]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoSet HAVE_STRUCT_IFREQ_IFR_ADDR, HAVE_STRUCT_FHSTATUS_FHS_FH to 1.
mbr [Tue, 2 Sep 2003 15:56:05 +0000 (15:56 +0000)]
Set HAVE_STRUCT_IFREQ_IFR_ADDR, HAVE_STRUCT_FHSTATUS_FHS_FH to 1.

21 years agocosmetics
sos [Tue, 2 Sep 2003 15:53:01 +0000 (15:53 +0000)]
cosmetics

21 years agoResolve conflicts.
mbr [Tue, 2 Sep 2003 15:42:57 +0000 (15:42 +0000)]
Resolve conflicts.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r119679,
mbr [Tue, 2 Sep 2003 15:27:38 +0000 (15:27 +0000)]
This commit was generated by cvs2svn to compensate for changes in r119679,
which included commits to RCS files with non-trunk default branches.

21 years agoVirgin import of AMD (am-utils) 20030828 (6.0.9)
mbr [Tue, 2 Sep 2003 15:27:38 +0000 (15:27 +0000)]
Virgin import of AMD (am-utils) 20030828 (6.0.9)

21 years agoAdjust the max transfer size used.
sos [Tue, 2 Sep 2003 13:26:02 +0000 (13:26 +0000)]
Adjust the max transfer size used.

21 years agoImport ISC DHCP 3.0.1 RC12 client.
mbr [Tue, 2 Sep 2003 11:15:03 +0000 (11:15 +0000)]
Import ISC DHCP 3.0.1 RC12 client.

- s/include/includes
- remove already merged files

21 years agoResolve conflicts.
mbr [Tue, 2 Sep 2003 11:13:21 +0000 (11:13 +0000)]
Resolve conflicts.

21 years agoResolve conflicts.
mbr [Tue, 2 Sep 2003 11:10:29 +0000 (11:10 +0000)]
Resolve conflicts.

* rev. 1.22

21 years agoResolve conflicts:
mbr [Tue, 2 Sep 2003 11:09:45 +0000 (11:09 +0000)]
Resolve conflicts:

* rev. 1.33/1.25

21 years agoThis commit was generated by cvs2svn to compensate for changes in r119669,
mbr [Tue, 2 Sep 2003 11:01:27 +0000 (11:01 +0000)]
This commit was generated by cvs2svn to compensate for changes in r119669,
which included commits to RCS files with non-trunk default branches.

21 years agoImport ISC DHCP 3.0.1 RC12 client.
mbr [Tue, 2 Sep 2003 11:01:27 +0000 (11:01 +0000)]
Import ISC DHCP 3.0.1 RC12 client.

21 years agoCheck an arguments count before proceed in sysctl_handler().
maxim [Tue, 2 Sep 2003 10:36:40 +0000 (10:36 +0000)]
Check an arguments count before proceed in sysctl_handler().

PR: bin/56298
Submitted by: Kang Liu <liukang@bjpu.edu.cn>
MFC after: 2 weeks

# We need a regression test suit for ipfw(2)/ipfw(8) badly.

21 years agoSync with ISC repo. Use SIGTERM instead of SIGKILL, make oldpid a long and
mbr [Tue, 2 Sep 2003 09:09:53 +0000 (09:09 +0000)]
Sync with ISC repo. Use SIGTERM instead of SIGKILL, make oldpid a long and
cast it into a pid_t.

21 years agoPrepare for locking mlx(4) by cleaning up the use of busdma. No real
scottl [Tue, 2 Sep 2003 08:30:31 +0000 (08:30 +0000)]
Prepare for locking mlx(4) by cleaning up the use of busdma.  No real
functional changes should result from this.

21 years agoTeach rescue about NOATM, NO_VINUM and NOINET6
phk [Tue, 2 Sep 2003 06:43:58 +0000 (06:43 +0000)]
Teach rescue about NOATM, NO_VINUM and NOINET6

21 years agoDon't open with exclusive bit, swapon(8) wants to trash our swapdev.
phk [Tue, 2 Sep 2003 05:53:44 +0000 (05:53 +0000)]
Don't open with exclusive bit, swapon(8) wants to trash our swapdev.

