]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoUse pci_get_devid().
Hidetoshi Shimokawa [Tue, 18 Feb 2003 10:01:44 +0000 (10:01 +0000)]
Use pci_get_devid().

21 years agoAdd 'uPD72871/2 IEEE1394 1-Chip OHCI Host Controller'.
Hidetoshi Shimokawa [Tue, 18 Feb 2003 09:46:52 +0000 (09:46 +0000)]
Add 'uPD72871/2 IEEE1394 1-Chip OHCI Host Controller'.

21 years ago* firewire.c
Hidetoshi Shimokawa [Tue, 18 Feb 2003 09:39:48 +0000 (09:39 +0000)]
* firewire.c
- Fix memory leak in detaching.
- Initialize fc->status to other than FWBUSREST.

* fwohci.c
- Ignore BUS reset events while BUS reset phase. We can't clear that flag
during bus reset phase.

21 years agoAdd central and fhc.
Jake Burkholder [Tue, 18 Feb 2003 09:38:51 +0000 (09:38 +0000)]
Add central and fhc.

21 years agoAdd drivers for the central and fhc busses found in enterprise class
Jake Burkholder [Tue, 18 Feb 2003 09:01:01 +0000 (09:01 +0000)]
Add drivers for the central and fhc busses found in enterprise class
UltraSPARCs, and an eeprom attachment for fhc, which allows the date
to be set properly on these machines.  Central is a wierd bus which
seems to only ever have 1 fhc attached to it.   FHC (FireHose Controller)
is another wierd bus with various things on it depending where its attached.
The fhc attached to central has eeprom and zs, and the fhcs which attach
directly to nexus have simm-status, environment and other nodes, none of
which I'll probably ever have documentation for.

Thanks to Ade Lovett for providing access to an 8 cpu e4500.

21 years agoremove stray debugging printf
Sam Leffler [Tue, 18 Feb 2003 06:25:46 +0000 (06:25 +0000)]
remove stray debugging printf

Noted by: Kasper Steensgaard <steensgaard@person.dk>

21 years agoM_MOVE_PKTHDR must happen before any cluster is attached
Sam Leffler [Tue, 18 Feb 2003 06:20:16 +0000 (06:20 +0000)]
M_MOVE_PKTHDR must happen before any cluster is attached

Submitted by: Harti Brandt <brandt@fokus.fraunhofer.de>
MFC after: 1 day

21 years agoFix stupid whitespace typo.
Bruce A. Mah [Tue, 18 Feb 2003 05:16:54 +0000 (05:16 +0000)]
Fix stupid whitespace typo.

21 years agoInitiate de-orbit burn for USE_PCI_BIOS_FOR_READ_WRITE. This has been
Peter Wemm [Tue, 18 Feb 2003 03:36:49 +0000 (03:36 +0000)]
Initiate de-orbit burn for USE_PCI_BIOS_FOR_READ_WRITE.  This has been
#if'ed out for a while.  Complete the deed and tidy up some other bits.

We need to be able to call this stuff from outer edges of interrupt
handlers for devices that have the ISR bits in pci config space.  Making
the bios code mpsafe was just too hairy.  We had also stubbed it out some
time ago due to there simply being too much brokenness in too many systems.
This adds a leaf lock so that it is safe to use pci_read_config() and
pci_write_config() from interrupt handlers.  We still will use pcibios
to do interrupt routing if there is no acpi.. [yes, I tested this]

Briefly glanced at by:  imp

21 years agoRemove contractions.
Tom Rhodes [Tue, 18 Feb 2003 03:34:32 +0000 (03:34 +0000)]
Remove contractions.
Use `Force' in place of `Cause' which sounds better.

21 years agoInclude class designation in pnpinfo for generic driver loading
Warner Losh [Tue, 18 Feb 2003 03:25:57 +0000 (03:25 +0000)]
Include class designation in pnpinfo for generic driver loading

21 years agoNew release notes: Linux IPv6 support, cm support for IPX (+MFC),
Bruce A. Mah [Tue, 18 Feb 2003 02:51:53 +0000 (02:51 +0000)]
New release notes: Linux IPv6 support, cm support for IPX (+MFC),
FAST_IPSEC and INET6 can coexist (sort of), wfd/wst removal, compat4x
adds libcrypto/libgmp/libssl libraries, config(8) nodevice,
mount_nfs(8) -c (+MFC), ping(8) -M.

