]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years ago/home/green/tmp/cvsSFosXg
green [Tue, 17 Dec 2002 20:39:38 +0000 (20:39 +0000)]
/home/green/tmp/cvsSFosXg

21 years agoThe code uses trapsasync, however the manual page uses asynctraps. Fix the
trhodes [Tue, 17 Dec 2002 20:27:48 +0000 (20:27 +0000)]
The code uses trapsasync, however the manual page uses asynctraps.  Fix the
manual page to reflect the code.

PR: 45820
Submitted by: Marco Molteni <molter@tin.it>
Discussed with: tjr

21 years agoRemove unused lockcnt variable.
phk [Tue, 17 Dec 2002 20:23:51 +0000 (20:23 +0000)]
Remove unused lockcnt variable.

Approved by: mckusick

21 years agoSplit files into 1.44MB chunks rather than a tiny, high-ftp overhead size.
obrien [Tue, 17 Dec 2002 20:19:37 +0000 (20:19 +0000)]
Split files into 1.44MB chunks rather than a tiny, high-ftp overhead size.

Approved by: rwatson

21 years agoStrip out any .comment ELF sections from crunchgen'ed binaries.
obrien [Tue, 17 Dec 2002 20:18:13 +0000 (20:18 +0000)]
Strip out any .comment ELF sections from crunchgen'ed binaries.

21 years agoHold the page queues lock when performing vm_page_flag_set().
alc [Tue, 17 Dec 2002 19:55:28 +0000 (19:55 +0000)]
Hold the page queues lock when performing vm_page_flag_set().

21 years agoAdd IPP (Internet Printing Protocol) and remove blank lines.
eric [Tue, 17 Dec 2002 19:42:07 +0000 (19:42 +0000)]
Add IPP (Internet Printing Protocol) and remove blank lines.

Obtained from: IANA
Sponsored by: Apple

21 years agoStraighten out a compound if() to improve readability marginally.
phk [Tue, 17 Dec 2002 19:33:44 +0000 (19:33 +0000)]
Straighten out a compound if() to improve readability marginally.

21 years agoIndent properly.
phk [Tue, 17 Dec 2002 19:31:26 +0000 (19:31 +0000)]
Indent properly.

21 years agoRemove unused variable cn_devfsdev.
phk [Tue, 17 Dec 2002 19:30:50 +0000 (19:30 +0000)]
Remove unused variable cn_devfsdev.

21 years agoRemove unused and incorrectly maintained variable "in_interfaces"
phk [Tue, 17 Dec 2002 19:30:04 +0000 (19:30 +0000)]
Remove unused and incorrectly maintained variable "in_interfaces"

21 years agoA couple of stylistic improvements.
phk [Tue, 17 Dec 2002 19:16:10 +0000 (19:16 +0000)]
A couple of stylistic improvements.

21 years agoDon't cast a pointer to (intptr_t) and then on to (int) when we cannot
phk [Tue, 17 Dec 2002 19:13:03 +0000 (19:13 +0000)]
Don't cast a pointer to (intptr_t) and then on to (int) when we cannot
be sure that (int) is large enough.  Instead cast only to (intptr_t) and
cast the switch/case values to (intptr_t) as well.

21 years agoDocument `trap EXIT` and `trap SIGNAME`.
fanf [Tue, 17 Dec 2002 17:31:44 +0000 (17:31 +0000)]
Document `trap EXIT` and `trap SIGNAME`.

21 years agoTry to fix the problem with burncd blank not seeing the end of
sos [Tue, 17 Dec 2002 16:27:45 +0000 (16:27 +0000)]
Try to fix the problem with burncd blank not seeing the end of
the blank operation.

21 years agoThis should be the way PC98 ATA disks are seen geometry wise.
sos [Tue, 17 Dec 2002 16:26:22 +0000 (16:26 +0000)]
This should be the way PC98 ATA disks are seen geometry wise.

