]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoThe new order of things is that dwlpxN is now called pcibN- so hack around
mjacob [Thu, 10 May 2001 07:08:03 +0000 (07:08 +0000)]
The new order of things is that dwlpxN is now called pcibN- so hack around
*that* whilst we ponder the best way to decide how to register dwlpx interrupts
with TLSB.

23 years agoFix the panics for real this time. When something can't be allocated,
imp [Thu, 10 May 2001 06:55:39 +0000 (06:55 +0000)]
Fix the panics for real this time.  When something can't be allocated,
we need to delete the info from the list as well as zero out the res
pointer we saved in the code.

Also made a few style(9) changes while I was at it.  Don't use if
(ptr) or if (!ptr), but compare against NULL.  Compare against NULL
rather than 0.  Don't have useless blocks.

There are likely other problems as well, but at least the wi based
wireless card with memory listed in its cis doesn't panic the system
when the card is inserted.

23 years agoadd alpha_pci_route_interrupt method
mjacob [Thu, 10 May 2001 06:52:56 +0000 (06:52 +0000)]
add alpha_pci_route_interrupt method

23 years agoQuick hack to reintroduce the notion that there might be alpha platforms
mjacob [Thu, 10 May 2001 05:23:58 +0000 (05:23 +0000)]
Quick hack to reintroduce the notion that there might be alpha platforms
without an i8254 timer/counter. This really needs to be cleaned up.

23 years agoFICL 2.05 has a flawed definition of BASE. Fix it.
dcs [Thu, 10 May 2001 05:13:09 +0000 (05:13 +0000)]
FICL 2.05 has a flawed definition of BASE. Fix it.

Spotted by: Peter Jeremy <peter.jeremy@alcatel.com.au>

23 years agoThis file is not needed, use the default one
luigi [Thu, 10 May 2001 05:03:43 +0000 (05:03 +0000)]
This file is not needed, use the default one

23 years agoInclude sys/lock.h for witness_list_locks() and axe unneeded sys/mutex.h
jhb [Wed, 9 May 2001 22:08:28 +0000 (22:08 +0000)]
Include sys/lock.h for witness_list_locks() and axe unneeded sys/mutex.h
include.

23 years agoAvoid flushing IPv6 routes. `route flush' removes necessary
ume [Wed, 9 May 2001 20:30:55 +0000 (20:30 +0000)]
Avoid flushing IPv6 routes.  `route flush' removes necessary
routes for IPv6 as well as IPv4 routes.

23 years agoAdd in commented out entries for NEWCARD so that they are at least
jhb [Wed, 9 May 2001 19:37:25 +0000 (19:37 +0000)]
Add in commented out entries for NEWCARD so that they are at least
documented.  They cannot be turned on by default due to conflicting
symbols at link time between OLDCARD and NEWCARD.

Approved by: imp

23 years agoThe sk driver developed a bug when the multicast code was changed to
wpaul [Wed, 9 May 2001 18:22:42 +0000 (18:22 +0000)]
The sk driver developed a bug when the multicast code was changed to
use TAILQ macros. The sk_attach_xmac() routine calls sk_init_xmac()
before doing the transceiver probe, but *before* ether_ifattach()
is called. This causes sk_init_xmac() to call sk_setmulti(), which
tries to do a TAILQ_FOREACH(), which it can't do because ether_ifattach()
hasn't done a TAILQ_INIT() yet. This causes a NULL pointer dereference
and panic in sk_setmulti() at driver load/initialization time.

Fixed by calling ether_ifattach() before the MII probe.

The code in RELENG_4 still uses the old way of enumerating the
multicast list and doesn't have this problem. Yet.

23 years agoAdd include of sys/mutex.h and resort include of sys/lock.h.
jhb [Wed, 9 May 2001 16:56:48 +0000 (16:56 +0000)]
Add include of sys/mutex.h and resort include of sys/lock.h.

