23 +0000)]
This commit was generated by cvs2svn to compensate for changes in r88237,
which included commits to RCS files with non-trunk default branches.

22 years agoFix the 'nm' -> segfault bug (on the vendor branch). This has been fixed
Peter Wemm [Thu, 20 Dec 2001 02:23:44 +0000 (02:23 +0000)]
Fix the 'nm' -> segfault bug (on the vendor branch).  This has been fixed
the same way in the cygnus base source.  The problem was that bfd was
writing to memory obtained from malloc(0).    The next import will update
this if necessary.  I'm not sure of the origins of this patch.

Discussed with: obrien

22 years agoMention that the Xircom and Conexant chips are supported in
Bill Paul [Thu, 20 Dec 2001 00:02:13 +0000 (00:02 +0000)]
Mention that the Xircom and Conexant chips are supported in
the dc(4) man page.

22 years agoPseudofs was leaking VFS cache entries badly due to its cache and use of
Matthew Dillon [Wed, 19 Dec 2001 23:58:09 +0000 (23:58 +0000)]
Pseudofs was leaking VFS cache entries badly due to its cache and use of
the wrong VOP descriptor.  This misuse caused VFS-cached vnodes to be
re-cached, resulting in the leak.  This commit is an interim fix until DES
has a chance to rework the code involved.

22 years agoClarify dirprefs release note.
Bruce A. Mah [Wed, 19 Dec 2001 23:38:11 +0000 (23:38 +0000)]
Clarify dirprefs release note.

22 years agoAdd missing {} to 'for'. It fix coredump (result of NetBSD changes merge)
Andrey A. Chernov [Wed, 19 Dec 2001 22:25:11 +0000 (22:25 +0000)]
Add missing {} to 'for'. It fix coredump (result of NetBSD changes merge)

22 years agoAdd (unsigned char) cast to ctype macros
Andrey A. Chernov [Wed, 19 Dec 2001 21:50:22 +0000 (21:50 +0000)]
Add (unsigned char) cast to ctype macros
ftell->ftello, fseek->fseeko
File 'newsize' type int->off_t
Add visible (long) cast to fsize() when it called to small one message file and
result is assigned to long.

22 years agoFix
Poul-Henning Kamp [Wed, 19 Dec 2001 21:19:43 +0000 (21:19 +0000)]
Fix
        md5 -q -s foo
to not expect input on stdin after hashing the string.

MFC after:      1 week

22 years agoComment out DEVICE_POLLING so that LINT compiles again.
Poul-Henning Kamp [Wed, 19 Dec 2001 20:53:54 +0000 (20:53 +0000)]
Comment out DEVICE_POLLING so that LINT compiles again.

22 years agoMake this driver a better citizen by moving dev creation and
Warner Losh [Wed, 19 Dec 2001 19:37:31 +0000 (19:37 +0000)]
Make this driver a better citizen by moving dev creation and
other initialization into attach from probe.  Also hide a few
printfs behind a bootverbose.

approved in principle by: phk

22 years agoCalculate whether the sbuf is dynamic *before* bzero()ing the
Matthew Dillon [Wed, 19 Dec 2001 19:04:57 +0000 (19:04 +0000)]
Calculate whether the sbuf is dynamic *before* bzero()ing the
structure.  This fixes a serious memory leak in the sbuf code.

MFC after: 3 days

22 years agoCorrect unexpected interrupt detection.
Orion Hodson [Wed, 19 Dec 2001 18:26:53 +0000 (18:26 +0000)]
Correct unexpected interrupt detection.

22 years agoFix the "conexant chips don't work in full duplexmode" problem. According
Bill Paul [Wed, 19 Dec 2001 18:23:45 +0000 (18:23 +0000)]
Fix the "conexant chips don't work in full duplexmode" problem. According
to Phil Kernick:

"The problem is that in full duplex mode, the Conexant chip always reports a
carrier lost error, even when the frame is successfully sent.  So, if we
have a Conexant chip, then ignore carrier lost when in full duplex
mode."

Since the Xircom chips seem to have the same issue and since we already
have a workaround for this, just expand the workaround test to also
check for DC_IS_CONEXANT().

