]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoAllow alpha kernels to compile again- make sure opt_ddb.h is included
Matt Jacob [Thu, 17 Jan 2002 02:16:35 +0000 (02:16 +0000)]
Allow alpha kernels to compile again- make sure opt_ddb.h is included
and the reference to db_regs is *extern* from alpha/include/db_machdep.h
(put it in alpha/alpha/machdep.c)- this avoids the problems we've had
about different 'common' sizes prohibiting the kernel from linking.

22 years agopid is 'long' on alpha.
Matt Jacob [Thu, 17 Jan 2002 02:14:44 +0000 (02:14 +0000)]
pid is 'long' on alpha.

22 years agoMerge the changes in the English version (1.254 --> 1.255).
Jun Kuriyama [Thu, 17 Jan 2002 02:09:43 +0000 (02:09 +0000)]
Merge the changes in the English version (1.254 --> 1.255).

22 years agoforced commit for previous revision:
Alfred Perlstein [Thu, 17 Jan 2002 00:12:56 +0000 (00:12 +0000)]
forced commit for previous revision:

PR: bin/33897

22 years agoFix boundry condition in lock management:
Alfred Perlstein [Thu, 17 Jan 2002 00:12:05 +0000 (00:12 +0000)]
Fix boundry condition in lock management:

 Alfred, I took a look at retry_blockingfilelocklist() and the
 solution seemed  simple enough. Please correct me if I am wrong.
 It seems said routine doesn't  take into account boundary conditions
 when putting back file_lock entries into the blocked lock-list.
 Specifically, it fails when the file_lock being put back is the
 last element in the list, and when it is the only element in the
 list.  I've included a patch below.

 Basically, it introduces another variable: pfl, which keeps track
 of the list  item before ifl. That way if nfl is NULL, ifl gets
 inserted after pfl. If pfl  is also NULL, then it gets inserted
 at the head of the list (since it was the  only element in the
 list).

Submitted by: Mike Makonnen <mike_makonnen@yahoo.com>
Tested by: Thomas Quinot <thomas@cuivre.fr.eu.org>

22 years agoo Improve the line-wrapping of additional comments, some of which appeared
Robert Watson [Thu, 17 Jan 2002 00:10:28 +0000 (00:10 +0000)]
o Improve the line-wrapping of additional comments, some of which appeared
  to be wrapped around 60, others around 40 columns.

22 years agoo Remove a somewhat less comprehensible comment about modifying /etc/rc.
Robert Watson [Thu, 17 Jan 2002 00:08:44 +0000 (00:08 +0000)]
o Remove a somewhat less comprehensible comment about modifying /etc/rc.
o Improve line-wrapping of another comment for consistency.

22 years agoForgot one part of the VLAN support for the dc(4) driver.
Doug Ambrisko [Wed, 16 Jan 2002 21:34:11 +0000 (21:34 +0000)]
Forgot one part of the VLAN support for the dc(4) driver.

Pointed out by: Shin-ichi YOSHIMOTO <yosimoto@waishi.jp>

22 years agoFix giant handling in pwrite(2), I forgot to release it when finishing
Alfred Perlstein [Wed, 16 Jan 2002 21:33:41 +0000 (21:33 +0000)]
Fix giant handling in pwrite(2), I forgot to release it when finishing
the syscall.

22 years agoFix typo by s/advertise/advertises/
Jens Schweikhardt [Wed, 16 Jan 2002 19:19:09 +0000 (19:19 +0000)]
Fix typo by s/advertise/advertises/
MFC after: 4 weeks

22 years agoUpdate release note: kern.security.* sysctls have moved to
Bruce A. Mah [Wed, 16 Jan 2002 16:45:29 +0000 (16:45 +0000)]
Update release note:  kern.security.* sysctls have moved to
security.*.

Requested by: arr

22 years agoAdd VLAN for the dc(4) driver (ie long frame). The patch is 2 parts.
Doug Ambrisko [Wed, 16 Jan 2002 16:33:58 +0000 (16:33 +0000)]
Add VLAN for the dc(4) driver (ie long frame).  The patch is 2 parts.
One to notify the system that the MTU for VLAN can be 1500 so the vlan
will automatically be configured with a 1500 MTU the other is to ignore
the error case if the received frame is to long.

The frame size notification came from code in the SIS driver, and
the support for long frames derived from the NetBSD Tulip driver.