Add XXX comment with a rating of this concept.

21 years agoCardBus instead of Cardbus. Add I/O Data WN-AG/CB and Prixim Skyline
imp [Mon, 1 Sep 2003 23:09:09 +0000 (23:09 +0000)]
CardBus instead of Cardbus.  Add I/O Data WN-AG/CB and Prixim Skyline
4030 cards.

Special Thanks: shigeru yamamoto-san for the kind donation of I/O Data and
other wireless cards

21 years agoSimplify the ioctl handling in GEOM.
phk [Mon, 1 Sep 2003 20:45:32 +0000 (20:45 +0000)]
Simplify the ioctl handling in GEOM.

This replaces the current ioctl processing with a direct call path
from geom_dev() where the ioctl arrives (from SPECFS) to any directly
connected GEOM class.

The inverse of the above is no longer supported.  This is the
situation were you have one or more intervening GEOM classes, for
instance a BSDlabel on top of a MBR or PC98.  If you want to issue
MBR or PC98 specific ioctls, you will need to issue them on a MBR
or PC98 providers.

This paves the way for inviting CD's, FD's and other special cases
inside GEOM.

21 years agoCommands submitted through the management interface won't have scatter/
scottl [Mon, 1 Sep 2003 20:44:18 +0000 (20:44 +0000)]
Commands submitted through the management interface won't have scatter/
gather lists.  Stop ignoring them and instead call the callback directly.
This unbreaks the management interface.

21 years agoChange clean_map from a global to an auto variable
eivind [Mon, 1 Sep 2003 16:46:47 +0000 (16:46 +0000)]
Change clean_map from a global to an auto variable

21 years agoDon't try to enable io or memory access for non-standard resource
dfr [Mon, 1 Sep 2003 15:01:49 +0000 (15:01 +0000)]
Don't try to enable io or memory access for non-standard resource
addresses. This stops resource allocations for e.g. amdpm failing - this
has its own special ways of enabling access.

21 years agoAdd support for AMD766 and AMD768 chipsets.
dfr [Mon, 1 Sep 2003 14:58:34 +0000 (14:58 +0000)]
Add support for AMD766 and AMD768 chipsets.

PR: 41812

21 years agoUpdate the kern.osreldate documentation to document the present format
eivind [Mon, 1 Sep 2003 14:26:33 +0000 (14:26 +0000)]
Update the kern.osreldate documentation to document the present format
used, and refer to <osreldate.h> to get userland date.

Submitted by: ru

21 years agoRetire 'c' partition for a CD device.
maxim [Mon, 1 Sep 2003 12:50:02 +0000 (12:50 +0000)]
Retire 'c' partition for a CD device.

Submitted by: Craig Rodrigues <rodrigc@crodrigues.org>
Not objected by: -current

21 years agoTry to close the race between disk_destroy() and a subsequent disk_create().
phk [Mon, 1 Sep 2003 12:03:13 +0000 (12:03 +0000)]
Try to close the race between disk_destroy() and a subsequent disk_create().

21 years agoRearrange the probe code yet again.
sos [Mon, 1 Sep 2003 11:13:21 +0000 (11:13 +0000)]
Rearrange the probe code yet again.

21 years agoTry a port reset if initial contact to a device failed.
ticso [Mon, 1 Sep 2003 07:47:42 +0000 (07:47 +0000)]
Try a port reset if initial contact to a device failed.

tested by: Lee Damon <nomad@castle.org>

21 years agoUse pmap_steal_memory() for the msgbuf instead of trying to squeeze
marcel [Mon, 1 Sep 2003 07:06:57 +0000 (07:06 +0000)]
Use pmap_steal_memory() for the msgbuf instead of trying to squeeze
it in the last chunk (phys_avail block). The last chunk very often is
not larger than one or two pages, resulting in a msgbuf that's too
small to hold a complete verbose boot.
Note that pmap_steal_memory() will bzero the memory it "allocates".
Consequently, ia64 will never preserve previous msgbufs. This is not
a noticable difference in practice. If the msgbuf could be reused,
it was invariably too small to have anything preserved anyway.