Modified release notes:  Explicitly name wlan module.

21 years agoRevert ALIGN change for the second and last time. I can't figure out
Tim J. Robbins [Tue, 18 Feb 2003 02:08:20 +0000 (02:08 +0000)]
Revert ALIGN change for the second and last time. I can't figure out
why this is breaking sparc64.

21 years agoUse arc4random() instead of random() when generating the master key.
Kris Kennaway [Tue, 18 Feb 2003 01:35:58 +0000 (01:35 +0000)]
Use arc4random() instead of random() when generating the master key.

MFC after: 1 week

21 years agoNuke xargs, I haven't had to help anyone with it, so I assume it stands on
Juli Mallett [Tue, 18 Feb 2003 00:07:06 +0000 (00:07 +0000)]
Nuke xargs, I haven't had to help anyone with it, so I assume it stands on
its own just fine.  Nuke request for heads-up on libufs, most of my work on
it these days is gradual, and mention that I'm willing to help with work in
it, so that others can feel free to make it suck less, and get feedback from
me in a purely "I understand this crud" sort of way.

21 years agoSanity check the BAR length reported by the CIS with the BAR length that
Scott Long [Mon, 17 Feb 2003 23:47:31 +0000 (23:47 +0000)]
Sanity check the BAR length reported by the CIS with the BAR length that
is encoded in the PCI BAR.  The latter is more reliable.

This allows the sio/modem function of the Xircom RealPort ethernet+modem
card to work.  Note that there still seem to be issues with sio_pci not
releasing resources on detach.

21 years agoBack out revision 1.34 until we can come up with a more accurate patch.
Chris Costello [Mon, 17 Feb 2003 22:31:27 +0000 (22:31 +0000)]
Back out revision 1.34 until we can come up with a more accurate patch.

21 years agoAdd a section on Giant lock manipulation macros.
Chris Costello [Mon, 17 Feb 2003 22:21:41 +0000 (22:21 +0000)]
Add a section on Giant lock manipulation macros.

Submitted by: Hiten M. Pandya <hiten@angelica.unixdaemons.com>
Reviewed by: bmilekic

21 years agoXref miibus(4).
Tom Rhodes [Mon, 17 Feb 2003 21:22:56 +0000 (21:22 +0000)]
Xref miibus(4).

21 years agoMove the pnp and location info into the common pci bus. Make all known
Warner Losh [Mon, 17 Feb 2003 21:20:35 +0000 (21:20 +0000)]
Move the pnp and location info into the common pci bus.  Make all known
pci busses implement this.

Also minor comment smithing in cardbus.  Fix copyright to this year
with my name on it since I've been doing a lot to this file.

Reviewed by: jhb

21 years agoDo not remove object from the lists at the unref_dag() stage.
Alexander Kabaev [Mon, 17 Feb 2003 20:58:27 +0000 (20:58 +0000)]
Do not remove object from the lists at the unref_dag() stage.
Introduce a new unlink_object() function and call it in
unload_object() instead. Removing the object in unref_dag() is
too early, rtld calls _fini() function after that and shared
objects might fail resolve their own symbols.

21 years agoo Add a note explaining the meaning of mls/equal beyond "equal to all
Chris Costello [Mon, 17 Feb 2003 20:11:09 +0000 (20:11 +0000)]
o Add a note explaining the meaning of mls/equal beyond "equal to all
  labels"
o Remove the ++ compartment range notation example as this has not yet
  been merged into CVS.
o Include a "Runtime Configuration" section listing all of the relevant
  sysctl knobs for this policy.

Sponsored by: DARPA, Network Associates Laboratories
Obtained from: TrustedBSD Project

21 years ago- Document some of the sysctl parameters. The read-only ones will go into
Chris Costello [Mon, 17 Feb 2003 20:04:06 +0000 (20:04 +0000)]
- Document some of the sysctl parameters.  The read-only ones will go into
  another subsection, but I'm not quite sure where yet.  Right now it's just
  the configurable bits.
- Move the label format off into another subsection to match other labeling
  policy man pages.
- Make the sample range label look like the form specified.

