]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoMove the gallant 12 x 22 font data from a .h to a .c so it doesn't need
marius [Sat, 16 Jun 2007 21:48:50 +0000 (21:48 +0000)]
Move the gallant 12 x 22 font data from a .h to a .c so it doesn't need
to be compiled into every driver making use of it. Use a const instance
of struct gfb_font for this as the font isn't intended to be changed at
run-time and in order to accompany the font data with height and width
info.

17 years ago- Remove dupe and unused declarations and prototypes.
marius [Sat, 16 Jun 2007 21:46:15 +0000 (21:46 +0000)]
- Remove dupe and unused declarations and prototypes.
- Add missing prototypes.
- Define global variables not used outside of this module as static.
- Replace some outdated hard-coded functions names in panic strings
  with __func__.
- Fix some style(9) bugs.

17 years agoSync the styles of sys/boot/ofw/libofw/openfirm.c and
marius [Sat, 16 Jun 2007 21:38:04 +0000 (21:38 +0000)]
Sync the styles of sys/boot/ofw/libofw/openfirm.c and
sys/dev/ofw/openfirm.c with themselves, with each-other
and with style(9).

17 years ago- Define data of struct gfb_font a const as it's only used to supply font
marius [Sat, 16 Jun 2007 21:31:53 +0000 (21:31 +0000)]
- Define data of struct gfb_font a const as it's only used to supply font
  data and remove the array size from the definition as f.e. the gallant
  12 x 22 font data is 256 * 44 in size, exceeding the previously hard-
  coded size.
- Declare the bold8x16 instance of struct gfb_font as const as it's not
  intended to be changed at run-time as a whole either.
- Use __FBSDID in xboxfb.c

Tested by: rink

17 years agoIf attempting to cache a "busy", panic instead of printing a diagnostic
alc [Sat, 16 Jun 2007 21:07:51 +0000 (21:07 +0000)]
If attempting to cache a "busy", panic instead of printing a diagnostic
message and returning.

17 years agoForced commit to clarify preceding unclear commit message:
wilko [Sat, 16 Jun 2007 20:49:49 +0000 (20:49 +0000)]
Forced commit to clarify preceding unclear commit message:
- core was contacted by Joerg Peter Schley requesting 2 comment lines
  be removed with his name & email address in it.

17 years agoUse signed int, since all the samples are in that form.
ariff [Sat, 16 Jun 2007 20:36:39 +0000 (20:36 +0000)]
Use signed int, since all the samples are in that form.
This will help future implementation of soft amplifier
with variable/relative 0db.

17 years agoo Do not warn if the process exits before we get its name.
maxim [Sat, 16 Jun 2007 20:24:55 +0000 (20:24 +0000)]
o Do not warn if the process exits before we get its name.

PR: bin/113777
Submitted by: Dmitrij Tejblum
MFC after: 1 week

17 years agoPer request (to core) by the person refered to in the comment.
wilko [Sat, 16 Jun 2007 19:23:14 +0000 (19:23 +0000)]
Per request (to core) by the person refered to in the comment.

17 years agoRevert an accidental change from the previous rev.
scottl [Sat, 16 Jun 2007 18:22:19 +0000 (18:22 +0000)]
Revert an accidental change from the previous rev.

17 years agoWork around the other uses of M_WAITOK.
scottl [Sat, 16 Jun 2007 18:20:29 +0000 (18:20 +0000)]
Work around the other uses of M_WAITOK.

17 years agoWork around a malloc locking problem.
scottl [Sat, 16 Jun 2007 18:13:26 +0000 (18:13 +0000)]
Work around a malloc locking problem.

17 years agoAdd xfs(5) man page.
rodrigc [Sat, 16 Jun 2007 16:50:53 +0000 (16:50 +0000)]
Add xfs(5) man page.

17 years agoThe iop34x has 128 interrupts.
cognet [Sat, 16 Jun 2007 15:03:33 +0000 (15:03 +0000)]
The iop34x has 128 interrupts.

17 years ago- Bump ncurses share library version for the coming RELENG_7
rafan [Sat, 16 Jun 2007 14:42:08 +0000 (14:42 +0000)]
- Bump ncurses share library version for the coming RELENG_7

Pointed out by:      jhb
Approved by:      delphij (mentor)

17 years agoNever define the generated rpc server main function with storage type
se [Sat, 16 Jun 2007 14:39:25 +0000 (14:39 +0000)]
Never define the generated rpc server main function with storage type
"static". The header file generated by "rpcgen -h" will always declare
it as extern, leading to a "static after extern" error being issued by
gcc-4.2. This caused only a warning in gcc-3.x, but it has been wrong
all the time.

