]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoDeclare variables as static, like I intended.
marcel [Mon, 25 Oct 2004 03:39:31 +0000 (03:39 +0000)]
Declare variables as static, like I intended.

Obtained from: Dan Markarian <markarian at apple dot com>

19 years agoFix bogon. Create the BSD label in the FreeBSD slice.
marcel [Mon, 25 Oct 2004 03:29:21 +0000 (03:29 +0000)]
Fix bogon. Create the BSD label in the FreeBSD slice.

19 years agoCatch up with the BSD label changes. The partition offsets are relative
marcel [Mon, 25 Oct 2004 03:27:31 +0000 (03:27 +0000)]
Catch up with the BSD label changes. The partition offsets are relative
to the partition now. While here, make sure we skip unused partition
entries.

19 years agoBuild gpt(8) on all platforms.
marcel [Mon, 25 Oct 2004 02:25:17 +0000 (02:25 +0000)]
Build gpt(8) on all platforms.

19 years agoAdd byte swapping and UUID encoding/decoding to allow gpt to be compiled
marcel [Mon, 25 Oct 2004 02:23:39 +0000 (02:23 +0000)]
Add byte swapping and UUID encoding/decoding to allow gpt to be compiled
on big-endian machines.

Obtained from: Dan Markarian <markarian at apple dot com>

19 years agoAssert that the containing vm object is locked in vm_page_busy() and
alc [Sun, 24 Oct 2004 23:53:47 +0000 (23:53 +0000)]
Assert that the containing vm object is locked in vm_page_busy() and
vm_page_wakeup().

19 years agoMove from using the socket reference count to the file reference
rwatson [Sun, 24 Oct 2004 23:45:01 +0000 (23:45 +0000)]
Move from using the socket reference count to the file reference
count to prevent sockets from being garbage collected during
socket-specific system calls.  This is the same approach used in
most VFS-specific system calls, as well as generic file descriptor
system calls such as read() and write().

To do this, add a utility function getsock(), which is logically
identical to getvnode() used for the same purpose in VFS.  Unlike
fgetsock(), it returns with the file reference count elevated, but
no bump of the socket reference count.  Replace matching calls to
fputsock() with fdrop().

This change is made to all socket system calls other than
sendfile() and accept(), but the approach should be applicable to
those system calls also.

This shaves about four mutex operations off of each of these
system calls, including send() and recv() variants, adding about
1% to pps on minimal UDP packets for UP using netblast, and 4% on
SMP.

Reviewed by: pjd

19 years agoLog the actual number of bytes sent on the wire to /var/log/ftpd
yar [Sun, 24 Oct 2004 20:12:08 +0000 (20:12 +0000)]
Log the actual number of bytes sent on the wire to /var/log/ftpd
instead of the disk size of the file sent.   Since the log file
is intended to provide data for anonymous ftp traffic accounting,
the disk size of the file isn't really informative in this case.

PR: bin/72687
Submitted by: Oleg Koreshkov
MFC after: 1 week

19 years agoUse VM_ALLOC_NOBUSY instead of calling vm_page_wakeup().
alc [Sun, 24 Oct 2004 20:09:59 +0000 (20:09 +0000)]
Use VM_ALLOC_NOBUSY instead of calling vm_page_wakeup().

19 years agoMove the buffer method vector (buf->b_op) to the bufobj.
phk [Sun, 24 Oct 2004 20:03:41 +0000 (20:03 +0000)]
Move the buffer method vector (buf->b_op) to the bufobj.

Extend it with a strategy method.

Add bufstrategy() which do the usual VOP_SPECSTRATEGY/VOP_STRATEGY
song and dance.

Rename ibwrite to bufwrite().

Move the two NFS buf_ops to more sensible places, add bufstrategy
to them.

Add inlines for bwrite() and bstrategy() which calls through
buf->b_bufobj->b_ops->b_{write,strategy}().

Replace almost all VOP_STRATEGY()/VOP_SPECSTRATEGY() calls with bstrategy().

19 years agoMention fdescfs(5) and why it might be useful.
rwatson [Sun, 24 Oct 2004 19:44:23 +0000 (19:44 +0000)]
Mention fdescfs(5) and why it might be useful.

