]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years ago_gethostbynis{addr,name}() can handle an IPv6, now.
ume [Thu, 28 Apr 2005 17:44:44 +0000 (17:44 +0000)]
_gethostbynis{addr,name}() can handle an IPv6, now.

19 years agoPatches from Ruslan Ermilov to address problems compiling LINT
darrenr [Thu, 28 Apr 2005 16:33:15 +0000 (16:33 +0000)]
Patches from Ruslan Ermilov to address problems compiling LINT

19 years agoPatches from Ruslam Ermilov to remove NetBSD bits from Makefiles and cleanup
darrenr [Thu, 28 Apr 2005 16:26:35 +0000 (16:26 +0000)]
Patches from Ruslam Ermilov to remove NetBSD bits from Makefiles and cleanup
build problems with rescue.

19 years agoWe need to pass NGROUPS + 1 to getgrouplist(3) to display
robert [Thu, 28 Apr 2005 16:11:37 +0000 (16:11 +0000)]
We need to pass NGROUPS + 1 to getgrouplist(3) to display
NGROUPS groups.  getgrouplist(3) may put a duplicate group
id into the passed array (it sets [0] and [1] to the value
of the gid argument), but id_print() sorts them out.

19 years ago - Merge two functions for printing `id' output.
robert [Thu, 28 Apr 2005 15:55:54 +0000 (15:55 +0000)]
 - Merge two functions for printing `id' output.
   Showing the ids of both an user given by an argument to `id',
   and the current user, is now handled in a single function.
   Displaying the current user's ids was inaccurate because
   getgroups(2) had been used.  getgroups(2) returns the current
   kernel state of a user's groups, which may not always be
   correct if /etc/group was recently changed.
 - Fix a few style bugs.

PR: bin/78085

19 years agoIntroduce a flag to enable extended warnings (-x) and make them off
harti [Thu, 28 Apr 2005 15:37:25 +0000 (15:37 +0000)]
Introduce a flag to enable extended warnings (-x) and make them off
by default. This should fix the problem of getting lots of errors
when building with an up-to-date make and old *.mk files.

19 years agomake getnetby*() thread-safe.
ume [Thu, 28 Apr 2005 15:32:55 +0000 (15:32 +0000)]
make getnetby*() thread-safe.

19 years agoBe resistant to DESTDIR being set to some unkosher value, e.g. "/".
ru [Thu, 28 Apr 2005 14:45:57 +0000 (14:45 +0000)]
Be resistant to DESTDIR being set to some unkosher value, e.g. "/".

Noticed by: Steve Ames

19 years agoAdd support for the P600 and name the E400.
ps [Thu, 28 Apr 2005 14:40:23 +0000 (14:40 +0000)]
Add support for the P600 and name the E400.

19 years agoAdd an initial regression test suite for make(1).
harti [Thu, 28 Apr 2005 13:20:48 +0000 (13:20 +0000)]
Add an initial regression test suite for make(1).

Submitted by: Max Okumoto <okumotu@ucsd.edu> (partly, initial version)

19 years agoOnly allow RAID5 plexes to be parity checked.
le [Thu, 28 Apr 2005 13:09:00 +0000 (13:09 +0000)]
Only allow RAID5 plexes to be parity checked.

PR:           kern/80427
Submitty by:  Stijn Hoop <stijn@win.tue.nl>

19 years ago - Act according to the documentation (man page):
robert [Thu, 28 Apr 2005 12:44:02 +0000 (12:44 +0000)]
 - Act according to the documentation (man page):
   When adding users from a preformatted file, do not exit
   silently when empty lines or lines starting with a '#'
   are encountered - ignore them instead.
 - Fix a spelling error in a comment.

PR: bin/80058

19 years ago - Set negative row or column argument values to zero.
robert [Thu, 28 Apr 2005 12:37:15 +0000 (12:37 +0000)]
 - Set negative row or column argument values to zero.
   Negative values would produce undefined behaviour including
   a possible segmentation fault.
 - Explicitly initialize the global row and column variables
   to zero.

