]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoAdd descriptions for new path, info, maint, cat, bdeps, and
krion [Wed, 16 Jun 2004 07:29:21 +0000 (07:29 +0000)]
Add descriptions for new path, info, maint, cat, bdeps, and
rdeps variables and some new examples for "make search".

Approved by: josef (mentor)
MFC after: 2 days

20 years agoUpdated Bmake framework for Binutils 2.15.
obrien [Wed, 16 Jun 2004 07:09:44 +0000 (07:09 +0000)]
Updated Bmake framework for Binutils 2.15.

20 years agomdoc(7) police
bms [Wed, 16 Jun 2004 07:03:54 +0000 (07:03 +0000)]
mdoc(7) police

Submitted by: ru

20 years agoMFamd64
alc [Wed, 16 Jun 2004 07:03:15 +0000 (07:03 +0000)]
MFamd64
 Introduce pmap locking to many of the pmap functions.

20 years agoMake netstat(1) more closely follow documented behaviour. If a TCP
bms [Wed, 16 Jun 2004 07:00:50 +0000 (07:00 +0000)]
Make netstat(1) more closely follow documented behaviour. If a TCP
socket in LISTEN state happens to be bound to an interface, it will
show up in netstat(1) output even without the -a switch.

As the definition of "sockets used by server processes" is a
difficult one to qualify with regards to UDP, do not change the
output behaviour for UDP sockets.

PR: bin/26359

20 years agoUpdate HEAD with the stock binutils_2_15_20040523 file.
obrien [Wed, 16 Jun 2004 06:55:49 +0000 (06:55 +0000)]
Update HEAD with the stock binutils_2_15_20040523 file.

20 years agoWhen parsing an argument as an address in getaddr(), heed the -host option
bms [Wed, 16 Jun 2004 06:29:41 +0000 (06:29 +0000)]
When parsing an argument as an address in getaddr(), heed the -host option
in all cases.  The failure mode in the PR is easily reproducible without
this patch.

PR: bin/43139

20 years agoDocument the behaviour of the net.link.ether.inet.proxyall sysctl MIB
bms [Wed, 16 Jun 2004 06:16:29 +0000 (06:16 +0000)]
Document the behaviour of the net.link.ether.inet.proxyall sysctl MIB
variable.

PR: docs/21826
Reviewed by: ru
Requested by: Bernd Luevelsmeyer

20 years agoBlah, these came back. Use a bigger hammer.
obrien [Wed, 16 Jun 2004 06:14:51 +0000 (06:14 +0000)]
Blah, these came back.  Use a bigger hammer.

20 years agoMerge rev 1.2 (configuration changes to support FreeBSD) into Binutils 2.15.
obrien [Wed, 16 Jun 2004 06:13:06 +0000 (06:13 +0000)]
Merge rev 1.2 (configuration changes to support FreeBSD) into Binutils 2.15.

20 years agoMerge rev 1.2 (teach `ld' how to access FreeBSD's ld ELF hints) into
obrien [Wed, 16 Jun 2004 06:09:06 +0000 (06:09 +0000)]
Merge rev 1.2 (teach `ld' how to access FreeBSD's ld ELF hints) into
Binutils 2.15.

20 years agoMerge rev 1.2 (FreeBSD a.out configuration support) into Binutils 2.15.
obrien [Wed, 16 Jun 2004 06:03:34 +0000 (06:03 +0000)]
Merge rev 1.2 (FreeBSD a.out configuration support) into Binutils 2.15.

20 years agoMerge our ELF branding into Binutils 2.15.
obrien [Wed, 16 Jun 2004 06:02:14 +0000 (06:02 +0000)]
Merge our ELF branding into Binutils 2.15.

20 years agoThis commit was generated by cvs2svn to compensate for changes in r130561,
obrien [Wed, 16 Jun 2004 05:45:41 +0000 (05:45 +0000)]
This commit was generated by cvs2svn to compensate for changes in r130561,
which included commits to RCS files with non-trunk default branches.