Sponsored by: DARPA, Network Associates Laboratories
Obtained from: TrustedBSD Project

21 years agoan driver now needs wlan, so put the required explicit depends
Warner Losh [Mon, 17 Feb 2003 19:57:32 +0000 (19:57 +0000)]
an driver now needs wlan, so put the required explicit depends
into the source.

Reviewed by: ambrisko

21 years ago- build powerpc world on 9ball
Dag-Erling Smørgrav [Mon, 17 Feb 2003 19:50:58 +0000 (19:50 +0000)]
- build powerpc world on 9ball
- log directly into two files: a full log and a summary
- set timezone to GMT before starting

21 years agoKill a now-bogus comment
Warner Losh [Mon, 17 Feb 2003 19:48:39 +0000 (19:48 +0000)]
Kill a now-bogus comment

21 years agoThese don't need to be semi-public after all.
Warner Losh [Mon, 17 Feb 2003 19:47:02 +0000 (19:47 +0000)]
These don't need to be semi-public after all.

21 years agoBump pnpinfo and location size to 128 bytes. Since these are computed
Warner Losh [Mon, 17 Feb 2003 19:01:33 +0000 (19:01 +0000)]
Bump pnpinfo and location size to 128 bytes.  Since these are computed
on the fly and read into userland one at a time, this costs very
little total memory.  The pnpinfo sizes of pccard is more than 64
bytes due to the length of the strings that man cards have in their
CIS.

21 years ago-v now also prints the pnpinfo and location information for the devices
Warner Losh [Mon, 17 Feb 2003 18:56:54 +0000 (18:56 +0000)]
-v now also prints the pnpinfo and location information for the devices
whose bus' provide this information.

21 years agoExpand length of pnpinfo and length fields since pccard pnpinfo can
Warner Losh [Mon, 17 Feb 2003 18:56:00 +0000 (18:56 +0000)]
Expand length of pnpinfo and length fields since pccard pnpinfo can
easily be longer than 64 characters.

21 years agoAdd a bunch of cards that we've added IDs for in the past few months:
Warner Losh [Mon, 17 Feb 2003 18:21:51 +0000 (18:21 +0000)]
Add a bunch of cards that we've added IDs for in the past few months:

+Allied Telesis WR211PCM        Prism?  PCMCIA
+ArTem OnAir    Prism?  PCMCIA
+Bay eMobility 11B      Prism-2.5?      PCMCIA
+Blue Concentric Circle WL-379F Prism-II        CF
+Linksys WCF11  Prism-3 PCMCIA
+YIS TWL-11B    Prism-II        PCMCIA

21 years agoAdd libgmp.so.3 from 4.7-RELEASE CDs.
David E. O'Brien [Mon, 17 Feb 2003 17:26:03 +0000 (17:26 +0000)]
Add libgmp.so.3 from 4.7-RELEASE CDs.

Requested by: des

21 years agoFurther fix PS_NEEDSIGCHK
David Xu [Mon, 17 Feb 2003 14:54:57 +0000 (14:54 +0000)]
Further fix PS_NEEDSIGCHK

21 years agoMove code for detecting PS_NEEDSIGCHK into thread_schedule_upcall,
David Xu [Mon, 17 Feb 2003 14:41:22 +0000 (14:41 +0000)]
Move code for detecting PS_NEEDSIGCHK into thread_schedule_upcall,
I think it is a better place to handle it.

21 years ago* firewire.c
Hidetoshi Shimokawa [Mon, 17 Feb 2003 14:24:06 +0000 (14:24 +0000)]
* firewire.c
- Don't initiate bus reset even if probe failed for some nodes to prevent
infinite bus reset loop.

Problem Reported by: Pierre Beyssac <pb@fasterix.frmug.org>

- Protect timeout routine with splfw() for 4-stable.

* sbp.c

- Make sure to release devq when start request.