Should go into 5_0_RELEASE as well.

21 years agoDon't mangle geometry for pc98, this will happen in the ata driver.
phk [Tue, 17 Dec 2002 15:50:51 +0000 (15:50 +0000)]
Don't mangle geometry for pc98, this will happen in the ata driver.

21 years agoUnspam some experimental changes which should not have been committed.
phk [Tue, 17 Dec 2002 13:37:08 +0000 (13:37 +0000)]
Unspam some experimental changes which should not have been committed.

21 years agoRemember to hold topology lock when we change things.
phk [Tue, 17 Dec 2002 09:44:10 +0000 (09:44 +0000)]
Remember to hold topology lock when we change things.

Spotted by: kuriyama

21 years agoConstify the dumpconf() function.
phk [Tue, 17 Dec 2002 07:22:48 +0000 (07:22 +0000)]
Constify the dumpconf() function.

21 years agoAdd PERIPH and XPT debug options to camcontrol. This makes all CAM debug
njl [Tue, 17 Dec 2002 06:05:21 +0000 (06:05 +0000)]
Add PERIPH and XPT debug options to camcontrol. This makes all CAM debug
options tunable from userland.

Approved by: ken
MFC after: 3 days

21 years agoBack out 1.19 to rethink approach
njl [Tue, 17 Dec 2002 05:55:36 +0000 (05:55 +0000)]
Back out 1.19 to rethink approach

Requested by: julian@

21 years agoCheck the return status of chdir() when using the -C option.
seanc [Tue, 17 Dec 2002 04:26:22 +0000 (04:26 +0000)]
Check the return status of chdir() when using the -C option.

Reviewed by: bright && jmallet
MFC after: 1 day

21 years agoCheck that the dump device is large enough. Otherwise we could
marcel [Tue, 17 Dec 2002 02:51:56 +0000 (02:51 +0000)]
Check that the dump device is large enough. Otherwise we could
end up with a dump offset that's smaller than the start of the
dump device and either clobber data in preceding partitions or
try to write beyond the end of the medium (unsigned wrap).

Implement legacy behaviour to never write to the first 64KB as
that is where metadata (ie disklabels) may reside.

21 years agoAutomatically issue a "continue" along with the "detach" command. This
njl [Tue, 17 Dec 2002 01:32:11 +0000 (01:32 +0000)]
Automatically issue a "continue" along with the "detach" command.  This
fixes the problem of cleanly restarting a target after entering gdb mode.

Reviewed by: archie@

21 years agoFix syntax in last commit.
dillon [Tue, 17 Dec 2002 00:24:48 +0000 (00:24 +0000)]
Fix syntax in last commit.

21 years agoConvert the xl(4) driver to the busdma API. This should make
mux [Tue, 17 Dec 2002 00:08:49 +0000 (00:08 +0000)]
Convert the xl(4) driver to the busdma API.  This should make
it possible to use this driver under ia64, sparc64 (though
there may be endianness issues with this one) and other archs.

Tested on: i386, alpha (gallatin)

21 years agoReformat last change
julian [Mon, 16 Dec 2002 23:25:12 +0000 (23:25 +0000)]
Reformat last change
Requested by: nate@

21 years agoGet rid of g_slice_addslice() and use g_slice_config() instead.
phk [Mon, 16 Dec 2002 23:08:48 +0000 (23:08 +0000)]
Get rid of g_slice_addslice() and use g_slice_config() instead.

Tested with: i386 + src/tools/regression/geom

21 years agoDon't dump core into a partition that is too small for it.
julian [Mon, 16 Dec 2002 23:04:54 +0000 (23:04 +0000)]
Don't dump core into a partition that is too small for it.
If we do, we usually wrote backwareds into the proceeding partititon
which is usually the root partition.

21 years agoGoto WARNS 5 with assorted changes.
phk [Mon, 16 Dec 2002 22:35:56 +0000 (22:35 +0000)]
Goto WARNS 5 with assorted changes.