20 years agoImport of Binutils from the FSF 2.15 branch (just post-.0 release).
obrien [Wed, 16 Jun 2004 05:45:41 +0000 (05:45 +0000)]
Import of Binutils from the FSF 2.15 branch (just post-.0 release).
These bits are taken from the FSF anoncvs repo on 23-May-2004 04:41:00 UTC.

20 years agoMFamd64
alc [Wed, 16 Jun 2004 05:42:44 +0000 (05:42 +0000)]
MFamd64
 Remove dead or unneeded code, e.g., spl calls.

20 years agoDisconnect a temporarily-connected UDP socket in out-of-mbufs case. This
bms [Wed, 16 Jun 2004 05:41:00 +0000 (05:41 +0000)]
Disconnect a temporarily-connected UDP socket in out-of-mbufs case. This
fixes the problem of UDP sockets getting wedged in a connected state (and
bound to their destination) under heavy load.
Temporary bind/connect should probably be deleted in future
as an optimization, as described in "A Faster UDP" [Partridge/Pink 1993].

Notes:
 - INP_LOCK() is already held in udp_output(). The connection is in effect
   happening at a layer lower than the socket layer, therefore in theory
   socket locking should not be needed.
 - Inlining the in_pcbdisconnect() operation buys us nothing (in the case
   of the current state of the code), as laddr is not part of the
   inpcb hash or the udbinfo hash. Therefore there should be no need
   to rehash after restoring laddr in the error case (this was a
   concern of the original author of the patch).

PR: kern/41765
Requested by: gnn
Submitted by: Jinmei Tatuya (with cleanups)
Tested by: spray(8)

20 years agoDisable gdb from the build due to the impending binutils 2.15 import.
obrien [Wed, 16 Jun 2004 05:19:00 +0000 (05:19 +0000)]
Disable gdb from the build due to the impending binutils 2.15 import.
GDB 5.12 cannot consume the 2.15 bfd and libiberty.

20 years agoTurn on RFC 1724 compliant behaviour with regards to IP_ADD_MEMBERSHIP;
bms [Wed, 16 Jun 2004 05:08:57 +0000 (05:08 +0000)]
Turn on RFC 1724 compliant behaviour with regards to IP_ADD_MEMBERSHIP;
routed should be able to specify multicast memberships to be added by
interface index. This should fix the unnumbered / point-to-point case
for RIPv2.

PR: bin/51927
Requested by: Eugene Grosbein

20 years agoDocument the behaviour of IP_ADD_MEMBERSHIP as per RFC 1724.
bms [Wed, 16 Jun 2004 05:04:17 +0000 (05:04 +0000)]
Document the behaviour of IP_ADD_MEMBERSHIP as per RFC 1724.

PR: bin/51927

20 years agoConvert GIANT_REQUIRED to NET_ASSERT_GIANT for socket access.
rwatson [Wed, 16 Jun 2004 03:36:06 +0000 (03:36 +0000)]
Convert GIANT_REQUIRED to NET_ASSERT_GIANT for socket access.

20 years agoConvert GIANT_REQUIRED to NET_ASSERT_GIANT where Giant is used to
rwatson [Wed, 16 Jun 2004 03:12:50 +0000 (03:12 +0000)]
Convert GIANT_REQUIRED to NET_ASSERT_GIANT where Giant is used to
protect socket operations.  Leave one "as-is" as it also frobs
rootvp.

20 years agoCorrect an error in the implementation of pmap_is_prefaultable(). When I
alc [Wed, 16 Jun 2004 03:11:24 +0000 (03:11 +0000)]
Correct an error in the implementation of pmap_is_prefaultable().  When I
introduced this function in revision 1.441, I inverted one of the
comparisons.

20 years agoThis library has to change whenever the kernel process structure changes.
julian [Wed, 16 Jun 2004 00:34:31 +0000 (00:34 +0000)]
This library has to change whenever the kernel process structure changes.