This change does not modify the code generated by "rpcgen -m", it only
affects rpcgen used to generate server stubs with a local main function.

This is the minimal patch. It does not remove the now obsolete "storage"
parameter from write_program() and write_programs() in an attempt to keep
differences to other systems' versions of rpcgen as small as possible.

17 years ago- Better handle sending large pkt-drops. We were not triming
rrs [Sat, 16 Jun 2007 14:03:15 +0000 (14:03 +0000)]
- Better handle sending large pkt-drops. We were not triming
  the data with m_adj if a large pkt arrived with a bad csum
  some systems can't handle you not triming the tail (think panda :-D)

17 years ago'spi' and the return value of ntohl are unsigned. Remove the extra >=0
bz [Sat, 16 Jun 2007 09:25:23 +0000 (09:25 +0000)]
'spi' and the return value of ntohl are unsigned. Remove the extra >=0
check which was always true.
Document the special meaning of spi values of 0 and 1-255 with a comment.

Found with: Coverity Prevent(tm)
CID: 2047

17 years agoo Add an example how to create /etc/mtree style mtree(8) files.
maxim [Sat, 16 Jun 2007 08:26:00 +0000 (08:26 +0000)]
o Add an example how to create /etc/mtree style mtree(8) files.

PR: docs/113667
Submitted by: edwin
MFC after: 1 week

17 years agoDocument default values for timeout and retry count.
danfe [Sat, 16 Jun 2007 07:01:15 +0000 (07:01 +0000)]
Document default values for timeout and retry count.

Approved by: fjoe

17 years agoCheck for a NULL return from rpcclnt_buildheader- it can fail if
mjacob [Sat, 16 Jun 2007 05:42:26 +0000 (05:42 +0000)]
Check for a NULL return from rpcclnt_buildheader- it can fail if
the passed in auth_type is unacceptable to rpcauth_buildheader-
this avoids a null pointer panic. Clean up allocations if this
happens. This also quiets a gcc 4.2 complaint about ussing mheadend
without it being initialized.

Reviewed by: alfred

17 years agoUpdate a comment.
alc [Sat, 16 Jun 2007 05:25:53 +0000 (05:25 +0000)]
Update a comment.

17 years agoCorrect state machine handling of AUTH -> AUTH transitions that pass
sam [Sat, 16 Jun 2007 05:13:48 +0000 (05:13 +0000)]
Correct state machine handling of AUTH -> AUTH transitions that pass
through wpa_supplcant.  If a sta is deauth'd (e.g. due to inactivity)
with roaming mode set to manual then a subsequent MLME assoc request
will be incorrectly handled and the station will never reauthenticate.
To fix this interpret a reason code of zero as sufficient to send an
auth request frame.

17 years agoEnable the new physical memory allocator.
alc [Sat, 16 Jun 2007 04:57:06 +0000 (04:57 +0000)]
Enable the new physical memory allocator.

This allocator uses a binary buddy system with a twist.  First and
foremost, this allocator is required to support the implementation of
superpages.  As a side effect, it enables a more robust implementation
of contigmalloc(9).  Moreover, this reimplementation of
contigmalloc(9) eliminates the acquisition of Giant by
contigmalloc(..., M_NOWAIT, ...).

The twist is that this allocator tries to reduce the number of TLB
misses incurred by accesses through a direct map to small, UMA-managed
objects and page table pages.  Roughly speaking, the physical pages
that are allocated for such purposes are clustered together in the
physical address space.  The performance benefits vary.  In the most
extreme case, a uniprocessor kernel running on an Opteron, I measured
an 18% reduction in system time during a buildworld.

This allocator does not implement page coloring.  The reason is that
superpages have much the same effect.  The contiguous physical memory
allocation necessary for a superpage is inherently colored.

Finally, the one caveat is that this allocator does not effectively
support prezeroed pages.  I hope this is temporary.  On i386, this is
a slight pessimization.  However, on amd64, the beneficial effects of
the direct-map optimization outweigh the ill effects.  I speculate
that this is true in general of machines with a direct map.

Approved by: re

17 years agoLast (again ?!?) major commit for RELENG_7, featuring total Giant
ariff [Sat, 16 Jun 2007 03:37:28 +0000 (03:37 +0000)]
Last (again ?!?) major commit for RELENG_7, featuring total Giant
eradication in/from userland path, countless locking fixes, etc.