Tested on: 4 port D-Link adapter DFE-570TX 4 Intel 21143
Netgear card with 82c169 PNIC 10/100BaseTX

Reviewed by: ru (manpage), wpaul (not objected to), archie
Approved by: imp
Obtained from: NetBSD

22 years agoDocument fetch(1) reaction to SIGINFO.
Ruslan Ermilov [Wed, 16 Jan 2002 16:16:08 +0000 (16:16 +0000)]
Document fetch(1) reaction to SIGINFO.

PR: docs/33640
Submitted by: Maxim Konovalov <maxim@macomnet.ru>
Approved by: des

22 years agomdoc(7) police: tidy up OpenBSD fixes.
Ruslan Ermilov [Wed, 16 Jan 2002 15:21:39 +0000 (15:21 +0000)]
mdoc(7) police: tidy up OpenBSD fixes.

22 years agoReplace reference to ANSI C draft with standard.
Ruslan Ermilov [Wed, 16 Jan 2002 14:55:18 +0000 (14:55 +0000)]
Replace reference to ANSI C draft with standard.

PR: docs/30731

22 years agoAdd extern to avoid sloppy common style declarations.
Jake Burkholder [Wed, 16 Jan 2002 14:28:50 +0000 (14:28 +0000)]
Add extern to avoid sloppy common style declarations.

Tripped over by: jhb, mux@sneakerz.org

22 years agoDocument the -x debugging option.
Ruslan Ermilov [Wed, 16 Jan 2002 14:12:02 +0000 (14:12 +0000)]
Document the -x debugging option.

PR: docs/33784
Submitted by: Mike Makonnen <mike_makonnen@yahoo.com>

22 years agosocket()s first argument should be a protocol family rather than an
Brian Somers [Wed, 16 Jan 2002 14:03:52 +0000 (14:03 +0000)]
socket()s first argument should be a protocol family rather than an
address family.

22 years agosocket's first argument is an address family, not a protocol family.
Brian Somers [Wed, 16 Jan 2002 13:15:47 +0000 (13:15 +0000)]
socket's first argument is an address family, not a protocol family.

22 years agoAdded this makefile. This is not attached to the build yet. I often
Bruce Evans [Wed, 16 Jan 2002 12:18:22 +0000 (12:18 +0000)]
Added this makefile.  This is not attached to the build yet.  I often
install parts of /etc manually and it helps to have a makefile for
each subdir even if the main makefile doesn't invoke it.

22 years agoMake `options NETATALK' compile after the -fno-common changes.
Ian Dowse [Wed, 16 Jan 2002 11:26:50 +0000 (11:26 +0000)]
Make `options NETATALK' compile after the -fno-common changes.

Tested by: Emiel Kollof <coolvibe@hackerheaven.org>

22 years agoDon't print out 0xaa55 at the end of a vnconfig disk and use the
John Baldwin [Wed, 16 Jan 2002 07:24:38 +0000 (07:24 +0000)]
Don't print out 0xaa55 at the end of a vnconfig disk and use the
"canonical" disk name when calling disklabel so that disklabel auto works.

22 years ago- Attempt to help declutter kern. sysctl by moving security out from
Andrew R. Reiter [Wed, 16 Jan 2002 06:55:30 +0000 (06:55 +0000)]
- Attempt to help declutter kern. sysctl by moving security out from
  beneath it.

Reviewed by: rwatson

22 years agoPut write on read-only filesystem panic after we have weeded out
Kirk McKusick [Wed, 16 Jan 2002 04:59:09 +0000 (04:59 +0000)]
Put write on read-only filesystem panic after we have weeded out
block and character devices, fifo's, etc.

Submitted by: Bruce Evans <bde@zeta.org.au>

22 years agoChange <b28> to HTT (Hyperthreading technology). If this flag is set then
Peter Wemm [Wed, 16 Jan 2002 02:22:19 +0000 (02:22 +0000)]
Change <b28> to HTT (Hyperthreading technology).  If this flag is set then
cpuid with %eax=1 will return a logical cpu count in bits 16-23 of %ebx.
Bit 29 is actually 'TM' according to AP-485.  This signifies the presence
of the thermal control circuit (which I believe can slow the clock down
to reduce core temperature).

22 years agofix typo, there's uap, just fd
Alfred Perlstein [Wed, 16 Jan 2002 01:31:19 +0000 (01:31 +0000)]
fix typo, there's uap, just fd

22 years agoEnsure that we set all the %cr0 bits to a known state for the AP's before
Peter Wemm [Wed, 16 Jan 2002 00:44:29 +0000 (00:44 +0000)]
Ensure that we set all the %cr0 bits to a known state for the AP's before
they make it through to userland.  This should fix the p5-smp problem
without affecting the other cpus (eg: cyrix, see initcpu.c and the special
cache handling for these cpu types).