21 years agoConstification and some s/int/u_int/ changes.
phk [Mon, 16 Dec 2002 22:33:27 +0000 (22:33 +0000)]
Constification and some s/int/u_int/ changes.

21 years agoChange the name for the local unix-socket based protocol
mbr [Mon, 16 Dec 2002 22:24:26 +0000 (22:24 +0000)]
Change the name for the local unix-socket based protocol
from "unix" back to "local".  Add some compat stuff so both
ways work for some time.

Reviewed by:    phk
Approved by:    imp (UPDATING)
Requested by:   iedowse, lukem@netbsd.org

21 years agoUpdate this one too.
phk [Mon, 16 Dec 2002 22:01:25 +0000 (22:01 +0000)]
Update this one too.

21 years agoGet the test-suite into sync with the kernel again.
phk [Mon, 16 Dec 2002 21:39:16 +0000 (21:39 +0000)]
Get the test-suite into sync with the kernel again.

21 years agoSpelling: s/ingrity/integrity/
trhodes [Mon, 16 Dec 2002 21:13:25 +0000 (21:13 +0000)]
Spelling: s/ingrity/integrity/

21 years agoChange the way ELF coredumps are handled. Instead of unconditionally
dillon [Mon, 16 Dec 2002 19:24:43 +0000 (19:24 +0000)]
Change the way ELF coredumps are handled.  Instead of unconditionally
skipping read-only pages, which can result in valuable non-text-related
data not getting dumped, the ELF loader and the dynamic loader now mark
read-only text pages NOCORE and the coredump code only checks (primarily) for
complete inaccessibility of the page or NOCORE being set.

Certain applications which map large amounts of read-only data will
produce much larger cores.  A new sysctl has been added,
debug.elf_legacy_coredump, which will revert to the old behavior.

This commit represents collaborative work by all parties involved.
The PR contains a program demonstrating the problem.

PR: kern/45994
Submitted by: "Peter Edwards" <pmedwards@eircom.net>, Archie Cobbs <archie@dellroad.org>
Reviewed by: jdp, dillon
MFC after: 7 days

21 years agoAdd the trm(4) driver.
cognet [Mon, 16 Dec 2002 18:47:37 +0000 (18:47 +0000)]
Add the trm(4) driver.

MFC after: 1 day

21 years agoImport of a minimalist trm(4) man page.
cognet [Mon, 16 Dec 2002 18:44:24 +0000 (18:44 +0000)]
Import of a minimalist trm(4) man page.
Anyone with more man pages knowledge than I have should feel free to complete
it.

21 years agoThe HP DAT 40 tape drive should be able to handle variable block sizes.
trhodes [Mon, 16 Dec 2002 17:40:17 +0000 (17:40 +0000)]
The HP DAT 40 tape drive should be able to handle variable block sizes.
But for some reason the block size is different when a different type of
tape is placed in the drive.  This commit fixes that.

PR: 46209
Submitted by: Alex Wang <alex@alexwang.com>
Approved by: mjacob

21 years agoRemove the hto(be|le)[slq] and (be|le)toh[slq] macros defined in
robert [Mon, 16 Dec 2002 16:20:06 +0000 (16:20 +0000)]
Remove the hto(be|le)[slq] and (be|le)toh[slq] macros defined in
_KERNEL scope from "src/sys/sys/mchain.h".

Replace each occurrence of the above in _KERNEL scope with the
appropriate macro from the set of hto(be|le)(16|32|64) and
(be|le)toh(16|32|64) from "src/sys/sys/endian.h".

Tested by: tjr
Requested by: comment marked with XXX

21 years agoThis commit was generated by cvs2svn to compensate for changes in r107937,
des [Mon, 16 Dec 2002 15:28:05 +0000 (15:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r107937,
which included commits to RCS files with non-trunk default branches.

