]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
24 years agoWhen files are given to users by root, the quota system failed to
Kirk McKusick [Fri, 28 Apr 2000 06:12:56 +0000 (06:12 +0000)]
When files are given to users by root, the quota system failed to
reset their grace timer as their ownership crossed the soft limit
threshhold. Thus if they had been over their limit in the past,
they were suddenly penalized as if they had been over their limit
ever since. The fix is to check when root gives away files, that
when the receiving user crosses their soft limit, their grace timer
is reset. See the PR report for a detailed method of reproducing
the bug.

PR: kern/17128
Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
Reviewed by: Kirk McKusick <mckusick@mckusick.com>

24 years agosanpai-san (sanpai@sanpai.org) suggested that we put the novel probe
Warner Losh [Fri, 28 Apr 2000 05:01:35 +0000 (05:01 +0000)]
sanpai-san (sanpai@sanpai.org) suggested that we put the novel probe
first.  This will fix a few cards that hang on the WD probe.  He tells
me that PAO went one step farther and removed the WD proble completely
and none of the cards in the 2.x database broke in PAO3.  Since I'm
more conservative in this code, I'm just swapping the order, which he
said also fixed his problem.

Reviewed by: mdodd, iwasaki
Submitted by: sanpai@sanpai.org

24 years agoAdd a new macro CIRCLEQ_FOREACH_REVERSE for traversing through a
Archie Cobbs [Thu, 27 Apr 2000 22:50:12 +0000 (22:50 +0000)]
Add a new macro CIRCLEQ_FOREACH_REVERSE for traversing through a
circle queue in the reverse direction (from tail to head).

24 years agoDescribe drainwait
Andrey A. Chernov [Thu, 27 Apr 2000 22:08:29 +0000 (22:08 +0000)]
Describe drainwait

24 years agoremove last empty line
Andrey A. Chernov [Thu, 27 Apr 2000 21:58:46 +0000 (21:58 +0000)]
remove last empty line

24 years agopart of gfmt really changed commit
Andrey A. Chernov [Thu, 27 Apr 2000 21:48:48 +0000 (21:48 +0000)]
part of gfmt really changed commit

24 years agogfmt: set 'changed' flags only if something really changed.
Andrey A. Chernov [Thu, 27 Apr 2000 21:45:41 +0000 (21:45 +0000)]
gfmt: set 'changed' flags only if something really changed.

24 years agoAdd ability to manipulate with drain wait time
Andrey A. Chernov [Thu, 27 Apr 2000 21:31:23 +0000 (21:31 +0000)]
Add ability to manipulate with drain wait time

24 years agoMake gperf be a bootstrap-tool, since we now depend on features of
Garrett Wollman [Thu, 27 Apr 2000 21:24:43 +0000 (21:24 +0000)]
Make gperf be a bootstrap-tool, since we now depend on features of
the specific version in -current.

Approved in principle by: marcel

24 years agomake the example compile again.
Nick Hibma [Thu, 27 Apr 2000 21:08:24 +0000 (21:08 +0000)]
make the example compile again.

The makefile contains a reference to /sys/dev/ppbus. What really should
be done is copy the header files to /usr/include/sys/dev/ppbus.

PR:             kern/16767
Submitted by:    Jin Guojun (FTG staff) <jin@gracie.lbl.gov>

24 years agoRemove a warning
Nick Hibma [Thu, 27 Apr 2000 21:04:17 +0000 (21:04 +0000)]
Remove a warning

24 years agoRemove PC164UX from 'might work/might have SRM'. Add warning for
Wilko Bulte [Thu, 27 Apr 2000 20:19:48 +0000 (20:19 +0000)]
Remove PC164UX from 'might work/might have SRM'. Add warning for
non-SRM for PC164UX and XL/XLT DEC machines

Submitted by: <naddy@mips.rhein-neckar.de>

24 years agoAdd default 5min timeout for output drain to stop hanging on exit or in other
Andrey A. Chernov [Thu, 27 Apr 2000 20:14:21 +0000 (20:14 +0000)]
Add default 5min timeout for output drain to stop hanging on exit or in other
places when connection dropped

24 years agoUse profiling on LINT.
Poul-Henning Kamp [Thu, 27 Apr 2000 19:58:57 +0000 (19:58 +0000)]
Use profiling on LINT.
cosmetics.