22 years agoRevise the nfsiod auto tuning code. Now both the upper and lower limits
Peter Wemm [Tue, 15 Jan 2002 20:57:21 +0000 (20:57 +0000)]
Revise the nfsiod auto tuning code.  Now both the upper and lower limits
are specifyable by sysctl and are respected.

Submitted by: Maxime Henrion <mux@sneakerz.org>

22 years agobetter namespsace of static functions
Warner Losh [Tue, 15 Jan 2002 20:41:39 +0000 (20:41 +0000)]
better namespsace of static functions

22 years agoo Sync the device list with OLDCARD.
Warner Losh [Tue, 15 Jan 2002 20:21:43 +0000 (20:21 +0000)]
o Sync the device list with OLDCARD.
o Remove bogus flags that aren't used (if we need them in the future, we can
  add them back).
o Add support for the TI-1031.  This is the only YENTA compatible PCI-PCMCIA
  bridge that I'm aware of (all the others are PCIC on a PCI bus, which is
  different).

22 years agoThis commit was generated by cvs2svn to compensate for changes in r89402,
Jacques Vidrine [Tue, 15 Jan 2002 19:25:55 +0000 (19:25 +0000)]
This commit was generated by cvs2svn to compensate for changes in r89402,
which included commits to RCS files with non-trunk default branches.

22 years agoDon't use getlogin() to determine whether we are root.
Jacques Vidrine [Tue, 15 Jan 2002 19:25:55 +0000 (19:25 +0000)]
Don't use getlogin() to determine whether we are root.
(Import of vendor fix.)

22 years agoAdd a few cleanups from rev 1.1:
Mike Barcroft [Tue, 15 Jan 2002 17:52:21 +0000 (17:52 +0000)]
Add a few cleanups from rev 1.1:
o Restore vendor ID.
o Order variable types by size.
o Remove a gratuitous temporary variable.

Submitted by: bde

22 years agoThis commit was generated by cvs2svn to compensate for changes in r89397,
Ruslan Ermilov [Tue, 15 Jan 2002 17:15:28 +0000 (17:15 +0000)]
This commit was generated by cvs2svn to compensate for changes in r89397,
which included commits to RCS files with non-trunk default branches.

22 years agoSatisfy pedantic readers.
Ruslan Ermilov [Tue, 15 Jan 2002 17:15:28 +0000 (17:15 +0000)]
Satisfy pedantic readers.

PR: docs/31260

From ChangeLog:

: 2002-01-14  Ruslan Ermilov  <ru@FreeBSD.org>
:
:         * tmac/groff_mdoc.man: Minor fixes.

22 years agoBack out part of the revision 1.2 changes -- sendto(2) can
Ruslan Ermilov [Tue, 15 Jan 2002 17:07:56 +0000 (17:07 +0000)]
Back out part of the revision 1.2 changes -- sendto(2) can
not return ENOBUFS for unreliable protocols like divert.

This should fix an issue when natd(8) keeps spamming already
full dummynet(4) queues with the same packet forever.

Spotted by: chkno@dork.com
Explained by: luigi
Reviewed by: Ari Suutari <ari.suutari@syncrontech.com>
MFC after: 2 weeks

22 years agoAdd entries for the DEC VT400 and and VT420 terminals.
Sheldon Hearn [Tue, 15 Jan 2002 16:19:37 +0000 (16:19 +0000)]
Add entries for the DEC VT400 and and VT420 terminals.

Obtained from: comp.terminals
Reported by: tedm

22 years agoFirst ping after a preload (-l) was sent undelayed.
Ruslan Ermilov [Tue, 15 Jan 2002 15:10:52 +0000 (15:10 +0000)]
First ping after a preload (-l) was sent undelayed.

PR: bin/32354
Obtained from: ping.c,v 1.61

22 years agoAdd pkg_update to the SEE_ALSO manpage of pkg_create.
Giorgos Keramidas [Tue, 15 Jan 2002 14:39:31 +0000 (14:39 +0000)]
Add pkg_update to the SEE_ALSO manpage of pkg_create.