21 years agoVendor import of OpenPAM Cyclamen.
des [Mon, 16 Dec 2002 15:28:05 +0000 (15:28 +0000)]
Vendor import of OpenPAM Cyclamen.

21 years agoMerge in most non-style differences from Andrew Korty's pam_ssh 1.7.
des [Mon, 16 Dec 2002 14:33:18 +0000 (14:33 +0000)]
Merge in most non-style differences from Andrew Korty's pam_ssh 1.7.

21 years agoIgnore IEEE1284 descriptors when looking for bidirectional mode. We don't
des [Mon, 16 Dec 2002 14:22:59 +0000 (14:22 +0000)]
Ignore IEEE1284 descriptors when looking for bidirectional mode.  We don't
really know how to talk IEEE1284, so attaching to that interface makes the
printer unusable.

Approved by: joe

21 years agoMFNetBSD: 1.52-1.55
des [Mon, 16 Dec 2002 14:20:26 +0000 (14:20 +0000)]
MFNetBSD: 1.52-1.55

Mostly OpenBSD-related changes.  Remove newline from panic string.  Remove
an unreachable break statement.

21 years agoCatch up with the English version:
hrs [Mon, 16 Dec 2002 14:11:42 +0000 (14:11 +0000)]
Catch up with the English version:

1.115 -> 1.119 hardware/common/dev.sgml

21 years agoo getsockopt(2) 'level' argument should be 0, not SOCK_STREAM. It
maxim [Mon, 16 Dec 2002 13:42:13 +0000 (13:42 +0000)]
o getsockopt(2) 'level' argument should be 0, not SOCK_STREAM. It
does not hurt anything because uipc_ctloutput() does not check
sopt->sopt_level.

Pointed out by: ru
MFC after: 1 week

21 years agoFix the panic that occurs when kldunloading trm by using the correct
cognet [Mon, 16 Dec 2002 13:38:22 +0000 (13:38 +0000)]
Fix the panic that occurs when kldunloading trm by using the correct
prototype for trm_detach and freeing all resources.
While I'm there, handle better errors in trm_attach and remove the
PCI_BASE_ADDR0 definition, since it's what PCIR_MAPS is used for.

MFC after: 3 days

21 years agofixed a bug that IPv6 multicast packet is not forwarded if its packet size is equal...
suz [Mon, 16 Dec 2002 01:41:07 +0000 (01:41 +0000)]
fixed a bug that IPv6 multicast packet is not forwarded if its packet size is equal to the outgoing interface's MTU

Approved by: re
Obtained from: KAME
MFC after: 3 days

21 years agoTypo: devices numbers -> device numbers.
keramida [Mon, 16 Dec 2002 00:52:31 +0000 (00:52 +0000)]
Typo: devices numbers -> device numbers.

PR: docs/46279
Submitted by: Rich Morin <rdm@cfcl.com>
MFC after: 3 days

21 years agoRegen: swapoff
marcel [Mon, 16 Dec 2002 00:49:36 +0000 (00:49 +0000)]
Regen: swapoff

21 years agoChange swapoff from MNOPROTO to UNIMPL. The former doesn't work.
marcel [Mon, 16 Dec 2002 00:48:52 +0000 (00:48 +0000)]
Change swapoff from MNOPROTO to UNIMPL. The former doesn't work.

21 years agounifdef -DPC98 -U__alpha__
phk [Sun, 15 Dec 2002 22:15:20 +0000 (22:15 +0000)]
unifdef -DPC98 -U__alpha__

21 years agoBring in the settings from /etc/rc.devfs, which include the vga symlink
dougb [Sun, 15 Dec 2002 21:56:53 +0000 (21:56 +0000)]
Bring in the settings from /etc/rc.devfs, which include the vga symlink
that's being replaced here. This is what /etc/rc does now.

There is no NetBSD devfs, so there is no compatibility issue.