PR: bin/80348

19 years agoUse a minimal perfect hash for the special sources/targets too. Add
harti [Thu, 28 Apr 2005 12:05:43 +0000 (12:05 +0000)]
Use a minimal perfect hash for the special sources/targets too. Add
the corresponding magic to create the hash function to the Makefile.

19 years agoReturn the correct register number in the 'get_msr()' MD function.
jkoshy [Thu, 28 Apr 2005 08:13:19 +0000 (08:13 +0000)]
Return the correct register number in the 'get_msr()' MD function.

Only allow a process to use the x86 RDPMC instruction if it has
allocated and attached a PMC to itself.

Inform the MD layer of the "pseudo context switch out" that needs
to be done when the last thread of a process is exiting.

19 years agoFix the $$FreeBSD$$ that should be written verbatim to directive_hash.c
harti [Thu, 28 Apr 2005 08:04:46 +0000 (08:04 +0000)]
Fix the $$FreeBSD$$ that should be written verbatim to directive_hash.c
so that it doesn't get replaced with the Makefile's revision.

19 years agoX logins should be recorded in lastlog / wtmp / utmp. I have no idea why
des [Thu, 28 Apr 2005 07:59:09 +0000 (07:59 +0000)]
X logins should be recorded in lastlog / wtmp / utmp.  I have no idea why
this wasn't there already...  it makes much more sense this way.

MFC after: 2 weeks

19 years agoMove the hash function for directives into its own file and add
harti [Thu, 28 Apr 2005 07:55:55 +0000 (07:55 +0000)]
Move the hash function for directives into its own file and add
a Makefile target to re-created this file. Note, that there is no
explicite dependency to automatically re-create the file, because this
is needed only when the directive table changes and it requires the
(yet to come) devel/mph port.

Submitted by: Max Okumoto <okumoto@ucsd.edu> (first version)

19 years agoUpdate comment to direct the reader to libkern.h instead of systm.h.
marcel [Thu, 28 Apr 2005 05:50:18 +0000 (05:50 +0000)]
Update comment to direct the reader to libkern.h instead of systm.h.
The functions were moved.

Pointed out by: johan@

19 years agoSlightly reformat apb_alloc_resource() to create some horizontal space
marcel [Thu, 28 Apr 2005 03:33:46 +0000 (03:33 +0000)]
Slightly reformat apb_alloc_resource() to create some horizontal space
for enhancements. Shorten apb_map_checkrange() to apb_checkrange() for
the same reason. No functional change.

19 years agoInline functions belong in <sys/libkern.h>, not in <sys/systm.h>.
marcel [Thu, 28 Apr 2005 03:19:50 +0000 (03:19 +0000)]
Inline functions belong in <sys/libkern.h>, not in <sys/systm.h>.
Move crc32() and crc32_raw() from the latter to the former. Move
the declaration of crc32_tab[] to <sys/libkern.h> as well.

Pointed out by: bde@
Tested on: ia64, sparc64

19 years agoDelete stale files post their move to "sys/dev/hwpmc/".
jkoshy [Thu, 28 Apr 2005 01:32:58 +0000 (01:32 +0000)]
Delete stale files post their move to "sys/dev/hwpmc/".

19 years agoUse the new path (post repo-copy) to our sources.
jkoshy [Thu, 28 Apr 2005 00:53:45 +0000 (00:53 +0000)]
Use the new path (post repo-copy) to our sources.

19 years agoRefactor the CRC-32 code to enhance its usability. Move the actual
marcel [Wed, 27 Apr 2005 22:26:45 +0000 (22:26 +0000)]
Refactor the CRC-32 code to enhance its usability. Move the actual
CRC logic to a new function: crc32_raw() that obtains the initial
CRC value as well as leaves any post-processing to the caller. As
such, it can be used when the initial CRC value is not ~0U or when
the final CRC value does need to be inverted (bitwise). It also
means that crc32_raw() can be called repeatedly when the data is
not available as a single block, such as for scatter/gather lists
and the likes.

Avoid the additional call overhead incured by the refactoring by
moving the implementation off crc32() to sys/systm.h and making it
inlinable. Since crc32_raw() is itself trivial and since it may
be used in loops that iterate over fragments, having it available
for inlining can be beneficial. Hence, move its implementation
to sys/systm.h as well.