21 years agoAdd libgmp.so.3 from 4.7-RELEASE CDs.
Jacques Vidrine [Mon, 17 Feb 2003 13:45:13 +0000 (13:45 +0000)]
Add libgmp.so.3 from 4.7-RELEASE CDs.

Requested by: des

21 years agoo Restore an interrupt priority level before return.
Maxim Konovalov [Mon, 17 Feb 2003 13:41:55 +0000 (13:41 +0000)]
o Restore an interrupt priority level before return.

Submitted by: Roman Kurakin <rik@cronyx.ru>
Reviewed by: joerg
MFC after: 5 days

21 years agoo Fix ipfw uid rules: socheckuid() returns 0 when uid matches a socket
Maxim Konovalov [Mon, 17 Feb 2003 13:39:57 +0000 (13:39 +0000)]
o Fix ipfw uid rules: socheckuid() returns 0 when uid matches a socket
cr_uid.

Note: we do not have socheckuid() in RELENG_4, ip_fw2.c uses its
own macro for a similar purpose that is why ipfw2 in RELENG_4 processes
uid rules correctly. I will MFC the diff for code consistency.

Reported by: Oleg Baranov <ol@csa.ru>
Reviewed by: luigi
MFC after: 1 month

21 years agoFix missed patch in last commit
Julian Elischer [Mon, 17 Feb 2003 10:21:32 +0000 (10:21 +0000)]
Fix missed patch in last commit

21 years agoDeliver signals posted via an upcall to the appropriate thread.
Jonathan Mini [Mon, 17 Feb 2003 10:05:18 +0000 (10:05 +0000)]
Deliver signals posted via an upcall to the appropriate thread.

21 years agoUse the proc lock to protect p_realtimer instead of Giant, and obtain
Tim J. Robbins [Mon, 17 Feb 2003 10:03:02 +0000 (10:03 +0000)]
Use the proc lock to protect p_realtimer instead of Giant, and obtain
sched_lock around accesses to p_stats->p_timer[] to avoid a potential
race with hardclock. getitimer(), setitimer() and the realitexpire()
callout are now Giant-free.

21 years ago - Add a new function, thread_signal_add(), that is called from postsig to
Jeff Roberson [Mon, 17 Feb 2003 09:58:11 +0000 (09:58 +0000)]
 - Add a new function, thread_signal_add(), that is called from postsig to
   add a signal to a mailbox's pending set.
 - Add a new function, thread_signal_upcall(), this causes the current thread
   to upcall so that we can deliver pending signals.

Reviewed by: mini

21 years agoMove a bunch of flags from the KSE to the thread.
Julian Elischer [Mon, 17 Feb 2003 09:55:10 +0000 (09:55 +0000)]
Move a bunch of flags from the KSE to the thread.
I was in two minds as to where to put them in the first case..
I should have listenned to the other mind.

Submitted by:  parts by davidxu@
Reviewed by: jeff@ mini@

21 years agoDefine _ALIGNBYTES to be 15. This should have been done right away.
Marcel Moolenaar [Mon, 17 Feb 2003 09:53:29 +0000 (09:53 +0000)]
Define _ALIGNBYTES to be 15. This should have been done right away.

21 years agoPrint two new processor features:
Marcel Moolenaar [Mon, 17 Feb 2003 08:17:26 +0000 (08:17 +0000)]
Print two new processor features:
o  Spontaneous deferral (A feature required by dutch railways :-)
o  16-byte atomic operations (ld, st, cmpxchg)

21 years agoCall the weak symbol for sigprocmask, so that it can be overridden.
Jonathan Mini [Mon, 17 Feb 2003 07:47:13 +0000 (07:47 +0000)]
Call the weak symbol for sigprocmask, so that it can be overridden.

21 years ago - Split the struct kse into struct upcall and struct kse. struct kse will
Jeff Roberson [Mon, 17 Feb 2003 05:14:26 +0000 (05:14 +0000)]
 - Split the struct kse into struct upcall and struct kse.  struct kse will
   soon be visible only to schedulers.  This greatly simplifies much the
   KSE code.

