]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoRemove this binutils 2.8.1 file that was never part of any FreeBSD release.
obrien [Sat, 25 Nov 2000 02:08:34 +0000 (02:08 +0000)]
Remove this binutils 2.8.1 file that was never part of any FreeBSD release.

23 years agoRemove this binutils 2.8.1 file that was never part of any FreeBSD release.
obrien [Sat, 25 Nov 2000 02:02:01 +0000 (02:02 +0000)]
Remove this binutils 2.8.1 file that was never part of any FreeBSD release.

23 years agoIn env_destroy(), it is a bad idea to env_swap(self, 0) to switch
green [Sat, 25 Nov 2000 02:00:35 +0000 (02:00 +0000)]
In env_destroy(), it is a bad idea to env_swap(self, 0) to switch
back to the original environ unconditionally.  The setting of the
variable to save the previous environ is conditional; it happens when
ENV.e_committed is set.  Therefore, don't try to swap the env back
unless the previous env has been initialized.

PR: bin/22670
Submitted by: Takanori Saneto <sanewo@ba2.so-net.ne.jp>

23 years agoCorrect an arguement to ssh_add_identity, this matches what is currently
billf [Sat, 25 Nov 2000 01:55:42 +0000 (01:55 +0000)]
Correct an arguement to ssh_add_identity, this matches what is currently
in ports/security/openssh/files/pam_ssh.c

PR: 22164
Submitted by: Takanori Saneto <sanewo@ba2.so-net.ne.jp>
Reviewed by: green
Approved by: green

23 years agoDon't build libgcc_r if "NOLIBC_R" defined. If you don't have a threads
obrien [Sat, 25 Nov 2000 01:25:35 +0000 (01:25 +0000)]
Don't build libgcc_r if "NOLIBC_R" defined.  If you don't have a threads
safe libc, having a threads safe libgcc isn't of much use.

23 years agoPrint a warning when we detect a Realtek 8139B chip
roger [Fri, 24 Nov 2000 17:36:07 +0000 (17:36 +0000)]
Print a warning when we detect a Realtek 8139B chip
(identified by the IO map being 256 bytes long instead of 128)

This chip works very unreliably on my Lanner embedded PC with the rl driver.
Lots of watchdog timeouts or poor performance.

Forcing the media type to 10 Meg (ifconfig rl0 media 10baseT/UTP) is a good
workaround.

This looks very similar to the problem reported in PR kern/18790

It is interesting to note that the linux driver has lots of special
case code for this chip.

23 years agoAdd a module makefile for if_ep.
markm [Fri, 24 Nov 2000 17:04:51 +0000 (17:04 +0000)]
Add a module makefile for if_ep.

This is not enabled by default, as the module is missing a bunch of
useful stuff (like the ability to unload).

Not objected to by: mdodd

23 years agoEliminate groff(1) warnings.
ru [Fri, 24 Nov 2000 16:58:17 +0000 (16:58 +0000)]
Eliminate groff(1) warnings.

23 years agoThe default format for troff documentation is set with PRINTERDEVICE.
ru [Fri, 24 Nov 2000 14:08:07 +0000 (14:08 +0000)]
The default format for troff documentation is set with PRINTERDEVICE.

23 years agoMake this paper buildable in nroff mode, eliminate warnings,
ru [Fri, 24 Nov 2000 13:58:21 +0000 (13:58 +0000)]
Make this paper buildable in nroff mode, eliminate warnings,
do not force default printing device and compression mode.

23 years agomdoc(7) police: minor fixes.
ru [Fri, 24 Nov 2000 11:39:39 +0000 (11:39 +0000)]
mdoc(7) police: minor fixes.