20 years agoNice, is a property of a process as a whole..
julian [Wed, 16 Jun 2004 00:26:31 +0000 (00:26 +0000)]
Nice, is a property of a process as a whole..
I mistakenly moved it to the ksegroup when breaking up the process
structure. Put it back in the proc structure.

20 years agoDisable "bulk dequeue" when enabling ALTQ so it does not irritate the
mlaier [Tue, 15 Jun 2004 23:59:37 +0000 (23:59 +0000)]
Disable "bulk dequeue" when enabling ALTQ so it does not irritate the
timing.

20 years agoReplace IF_HANDOFF with new IFQ_HANDOFF to enqueue with ALTQ once enabled on
mlaier [Tue, 15 Jun 2004 23:57:42 +0000 (23:57 +0000)]
Replace IF_HANDOFF with new IFQ_HANDOFF to enqueue with ALTQ once enabled on
the respective drivers.

20 years agoChange strategy based on a suggestion from Ian Dowse. Instead of trying
peter [Tue, 15 Jun 2004 23:57:02 +0000 (23:57 +0000)]
Change strategy based on a suggestion from Ian Dowse.  Instead of trying
to keep track of different section base addresses at a symbol-by-symbol
level, just set the symbol values at load time.

20 years agoAvoid to touch red zone, in libpthread, ucontext may be saved by kernel's
davidxu [Tue, 15 Jun 2004 21:46:36 +0000 (21:46 +0000)]
Avoid to touch red zone, in libpthread, ucontext may be saved by kernel's
get_mcontext, and libpthread will use signalcontext to deliver signal in
userland, it looks same as kernel's send_sig does.

Reviewed by: deischen, tjr

20 years agoFix several bugs related to subdisk drive_offset calculation.
le [Tue, 15 Jun 2004 20:56:25 +0000 (20:56 +0000)]
Fix several bugs related to subdisk drive_offset calculation.

20 years agoOops. My last commit included a bug that would make "su -m" always
markm [Tue, 15 Jun 2004 20:23:02 +0000 (20:23 +0000)]
Oops. My last commit included a bug that would make "su -m" always
use /bin/sh. Fix this.

20 years agoIf a device is not ONLINE, do not attempt to submit any commands
ps [Tue, 15 Jun 2004 19:40:47 +0000 (19:40 +0000)]
If a device is not ONLINE, do not attempt to submit any commands
or I/O to the volume.  This solves a panic when removing a disk in
a RAID 0 volume.

Reported by: le

20 years agoRemove a stale comment.
alc [Tue, 15 Jun 2004 19:28:40 +0000 (19:28 +0000)]
Remove a stale comment.

20 years agoOverhaul the markup.
ru [Tue, 15 Jun 2004 19:26:11 +0000 (19:26 +0000)]
Overhaul the markup.

20 years agoConnect geom(8) to the build.
pjd [Tue, 15 Jun 2004 17:31:35 +0000 (17:31 +0000)]
Connect geom(8) to the build.

Reminded by: des

20 years agos/device_get_handle/acpi_get_handle
njl [Tue, 15 Jun 2004 16:49:20 +0000 (16:49 +0000)]
s/device_get_handle/acpi_get_handle

Submitted by: Hiroyuki Aizu

20 years agomdoc(7) police
bms [Tue, 15 Jun 2004 16:48:42 +0000 (16:48 +0000)]
mdoc(7) police

Submitted by: ru

20 years agoFix spelling.
dfr [Tue, 15 Jun 2004 16:03:49 +0000 (16:03 +0000)]
Fix spelling.

20 years agoFill in locking annotation for additional socket fields:
rwatson [Tue, 15 Jun 2004 13:43:11 +0000 (13:43 +0000)]
Fill in locking annotation for additional socket fields:

so_timeo        Used as a sleep/wakeup address, no locking.
sb_*            Almost all socket buffer fields locked with
                sockbuf lock for the oskcet buffer.
so_cred         Static after socket creation.

20 years agoOverhaul the markup.
ru [Tue, 15 Jun 2004 12:48:50 +0000 (12:48 +0000)]
Overhaul the markup.