Submitted by: davidxu

21 years agoMove call to pci_print_verbose until after all the variables that it
Warner Losh [Mon, 17 Feb 2003 04:13:44 +0000 (04:13 +0000)]
Move call to pci_print_verbose until after all the variables that it
depends on.

Pointy hat to: imp (anybody know if these things are accepted at Eco-Cycle?)

21 years agoBack out "drop first N values" method of removing monotonically increased
Andrey A. Chernov [Mon, 17 Feb 2003 03:52:35 +0000 (03:52 +0000)]
Back out "drop first N values" method of removing monotonically increased
seed->first value correlation. It breaks rand_r()... Other possible methods
like shuffling inside aray will breaks rand_r() too, because it assumes
only one word state, i.e. nothing extra can be added after seed assignment
in srand().

BTW, for old formulae seed->first value correlation is not so monotonically
increased as with other Linear Congruential Generators of this type only
becase arithmetic overflow happens. But overflow affects distribution
and lower bits very badly, as many articles says, such type of overflow
not improves PRNG.

So, monotonically increased seed->first value correlation problem remains...

21 years agoSecond attempt at removing machdep.h and using ALIGN from <sys/param.h>.
Tim J. Robbins [Mon, 17 Feb 2003 03:51:44 +0000 (03:51 +0000)]
Second attempt at removing machdep.h and using ALIGN from <sys/param.h>.
The problem with the previous attempt, as noticed by Marcel, was that
stacknxt was being aligned to a pointer boundary instead of an
ALIGNBYTES + 1 boundary, which broke sparc64.

21 years ago - Move ke_sticks, ke_iticks, ke_uticks, ke_uu, ke_su, and ke_iu back into
Jeff Roberson [Mon, 17 Feb 2003 02:19:58 +0000 (02:19 +0000)]
 - Move ke_sticks, ke_iticks, ke_uticks, ke_uu, ke_su, and ke_iu back into
   the proc.  These counters are only examined through calcru.

Submitted by: davidxu
Tested on: x86, alpha, UP/SMP

21 years agoCorrect the BSD/OS macro.
Tom Rhodes [Sun, 16 Feb 2003 22:30:00 +0000 (22:30 +0000)]
Correct the BSD/OS macro.

21 years agoNew release notes: FILE 3.40, IPFilter 3.4.31.
Bruce A. Mah [Sun, 16 Feb 2003 21:30:30 +0000 (21:30 +0000)]
New release notes:  FILE 3.40, IPFilter 3.4.31.

MFC noted:  OpenSSL 0.9.7.

21 years agoWe stopped reloading rc.conf for each script a long time ago. Things
Mike Makonnen [Sun, 16 Feb 2003 20:46:08 +0000 (20:46 +0000)]
We stopped reloading rc.conf for each script a long time ago. Things
shouldn't depend on that any more.

PR: conf/48332
Approved by: markm (mentor)(implicit)

21 years agozero the end of the memory cluster we're disposing of. Otherwise teh
Andrew Gallatin [Sun, 16 Feb 2003 19:25:04 +0000 (19:25 +0000)]
zero the end of the memory cluster we're disposing of.  Otherwise teh
vm page startup code finds a 20GB cluster on this wacky alphaserver I'm
working on..

MFC After: 7 days

21 years agoChange "dev_t gdbdev" to "void *gdb_arg", some possible paths for GDB
Poul-Henning Kamp [Sun, 16 Feb 2003 19:22:21 +0000 (19:22 +0000)]
Change "dev_t gdbdev" to "void *gdb_arg", some possible paths for GDB
will not have a dev_t.

21 years agoCorrect typos.
Jens Schweikhardt [Sun, 16 Feb 2003 18:38:12 +0000 (18:38 +0000)]
Correct typos.

Submitted by: Seth Hieronymus <sethh@principia.edu>
X-MFC after: 4.8 is out the door

21 years agoUpdate copyright.
Mike Makonnen [Sun, 16 Feb 2003 18:06:04 +0000 (18:06 +0000)]
Update copyright.

21 years agoWhatever dumbass wrote sio_pci.c didn't include a detach method.
Warner Losh [Sun, 16 Feb 2003 18:05:54 +0000 (18:05 +0000)]
Whatever dumbass wrote sio_pci.c didn't include a detach method.
Wearing said pointy hat, correct the oversight and hope nobody
notices.