Not objected to by: FreeBSD-rc@yahoogroups.com
Approved by: re (murray)

21 years agoPerform vm_object_lock() and vm_object_unlock() on kmem_object
alc [Sun, 15 Dec 2002 21:09:09 +0000 (21:09 +0000)]
Perform vm_object_lock() and vm_object_unlock() on kmem_object
around vm_page_lookup() and vm_page_free().

21 years agoAdd a new `W' flag, that when used in conjunction with the `Z' flag or the
sobomax [Sun, 15 Dec 2002 19:28:02 +0000 (19:28 +0000)]
Add a new `W' flag, that when used in conjunction with the `Z' flag or the
`J' flag means that newsyslog should wait for previously started compression
jobs to complete before starting a new one for this entry.  When it is used
along with the `G' flag, in the case when several logfiles match the pattern
and should be compressed, the newsyslog will compress logs one by one,
ensuring that only one compression job is running at a time.

This prevents newsyslog(8) from overloading system by starting several
compression jobs on big files simultaneously.

Sponsored by: Porta Software Ltd
MFC after: 2 weeks

21 years agoUpdate to previous change (1.54) to use an approperly wide inode field
mckusick [Sun, 15 Dec 2002 19:25:59 +0000 (19:25 +0000)]
Update to previous change (1.54) to use an approperly wide inode field
so as to work correctly on 64-bit platforms.

Reported-by: Jake Burkholder <jake@locore.ca>
Sponsored by:   DARPA & NAI Labs.
Approved by: Ian Dowse <iedowse@maths.tcd.ie>

21 years agoRegenerate system calls (swapoff added)
dillon [Sun, 15 Dec 2002 19:19:15 +0000 (19:19 +0000)]
Regenerate system calls (swapoff added)

21 years agoThis is David Schultz's swapoff code which I am finally able to commit.
dillon [Sun, 15 Dec 2002 19:17:57 +0000 (19:17 +0000)]
This is David Schultz's swapoff code which I am finally able to commit.
This should be considered highly experimental for the moment.

Submitted by: David Schultz <dschultz@uclink.Berkeley.EDU>
MFC after: 3 weeks

21 years agoFix a refcount race with the vmspace structure. In order to prevent
dillon [Sun, 15 Dec 2002 18:50:04 +0000 (18:50 +0000)]
Fix a refcount race with the vmspace structure.  In order to prevent
resource starvation we clean-up as much of the vmspace structure as we
can when the last process using it exits.  The rest of the structure
is cleaned up when it is reaped.  But since exit1() decrements the ref
count it is possible for a double-free to occur if someone else, such as
the process swapout code, references and then dereferences the structure.
Additionally, the final cleanup of the structure should not occur until
the last process referencing it is reaped.

This commit solves the problem by introducing a secondary reference count,
calling 'vm_exitingcnt'.  The normal reference count is decremented on exit
and vm_exitingcnt is incremented.  vm_exitingcnt is decremented when the
process is reaped.  When both vm_exitingcnt and vm_refcnt are 0, the
structure is freed for real.

MFC after: 3 weeks

21 years agoPC164 is 21172
wilko [Sun, 15 Dec 2002 14:18:19 +0000 (14:18 +0000)]
PC164 is 21172

Approved by: re (bmah)

21 years agoCorrect chipset information for the 164* series.
wilko [Sun, 15 Dec 2002 14:01:07 +0000 (14:01 +0000)]
Correct chipset information for the 164* series.

Submitted by: Christian Weisgerber <naddy@mips.inka.de>
Approved by: re (bmah)

21 years agoUpdate ROOT_MIN_SIZE for i386 to 118MB (and other ROOT_*_SIZE).
kuriyama [Sun, 15 Dec 2002 12:05:00 +0000 (12:05 +0000)]
Update ROOT_MIN_SIZE for i386 to 118MB (and other ROOT_*_SIZE).