21 years agoClarify the numbering of some of the build stages.
scottl [Mon, 1 Sep 2003 06:43:24 +0000 (06:43 +0000)]
Clarify the numbering of some of the build stages.

21 years agoImplement MBUF_STRESS_TEST mark II.
silby [Mon, 1 Sep 2003 05:55:37 +0000 (05:55 +0000)]
Implement MBUF_STRESS_TEST mark II.

Changes from the original implementation:

- Fragmentation is handled by the function m_fragment, which can
be called from whereever fragmentation is needed.  Note that this
function is wrapped in #ifdef MBUF_STRESS_TEST to discourage non-testing
use.

- m_fragment works slightly differently from the old fragmentation
code in that it allocates a seperate mbuf cluster for each fragment.
This defeats dma_map_load_mbuf/buffer's feature of coalescing adjacent
fragments.  While that is a nice feature in practice, it nerfed the
usefulness of mbuf_stress_test.

- Add two modes of random fragmentation.  Chains with fragments all of
the same random length and chains with fragments that are each uniquely
random in length may now be requested.

21 years agoLocking and misc cleanups; most of which I've been running for >4 months:
sam [Mon, 1 Sep 2003 05:35:55 +0000 (05:35 +0000)]
Locking and misc cleanups; most of which I've been running for >4 months:

o add locking
o strip irrelevant spl's
o split malloc types to better account for memory use
o remove unused IPSEC_NONBLOCK_ACQUIRE code
o remove dead code

Sponsored by: FreeBSD Foundation

21 years agoBackout rev 1.10.
deischen [Mon, 1 Sep 2003 05:31:33 +0000 (05:31 +0000)]
Backout rev 1.10.

Requested by: obrien

21 years agoadd locking
sam [Mon, 1 Sep 2003 05:12:36 +0000 (05:12 +0000)]
add locking

NB: There is a known LOR on the forwarding path; this needs to be resolved
    together with a similar issue in the bridge.  For the moment it is
    believed to be benign.

Sponsored by: FreeBSD Fondation

21 years agoo interlock domain list when adding domains
sam [Mon, 1 Sep 2003 05:01:55 +0000 (05:01 +0000)]
o interlock domain list when adding domains
o remove irrlevant spl

Notes:

1. We don't lock domain list traversals as this is safe until we start
   removing domains.
2. The calculation of max_datalen in net_init_domain appears safe as
   noone depends on max_hdr and max_datalen having consistent values.
3. Giant is still held for fast and slow timeouts; this must stay until
   each timeout routine is properly locked (coming soon).

Sponsored by: FreeBSD Fondation

21 years agoremove warning about use of old divert sockets; this was marked
sam [Mon, 1 Sep 2003 04:27:34 +0000 (04:27 +0000)]
remove warning about use of old divert sockets; this was marked
for removal before 5.2

Reviewed by: silence on -net and -arch

21 years agoadd locking
sam [Mon, 1 Sep 2003 04:23:48 +0000 (04:23 +0000)]
add locking

Sponsored by: FreeBSD Foundation

21 years agoEliminate last three uses of varargs.h in the tree. These three files
kan [Mon, 1 Sep 2003 04:12:18 +0000 (04:12 +0000)]
Eliminate last three uses of varargs.h in the tree. These three files
were including varargs.h file but did not use any of its macros,
so they escaped the clean-up before.

21 years agoBuild the submit.cf file instead of installing the version that ships
gshapiro [Mon, 1 Sep 2003 03:40:33 +0000 (03:40 +0000)]
Build the submit.cf file instead of installing the version that ships
with the vendor sendmail distribution.  By doing so, we avoid mergemaster
warnings after every 'make' in /etc/mail/.