# this should make xircom modems happier to detach once other bugs with
# the cardbus layer are fixed.

Noticed by: scottl
Conical Hat to: imp

21 years agoCondense the output to one line per removed user. For user's who
Mike Makonnen [Sun, 16 Feb 2003 18:04:06 +0000 (18:04 +0000)]
Condense the output to one line per removed user. For user's who
want the more verbose version, there's a -v option.

Approved by: markm (mentor)

21 years agoNo need to wrap siodetach() with sio_pccard_detach.
Warner Losh [Sun, 16 Feb 2003 18:03:15 +0000 (18:03 +0000)]
No need to wrap siodetach() with sio_pccard_detach.

21 years agoWhen trimming memory to fit Maxmem, and we discard an entire cluster,
Andrew Gallatin [Sun, 16 Feb 2003 17:30:49 +0000 (17:30 +0000)]
When trimming memory to fit Maxmem, and we discard an entire cluster,
make certain to subtract its pages from physmem.  Otherwise we end up
running phys_avail_cnt negative, and bad stuff happens.

MFC After: 1 week

21 years agoEliminate 61 warnings emitted at WARNS=2 (leaving 53 to go).
Jacques Vidrine [Sun, 16 Feb 2003 17:29:11 +0000 (17:29 +0000)]
Eliminate 61 warnings emitted at WARNS=2 (leaving 53 to go).
Only warnings that could be fixed without changing the generated object
code and without restructuring the source code have been handled.

Reviewed by: /sbin/md5

21 years agoFix logic in loop so it actually executes.
Alfred Perlstein [Sun, 16 Feb 2003 16:12:10 +0000 (16:12 +0000)]
Fix logic in loop so it actually executes.

Pointed out by: fjoe

21 years agoAdd #include <sys/resource.h>
Poul-Henning Kamp [Sun, 16 Feb 2003 16:01:52 +0000 (16:01 +0000)]
Add #include <sys/resource.h>

My apologies for missing these #includes, I must have confused the
dependencies with a wrong timestamp or something.

21 years agoAdd #include <sys/resource.h>
Poul-Henning Kamp [Sun, 16 Feb 2003 15:46:25 +0000 (15:46 +0000)]
Add #include <sys/resource.h>

21 years agoAdd #include <sys/resource.h>
Poul-Henning Kamp [Sun, 16 Feb 2003 15:31:26 +0000 (15:31 +0000)]
Add #include <sys/resource.h>

21 years agoAdd #include <sys/resource.h>
Poul-Henning Kamp [Sun, 16 Feb 2003 15:21:26 +0000 (15:21 +0000)]
Add #include <sys/resource.h>

21 years agoPointy hat- bad commit.
Matt Jacob [Sun, 16 Feb 2003 14:26:23 +0000 (14:26 +0000)]
Pointy hat- bad commit.

21 years agoThis file is now empty and should no longer be used anywhere.
Poul-Henning Kamp [Sun, 16 Feb 2003 14:15:03 +0000 (14:15 +0000)]
This file is now empty and should no longer be used anywhere.

Put a #warning here to alert people to the fact.

21 years agoRemove #include <sys/dkstat.h>
Poul-Henning Kamp [Sun, 16 Feb 2003 14:13:23 +0000 (14:13 +0000)]
Remove #include <sys/dkstat.h>

21 years agoRemove <sys/dkstat.h> #include
Poul-Henning Kamp [Sun, 16 Feb 2003 14:09:16 +0000 (14:09 +0000)]
Remove <sys/dkstat.h> #include

21 years agoRemove include of <sys/dkstat.h>
Poul-Henning Kamp [Sun, 16 Feb 2003 14:08:52 +0000 (14:08 +0000)]
Remove include of <sys/dkstat.h>

21 years agoMove cp_time[CPUSTATES] and related #defines to <sys/resource.h>
Poul-Henning Kamp [Sun, 16 Feb 2003 13:30:29 +0000 (13:30 +0000)]
Move cp_time[CPUSTATES] and related #defines to <sys/resource.h>

21 years agoPremptively include <sys/resource.h> to get the cp_time[CPUSTATES].
Poul-Henning Kamp [Sun, 16 Feb 2003 13:29:28 +0000 (13:29 +0000)]
Premptively include <sys/resource.h> to get the cp_time[CPUSTATES].