24 years agoRevert my changes to make gperf a build-tool; that was the wrong thing.
Garrett Wollman [Thu, 27 Apr 2000 19:38:57 +0000 (19:38 +0000)]
Revert my changes to make gperf a build-tool; that was the wrong thing.
All builds had been broken; now just upgrade builds are until I or
someone else can figure out the Right Thing.

24 years agoSpell PacketAliasRedirectAddr() correctly.
Ruslan Ermilov [Thu, 27 Apr 2000 18:06:05 +0000 (18:06 +0000)]
Spell PacketAliasRedirectAddr() correctly.

24 years agoLoad Sharing using IP Network Address Translation (RFC 2391, LSNAT).
Ruslan Ermilov [Thu, 27 Apr 2000 17:55:17 +0000 (17:55 +0000)]
Load Sharing using IP Network Address Translation (RFC 2391, LSNAT).

24 years agoLoad Sharing using IP Network Address Translation (RFC 2391, LSNAT).
Ruslan Ermilov [Thu, 27 Apr 2000 17:37:03 +0000 (17:37 +0000)]
Load Sharing using IP Network Address Translation (RFC 2391, LSNAT).

LSNAT links are first created by either PacketAliasRedirectPort() or
PacketAliasRedirectAddress() and then set up by one or more calls to
PacketAliasAddServer().

24 years agoUpdate periodic to use the function source_rc_confs that
Neil Blakey-Milner [Thu, 27 Apr 2000 17:11:03 +0000 (17:11 +0000)]
Update periodic to use the function source_rc_confs that
/etc/defaults/rc.conf now exports.

24 years agoAdd gperf(1) to build tools. It should have been one before, but
Garrett Wollman [Thu, 27 Apr 2000 16:43:45 +0000 (16:43 +0000)]
Add gperf(1) to build tools.  It should have been one before, but
gcc does not depend on version-specific gperf behavior (yet).

24 years agoAdd a bpfdetach() stub routine to bpf.c. Without this, you'll get an
Bill Paul [Thu, 27 Apr 2000 15:11:41 +0000 (15:11 +0000)]
Add a bpfdetach() stub routine to bpf.c. Without this, you'll get an
unresolved symbol error if you try to load a network driver into a kernel
which doesn't have bpf enabled.

Forgotten by: rwatson
Found by: peter

24 years agoAdded PC-98 supports.
Yoshihiro Takahashi [Thu, 27 Apr 2000 13:37:40 +0000 (13:37 +0000)]
Added PC-98 supports.

Submitted by: Chiharu Shibata <chi@bd.mbn.or.jp>,
Tomokazu HARADA <tkhara@osk4.3web.ne.jp> and
yokota

24 years agoSupported EGC 640x400, PEGC 640x400 and PEGC 640x480 graphics modes.
Yoshihiro Takahashi [Thu, 27 Apr 2000 13:34:32 +0000 (13:34 +0000)]
Supported EGC 640x400, PEGC 640x400 and PEGC 640x480 graphics modes.

Submitted by: Chiharu Shibata <chi@bd.mbn.or.jp> and
Tomokazu HARADA <tkhara@osk4.3web.ne.jp>

24 years agomachine/random.h -> sys/random.h
Yoshihiro Takahashi [Thu, 27 Apr 2000 13:13:43 +0000 (13:13 +0000)]
machine/random.h -> sys/random.h

24 years agoAdd wormio.h. The wd driver needs it.
Yoshihiro Takahashi [Thu, 27 Apr 2000 13:09:07 +0000 (13:09 +0000)]
Add wormio.h. The wd driver needs it.

24 years agoMake PXE do a dhcp request to find out where its files are. The
Paul Saab [Thu, 27 Apr 2000 12:30:28 +0000 (12:30 +0000)]
Make PXE do a dhcp request to find out where its files are.  The
DHCP server is no longer required to also act as the NFS/TFTP
server.

24 years agoFix language in description of firewall_script.
Dag-Erling Smørgrav [Thu, 27 Apr 2000 10:41:53 +0000 (10:41 +0000)]
Fix language in description of firewall_script.

Submitted by: sheldonh

24 years agoCalculate checksum properly for propagated IPX/NetBIOS packets.
Boris Popov [Thu, 27 Apr 2000 10:29:14 +0000 (10:29 +0000)]
Calculate checksum properly for propagated IPX/NetBIOS packets.