Reported by: Jeremy Chadwick <koitsu@parodius.com>
MFC after: 3 days
X-MFC after: and re approval

21 years agoAdd a new machine independent varargs.h and use it as a central place
kan [Mon, 1 Sep 2003 03:28:25 +0000 (03:28 +0000)]
Add a new machine independent varargs.h and use it as a central place
to announce the demise of varargs support in GCC versions 3.3+ and to
direct users to stdarg.h instead.

Fall back to machine/varargs.h for older GCC versions.

21 years agoExplicitly enable probe request frame reception when not in station mode;
sam [Mon, 1 Sep 2003 03:12:19 +0000 (03:12 +0000)]
Explicitly enable probe request frame reception when not in station mode;
this is needed for the 5212 which a separate filter bit for these frames.

Submitted by:   Stephane Laroche <stephane.laroche@colubris.com>

21 years agoStandardize idempotentcy ifdefs. Consistently use _MACHINE_VARARGS_H_
kan [Mon, 1 Sep 2003 03:01:45 +0000 (03:01 +0000)]
Standardize idempotentcy ifdefs. Consistently use _MACHINE_VARARGS_H_
symbol.

21 years agoo correct logic that checks frame size to decide if a cluaster is needed
sam [Mon, 1 Sep 2003 02:55:09 +0000 (02:55 +0000)]
o correct logic that checks frame size to decide if a cluaster is needed
o add an assertion to check the max possible packet size

Noticed by: David Young <dyoung@pobox.com>

21 years agoeliminate casts from the DMAADDR macro. This depends upon bus_addr_t being
jmg [Mon, 1 Sep 2003 01:07:24 +0000 (01:07 +0000)]
eliminate casts from the DMAADDR macro.  This depends upon bus_addr_t being
a type that you can do arithmetic with.  This eliminates many warnings when
compiling with PAE.

Various by: scottl

21 years agoIf ~ chars were pennies, I'd be pennyless. This should fix all of the
scottl [Mon, 1 Sep 2003 00:20:29 +0000 (00:20 +0000)]
If ~ chars were pennies, I'd be pennyless.  This should fix all of the
'command not in queue' panics.  Also fix a nearby style problem.

21 years agoUse direct mapped KVA for the sf_buf allocator, as made possible
marcel [Mon, 1 Sep 2003 00:12:27 +0000 (00:12 +0000)]
Use direct mapped KVA for the sf_buf allocator, as made possible
by the previous commit. While here, fix a typo, reformat comments
and fix a long line.

Tested with: ftpd

21 years ago* Fix a duplicate "how how"
dougb [Sun, 31 Aug 2003 23:05:39 +0000 (23:05 +0000)]
* Fix a duplicate "how how"
* Fix pedstals -> pedestals

21 years agoImplement cpu_set_upcall_kse. May need tweaking.
jake [Sun, 31 Aug 2003 22:58:56 +0000 (22:58 +0000)]
Implement cpu_set_upcall_kse.  May need tweaking.

21 years agoRemove -pthread as a compiler option. It was deprecated 2.5 years
deischen [Sun, 31 Aug 2003 22:38:52 +0000 (22:38 +0000)]
Remove -pthread as a compiler option.  It was deprecated 2.5 years
ago, but not removed.

No reply from: threads, kan, obrien

21 years agoCatch up my(4) with hardware notes by adding Myson MTD89X to list of
simon [Sun, 31 Aug 2003 19:37:48 +0000 (19:37 +0000)]
Catch up my(4) with hardware notes by adding Myson MTD89X to list of
supported devices.