21 years agoDeclare tty related counters in a tty related .h file.
Poul-Henning Kamp [Sun, 16 Feb 2003 13:23:29 +0000 (13:23 +0000)]
Declare tty related counters in a tty related .h file.

21 years agoMove the tty related statistics counters to live with the tty code.
Poul-Henning Kamp [Sun, 16 Feb 2003 13:22:15 +0000 (13:22 +0000)]
Move the tty related statistics counters to live with the tty code.

21 years agoAlphabetize.
Dag-Erling Smørgrav [Sun, 16 Feb 2003 13:07:20 +0000 (13:07 +0000)]
Alphabetize.

21 years agoBack out previous commit now that pam_opieaccess(8) has an allow_local
Dag-Erling Smørgrav [Sun, 16 Feb 2003 13:04:27 +0000 (13:04 +0000)]
Back out previous commit now that pam_opieaccess(8) has an allow_local
option.

21 years agoAdd the allow_local option to all pam_opieaccess entries.
Dag-Erling Smørgrav [Sun, 16 Feb 2003 13:02:39 +0000 (13:02 +0000)]
Add the allow_local option to all pam_opieaccess entries.

21 years agoAdd the want_agent option to the commented-out "session" pam_ssh entry.
Dag-Erling Smørgrav [Sun, 16 Feb 2003 13:02:03 +0000 (13:02 +0000)]
Add the want_agent option to the commented-out "session" pam_ssh entry.

21 years agoAdd an "allow_local" option which forces historical behaviour.
Dag-Erling Smørgrav [Sun, 16 Feb 2003 13:01:03 +0000 (13:01 +0000)]
Add an "allow_local" option which forces historical behaviour.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r110989,
Dag-Erling Smørgrav [Sun, 16 Feb 2003 11:28:55 +0000 (11:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r110989,
which included commits to RCS files with non-trunk default branches.

21 years agoMFP4: guard against null pointer dereference which sshd can cause (#25271)
Dag-Erling Smørgrav [Sun, 16 Feb 2003 11:28:55 +0000 (11:28 +0000)]
MFP4: guard against null pointer dereference which sshd can cause (#25271)

21 years agoParanoia: instead of a NULL conversation function, use one that always
Dag-Erling Smørgrav [Sun, 16 Feb 2003 11:03:55 +0000 (11:03 +0000)]
Paranoia: instead of a NULL conversation function, use one that always
returns PAM_CONV_ERR; moreover, make sure we always have the right
conversation function installed before calling PAM service functions.
Also unwrap some not-so-long lines.

MFC after: 3 days

21 years ago - Introduce a new function bremfreel() that does a bremfree with the buf
Jeff Roberson [Sun, 16 Feb 2003 10:43:06 +0000 (10:43 +0000)]
 - Introduce a new function bremfreel() that does a bremfree with the buf
   queue lock already held.
 - In getblk() and flushbufqueues() use bremfreel() while we still have the
   buf queue lock held to keep the lists consistent.
 - Add LK_NOWAIT to two cases where we're essentially asserting that the bufs
   are not locked while acquiring the locks.  This will make sure that we get
   the appropriate panic() and not another one for sleeping with a lock held.

21 years ago - Add a WITNESS_SLEEP() for the appropriate cases in lockmgr().
Jeff Roberson [Sun, 16 Feb 2003 10:39:49 +0000 (10:39 +0000)]
 - Add a WITNESS_SLEEP() for the appropriate cases in lockmgr().

21 years agoPass FTP_PASSIVE_MODE, FTP_PROXY, and HTTP_PROXY variables to chroot sandbox.
Makoto Matsushita [Sun, 16 Feb 2003 08:34:11 +0000 (08:34 +0000)]
Pass FTP_PASSIVE_MODE, FTP_PROXY, and HTTP_PROXY variables to chroot sandbox.
These variables will be used by pkg_add, which installs mkisofs package
while MAKE_ISOS=YES.  See src/release/${arch}/mkisoimages.sh if you doubt.