- General sleep call through msleep(9) has been converted to condvar(9)
  with better consistencies.
- Heavily guard every possible "slow path" entries (open(), close(),
  few ioctl()s, sysctls), but once it entering "fast path" (io, interrupt
  started), they are free to fly on their own.
- Rearrange locking sequences, resulting better concurrency and
  serialization. Large part doesn't even need locking at all, and will be
  removed in future. Less clutter, except in few places due to lock
  ordering.
- Anonymous mixer object creation/deletion to simplify mixer handling
  beyond typical mixer ioctls.
  Submitted by: chibis (with modifications)
- Add few mix_[get|set|..] functions to avoid calling mixer_ioctl()
  directly using cryptic arguments.
- Locking fixes to avoid possible deadlock with (still under Giant) USB.
- Better simplex/duplex device handling.
- Recover mmap() functionality for recording, which has been lost
  since 2.2.x - 3.x (the introduction of newpcm). Full-duplex mmap still
  doesn't work (due to VM/page design), but people still can mmap
  both by opening each direction separately. mmaped playback is guarantee
  to work either way.
- New sysctl: "hw.snd.compat_linux_mmap" to allow PROT_EXEC page
  mapping, due to recent changes in linux compatibility layer which
  require it. All linux applications that using sound + mmap() (mostly games)
  require this to be enabled. Disabled by default.
- Other goodies.. too many, that will increase releng7 shareholder value
  and make users of releng6 (and below) cry ;)

* This commit should be atomic. If anything goes wrong (not counting problem
  originated from elsewhere), I will not hesitate to revert everything back
  within 12 hours. This substantial changes itself not a rocket science
  and the process has begun for almost 2 years, and lots of incremental
  changes are already in place during that period of time.
* Some issues does occur in snd_emu10kx (note the 'x') due to various
  internal locking issues and it is currently being worked on by chibis.

Tested by: chibis (Yuriy Tsibizov), joel, Alexandre Vieira,
           many innocent souls...

17 years ago- Raise max range of sctp_logging sysctl so panda does not disallow
rrs [Sat, 16 Jun 2007 03:28:18 +0000 (03:28 +0000)]
- Raise max range of sctp_logging sysctl so panda does not disallow
  us to turn on logging levels.

17 years agoDisable TSO support.
yongari [Sat, 16 Jun 2007 02:54:19 +0000 (02:54 +0000)]
Disable TSO support.
Without bus_dma clean up and increment of number of Tx descriptors
it's hard to guarantee correct Tx operation in TSO case. The TSO
support would be enabled again when I get more feeback from re(4)
patch posted to current.

17 years agoIn previous releases of FreeBSD we have enforced -EfGm, and the
delphij [Sat, 16 Jun 2007 02:43:44 +0000 (02:43 +0000)]
In previous releases of FreeBSD we have enforced -EfGm, and the
new less(1) version enforced only -Em.  Restore old behavior in
respect with POLA.

Submitted by: ru (on -current@)

17 years agoAdd a manual page for tmpfs(5).
delphij [Sat, 16 Jun 2007 02:35:29 +0000 (02:35 +0000)]
Add a manual page for tmpfs(5).

17 years ago- Migrated IPMI fix from RELENG_6.
davidch [Sat, 16 Jun 2007 02:27:03 +0000 (02:27 +0000)]
- Migrated IPMI fix from RELENG_6.
- Added additional debug code.

17 years agoMFp4: Add regression tests for tmpfs.
delphij [Sat, 16 Jun 2007 02:04:44 +0000 (02:04 +0000)]
MFp4: Add regression tests for tmpfs.

Obtained from: NetBSD via p4
Submitted by: Howard Su

17 years agoMFp4: Add tmpfs, an efficient memory file system.
delphij [Sat, 16 Jun 2007 01:56:05 +0000 (01:56 +0000)]
MFp4: Add tmpfs, an efficient memory file system.

Please note that, this is currently considered as an
experimental feature so there could be some rough
edges.  Consult http://wiki.freebsd.org/TMPFS for
more information.

For now, connect tmpfs to build on i386 and amd64
architectures only.  Please let us know if you have
success with other platforms.

This work was developed by Julio M. Merino Vidal
for NetBSD as a SoC project; Rohit Jalan ported it
from NetBSD to FreeBSD.  Howard Su and Glen Leeder
are worked on it to continue this effort.