Submitted by: Lukas Ertl <l.ertl@univie.ac.at>
PR: docs/56244
Confirmed by: driver source code

21 years agoUpdate the comment to indicate ICH5 support
njl [Sun, 31 Aug 2003 19:23:00 +0000 (19:23 +0000)]
Update the comment to indicate ICH5 support

21 years agoUpgrade to 4.3
ache [Sun, 31 Aug 2003 19:15:35 +0000 (19:15 +0000)]
Upgrade to 4.3

21 years agoMerge local changes
ache [Sun, 31 Aug 2003 18:47:12 +0000 (18:47 +0000)]
Merge local changes

21 years agoMerge local changes
ache [Sun, 31 Aug 2003 18:29:38 +0000 (18:29 +0000)]
Merge local changes

21 years agoThis commit was generated by cvs2svn to compensate for changes in r119610,
ache [Sun, 31 Aug 2003 17:55:46 +0000 (17:55 +0000)]
This commit was generated by cvs2svn to compensate for changes in r119610,
which included commits to RCS files with non-trunk default branches.

21 years agoVirgin import of GNU Readline 4.3
ache [Sun, 31 Aug 2003 17:55:46 +0000 (17:55 +0000)]
Virgin import of GNU Readline 4.3

21 years agoDetect Geode CPUs and initialize the 27MHz timecounter "Geode".
phk [Sun, 31 Aug 2003 16:20:34 +0000 (16:20 +0000)]
Detect Geode CPUs and initialize the 27MHz timecounter "Geode".

This timecounter is 2usec faster than the i8254 and has 22 times
better resolution.

21 years agoWhitespace diff reduction between bsd.prog.mk and bsd.lib.mk outputs.
ru [Sun, 31 Aug 2003 15:20:17 +0000 (15:20 +0000)]
Whitespace diff reduction between bsd.prog.mk and bsd.lib.mk outputs.

21 years agoOnboard Adaptec SCSI HBAs are bootable on UP2000.
wilko [Sun, 31 Aug 2003 12:21:10 +0000 (12:21 +0000)]
Onboard Adaptec SCSI HBAs are bootable on UP2000.

Obtained from: obrien

21 years ago - Backout rev 1.142. This caused a deadlock that I do not understand. More
jeff [Sun, 31 Aug 2003 11:26:52 +0000 (11:26 +0000)]
 - Backout rev 1.142.  This caused a deadlock that I do not understand.  More
   investigation is required.

21 years ago - Define a new flag for getblk(): GB_NOCREAT. This flag causes getblk() to
jeff [Sun, 31 Aug 2003 08:50:11 +0000 (08:50 +0000)]
 - Define a new flag for getblk(): GB_NOCREAT.  This flag causes getblk() to
   bail out if the buffer is not already present.
 - The buffer returned by incore() is not locked and should not be sent to
   brelse().  Use getblk() with the new GB_NOCREAT flag to preserve the
   desired semantics.

21 years agoAdd a note that the -u option can be overridden by settings in login.conf(5).
ceri [Sun, 31 Aug 2003 07:45:48 +0000 (07:45 +0000)]
Add a note that the -u option can be overridden by settings in login.conf(5).

PR: docs/56017
Submitted by: Josef El-Rayes <j.el-rayes@daemon.li>

21 years ago - Don't acquire the vnode interlock in drain_output(). Instead, require the
jeff [Sun, 31 Aug 2003 07:29:34 +0000 (07:29 +0000)]
 - Don't acquire the vnode interlock in drain_output().  Instead, require the
   caller to acquire it.  This permits drain_output() to be done atomically
   with other operations as well as reducing the number of lock operations.
 - Assert that the proper locks are held in drain_output().
 - Change getdirtybuf() to accept a mutex as an argument.  This mutex is used
   to protect the vnode's buf list and the BKGRDWAIT flag.  This lock is
   dropped when we successfully acquire a buffer and held on return
   otherwise.  These semantics reduce the number of cumbersome cases in
   calling code.
 - Pass the mtx from getdirtybuf() into interlocked_sleep() and allow this
   mutex to be used as the interlock argument to BUF_LOCK() in the LOCKBUF
   case of interlocked_sleep().
 - Change the return value of getdirtybuf() to be the resulting locked buffer
   or NULL otherwise.  This is for callers who pass in a list head that
   requires a lock.  It is necessary since the lock that protects the list
   head must be dropped in getdirtybuf() so that we don't have a lock order
   reversal with the buf queues lock in bremfree().
 - Adjust all callers of getdirtybuf() to match the new semantics.
 - Add a comment in indir_trunc() that points at unlocked access to a buf.
   This may also be one of the last instances of incore() in the tree.