24 years agoFix support for 802.2 and SNAP frames. Bug was introduced during
Boris Popov [Thu, 27 Apr 2000 10:13:52 +0000 (10:13 +0000)]
Fix support for 802.2 and SNAP frames. Bug was introduced during
initial import.

Tested by: Jorge P Vasquez <jorge@acron.ind.br>

24 years agoCreate ${KRNLOBJDIR} before running config(8), since config(8) does
Sheldon Hearn [Thu, 27 Apr 2000 09:20:18 +0000 (09:20 +0000)]
Create ${KRNLOBJDIR} before running config(8), since config(8) does
not create required parent directories of the kernel compile
directory specified with its -d option.

24 years agoDocument firewall_script
Dag-Erling Smørgrav [Thu, 27 Apr 2000 09:10:53 +0000 (09:10 +0000)]
Document firewall_script

24 years agoSuggest to use -pthreads instead of -lc_r
Alexey Zelkin [Thu, 27 Apr 2000 08:54:09 +0000 (08:54 +0000)]
Suggest to use -pthreads instead of -lc_r

Noticed by: John-Mark Gurney <gurney_j@efn.org>

24 years agoAdd to defaults/rc.conf a new function source_rc_confs which rc
Sheldon Hearn [Thu, 27 Apr 2000 08:43:49 +0000 (08:43 +0000)]
Add to defaults/rc.conf a new function source_rc_confs which rc
scripts may use to source safely overrides in ${rc_conf_files}
files.

This protects users who insist on the bad practice of copying
/etc/defaults/rc.conf to /etc/rc.conf from a recursive loop
that exhausts available file descriptors.

Several people have expressed interest in breaking this function
out into its own shell script.  Anyone who wants to embark on
such an undertaking would do well to study the attributed PR.

PR: 17595
Reported by: adrian
Submitted by: Doug Barton <Doug@gorean.org>

24 years agoDon't report errors when we UDP_WRITE fails with a status of 1.
Paul Saab [Thu, 27 Apr 2000 03:23:50 +0000 (03:23 +0000)]
Don't report errors when we UDP_WRITE fails with a status of 1.

24 years agoAdd a wrapper for the sendfile() system call.
Jason Evans [Thu, 27 Apr 2000 00:59:44 +0000 (00:59 +0000)]
Add a wrapper for the sendfile() system call.

PR: bin/17366

24 years agoBack out the hook to execute the file ${firewall_type}. The intended
Brian S. Dean [Thu, 27 Apr 2000 00:48:59 +0000 (00:48 +0000)]
Back out the hook to execute the file ${firewall_type}.  The intended
purpose of the hook was to provide the ability for a shell program to
instantiate the firewall rules instead of forcing them to be
statically coded.  This functionality was already present through the
use of ${firewall_script}, and I see no need to keep the
${firewall_type} hook around.

Reminded by: Dag-Erling Smorgrav <des@freebsd.org>

24 years agoAdd test to detect propagation of cancellation points within libc_r.
Jason Evans [Wed, 26 Apr 2000 23:25:58 +0000 (23:25 +0000)]
Add test to detect propagation of cancellation points within libc_r.

24 years agoRemove cancellation point propagation.
Jason Evans [Wed, 26 Apr 2000 23:17:17 +0000 (23:17 +0000)]
Remove cancellation point propagation.

24 years agoduh, i forgot to change a bitmask, sorry alexander
Cameron Grant [Wed, 26 Apr 2000 23:08:12 +0000 (23:08 +0000)]
duh, i forgot to change a bitmask, sorry alexander

Submitted by: Alexander Matey <matey@cis.ohio-state.edu>

24 years ago Remove synchronizing instruction in MP unlock code. It turns out
Matthew Dillon [Wed, 26 Apr 2000 21:16:54 +0000 (21:16 +0000)]
Remove synchronizing instruction in MP unlock code.  It turns out
    not to be necessary.

24 years ago Fix #! script exec under linux emulation. If a script is exec'd from a
Matthew Dillon [Wed, 26 Apr 2000 20:58:40 +0000 (20:58 +0000)]
Fix #! script exec under linux emulation.  If a script is exec'd from a
    program running under linux emulation, the script binary is checked for
    in /compat/linux first.  Without this patch the wrong script binary
    (i.e. the FreeBSD binary) will be run instead of the linux binary.
    For example, #!/bin/sh, thus breaking out of linux compatibility mode.