PR: docs/33666
Submitted by: Tom Rhodes <darklogik@pittgoth.com>

22 years agoBump the limits for determining if we've held a spinlock too long as they
John Baldwin [Tue, 15 Jan 2002 14:20:33 +0000 (14:20 +0000)]
Bump the limits for determining if we've held a spinlock too long as they
seem to be too short for the 500 Mhz DS20 I'm testing on.  The rather
arbitrary numbers are rather bogus anyways.  We should probably have
variables for these limits that are calibrated in the MD startup code
somehow.

22 years ago- Catch up printtrap() to KSE by using curthread and testing it against
John Baldwin [Tue, 15 Jan 2002 14:17:07 +0000 (14:17 +0000)]
- Catch up printtrap() to KSE by using curthread and testing it against
  NULL rather than curproc.  Without this, if we trap early before
  curthread is set, we recursively panic.
- In an SMP kernel, if we trap before curthread is set, then trap is going
  to recursively panic trying to bump td->td_md.md_kernnest.  The trap is
  fatal anyways, so to make debugging easier just call printtrap() to
  dump the trap info to the console and then halt.

22 years agoDo not install man(1) setuid ``man''.
Ruslan Ermilov [Tue, 15 Jan 2002 14:11:05 +0000 (14:11 +0000)]
Do not install man(1) setuid ``man''.

The catpaging and setuidness features of man(1) combined make
it vulnerable to a number of security attacks.  Specifically,
it was possible to overwrite system catpages with arbitrarily
contents by either setting up a symlink to a directory holding
system catpages, or by writing custom -mdoc or -man groff(1)
macro packages and setting up GROFF_TMAC_PATH in environment
to point to them.  (See PR below for details).

This means man(1) can no longer create system catpages on a
regular user's behalf.  (It is still able to if the user has
write permissions to the directory holding catpages, e.g.,
user's own manpages, or if the running user is ``root''.)

To create and install catpages during ``make world'', please
set MANBUILDCAT=YES in /etc/make.conf.  To rebuild catpages
on a weekly basis, please set weekly_catman_enable="YES" in
/etc/periodic.conf.

PR: bin/32791

22 years agoAdd missed includes.
Maxim Sobolev [Tue, 15 Jan 2002 12:23:52 +0000 (12:23 +0000)]
Add missed includes.

Reviewed by: md5

22 years agoo Add prototype for printf(3).
Mike Barcroft [Tue, 15 Jan 2002 08:50:28 +0000 (08:50 +0000)]
o Add prototype for printf(3).

style(9):
o Order variables in declarations.
o Move initialization out of declaration.
o Fix over-indents in previous delta.

22 years agoMerge the changes in the English version (1.253 --> 1.254).
Jun Kuriyama [Tue, 15 Jan 2002 08:49:44 +0000 (08:49 +0000)]
Merge the changes in the English version (1.253 --> 1.254).

22 years agostyle(9)
Mike Barcroft [Tue, 15 Jan 2002 08:26:58 +0000 (08:26 +0000)]
style(9)

Submitted by: Joseph Mallett <jmallett@xmach.org>
Reviewed by: md5(1)

22 years agoWhen downgrading a filesystem from read-write to read-only, operations
Kirk McKusick [Tue, 15 Jan 2002 07:17:12 +0000 (07:17 +0000)]
When downgrading a filesystem from read-write to read-only, operations
involving file removal or file update were not always being fully
committed to disk. The result was lost files or corrupted file data.
This change ensures that the filesystem is properly synced to disk
before the filesystem is down-graded.

This delta also fixes a long standing bug in which a file open for
reading has been unlinked. When the last open reference to the file
is closed, the inode is reclaimed by the filesystem. Previously,
if the filesystem had been down-graded to read-only, the inode could
not be reclaimed, and thus was lost and had to be later recovered
by fsck.  With this change, such files are found at the time of the
down-grade.  Normally they will result in the filesystem down-grade
failing with `device busy'. If a forcible down-grade is done, then
the affected files will be revoked causing the inode to be released
and the open file descriptors to begin failing on attempts to read.

Submitted by: "Sam Leffler" <sam@errno.com>

22 years agoMake PCI_ALLOW_UNSUPPORTED_IO_RANGE an option until the ISA address
Warner Losh [Tue, 15 Jan 2002 06:46:59 +0000 (06:46 +0000)]
Make PCI_ALLOW_UNSUPPORTED_IO_RANGE an option until the ISA address
problem is fixed at the bridge level.  This is needed for some newer
laptops that have the cardbus bridge not on pci0.