22 years agoFix compiler warning in dc_intr(): if the only code that does a "goto"
Bill Paul [Wed, 19 Dec 2001 18:13:44 +0000 (18:13 +0000)]
Fix compiler warning in dc_intr(): if the only code that does a "goto"
to a label is inside an #ifdef block, then the label should *also* be
inside an #ifdef block. Hide the "done:" label which is only used if
DEVICE_POLLING is enabled under #ifdef DEVICE_POLLING.

22 years agoNitlets.
Orion Hodson [Wed, 19 Dec 2001 18:04:04 +0000 (18:04 +0000)]
Nitlets.

22 years agoClear resume interrupts - these can occur during initialization and
Orion Hodson [Wed, 19 Dec 2001 17:38:38 +0000 (17:38 +0000)]
Clear resume interrupts - these can occur during initialization and
must be cleared to prevent machine hanging (presently aflicts -current
and -stable).

Problem reported by Bruce Montague <brucem@cse.iitkgp.ernet.in>

PR: kern/29769 (probably)

22 years agoDon't assume that /var/run and /var/spool/lock are on the same
Brian Somers [Wed, 19 Dec 2001 16:10:25 +0000 (16:10 +0000)]
Don't assume that /var/run and /var/spool/lock are on the same
partition.

Submitted by: ru

22 years agoIt's no longer necessary to ensure that ``gate'' is set when RTF_GATEWAY
Brian Somers [Wed, 19 Dec 2001 16:05:27 +0000 (16:05 +0000)]
It's no longer necessary to ensure that ``gate'' is set when RTF_GATEWAY
is passed, as subsequent code does that check now anyway.

Submitted by: ru

22 years agoOnly call rt_getifa() if we've either been passed a gateway or
Brian Somers [Wed, 19 Dec 2001 16:03:27 +0000 (16:03 +0000)]
Only call rt_getifa() if we've either been passed a gateway or
if we've been given an RTA_IFP or changed RTA_IFA sockaddr.

This fixes the following bug:
  >/dev/tun100
  >/dev/tun101
  ifconfig tun100 1.2.3.4 5.6.7.8
  ifconfig tun101 1.2.3.4 6.7.8.9
  route change 6.7.8.9 -ifa 1.2.3.4 -iface -mtu 500
which erroneously changed tun101's host route to have an ifp of tun100
(rt_getifa() sets the ifp after calling ifa_ifwithnet(1.2.3.4))

This incarnation submitted by: ru

22 years agoWhen storing an int value in a void *, use intptr_t as the cast type
Jonathan Lemon [Wed, 19 Dec 2001 15:57:43 +0000 (15:57 +0000)]
When storing an int value in a void *, use intptr_t as the cast type
(instead of int) to keep the 64 bit platforms happy.

22 years agoDon't try to free a NULL route when doing IPFIREWALL_FORWARD.
Yaroslav Tykhiy [Wed, 19 Dec 2001 14:54:13 +0000 (14:54 +0000)]
Don't try to free a NULL route when doing IPFIREWALL_FORWARD.
An old route will be NULL at that point if a packet were initially
routed to an interface (using the IP_ROUTETOIF flag.)

Submitted by: Igor Timkin <ivt@gamma.ru>

22 years agoAdd another PS2 PNP id.
Sheldon Hearn [Wed, 19 Dec 2001 13:32:21 +0000 (13:32 +0000)]
Add another PS2 PNP id.

PR: 32973
Submitted by: KT Sin <ktsin@acm.org>

22 years agoAllow retrieval of the virtual address of the AGP aperture
Matthew N. Dodd [Wed, 19 Dec 2001 08:54:29 +0000 (08:54 +0000)]
Allow retrieval of the virtual address of the AGP aperture
using agp_get_info().

MFC after: 1 week

22 years agoDon't put variable declarations in header files, put prototypes.
Matthew N. Dodd [Wed, 19 Dec 2001 08:49:11 +0000 (08:49 +0000)]
Don't put variable declarations in header files, put prototypes.

'pci_devq' provides useful information now.

22 years agoMake ``df -l'' work when no network filesystems are loaded.
Ruslan Ermilov [Wed, 19 Dec 2001 08:38:02 +0000 (08:38 +0000)]
Make ``df -l'' work when no network filesystems are loaded.