23 years agoFix a layout problem (.It Li -> .Fx)
ben [Fri, 24 Nov 2000 11:17:28 +0000 (11:17 +0000)]
Fix a layout problem (.It Li -> .Fx)

PR: 23055
Submitted by: Jimmy Olgeni <olgeni@uli.it>

23 years ago(forced commit)
sheldonh [Fri, 24 Nov 2000 11:13:15 +0000 (11:13 +0000)]
(forced commit)

More detail describing why chmod is installed into ${INSTALLTMP} since
rev 1.176:

We use Perl's MakeMaker-generated makefiles to install Perl.  Since we
can't hack up MakeMaker's MM_Unix.pm to suit our build, we need chmod
to be around until we use our own makefiles to install perl.

PR: gnu/22025

23 years agoRemove the only hard sentence break in the file.
sheldonh [Fri, 24 Nov 2000 10:48:02 +0000 (10:48 +0000)]
Remove the only hard sentence break in the file.

23 years agoDocument EI_OSABI and EI_ABIVERSION.
asmodai [Fri, 24 Nov 2000 10:42:21 +0000 (10:42 +0000)]
Document EI_OSABI and EI_ABIVERSION.

Requested by: obrien

23 years agoMention mergemaster(8).
ben [Fri, 24 Nov 2000 10:22:08 +0000 (10:22 +0000)]
Mention mergemaster(8).

Submitted by: obrien

23 years agoLet the pwd program double as realpath(1).
phk [Fri, 24 Nov 2000 10:18:52 +0000 (10:18 +0000)]
Let the pwd program double as realpath(1).

This lets you resolve pathnames to their underlying physical path:

critter# realpath /sys/kern/subr_disk.c
/freebsd/src/sys/kern/subr_disk.c

Update the pwd man-page slightly.

23 years agoReimplement the groff(1) warnings elimination fixes in a better way.
ru [Fri, 24 Nov 2000 10:05:30 +0000 (10:05 +0000)]
Reimplement the groff(1) warnings elimination fixes in a better way.

23 years agoEliminate groff(1) warnings.
ru [Fri, 24 Nov 2000 09:33:37 +0000 (09:33 +0000)]
Eliminate groff(1) warnings.

23 years ago[rsh.1] Document the `-4' and `-6' options.
jkoshy [Fri, 24 Nov 2000 09:25:44 +0000 (09:25 +0000)]
[rsh.1] Document the `-4' and `-6' options.
[rsh.c] Make usage message match the code.

23 years ago[rlogin.1] Document the `-4' and `-6' options. Correct a typo.
jkoshy [Fri, 24 Nov 2000 09:14:09 +0000 (09:14 +0000)]
[rlogin.1] Document the `-4' and `-6' options.   Correct a typo.
[rlogin.c] Make the usage message match the code.

23 years agoFixup (hopefully) bridging + ipfw + dummynet together...
bmilekic [Thu, 23 Nov 2000 22:25:03 +0000 (22:25 +0000)]
Fixup (hopefully) bridging + ipfw + dummynet together...

* Some dummynet code incorrectly handled a malloc()-allocated pseudo-mbuf
  header structure, called "pkt," and could consequently pollute the mbuf
  free list if it was ever passed to m_freem(). The fix involved passing not
  pkt, but essentially pkt->m_next (which is a real mbuf) to the mbuf
  utility routines.

* Also, for dummynet, in bdg_forward(), made the code copy the ethernet header
  back into the mbuf (prepended) because the dummynet code that follows expects
  it to be there but it is, unfortunately for dummynet, passed to bdg_forward
  as a seperate argument.

PRs: kern/19551 ; misc/21534 ; kern/23010
Submitted by: Thomas Moestl <tmoestl@gmx.net>
Reviewed by: bmilekic
Approved by: luigi

23 years agoAdd a brief warning about devclass and multiple device driver attachments.
dcs [Thu, 23 Nov 2000 17:19:54 +0000 (17:19 +0000)]
Add a brief warning about devclass and multiple device driver attachments.

