]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoWARNS=4, de-__P()
alfred [Thu, 11 Jul 2002 22:29:11 +0000 (22:29 +0000)]
WARNS=4, de-__P()

22 years agoUpdate manpages to reference 'timed' rpc functions
alfred [Thu, 11 Jul 2002 22:25:16 +0000 (22:25 +0000)]
Update manpages to reference 'timed' rpc functions

22 years agoFix indentation broken in previous __P removal.
alfred [Thu, 11 Jul 2002 22:18:26 +0000 (22:18 +0000)]
Fix indentation broken in previous __P removal.

Grumbled about by: bde

22 years agothread_exit() requires PROC_LOCK to be held, so lock it.
jhb [Thu, 11 Jul 2002 22:13:33 +0000 (22:13 +0000)]
thread_exit() requires PROC_LOCK to be held, so lock it.

22 years agoWARNS=4, de-__P()
alfred [Thu, 11 Jul 2002 22:11:20 +0000 (22:11 +0000)]
WARNS=4, de-__P()

22 years agoAdd a missing newline during panic printf's for SMP systems that don't
jhb [Thu, 11 Jul 2002 21:56:37 +0000 (21:56 +0000)]
Add a missing newline during panic printf's for SMP systems that don't
have APICS.  (Like all the !i386 archs).

22 years agoRemove unused variables due to libufs(3) commit. Clean up extraneous use of
jmallett [Thu, 11 Jul 2002 21:46:12 +0000 (21:46 +0000)]
Remove unused variables due to libufs(3) commit.  Clean up extraneous use of
semicolon at closing paren of a function body.

22 years agoConvert dumpfs(8) to libufs(3), rather than rolling local functions for a
jmallett [Thu, 11 Jul 2002 21:44:03 +0000 (21:44 +0000)]
Convert dumpfs(8) to libufs(3), rather than rolling local functions for a
number of things.

22 years agoWARNS=4, de-__P()
alfred [Thu, 11 Jul 2002 21:40:15 +0000 (21:40 +0000)]
WARNS=4, de-__P()

22 years agode-__P()
alfred [Thu, 11 Jul 2002 21:28:16 +0000 (21:28 +0000)]
de-__P()

22 years agoFill out (zero) and fill in (when doing getino()) the minimum and maximum
jmallett [Thu, 11 Jul 2002 21:27:26 +0000 (21:27 +0000)]
Fill out (zero) and fill in (when doing getino()) the minimum and maximum
inodes in our inoblock (disk->d_ino{min,max}) appropriately.

22 years agoWARNS=4, de-__P()
alfred [Thu, 11 Jul 2002 21:26:41 +0000 (21:26 +0000)]
WARNS=4, de-__P()

22 years agoComplete rewrite, once again.
joerg [Thu, 11 Jul 2002 21:20:54 +0000 (21:20 +0000)]
Complete rewrite, once again.

This is basically a ``C compilation'' of the former whereis.pl file,
employing the same algorithms, and aiming at being mostly
UI-compatible to the old (legally tainted) 4.3BSD whereis(1).  In
comparision, the 4.4BSD-Lite version is just another variant of
which(1) only, where in particular the option to search for source
directories is sorely missing.

While i was at it, i added two more options which i contemplated doing
long since.  -x will suppress the run of locate(1) to find sources
that could not be found otherwise, potentially saving a lot of time
(but obviously, risking to not find some sources that are well hidden
in the tree).  -q will omit the leading name of the query, so in
particular, you can now do something like:

cd `whereis -qs ls`

I'd explicitly like to thank johan for his review which was quite a
bit more than an average review, including sending me a lot of diffs.

Reviewed by: johan

22 years agoWARNS=4
alfred [Thu, 11 Jul 2002 21:18:55 +0000 (21:18 +0000)]
WARNS=4

22 years agoWARNS=4 cleanup, de-__P()
alfred [Thu, 11 Jul 2002 21:16:52 +0000 (21:16 +0000)]
WARNS=4 cleanup, de-__P()

22 years agode-__P()
alfred [Thu, 11 Jul 2002 21:13:37 +0000 (21:13 +0000)]
de-__P()

22 years agoAdd a bit of information on FFS snapshots to the mount(8) manual page.
trhodes [Thu, 11 Jul 2002 21:12:45 +0000 (21:12 +0000)]
Add a bit of information on FFS snapshots to the mount(8) manual page.