PR: bin/32397

22 years agoExtend the SYN DoS defense by adding syncookies to the syncache.
Jonathan Lemon [Wed, 19 Dec 2001 06:12:14 +0000 (06:12 +0000)]
Extend the SYN DoS defense by adding syncookies to the syncache.
All TCP ISNs that are sent out are valid cookies, which allows entries
in the syncache to be dropped and still have the ACK accepted later.
As all entries pass through the syncache, there is no sudden switchover
from cache -> cookies when the cache is full; instead, syncache entries
simply have a reduced lifetime.  More details may be found in the
"Resisting DoS attacks with a SYN cache" paper in the Usenix BSDCon 2002
conference proceedings.

Sponsored by: DARPA, NAI Labs

22 years agoxinstall is part of the upgrade path for 4.4-RELEASE as well.
Warner Losh [Wed, 19 Dec 2001 06:05:42 +0000 (06:05 +0000)]
xinstall is part of the upgrade path for 4.4-RELEASE as well.
Do the usualy ifdef trick here.

Reported by: dworkin muller <addr-removed> and faried nawaz <fn@hungry.com>

22 years agoForced commit: Previous commit message should have read:
Bruce A. Mah [Wed, 19 Dec 2001 05:34:40 +0000 (05:34 +0000)]
Forced commit:  Previous commit message should have read:

MFCs noted:  sh(1) "test" built-in, ftpd(8) -o/-O

22 years agoMFC: ftpd(8) -o/-O.
Bruce A. Mah [Wed, 19 Dec 2001 05:30:35 +0000 (05:30 +0000)]
MFC:  ftpd(8) -o/-O.

Requested by: sheldonh

22 years agoNew release note: i386 performance enhancements for dc(4) and sis(4).
Bruce A. Mah [Wed, 19 Dec 2001 05:21:43 +0000 (05:21 +0000)]
New release note:  i386 performance enhancements for dc(4) and sis(4).

Suggested by: luigi

22 years agoDisplay the right message for the SiS 645 chipset.
John Baldwin [Wed, 19 Dec 2001 01:53:11 +0000 (01:53 +0000)]
Display the right message for the SiS 645 chipset.

Submitted by: Kent Stewart <kstewart@owt.com>

22 years agoDo not initialize static/global variables to 0. Use bss instead of
Peter Wemm [Wed, 19 Dec 2001 01:35:18 +0000 (01:35 +0000)]
Do not initialize static/global variables to 0. Use bss instead of
taking up space in the data section.

22 years agoUse a different mechanism to get the vnlru process to wake up and notice
Peter Wemm [Wed, 19 Dec 2001 01:31:12 +0000 (01:31 +0000)]
Use a different mechanism to get the vnlru process to wake up and notice
the shutdown request at reboot/halt time.
Disable the printf 'vnlru process getting nowhere, pausing...' and instead
export the count to the debug.vnlru_nowhere sysctl.

22 years agoFix comment for the SiS 645 chipset to be 645 instead of 635.
John Baldwin [Wed, 19 Dec 2001 01:23:32 +0000 (01:23 +0000)]
Fix comment for the SiS 645 chipset to be 645 instead of 635.

22 years agoDocument that rfork() will return EINVAL if flags not listed in the
John Baldwin [Wed, 19 Dec 2001 00:59:54 +0000 (00:59 +0000)]
Document that rfork() will return EINVAL if flags not listed in the
manpage are passed in.

22 years agoMake RFPPWAIT a kernel-only flag. Programs desiring this feature should
John Baldwin [Wed, 19 Dec 2001 00:55:13 +0000 (00:55 +0000)]
Make RFPPWAIT a kernel-only flag.  Programs desiring this feature should
use vfork() rather than manually invoking rfork().

22 years agoComplete the device polling support by adding a thread in charge
Luigi Rizzo [Wed, 19 Dec 2001 00:53:24 +0000 (00:53 +0000)]
Complete the device polling support by adding a thread in charge
of polling interfaces at the lowest possible priority
(this might result in softnetisr being scheduled, but there is
no risk of livelock because they have a higher priority than
this thread).

22 years agoReturn EINVAL if kernel only flags are passed to the rfork syscall rather
John Baldwin [Wed, 19 Dec 2001 00:53:23 +0000 (00:53 +0000)]
Return EINVAL if kernel only flags are passed to the rfork syscall rather
than silently masking them.