Keep the original implementation of crc32() in libkern/crc32.c for
documentation purposes (as a comment of course).

Triggered by: Jose M Rodriguez (josemi at freebsd dot jazztel dot es)
Discussed on: current@
Tested on: amd64, ia64 (BVO having GPT partitions)
Jargon file candidate: BVO = By Virtue Of :-)

19 years agoMake the Z8530 more reliable as low-level console by making use of the
marcel [Wed, 27 Apr 2005 21:57:51 +0000 (21:57 +0000)]
Make the Z8530 more reliable as low-level console by making use of the
fact that access to RR0 does not need a prior write to the register
index because the index always reverts to 0 after the indexed register
has been accessed.

Typically when a RR or WR is to accessed, one programs the index (which
is a write to the control register), followed by a read or write to the
actual indexed register (a read pr write to the same control register).
When this non-atomic sequence is interrupted after having written the
index and low-level console I/O is done in that situation, the write to
program the index will actually write to the indexed register and nuke
state. This almost always yields a wedge.

By not programming the index register and instead just reading from RR0,
the worst case scenario is non-fatal. For if we don't actually read from
RR0 but some other register we get an invalid status, which may lead us
to conclude that the transit data register is empty when it's not or that
the receive data register contains data when it doesn't. Hence, we may
lose an output character or get a sporadic input character, but given
the situation this is a non-issue.

Full serialization is not possible due to the fact that this code needs
to work from DDB and before mutex initialization has happened.

In collaboration with: kris@, marius@
Tested by: kris@
MFC after: 1 day
X-MFC: 5.4-RELEASE candidate

19 years agohide implementation specific internal functions from netdb.h.
ume [Wed, 27 Apr 2005 19:12:57 +0000 (19:12 +0000)]
hide implementation specific internal functions from netdb.h.
it is needed to make get{host,net}by*() thread-safe.

19 years agoAdd code from Kazuhito HONDA that allows the user to see
julian [Wed, 27 Apr 2005 17:16:27 +0000 (17:16 +0000)]
Add code from Kazuhito HONDA that allows the user to see
the available modes in /dev/sndstat.
e.g.
pcm1: <USB Audio> at addr ? (0p/1r/0v channels duplex)
        mode 1:(input) 1ch, 16/16bit, pcm, 44100Hz
mode 2:(input) 1ch, 16/16bit, pcm, 22050Hz
mode 3:(input) 1ch, 16/16bit, pcm, 11025Hz
mode 4:(input) 1ch, 16/16bit, pcm, 8000Hz

19 years agoWhen NFS was converted to the new mount syscall, code was written that sets
des [Wed, 27 Apr 2005 14:46:02 +0000 (14:46 +0000)]
When NFS was converted to the new mount syscall, code was written that sets
the MNT_RDONLY flag if the "ro" option was passed in from userland, and
clears it otherwise.  In the diskless case, the MNT_RDONLY flag is already
set when this code is reached, but there are no mount options, so it was
incorrectly cleared.  Change the logic so the MNT_RDONLY flag is set if the
"ro" option was specified, and left alone otherwise.

Note that the NFS code will still happily let you mount a filesystem RW
even if the server exports it RO.  I'm not sure how to fix that.

19 years agoDisable the IPFILTER bits until they compile again. This can be overriden
scottl [Wed, 27 Apr 2005 14:13:55 +0000 (14:13 +0000)]
Disable the IPFILTER bits until they compile again.  This can be overriden
by adding WANT_IPFILTER to /etc/make.conf.  Note that this is only a partial
hack and only works when building the world and kernel the 'sanctioned' way.
I hope that this hack is only temporary and can be reverted soon.

19 years agoAllocating the memory for the kernel stack one time is enough.
cognet [Wed, 27 Apr 2005 13:29:54 +0000 (13:29 +0000)]
Allocating the memory for the kernel stack one time is enough.

19 years agoFix compilation problem.
davidxu [Wed, 27 Apr 2005 13:17:23 +0000 (13:17 +0000)]
Fix compilation problem.