21 years agoAdd support for the 82801EB (ICH5) SMBus controller.
njl [Sun, 31 Aug 2003 01:28:02 +0000 (01:28 +0000)]
Add support for the 82801EB (ICH5) SMBus controller.

PR: kern/55485
Submitted by: Shin-ichi Yoshimoto <yosimoto@waishi.jp>
MFC after: 1 day

21 years ago - If there is no vp assume that BKGRDINPROG is not set and set RELPBUF in
jeff [Sun, 31 Aug 2003 01:07:45 +0000 (01:07 +0000)]
 - If there is no vp assume that BKGRDINPROG is not set and set RELPBUF in
   brelse().

21 years ago - In some cases bp->b_vp can be NULL in brelse, don't try to lock the
jeff [Sun, 31 Aug 2003 00:06:07 +0000 (00:06 +0000)]
 - In some cases bp->b_vp can be NULL in brelse, don't try to lock the
   interlock in that case.

Found by: alc

21 years ago - Add vm object locking to the part of vm_pageout_scan() that launders
alc [Sun, 31 Aug 2003 00:00:46 +0000 (00:00 +0000)]
 - Add vm object locking to the part of vm_pageout_scan() that launders
   dirty pages.
 - Remove some unused variables.

21 years agoIntroduce MAP_ENTRY_GROWS_DOWN and MAP_ENTRY_GROWS_UP to allow for
marcel [Sat, 30 Aug 2003 21:25:23 +0000 (21:25 +0000)]
Introduce MAP_ENTRY_GROWS_DOWN and MAP_ENTRY_GROWS_UP to allow for
growable (stack) entries that not only grow down, but also grow up.
Have vm_map_growstack() take these flags into account when growing
an entry.

This is the first step in adding support for upward growable stacks.
It is a required feature on ia64 to support the register stack (or
rstack as I like to call it -- it also means reverse stack). We do
not currently create rstacks, so the upward growing is not exercised
and the change should be a functional no-op.

Reviewed by: alc

21 years agoLabel the uarea address as such in DDB's ps output
phk [Sat, 30 Aug 2003 19:06:57 +0000 (19:06 +0000)]
Label the uarea address as such in DDB's ps output

21 years agoAdd the new g_dev_getprovider() function, the swap_pager needs it now.
phk [Sat, 30 Aug 2003 18:33:55 +0000 (18:33 +0000)]
Add the new g_dev_getprovider() function, the swap_pager needs it now.

Spotted by: mr

21 years agoAdd a close() method to a swapdev.
phk [Sat, 30 Aug 2003 16:44:26 +0000 (16:44 +0000)]
Add a close() method to a swapdev.

Add a GEOM based backend.

Remove the device/VOP_SPECSTRATEGY() based backend.

21 years agoProtect the swapdevice tailq with a mutex.
phk [Sat, 30 Aug 2003 16:10:28 +0000 (16:10 +0000)]
Protect the swapdevice tailq with a mutex.

Store the udev_t we will report to userland in the swdevt.

21 years agoReturn (-1) not (ENOENT) for mac_prepare_type(), and set errno to
rwatson [Sat, 30 Aug 2003 14:51:01 +0000 (14:51 +0000)]
Return (-1) not (ENOENT) for mac_prepare_type(), and set errno to
ENOENT instead.