No objection heard from: current@FreeBSD.org
Suggested by: des, nectar (about *_PROXY variables)

21 years agoRemove GIANT_REQUIRED from vm_pageq_remove().
Alan Cox [Sun, 16 Feb 2003 06:36:48 +0000 (06:36 +0000)]
Remove GIANT_REQUIRED from vm_pageq_remove().

21 years agoprevent overflow in shminfo.shmmax
Alfred Perlstein [Sun, 16 Feb 2003 06:08:55 +0000 (06:08 +0000)]
prevent overflow in shminfo.shmmax

21 years agoTemporarily back out machdep.h/ALIGN changes. It seems that on sparc64,
Tim J. Robbins [Sun, 16 Feb 2003 03:28:11 +0000 (03:28 +0000)]
Temporarily back out machdep.h/ALIGN changes. It seems that on sparc64,
using the alignment from sys/param.h (16) instead of the alignment
from machdep.h (8) tickled a nasty bug in the memory allocator that I
haven't been able to track down yet.

21 years agoAdd MPSAFE comment to linux_sigpending().
Tim J. Robbins [Sun, 16 Feb 2003 02:33:12 +0000 (02:33 +0000)]
Add MPSAFE comment to linux_sigpending().

21 years agoRegen from syscalls.master 1.52.
Tim J. Robbins [Sun, 16 Feb 2003 02:32:17 +0000 (02:32 +0000)]
Regen from syscalls.master 1.52.

21 years agoMark linux_sigpending() as MPSAFE.
Tim J. Robbins [Sun, 16 Feb 2003 02:31:05 +0000 (02:31 +0000)]
Mark linux_sigpending() as MPSAFE.

21 years agoRegen from syscalls.master 1.49.
Tim J. Robbins [Sun, 16 Feb 2003 02:28:35 +0000 (02:28 +0000)]
Regen from syscalls.master 1.49.

21 years agoRemove extraneous FILEDESC_LOCK around atomic read.
Jeffrey Hsu [Sun, 16 Feb 2003 02:15:15 +0000 (02:15 +0000)]
Remove extraneous FILEDESC_LOCK around atomic read.

21 years agoCheckpoint a work in progress:
Warner Losh [Sun, 16 Feb 2003 02:06:50 +0000 (02:06 +0000)]
Checkpoint a work in progress:
o Use the common pci_* routines in preference to the copied and hacked
  routines from an ancient pci.c.

This saves 509 lines in cardbus.c.  More savings to follow when I
convert the resource code over.  In the past when I've done this the
resource code conversion breaks cardbus in subtle ways so I'm doing a
1/2 way checkpoint this time.  cardbus still works for me the same as
it did before.

It also looks like cardbus devices now show up as pci bus devices to
pciconf -l, but maybe that was happening before.

Inspired by a patch from Justin Gibbs many moons ago.  When he
finishes his kobj multiple inheritance work, we can transition the
finished version of this work to that fairly easily.

21 years agoUse rman_get_device rather than rle->resl->r_dev.
Warner Losh [Sun, 16 Feb 2003 02:02:44 +0000 (02:02 +0000)]
Use rman_get_device rather than rle->resl->r_dev.
make pci_hdrtypedata and pci_read_extcap accessible (but maybe in the end
   we'll make them private again).

21 years agoMake sure we propagate rxid && iid in isp_target_put_atio
Matt Jacob [Sun, 16 Feb 2003 01:37:51 +0000 (01:37 +0000)]
Make sure we propagate rxid && iid in isp_target_put_atio

21 years agoPick up some compilation warning fixes from NetBSD.
Matt Jacob [Sun, 16 Feb 2003 01:32:52 +0000 (01:32 +0000)]
Pick up some compilation warning fixes from NetBSD.

If we don't have ISP_FW_CRASH_DUMP defined, we have to do
a isp_reinit in the core code- not the platform code- so
fix the ISP_CONN_FATAL case.