MFC after: 1 week
Suggested by: Stephen Montgomery-Smith <stephen at math dot missouri dot edu>

19 years agoFix a panic discovered with some apache2 configure test (that seemed to
mlaier [Sun, 24 Oct 2004 19:35:02 +0000 (19:35 +0000)]
Fix a panic discovered with some apache2 configure test (that seemed to
trigger a socket creation race some some kind). Checking for non-NULL socket
and credential is not a bad idea anyway. Unfortunatly too late for the
release.

Reported & tested by: Gilbert Cao
MFC after: 2 weeks

19 years agoAcquire the vm object lock before rather than after calling
alc [Sun, 24 Oct 2004 19:32:19 +0000 (19:32 +0000)]
Acquire the vm object lock before rather than after calling
vm_page_sleep_if_busy().  (The motivation being to transition
synchronization of the vm_page's PG_BUSY flag from the global page queues
lock to the per-object lock.)

19 years agoUse VM_ALLOC_NOBUSY instead of calling vm_page_wakeup().
alc [Sun, 24 Oct 2004 18:46:32 +0000 (18:46 +0000)]
Use VM_ALLOC_NOBUSY instead of calling vm_page_wakeup().

19 years agoFor variables that are only checked with defined(), don't provide
ru [Sun, 24 Oct 2004 16:20:59 +0000 (16:20 +0000)]
For variables that are only checked with defined(), don't provide
any fake value.

19 years agoOnly check if ENABLE_SUID_NEWGRP is defined, for consistency with
ru [Sun, 24 Oct 2004 16:02:38 +0000 (16:02 +0000)]
Only check if ENABLE_SUID_NEWGRP is defined, for consistency with
other ENABLE_SUID_* variables.

19 years agoFor variables that are only checked with defined(), don't provide
ru [Sun, 24 Oct 2004 15:52:04 +0000 (15:52 +0000)]
For variables that are only checked with defined(), don't provide
any fake value.

While here, sort NO*'s in dictionary order to ease searching by a
human being.

19 years agoFor variables that are only checked with defined(), don't provide
ru [Sun, 24 Oct 2004 15:46:50 +0000 (15:46 +0000)]
For variables that are only checked with defined(), don't provide
any fake value.

19 years agoFor variables that are only checked with defined(), don't provide
ru [Sun, 24 Oct 2004 15:33:08 +0000 (15:33 +0000)]
For variables that are only checked with defined(), don't provide
any fake value.

19 years agoThere's no such knob as WERROR.
ru [Sun, 24 Oct 2004 14:55:35 +0000 (14:55 +0000)]
There's no such knob as WERROR.

19 years agoAdd a Japanese override of dev-auto.sgml.
hrs [Sun, 24 Oct 2004 14:31:27 +0000 (14:31 +0000)]
Add a Japanese override of dev-auto.sgml.

Based on work by: Ryusuke Suzuki (ryusuke at nucl.sci.hokudai.ac.jp)

19 years agoFixed misspelling of DPADD.
ru [Sun, 24 Oct 2004 14:25:53 +0000 (14:25 +0000)]
Fixed misspelling of DPADD.

19 years ago- use `realpath /dev/dumpdev` instead of just /dev/dumpdev so messages
des [Sun, 24 Oct 2004 13:04:09 +0000 (13:04 +0000)]
- use `realpath /dev/dumpdev` instead of just /dev/dumpdev so messages
  will show the real device name
- show different error messages for missing dump device and directory

19 years ago- Use quad_t for dcons buffer address and size.
simokawa [Sun, 24 Oct 2004 12:41:04 +0000 (12:41 +0000)]
- Use quad_t for dcons buffer address and size.
- Allow read/write access to dcons buffer passed by loader(8).

19 years agoPreserve dcons(4) buffer passed by loader(8).
simokawa [Sun, 24 Oct 2004 12:37:47 +0000 (12:37 +0000)]
Preserve dcons(4) buffer passed by loader(8).

19 years agoNEED_LIBNAMES is gone long ago.
ru [Sun, 24 Oct 2004 12:36:21 +0000 (12:36 +0000)]
NEED_LIBNAMES is gone long ago.

19 years agoCollapse several adjacent .if's into .if/.elif.
ru [Sun, 24 Oct 2004 12:32:41 +0000 (12:32 +0000)]
Collapse several adjacent .if's into .if/.elif.