    This solves a number of problems people have had installing linux
    software on FreeBSD boxes.

24 years agoDriver for DEC "Tulip" based WAN cards from LanMedia Corporation.
Poul-Henning Kamp [Wed, 26 Apr 2000 20:16:56 +0000 (20:16 +0000)]
Driver for DEC "Tulip" based WAN cards from LanMedia Corporation.

This driver should support both the SSI (V.35 etc) E1/T1 unchannelized,
DS3 and HSSI cards.  Only tested on the SSI card.

More info at: http://www.lanmedia.com

Thanks to LanMedia for donating two LMC1000P cards.

if_de.c driver modified by:     LanMedia
NetGraphification by:    Stephen Kiernan <sk-ports@vegamuse.org>

24 years agofix minor numbers for multi-channel devices
Cameron Grant [Wed, 26 Apr 2000 20:06:52 +0000 (20:06 +0000)]
fix minor numbers for multi-channel devices

Submitted by: Alexander Matey <matey@cis.ohio-state.edu>

24 years agoNow that we are compiling PXE into libi386, we don't need ../libi386 in
John Baldwin [Wed, 26 Apr 2000 19:54:49 +0000 (19:54 +0000)]
Now that we are compiling PXE into libi386, we don't need ../libi386 in
.PATH anymore.

24 years agorepo-copied to sys/gnu/dev/sound/pci/emu10k1.h
Cameron Grant [Wed, 26 Apr 2000 19:53:46 +0000 (19:53 +0000)]
repo-copied to sys/gnu/dev/sound/pci/emu10k1.h

24 years agoupdate for new location of emu10k1.h under sys/gnu
Cameron Grant [Wed, 26 Apr 2000 19:51:19 +0000 (19:51 +0000)]
update for new location of emu10k1.h under sys/gnu

24 years agoMFPAO3. Improvement of of pccard cis tuple parsing capability.
Mitsuru IWASAKI [Wed, 26 Apr 2000 15:36:43 +0000 (15:36 +0000)]
MFPAO3.  Improvement of of pccard cis tuple parsing capability.
 - Fixed bogus CIS tuple dumping (Network node ID, IRQ modes and etc.)
 - Include telling drivers ethernet address if Network node ID
   tuple is available.  This is usefull for some bogus ehter cards which
   can't get correct ethernet address from CIS tupple.

Obtained from: PAO3

24 years agoMFPAO3. Add support multi io window. This code support following cards;
Mitsuru IWASAKI [Wed, 26 Apr 2000 15:11:17 +0000 (15:11 +0000)]
MFPAO3.  Add support multi io window.  This code support following cards;
Accton UE2212
PLANET-SMART-COM-CREDITCARD-2000
Melco LPC-T
ME-3000II
Laneed LD-CDY
Melco LPC3-TX

Submitted by: MIHIRA Sanpei Yoshiro <sanpei@sanpei.org>
Obtained from: PAO3

24 years ago"r" (raw) devices aren't named the traditional way any longer.
David E. O'Brien [Wed, 26 Apr 2000 12:54:51 +0000 (12:54 +0000)]
"r" (raw) devices aren't named the traditional way any longer.

24 years agoThe worm device is dead in -CURRENT, and wormio.h just went away.
David E. O'Brien [Wed, 26 Apr 2000 11:58:32 +0000 (11:58 +0000)]
The worm device is dead in -CURRENT, and wormio.h just went away.

Reviewed by: commiters on IRC

24 years agoMove procfs_fullpath() to vfs_cache.c, with a rename to textvp_fullpath().
Brian Feldman [Wed, 26 Apr 2000 11:57:45 +0000 (11:57 +0000)]
Move procfs_fullpath() to vfs_cache.c, with a rename to textvp_fullpath().
There's no excuse to have code in synthetic filestores that allows direct
references to the textvp anymore.

Feature requested by: msmith
Feature agreed to by: warner
Move requested by: phk
Move agreed to by: bde

24 years ago/dev/r<FOO> => /dev/<FOO>
David E. O'Brien [Wed, 26 Apr 2000 11:52:44 +0000 (11:52 +0000)]
/dev/r<FOO> => /dev/<FOO>

24 years ago"r" devices are on their way out.
David E. O'Brien [Wed, 26 Apr 2000 11:38:10 +0000 (11:38 +0000)]
"r" devices are on their way out.