22 years agoSync to GENERIC: add UFS_DIRHASH
Warner Losh [Tue, 15 Jan 2002 06:43:59 +0000 (06:43 +0000)]
Sync to GENERIC: add UFS_DIRHASH

Submitted by: Anders Andersson <anders@hack.org>

22 years agoCompilers are very picky about matching opeining and closing brackets.
John Hay [Tue, 15 Jan 2002 05:43:37 +0000 (05:43 +0000)]
Compilers are very picky about matching opeining and closing brackets.

22 years agoMake man page and usage statement match reality submitted by Ruslan,
Doug Ambrisko [Tue, 15 Jan 2002 04:31:08 +0000 (04:31 +0000)]
Make man page and usage statement match reality submitted by Ruslan,
Clean up "n to m" type options with "n-m" and some other improvements
suggested by Ruslan.

Change -C option to report the transmit key "4" if in "Home" mode.

Submitted by: ru
Approved by: imp, ru

22 years agoReinstate linux_ifname. Although the Linuxulator doesn't use it
Marcel Moolenaar [Tue, 15 Jan 2002 03:10:33 +0000 (03:10 +0000)]
Reinstate linux_ifname. Although the Linuxulator doesn't use it
itself, it's used outside the Linuxulator. Reimplement the
function so that its behaviour matches the current renaming
scheme. It's probably better to formalize these interdependencies.

22 years agoPush down Giant in dup(2) and dup2(2), Giant is only needed when
Alfred Perlstein [Tue, 15 Jan 2002 00:58:40 +0000 (00:58 +0000)]
Push down Giant in dup(2) and dup2(2), Giant is only needed when
calling closef() in the case of dup2(2) duping over a descriptor
and when fdalloc must grow or free a filedesc.

22 years agoFix select on fifos.
Alfred Perlstein [Mon, 14 Jan 2002 22:03:48 +0000 (22:03 +0000)]
Fix select on fifos.

Backout revision 1.56 and 1.57 of fifo_vnops.c.

Introduce a new poll op "POLLINIGNEOF" that can be used to ignore
EOF on a fifo, POLLIN/POLLRDNORM is converted to POLLINIGNEOF within
the FIFO implementation to effect the correct behavior.

This should allow one to view a fifo pretty much as a data source
rather than worry about connections coming and going.

Reviewed by: bde

22 years agoFix mind-o: compare sc->sis_rev instead of 'command' when trying to
Bill Paul [Mon, 14 Jan 2002 20:52:31 +0000 (20:52 +0000)]
Fix mind-o: compare sc->sis_rev instead of 'command' when trying to
decide how to read the station address.

22 years agofix MLINKS for timeradd.3
Max Khon [Mon, 14 Jan 2002 20:36:34 +0000 (20:36 +0000)]
fix MLINKS for timeradd.3

22 years agoCommit a know fix for hpfs to use vop_defaultop plug instead of wrong
Semen Ustimenko [Mon, 14 Jan 2002 20:13:42 +0000 (20:13 +0000)]
Commit a know fix for hpfs to use vop_defaultop plug instead of wrong
hpfs_bypass() routine.

MFC after: 1 day

22 years agoRemove a bogus FILEDESC_UNLOCK.
Alfred Perlstein [Mon, 14 Jan 2002 19:45:03 +0000 (19:45 +0000)]
Remove a bogus FILEDESC_UNLOCK.

Submitted by: tanimura

22 years agoUnbreak installation for the CD-ROM and possiblhy other media types.
Brian Feldman [Mon, 14 Jan 2002 19:42:11 +0000 (19:42 +0000)]
Unbreak installation for the CD-ROM and possiblhy other media types.

Obtained from: LOMAC project
Sponsored by: DARPA, NAI Labs

22 years agoMake the i386 boot2 fully blocksize-agnostic, as has already been
Ian Dowse [Mon, 14 Jan 2002 19:39:19 +0000 (19:39 +0000)]
Make the i386 boot2 fully blocksize-agnostic, as has already been
done with boot1 on the alpha. We use 4k buffers regardless of the
actual filesystem block size.

Remove the simple malloc() implementation, as it is no longer used.

22 years agoThe vnode was not being vput()'d in the EEXIST mknod case on the nfs
Matthew Dillon [Mon, 14 Jan 2002 19:14:08 +0000 (19:14 +0000)]
The vnode was not being vput()'d in the EEXIST mknod case on the nfs
server side.  This can lead to a system deadlock.