21 years agoWhen doing tail -F, return if the freopen() of the file fails instead of
tjr [Sun, 15 Dec 2002 11:26:57 +0000 (11:26 +0000)]
When doing tail -F, return if the freopen() of the file fails instead of
trying to use a null file pointer.

21 years agoo Trim EOL whitespaces.
maxim [Sun, 15 Dec 2002 10:24:36 +0000 (10:24 +0000)]
o Trim EOL whitespaces.

MFC after: 1 week

21 years agoo s/if_name[16]/if_name[IFNAMSIZ]/
maxim [Sun, 15 Dec 2002 10:23:02 +0000 (10:23 +0000)]
o s/if_name[16]/if_name[IFNAMSIZ]/

Reviewed by: luigi
MFC after: 1 week

21 years agoo M_DONTWAIT is mbuf(9) flag: malloc(M_DONTWAIT) -> malloc(M_NOWAIT).
maxim [Sun, 15 Dec 2002 10:21:30 +0000 (10:21 +0000)]
o M_DONTWAIT is mbuf(9) flag: malloc(M_DONTWAIT) -> malloc(M_NOWAIT).
The bug does not affect anything because M_NOWAIT == M_DONTWAIT.

Reviewed by: luigi
MFC after: 1 week

21 years agoo Fix byte order logging issue: sa.sin_port is already in host byte order.
maxim [Sun, 15 Dec 2002 09:44:02 +0000 (09:44 +0000)]
o Fix byte order logging issue: sa.sin_port is already in host byte order.

PR: kern/45964
Submitted by: Sascha Blank <sblank@tiscali.de>
Reviewed by: luigi
MFC after: 1 week

21 years agoo Clear a high bit of ipc_perm.seq so msgget(3) never returns a
maxim [Sun, 15 Dec 2002 09:41:46 +0000 (09:41 +0000)]
o Clear a high bit of ipc_perm.seq so msgget(3) never returns a
negative message queue id.

PR: kern/46122
Submitted by: Vladimir B.Grebenschikov <vova@sw.ru>
MFC after: 2 weeks

21 years agoo Fix bogus getsockopt(2) call: swap SOCK_STREAM and LOCAL_PEERCRED.
maxim [Sun, 15 Dec 2002 09:37:54 +0000 (09:37 +0000)]
o Fix bogus getsockopt(2) call: swap SOCK_STREAM and LOCAL_PEERCRED.
The bug does not affect anything because SOCK_STREAM == LOCAL_PEERCRED == 0x1.

PR: bin/46165
Submitted by: Alain Thivillon <at@rominet.net>
Reviewed by: dd
MFC after: 1 week

21 years agoAs per the comments, vm_object_page_remove() now expects its caller to lock
alc [Sun, 15 Dec 2002 07:30:51 +0000 (07:30 +0000)]
As per the comments, vm_object_page_remove() now expects its caller to lock
the object (i.e., acquire Giant).

21 years agoPerform vm_object_lock() and vm_object_unlock() around
alc [Sun, 15 Dec 2002 07:16:51 +0000 (07:16 +0000)]
Perform vm_object_lock() and vm_object_unlock() around
vm_object_page_remove().

21 years agoPerform vm_object_lock() and vm_object_unlock() around
alc [Sun, 15 Dec 2002 05:41:56 +0000 (05:41 +0000)]
Perform vm_object_lock() and vm_object_unlock() around
vm_object_page_remove().

21 years agoRemove redundant check for negative or zero v_usecount; vrele() already
tjr [Sun, 15 Dec 2002 04:32:41 +0000 (04:32 +0000)]
Remove redundant check for negative or zero v_usecount; vrele() already
checks that.

21 years agoRemove unneeded casts. Add others to make WARNS=5 happy.
obrien [Sun, 15 Dec 2002 02:15:19 +0000 (02:15 +0000)]
Remove unneeded casts.  Add others to make WARNS=5 happy.