24 years agoDefault device not longer uses the "r" raw prefix.
David E. O'Brien [Wed, 26 Apr 2000 11:34:52 +0000 (11:34 +0000)]
Default device not longer uses the "r" raw prefix.

24 years ago/dev/r<FOO> => /dev/<FOO>
David E. O'Brien [Wed, 26 Apr 2000 11:24:44 +0000 (11:24 +0000)]
/dev/r<FOO> => /dev/<FOO>

24 years agoThis file isn't needed anymore, has symbols that conflicts with cdrio.h
David E. O'Brien [Wed, 26 Apr 2000 11:18:16 +0000 (11:18 +0000)]
This file isn't needed anymore, has symbols that conflicts with cdrio.h
which produces warnings when compiling kdump.

24 years agoClarify the use of the auth service's -d option for specifying
Sheldon Hearn [Wed, 26 Apr 2000 10:40:35 +0000 (10:40 +0000)]
Clarify the use of the auth service's -d option for specifying
a fallback username.

Reviewed by: green

24 years agoAdd braces to avoid ambiguity in a nested conditional (silences a gcc
Sheldon Hearn [Wed, 26 Apr 2000 10:38:19 +0000 (10:38 +0000)]
Add braces to avoid ambiguity in a nested conditional (silences a gcc
warning).

24 years agoBring support in for Intel Wired for Management 1.1 (PXE 0.99 and
Paul Saab [Wed, 26 Apr 2000 07:38:40 +0000 (07:38 +0000)]
Bring support in for Intel Wired for Management 1.1 (PXE 0.99 and
below).  This did not work previously because interrupts were
disabled when PXE calls were being made, and they must be enabled.
This should also allow us to be compliant with all newer PXE rom's
from Intel.

For PXE 0.99, this has been tested using the Intel N440BX motherboard
and I am confident it will work on the Intel L440GX motherboard.

Lots of help/information from: jhb, peter

I would like to thank Michael Johnston <michael.johnston@intel.com>,
Mike Henry <mike.henry@intel.com>, and all the other PXE developers
at Intel for their help, and information in helping solve this
problem.

24 years agorandom.h moved to sys/
David E. O'Brien [Wed, 26 Apr 2000 07:17:26 +0000 (07:17 +0000)]
random.h moved to sys/

24 years agoUnbreak for the case where ``make obj'' was run first.
David E. O'Brien [Wed, 26 Apr 2000 07:00:46 +0000 (07:00 +0000)]
Unbreak for the case where ``make obj'' was run first.

24 years agoRemove dummy read_random function since we now have a real one.
Matt Jacob [Wed, 26 Apr 2000 06:49:39 +0000 (06:49 +0000)]
Remove dummy read_random function since we now have a real one.

24 years agoWell, make intrhand2_t take a void * arg then. Matches i386.
Matt Jacob [Wed, 26 Apr 2000 06:48:54 +0000 (06:48 +0000)]
Well, make intrhand2_t take a void * arg then. Matches i386.

24 years agoReplace the /dev/stdin, /dev/stdout, and /dev/stderr nodes with
Chris Costello [Wed, 26 Apr 2000 06:12:42 +0000 (06:12 +0000)]
Replace the /dev/stdin, /dev/stdout, and /dev/stderr nodes with
symlinks to fd/0, fd/1, and fd/2 respectively.  This will make
things easier for upcoming changes to fdesc (and does not break
anything for the current state of things) and hopefully its
position as the replacement for our existing static /dev/fd nodes.

Suggested (and reviewed) by: phk

24 years agoLanMedias driver sources for the LMC1000 and similar cards.
Poul-Henning Kamp [Wed, 26 Apr 2000 06:02:13 +0000 (06:02 +0000)]
LanMedias driver sources for the LMC1000 and similar cards.

Does not yet compile yet.

24 years agoCorrect typo in description string: `CANNOT' -> `CAN'.
Joseph Koshy [Wed, 26 Apr 2000 05:21:54 +0000 (05:21 +0000)]
Correct typo in description string: `CANNOT' -> `CAN'.

PR: docs/18177
Submitted by: MIHIRA Yoshiro <sanpei@sanpei.org>

24 years agoFix typo. Use `.Fa' to denote a function argument.
Joseph Koshy [Wed, 26 Apr 2000 05:09:22 +0000 (05:09 +0000)]
Fix typo.  Use `.Fa' to denote a function argument.