Reviewed by:    iedowse
Tested by:      Alexey G Misurenko <mag@caravan.ru>, iedowse
Bug found with help by: Alexey G Misurenko <mag@caravan.ru>
MFC at:         earliest convenience

22 years agoMove bootloader-related changes into a new sect3-level section
Bruce A. Mah [Mon, 14 Jan 2002 19:02:19 +0000 (19:02 +0000)]
Move bootloader-related changes into a new sect3-level section
underneath the Kernel Changes.

22 years agoEverybody (for suitable values of "everybody") seems to think pam.conf should
Dag-Erling Smørgrav [Mon, 14 Jan 2002 17:15:53 +0000 (17:15 +0000)]
Everybody (for suitable values of "everybody") seems to think pam.conf should
be removed outright.

Sponsored by: DARPA, NAI Labs

22 years agoyp(4) -> yp(8).
Ruslan Ermilov [Mon, 14 Jan 2002 16:59:03 +0000 (16:59 +0000)]
yp(4) -> yp(8).

PR: docs/30797

22 years agoConvert to man4/Makefile style.
Ruslan Ermilov [Mon, 14 Jan 2002 16:53:43 +0000 (16:53 +0000)]
Convert to man4/Makefile style.

22 years agoRe-add pam.conf so it will get installed so people who don't know about
Dag-Erling Smørgrav [Mon, 14 Jan 2002 16:30:22 +0000 (16:30 +0000)]
Re-add pam.conf so it will get installed so people who don't know about
pam.d will find out about it by reading pam.conf.

Sponsored by: DARPA, NAI Labs

22 years agomdoc(7) police: tidy up the markup.
Ruslan Ermilov [Mon, 14 Jan 2002 16:07:10 +0000 (16:07 +0000)]
mdoc(7) police: tidy up the markup.

22 years agoThis commit was generated by cvs2svn to compensate for changes in r89354,
Andrey A. Chernov [Mon, 14 Jan 2002 15:58:32 +0000 (15:58 +0000)]
This commit was generated by cvs2svn to compensate for changes in r89354,
which included commits to RCS files with non-trunk default branches.

22 years agoImport official baudrate fix into the vendor branch
Andrey A. Chernov [Mon, 14 Jan 2002 15:58:32 +0000 (15:58 +0000)]
Import official baudrate fix into the vendor branch

22 years agoMerge the changes in the English version (1.252 --> 1.253).
Jun Kuriyama [Mon, 14 Jan 2002 15:54:18 +0000 (15:54 +0000)]
Merge the changes in the English version (1.252 --> 1.253).

22 years agoRelease note deleted: sysinstall newfs blocksize parameters on
Bruce A. Mah [Mon, 14 Jan 2002 15:46:58 +0000 (15:46 +0000)]
Release note deleted:  sysinstall newfs blocksize parameters on
alpha.

MFCs noted:  Bootloaders support larger filesystem block sizes.

22 years agoPoint people towards /etc/pam.d/README.
Dag-Erling Smørgrav [Mon, 14 Jan 2002 15:08:02 +0000 (15:08 +0000)]
Point people towards /etc/pam.d/README.

22 years agoFixed two bugs with the "-l preload" option:
Ruslan Ermilov [Mon, 14 Jan 2002 14:33:35 +0000 (14:33 +0000)]
Fixed two bugs with the "-l preload" option:

- first ping after a preload was sent undelayed
- we could send more than -c packets in preload

PR: bin/32354

22 years agoThe datasheet for the 6710/6722 says that you must use one method to
Warner Losh [Mon, 14 Jan 2002 14:10:13 +0000 (14:10 +0000)]
The datasheet for the 6710/6722 says that you must use one method to
detect 3.3V cards for the 6710 and another method for the 6722.  This
latter method is also how the 6729/6730 is supposed to detect 3.3V
cards.  This method works great on my Fujitsu Stylistic 500.  Sadly,
it appears that not all laptop makers are as detail oriented as the
folks that made the Stylistic.  IBM Thinkpad 701C and AST Asentia 810N
both hang hard when the 6729 method is used, but at least the thinkpad
works when the 6710 method is used.  The failure mode appears to be any
access to the memory that we've mapped the CIS in causes the machine to
hang until you eject the card.  The Thinkpad, at least, works with this
change, and it doesn't break my Stylistic.

MFC after: 4 days

22 years agoAdd recognition of the Omega 83C094 PCI <-> PCMCIA bridge chip set.
Warner Losh [Mon, 14 Jan 2002 13:50:45 +0000 (13:50 +0000)]
Add recognition of the Omega 83C094 PCI <-> PCMCIA bridge chip set.