19 years ago - Don't vgonel() via vgone() or vrecycle() if the vnode is already doomed.
jeff [Wed, 27 Apr 2005 10:03:21 +0000 (10:03 +0000)]
 - Don't vgonel() via vgone() or vrecycle() if the vnode is already doomed.
   This fixes forced unmounts via nullfs.

Reported by: kkenn
Sponsored by: Isilon Systems, Inc.

19 years ago - Changes to vgone() and related teardown code have meant that the vxthread
jeff [Wed, 27 Apr 2005 09:18:10 +0000 (09:18 +0000)]
 - Changes to vgone() and related teardown code have meant that the vxthread
   pointer is no longer needed.

19 years ago - Stop setting vxthread, we've asserted that it was useless for several
jeff [Wed, 27 Apr 2005 09:17:33 +0000 (09:17 +0000)]
 - Stop setting vxthread, we've asserted that it was useless for several
   weeks now.

19 years ago - Stop checking vxthread, we've asserted that it was useless for several
jeff [Wed, 27 Apr 2005 09:17:11 +0000 (09:17 +0000)]
 - Stop checking vxthread, we've asserted that it was useless for several
   weeks.

19 years ago - Fix several locking problems in unionfs_mount so that it will come
jeff [Wed, 27 Apr 2005 09:07:13 +0000 (09:07 +0000)]
 - Fix several locking problems in unionfs_mount so that it will come
   closer to passing DEBUG_VFS_LOCKS.

19 years ago - Pass the ISOPEN flag down to our lower filesystems.
jeff [Wed, 27 Apr 2005 09:06:06 +0000 (09:06 +0000)]
 - Pass the ISOPEN flag down to our lower filesystems.
 - Remove an erroneous VOP lock assert.

19 years ago - Pass the ISOPEN flag to namei so filesystems will know we're about to
jeff [Wed, 27 Apr 2005 09:05:19 +0000 (09:05 +0000)]
 - Pass the ISOPEN flag to namei so filesystems will know we're about to
   open them or otherwise access the data.

19 years ago - Add an ISOPEN flag that filesystems can use to determine if a namei()
jeff [Wed, 27 Apr 2005 09:00:47 +0000 (09:00 +0000)]
 - Add an ISOPEN flag that filesystems can use to determine if a namei()
   caller will be interested in the actual data contents of a vnode after
   a successful lookup.  This intended to help deal with lifetime issues
   for device cloning and to alert autofs when filesystems need to be
   mounted.

19 years agoThe current H.323 proxy source has a license that isn't suitable for inclusion
darrenr [Wed, 27 Apr 2005 05:53:12 +0000 (05:53 +0000)]
The current H.323 proxy source has a license that isn't suitable for inclusion
with FreeBSD so we shouldn't be trying to include it here.

19 years agoForced commit to note the move from "src/sys/hwpmc" to "src/sys/dev/hwpmc".
jkoshy [Wed, 27 Apr 2005 05:51:13 +0000 (05:51 +0000)]
Forced commit to note the move from "src/sys/hwpmc" to "src/sys/dev/hwpmc".

19 years ago- Comment out duplicate rcsid strings in *.c files
darrenr [Wed, 27 Apr 2005 03:48:10 +0000 (03:48 +0000)]
- Comment out duplicate rcsid strings in *.c files
- Move SIOCPROXY from ip_nat.h to ip_proxy.h and fix ip_proxy.h so that it
  can be easily compiled into kdump, et al.

19 years agoRemove COMPAT_32BIT, it is no longer needed.
davidxu [Wed, 27 Apr 2005 01:29:03 +0000 (01:29 +0000)]
Remove COMPAT_32BIT, it is no longer needed.

19 years agoProvide stub functions for i386_set_ldt() and i386_get_ldt() even when
peter [Tue, 26 Apr 2005 20:51:15 +0000 (20:51 +0000)]
Provide stub functions for i386_set_ldt() and i386_get_ldt() even when
compiling as an amd64 support binary.  They will return EINVAL on an amd64
kernel, but this simplifies other #ifdefs that were getting a bit nasty.