20 years agoOops, backout debug code..
sos [Tue, 15 Jun 2004 11:38:48 +0000 (11:38 +0000)]
Oops, backout debug code..

20 years agoDont set prefetch etc on VIA chips, causes problems on newer chips and
sos [Tue, 15 Jun 2004 11:16:36 +0000 (11:16 +0000)]
Dont set prefetch etc on VIA chips, causes problems on newer chips and
ATAPI devices.

20 years agoIncrease robustness of SATA handling.
sos [Tue, 15 Jun 2004 11:02:09 +0000 (11:02 +0000)]
Increase robustness of SATA handling.

20 years agoAdd pmap locking to pmap_extract(), pmap_mincore(), and pmap_remove().
alc [Tue, 15 Jun 2004 07:41:44 +0000 (07:41 +0000)]
Add pmap locking to pmap_extract(), pmap_mincore(), and pmap_remove().

20 years agoUpdate some internal comments about the --no-same-permissions option.
kientzle [Tue, 15 Jun 2004 07:16:44 +0000 (07:16 +0000)]
Update some internal comments about the --no-same-permissions option.
Thanks to: Kris Kennaway for doing some gtar research for me.

20 years agoBe a little more careful about creating directories:
kientzle [Tue, 15 Jun 2004 07:12:04 +0000 (07:12 +0000)]
Be a little more careful about creating directories:
 * Restore directories with 0700 permissions initially,
   then use the fixup pass to correct the permissions
 * Trim trailing "/" and "/." in mkdirpath()

Suggested by: Garrett Wollman

20 years agoAs near as I can tell, --no-same-permissions is a no-op in
kientzle [Tue, 15 Jun 2004 06:44:41 +0000 (06:44 +0000)]
As near as I can tell, --no-same-permissions is a no-op in
gtar, so that makes it easy to implement.

Required by: audio/timidity port
Thanks to: Kris Kennaway

20 years agoOops. bsdtar's old -X option didn't take an argument; the new
kientzle [Tue, 15 Jun 2004 06:24:54 +0000 (06:24 +0000)]
Oops.  bsdtar's old -X option didn't take an argument; the new
(gtar-compatible) one does require an argument.

20 years agoAdd gtar-compatible -X/--exclude-from
kientzle [Tue, 15 Jun 2004 05:55:41 +0000 (05:55 +0000)]
Add gtar-compatible -X/--exclude-from

20 years agoLock down rawcb_list, a global list of control blocks for raw sockets,
rwatson [Tue, 15 Jun 2004 04:13:59 +0000 (04:13 +0000)]
Lock down rawcb_list, a global list of control blocks for raw sockets,
using rawcb_mtx.  Hold this mutex while modifying or iterating over
the control list; this means that the mutex is held over calls into
socket delivery code, which no longer causes a lock order reversal as
the routing socket code uses a netisr to avoid recursing socket ->
routing -> socket.

Note: Locking of IPsec consumers of rawcb_list is not included in this
commit.

20 years agoGrab the socket buffer send or receive mutex when performing a
rwatson [Tue, 15 Jun 2004 03:51:44 +0000 (03:51 +0000)]
Grab the socket buffer send or receive mutex when performing a
read-modify-write on the sb_state field.  This commit catches only
the "easy" ones where it doesn't interact with as yet unmerged
locking.

20 years agoFix a typeo in IFQ_HANDOFF.
mlaier [Tue, 15 Jun 2004 03:40:39 +0000 (03:40 +0000)]
Fix a typeo in IFQ_HANDOFF.

20 years agoCatch one more use of acpi_MatchHid and update for new API.
njl [Tue, 15 Jun 2004 02:17:23 +0000 (02:17 +0000)]
Catch one more use of acpi_MatchHid and update for new API.

20 years agoWe only need the devclass_find() result, not the softc.
njl [Tue, 15 Jun 2004 02:12:12 +0000 (02:12 +0000)]
We only need the devclass_find() result, not the softc.

20 years agoFix build.
kientzle [Tue, 15 Jun 2004 01:51:45 +0000 (01:51 +0000)]
Fix build.