PR: 30139
Submitted by: Chris Knight <chris@aims.com.au?
MFC after: 3 days

22 years agoWARNS?=4 cleanup.
alfred [Thu, 11 Jul 2002 21:12:18 +0000 (21:12 +0000)]
WARNS?=4 cleanup.

22 years agoAdd note about the info documentation of diff(1).
keramida [Thu, 11 Jul 2002 21:11:26 +0000 (21:11 +0000)]
Add note about the info documentation of diff(1).
It contains a detailed description of the =format option part.

PR: docs/36464
Submitted by: Gary W. Swearingen <swear@blarg.net>

22 years agode-__P()
alfred [Thu, 11 Jul 2002 20:53:56 +0000 (20:53 +0000)]
de-__P()

22 years agoFixed misspelling of "hint." as "hints." in rp's hints.
bde [Thu, 11 Jul 2002 20:47:56 +0000 (20:47 +0000)]
Fixed misspelling of "hint." as "hints." in rp's hints.

22 years agoFixed misspelling of "hint." as "hints." in the description of the "hint."
bde [Thu, 11 Jul 2002 20:43:37 +0000 (20:43 +0000)]
Fixed misspelling of "hint." as "hints." in the description of the "hint."
keyword and in the description of rp's hints.

Didn't fix rp's hints being mostly in comments so that they are harder to
use (they don't get linted either way because makeLINT.sh strips them and
there is no compile-time syntax checking of hints anyway).

22 years agoRELENG_4_6 PL 1 libc bits with non-vulnerable resolver.
obrien [Thu, 11 Jul 2002 20:40:24 +0000 (20:40 +0000)]
RELENG_4_6 PL 1 libc bits with non-vulnerable resolver.

Security Advisory: FreeBSD-SA-02:28.resolv

22 years agoImproved wording in the descriptions of hints. In particular, don't start
bde [Thu, 11 Jul 2002 20:24:58 +0000 (20:24 +0000)]
Improved wording in the descriptions of hints.  In particular, don't start
a new paragraph for describing the hints associated with each card type.

Reviewed by: ru

22 years agoMention zegrep and zfgrep too,
keramida [Thu, 11 Jul 2002 20:12:15 +0000 (20:12 +0000)]
Mention zegrep and zfgrep too,
and what their equivalent grep options are.

PR: docs/35603
Submitted by: "Gary W. Swearingen" <swear@blarg.net>

22 years agoUse the length modifier 'l' when supplying an argument of
robert [Thu, 11 Jul 2002 20:10:07 +0000 (20:10 +0000)]
Use the length modifier 'l' when supplying an argument of
type long to printf(3).

22 years ago - Remove unnecessary inclusion of <utmp.h>.
robert [Thu, 11 Jul 2002 20:01:36 +0000 (20:01 +0000)]
 - Remove unnecessary inclusion of <utmp.h>.
 - Cast a value to time_t before comparing it to another
   time_t to fix a warning.

22 years agode-__P()
alfred [Thu, 11 Jul 2002 19:07:35 +0000 (19:07 +0000)]
de-__P()

22 years ago o Lock accesses to the page queues.
alc [Thu, 11 Jul 2002 18:48:05 +0000 (18:48 +0000)]
 o Lock accesses to the page queues.

22 years agomark unused variable.
alfred [Thu, 11 Jul 2002 18:45:59 +0000 (18:45 +0000)]
mark unused variable.

22 years agoconditionally set WARNS via ?=.
alfred [Thu, 11 Jul 2002 18:45:14 +0000 (18:45 +0000)]
conditionally set WARNS via ?=.

22 years agofixup for WARNS=4.
alfred [Thu, 11 Jul 2002 18:42:53 +0000 (18:42 +0000)]
fixup for WARNS=4.

22 years agode-__P()
alfred [Thu, 11 Jul 2002 18:31:16 +0000 (18:31 +0000)]
de-__P()

22 years agoSet NO_WERROR to ignore the following warning which is emitted on
bde [Thu, 11 Jul 2002 18:06:09 +0000 (18:06 +0000)]
Set NO_WERROR to ignore the following warning which is emitted on
alphas:
    .../elf2aout.c:130: warning: cast increases required alignment of
    target type
The warning is about casting ((char *)e + phoff) to a struct pointer,
where e is aligned but phoff might be garbage, so I think the warning
should be emitted on most machines (even on i386's, alignment checking
might be on) and the correct fix would involve validation phoff before
using it.