19 years agoWhile I'm here, list the new kenv (boot.netif.name) along with the others.
des [Tue, 26 Apr 2005 20:47:59 +0000 (20:47 +0000)]
While I'm here, list the new kenv (boot.netif.name) along with the others.

19 years agoNo longer use _amd64_set_gsbase(). Use i386_set_gsbase() even when
peter [Tue, 26 Apr 2005 20:46:11 +0000 (20:46 +0000)]
No longer use _amd64_set_gsbase().  Use i386_set_gsbase() even when
building for amd64.

19 years agoWhen netbooting, as soon as we've figured out which interface we booted
des [Tue, 26 Apr 2005 20:45:29 +0000 (20:45 +0000)]
When netbooting, as soon as we've figured out which interface we booted
from, store its name in a kenv variable.

19 years agoRemove the special _amd64_set_gsbase() code for #ifdef COMPAT_32BIT, now
peter [Tue, 26 Apr 2005 20:41:48 +0000 (20:41 +0000)]
Remove the special _amd64_set_gsbase() code for #ifdef COMPAT_32BIT, now
that the amd64 kernel implements i386_get/set_gsbase().  All the rest of
the ldt backwards compat code should go away soon.

19 years agoStop calling _amd64_set_gsbase() for COMPAT_32BIT. The amd64 kernel
peter [Tue, 26 Apr 2005 20:38:44 +0000 (20:38 +0000)]
Stop calling _amd64_set_gsbase() for COMPAT_32BIT.  The amd64 kernel
implements i386_set_gsbase(), so there is no need for the variation.

19 years agoFix a the previous commit. I wanted to remove the if and always run the
brooks [Tue, 26 Apr 2005 20:22:31 +0000 (20:22 +0000)]
Fix a the previous commit.  I wanted to remove the if and always run the
body not remove both.

Reported by: ceri
Pointy hat: brooks

19 years agoDon't force IPv6 proto to be printed numericaly.
brooks [Tue, 26 Apr 2005 20:00:07 +0000 (20:00 +0000)]
Don't force IPv6 proto to be printed numericaly.

Noticed by: ceri

19 years agoIntroduce a struct icmphdr which contains the type, code, and cksum
brooks [Tue, 26 Apr 2005 18:10:21 +0000 (18:10 +0000)]
Introduce a struct icmphdr which contains the type, code, and cksum
fields of an ICMP packet.

Use this to allow ipfw to pullup only these values since it does not use
the rest of the packet and it was failed on ICMP packets because they
were not long enough.

struct icmp should probably be modified to use these at some point, but
that will break a fair bit of code so it can wait for another day.

On the off chance that adding this struct breaks something in ports,
bump __FreeBSD_version.

Reported by: Randy Bush <randy at psg dot com>
Tested by: Randy Bush <randy at psg dot com>

19 years agoour get{proto,serv}by*() use a thread-specific data space.
ume [Tue, 26 Apr 2005 18:04:09 +0000 (18:04 +0000)]
our get{proto,serv}by*() use a thread-specific data space.

19 years agoAdd an include for netinet/ip_compat.h directly so that we don't need to
darrenr [Tue, 26 Apr 2005 17:58:05 +0000 (17:58 +0000)]
Add an include for netinet/ip_compat.h directly so that we don't need to
add another special file in the creation of ioctls.c for kdump.

19 years agoDrop the CURPROC, curkse, and curksegrp aliases as they aren't used
jhb [Tue, 26 Apr 2005 17:07:40 +0000 (17:07 +0000)]
Drop the CURPROC, curkse, and curksegrp aliases as they aren't used
anywhere.

19 years agoEnable tcp-wrapper support in the SNMP daemon.
harti [Tue, 26 Apr 2005 16:49:15 +0000 (16:49 +0000)]
Enable tcp-wrapper support in the SNMP daemon.

Submitted by: glebius

19 years agoThis commit was generated by cvs2svn to compensate for changes in r145557,
harti [Tue, 26 Apr 2005 16:43:22 +0000 (16:43 +0000)]
This commit was generated by cvs2svn to compensate for changes in r145557,
which included commits to RCS files with non-trunk default branches.