Obtained from: NetBSD via p4
Submitted by: Howard Su (with some minor changes)
Approved by: re (kensmith)

17 years ago- Lock sbp_write_cmd() and ORB_POINTER_ACTIVE flag.
simokawa [Sat, 16 Jun 2007 00:59:41 +0000 (00:59 +0000)]
- Lock sbp_write_cmd() and ORB_POINTER_ACTIVE flag.
- Remove unnecessary timestamps.
- Return CAM_RESRC_UNAVAIL for ORB shortage.
- Fix a lock problem when doorbell is used.
- Fix a potential bug for unordered execution.

17 years ago- Matthew's changes to get inlines out, plus a few of my own
rrs [Sat, 16 Jun 2007 00:33:47 +0000 (00:33 +0000)]
-  Matthew's changes to get inlines out, plus a few of my own
   to deal with the VRF inline function -> becomes a macro now.
Submitted by: Matthew Jacobs

17 years agoIn case of failure we can directly return ENOBUFS because
bz [Sat, 16 Jun 2007 00:15:14 +0000 (00:15 +0000)]
In case of failure we can directly return ENOBUFS because
'result' is still NULL and we do not need to free anything.
That allows us to gc the entire goto parts and a now unused variable.

Found with: Coverity Prevent(tm)
CID: 2519

17 years agoInitialize some variables that GCC4.2 thinks might possibly be used without
mjacob [Fri, 15 Jun 2007 23:49:54 +0000 (23:49 +0000)]
Initialize some variables that GCC4.2 thinks might possibly be used without
being initialized.

17 years agoAdd a missing return so that we drop out in case of an error and
bz [Fri, 15 Jun 2007 23:45:39 +0000 (23:45 +0000)]
Add a missing return so that we drop out in case of an error and
do not continue with a NULL pointer. [1]

While here change the return of the error handling code path above.
I cannot see why we should always return 0 there. Neither does KAME
nor do we in here for the similar check in all the other functions.

Found with: Coverity Prevent(tm) [1]
CID: 2521

17 years agoPrototype (but functional) Linux-ish /dev/nvram interface to the extra
peter [Fri, 15 Jun 2007 22:58:14 +0000 (22:58 +0000)]
Prototype (but functional) Linux-ish /dev/nvram interface to the extra
114 bytes of cmos ram in the PC clock chip.  The big difference between
this and the Linux version is that we do not recalculate the checksums
for bytes 16..31.

We use this at work when cloning identical machines - we can copy the
bios settings as well.  Reading /dev/nvram gives 114 bytes of data but
you can seek/read/write whichever bytes you like.

Yes, this is a "foot, gun, fire!" type of device.

17 years agoGarbage collect unused variables.
mjacob [Fri, 15 Jun 2007 22:56:12 +0000 (22:56 +0000)]
Garbage collect unused variables.

17 years agoGarbage collect some debug code that not only no longer could
mjacob [Fri, 15 Jun 2007 22:54:11 +0000 (22:54 +0000)]
Garbage collect some debug code that not only no longer could
work but in fact probably causes a random pointer dereferences.
Garbage collect the tp variable too.

17 years agoWith the current code 'src' is never NULL. Nevertheless move the check for
bz [Fri, 15 Jun 2007 22:35:59 +0000 (22:35 +0000)]
With the current code 'src' is never NULL. Nevertheless move the check for
NULL before dereferencing the pointer.

Found with: Coverity Prevent(tm)
CID: 2528

17 years agoAdd information about the implications of using mmap(2) instead of sbrk(2).
jasone [Fri, 15 Jun 2007 22:32:33 +0000 (22:32 +0000)]
Add information about the implications of using mmap(2) instead of sbrk(2).

Submitted by: bmah, jhb

17 years agoLooking at {ah,esp}_input_cb it seems we might be able to end up
bz [Fri, 15 Jun 2007 22:23:33 +0000 (22:23 +0000)]
Looking at {ah,esp}_input_cb it seems we might be able to end up
without an mtag in ipsec4_common_input_cb.
So in case of !IPCOMP (AH,ESP) only change the m_tag_id if an mtag
was passed to ipsec4_common_input_cb.

Found with: Coverity Prevent(tm)
CID: 2523

17 years agoFix junk/zero filling for realloc(). Junk filling was missing in one case,
jasone [Fri, 15 Jun 2007 22:00:16 +0000 (22:00 +0000)]
Fix junk/zero filling for realloc().  Junk filling was missing in one case,
and zero filling was broken in a way that could cause memory corruption.

Update comments.