22 years agomake mostly WARNS=4 clean.
alfred [Thu, 11 Jul 2002 17:55:14 +0000 (17:55 +0000)]
make mostly WARNS=4 clean.
constify, mark unsued args, fixup prototypes.

22 years agoConvert old style (type foo *)0 casts to NULLs
dillon [Thu, 11 Jul 2002 17:54:58 +0000 (17:54 +0000)]
Convert old style (type foo *)0 casts to NULLs

PR: kern/40360
Requested by: Hiten PAndya via direct email

22 years agoFixed 4 printf format errors that were fatal on alphas. %qd is not even
bde [Thu, 11 Jul 2002 17:49:41 +0000 (17:49 +0000)]
Fixed 4 printf format errors that were fatal on alphas.  %qd is not even
suitable for printing quad_t's since it is equivalent to %lld but quad_t
is unsigned long on alphas.  quad_t shouldn't be used anyway.

22 years agoLog the IP of the machine if an invalid host is passed over the wire.
alfred [Thu, 11 Jul 2002 17:44:59 +0000 (17:44 +0000)]
Log the IP of the machine if an invalid host is passed over the wire.

PR: bin/29406

22 years agoFix return values in the sm_notify_1 service routine to return an answer
alfred [Thu, 11 Jul 2002 17:39:50 +0000 (17:39 +0000)]
Fix return values in the sm_notify_1 service routine to return an answer
most of the time (unless fork fails).  This should fix the problem where
FreeBSD won't respond to a remote host and therefor the remote hosts
tries indefinitely to contact the FreeBSD hosts thereby irritating the
system administrator.

PR: misc/27810

22 years agoFix syslog format errors introduced in 1.6.
alfred [Thu, 11 Jul 2002 17:36:09 +0000 (17:36 +0000)]
Fix syslog format errors introduced in 1.6.

22 years agoadd includes for inet_ntoa prototype.
alfred [Thu, 11 Jul 2002 17:33:47 +0000 (17:33 +0000)]
add includes for inet_ntoa prototype.

22 years agoUncommented WARNS=0. ipfw2.c is full of printf format errors that are
bde [Thu, 11 Jul 2002 17:33:37 +0000 (17:33 +0000)]
Uncommented WARNS=0.  ipfw2.c is full of printf format errors that are
fatal on alphas.

Fixed setting of WARNS.  WARNS should never be set unconditionally, since
this breaks testing of different WARNS values by setting it at a higher
level (e.g., on the command line).

22 years agoinclude string.h for strcmp(3) prototype.
alfred [Thu, 11 Jul 2002 17:32:16 +0000 (17:32 +0000)]
include string.h for strcmp(3) prototype.

22 years agoinclude stdlib.h for exit(3) prototype.
alfred [Thu, 11 Jul 2002 17:31:15 +0000 (17:31 +0000)]
include stdlib.h for exit(3) prototype.

22 years agoFixed a printf format error that was fatal on alphas. Adding WFORMAT=0
bde [Thu, 11 Jul 2002 17:28:29 +0000 (17:28 +0000)]
Fixed a printf format error that was fatal on alphas.  Adding WFORMAT=0
to the Makefile didn't affect this bug because WFORMAT only controls
higher- level format checking (not the -Wformat that is implicit in
-Wall).

Fixed a nearby printf format error that was benign and 3 nearby style bugs.

22 years agoDo not allow unsafe characters in monitored hostnames.
alfred [Thu, 11 Jul 2002 17:11:42 +0000 (17:11 +0000)]
Do not allow unsafe characters in monitored hostnames.

This is modified version of the fix posted in the PR, I've taken Wollman's
advice and used strvis(3) to properly show invalid hostnames.

PR: bin/39815

22 years agoem(4) update.
bmah [Thu, 11 Jul 2002 17:08:07 +0000 (17:08 +0000)]
em(4) update.

22 years agoNew release notes: acct(2) appends to files, basename(1) -a/-s, GDB
bmah [Thu, 11 Jul 2002 17:03:36 +0000 (17:03 +0000)]
New release notes:  acct(2) appends to files, basename(1) -a/-s, GDB
5.2.

Updated release notes: em(4) update, ACPI CA 20020611,
mountd(8)/nfsd(8) moved to /usr/sbin, vidcontrol(1) -S.