21 years agoUnbreak support for systems that have no limit on pathname length.
tjr [Sun, 15 Dec 2002 00:40:47 +0000 (00:40 +0000)]
Unbreak support for systems that have no limit on pathname length.

Noticed by: bde

21 years agoAssert that the page queues lock is held in vm_page_unhold(),
alc [Sun, 15 Dec 2002 00:06:02 +0000 (00:06 +0000)]
Assert that the page queues lock is held in vm_page_unhold(),
vm_page_remove(), and vm_page_free_toq().

21 years agoMake minimal changes to point the users to the Texinfo documentation
keramida [Sat, 14 Dec 2002 23:58:49 +0000 (23:58 +0000)]
Make minimal changes to point the users to the Texinfo documentation
for more detailed information about sdiff(1).  Hopefully this will
make it easier to find how `interactive merging' works with sdiff.

PR: docs/30618
Submitted by: mark@summersault.com

21 years agoNetBSD manpages use .Nm "" to typeset the `name' of the manpage, but
keramida [Sat, 14 Dec 2002 23:45:54 +0000 (23:45 +0000)]
NetBSD manpages use .Nm "" to typeset the `name' of the manpage, but
this doesn't work quite right with our groff.  They have an older
groff version in the base system, but no real plans to update both
groff and manpages.  Fix the .Nm usage locally in our manpages.

PR: docs/41972
Submitted by: Yoshiteru Kageyama <yt-kage@cb3.so-net.ne.jp>
Discussed with: ru and obrien

21 years agoBuild kernels for pc98 just like the other platforms. Also, remove
imp [Sat, 14 Dec 2002 22:36:56 +0000 (22:36 +0000)]
Build kernels for pc98 just like the other platforms.  Also, remove
one space to make the dates line up for easy mental math.

# Too bad only i386 and pc98 are the only parts of the universe
# currently buildable on my box :-(

21 years agoMFi386 revisions 1.375 and 1.376
alc [Sat, 14 Dec 2002 22:25:35 +0000 (22:25 +0000)]
MFi386 revisions 1.375 and 1.376
 Hold the page queues lock when calling pmap_unwire_pte_hold() or
 pmap_remove_pte().  Use vm_page_sleep_if_busy() in
 _pmap_unwire_pte_hold() so that the page queues lock is released
 when sleeping.

21 years agoChange tcp.inflight_min from 1024 to a production default of 6144. Create
dillon [Sat, 14 Dec 2002 21:00:17 +0000 (21:00 +0000)]
Change tcp.inflight_min from 1024 to a production default of 6144.  Create
a sysctl for the stabilization value for the bandwidth delay product (inflight)
algorithm and document it.

MFC after: 3 days

21 years agoEmploy the unused bytes after the disklabel in the second sector. This makes
phk [Sat, 14 Dec 2002 19:44:13 +0000 (19:44 +0000)]
Employ the unused bytes after the disklabel in the second sector.  This makes
it possible to make UFS1_ONLY and UFS2_ONLY versions which fit inside the
traditional 16 sectors.

Remove assorted now unneeded hackery.

UFS1_AND_UFS2 still needs another 150 bytes to work, and that is probably
not within our reach, ever.

21 years agoRemove unnecessary call to fsread().
phk [Sat, 14 Dec 2002 19:40:35 +0000 (19:40 +0000)]
Remove unnecessary call to fsread().

21 years agoRestructure so we can compile UFS1_ONLY, UFS2_ONLY or UFS1_AND_UFS2
phk [Sat, 14 Dec 2002 19:39:44 +0000 (19:39 +0000)]
Restructure so we can compile UFS1_ONLY, UFS2_ONLY or UFS1_AND_UFS2
versions from the same basic function.

21 years agoConvert the use of vtophys() for doing autosense to use busdma. Also correct
scottl [Sat, 14 Dec 2002 19:20:57 +0000 (19:20 +0000)]
Convert the use of vtophys() for doing autosense to use busdma.  Also correct
some error codes that get returned to CAM.