Reported by: "Kenneth D. Merry" <ken@kdm.org>
Submitted by: Bryan Liesner <bleez@comcast.net>

21 years agoUpgrade to textutils 2.1 sort
ache [Sat, 30 Aug 2003 14:28:14 +0000 (14:28 +0000)]
Upgrade to textutils 2.1 sort

21 years agoMerge local fixes
ache [Sat, 30 Aug 2003 14:19:09 +0000 (14:19 +0000)]
Merge local fixes

21 years agoRemove file no longer in the dist
ache [Sat, 30 Aug 2003 14:08:22 +0000 (14:08 +0000)]
Remove file no longer in the dist

21 years agoThis commit was generated by cvs2svn to compensate for changes in r119582,
ache [Sat, 30 Aug 2003 14:05:55 +0000 (14:05 +0000)]
This commit was generated by cvs2svn to compensate for changes in r119582,
which included commits to RCS files with non-trunk default branches.

21 years agoVirgin import (trimmed) of GNU Sort, textutils 2.1
ache [Sat, 30 Aug 2003 14:05:55 +0000 (14:05 +0000)]
Virgin import (trimmed) of GNU Sort, textutils 2.1

21 years agoUnbreak the NOINSTALLLIB install.
ru [Sat, 30 Aug 2003 13:57:31 +0000 (13:57 +0000)]
Unbreak the NOINSTALLLIB install.

PR: 50945
Submitted by: Rene de Vries <rene@tunix.nl>
Reminded by: jmallett

21 years agoRevert rev. 1.378, and restore the correct arithmetic expression
ru [Sat, 30 Aug 2003 13:48:16 +0000 (13:48 +0000)]
Revert rev. 1.378, and restore the correct arithmetic expression
syntax.  The

make buildworld
mv /usr/include /usr/include.old
make installworld

issue has been fixed a month ago in Makefile,v 1.285, and there
is no valid reason to continue to keep the wrong syntax here --
buildworld takes care of upgrading a make for you if necessary.
But if you find yourself in an environment with an old make(1)
binary that breaks on this, and this is because you attempted
to run a target other than buildworld, don't whine but try again
with -DALWAYS_CHECK_MAKE defined -- it should do the trick.
Otherwise, if you still have a problem, please report it as a
bug and attach the ``make -dl ...'' output.

Reviewed by: marcel

21 years agoCosmetics: folded BITGTS into TGTS.
ru [Sat, 30 Aug 2003 13:33:41 +0000 (13:33 +0000)]
Cosmetics: folded BITGTS into TGTS.

21 years agoImplement missing shell arithmetic operators in $(()) expansion
schweikh [Sat, 30 Aug 2003 12:31:44 +0000 (12:31 +0000)]
Implement missing shell arithmetic operators in $(()) expansion
and variable recognition.

PR: standards/52972
Submitted by: Wartan Hachaturow <wart@tepkom.ru>
Reviewed by: tjr (improved on original patch)
Tested by: buildworld on CURRENT.
MFC after: 6 weeks

21 years agoAllow the concurrency level to be reduced.
deischen [Sat, 30 Aug 2003 12:09:16 +0000 (12:09 +0000)]
Allow the concurrency level to be reduced.

Reviewed by: davidxu

21 years agoContinue the objectification of the swapdev backends:
phk [Sat, 30 Aug 2003 11:33:25 +0000 (11:33 +0000)]
Continue the objectification of the swapdev backends:

Remove the vnode and dev_t fields and replace them with a void *.

Introduce separate strategy functions for devices and regular (NFS)
vnodes.

For devices we don't need the vnode v_numoutput stuff.

Add a generic swaponsomething() function to add a swapdevice and
split the remainder of swaponvp() into swaponvp() and swapondev()
which calls this backend.

21 years agoMake the strategy function a method of the individual swapdev.
phk [Sat, 30 Aug 2003 09:42:00 +0000 (09:42 +0000)]
Make the strategy function a method of the individual swapdev.

21 years agoConsistent use modern function definitions
phk [Sat, 30 Aug 2003 08:32:42 +0000 (08:32 +0000)]
Consistent use modern function definitions

21 years agoStyle fixes of the whitespace variety. Fix long lines and tabs.
markm [Sat, 30 Aug 2003 08:10:58 +0000 (08:10 +0000)]
Style fixes of the whitespace variety. Fix long lines and tabs.