MFCs noted:  dump(8) -S.

22 years agoPass -DBOOTSTRAPPING to the kernel's build tool (sys/dev/aic7xxx/aicasm).
ru [Thu, 11 Jul 2002 16:43:59 +0000 (16:43 +0000)]
Pass -DBOOTSTRAPPING to the kernel's build tool (sys/dev/aic7xxx/aicasm).
This way, it has a chance to be built with gcc 2.95.x (using the bandaid
in share/mk/bsd.cpu.mk,v 1.11).

Feedback timeout from: gordon

22 years agoDo not override the CFLAGS with its standard value. At some point,
ru [Thu, 11 Jul 2002 16:27:04 +0000 (16:27 +0000)]
Do not override the CFLAGS with its standard value.  At some point,
different architectures may choose to use different default values
for CFLAGS, for example.  (It was added in rev. 1.200 as a measure
to make boot images fit the floppies, and was never reverted.)

22 years agoAdd functions allowing for the user to specify a timeout for rpc functions.
alfred [Thu, 11 Jul 2002 16:23:04 +0000 (16:23 +0000)]
Add functions allowing for the user to specify a timeout for rpc functions.

Update copyrights to reflect where this code was lifted from. (tirpc '99)

Submitted by: mbr

22 years agoAdd -h option to rpcbind, used to specify what address to bind to for
alfred [Thu, 11 Jul 2002 16:19:43 +0000 (16:19 +0000)]
Add -h option to rpcbind, used to specify what address to bind to for
UDP requests.

Submitted by: mbr

22 years agoAdd missing ret instruction to the ptrace() syscall wrapper.
tmm [Thu, 11 Jul 2002 15:48:02 +0000 (15:48 +0000)]
Add missing ret instruction to the ptrace() syscall wrapper.

22 years agoDo not override the standard `distribute' target that is currently
ru [Thu, 11 Jul 2002 14:13:37 +0000 (14:13 +0000)]
Do not override the standard `distribute' target that is currently
available from bsd.obj.mk.

The native version was identical (and pretty much unused except in
the -DMODULES_WITH_WORLD case, which it is not for "make release")
except that the "bin" -> "base" change of the default DISTRIBUTION
name did not propagate here.

22 years agoUpdate the SUBDIR hacks to work with the current bsd.subdir.mk.
ru [Thu, 11 Jul 2002 13:38:05 +0000 (13:38 +0000)]
Update the SUBDIR hacks to work with the current bsd.subdir.mk.

22 years agoRemoved the (never used) help-distribute target from here.
ru [Thu, 11 Jul 2002 13:31:52 +0000 (13:31 +0000)]
Removed the (never used) help-distribute target from here.

(Similar targets were once used during the release building
process for kerberosIV and kerberos5.)

22 years agoUse realhostname_sa(3) so the IP address will be used instead of the
des [Thu, 11 Jul 2002 10:36:10 +0000 (10:36 +0000)]
Use realhostname_sa(3) so the IP address will be used instead of the
hostname if the latter is too long for utmp.

Submitted by: ru
MFC after: 3 days

22 years agoSet WFORMAT=0, because our compiler currently complains about NULL
sheldonh [Thu, 11 Jul 2002 08:45:38 +0000 (08:45 +0000)]
Set WFORMAT=0, because our compiler currently complains about NULL
arg 2 to err() and friends, and warnings are promoted to errors.

This allows the following revisions to be reverted:

rev 1.39 src/bin/cp/cp.c
rev 1.26 src/bin/chmod/chmod.c
rev 1.40 src/bin/rm/rm.c

The following revisions can already be reverted, because they were
already covered by WFORMAT=0:

rev 1.8 src/bin/ls/lomac.c
rev 1.63 src/bin/ls/ls.c
rev 1.8 src/bin/ps/lomac.c
rev 1.34 src/bin/rcp/rcp.c

22 years agoBah, move the invltlb counter to C code and hook a debug sysctl onto it.
peter [Thu, 11 Jul 2002 08:31:10 +0000 (08:31 +0000)]
Bah, move the invltlb counter to C code and hook a debug sysctl onto it.

22 years agos/NCPU/MAXCPU/ to try and get this to compile.
peter [Thu, 11 Jul 2002 08:24:33 +0000 (08:24 +0000)]
s/NCPU/MAXCPU/ to try and get this to compile.