21 years agoAlways use the smaller GCC builtin memcpy
phk [Sat, 14 Dec 2002 19:15:27 +0000 (19:15 +0000)]
Always use the smaller GCC builtin memcpy

21 years agoRemove unused variable.
phk [Sat, 14 Dec 2002 19:09:37 +0000 (19:09 +0000)]
Remove unused variable.

21 years agoAdd a list of currently supported modes && an EXAMPLES section.
trhodes [Sat, 14 Dec 2002 18:38:34 +0000 (18:38 +0000)]
Add a list of currently supported modes && an EXAMPLES section.

PR: 40465
Submitted by: Cyrille Lefevre <cyrille.lefevre@laposte.net>

21 years agoAdd prototype for sockatmark().
fenner [Sat, 14 Dec 2002 18:14:34 +0000 (18:14 +0000)]
Add prototype for sockatmark().

21 years agoMarkup fixes. Document more correctly.
trhodes [Sat, 14 Dec 2002 18:07:13 +0000 (18:07 +0000)]
Markup fixes.  Document more correctly.

PR: 43678
Submitted by: ru

21 years agoDon't fill in the table with the BIOS idea about disk-geometry, we don't
phk [Sat, 14 Dec 2002 17:34:28 +0000 (17:34 +0000)]
Don't fill in the table with the BIOS idea about disk-geometry, we don't
use it.  This saves a surprising number of bytes.

21 years agoUndo the adjustment of the total memory used by dirhash in the case
iedowse [Sat, 14 Dec 2002 17:16:16 +0000 (17:16 +0000)]
Undo the adjustment of the total memory used by dirhash in the case
where allocating the dirhash structure fails. Fix a few typos in
comments and update copyright.

MFC after: 1 week

21 years agoOnly dump the BIOS geometry table from bootinfo on PC98, we don't use
phk [Sat, 14 Dec 2002 16:11:28 +0000 (16:11 +0000)]
Only dump the BIOS geometry table from bootinfo on PC98, we don't use
the contents on i386 anymore.

21 years agoDocument NO_CPU_CFLAGS && NO_CPU_COPTFLAGS in the make.conf(5) manual
trhodes [Sat, 14 Dec 2002 15:53:32 +0000 (15:53 +0000)]
Document NO_CPU_CFLAGS && NO_CPU_COPTFLAGS in the make.conf(5) manual
page.

PR: 45418
Reviewed by: jhb (a slightly older version)

21 years agomdoc(7) police: markup nits + fixed two misspellings of drivers.flp.
ru [Sat, 14 Dec 2002 15:46:50 +0000 (15:46 +0000)]
mdoc(7) police: markup nits + fixed two misspellings of drivers.flp.

21 years agomdoc(7) police: markup overhaul.
ru [Sat, 14 Dec 2002 15:29:27 +0000 (15:29 +0000)]
mdoc(7) police: markup overhaul.

21 years agoDon't build auth-pam.c and auth2-pam.c, auth2-pam-freebsd.c is all we need.
des [Sat, 14 Dec 2002 13:54:57 +0000 (13:54 +0000)]
Don't build auth-pam.c and auth2-pam.c, auth2-pam-freebsd.c is all we need.
Use pthreads for PAM if the platform supports it and the user asked for it
(by setting OPENSSH_USE_POSIX_THREADS)

Sponsored by: DARPA, NAI Labs

21 years agoIf possible, use pthreads instead of a child process for PAM.
des [Sat, 14 Dec 2002 13:52:39 +0000 (13:52 +0000)]
If possible, use pthreads instead of a child process for PAM.

Reimplement the necessary bits from auth_pam.c and auth2_pam.c so that
they share the PAM context used by the keyboard-interactive thread.  If
a child process is used instead, they will (necessarily) use a separate
context.

Constify do_pam_account() and do_pam_session().

Sponsored by: DARPA, NAI Labs