MFC after: 5 days

22 years ago'uname -p' prints the value of hw.machine_arch instead of hw.machine.
Yoshihiro Takahashi [Mon, 14 Jan 2002 12:49:46 +0000 (12:49 +0000)]
'uname -p' prints the value of hw.machine_arch instead of hw.machine.

Reviewed by: imp
No response from: -arch
MFC after: 3 weeks

22 years agoAdded 'MACHINE=i386' to MKMODULESENV variable.
Yoshihiro Takahashi [Mon, 14 Jan 2002 12:48:18 +0000 (12:48 +0000)]
Added 'MACHINE=i386' to MKMODULESENV variable.
This enables to make modules for PC/AT on PC-98 boxes.

MFC after: 3 weeks

22 years agoBackout inclusion of sys/queue.h
Alfred Perlstein [Mon, 14 Jan 2002 09:19:18 +0000 (09:19 +0000)]
Backout inclusion of sys/queue.h

22 years agoRevert rev 1.316 now that the bootstrap issues with filesystems using
Sheldon Hearn [Mon, 14 Jan 2002 09:16:56 +0000 (09:16 +0000)]
Revert rev 1.316 now that the bootstrap issues with filesystems using
block sizees larger than 8192 bytes have been resolved, as per the
following deltas:

rev 1.34 src/sys/boot/i386/boot2/boot2.c
rev 1.5 src/sys/boot/alpha/boot1/sys.c

22 years agoBackout inclusion of queue.h since rev 1.38 sys/file.h now has it
Alfred Perlstein [Mon, 14 Jan 2002 09:07:15 +0000 (09:07 +0000)]
Backout inclusion of queue.h since rev 1.38 sys/file.h now has it
included in the right order.

22 years agoRemove requirement for queue.h by consumers by moving its inclusion
Alfred Perlstein [Mon, 14 Jan 2002 09:02:33 +0000 (09:02 +0000)]
Remove requirement for queue.h by consumers by moving its inclusion
before other headers that require it.

Pointed out by: ru, bde

22 years agoMerge the changes in the English version (1.251 --> 1.252).
Jun Kuriyama [Mon, 14 Jan 2002 08:27:35 +0000 (08:27 +0000)]
Merge the changes in the English version (1.251 --> 1.252).

22 years ago o Correct the initialization of aiolio_zone: Each entry was 16 times larger
Alan Cox [Mon, 14 Jan 2002 07:26:33 +0000 (07:26 +0000)]
 o Correct the initialization of aiolio_zone: Each entry was 16 times larger
   than necessary.
 o Move a rarely-used goto label inside a critical section so that we don't
   perform an splnet() for which there is no corresponding splx().
 o Remove unnecessary splnet()/splx() around accesses to kaioinfo::kaio_jobdone
   in aio_return().
 o Use TAILQ_FOREACH for simple cases of iteration over kaioinfo::kaio_jobdone.

22 years agoadd queue.h, now needed for file.h
Alfred Perlstein [Mon, 14 Jan 2002 04:12:54 +0000 (04:12 +0000)]
add queue.h, now needed for file.h

22 years agoNew release notes: boot loaders support larger filesystem block
Bruce A. Mah [Mon, 14 Jan 2002 03:59:36 +0000 (03:59 +0000)]
New release notes:  boot loaders support larger filesystem block
sizes, new sis(4) support, tcpd(8) now built and installed.

22 years agoMention sis(4) support for SiS 635 and 735 motherboard chipsets.
Bruce A. Mah [Mon, 14 Jan 2002 03:54:48 +0000 (03:54 +0000)]
Mention sis(4) support for SiS 635 and 735 motherboard chipsets.

22 years agouse bus_space(9) rather than the memory mapped register set. This is
Warner Losh [Mon, 14 Jan 2002 02:59:06 +0000 (02:59 +0000)]
use bus_space(9) rather than the memory mapped register set.  This is
the first step towrads merging pcic and pccbb for 16-bit cards.

22 years agodon't initialize the mutex in the temporary struct file, the soo_*
Alfred Perlstein [Mon, 14 Jan 2002 02:18:59 +0000 (02:18 +0000)]
don't initialize the mutex in the temporary struct file, the soo_*
functions just grab f_data and don't muck with anything else so this
should be ok.

this fixes a panic with invariants where it thinks we've doubly initialized
the filetmp mutex even though all we've done is neglect to bzero it.