22 years agoBring getconf back in from the cold.
peter [Thu, 11 Jul 2002 07:28:21 +0000 (07:28 +0000)]
Bring getconf back in from the cold.

22 years agoRemoved no longer used share/examples/diskless/* dirs (forgotten
ru [Thu, 11 Jul 2002 07:15:36 +0000 (07:15 +0000)]
Removed no longer used share/examples/diskless/* dirs (forgotten
by luigi@) and never used share/examples/kld/dyn_sysctl/module.

22 years agoDon't assume the shell's controlling terminal is attached to file descriptor
tjr [Thu, 11 Jul 2002 06:42:11 +0000 (06:42 +0000)]
Don't assume the shell's controlling terminal is attached to file descriptor
2. Instead, open /dev/tty. This problem stopped commands in subshells from
being executed correctly if standard error was redirected.

PR: 36671
Obtained from: NetBSD (but simplified)

22 years agoRemove support for the "old" tty driver by unifdef -UOLD_TTY_DRIVER;
tjr [Thu, 11 Jul 2002 06:26:48 +0000 (06:26 +0000)]
Remove support for the "old" tty driver by unifdef -UOLD_TTY_DRIVER;
many other parts of the shell are no longer compatible with this, and it
makes jobs.c quite cluttered with #ifdef's.

22 years agoWhen growing the job table, don't relocate the jobmru pointer if it's NULL.
tjr [Thu, 11 Jul 2002 04:22:41 +0000 (04:22 +0000)]
When growing the job table, don't relocate the jobmru pointer if it's NULL.

22 years agoMove the MSIZE and MCLSHIFT options out of the undocumented section in
ken [Thu, 11 Jul 2002 04:15:53 +0000 (04:15 +0000)]
Move the MSIZE and MCLSHIFT options out of the undocumented section in
NOTES.  Add some comments about the potential problems associated with NIC
driver modules and changing these options.

Fix sorting problems in sys/conf/options with the MSIZE and MCLSHIFT
options.

Reviewed by: bde

22 years agoRemove a couple of debugging lines.
mjacob [Thu, 11 Jul 2002 03:27:30 +0000 (03:27 +0000)]
Remove a couple of debugging lines.

22 years agoEnable ISP SBus support.
mjacob [Thu, 11 Jul 2002 03:26:39 +0000 (03:26 +0000)]
Enable ISP SBus support.

22 years ago'Support' for ISP SBus cards.
mjacob [Thu, 11 Jul 2002 03:25:04 +0000 (03:25 +0000)]
'Support' for ISP SBus cards.

This code does not imply that SBus cards work yet. They hang for me.
But I can't netboot the latest snapshot on my ultra1e, and things
hang at bus_setup_intr time.

Since I'm offline for a while, I thought I'd toss this in in case somebody
else who has a bit better luck wants to fart around with it. Please try
and wait until I get back to check things in.

22 years agoDisable format checking for NULL-passing prototypes (really only err*(),
dillon [Thu, 11 Jul 2002 02:51:02 +0000 (02:51 +0000)]
Disable format checking for NULL-passing prototypes (really only err*(),
warn*(), and setproctitle() functions) to buildworld work again.  This
can be cleaned up later if/when a new GCC supports the feature (but personally
I think it's a waste of time to keep mod'ing imported GCC sources for this
since only three procedures are involved).

Suggested by: peter

22 years ago o Add a "needs wakeup" flag to the vm_map for use by kmem_alloc_wait()
alc [Thu, 11 Jul 2002 02:39:24 +0000 (02:39 +0000)]
 o Add a "needs wakeup" flag to the vm_map for use by kmem_alloc_wait()
   and kmem_free_wakeup().  Previously, kmem_free_wakeup() always
   called wakeup().  In general, no one was sleeping.
 o Export vm_map_unlock_and_wait() and vm_map_wakeup() from vm_map.c
   for use in vm_kern.c.

22 years agoRevert removal of cred_free_thread(): It is used to ensure that a thread's
mini [Thu, 11 Jul 2002 02:18:33 +0000 (02:18 +0000)]
Revert removal of cred_free_thread(): It is used to ensure that a thread's
credentials are not improperly borrowed when the thread is not current in
the kernel.