17 years agoRemove stale 'XXX implement' comments for syscalls which have since been
emaste [Fri, 15 Jun 2007 21:54:26 +0000 (21:54 +0000)]
Remove stale 'XXX implement' comments for syscalls which have since been
implemented.

17 years agoThe mac address must be written a word length at a time, it was having no effect...
thompsa [Fri, 15 Jun 2007 21:45:41 +0000 (21:45 +0000)]
The mac address must be written a word length at a time, it was having no effect before.

MFC after: 1 week

17 years agos,#,*, in a multi-line comment. This is C.
bz [Fri, 15 Jun 2007 21:34:12 +0000 (21:34 +0000)]
s,#,*, in a multi-line comment. This is C.
No functional change.

17 years agoThough we are only called for the three security protocols we can
bz [Fri, 15 Jun 2007 21:32:51 +0000 (21:32 +0000)]
Though we are only called for the three security protocols we can
handle, document those sprotos using an IPSEC_ASSERT so that it will
be clear that 'spi' will always be initialized when used the first time.

Found with: Coverity Prevent(tm)
CID: 2533

17 years agoName change SCTP_KTR_SUBSYS -> KTR_SCTP
rrs [Fri, 15 Jun 2007 20:54:12 +0000 (20:54 +0000)]
Name change SCTP_KTR_SUBSYS -> KTR_SCTP

17 years agoRemove extraneous extern (its gotten from sctp_sysctl.h)
rrs [Fri, 15 Jun 2007 20:23:41 +0000 (20:23 +0000)]
Remove extraneous extern (its gotten from sctp_sysctl.h)

17 years agoFix build warnings
kmacy [Fri, 15 Jun 2007 20:02:02 +0000 (20:02 +0000)]
Fix build warnings
Submitted by: mjacob@

17 years agoWhen removing a stream from the output-stream-wheel, if its the
rrs [Fri, 15 Jun 2007 19:49:13 +0000 (19:49 +0000)]
When removing a stream from the output-stream-wheel, if its the
first stream we saw we must update the starting point in the
wheel, else we may loop in an endless loop.

17 years ago - Fix an off by one error in sched_pri_range.
jeff [Fri, 15 Jun 2007 19:33:58 +0000 (19:33 +0000)]
 - Fix an off by one error in sched_pri_range.
 - In tdq_choose() only assert that a thread does not have too high a
   priority (low value) for the queue we removed it from.  This will catch
   bugs in priority elevation.  It's not a serious error for the thread
   to have too low a priority as we don't change queues in this case as
   an optimization.

Reported by: kris

17 years ago- Update the comment lines in sctp_input.c
rrs [Fri, 15 Jun 2007 19:28:58 +0000 (19:28 +0000)]
- Update the comment lines in sctp_input.c
- We need to init the INP_LOCK since otherwise for
  non-SMP kernels you crash when you set the TOS.

17 years agoStub out imported IGMPv3 definitions which clash with those of
bms [Fri, 15 Jun 2007 18:59:10 +0000 (18:59 +0000)]
Stub out imported IGMPv3 definitions which clash with those of
the XORP router; the IGMPv3 definitions will be updated at a later
point in time when IGMPv3/MLDv2 support is fully merged.

17 years agoExpand TCP counters from 9 digits to 12.
jhb [Fri, 15 Jun 2007 18:12:58 +0000 (18:12 +0000)]
Expand TCP counters from 9 digits to 12.

MFC after: 1 week
PR: bin/112881

17 years agoConvert magic to a uintptr_t. This should get rid of some warnings on
njl [Fri, 15 Jun 2007 18:02:34 +0000 (18:02 +0000)]
Convert magic to a uintptr_t.  This should get rid of some warnings on
gcc4.

17 years agoConsistently indent the R() macro for fields on the right-side of the
jhb [Fri, 15 Jun 2007 18:00:10 +0000 (18:00 +0000)]
Consistently indent the R() macro for fields on the right-side of the
display to make the code easier to read.

PR: bin/112881

17 years ago- Issue one, new stack reduction left packet_drop handling still
rrs [Fri, 15 Jun 2007 17:59:57 +0000 (17:59 +0000)]
- Issue one, new stack reduction left packet_drop handling still
  thinking it had the whole chunk. This could cause a crash if
  a large packet drop came in. Fixed by adjusting the trunc length
  down to the limit.
- Large sacks with lots of segments could also have same issue. Changed
  duplicate and segment handling to use proper get_m_ptr function to
  pull each block from mbuf chains.