19 years agoVirgin import of bsnmp 1.9
harti [Tue, 26 Apr 2005 16:43:22 +0000 (16:43 +0000)]
Virgin import of bsnmp 1.9

19 years agoI missed a s/nv/nve/.
obrien [Tue, 26 Apr 2005 16:07:50 +0000 (16:07 +0000)]
I missed a s/nv/nve/.

Submitted by: Tai-hwa Liang <avatar@mmlab.cse.yzu.edu.tw>

19 years agoNot looking for ipfilter source files in the right place
darrenr [Tue, 26 Apr 2005 15:35:50 +0000 (15:35 +0000)]
Not looking for ipfilter source files in the right place

19 years agoFix include directives that were missing the netinet include directory, where
darrenr [Tue, 26 Apr 2005 15:18:45 +0000 (15:18 +0000)]
Fix include directives that were missing the netinet include directory, where
the ipl.h file is found.  Also add back in ip_fil.c, which somehow went missing
in action.

19 years agoUse __func__ throughout devstat.c instead of defining a func_name
mux [Tue, 26 Apr 2005 15:02:01 +0000 (15:02 +0000)]
Use __func__ throughout devstat.c instead of defining a func_name
variable in each function.

19 years agoadd IPv6 awareness for NIS query of gethostby*().
ume [Tue, 26 Apr 2005 14:55:47 +0000 (14:55 +0000)]
add IPv6 awareness for NIS query of gethostby*().

Inspired by: NetBSD

19 years agoMake this compile with GCC4 by fixing a signedness issue.
mux [Tue, 26 Apr 2005 14:51:09 +0000 (14:51 +0000)]
Make this compile with GCC4 by fixing a signedness issue.

Reviewed by: md5(1)

19 years agoFix problems with building libipf:
darrenr [Tue, 26 Apr 2005 14:27:12 +0000 (14:27 +0000)]
Fix problems with building libipf:
ipf_dontuning.c - change the include to look in netinet for ipl.h
ipft_tx.c - make the private use of arrays with tcp flags info in them more
            not use names that can be "confusing"

19 years agoMake this compile with GCC4 by fixing a few signedness related warnings.
mux [Tue, 26 Apr 2005 14:25:11 +0000 (14:25 +0000)]
Make this compile with GCC4 by fixing a few signedness related warnings.

Reviewed by: md5(1)

19 years agoRemove an extra mutex unlock in the morpheus interrupt handler.
scottl [Tue, 26 Apr 2005 13:38:29 +0000 (13:38 +0000)]
Remove an extra mutex unlock in the morpheus interrupt handler.

PR: 80246
Submitted by: Dean Strik
MFC After: 3 days

19 years agoAdd 'curcpu', a shortcut to the current CPU ID, similar to curthread,
rwatson [Tue, 26 Apr 2005 13:15:49 +0000 (13:15 +0000)]
Add 'curcpu', a shortcut to the current CPU ID, similar to curthread,
curproc, et al.  Useful for indexing into per-CPU data structures.

MFC after: 2 weeks

19 years agoFix stripe size in example.
pjd [Tue, 26 Apr 2005 09:23:21 +0000 (09:23 +0000)]
Fix stripe size in example.

Found by: kris
MFC after: 3 days

19 years agoFix a bug introduced in r1.89 thats caused leak of requests, and possibly
sos [Tue, 26 Apr 2005 06:42:33 +0000 (06:42 +0000)]
Fix a bug introduced in r1.89 thats caused leak of requests, and possibly
bogus data to be written.

19 years agoConditionalize the ipfilter header files on NO_IPFILTER. While normally these
scottl [Tue, 26 Apr 2005 02:01:39 +0000 (02:01 +0000)]
Conditionalize the ipfilter header files on NO_IPFILTER.  While normally these
should be harmless, the kdump(1) build does evil things with collecting
system header files, and thus would unconditionally collect and process these.

MFC After: 3 days

19 years agoRemove the extra EEPROM reload step I added before. vge_reset()
wpaul [Mon, 25 Apr 2005 23:26:20 +0000 (23:26 +0000)]
Remove the extra EEPROM reload step I added before. vge_reset()
already does this anyway.