20 years agoTransform tbr_dequeue into a function pointer in order to build drivers with
mlaier [Tue, 15 Jun 2004 01:45:19 +0000 (01:45 +0000)]
Transform tbr_dequeue into a function pointer in order to build drivers with
ALTQ enabled versions of IFQ_* macros by default, as requested by serveral
others. This is a follow-up to the quick fix I committed yesterday which
turned off the ALTQ checks for non-ALTQ kernels.

20 years agoFix symbol lookups between modules. This caused modules that depend on
peter [Tue, 15 Jun 2004 01:35:57 +0000 (01:35 +0000)]
Fix symbol lookups between modules.  This caused modules that depend on
other modules to explode.  eg: snd_ich->snd_pcm and umass->usb.
The problem was that I was using the unified base address of the module
instead of finding the start address of the section in question.

20 years agoInsurance: cause a proper symbol lookup failure for symbol entries that
peter [Tue, 15 Jun 2004 01:33:39 +0000 (01:33 +0000)]
Insurance: cause a proper symbol lookup failure for symbol entries that
reference unknown sections.. rather than returning a small value.

20 years agoThe maximum value of iovcnt is IOV_MAX, not 16.
kbyanc [Tue, 15 Jun 2004 01:24:41 +0000 (01:24 +0000)]
The maximum value of iovcnt is IOV_MAX, not 16.

20 years agoAdd an MLINKS entry, rawip, to make the raw socket userland documentation
bms [Tue, 15 Jun 2004 01:18:56 +0000 (01:18 +0000)]
Add an MLINKS entry, rawip, to make the raw socket userland documentation
somewhat easier to find.

PR: docs/30873
Requested by: Douglas De Couto

20 years agoAdd a paragraph about the importance of byte order when using raw
bms [Tue, 15 Jun 2004 01:08:09 +0000 (01:08 +0000)]
Add a paragraph about the importance of byte order when using raw
sockets with the IP_HDRINCL option.
Add a list entry about the errors which IP_HDRINCL can return.
Bump .Dd.

PR: docs/30873
Submitted by: Douglas De Couto
Obtained from: BSD/OS (byte order), some old libnet stuff I did years ago.

20 years agoMake contigmalloc() more reliable:
green [Tue, 15 Jun 2004 01:02:00 +0000 (01:02 +0000)]
Make contigmalloc() more reliable:

1. Remove a race whereby contigmalloc() would deadlock against the
   running processes in the system if they kept reinstantiating
   the memory on the active and inactive page queues that it was
   trying to flush out.  The process doing the contigmalloc() would
   sit in "swwrt" forever and the swap pager would be going at full
   force, but never get anywhere.  Instead of doing it until the
   queues are empty, launder for as many iterations as there are
   pages in the queue.
2. Do all laundering to swap synchronously; previously, the vnode
   laundering was synchronous and the swap laundering not.
3. Increase the number of launder-or-allocate passes to three, from
   two, while failing without bothering to do all the laundering on
   the third pass if allocation was not possible.  This effectively
   gives exactly two chances to launder enough contiguous memory,
   helpful with high memory churn where a lot of memory from one pass
   to the next (and during a single laundering loop) becomes dirtied
   again.

I can now reliably hot-plug hardware requiring a 256KB contigmalloc()
without having the kldload/cbb ithread sit around failing to make
progress, while running a busy X session.  Previously, it took killing
X to get contigmalloc() to get further (that is, quiescing the system),
and even then contigmalloc() returned failure.

20 years agoAdd Device ID for: TI 802.11b/g Wireless Cardbus/PCI Adapter
bms [Tue, 15 Jun 2004 00:38:12 +0000 (00:38 +0000)]
Add Device ID for: TI 802.11b/g Wireless Cardbus/PCI Adapter

PR: conf/67700
Submitted by: Jukka A. Ukkonen
Reviewed by: http://www.pcidatabase.com/

20 years agoAdd Device ID for: Radeon 9200 Series - Secondary.
bms [Tue, 15 Jun 2004 00:36:08 +0000 (00:36 +0000)]
Add Device ID for: Radeon 9200 Series - Secondary.