22 years agoFix typo in usage (power same -> power save).
Pierre Beyssac [Tue, 18 Dec 2001 23:28:45 +0000 (23:28 +0000)]
Fix typo in usage (power same -> power save).

22 years agoAxe stale extern for a non-existent variable.
John Baldwin [Tue, 18 Dec 2001 22:42:09 +0000 (22:42 +0000)]
Axe stale extern for a non-existent variable.

22 years agoChange the test for _MACHINE_ARCH == i386 into #ifdef __i386__ since it
Joerg Wunsch [Tue, 18 Dec 2001 22:16:33 +0000 (22:16 +0000)]
Change the test for _MACHINE_ARCH == i386 into #ifdef __i386__ since it
otherwise breaks on the Alpha arch.  I think this is wrong since i'd
actually like to probe for a PC architecture, not for a particular CPU
type.  Anyway, now it's again the way it used to be.

22 years agoSync with most of NetBSD's changes, including:
Mike Heffner [Tue, 18 Dec 2001 20:52:09 +0000 (20:52 +0000)]
Sync with most of NetBSD's changes, including:

 *) Sync with 4.4BSD-Lite2
 *) Set usecs for utimes()
 *) Add 'inc' command and 'autoinc' option that check for new mail
    manually and automatically, respectively
 *) Use POSIX signal handling and tty semantics
 *) Handle long lines correctly when paging messages
 *) Add ability to explicitly search 'To:' line
 *) Various manpage cleanups
 *) Support overriding '~/.mailrc' with $MAILRC
 *) Support 'askbcc' and 'asksub' options
 *) Fix various bugs

Reviewed by: ru (mail.1)
Obtained from: NetBSD

22 years agoThis is a forward port of Peter's vlrureclaim() fix, with some minor mods
Matthew Dillon [Tue, 18 Dec 2001 20:48:54 +0000 (20:48 +0000)]
This is a forward port of Peter's vlrureclaim() fix, with some minor mods
by me to make it more efficient.  The original code had serious balancing
problems and could also deadlock easily.  This code relegates the vnode
reclamation to its own kproc and relaxes the vnode reclamation requirements
to better maintain kern.maxvnodes.  This code still doesn't balance as well
as it could, but it does a much better job then the original code.

Approved by: re@freebsd.org
Obtained from: ps, peter, dillon
MFS Assuming: Assuming no problems crop up in Yahoo testing
MFC after: 7 days

22 years agoset TMPDIR to /tmp
Brian Feldman [Tue, 18 Dec 2001 20:05:20 +0000 (20:05 +0000)]
set TMPDIR to /tmp

22 years agoMerge part of revision 1.18 of sys/i386/linux/linux_machdep.c: don't use
John Baldwin [Tue, 18 Dec 2001 18:50:02 +0000 (18:50 +0000)]
Merge part of revision 1.18 of sys/i386/linux/linux_machdep.c: don't use
RFTHREAD in linux_clone().

22 years agoIn a couple of places, we recalculated addresses we already had in local
Julian Elischer [Tue, 18 Dec 2001 18:46:32 +0000 (18:46 +0000)]
In a couple of places, we recalculated addresses we already had in local
pointer variables.

22 years agoDon't use RFTHREAD when creating a kernel process.
John Baldwin [Tue, 18 Dec 2001 18:42:02 +0000 (18:42 +0000)]
Don't use RFTHREAD when creating a kernel process.

22 years agoAdd LIBSMB.
Ruslan Ermilov [Tue, 18 Dec 2001 18:27:27 +0000 (18:27 +0000)]
Add LIBSMB.

22 years agoWhitespace police.
Ruslan Ermilov [Tue, 18 Dec 2001 18:21:51 +0000 (18:21 +0000)]
Whitespace police.

22 years agoChange the atomic_set_char to atomic_set_int and atomic_clear_char
Kirk McKusick [Tue, 18 Dec 2001 18:05:17 +0000 (18:05 +0000)]
Change the atomic_set_char to atomic_set_int and atomic_clear_char
to atomic_clear_int to ease the implementation for the sparc64.

Requested by: Jake Burkholder <jake@locore.ca>

22 years agoWhitespace fix.
John Baldwin [Tue, 18 Dec 2001 18:03:48 +0000 (18:03 +0000)]
Whitespace fix.