19 years agoELF64 support is not needed on pc98.
nyan [Sun, 24 Oct 2004 12:18:11 +0000 (12:18 +0000)]
ELF64 support is not needed on pc98.

19 years agoMFi386: revision 1.4.
nyan [Sun, 24 Oct 2004 12:16:18 +0000 (12:16 +0000)]
MFi386: revision 1.4.

19 years agoMFi386: Set bi_basemem and bi_extmem variables.
nyan [Sun, 24 Oct 2004 12:15:07 +0000 (12:15 +0000)]
MFi386: Set bi_basemem and bi_extmem variables.
        Respect RB_MULTIPLE flag.

19 years agoMFi386: Add support for preloading ELF relocatable object format modules.
nyan [Sun, 24 Oct 2004 12:14:05 +0000 (12:14 +0000)]
MFi386: Add support for preloading ELF relocatable object format modules.

19 years agoDisable ed1 - ed12.
nyan [Sun, 24 Oct 2004 12:07:02 +0000 (12:07 +0000)]
Disable ed1 - ed12.

19 years agoMerged from src/sbin/fdisk/fdisk.c revision 1.75.
nyan [Sun, 24 Oct 2004 12:05:10 +0000 (12:05 +0000)]
Merged from src/sbin/fdisk/fdisk.c revision 1.75.
(Don't prepend "/dev" if an argument given is an absolute pathname.)

19 years agoAdd identifiers for the HP Laserjet 2300d USB
rse [Sun, 24 Oct 2004 11:16:29 +0000 (11:16 +0000)]
Add identifiers for the HP Laserjet 2300d USB
printer and the HP ScanJet 4670v USB scanner.

MFC after: 1 week

19 years ago-O2 compile isn't quite ready for WARNS=2 yet.
ru [Sun, 24 Oct 2004 10:42:15 +0000 (10:42 +0000)]
-O2 compile isn't quite ready for WARNS=2 yet.

19 years agoAllow multiple devices to be specified on one device / nodevice line.
des [Sun, 24 Oct 2004 10:27:16 +0000 (10:27 +0000)]
Allow multiple devices to be specified on one device / nodevice line.
Also allow "device" / "nodevice" to be spelled "devices" / "nodevices".

MFC after: 2 weeks

19 years agoUpdate the family tree for FreeBSD 5.3
scottl [Sun, 24 Oct 2004 10:09:20 +0000 (10:09 +0000)]
Update the family tree for FreeBSD 5.3

19 years agoIf the file specified in an "include" line does not exist in the current
des [Sun, 24 Oct 2004 09:38:41 +0000 (09:38 +0000)]
If the file specified in an "include" line does not exist in the current
directory, and its name does not begin with a period or a forward slash,
go look for it in ../../conf.

Wished for by: scottl
MFC after: 2 weeks

19 years agoFix some warnings that only triggered in LINT.
scottl [Sun, 24 Oct 2004 09:23:07 +0000 (09:23 +0000)]
Fix some warnings that only triggered in LINT.

19 years agoUpdate for 5-STABLE.
scottl [Sun, 24 Oct 2004 09:07:06 +0000 (09:07 +0000)]
Update for 5-STABLE.

19 years agoHook the hptmv driver up to the build.
scottl [Sun, 24 Oct 2004 08:53:40 +0000 (08:53 +0000)]
Hook the hptmv driver up to the build.

19 years agoEmbed the correct name.
scottl [Sun, 24 Oct 2004 08:48:32 +0000 (08:48 +0000)]
Embed the correct name.

19 years agoClean up some messy Makefile stuff so that this can be built into the
scottl [Sun, 24 Oct 2004 08:38:29 +0000 (08:38 +0000)]
Clean up some messy Makefile stuff so that this can be built into the
kernel.

19 years agoHack around a problem with sys/tools/usbdevs2h.awk that generates
ru [Sun, 24 Oct 2004 08:26:48 +0000 (08:26 +0000)]
Hack around a problem with sys/tools/usbdevs2h.awk that generates
both usbdevs.h and usbdevs_data.h.  (The latter was not cleaned.)

19 years agoCurlies only work in target and sources specifications, but
ru [Sun, 24 Oct 2004 07:54:44 +0000 (07:54 +0000)]
Curlies only work in target and sources specifications, but
not when passed to Bourne shell.  This unbreaks "make clean".