17 years agoAdd a new counter for retransmitted packets due to SACK.
jhb [Fri, 15 Jun 2007 17:55:44 +0000 (17:55 +0000)]
Add a new counter for retransmitted packets due to SACK.

PR: bin/112881
Submitted by: Phil Rosenthal <pr  isprime com>

17 years agoSimplify/optimize RB_NFIND().
jasone [Fri, 15 Jun 2007 16:09:47 +0000 (16:09 +0000)]
Simplify/optimize RB_NFIND().

Submitted by: Andriy Gapon <avg@icyb.net.ua>

17 years agoNew release notes: 802.11 stack megacommit, IPX-over-IP gone,
bmah [Fri, 15 Jun 2007 15:41:22 +0000 (15:41 +0000)]
New release notes:  802.11 stack megacommit, IPX-over-IP gone,
SSM, pam_nologin(8) change, GNU Diffutils 2.8.7.

17 years agoIn setaudit_addr(), drop the process lock in error cases.
rwatson [Fri, 15 Jun 2007 15:20:56 +0000 (15:20 +0000)]
In setaudit_addr(), drop the process lock in error cases.

Submitted by: Peter Holm <peter@holm.cc> (BugMaster)

17 years agoThe i386 "fixit" floppy began overflowing, most likely due to the new
kensmith [Fri, 15 Jun 2007 14:41:11 +0000 (14:41 +0000)]
The i386 "fixit" floppy began overflowing, most likely due to the new
gcc.  Replace vi(1) with ed(1) as the editor to make the contents fit
on a floppy again.

17 years ago- Add an option to change escape character.
simokawa [Fri, 15 Jun 2007 12:09:16 +0000 (12:09 +0000)]
- Add an option to change escape character.
- Use CTRL macro.
- Make target reset work on telnet port.
- Add a key bind to invoke kgdb on the terminal. (experimental)

17 years agoThird stage of unbreaking printing of pseudo-nice values (realtime
bde [Fri, 15 Jun 2007 12:03:07 +0000 (12:03 +0000)]
Third stage of unbreaking printing of pseudo-nice values (realtime
priorities, etc.) in the NICE field:

Use a combination of pri_native and pri_user instead of pri_level to
guess the original realtime priority.  Using pri_level here has been
wrong since 2001/02/12.  Using only pri_native here would be correct
if the kernel actually initialized it reasonably.  (The kernel exports
its raw td_base_priority as pri_native, but userland mostly wants a
refined base priority).  Give up on waiting pri_native to work correctly
and only use it when there is nothing better (for kthreads).

This should reduce printing of bizarre pseudo-nice values.  Bizarre
values are still printed if we observe a transient borrowed priority
for a kthread (transient borrowing is the main thing that makes the
raw td_base_priority almost useless in userland), or if there is a
kernel bug.  One current kernel bug involves the kernel idprio thread
pagezero permanently changing its priority from PRI_MAX_IDLE (255) to
PUSER (160).  Then the bizarre value "ki-6" is printed instead of
"ki31".  Here "-6" is PRI_MIN_IDLE - PUSER = -64 truncated to 2
characters.  We are observing a transient borrowed priority that has
become permanent due to a bug.

ps/print.c:priorityr() needs similar changes (including ones in stage 2
here).

17 years agoAdd PAM support to atrun(8).
yar [Fri, 15 Jun 2007 12:02:16 +0000 (12:02 +0000)]
Add PAM support to atrun(8).

17 years agoRe-enable raw dump format support on i386 and amd64 for /dev/fwmem.
simokawa [Fri, 15 Jun 2007 11:35:11 +0000 (11:35 +0000)]
Re-enable raw dump format support on i386 and amd64 for /dev/fwmem.

17 years agoLocked out and expired accounts shouldn't be accessible via remote
yar [Fri, 15 Jun 2007 11:33:13 +0000 (11:33 +0000)]
Locked out and expired accounts shouldn't be accessible via remote
mailbox protocols.  Add pam_unix to the `account' function class, too,
for imap and pop3 to actually implement this policy.

17 years agoSplit the FILES list across multiple lines as in rc.d/Makefile
yar [Fri, 15 Jun 2007 11:22:10 +0000 (11:22 +0000)]
Split the FILES list across multiple lines as in rc.d/Makefile
so that the change history stays easily readable as the number
of PAM-aware services grows.

17 years agoActually use new features of perr() and perrx().
yar [Fri, 15 Jun 2007 10:34:36 +0000 (10:34 +0000)]
Actually use new features of perr() and perrx().