22 years agoThe smbutils import really only applies to the i386...fix this
Bruce A. Mah [Tue, 18 Dec 2001 16:26:57 +0000 (16:26 +0000)]
The smbutils import really only applies to the i386...fix this
(noticed by sheldonh).  While I'm here, note its MFC, also add some
cross-references from the kernel part.

22 years agoFixed the bug in transparent TCP proxying with the "encode_ip_hdr"
Ruslan Ermilov [Tue, 18 Dec 2001 16:13:45 +0000 (16:13 +0000)]
Fixed the bug in transparent TCP proxying with the "encode_ip_hdr"
option -- TcpAliasOut() did not catch the IP header length change.

Submitted by: Stepachev Andrey <aka50@mail.ru>

22 years agoAdd tbl and tbsp as contractions of tablespoon, and tsp for teaspoon.
Josef Karthauser [Tue, 18 Dec 2001 11:25:58 +0000 (11:25 +0000)]
Add tbl and tbsp as contractions of tablespoon, and tsp for teaspoon.

Submitted by: banshee <banshee@pork.abattoir.com>
MFC after: 7 days

22 years agomdoc(7) police: markup, spelling, and grammar fixes.
Ruslan Ermilov [Tue, 18 Dec 2001 09:40:22 +0000 (09:40 +0000)]
mdoc(7) police: markup, spelling, and grammar fixes.

PR: docs/32955

22 years ago- Change all callers of addupc_task() to check PS_PROFIL explicitly and
John Baldwin [Tue, 18 Dec 2001 09:06:10 +0000 (09:06 +0000)]
- Change all callers of addupc_task() to check PS_PROFIL explicitly and
  remove the check from addupc_task().  It would need sched_lock while
  testing the flag anyways.
- Always read sticks while holding sched_lock using a temporary variable
  where needed.
- Always init prticks to 0 in ast() to quiet a warning.

22 years agoVarious assembly fixes mostly in the form of using the "+" modifier for
John Baldwin [Tue, 18 Dec 2001 08:54:39 +0000 (08:54 +0000)]
Various assembly fixes mostly in the form of using the "+" modifier for
output operands to mark them as both input and output rather than listing
operands twice.

Reviewed by: bde

22 years agoAllow the ATOMIC_ASM() macro to pass in the constraints on the V parameter
John Baldwin [Tue, 18 Dec 2001 08:51:34 +0000 (08:51 +0000)]
Allow the ATOMIC_ASM() macro to pass in the constraints on the V parameter
since the char versions need to use either ax, bx, cx, or dx.

Submitted by: Peter Jeremy (mostly)
Recommended by: bde

22 years agoFix warnings: the bge_bpd_read*() functions are not used (#ifdef notdef)
Peter Wemm [Tue, 18 Dec 2001 08:03:25 +0000 (08:03 +0000)]
Fix warnings: the bge_bpd_read*() functions are not used (#ifdef notdef)

22 years agoFix warning; line 1640: ciss_abort_request defined but not used.
Peter Wemm [Tue, 18 Dec 2001 08:01:48 +0000 (08:01 +0000)]
Fix warning; line 1640: ciss_abort_request defined but not used.

22 years ago- Add a function to convert 6 byte SCSI commands for read, write, mode
John Baldwin [Tue, 18 Dec 2001 07:55:17 +0000 (07:55 +0000)]
- Add a function to convert 6 byte SCSI commands for read, write, mode
  sense, and mode select into their 10 byte equivalents.  Eventually the
  da(4) driver will become more intelligent about this, or at least allow
  umass(4) to pass quirks in directly.  However, this is a functional
  workaround until a better fix is implemented.
- Use the 6 to 10 conversion function to allow the ATAPI and UFI command
  sets to emulate 6 byte commands with 10 byte commands.
- Use the ATAPI command set rather than UFI for the ScanLogic SL11R-IDE
  as it supports the SYNCH_CACHE command.
- Enable ATAPI command set support.
- Pass READ/WRITE_12 commands through for UFI support as the UFI spec
  says they should be supported.
- Update a comment in the UFI translation function since we handle
  MODE_SELECT.

22 years agoThis commit was generated by cvs2svn to compensate for changes in r88109,
David E. O'Brien [Tue, 18 Dec 2001 04:13:58 +0000 (04:13 +0000)]
This commit was generated by cvs2svn to compensate for changes in r88109,
which included commits to RCS files with non-trunk default branches.