23 years agoEliminate groff(1) warnings.
ru [Thu, 23 Nov 2000 15:21:30 +0000 (15:21 +0000)]
Eliminate groff(1) warnings.

23 years agoUse a more generic name for the compiler cross-reference. (cc instead of
ben [Thu, 23 Nov 2000 14:44:45 +0000 (14:44 +0000)]
Use a more generic name for the compiler cross-reference.  (cc instead of
gcc.)

Suggested by: sheldonh (not sholdonh)

23 years ago* Say that ps, top, etc, "may not work" with the new kernel, rather than
ben [Thu, 23 Nov 2000 14:19:50 +0000 (14:19 +0000)]
* Say that ps, top, etc, "may not work" with the new kernel, rather than
  "will not work", as in many cases they will.

Suggested by: sholdonh

* Add cross-references to reboot(8) and shutdown(8)

23 years agoAdd an example to explain the basic procedure to build the world.
ben [Thu, 23 Nov 2000 14:00:26 +0000 (14:00 +0000)]
Add an example to explain the basic procedure to build the world.

Suggested by: obrien

23 years agoUpdate for the Bintuils 2.10.1 import.
obrien [Thu, 23 Nov 2000 12:28:19 +0000 (12:28 +0000)]
Update for the Bintuils 2.10.1 import.

23 years agokenv(1) first appeared in FreeBSD 4.1.1, not FreeBSD 5.0.
ben [Thu, 23 Nov 2000 11:21:35 +0000 (11:21 +0000)]
kenv(1) first appeared in FreeBSD 4.1.1, not FreeBSD 5.0.

PR: 23020
Submitted by: Simon Dick <simond@irrelevant.org>

23 years agoAdd the appropriate magic to make .Fx 4.1.1 work.
ben [Thu, 23 Nov 2000 11:17:33 +0000 (11:17 +0000)]
Add the appropriate magic to make .Fx 4.1.1 work.

23 years agoEliminate groff(1) warnings.
ru [Thu, 23 Nov 2000 11:08:17 +0000 (11:08 +0000)]
Eliminate groff(1) warnings.

23 years agoForgot to patch this file in file descriptor race fix commit
dillon [Thu, 23 Nov 2000 11:05:14 +0000 (11:05 +0000)]
Forgot to patch this file in file descriptor race fix commit

Submitted-by: "Danny J. Zerkel" <dzerkel@columbus.rr.com>
23 years agoEliminate groff(1) warnings.
ru [Thu, 23 Nov 2000 10:15:50 +0000 (10:15 +0000)]
Eliminate groff(1) warnings.

Obtained from: NetBSD

23 years agoo Change the argument of linux_sigreturn to be a pointer to a
marcel [Thu, 23 Nov 2000 08:55:30 +0000 (08:55 +0000)]
o  Change the argument of linux_sigreturn to be a pointer to a
   struct sigframe. We need more than only the signal context.

o  Properly convert the signal mask when setting up the signal
   frame in linux_sendsig and properly convert it back in
   linux_sigreturn.

Do some cleanups and improve style while here.

23 years agoChange the argument of linux_sigreturn to be a pointer to a
marcel [Thu, 23 Nov 2000 08:53:19 +0000 (08:53 +0000)]
Change the argument of linux_sigreturn to be a pointer to a
struct sigframe.

23 years agoStarting from groff(1) version 1.12, -msafer is now the default,
ru [Thu, 23 Nov 2000 08:51:33 +0000 (08:51 +0000)]
Starting from groff(1) version 1.12, -msafer is now the default,
but index building for 13.viref requires the use of .sy request.
Repair the index building by running groff(1) in unsafe -U mode.

Requested by: bde

23 years agomdoc(7) police: misc minor fixes.
ru [Thu, 23 Nov 2000 08:18:10 +0000 (08:18 +0000)]
mdoc(7) police: misc minor fixes.

23 years agoRemove extra ")".
kuriyama [Thu, 23 Nov 2000 08:09:45 +0000 (08:09 +0000)]
Remove extra ")".