Requested by: jhb, alfred

22 years agoAdd SBus firmware (compiled in only for sparc systems).
mjacob [Thu, 11 Jul 2002 01:55:15 +0000 (01:55 +0000)]
Add SBus firmware (compiled in only for sparc systems).

22 years agoDo not try to use PAM for password authentication, as it is
des [Wed, 10 Jul 2002 23:05:13 +0000 (23:05 +0000)]
Do not try to use PAM for password authentication, as it is
already (and far better) supported by the challenge/response
authentication mechanism.

22 years agoDon't forget to clear the buffer before reusing it.
des [Wed, 10 Jul 2002 23:04:07 +0000 (23:04 +0000)]
Don't forget to clear the buffer before reusing it.

22 years agofix a comment and note a problem with XXXSMP
julian [Wed, 10 Jul 2002 21:17:23 +0000 (21:17 +0000)]
fix a comment and note a problem with XXXSMP

22 years agoThis file has been included en-mass into i386/i386/exception.s
julian [Wed, 10 Jul 2002 21:07:47 +0000 (21:07 +0000)]
This file has been included en-mass into i386/i386/exception.s

22 years agoerr() is documented as allowing NULL for the format string but GCC isn't
dillon [Wed, 10 Jul 2002 20:44:55 +0000 (20:44 +0000)]
err() is documented as allowing NULL for the format string but GCC isn't
happy about it any more so change the usage to make buildworld work again.

22 years agoerr() is documented as allowing NULL for the format string but GCC isn't
dillon [Wed, 10 Jul 2002 20:36:25 +0000 (20:36 +0000)]
err() is documented as allowing NULL for the format string but GCC isn't
happy about it any more so change the useage so buildworld works again.

22 years agoRemove the critmode sysctl - the new method for critical_enter/exit (already
dillon [Wed, 10 Jul 2002 20:15:58 +0000 (20:15 +0000)]
Remove the critmode sysctl - the new method for critical_enter/exit (already
the default) is now the only method for i386.

Remove the paraphanalia that supported critmode.  Remove td_critnest, clean
up the assembly, and clean up (mostly remove) the old junk from
cpu_critical_enter() and cpu_critical_exit().

22 years agoConsistently line-up /**/ comments so they don't cause line wrappage.
obrien [Wed, 10 Jul 2002 19:36:04 +0000 (19:36 +0000)]
Consistently line-up /**/ comments so they don't cause line wrappage.

22 years agoOpen accounting file for appending, not general writing.
johan [Wed, 10 Jul 2002 17:31:58 +0000 (17:31 +0000)]
Open accounting file for appending, not general writing.
This allows accton(1) to be used with an append-only file.

PR: 7169
Reported by: Joao Carlos Mendes Luis <jonny@jonny.eng.br>
Reviewed by: bde
Approved by: sheldonh (mentor)
MFC after: 2 weeks

22 years agoAdd PCI ID and configuration hint for Netmos Nm9845 PCI 6-ports serial
sobomax [Wed, 10 Jul 2002 17:26:11 +0000 (17:26 +0000)]
Add PCI ID and configuration hint for Netmos Nm9845 PCI 6-ports serial
card.

Submitted by: Oleg Antoshin <oleh@at2000.kiev.ua>
MFC after: 3 days

22 years agoReplace the global buffer hash table with per-vnode splay trees using a
dillon [Wed, 10 Jul 2002 17:02:32 +0000 (17:02 +0000)]
Replace the global buffer hash table with per-vnode splay trees using a
methodology similar to the vm_map_entry splay and the VM splay that Alan
Cox is working on.  Extensive testing has appeared to have shown no
increase in overhead.

Disadvantages
    Dirties more cache lines during lookups.

    Not as fast as a hash table lookup (but still N log N and optimal
    when there is locality of reference).

Advantages
    vnode->v_dirtyblkhd is now perfectly sorted, making fsync/sync/filesystem
    syncer operate more efficiently.

    I get to rip out all the old hacks (some of which were mine) that tried
    to keep the v_dirtyblkhd tailq sorted.

    The per-vnode splay tree should be easier to lock / SMPng pushdown on
    vnodes will be easier.

    This commit along with another that Alan is working on for the VM page
    global hash table will allow me to implement ranged fsync(), optimize
    server-side nfs commit rpcs, and implement partial syncs by the
    filesystem syncer (aka filesystem syncer would detect that someone is
    trying to get the vnode lock, remembers its place, and skip to the
    next vnode).