PR: conf/67700
Submitted by: Jukka A. Ukkonen
Reviewed by: http://www.pcidatabase.com/

20 years agoRename -X to --one-file-system, as GNU tar uses -X for
kientzle [Tue, 15 Jun 2004 00:28:34 +0000 (00:28 +0000)]
Rename -X to --one-file-system, as GNU tar uses -X for
something else.  I would really like a short option for
this, but all of the obvious ones conflict with something else.

20 years agoIgnore kernel routes with the RTF_WASCLONED flag set. This is a slight
bms [Tue, 15 Jun 2004 00:23:26 +0000 (00:23 +0000)]
Ignore kernel routes with the RTF_WASCLONED flag set. This is a slight
modification of a patch which was already applied for BSD/OS in the
Rhyolite.com sources; this file is already off the vendor branch.

PR: bin/57484
Submitted by: Richard Perini

20 years agoClean up usage message(s):
kientzle [Tue, 15 Jun 2004 00:07:23 +0000 (00:07 +0000)]
Clean up usage message(s):
  * Usage goes to stderr, not stdout
  * Use correct argument markup
  * bsdtar --help  no longer exits with an error return code
  * ensure that the word "bsdtar" appears in the first
    line output from "bsdtar --help" (even if the program is
    invoked as "tar")

In particular, scripts can now test for the presence of bsdtar.
For example, in /bin/sh:

if (tar --help 2>&1 | grep bsdtar >/dev/null 2>&1) then \
          echo bsdtar; else echo not bsdtar; fi

20 years agoWhitespace.
bms [Mon, 14 Jun 2004 22:44:13 +0000 (22:44 +0000)]
Whitespace.

20 years agoDo not depend on the global 'sockt' being initialized to 0;
bms [Mon, 14 Jun 2004 22:43:05 +0000 (22:43 +0000)]
Do not depend on the global 'sockt' being initialized to 0;
instead, use the symbolic constant STDIN_FILENO, as this is
a daemon invoked from inetd.
Remove 'sockt' as it is not referenced.

20 years agoChange the default behaviour of talk(1) to use "localhost" as the
bms [Mon, 14 Jun 2004 22:34:13 +0000 (22:34 +0000)]
Change the default behaviour of talk(1) to use "localhost" as the
"machine name" in ntalkd(8) request packets, when the destination
and source are local.

This should make talk(1) use much more pleasant for those security-
conscious individuals who have chosen to bind talkd to "localhost".

Previous to this change, talk(1) would require that the hostname of
the machine, as retrieved by gethostname(3), resolved to a valid and
reachable IPv4 address, using gethostbyname(3).

This makes talk(1) dependent on a valid host entry for "localhost"
in /etc/hosts (or the Domain Name System).

PR: bin/23178
Submitted by: angui.sh admin (with cleanups)

20 years agoAdd HP Smart Array 6422 and V100 to the list of supported controllers.
simon [Mon, 14 Jun 2004 22:10:56 +0000 (22:10 +0000)]
Add HP Smart Array 6422 and V100 to the list of supported controllers.

20 years agoRemove unneeded '-' from comment header; this comment contains only
rwatson [Mon, 14 Jun 2004 22:03:14 +0000 (22:03 +0000)]
Remove unneeded '-' from comment header; this comment contains only
English text paragraphs that shouldn't have problems when run through
indent.

20 years agoUpdate for so_state->sb_state, SB_* flag renames.
rwatson [Mon, 14 Jun 2004 21:42:01 +0000 (21:42 +0000)]
Update for so_state->sb_state, SB_* flag renames.

Submitted by: rik

20 years agoUse %zu to print values with type size_t.
stefanf [Mon, 14 Jun 2004 20:55:30 +0000 (20:55 +0000)]
Use %zu to print values with type size_t.