19 years agoApply a torniquet to the problem of the drive unexpectedly disconnecting
scottl [Mon, 25 Apr 2005 22:11:43 +0000 (22:11 +0000)]
Apply a torniquet to the problem of the drive unexpectedly disconnecting
during a data phase.  Before, we would try to recover the autosense, but
the DMA engine would still be active with interrupted transfer, and we'd
quickly spiral out of control and cause massive data corruption.  For now,
just reset the chip and cancel everything.  The better solution is to
cancel the DMA operation, but there is no clear way to do that right now.
The data corruption problem is severe enough to warrant this fix in the
interim.  Thanks to Kris Kenneway to sacrificing countless filesystems to
this bug.

MFC After: 3 days

19 years agoRemove the ACPI_MAX_THREADS option.
scottl [Mon, 25 Apr 2005 19:59:09 +0000 (19:59 +0000)]
Remove the ACPI_MAX_THREADS option.

19 years agoAdd the vm.exec_map_entries tunable and read-only sysctl, which controls
kris [Mon, 25 Apr 2005 19:22:05 +0000 (19:22 +0000)]
Add the vm.exec_map_entries tunable and read-only sysctl, which controls
the number of entries in exec_map (maximum number of simultaneous execs
that can be handled by the kernel).  The default value of 16 is
insufficient on heavily loaded machines (particularly SMP machines), and
if it is exceeded then executing further processes will generate a SIGABRT.

This is a workaround until a better solution can be implemented.

Reviewed by: alc
MFC after: 3 days

19 years agothis file does not belong in a freebsd distribution
darrenr [Mon, 25 Apr 2005 19:15:19 +0000 (19:15 +0000)]
this file does not belong in a freebsd distribution

19 years agoall the ipfilter tools are now built from sbin/Makefile, so remove them from
darrenr [Mon, 25 Apr 2005 19:04:12 +0000 (19:04 +0000)]
all the ipfilter tools are now built from sbin/Makefile, so remove them from
this Makefile.

19 years agoRemove directories no longer used for building ipfilter tools
darrenr [Mon, 25 Apr 2005 19:02:05 +0000 (19:02 +0000)]
Remove directories no longer used for building ipfilter tools

19 years agoremove old directories used to build ipfilter tools
darrenr [Mon, 25 Apr 2005 18:59:29 +0000 (18:59 +0000)]
remove old directories used to build ipfilter tools

19 years agoThere's only one ipfilter directory now
darrenr [Mon, 25 Apr 2005 18:56:42 +0000 (18:56 +0000)]
There's only one ipfilter directory now

19 years agocreate a new build heirarchy for ipfilter tools
darrenr [Mon, 25 Apr 2005 18:55:52 +0000 (18:55 +0000)]
create a new build heirarchy for ipfilter tools

19 years agonew files and compile defines to build new ipfilter
darrenr [Mon, 25 Apr 2005 18:46:00 +0000 (18:46 +0000)]
new files and compile defines to build new ipfilter

19 years agoMerge the changes from 3.4.35 to 4.1.8 into the kernel source tree
darrenr [Mon, 25 Apr 2005 18:43:14 +0000 (18:43 +0000)]
Merge the changes from 3.4.35 to 4.1.8 into the kernel source tree

19 years agoCorrect the if_link_state_change() logic: when the link went down,
wpaul [Mon, 25 Apr 2005 18:37:27 +0000 (18:37 +0000)]
Correct the if_link_state_change() logic: when the link went down,
if_link_state_change() reported link up, and when the link went up,
if_link_state_change() reported link down. These should be swapped.

19 years agoReading the EEPROM to learn the station address doesn't seem to work
wpaul [Mon, 25 Apr 2005 18:29:42 +0000 (18:29 +0000)]
Reading the EEPROM to learn the station address doesn't seem to work
on boards with VIA gigE controllers that are embedded in VIA chipsets.
Presumably, they don't have an external EEPROM and store the MAC
address somewhere else. To get around this, force an autoload and
read the station address from the RX filter registers instead.
This has been tested to work on both embedded and standalone
controllers.