19 years agoUse VM_ALLOC_NOBUSY to eliminate an unneeded vm_page_wakeup() call and the
alc [Sun, 24 Oct 2004 07:31:07 +0000 (07:31 +0000)]
Use VM_ALLOC_NOBUSY to eliminate an unneeded vm_page_wakeup() call and the
synchronization that one entails.

19 years agoAvoid repeated acquisition and release of the vm object lock inside of
alc [Sun, 24 Oct 2004 07:12:13 +0000 (07:12 +0000)]
Avoid repeated acquisition and release of the vm object lock inside of
two loops in agp_generic_bind_memory().  As an intended side-effect, all
of the calls to vm_page_wakeup() are now performed with the containing
vm object lock held.

19 years agoSort SUBDIR.
ru [Sun, 24 Oct 2004 06:22:16 +0000 (06:22 +0000)]
Sort SUBDIR.

19 years agoIntroduce VM_ALLOC_NOBUSY, an option to vm_page_alloc() and vm_page_grab()
alc [Sun, 24 Oct 2004 06:15:36 +0000 (06:15 +0000)]
Introduce VM_ALLOC_NOBUSY, an option to vm_page_alloc() and vm_page_grab()
that indicates that the caller does not want a page with its busy flag set.
In many places, the global page queues lock is acquired and released just
to clear the busy flag on a just allocated page.  Both the allocation of
the page and the clearing of the busy flag occur while the containing vm
object is locked.  So, the busy flag might as well never be set.

19 years agoImport the HighPoint RocketRAID 182x driver. Thanks to HighPoint for
scottl [Sun, 24 Oct 2004 05:37:23 +0000 (05:37 +0000)]
Import the HighPoint RocketRAID 182x driver.  Thanks to HighPoint for
providing the original driver, and thanks to IronSystems for providing
hardware for testing.

19 years agoAdd a missing full-stop to match the rest of the items in a list.
keramida [Sun, 24 Oct 2004 00:26:35 +0000 (00:26 +0000)]
Add a missing full-stop to match the rest of the items in a list.

19 years agoCheck unhandled signals before thread marks itself as DEAD,
davidxu [Sat, 23 Oct 2004 23:37:54 +0000 (23:37 +0000)]
Check unhandled signals before thread marks itself as DEAD,
this reduces chances of signal losting problem found by
Peter Holm <peter@holm.cc>

19 years ago1. Move thread list flags into new separate member, and atomically
davidxu [Sat, 23 Oct 2004 23:28:36 +0000 (23:28 +0000)]
1. Move thread list flags into new separate member, and atomically
   put DEAD thread on GC list, this closes a race between pthread_join
   and thr_cleanup.
2. Introduce a mutex to protect tcb initialization, tls allocation and
   deallocation code in rtld seems no lock protection or it is broken,
   under stress testing, memory is corrupted.

Reviewed by: deischen
patch partly provided by: deischen

19 years agoAdd -m option, set default fwmem target.
simokawa [Sat, 23 Oct 2004 23:28:08 +0000 (23:28 +0000)]
Add -m option, set default fwmem target.

19 years agoUse errx() instead of fprintf()/exit() for conciseness.
rwatson [Sat, 23 Oct 2004 22:18:37 +0000 (22:18 +0000)]
Use errx() instead of fprintf()/exit() for conciseness.

Suggested by: ru (some time ago)

19 years agoUse errx() instead of perror()/exit() for conciseness.
rwatson [Sat, 23 Oct 2004 22:11:35 +0000 (22:11 +0000)]
Use errx() instead of perror()/exit() for conciseness.

Suggested by: ru (some time ago)

19 years agoAdd a regression test for the alternate shell specification.
ru [Sat, 23 Oct 2004 21:38:58 +0000 (21:38 +0000)]
Add a regression test for the alternate shell specification.

19 years ago"unset verbose" is 13 characters, not 10.
ru [Sat, 23 Oct 2004 21:36:55 +0000 (21:36 +0000)]
"unset verbose" is 13 characters, not 10.

19 years agoExpand the scope of the .SHELL specification to also cover
ru [Sat, 23 Oct 2004 21:34:41 +0000 (21:34 +0000)]
Expand the scope of the .SHELL specification to also cover
the compat mode of operation and the != operator.