23 years agoAdd needed sys/lock.h include.
jhb [Wed, 9 May 2001 16:55:59 +0000 (16:55 +0000)]
Add needed sys/lock.h include.

23 years ago * Simplify the population of the /etc memory filesystem. To avoid
bsd [Wed, 9 May 2001 15:13:51 +0000 (15:13 +0000)]
  * Simplify the population of the /etc memory filesystem.  To avoid
    the null mount, we currently create a temporary mfs on /tmp, copy
    /etc to /tmp, then mount /etc as mfs and copy everything back from
    /tmp, then delete the /tmp mfs.

    The patch eliminates the temporary /tmp mfs and the subsequent
    copying and simply populates the /etc mfs by copying from
    /conf/default/etc.  This requires that /conf/default/etc contain a
    complete copy of all the /etc stuff instead of just overrides.  I
    don't think that is too much of an extra step in setting up a
    diskless environment.

  * Provide the ability to make /tmp a memory filesystem independent
    of /var.  This removes the requirement that /tmp be a symlink to
    /var/tmp and this makes the diskless code work with the default
    filesystem layout.  If a seperate /tmp memory filesystem is
    created, the 'tmpsize' environment variable is used to determine
    its size (default to 10 Meg).

  * Reduce diffs between the -current and -stable versions of these
    files to a bare minimum.  Only the definition of the shell
    function 'mount_md' is different.

Not Objected to by: -arch@, -small@

MFC after: 2 days

23 years agoFixed world breakage in previous commit. -lpam must never be used
bde [Wed, 9 May 2001 14:30:49 +0000 (14:30 +0000)]
Fixed world breakage in previous commit.  -lpam must never be used
directly (except in the definition of MINUSLPAM in bsd.libnames.mk)
since it doesn't give all the libraries necessary for static linkage.

Fixed missing ${LIBPAM} in DPADD.

Fixed some style bugs in DPADD and LDADD.

23 years agoFixed world breakage in previous commit. -lpam must never be used
bde [Wed, 9 May 2001 14:23:54 +0000 (14:23 +0000)]
Fixed world breakage in previous commit.  -lpam must never be used
directly (except in the definition of MINUSLPAM in bsd.libnames.mk)
since it doesn't give all the lbraries necessary for static linkage.

Fixed new and old bugs in DPADD.  ${LIBPAM} was missing, and the
library order was different from that in LDADD so `make checkdpadd'
reported a non-bug.

23 years agoRemove the error var, it hides the real one.
sos [Wed, 9 May 2001 13:01:10 +0000 (13:01 +0000)]
Remove the error var, it hides the real one.

PR 27213.

BTW the CDIOCREADAUDIO ioctl is deprecated, its not longer needed
and was an ugly hack from start on.

23 years agoAvoid overflow when converting ticks to jiffies.
des [Wed, 9 May 2001 11:41:54 +0000 (11:41 +0000)]
Avoid overflow when converting ticks to jiffies.

PR: 27215
Submitted by: Jim Pirzyk <Jim.Pirzyk@disney.com>

23 years agoFix operation of df on unmounted filesystems, and add the ability to run df
kris [Wed, 9 May 2001 08:44:15 +0000 (08:44 +0000)]
Fix operation of df on unmounted filesystems, and add the ability to run df
on unmounted non-UFS filesystem using '-t'

Submitted by: bde

23 years agoPatches from OpenBSD:
kris [Wed, 9 May 2001 08:37:18 +0000 (08:37 +0000)]
Patches from OpenBSD:
        - check the msg.tsp_type value prior to using it as an
          index into char *tsptype[]
        - use strlcpy's instead of strcpy's
        - & handle short packets properly.

Submitted by: "Andrew R. Reiter" <arr@watson.org>
Obtained from: OpenBSD

23 years agoFix what was clearly a 3am brain-o; Boolean should be signed, not
jkh [Wed, 9 May 2001 08:01:56 +0000 (08:01 +0000)]
Fix what was clearly a 3am brain-o; Boolean should be signed, not
unsigned.  C is kinda loose about this sort of thing but it's no excuse.