23 years agoAdd a forgotten mutex_exit()/KUE_UNLOCK() to kue_init().
wpaul [Thu, 23 Nov 2000 05:58:48 +0000 (05:58 +0000)]
Add a forgotten mutex_exit()/KUE_UNLOCK() to kue_init().

23 years agoDon't unregister the ioctl handlers before we verified we
marcel [Thu, 23 Nov 2000 03:21:58 +0000 (03:21 +0000)]
Don't unregister the ioctl handlers before we verified we
can unload. Doing so leaves the linuxulator in a crippled
state (no ioctl support) when Linux binaries are run at
unload time.

While here, consistently spell ELF in capitals and perform
some minor style improvements.

ELF spelling submitted by: asmodai

23 years agoCorrect the pathname of the history file.
grog [Thu, 23 Nov 2000 01:41:11 +0000 (01:41 +0000)]
Correct the pathname of the history file.

Reported by: Crist J . Clark <cjclark@alum.mit.edu>

23 years agoNew release notes: Thread supports under Linux emulation on alpha,
bmah [Thu, 23 Nov 2000 01:22:59 +0000 (01:22 +0000)]
New release notes:  Thread supports under Linux emulation on alpha,
an(4)/ancontrol(8) WEP, ng_one2many(4), softupdates bugfixes/
enhancements, OpenSSH agent/X forwarding bug fixed, ssh(1) no longer
SUID root, OpenSSL->0.9.6, binutils->2.10.1, lastlogin(1),
newfs(1) write combining (+MFC), config(8) buffer overflows fixed,
pcm(4) drivers compiled into GENERIC by default.

Other MFCs:  Write combining for crash dumps.

Delete a duplicated note on mergemaster(8).

23 years agoFix the KTR tracepoints for mtx_enter/exit/try_enter to properly order the
jhb [Thu, 23 Nov 2000 00:38:45 +0000 (00:38 +0000)]
Fix the KTR tracepoints for mtx_enter/exit/try_enter to properly order the
parameters for the KTR_EXTEND case.

23 years agoUpdate the probe some more to deal with 16/32 bit issues. If the chip
wpaul [Thu, 23 Nov 2000 00:28:43 +0000 (00:28 +0000)]
Update the probe some more to deal with 16/32 bit issues. If the chip
is already in 32-bit mode, we need to be able to detect this and still
read the chip ID code. Detecting 32-bit mode is actually a little
tricky, since we want to avoid turning it on accidentally. The easiest
way to do it is to just try and read the PCI subsystem ID from the
bus control registers using 16-bit accesses and compare that with the
value read from PCI config space. If they match, then we know we're in
16-bit mode, otherwise we assume 32-bit mode.

23 years agoFinish the job of conditionalizing the PC98 stuff.
jkh [Wed, 22 Nov 2000 22:28:34 +0000 (22:28 +0000)]
Finish the job of conditionalizing the PC98 stuff.

23 years agoRemove "NODESCRYPTLINKS" it has been superseeded by "passwd_format"
obrien [Wed, 22 Nov 2000 20:37:40 +0000 (20:37 +0000)]
Remove "NODESCRYPTLINKS" it has been superseeded by "passwd_format"
in /etc/login.conf.

23 years agoAdd one2many subdirectory (previously forgotten).
archie [Wed, 22 Nov 2000 19:04:42 +0000 (19:04 +0000)]
Add one2many subdirectory (previously forgotten).