22 years agoVirgin import of GCC 3.0.2's libf2c
David E. O'Brien [Tue, 18 Dec 2001 04:13:58 +0000 (04:13 +0000)]
Virgin import of GCC 3.0.2's libf2c

22 years agoAdd calibration test to determine extent of AC97 overclocking (if any).
Orion Hodson [Tue, 18 Dec 2001 03:29:31 +0000 (03:29 +0000)]
Add calibration test to determine extent of AC97 overclocking (if any).

22 years agoWith GCC 3, we *must* use the GCC stdarg.h. We can no longer just make
David E. O'Brien [Tue, 18 Dec 2001 03:28:08 +0000 (03:28 +0000)]
With GCC 3, we *must* use the GCC stdarg.h.  We can no longer just make
_BSD_VA_LIST_ match what we think the compiler is doing.

22 years agoFix the retrieval of USRSTACK via sysctl so that it works for 64-bit
Daniel Eischen [Tue, 18 Dec 2001 02:02:59 +0000 (02:02 +0000)]
Fix the retrieval of USRSTACK via sysctl so that it works for 64-bit
archs.  This should fix libc_r on alpha.

Submitted by: Bernd Walter <ticso@cicely9.cicely.de>

22 years agoAvoid passing the variable `tl' to functions that just use it for
Ian Dowse [Tue, 18 Dec 2001 01:22:09 +0000 (01:22 +0000)]
Avoid passing the variable `tl' to functions that just use it for
temporary storage. In the old NFS code it wasn't at all clear if
the value of `tl' was used across or after macro calls, but I'm
fairly confident that the convention was to keep its use local.
Each ex-macro function now uses a local version of this variable,
so all of the double-indirection goes away.

The only exception to the `local use' rule for `tl' is nfsm_clget(),
which is left unchanged by this commit.

Reviewed by: peter

22 years agoConst'ify the CDB format string passed to the CDB parsing routines
Kelly Yancey [Tue, 18 Dec 2001 00:48:44 +0000 (00:48 +0000)]
Const'ify the CDB format string passed to the CDB parsing routines
(csio_decode_visit() and family).

Reviewed by: ken

22 years agoModify the critical section API as follows:
John Baldwin [Tue, 18 Dec 2001 00:27:18 +0000 (00:27 +0000)]
Modify the critical section API as follows:
- The MD functions critical_enter/exit are renamed to start with a cpu_
  prefix.
- MI wrapper functions critical_enter/exit maintain a per-thread nesting
  count and a per-thread critical section saved state set when entering
  a critical section while at nesting level 0 and restored when exiting
  to nesting level 0.  This moves the saved state out of spin mutexes so
  that interlocking spin mutexes works properly.
- Most low-level MD code that used critical_enter/exit now use
  cpu_critical_enter/exit.  MI code such as device drivers and spin
  mutexes use the MI wrappers.  Note that since the MI wrappers store
  the state in the current thread, they do not have any return values or
  arguments.
- mtx_intr_enable() is replaced with a constant CRITICAL_FORK which is
  assigned to curthread->td_savecrit during fork_exit().

Tested on: i386, alpha

22 years agoNew release note: Network device polling.
Bruce A. Mah [Mon, 17 Dec 2001 23:27:49 +0000 (23:27 +0000)]
New release note:  Network device polling.

Modified release note:  Smaller local slowstart flightsize now that
buffers are larger.

22 years agoSmall cleanups to the SMP code:
John Baldwin [Mon, 17 Dec 2001 23:14:35 +0000 (23:14 +0000)]
Small cleanups to the SMP code:
- Axe inlvtlb_ok as it was completely redundant with smp_active.
- Remove references to non-existent variable and non-existent file
  in i386/include/smp.h.
- Don't perform initializations local to each CPU while holding the
  ap boot lock on i386 while an AP bootstraps itself.
- Reorganize the AP startup code some to unify the latter half of the
  functions to bring an AP up.  Eventually this might be broken out into
  a MI function in subr_smp.c.