17 years agoCatch up with perr() becoming variadic.
yar [Fri, 15 Jun 2007 10:12:37 +0000 (10:12 +0000)]
Catch up with perr() becoming variadic.

17 years agoMake perr() variadic and add perrx() to use in cases where
yar [Fri, 15 Jun 2007 10:10:40 +0000 (10:10 +0000)]
Make perr() variadic and add perrx() to use in cases where
errno is irrelevant.  Some code duplication can be reduced
if perr() is variadic and perrx() is available.

17 years agoThis commit was generated by cvs2svn to compensate for changes in r170764,
delphij [Fri, 15 Jun 2007 07:49:35 +0000 (07:49 +0000)]
This commit was generated by cvs2svn to compensate for changes in r170764,
which included commits to RCS files with non-trunk default branches.

17 years agoApply vendor patch that implements 1/2 as aliases of l/r.
delphij [Fri, 15 Jun 2007 07:49:35 +0000 (07:49 +0000)]
Apply vendor patch that implements 1/2 as aliases of l/r.

Requested by:   dmalone, Chuck Swiger

17 years agoUpdate instructions about upgrading.
delphij [Fri, 15 Jun 2007 07:36:39 +0000 (07:36 +0000)]
Update instructions about upgrading.

17 years agoAdd a list of removed files during import
delphij [Fri, 15 Jun 2007 07:28:00 +0000 (07:28 +0000)]
Add a list of removed files during import

17 years agoRemove files that were taken off vendor branch. Difference
delphij [Fri, 15 Jun 2007 07:22:26 +0000 (07:22 +0000)]
Remove files that were taken off vendor branch.  Difference
against vendor branch is now maintained in patchsets.

17 years agoRemove old files that no longer exist in latest distribution or
delphij [Fri, 15 Jun 2007 07:17:50 +0000 (07:17 +0000)]
Remove old files that no longer exist in latest distribution or
removed during import.

17 years agoThis commit was generated by cvs2svn to compensate for changes in r170759,
delphij [Fri, 15 Jun 2007 07:17:50 +0000 (07:17 +0000)]
This commit was generated by cvs2svn to compensate for changes in r170759,
which included commits to RCS files with non-trunk default branches.

17 years agoUpdate build skeleton after diffutils 2.8.7 update.
delphij [Fri, 15 Jun 2007 07:09:34 +0000 (07:09 +0000)]
Update build skeleton after diffutils 2.8.7 update.

17 years agoLocalized regeneration of config.h.
delphij [Fri, 15 Jun 2007 07:08:19 +0000 (07:08 +0000)]
Localized regeneration of config.h.

17 years agoVirgin import of diffutils 2.8.7.
delphij [Fri, 15 Jun 2007 07:06:13 +0000 (07:06 +0000)]
Virgin import of diffutils 2.8.7.

17 years agoThis commit was generated by cvs2svn to compensate for changes in r170754,
delphij [Fri, 15 Jun 2007 07:06:13 +0000 (07:06 +0000)]
This commit was generated by cvs2svn to compensate for changes in r170754,
which included commits to RCS files with non-trunk default branches.

17 years agoAdd a man page for wlan_amrr
kevlo [Fri, 15 Jun 2007 04:41:15 +0000 (04:41 +0000)]
Add a man page for wlan_amrr

Reviewed by: sam, sephe

17 years agoYou can never have enough documentation.
gshapiro [Fri, 15 Jun 2007 03:21:20 +0000 (03:21 +0000)]
You can never have enough documentation.

Document the SENDMAIL_ALIASES, SENDMAIL_MAP_SRC, SENDMAIL_MAP_TYPE, and
SENDMAIL_START_SCRIPT
.Xr make.conf 5
variables.
These are used in
.Pa /etc/mail/Makefile .

PR: conf/40548
MFC after: 3 days

17 years ago- Add VRF id to sctp_ifa structure, needed mainly in panda but useful
rrs [Fri, 15 Jun 2007 03:16:48 +0000 (03:16 +0000)]
- Add VRF id to sctp_ifa structure, needed mainly in panda but useful
  during deletes of ifa's in diff VRF's when applicable.

17 years ago- Conditionally pickup Giant around the network interface
csjp [Fri, 15 Jun 2007 02:53:51 +0000 (02:53 +0000)]
- Conditionally pickup Giant around the network interface
  ioctl routines if we are running with !mpsafenet
- Change un-conditional Giant acquisition around ifpromisc
  to occur only if we are running with !mpsafenet

With these locking bits in place, we can now remove the Giant
requirement from BPF, so drop the D_NEEDGIANT device flag.
This change removes Giant acquisitions around BPF device
handlers (read, write, ioctl etc).