23 years agoo fix a typo in copyright part: mdoc comment starts with .\", not ,\"
phantom [Wed, 22 Nov 2000 18:00:50 +0000 (18:00 +0000)]
o fix a typo in copyright part: mdoc comment starts with .\", not ,\"
o few mdoc(7) fixes

PR: docs/13218

23 years agoremove bogus Xref to lsdev(8)
phantom [Wed, 22 Nov 2000 17:57:44 +0000 (17:57 +0000)]
remove bogus Xref to lsdev(8)

PR: docs/13218

23 years agoo get rid lfs mentions
phantom [Wed, 22 Nov 2000 17:56:21 +0000 (17:56 +0000)]
o get rid lfs mentions
o remove non-mdoc groff's instructions

23 years agoWe do not support lfs. Remove it from list of remountable fs'es.
phantom [Wed, 22 Nov 2000 17:54:56 +0000 (17:54 +0000)]
We do not support lfs. Remove it from list of remountable fs'es.

23 years agoUse correct macro for path name
phantom [Wed, 22 Nov 2000 17:53:17 +0000 (17:53 +0000)]
Use correct macro for path name

PR: docs/13218

23 years agomdoc(7) police: Er macro usage cleanup.
ru [Wed, 22 Nov 2000 16:11:48 +0000 (16:11 +0000)]
mdoc(7) police: Er macro usage cleanup.

23 years agomdoc(7) police: Er macro usage cleanup.
ru [Wed, 22 Nov 2000 16:02:00 +0000 (16:02 +0000)]
mdoc(7) police: Er macro usage cleanup.

23 years agomdoc(7) police: Er macro usage cleanup.
ru [Wed, 22 Nov 2000 15:55:01 +0000 (15:55 +0000)]
mdoc(7) police: Er macro usage cleanup.

23 years agoDon't go haywire if the server closes the connection in the middle of a
des [Wed, 22 Nov 2000 14:50:46 +0000 (14:50 +0000)]
Don't go haywire if the server closes the connection in the middle of a
multiline response (proper fix this time).

23 years agoRevert previous commit, it was somewhat hasty.
des [Wed, 22 Nov 2000 14:44:48 +0000 (14:44 +0000)]
Revert previous commit, it was somewhat hasty.

23 years agoReduce number of #ifdef nestings.
asmodai [Wed, 22 Nov 2000 14:38:30 +0000 (14:38 +0000)]
Reduce number of #ifdef nestings.

Submitted by: bde

23 years agoDon't go haywire if the server closes the connection during a multiline
des [Wed, 22 Nov 2000 14:30:28 +0000 (14:30 +0000)]
Don't go haywire if the server closes the connection during a multiline
response.

23 years agoAdd a note asking committers to update make.conf.5 when they update this.
ben [Wed, 22 Nov 2000 14:11:21 +0000 (14:11 +0000)]
Add a note asking committers to update make.conf.5 when they update this.

23 years agoAdd two new manual pages to document the various options available in
ben [Wed, 22 Nov 2000 14:04:26 +0000 (14:04 +0000)]
Add two new manual pages to document the various options available in
make.conf, and the /usr/src building process.

PR: 22676
Submitted by: Mike Meyer <mwm@mired.org>
Reviewed by: nik, sheldonh

23 years agomdoc(7) police: fixed typo.
ru [Wed, 22 Nov 2000 13:49:11 +0000 (13:49 +0000)]
mdoc(7) police: fixed typo.

23 years agomdoc(7) police: add the missing and required .Dd macro call.
ru [Wed, 22 Nov 2000 13:31:01 +0000 (13:31 +0000)]
mdoc(7) police: add the missing and required .Dd macro call.

23 years agoCreate temporary files in a secure directory, instead of using multiple
kris [Wed, 22 Nov 2000 11:09:30 +0000 (11:09 +0000)]
Create temporary files in a secure directory, instead of using multiple
filenames based on a single invocation of mktemp() in /tmp, which is
easily predictable after the first one.

Audited by:     markm

23 years agoUse secure temporary filenames during build.
kris [Wed, 22 Nov 2000 11:05:51 +0000 (11:05 +0000)]
Use secure temporary filenames during build.