20 years agoTeach fstat(1) about new location for socket state flags relating to
rwatson [Mon, 14 Jun 2004 20:54:18 +0000 (20:54 +0000)]
Teach fstat(1) about new location for socket state flags relating to
socket buffer state.

Submitted by: rik
Reminded by: le

20 years agomdoc(7) police
bms [Mon, 14 Jun 2004 19:31:10 +0000 (19:31 +0000)]
mdoc(7) police

Submitted by: ru

20 years agoAdd -b to usage.
bms [Mon, 14 Jun 2004 19:29:37 +0000 (19:29 +0000)]
Add -b to usage.

Pointed out by: ceri
Pointy hat to: bms

20 years agoRevert the removal of the initial_irq hack for now as this code is more
jhb [Mon, 14 Jun 2004 18:54:14 +0000 (18:54 +0000)]
Revert the removal of the initial_irq hack for now as this code is more
broken than I thought and doesn't do a good job of routing virgin
interrupts at all.

20 years agoDocument the location of the header file which declares the function
bms [Mon, 14 Jun 2004 18:41:24 +0000 (18:41 +0000)]
Document the location of the header file which declares the function
prototypes documented in this manual page.

PR: bin/4459

20 years agoAdd prototypes for the userland gmon support functions, which normally
bms [Mon, 14 Jun 2004 18:39:28 +0000 (18:39 +0000)]
Add prototypes for the userland gmon support functions, which normally
live in src/lib/libc/gmon/gmon.c. glibc puts these prototypes in the same
header, so put them here for the sake of consistency.

PR: bin/4459
Reviewed by: bde

20 years agoFix a couple of typos.
jhb [Mon, 14 Jun 2004 18:37:23 +0000 (18:37 +0000)]
Fix a couple of typos.

PR: doc/67894
Submitted by: Chris Pepper pepper at reppep dot com

20 years agoUpdate the .Dd date for previous commit.
jdp [Mon, 14 Jun 2004 18:33:21 +0000 (18:33 +0000)]
Update the .Dd date for previous commit.

Gently nudged by: krion

20 years agoChange the return value of sema_timedwait() so it returns 0 on
jdp [Mon, 14 Jun 2004 18:19:05 +0000 (18:19 +0000)]
Change the return value of sema_timedwait() so it returns 0 on
success and a proper errno value on failure.  This makes it
consistent with cv_timedwait(), and paves the way for the
introduction of functions such as sema_timedwait_sig() which can
fail in multiple ways.

Bump __FreeBSD_version and add a note to UPDATING.

Approved by: scottl (ips driver), arch

20 years agoThe socket field so_state is used to hold a variety of socket related
rwatson [Mon, 14 Jun 2004 18:16:22 +0000 (18:16 +0000)]
The socket field so_state is used to hold a variety of socket related
flags relating to several aspects of socket functionality.  This change
breaks out several bits relating to send and receive operation into a
new per-socket buffer field, sb_state, in order to facilitate locking.
This is required because, in order to provide more granular locking of
sockets, different state fields have different locking properties.  The
following fields are moved to sb_state:

  SS_CANTRCVMORE            (so_state)
  SS_CANTSENDMORE           (so_state)
  SS_RCVATMARK              (so_state)

Rename respectively to:

  SBS_CANTRCVMORE           (so_rcv.sb_state)
  SBS_CANTSENDMORE          (so_snd.sb_state)
  SBS_RCVATMARK             (so_rcv.sb_state)

This facilitates locking by isolating fields to be located with other
identically locked fields, and permits greater granularity in socket
locking by avoiding storing fields with different locking semantics in
the same short (avoiding locking conflicts).  In the future, we may
wish to coallesce sb_state and sb_flags; for the time being I leave
them separate and there is no additional memory overhead due to the
packing/alignment of shorts in the socket buffer structure.

20 years agoAdd whois.abuse.net to whois(1).
bms [Mon, 14 Jun 2004 17:30:46 +0000 (17:30 +0000)]
Add whois.abuse.net to whois(1).