While here, fixed a bug in the .SHELL directive processing
when only the name= attribute is specified and no built-in
shell matches this name, causing null pointer dereference.

Obtained from: NetBSD (except for bugs)

19 years agoDon't call g_waitidle(), it happens automagically now.
phk [Sat, 23 Oct 2004 20:52:15 +0000 (20:52 +0000)]
Don't call g_waitidle(), it happens automagically now.

19 years agoDon't explicitly call g_waitidle(), it happens automagically now.
phk [Sat, 23 Oct 2004 20:50:06 +0000 (20:50 +0000)]
Don't explicitly call g_waitidle(), it happens automagically now.

19 years agoAdd a new per-thread private flag: TDP_GEOM.
phk [Sat, 23 Oct 2004 20:49:17 +0000 (20:49 +0000)]
Add a new per-thread private flag: TDP_GEOM.

This flag gets set whenever the thread posts an event on the GEOM
event queue, and if the flag is set when the thread is prepared
to return to userland from the kernel, g_waitidle() will be called
to make sure that the posted events have completed.

This can replace an insufficient number of g_waitidle() calls in
various other places, and has the advantage of being failsafe:  Any
system call which does a VOP_OPEN()/VOP_CLOSE will now correctly
wait for any geom events it posted as part of spoils or tastes.

Assert that topology and Giant is not held in g_waitidle().

19 years agoMove the prototype for g_waitidle() to a more visible place.
phk [Sat, 23 Oct 2004 20:22:02 +0000 (20:22 +0000)]
Move the prototype for g_waitidle() to a more visible place.

19 years agoDrop Giant around the call to g_waitidle().
phk [Sat, 23 Oct 2004 20:21:05 +0000 (20:21 +0000)]
Drop Giant around the call to g_waitidle().
This is necessary to allow any geom events which need it to pick up Giant.

19 years agoRebuild from FreeBSD32 syscalls.master:1.42.
rwatson [Sat, 23 Oct 2004 20:05:42 +0000 (20:05 +0000)]
Rebuild from FreeBSD32 syscalls.master:1.42.

19 years ago32-bit FreeBSD ABI compatibility stubs from syscalls.master:1.178.
rwatson [Sat, 23 Oct 2004 20:04:56 +0000 (20:04 +0000)]
32-bit FreeBSD ABI compatibility stubs from syscalls.master:1.178.

19 years agoRebuild from syscalls.master:1.178.
rwatson [Sat, 23 Oct 2004 20:01:32 +0000 (20:01 +0000)]
Rebuild from syscalls.master:1.178.

19 years agoAdd system call place-holders for the following system calls
rwatson [Sat, 23 Oct 2004 20:00:43 +0000 (20:00 +0000)]
Add system call place-holders for the following system calls
implementing Sun's BSM Audit API on FreeBSD:

  audit()
  auditon()
  getauid()
  setauid()
  getaudit()
  setaudit()
  getaudit_addr()
  setaudit_addr()
  auditctl()

Submitted by: Wayne Salamon <wsalamon at computer dot org>
Obtained from: TrustedBSD Project

19 years agoDue to space constraints, use gnome2-lite for i386 and gnome2 for everything
scottl [Sat, 23 Oct 2004 19:54:18 +0000 (19:54 +0000)]
Due to space constraints, use gnome2-lite for i386 and gnome2 for everything
else.  Also add ltmdm to i386 by popular request.

19 years agosocreate() does an early abort if either the protocol cannot be found,
andre [Sat, 23 Oct 2004 19:06:43 +0000 (19:06 +0000)]
socreate() does an early abort if either the protocol cannot be found,
or pru_attach is NULL.  With loadable protocols the SPACER dummy protocols
have valid function pointers for all methods to functions returning just
EOPNOTSUPP.  Thus the early abort check would not detect immediately that
attach is not supported for this protocol.  Instead it would correctly
get the EOPNOTSUPP error later on when it calls the protocol specific
attach function.

Add testing against the pru_attach_notsupp() function pointer to the
early abort check as well.

19 years agoAquire GIANT in pf_proto_[un]register() before manipulating the protosw.
andre [Sat, 23 Oct 2004 18:52:06 +0000 (18:52 +0000)]
Aquire GIANT in pf_proto_[un]register() before manipulating the protosw.