Audited by: markm, cracauer

23 years ago(null commit)
ru [Wed, 22 Nov 2000 09:50:40 +0000 (09:50 +0000)]
(null commit)

Previous commit has wrong log message.  The correct one is:

mdoc(7) police: do not split author names in the AUTHORS section.

23 years agomdoc(7) police: do not split author names in the AUTHORS section.
ru [Wed, 22 Nov 2000 09:35:58 +0000 (09:35 +0000)]
mdoc(7) police: do not split author names in the AUTHORS section.

23 years agolog
ru [Wed, 22 Nov 2000 09:23:54 +0000 (09:23 +0000)]
log

23 years agomdoc(7) police: use the new feature of the An macro.
ru [Wed, 22 Nov 2000 08:47:35 +0000 (08:47 +0000)]
mdoc(7) police: use the new feature of the An macro.

23 years agoMake An macro understand two special arguments, -nosplit
ru [Wed, 22 Nov 2000 08:39:17 +0000 (08:39 +0000)]
Make An macro understand two special arguments, -nosplit
and -split.  See the mdoc.samples(7) manpage for details.

Reviewed by: sheldonh

23 years agomdoc(7) police: minor cleanup.
ru [Wed, 22 Nov 2000 08:19:15 +0000 (08:19 +0000)]
mdoc(7) police: minor cleanup.

23 years agoProtect the following with a lockmgr lock:
jake [Wed, 22 Nov 2000 07:42:04 +0000 (07:42 +0000)]
Protect the following with a lockmgr lock:

allproc
zombproc
pidhashtbl
proc.p_list
proc.p_hash
nextpid

Reviewed by: jhb
Obtained from: BSD/OS and netbsd

23 years agoProtect the bounce page list with a mutex instead of using splhigh().
dfr [Wed, 22 Nov 2000 07:16:51 +0000 (07:16 +0000)]
Protect the bounce page list with a mutex instead of using splhigh().

23 years agoInstall DELAY(9).
alfred [Wed, 22 Nov 2000 03:44:55 +0000 (03:44 +0000)]
Install DELAY(9).

23 years agoDocument the DELAY() function.
alfred [Wed, 22 Nov 2000 03:38:32 +0000 (03:38 +0000)]
Document the DELAY() function.

23 years agoAdd `password_format=md5' to the default settings. I've had this locally
obrien [Tue, 21 Nov 2000 23:25:39 +0000 (23:25 +0000)]
Add `password_format=md5' to the default settings.  I've had this locally
for a while, but a recent email to -stable suggests it should be spelled
out as the documentation of "password_format" is sparse.

Also add a `des_users' entry.
Submitted by: Sean O'Connell <sean@stat.Duke.EDU>

23 years agoFix a logic reversal: the RFFDG flag must be unset in order to share
jlemon [Tue, 21 Nov 2000 22:40:59 +0000 (22:40 +0000)]
Fix a logic reversal: the RFFDG flag must be unset in order to share
a file descriptor table.  (Thanks to Alan Cox)

Clarify the fact that the changelist and eventlist can be shared.

23 years agoAdd description of M_WRITABLE macro and new M_RDONLY flag. Try my best to
bmilekic [Tue, 21 Nov 2000 22:29:31 +0000 (22:29 +0000)]
Add description of M_WRITABLE macro and new M_RDONLY flag. Try my best to
conform to line break mdoc FreeBSD standards, although I am new to this.

Reviewed by: sheldonh

23 years agoAhem, fix the disclaimer portion of the copyright so it disclaim's the
jhb [Tue, 21 Nov 2000 21:10:15 +0000 (21:10 +0000)]
Ahem, fix the disclaimer portion of the copyright so it disclaim's the
voices in my head.  You can sue the voices in Bill Paul's head all you
want.