Spotted by: kenny

23 years agos/ssh_host_key/ssh_host_rsa_key/ since that is what openssh uses now
peter [Wed, 9 May 2001 07:46:44 +0000 (07:46 +0000)]
s/ssh_host_key/ssh_host_rsa_key/ since that is what openssh uses now
after a mergemaster.

23 years agoClarify that the Adaptec 2400A is an ATA controller.
scottl [Wed, 9 May 2001 05:58:01 +0000 (05:58 +0000)]
Clarify that the Adaptec 2400A is an ATA controller.

23 years agoAdd the Adaptec 2400A as an ATA RAID controller. Remove reference to the
scottl [Wed, 9 May 2001 05:56:49 +0000 (05:56 +0000)]
Add the Adaptec 2400A as an ATA RAID controller.  Remove reference to the
non-existant Adaptec 1400 controller.

23 years agoFix a typo (pasto?).
dd [Wed, 9 May 2001 05:03:35 +0000 (05:03 +0000)]
Fix a typo (pasto?).

23 years agoFix some of the handling in the pam module, don't unregister things
alfred [Wed, 9 May 2001 03:40:37 +0000 (03:40 +0000)]
Fix some of the handling in the pam module, don't unregister things
that were never registered.  At the same time handle a failure from
pam_setcreds with a bit more paranioa than the previous fix.

Sync a bit with the "Portable OpenSSH" work to make comparisons a easier.

23 years agoUnbreak world, IN_SHLOCK/IN_EXLOCK haven't existed in a while and
alfred [Wed, 9 May 2001 03:38:02 +0000 (03:38 +0000)]
Unbreak world, IN_SHLOCK/IN_EXLOCK haven't existed in a while and
Kirk finally has ditched them.  While I'm here also ditch FSHLOCK.

23 years agorcsid style fix.
obrien [Wed, 9 May 2001 01:38:28 +0000 (01:38 +0000)]
rcsid style fix.

23 years agoA WIP snuck in the rev 1.34 commit.
obrien [Wed, 9 May 2001 01:37:49 +0000 (01:37 +0000)]
A WIP snuck in the rev 1.34 commit.

23 years agoEliminate some panics for errors we can recover from.
dmlb [Wed, 9 May 2001 00:03:19 +0000 (00:03 +0000)]
Eliminate some panics for errors we can recover from.

Reduce the verbose memory map setup reports and work with pccardd to
set the common memory map up.

Use enumeration values for CARD_SET_RES_FLAGS.

Use DELAY when spinning waiting for the card to come free instead of a loop.

MFC: after 1 week

23 years agoUse enumeration values for CARD_SET_RES_FLAGS.
dmlb [Tue, 8 May 2001 23:59:13 +0000 (23:59 +0000)]
Use enumeration values for CARD_SET_RES_FLAGS.

Remove panic on out of range io window and return ENXIO. Add a similar
check for memory windows.

Approved by: imp

23 years agoUse enumeration types for CARD_SET_RES_FLAGS.
dmlb [Tue, 8 May 2001 23:57:32 +0000 (23:57 +0000)]
Use enumeration types for CARD_SET_RES_FLAGS.

Approved by: imp

23 years agoAdd additional enumeration types for CARD_SET_RES_FLAGS.
dmlb [Tue, 8 May 2001 23:56:47 +0000 (23:56 +0000)]
Add additional enumeration types for CARD_SET_RES_FLAGS.

Approved by: imp
MFC: after 1 week

23 years agoAdd a sysctl pair for the pcic memory allocation range
dmlb [Tue, 8 May 2001 22:51:05 +0000 (22:51 +0000)]
Add a sysctl pair for the pcic memory allocation range
machdep.pccard.pcic_mem_start
machdep.pccard.pcic_mem_end
and default the range to IOM_BEGIN/IOM_END.

This may prove useful to if_ray users (and others) on more modern
hardware that maps BIOS stuff into 0xd000-0xdffff.