19 years ago* Someone imported a lot of files with the wrong CVS tag, so lots of files need
darrenr [Mon, 25 Apr 2005 18:20:15 +0000 (18:20 +0000)]
* Someone imported a lot of files with the wrong CVS tag, so lots of files need
  that fixed in them....
* Keep unnecessary files out of the non-vendor part of this CVS repository.

19 years agoThis commit was generated by cvs2svn to compensate for changes in r145516,
darrenr [Mon, 25 Apr 2005 18:15:41 +0000 (18:15 +0000)]
This commit was generated by cvs2svn to compensate for changes in r145516,
which included commits to RCS files with non-trunk default branches.

19 years agoimport ipfilter 4.1.8 into the kernel source tree
darrenr [Mon, 25 Apr 2005 18:15:41 +0000 (18:15 +0000)]
import ipfilter 4.1.8 into the kernel source tree

19 years agoAllow BOOT_BOOT0_COMCONSOLE_SPEED to be derived from BOOT_COMCONSOLE_SPEED
ru [Mon, 25 Apr 2005 17:41:35 +0000 (17:41 +0000)]
Allow BOOT_BOOT0_COMCONSOLE_SPEED to be derived from BOOT_COMCONSOLE_SPEED
at compile time.

19 years agoThis commit was generated by cvs2svn to compensate for changes in r145513,
darrenr [Mon, 25 Apr 2005 17:40:37 +0000 (17:40 +0000)]
This commit was generated by cvs2svn to compensate for changes in r145513,
which included commits to RCS files with non-trunk default branches.

19 years agoensure parsing numeric address before any host query.
ume [Mon, 25 Apr 2005 17:36:28 +0000 (17:36 +0000)]
ensure parsing numeric address before any host query.

Inspired by: NetBSD

19 years agoThis commit was generated by cvs2svn to compensate for changes in r145510,
darrenr [Mon, 25 Apr 2005 17:31:50 +0000 (17:31 +0000)]
This commit was generated by cvs2svn to compensate for changes in r145510,
which included commits to RCS files with non-trunk default branches.

19 years agoimport ipfilter 4.1.8 into the vendor branch
darrenr [Mon, 25 Apr 2005 17:31:50 +0000 (17:31 +0000)]
import ipfilter 4.1.8 into the vendor branch

19 years agoNew release note: SA-05:05.cvs.
bmah [Mon, 25 Apr 2005 15:32:58 +0000 (15:32 +0000)]
New release note:  SA-05:05.cvs.

MFCs noted:  restore(8) of FreeBSD 1 images, syslog(8) expanded
hostname specifications.

19 years agoremove unused variable.
ume [Mon, 25 Apr 2005 14:52:13 +0000 (14:52 +0000)]
remove unused variable.

19 years agoMerge the following from the English version:
den [Mon, 25 Apr 2005 12:22:25 +0000 (12:22 +0000)]
Merge the following from the English version:

   1.5 -> 1.7 hardware/common/artheader.sgml
   1.276 -> 1.278 hardware/common/dev.sgml

Update also Russian override of dev-auto.sgml

Obtained from: The FreeBSD Russian Documentation Project

19 years agoFix provider's size check for 'insert' command.
pjd [Mon, 25 Apr 2005 10:41:26 +0000 (10:41 +0000)]
Fix provider's size check for 'insert' command.
Before this fix one was able to insert one sector too small provider.

MFC after: 3 days

19 years agoDeal with failed malloc calls[1].
bz [Mon, 25 Apr 2005 10:18:24 +0000 (10:18 +0000)]
Deal with failed malloc calls[1].
While there also check for failed device_add_child calls.

Found by: Coventry Analysis tool[1].
Submitted by: sam[1]
Approved by: pjd (mentor)
MFC after: 1 week

19 years agoTypeo.
mlaier [Mon, 25 Apr 2005 10:14:26 +0000 (10:14 +0000)]
Typeo.

Submitted by: Emil Mikulic

19 years agoCosmetics
sos [Mon, 25 Apr 2005 07:57:04 +0000 (07:57 +0000)]
Cosmetics