Noticed by: jhb

23 years agoDevices that are running at splimp MUST use ng_queue_data()
julian [Tue, 21 Nov 2000 20:48:54 +0000 (20:48 +0000)]
Devices that are running at splimp MUST use ng_queue_data()
instead of ng_send_data().
The latter could lead to running the IP stack at splimp
instead of splnet, (among other problems) (that MAY be safe
but I wouldn't count on it).

Noticed while preparing a new set of netgraph stuff.

23 years agoAdd new man page to the mix.
markm [Tue, 21 Nov 2000 20:31:51 +0000 (20:31 +0000)]
Add new man page to the mix.

23 years agoWhitespace-only change. Tidy up, alphabetise and wrap.
markm [Tue, 21 Nov 2000 20:28:48 +0000 (20:28 +0000)]
Whitespace-only change. Tidy up, alphabetise and wrap.

23 years agoProtect p_wchan with sched_lock in selwakeup().
jlemon [Tue, 21 Nov 2000 20:22:34 +0000 (20:22 +0000)]
Protect p_wchan with sched_lock in selwakeup().

23 years agoAssembler fixes.
markm [Tue, 21 Nov 2000 20:16:49 +0000 (20:16 +0000)]
Assembler fixes.

Fix opcodes that were typed as ".byte 0xNN, 0xMM" when an older
assembler could not recognise the newer Pentium instructions.
Reviewed by: jhb

23 years agoFix buffer overflows in filenames. If you had a path > 80 characters
imp [Tue, 21 Nov 2000 19:58:55 +0000 (19:58 +0000)]
Fix buffer overflows in filenames.  If you had a path > 80 characters
for your /usr/obj/path/to/my/files path to the kernel, then weird
things happened.  make buildkernel would fail because config was
dumping core or generating bad file names (depending on the lenght of
the path).

While I was here, also use strlcpy, strlcat and snprintf (or asprintf)
as necessary.  Minor format policing for the snprintf calls as well.

23 years agoAdd a consistent API to a feature that most modern CPUs have; a fast
markm [Tue, 21 Nov 2000 19:55:21 +0000 (19:55 +0000)]
Add a consistent API to a feature that most modern CPUs have; a fast
counter register in-CPU.

This is to be used as a fast "timer", where linearity is more important
than time, and multiple lines in the linearity caused by multiple CPUs
in an SMP machine is not a problem.

This adds no code whatsoever to the FreeBSD kernel until it is actually
used, and then as a single-instruction inline routine (except for the
80386 and 80486 where it is some more inline code around nanotime(9).

Reviewed by: bde, kris, jhb

23 years agoProvide a new interface for the user of aio_read() and aio_write() to request
alc [Tue, 21 Nov 2000 19:36:36 +0000 (19:36 +0000)]
Provide a new interface for the user of aio_read() and aio_write() to request
a kevent upon completion of the I/O.  Specifically, introduce a new type
of sigevent notification, SIGEV_EVENT.  If sigev_notify is SIGEV_EVENT,
then sigev_notify_kqueue names the kqueue that should receive the event
and sigev_value contains the "void *" is copied into the kevent's udata
field.

In contrast to the existing interface, this one: 1) works on
the Alpha 2) avoids the extra copyin() call for the kevent because all
of the information needed is in the sigevent and 3) could be
applied to request a single kevent upon completion of an entire lio_listio().

Reviewed by: jlemon

23 years agoStop handcoding a couple of instructions since gas 2.10 can properly
jhb [Tue, 21 Nov 2000 18:36:10 +0000 (18:36 +0000)]
Stop handcoding a couple of instructions since gas 2.10 can properly
assemble 16-bit code.

Noticed by: markm

23 years agomdoc(7) police: Add a missing `.Sm on' request.
ru [Tue, 21 Nov 2000 18:18:46 +0000 (18:18 +0000)]
mdoc(7) police: Add a missing `.Sm on' request.

23 years agoClarify the time parameter of the kevent() system call.
asmodai [Tue, 21 Nov 2000 15:22:36 +0000 (15:22 +0000)]
Clarify the time parameter of the kevent() system call.