Note that the buffer cache splay is somewhat more complex then other splays
due to special handling of background bitmap writes (multiple buffers with
the same lblkno in the same vnode), and B_INVAL discontinuities between the
old hash table and the existence of the buffer on the v_cleanblkhd list.

Suggested by: alc

22 years ago- fixed the order of searching SA table for packets.
ume [Wed, 10 Jul 2002 16:39:38 +0000 (16:39 +0000)]
- fixed the order of searching SA table for packets.
- comment about deletion of SA that has not been used by reaching
  soft lifetime.

Obtained from: KAME
MFC after: 2 week

22 years agoFix some comments.
wollman [Wed, 10 Jul 2002 16:35:02 +0000 (16:35 +0000)]
Fix some comments.

22 years agoRename 'usio' to 'ucom'.
joe [Wed, 10 Jul 2002 16:07:33 +0000 (16:07 +0000)]
Rename 'usio' to 'ucom'.

Spotted by: akiyama

22 years agoRemove label_t and physadr, which seem to have never been used in
mike [Wed, 10 Jul 2002 15:47:59 +0000 (15:47 +0000)]
Remove label_t and physadr, which seem to have never been used in
FreeBSD.

Submitted by: bde

22 years agoExpress my interest about being notified for fdc(4) and sppp(4) changes.
joerg [Wed, 10 Jul 2002 14:51:48 +0000 (14:51 +0000)]
Express my interest about being notified for fdc(4) and sppp(4) changes.

This file is getting unparseable.  IMHO we should at least sort it by
maintainer, or even better by subsystem.  Opinions?

22 years agoAdd setjmp (needed for DDB).
benno [Wed, 10 Jul 2002 12:26:17 +0000 (12:26 +0000)]
Add setjmp (needed for DDB).

22 years agoAdd DDB support.
benno [Wed, 10 Jul 2002 12:21:54 +0000 (12:21 +0000)]
Add DDB support.

22 years ago- Make sure we don't trample our metadata pointer in our initial bootstrap.
benno [Wed, 10 Jul 2002 12:16:48 +0000 (12:16 +0000)]
- Make sure we don't trample our metadata pointer in our initial bootstrap.
- Load metadata parameters.

22 years agoMetadata definitions.
benno [Wed, 10 Jul 2002 12:14:23 +0000 (12:14 +0000)]
Metadata definitions.

22 years agoAdd support for passing metadata.
benno [Wed, 10 Jul 2002 12:13:16 +0000 (12:13 +0000)]
Add support for passing metadata.

22 years agoIncorporate changes made to the NetBSD version of this driver.
benno [Wed, 10 Jul 2002 10:24:23 +0000 (10:24 +0000)]
Incorporate changes made to the NetBSD version of this driver.

- Remove some obsolete code (NetBSD gem.c r1.12)
- Clean up how the local MAC address is programmed (NetBSD gem.c r1.13)
- Make the driver work on PowerMacs with gigabit interfaces
  (NetBSD gem.c r1.14 and r1.15, gemreg.h r1.3 and r1.4, gemvar.h r1.6 and 1.7)
- Suppress RX_MAC interrutps regarding the FRAME_COUNT register.
  (NetBSD gem.c r1.16 and r1.17)
- Fix receiver lockups. (NetBSD gem.c r1.18, gemvar.h r1.8)
- Distinguish between Apple and Sun variants (NetBSD if_gem_pci.c r1.9)

Reviewed by: tmm
Obtained from: NetBSD

22 years agoRemove some diagnostic code that snuck in.
benno [Wed, 10 Jul 2002 09:48:43 +0000 (09:48 +0000)]
Remove some diagnostic code that snuck in.

22 years agoRemove some dead code.
benno [Wed, 10 Jul 2002 09:46:49 +0000 (09:46 +0000)]
Remove some dead code.

22 years agoRemove some unused includes.
benno [Wed, 10 Jul 2002 09:46:24 +0000 (09:46 +0000)]
Remove some unused includes.

22 years agoFix a typo.
maxim [Wed, 10 Jul 2002 09:06:42 +0000 (09:06 +0000)]
Fix a typo.

MFC after: 3 days

22 years agoMake our i386 hacks in a less vendor file disturbing manner.
obrien [Wed, 10 Jul 2002 07:48:45 +0000 (07:48 +0000)]
Make our i386 hacks in a less vendor file disturbing manner.