PR: docs/18214
Submitted by: Ben Smithurst <ben@scientia.demon.co.uk>

24 years agoDon't disable interrupts when calling a vm86 mode interrupt or routine
John Baldwin [Wed, 26 Apr 2000 04:35:25 +0000 (04:35 +0000)]
Don't disable interrupts when calling a vm86 mode interrupt or routine
from user mode.  Don't disable interrupts when returning from vm86 mode
to user mode either.  Now, we only disable interrupts before calling a
hardware interrupt handler, which is the only time we _should_ be
disabling interrupts.

Because of this, err, feature, any routine that one called in vm86 mode
had to re-enable interrupts by setting the interrupt flag or interrupts
would remain disabled even after the routine returned.  For example, I
have a simple debugging routine that uses a vm86 mode function to dump
any arbitrary memory word that I use to read the BIOS timer or any other
memory location.  This function does 1 load instruction from memory and
then returns.  Since it didn't re-enable interrupts, the first time I
called it to read the BIOS timer, it disabled interrupts.   This also
affected the PXE bootstrap as it needs interrupts enabled while it is
processing.  This patch fixes both of those situations so that those
functions do not worry about having to enable interrupts.  Hardware
interrupt handlers worked fine with the old code because they always
enable interrupts as part of their routine.

If you have any problems with the loader after this commit, please
let me know.  I'd like to MFC it in a week or two since PXE support
needs it.

Noticed by: ps, Michael Johnston <michael.johnston@intel.com>

24 years agoremove "register" specifiers to supress compiler warning.
Kenjiro Cho [Wed, 26 Apr 2000 02:40:33 +0000 (02:40 +0000)]
remove "register" specifiers to supress compiler warning.

24 years agoHello, getconf. This is a slight reinvention of the
Garrett Wollman [Wed, 26 Apr 2000 02:36:54 +0000 (02:36 +0000)]
Hello, getconf.  This is a slight reinvention of the
wheel^H^H^H^H^HPOSIX.2 and X/Open utility, and rather
more complicated than necessary.

24 years agoRemove unused variable.
Matt Jacob [Wed, 26 Apr 2000 00:20:01 +0000 (00:20 +0000)]
Remove unused variable.

24 years agoIgnore tulip chips on LanMedia WAN cards.
Poul-Henning Kamp [Tue, 25 Apr 2000 21:45:45 +0000 (21:45 +0000)]
Ignore tulip chips on LanMedia WAN cards.

24 years agoRemove mistakenly added {o}init. The build system adds it automatically.
Doug White [Tue, 25 Apr 2000 19:03:31 +0000 (19:03 +0000)]
Remove mistakenly added {o}init.  The build system adds it automatically.

24 years agoCommit PicoBSD mega-patch:
Doug White [Tue, 25 Apr 2000 19:00:08 +0000 (19:00 +0000)]
Commit PicoBSD mega-patch:
. Bring PicoBSD up to date with -CURRENT reality.
. Make PicoBSD smart and not spam /dev/vn0 and /mnt.  Now it uses the first
  unused vn device and creates a temporary mountpoint in /tmp.
. Miscellaneous build cleanups and optimizations.

Unfortunately the bridge, isp, and router floppies are too big and need
the axe treatment.  The install floppy needs updating to -CURRENT. Dial and
net build and (appear to) run OK.  I will be adressing these in the
near future (unless someone beats me to it :-) ).

PR: misc/17737
Submitted by: Omachonu Ogali <oogali@intranova.net>

24 years agoAdd punctuation to make a sentence more readable.
Garrett Wollman [Tue, 25 Apr 2000 16:40:57 +0000 (16:40 +0000)]
Add punctuation to make a sentence more readable.

24 years agoReassemble a sentence that has been botched in rev 1.4.
Joerg Wunsch [Tue, 25 Apr 2000 13:53:48 +0000 (13:53 +0000)]
Reassemble a sentence that has been botched in rev 1.4.

Reported by: Alexander Leidinger <Alexander@leidinger.net>

24 years agoClone the {b|bio}_offset field, and make sure it is always initialized
Poul-Henning Kamp [Tue, 25 Apr 2000 10:51:18 +0000 (10:51 +0000)]
Clone the {b|bio}_offset field, and make sure it is always initialized
in struct bio.  Eventually, bio_offset will probably obsolete the
bio_blkno and bio_pblkno fields.

Remove the special hack in atapi-cd.c to determine of bio_offset was valid.