Inspired by PR: 21708
Submitted by: Ronald F. Guilmette <rfg@monkeys.com>
Reviewed by: jlemon

23 years agoReflect reality about number of developers.
kuriyama [Tue, 21 Nov 2000 14:31:10 +0000 (14:31 +0000)]
Reflect reality about number of developers.

Submitted by: sheldonh

23 years agoBump the buffersize from 0x1000 to 0x4000.
julian [Tue, 21 Nov 2000 12:22:11 +0000 (12:22 +0000)]
Bump the buffersize from 0x1000 to 0x4000.
0x1000 gets underflows on my dell 7500 whenever I moved the mouse.
(reported as "hwptr went backwards...", with jumpy sound)
Sounds great now....

23 years agoRemove text saying "this is available from the compatibility library,
ben [Tue, 21 Nov 2000 12:03:30 +0000 (12:03 +0000)]
Remove text saying "this is available from the compatibility library,
libcompat" in favour of a .Sh LIBRARY section using the .Lb macro.  Also add
.Bf -symbolic around the text saying "this is obsolete" in re_comp.3.

PR: 22675
Submitted by: Mike Meyer <mwm@mired.org>
Reviewed by: sheldonh

23 years agomdoc(7) police: fixed the section reference to the mdoc.samples(7).
ru [Tue, 21 Nov 2000 08:12:23 +0000 (08:12 +0000)]
mdoc(7) police: fixed the section reference to the mdoc.samples(7).

23 years agoadd Linksts EtherFast 10/100 PC Card (PCMPC100 V2)
sanpei [Tue, 21 Nov 2000 07:01:24 +0000 (07:01 +0000)]
add Linksts EtherFast 10/100 PC Card (PCMPC100 V2)

Submitted by: Ben Lutgens <blutgens@usinternet.com>

23 years agoDon't test the return value from the actual command when determining
msmith [Tue, 21 Nov 2000 06:58:24 +0000 (06:58 +0000)]
Don't test the return value from the actual command when determining
success/failure for submission of the command.  The caller will want
to deal with this itself.

23 years agoIf a user is in the database, rmuser ought to be able to remove them.
dougb [Tue, 21 Nov 2000 05:52:35 +0000 (05:52 +0000)]
If a user is in the database, rmuser ought to be able to remove them.
Address this by using getpwnam(), thus killing several birds with
the same stone. My fix is slightly more aggressive than the
originators. :)

PR: misc/22278

23 years agoFix release, broken by the perl cross-build fixes.
marcel [Tue, 21 Nov 2000 04:37:30 +0000 (04:37 +0000)]
Fix release, broken by the perl cross-build fixes.

The distribute target is basicly the same as an install. For
perl, this means that miniperl is needed. Since miniperl is
only present in the object directory, we need to make sure
the path is set correctly. To do this, we have make release
use a new distribworld target that sets the path before doing
a make distribute.

23 years agoDo a better job of preserving the value rc is called with
dougb [Tue, 21 Nov 2000 04:21:54 +0000 (04:21 +0000)]
Do a better job of preserving the value rc is called with
in the face of possible corruption by rc.diskless*.

PR: misc/21904
Submitted by: Anton Voronin, anton@urc.ac.ru

23 years agofix isa DMA on pyxis based machines:
gallatin [Tue, 21 Nov 2000 03:25:31 +0000 (03:25 +0000)]
fix isa DMA on pyxis based machines:
- move the call to cia_init_sgmap() to after we've determined if we're a pyxis
- convert needed splhigh() in cia_sgmap_invalidate_pyxis() to disable_intr()

Previously, any isa DMA on a pyxis based machine would cause a panic
in cia_sgmap_invalidate_pyxis() because the pyxis workaround was never
setup.

- while i'm at it, convert needed splhigh() in cia_swiz_set_hae_mem to
disable_intr()