19 years agoo Remove entire section talking about RTF_PRCLONING that got nuked in 5.2
andre [Sat, 23 Oct 2004 18:45:53 +0000 (18:45 +0000)]
o Remove entire section talking about RTF_PRCLONING that got nuked in 5.2
  already.
o Better description of IP fastforwarding.  It can do the full thing now
  including firewalling.
o Reword the description of ICMP redirects.
o Add references to route(4) and pfil(9).

19 years agoRephrase certain centences and sync description of default values to reality.
andre [Sat, 23 Oct 2004 18:37:23 +0000 (18:37 +0000)]
Rephrase certain centences and sync description of default values to reality.

19 years agoBump document date that was forgotten in the last commit.
andre [Sat, 23 Oct 2004 18:35:35 +0000 (18:35 +0000)]
Bump document date that was forgotten in the last commit.

Reminded by: ru

19 years agoUse %zu to format size_t instead of %jd and an intmax_t cast.
des [Sat, 23 Oct 2004 15:58:50 +0000 (15:58 +0000)]
Use %zu to format size_t instead of %jd and an intmax_t cast.

Approved by: hamlet

19 years ago/%x/%s/ -- mismerged DEBUGGER() printf() format stirng from the
rwatson [Sat, 23 Oct 2004 15:12:34 +0000 (15:12 +0000)]
/%x/%s/ -- mismerged DEBUGGER() printf() format stirng from the
TrustedBSD branch.

Submitted by: bde

19 years agoMention pf.conf(5), pfctl(8) and pflogd(8) as ADDITIONAL READINGs.
mlaier [Sat, 23 Oct 2004 15:05:46 +0000 (15:05 +0000)]
Mention pf.conf(5), pfctl(8) and pflogd(8) as ADDITIONAL READINGs.

Requested by: Chris Laverdure <dashevil sympatico ca>

19 years agouse bioq_takefirst()
phk [Sat, 23 Oct 2004 12:45:39 +0000 (12:45 +0000)]
use bioq_takefirst()

19 years agoUse bioq_takefirst()
phk [Sat, 23 Oct 2004 12:44:19 +0000 (12:44 +0000)]
Use bioq_takefirst()

19 years agoStop amd64 warnings.
phk [Sat, 23 Oct 2004 12:42:18 +0000 (12:42 +0000)]
Stop amd64 warnings.

19 years agoRemove P_STOPPED_TRACE bit if debugger dies without a chance to
davidxu [Sat, 23 Oct 2004 11:20:26 +0000 (11:20 +0000)]
Remove P_STOPPED_TRACE bit if debugger dies without a chance to
detach debugged process.

19 years agoMake documentation up to date with recent changes.
glebius [Sat, 23 Oct 2004 11:14:30 +0000 (11:14 +0000)]
Make documentation up to date with recent changes.

Reviewed by: ru

19 years agoAdd some basic KTR tracing to busdma on i386. This is likely not
rwatson [Sat, 23 Oct 2004 10:34:27 +0000 (10:34 +0000)]
Add some basic KTR tracing to busdma on i386.  This is likely not
the final set of traces -- someone with more busdma background
will probably want to review and expand this, as well as port to
other platforms.  This tracing is sufficient to identify key
busdma events on i386, and in particular to draw attention to
bounce buffering events that may have a substantial performance
impact.

19 years agoLocking cleanups to remove the need for a recursive mutex
mtm [Sat, 23 Oct 2004 08:33:10 +0000 (08:33 +0000)]
Locking cleanups to remove the need for a recursive mutex
o Instead of locking and unlocking all over the place, use
  lock assertions to make certain that the bfe lock is held
  where necessary.
o Create locked and unlocked versions of bfe_init and bfe_start. These
  functions can be called from outside the module and by functions
  within the bfe module. The calls from outside the module don't
  hold the bfe lock so the unlocked versions called by these functions
  simple obtain the bfe lock and call the locked version.

- Fix a typo (scp) in the locking macros that only worked because in all the
  instances in which it was called the softc pointer happened to be named 'sc'.