24 years agoFix excessive use of parenthesis in previous commit.
Sheldon Hearn [Tue, 25 Apr 2000 08:59:51 +0000 (08:59 +0000)]
Fix excessive use of parenthesis in previous commit.

Requested by: bde

24 years agoFirst cut at adding CIS reading to xe driver. I don't have a card
Warner Losh [Tue, 25 Apr 2000 06:09:54 +0000 (06:09 +0000)]
First cut at adding CIS reading to xe driver.  I don't have a card
that seems to be working (I have a MF card that has a 336 modem and
ethernet that the probe routine finds, but the attach fails on because
pccardd doesn't do what you'd like with MF cards all the time).

24 years agoCheckin my first batch of New Mexico changes:
Warner Losh [Tue, 25 Apr 2000 06:07:27 +0000 (06:07 +0000)]
Checkin my first batch of New Mexico changes:
o minor whitespace things (bad because this is also a functional commit)
o Backport reading in of CIS entries from the driver level.

24 years agoMore machdep/random.h -> sys/random.h fixes.
Boris Popov [Tue, 25 Apr 2000 05:06:31 +0000 (05:06 +0000)]
More machdep/random.h -> sys/random.h fixes.

24 years agoFix a long-standing bug which caused massive character loss in remote
Greg Lehey [Tue, 25 Apr 2000 01:41:57 +0000 (01:41 +0000)]
Fix a long-standing bug which caused massive character loss in remote
serial gdb: interrupts were causing either overruns or stealing
characters.  Put splhigh() around the routines which transfer packets
across the line.  Since this happens when the system is halted in
debug, this doesn't cause any particular problem.  Now it is possible
to run the link at 115,200 bps.

PR:    (not assigned yet, must be in limbo somewhere)

Add partial support for detecting non-existent gdb devices.

Add $FreeBSD$ tag.

24 years agoAdd linker_if.m to the mix.
David E. O'Brien [Mon, 24 Apr 2000 23:08:23 +0000 (23:08 +0000)]
Add linker_if.m to the mix.

24 years agoUse kern/kern_random.c.
David E. O'Brien [Mon, 24 Apr 2000 21:53:54 +0000 (21:53 +0000)]
Use kern/kern_random.c.

24 years agoAutomated regression test harness for libc_r. Existing tests are integrated,
Jason Evans [Mon, 24 Apr 2000 21:07:45 +0000 (21:07 +0000)]
Automated regression test harness for libc_r.  Existing tests are integrated,
a new test for POSIX semaphores was added, and examples of harness usage are
included.

24 years agoDisable the returning of "wedged" commands; this is wrong and needs to be
Mike Smith [Mon, 24 Apr 2000 19:56:26 +0000 (19:56 +0000)]
Disable the returning of "wedged" commands; this is wrong and needs to be
rethought.

24 years agoUpdate the Tigon firmware to 12.3.21. This fixes a few bugs and adds support
Bill Paul [Mon, 24 Apr 2000 17:44:45 +0000 (17:44 +0000)]
Update the Tigon firmware to 12.3.21. This fixes a few bugs and adds support
for cards with 2MB of on-board SRAM.

24 years ago* Use sys/sys/random.h rather than a i386 specific one.
David E. O'Brien [Mon, 24 Apr 2000 17:30:08 +0000 (17:30 +0000)]
* Use sys/sys/random.h rather than a i386 specific one.
* There was nothing that should be machine dependant about
  i386/isa/random_machdep.c, so it is now sys/kern/kern_random.c.

24 years ago* Rewrite to use kobj(9) instead of hard-coded function tables.
Doug Rabson [Mon, 24 Apr 2000 17:08:04 +0000 (17:08 +0000)]
* Rewrite to use kobj(9) instead of hard-coded function tables.
* Report link errors to stdout with uprintf() so that the user can see
  what went wrong (PR kern/9214).
* Add support code to allow module symbols to be loaded into GDB using
  the debugger's "sharedlibrary" command.

24 years agoRemove a debugging define which I left by mistake.
Doug Rabson [Mon, 24 Apr 2000 16:59:46 +0000 (16:59 +0000)]
Remove a debugging define which I left by mistake.

24 years agoCosmetic to make errors messages look similar.
Duncan Barclay [Mon, 24 Apr 2000 15:49:20 +0000 (15:49 +0000)]
Cosmetic to make errors messages look similar.