22 years ago1) Localize (LC_CTYPE)
Andrey A. Chernov [Mon, 17 Dec 2001 23:14:14 +0000 (23:14 +0000)]
1) Localize (LC_CTYPE)
2) Catch "" to 0 conversion for OSes that not catch it in strto*()
   (f.e. -stable). It is needed because POSIX agrees with both variants.

22 years agoFix a problem where stats overflow interrupts would cause
Mike Silbersack [Mon, 17 Dec 2001 22:24:19 +0000 (22:24 +0000)]
Fix a problem where stats overflow interrupts would cause
a major slowdown, and re-enable stats overflow interrupts.

For future reference, the bug was in our code, and not
some bug in the 3com chips.

Reviewed by: wpaul
MFC after: 2 days

22 years agoCleaning up some entries missed on the first sweep. Add a directory in
Crist J. Clark [Mon, 17 Dec 2001 22:05:23 +0000 (22:05 +0000)]
Cleaning up some entries missed on the first sweep. Add a directory in
/var/run to store ppp(8) command sockets.

PR: bin/29966
MFC after: 4

22 years agoNew release note: smbfs userland import.
Bruce A. Mah [Mon, 17 Dec 2001 21:32:36 +0000 (21:32 +0000)]
New release note:  smbfs userland import.

Modified release notes:  bge(4) input checksum offload disabled,
ncv/nsp/stg modules.

MFCs noted:  bootloader -p

22 years agoBack out cometic changes. This is for easily syncing with KAME in other BSDs.
Munechika SUMIKAWA [Mon, 17 Dec 2001 18:13:12 +0000 (18:13 +0000)]
Back out cometic changes.  This is for easily syncing with KAME in other BSDs.

22 years agoInstall script via SCRIPTS.
Ruslan Ermilov [Mon, 17 Dec 2001 16:52:32 +0000 (16:52 +0000)]
Install script via SCRIPTS.

22 years agoInstall files via FILES.
Ruslan Ermilov [Mon, 17 Dec 2001 16:45:09 +0000 (16:45 +0000)]
Install files via FILES.

22 years agoInstall files via FILES.
Ruslan Ermilov [Mon, 17 Dec 2001 16:21:59 +0000 (16:21 +0000)]
Install files via FILES.

22 years agoOops, forgot to remove the `beforeinstall' block.
Ruslan Ermilov [Mon, 17 Dec 2001 16:04:40 +0000 (16:04 +0000)]
Oops, forgot to remove the `beforeinstall' block.

22 years agoInstall files via FILES.
Ruslan Ermilov [Mon, 17 Dec 2001 16:02:15 +0000 (16:02 +0000)]
Install files via FILES.

22 years agoAdd Italian translations to the example for refusing docproj
Michael Lucas [Mon, 17 Dec 2001 15:41:13 +0000 (15:41 +0000)]
Add Italian translations to the example for refusing docproj
translations.  This will once again allow docproj trackers to use the
sample out-of-the-box to only download English.

Rapid MFC to avoid the coming freeze.

PR: 32329
Approved by: bmah
MFC after: 1 day

22 years agoInstall files via FILES.
Ruslan Ermilov [Mon, 17 Dec 2001 15:23:57 +0000 (15:23 +0000)]
Install files via FILES.

22 years agoFix some style bugs
Alexey Zelkin [Mon, 17 Dec 2001 15:11:29 +0000 (15:11 +0000)]
Fix some style bugs

Prompted by: mike

22 years agoFILES support for bsd.prog.mk. See bsd.README for details.
Ruslan Ermilov [Mon, 17 Dec 2001 13:59:35 +0000 (13:59 +0000)]
FILES support for bsd.prog.mk.  See bsd.README for details.

Stolen from: NetBSD

22 years agoRemoved a no-op FILES from bsd.doc.mk.
Ruslan Ermilov [Mon, 17 Dec 2001 13:45:56 +0000 (13:45 +0000)]
Removed a no-op FILES from bsd.doc.mk.

22 years agoAdd support for devfs. In other words, for -CURRENT, make work at all
Brian Feldman [Mon, 17 Dec 2001 13:42:46 +0000 (13:42 +0000)]
Add support for devfs.  In other words, for -CURRENT, make work at all
again for normal device nodes.

22 years agoInstall devfs includes.
Brian Feldman [Mon, 17 Dec 2001 13:38:58 +0000 (13:38 +0000)]
Install devfs includes.