- Mark the interrupt MPSAFE

Tested by: matusita, Dario Freni <saturnero@gufi.org>
Silence from: -net, wpaul

19 years agoMove devfs earlier in the boot sequence. Some system daemons and other
mtm [Sat, 23 Oct 2004 06:50:50 +0000 (06:50 +0000)]
Move devfs earlier in the boot sequence. Some system daemons and other
programs may need to use the symlinks and permissions that it sets up.

Discussed on: -current

19 years agoIPDIVERT will automatically load ipdivert.ko if not compiled into the
andre [Fri, 22 Oct 2004 23:07:52 +0000 (23:07 +0000)]
IPDIVERT will automatically load ipdivert.ko if not compiled into the
kernel.

19 years ago- Turn KASSERT()s into warning printf()'s in the g_class_load() routine.
arr [Fri, 22 Oct 2004 22:16:24 +0000 (22:16 +0000)]
- Turn KASSERT()s into warning printf()'s in the g_class_load() routine.
  This removes a panic that will occur if you build with GENERIC and
  attempt to kldload a GEOM module that is already in the kernel.

Reviewed by: phk

19 years agoImprove accuracy of last addition, as best Google advices me.
phk [Fri, 22 Oct 2004 19:59:37 +0000 (19:59 +0000)]
Improve accuracy of last addition, as best Google advices me.

19 years agoShave 40 unused bytes from struct tcpcb.
andre [Fri, 22 Oct 2004 19:55:04 +0000 (19:55 +0000)]
Shave 40 unused bytes from struct tcpcb.

19 years agoAutomatically load the ipdivert module if it was not compiled into the kernel
andre [Fri, 22 Oct 2004 19:36:03 +0000 (19:36 +0000)]
Automatically load the ipdivert module if it was not compiled into the kernel
and natd_enable is true.

19 years agoWhen printing the initialization string and IPDIVERT is not compiled into the
andre [Fri, 22 Oct 2004 19:18:06 +0000 (19:18 +0000)]
When printing the initialization string and IPDIVERT is not compiled into the
kernel refer to it as "loadable" instead of "disabled".

19 years agoRefuse to unload the ipdivert module unless the 'force' flag is given to kldunload.
andre [Fri, 22 Oct 2004 19:12:01 +0000 (19:12 +0000)]
Refuse to unload the ipdivert module unless the 'force' flag is given to kldunload.

Reflect the fact that IPDIVERT is a loadable module in the divert(4) and ipfw(8)
man pages.

19 years agoProperly handle failure to allocate isadma bounce buffer
phk [Fri, 22 Oct 2004 19:01:10 +0000 (19:01 +0000)]
Properly handle failure to allocate isadma bounce buffer

19 years agoRemove a "needs Giant" flag from the /dev/apm compat device.
njl [Fri, 22 Oct 2004 17:17:12 +0000 (17:17 +0000)]
Remove a "needs Giant" flag from the /dev/apm compat device.

MFC after: 2 weeks

19 years agoAdd punctuation which appears to have been accidentally removed in
cperciva [Fri, 22 Oct 2004 16:44:39 +0000 (16:44 +0000)]
Add punctuation which appears to have been accidentally removed in
revision 1.6.

19 years agoI'm surprised this one was not here yet.
phk [Fri, 22 Oct 2004 16:18:15 +0000 (16:18 +0000)]
I'm surprised this one was not here yet.

To save you all a trip to CVSWEB:

Experts have correctly predicted nine out of the last five recessions.
               -- Paul Samuelson,  Nobel laureate in economics.

19 years agoRespect _BOOT flag.
simokawa [Fri, 22 Oct 2004 15:39:39 +0000 (15:39 +0000)]
Respect _BOOT flag.

19 years agoCheck _BOOT flag.
simokawa [Fri, 22 Oct 2004 15:03:22 +0000 (15:03 +0000)]
Check _BOOT flag.

19 years agoRespect RB_MULTIPLE flag.
simokawa [Fri, 22 Oct 2004 14:57:28 +0000 (14:57 +0000)]
Respect RB_MULTIPLE flag.

19 years ago- Add FireWire subclass and OHCI interface.
simokawa [Fri, 22 Oct 2004 14:56:23 +0000 (14:56 +0000)]
- Add FireWire subclass and OHCI interface.
- Add some PCI BIOS function calls.
(find_devclass, read_config, write_config)