Submitted by: ceri (with cleanups)

20 years agoDon't free a VINUMDRIVE softc when it's orphaned or spoiled. All
le [Mon, 14 Jun 2004 17:12:32 +0000 (17:12 +0000)]
Don't free a VINUMDRIVE softc when it's orphaned or spoiled.  All
allocated ressouces should be ultimately freed in gv_destroy_geom()
(when unloading the module and not earlier), but I need to look at this
more closely.

20 years agoCorrectly calculate subdisk offset in RAID5 plexes.
le [Mon, 14 Jun 2004 17:06:55 +0000 (17:06 +0000)]
Correctly calculate subdisk offset in RAID5 plexes.

20 years agoFactor out some duplicated code and fix some style(9) issues.
mux [Mon, 14 Jun 2004 16:53:20 +0000 (16:53 +0000)]
Factor out some duplicated code and fix some style(9) issues.

Submitted by: Liam J. Foy <liamfoy@sepulcrum.org>

20 years agoRemove some more leftover from the old pfaltq_module hack to allow for
mlaier [Mon, 14 Jun 2004 16:13:05 +0000 (16:13 +0000)]
Remove some more leftover from the old pfaltq_module hack to allow for
kernels w/ pf, but w/o altq.

Reported-by:  Xin LI

20 years agoWordsmith the BUGS section re: IPv6 link-local addresses.
bms [Mon, 14 Jun 2004 16:04:59 +0000 (16:04 +0000)]
Wordsmith the BUGS section re: IPv6 link-local addresses.

PR: docs/66541
Submitted by: Michel Lavondes (with some cleanups/additions)

20 years agoFixed some style bugs:
bde [Mon, 14 Jun 2004 15:37:49 +0000 (15:37 +0000)]
Fixed some style bugs:
- the comments on the rune/wide char types were lost
- struct cdev was bogusly forward declared
- some comments were excessively indented.

20 years agoImprove mapping of relative to absolute volume.
josef [Mon, 14 Jun 2004 15:01:16 +0000 (15:01 +0000)]
Improve mapping of relative to absolute volume.

I added bounds checking to the patch and cg improved
the formular.

Submitted by: Andriy Gapon <avg@icyb.net.ua>
PR: kern/65485
Approved by: cg
Reviewed by: imp, rwatson, le

20 years agoDate bump.
bms [Mon, 14 Jun 2004 14:49:44 +0000 (14:49 +0000)]
Date bump.

Nudged by: ru
Pointy hat: bms

20 years agoRemove a left over from userland buffer-cache access to disks.
phk [Mon, 14 Jun 2004 14:25:03 +0000 (14:25 +0000)]
Remove a left over from userland buffer-cache access to disks.

20 years agoDate bump.
bms [Mon, 14 Jun 2004 14:24:18 +0000 (14:24 +0000)]
Date bump.

Requested by: krion

20 years agoDocument some return values.
bms [Mon, 14 Jun 2004 14:18:41 +0000 (14:18 +0000)]
Document some return values.

PR: bin/22198
Submitted by: Nick Johnson (with cleanups)

20 years agoFixes from the mdoc(7) police.
dfr [Mon, 14 Jun 2004 14:12:29 +0000 (14:12 +0000)]
Fixes from the mdoc(7) police.

Submitted by: ru

20 years agoAdd whois.iana.org to the whois(1) utility, under the -I option.
bms [Mon, 14 Jun 2004 14:03:10 +0000 (14:03 +0000)]
Add whois.iana.org to the whois(1) utility, under the -I option.

PR: bin/48914
Submitted by: James Raftery

20 years agoFix formatting of AUTHORS section.
dfr [Mon, 14 Jun 2004 12:16:43 +0000 (12:16 +0000)]
Fix formatting of AUTHORS section.

20 years agoAdd a JOBS config keyword, which corresponds to the --jobs tinderbox(1)
des [Mon, 14 Jun 2004 12:10:48 +0000 (12:10 +0000)]
Add a JOBS config keyword, which corresponds to the --jobs tinderbox(1)
command line option.