21 years agoAdd pst as a module
sos [Sat, 30 Aug 2003 08:01:05 +0000 (08:01 +0000)]
Add pst as a module

21 years agoDocument that read(2) can also return EPERM
dds [Sat, 30 Aug 2003 07:59:05 +0000 (07:59 +0000)]
Document that read(2) can also return EPERM

See e.g. nfsclient/nfs_vnops.c

static int
nfs_read(struct vop_read_args *ap)
{
        struct vnode *vp = ap->a_vp;

        if (vp->v_type != VREG)
                return (EPERM);
        return (nfs_bioread(vp, ap->a_uio, ap->a_ioflag, ap->a_cred));
}

Approved by: schweikh (mentor)
MFC after: 6 weeks

21 years agoBackout Rev. 1.24
brueffer [Sat, 30 Aug 2003 07:49:42 +0000 (07:49 +0000)]
Backout Rev. 1.24

English lessons provided by: jhb

21 years agoTypo in last commit.
phk [Sat, 30 Aug 2003 06:39:56 +0000 (06:39 +0000)]
Typo in last commit.

Spotted by: tjr

21 years agoAdd a reference to bn(3) for those looking for functional multiprecision
seanc [Sat, 30 Aug 2003 05:35:46 +0000 (05:35 +0000)]
Add a reference to bn(3) for those looking for functional multiprecision
integer arithmetic.

21 years agoMore debug under boot verbose.
imp [Fri, 29 Aug 2003 23:25:00 +0000 (23:25 +0000)]
More debug under boot verbose.

21 years agoFix comment.
imp [Fri, 29 Aug 2003 23:24:41 +0000 (23:24 +0000)]
Fix comment.

21 years agoGrammar fix
brueffer [Fri, 29 Aug 2003 20:12:21 +0000 (20:12 +0000)]
Grammar fix

21 years agoMigrate the sf_buf allocator that is used by sendfile(2) and zero-copy
alc [Fri, 29 Aug 2003 20:04:10 +0000 (20:04 +0000)]
Migrate the sf_buf allocator that is used by sendfile(2) and zero-copy
sockets into machine-dependent files.  The rationale for this
migration is illustrated by the modified amd64 allocator.  It uses the
amd64's direct map to avoid emphemeral mappings in the kernel's
address space.  On an SMP, the emphemeral mappings result in an IPI
for TLB shootdown for each transmitted page.  Yuck.

Maintainers of other 64-bit platforms with direct maps should be able
to use the amd64 allocator as a reference implementation.

21 years agoupdate UP2000 section with info gleaned from obrien's newly acquired
wilko [Fri, 29 Aug 2003 19:43:08 +0000 (19:43 +0000)]
update UP2000 section with info gleaned from obrien's newly acquired
UP2000 dual 750MHz machine.

21 years agoMore checking of M_PREPEND() return values: pass through link-layer
rwatson [Fri, 29 Aug 2003 19:27:09 +0000 (19:27 +0000)]
More checking of M_PREPEND() return values: pass through link-layer
Appletalk code.  These changes are untested since I don't have
an Appletalk environment.

21 years agoIntroduce error checking for calls to M_PREPEND():
rwatson [Fri, 29 Aug 2003 19:12:18 +0000 (19:12 +0000)]
Introduce error checking for calls to M_PREPEND():

  ether_output() when prepending netatalk AFA_PHASE2 llc headers (TRYWAIT).
  ether_output() when prepending ethernet header to a frame (DONTWAIT).

21 years ago- Update the page title to include information about which cards
simon [Fri, 29 Aug 2003 17:59:01 +0000 (17:59 +0000)]
- Update the page title to include information about which cards
  are supported by the driver.
- Use a list to specify which cards are supported.
- Add the 3c592/3c597 EISA cards to the list of supported cards.

PR: docs/56086 (based on)
Submitted by: Lukas Ertl <l.ertl@univie.ac.at>