MFC after: 1 month
Discussed with: rwatson

17 years agoKTR_GEN -> KTR_SUBSYS (for Kris).
rrs [Fri, 15 Jun 2007 02:34:36 +0000 (02:34 +0000)]
KTR_GEN -> KTR_SUBSYS (for Kris).

17 years ago- Oppps, forgot to update out the notes file for LINT builds- purge
rrs [Fri, 15 Jun 2007 02:29:19 +0000 (02:29 +0000)]
- Oppps, forgot to  update out the notes file for LINT builds- purge
  old logging options that are no longer needed.

17 years agoRemove the restriction that rtprio(2) cannot be used to set the realtime
rwatson [Thu, 14 Jun 2007 23:31:52 +0000 (23:31 +0000)]
Remove the restriction that rtprio(2) cannot be used to set the realtime
or idle priority of another process owned by the same user.  This means
that privilege in rtprio(2) (and rtprio_thread(2)) is required indirectly
via p_cansched(9) or directly to set realtime/idle privilege, rather than
directly affecting target process authorization.

17 years ago- Fix so ifn's are properly deleted when the ref count goes to 0.
rrs [Thu, 14 Jun 2007 22:59:04 +0000 (22:59 +0000)]
- Fix so ifn's are properly deleted when the ref count goes to 0.
- Fix so VRF's will clean themselves up when no references are around.
- Allow sctp_ifa to be passed into inpcb_bind, addr_mgmt_ep_sa to bypass
  normal validation checks.
- turn auto-asconf off for subset bound sockets
- Moves all logging to use KTR. This gets rid of most
  of the logging #ifdef's with a few exceptions reducing
  the number of config options for SCTP.

17 years agoThe struct selinfo pointed to by *sip MUST be zeroed before calling
njl [Thu, 14 Jun 2007 22:38:50 +0000 (22:38 +0000)]
The struct selinfo pointed to by *sip MUST be zeroed before calling
selrecord() or selwake*().  Otherwise, a panic may occur.

MFC after: 1 day

17 years agoUse a single setusercontext(3) instead of a bunch of basic syscalls.
yar [Thu, 14 Jun 2007 22:16:21 +0000 (22:16 +0000)]
Use a single setusercontext(3) instead of a bunch of basic syscalls.
Besides aesthetic benefits, that makes at(1) jobs subject to such
login.conf(5) settings as resource limits.

17 years agoSpotted another copy of the loony error message.
yar [Thu, 14 Jun 2007 21:24:05 +0000 (21:24 +0000)]
Spotted another copy of the loony error message.

17 years agoMake the order of tests the same for user and kernel. Use the natural
harti [Thu, 14 Jun 2007 20:11:48 +0000 (20:11 +0000)]
Make the order of tests the same for user and kernel. Use the natural
layer order from bottom to top.

17 years agoUse an array of size NGROUP_MAX for the getgroups() call instead of NGRP.
harti [Thu, 14 Jun 2007 20:07:35 +0000 (20:07 +0000)]
Use an array of size NGROUP_MAX for the getgroups() call instead of NGRP.
When NGROUP_MAX is larger than NGRP the call used to fail. Now the call
succeedes, but only the first NGRP groups are actually used for authentication.

17 years agoAccording to the documentation mech_type for gss_init_sec_context() may
harti [Thu, 14 Jun 2007 19:58:24 +0000 (19:58 +0000)]
According to the documentation mech_type for gss_init_sec_context() may
be passed as GSS_C_NO_OID in which case a default mech should be used.
This case was not handled and leads to core dumps when using nss_ldap.
Now use the first mech in this case. When there is no mechanism available
return an error (this part is taken from the PR).

PR: 113266
Submitted by: Eirik Nygaard <eirikald@pvv.ntnu.no> (partly)

17 years agocorrect some limits on interrupt proccessing so that
gallatin [Thu, 14 Jun 2007 19:35:03 +0000 (19:35 +0000)]
correct some limits on interrupt proccessing so that
fast forwarding back out the same mxge interface works nicely.

17 years agoOnly require privilege to set the current time adjustment, not in order to
rwatson [Thu, 14 Jun 2007 18:37:58 +0000 (18:37 +0000)]
Only require privilege to set the current time adjustment, not in order to
query it.