24 years agoMoved debugging definitions to a header file.
Duncan Barclay [Mon, 24 Apr 2000 15:19:40 +0000 (15:19 +0000)]
Moved debugging definitions to a header file.

Removed NETBSDTX spares.

Renamed a load of functions - specifcally
start -> tx
init has download, sj and assoc as sub-functions
report_params and update params move to repparams/upparams

Tidied up old #define's

Got rid of old DPRINTF, made printf's RAY_PRINTF and panic's RAY_PANIC

All code KNF

Removed checking the ECF_TO_HOST area in ray_init - this gets hosed
when someone updates/read a parameter from the card so causes
unneeded grief. This required moving the tib check into ray_attach.

Changed handling of interface flags in ioctl so that promiscous mode
changes are only done as needed.

Sequences of comq entries can be added to an array and automatically
dealt with - used in ray_init_user and others.

Moved IFF_RUNNING checks from the comq commands to com_runq - still
not sure what to do so we PANIC - will be fixed.

ray_sj now checks to see if any of the parameters it can update can be
updated.

ray_sj_done now updates parameters to the current n/w set if we changed
them earlier. I was being a bit thick in earlier comments as to why this
check was done - the ECF never changes the parameters.

Assocication with APs is handled outside of ray_sj - need to add
WEP stuff if I ever get my hands on an AP.

ray_stop, ray_unload and ray_reset are currently broken - reset isn't used anyway.

ray_tx_XXX and ray_rx only have cosmetic changes.

Interrupt handler now gets command out of ccs. This is so ray_intr_ccs and
ray_intr_rcs don't need to remap CM (I'm trying to roll things up a bit as
it must be slowing us down).

ray_intr_ccs just vectors commands to functions or error checks - a
jump table could replace it in ray_intr.

Ditto for ray_intr_rcs.

mcast is currently broken - I need to do more work for ALLMULTI etc. This is reasonably easy to fix becuase of the comq array stuff.

Added a few more checks into repparams so that it is firmware version aware. It also uses the new comq framework.

Parameter updating is also much better with the update parameter return routine re-worked.

Added a couple of runq entry helper functions - one that malloc's a new
entry and fills it in with sensible defaults and another that adds and
runs a seqencue of entries.

24 years agoBit mask definitions for firmware versioning.
Duncan Barclay [Mon, 24 Apr 2000 14:51:25 +0000 (14:51 +0000)]
Bit mask definitions for firmware versioning.

Added versioning info to the MIB sizes array so that both raycontrol
and if_ray have a better chance of not sending duff data to the ECF.

24 years agoInitial import.
Duncan Barclay [Mon, 24 Apr 2000 14:50:01 +0000 (14:50 +0000)]
Initial import.

24 years agoMake the moved about version compile and run
Duncan Barclay [Mon, 24 Apr 2000 10:10:01 +0000 (10:10 +0000)]
Make the moved about version compile and run

24 years ago- Added support for 1bpp and 4bpp BMP files.
Kazutaka YOKOTA [Mon, 24 Apr 2000 10:09:42 +0000 (10:09 +0000)]
- Added support for 1bpp and 4bpp BMP files.
  (PC98 part of the commit will follow.)

Submitted (50%) by: Chiharu Shibata <chi@bd.mbn.or.jp>

24 years agoFix the grammar in my previous commit "lose" -> "loss".
Mike Pritchard [Mon, 24 Apr 2000 08:50:15 +0000 (08:50 +0000)]
Fix the grammar in my previous commit "lose" -> "loss".
I should have done it that way in the first place.

Pointed-out-by: bde
24 years agoActivate ja nls if WANT_KANJI
Andrey A. Chernov [Mon, 24 Apr 2000 08:14:16 +0000 (08:14 +0000)]
Activate ja nls if WANT_KANJI
Fix 'make release'
Suppress 'make distribute'

24 years agoAdd "device miibus" to SYNOPSIS section.
Bill Paul [Mon, 24 Apr 2000 02:01:56 +0000 (02:01 +0000)]
Add "device miibus" to SYNOPSIS section.

24 years agoMake FTP_PASSIVE_MODE check more specific.
Jordan K. Hubbard [Sun, 23 Apr 2000 21:23:34 +0000 (21:23 +0000)]
Make FTP_PASSIVE_MODE check more specific.
Suggested by: Eric D. Futch <efutch@nyct.net>