22 years agoImplement vfs.nfs.iodmin (minimum number of nfsiod's) and
Peter Wemm [Mon, 14 Jan 2002 02:13:46 +0000 (02:13 +0000)]
Implement vfs.nfs.iodmin (minimum number of nfsiod's) and
vfs.nfs.iodmaxidle (idle time before nfsiod's exit).  Make it adaptive
so that we create nfsiod's on demand and they go away after not being
used for a while.  The upper limit is NFS_MAXASYNCDAEMON (currently 20).
More will be done here, but this is a useful checkpoint.

Submitted by: Maxime Henrion <mux@qualys.com>

22 years agoMerge some updates and markup fixes from OpenBSD. This is mainly
Crist J. Clark [Mon, 14 Jan 2002 02:08:02 +0000 (02:08 +0000)]
Merge some updates and markup fixes from OpenBSD. This is mainly
motivated by the new "CAVEATS" section.

Inspired by: alfred noting NetBSD's merging OpenBSD's changes
Obtained from: OpenBSD

22 years agoAdd xref for timeradd(3).
Doug White [Mon, 14 Jan 2002 00:38:41 +0000 (00:38 +0000)]
Add xref for timeradd(3).

PR: 13079

22 years agoAdd manpage for timeradd macros and friends. A companion page for
Doug White [Mon, 14 Jan 2002 00:38:20 +0000 (00:38 +0000)]
Add manpage for timeradd macros and friends. A companion page for
timespecadd should be added at a future point...

PR: 13079

22 years agoReplace ffind_* with fget calls.
Alfred Perlstein [Mon, 14 Jan 2002 00:13:45 +0000 (00:13 +0000)]
Replace ffind_* with fget calls.

Make fget MPsafe.

Make fgetvp and fgetsock use the fget subsystem to reduce code bloat.

Push giant down in fpathconf().

22 years agoAdd parens around macro args.
Alfred Perlstein [Sun, 13 Jan 2002 23:51:47 +0000 (23:51 +0000)]
Add parens around macro args.

Forgotten by: dillon

22 years agoremove unused socket pointer
Alfred Perlstein [Sun, 13 Jan 2002 22:15:18 +0000 (22:15 +0000)]
remove unused socket pointer

22 years agoInclude sys/_lock.h and sys/_mutex.h to reduce namespace pollution.
Alfred Perlstein [Sun, 13 Jan 2002 21:37:49 +0000 (21:37 +0000)]
Include sys/_lock.h and sys/_mutex.h to reduce namespace pollution.

Requested by: jhb

22 years agoImplement the POSIX 1003.1-2001 -r and -t options in at(1). Fix some
Mike Barcroft [Sun, 13 Jan 2002 20:21:08 +0000 (20:21 +0000)]
Implement the POSIX 1003.1-2001 -r and -t options in at(1).  Fix some
minor issues in the rest of the source and manual.

Submitted by: Joe Halpin <joe.halpin@attbi.com>
Obtained from: touch(1) (partially)
MFC after: 1 month

22 years ago o Call the functions registered with at_exec() from exec_new_vmspace()
Alan Cox [Sun, 13 Jan 2002 19:36:35 +0000 (19:36 +0000)]
 o Call the functions registered with at_exec() from exec_new_vmspace()
   instead of execve().  Otherwise, the possibility still exists
   for a pending AIO to modify the new address space.

Reviewed by: alfred

22 years agoMake call_desc[] extern since it's also defined in layer4/i4b_l4mgmt.c
Gary Jennejohn [Sun, 13 Jan 2002 19:16:52 +0000 (19:16 +0000)]
Make call_desc[] extern since it's also defined in layer4/i4b_l4mgmt.c
which was causing the link of the kernel to fail. Since layer4/i4b_l4mgmt.c
is always required by i4b and layer3/i4b_q931.c is only needed when a
passive card is present it makes sense to have the declaration in the
former file only.

No MFC required since the problem only became apparent after a change to
the compile options in -current which AFAIK wasn't made in -stable.

22 years agoThe fix for >/dev/stdout, including Tor Egge's fix for the bug in the
Sergey Babkin [Sun, 13 Jan 2002 14:37:40 +0000 (14:37 +0000)]
The fix for >/dev/stdout, including Tor Egge's fix for the bug in the
original attempt of the fix. And yes, this time I've tried to build
world with it and it succeeded.

Submitted by: Tor Egge
MFC after: 1 week