MFC: after 1 week

Approved by: imp

23 years agoSince PAM is broken, let pam_setcred() failure be non-fatal.
green [Tue, 8 May 2001 22:30:18 +0000 (22:30 +0000)]
Since PAM is broken, let pam_setcred() failure be non-fatal.

23 years agoremove a reference to a nonexisting script, "clean", when
luigi [Tue, 8 May 2001 20:44:37 +0000 (20:44 +0000)]
remove a reference to a nonexisting script, "clean", when
building all images. Also, build "bridge" image as well.

Reported-by: Bruce Montague
23 years agoMFS: use absolute, not relative path for tinyware
luigi [Tue, 8 May 2001 20:38:52 +0000 (20:38 +0000)]
MFS: use absolute, not relative path for tinyware

23 years agoHere comes the forgotten C source, oopsy.
schweikh [Tue, 8 May 2001 20:27:14 +0000 (20:27 +0000)]
Here comes the forgotten C source, oopsy.
Turned the shell script into a binary fixing several minor buglets.
Mention _POSIX_SOURCE feature test macro in man page.
PR:           bin/19337
Submitted by: schweikh
Reviewed by:  joerg, bde
MFC after:    2 weeks

23 years agoTurned the shell script into a binary fixing several minor buglets.
schweikh [Tue, 8 May 2001 19:47:01 +0000 (19:47 +0000)]
Turned the shell script into a binary fixing several minor buglets.
Mention _POSIX_SOURCE feature test macro in man page.
PR: bin/19337
Submitted by: myself way back when I was a nobody :-) (schweikh)
Reviewed by: joerg, bde
MFC after: 2 weeks

23 years agoCleanups, more consistent use of <screen> etc.
wilko [Tue, 8 May 2001 18:42:46 +0000 (18:42 +0000)]
Cleanups, more consistent use of <screen> etc.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r76371,
assar [Tue, 8 May 2001 14:57:13 +0000 (14:57 +0000)]
This commit was generated by cvs2svn to compensate for changes in r76371,
which included commits to RCS files with non-trunk default branches.

23 years agomdoc(ng) fixes
assar [Tue, 8 May 2001 14:57:13 +0000 (14:57 +0000)]
mdoc(ng) fixes

Submitted by: ru

23 years agomdoc(ng) fixes
assar [Tue, 8 May 2001 14:57:13 +0000 (14:57 +0000)]
mdoc(ng) fixes

Submitted by: ru

23 years ago- Eliminate locks in functions called only during probe and attach.
tanimura [Tue, 8 May 2001 12:15:26 +0000 (12:15 +0000)]
- Eliminate locks in functions called only during probe and attach.
- Finish transmitting data to mpu when a buffer gets empty.

Submitted by: KUROSAWA Takahiro <fwkg7679@mb.infoweb.ne.jp>

23 years agogethostbyname2() can't do AF_INET6 lookups over NIS.
alex [Tue, 8 May 2001 11:21:15 +0000 (11:21 +0000)]
gethostbyname2() can't do AF_INET6 lookups over NIS.
getaddrinfo(3) must be used.

Submitted by: ume

23 years agoWhen opening the file to broadcast, do it with the user's gid, not the
kris [Tue, 8 May 2001 11:11:42 +0000 (11:11 +0000)]
When opening the file to broadcast, do it with the user's gid, not the
egid (tty).

Obtained from: OpenBSD
Submitted by: Maxime Henrion <mux@qualys.com>
Reviewed by: imp

23 years agoPolish error handling with biofinish().
phk [Tue, 8 May 2001 09:10:27 +0000 (09:10 +0000)]
Polish error handling with biofinish().

23 years agoPolish error handling code using biofinish()
phk [Tue, 8 May 2001 09:09:32 +0000 (09:09 +0000)]
Polish error handling code using biofinish()

23 years agoRemove an 'optimization' I hope to never see again.
alfred [Tue, 8 May 2001 09:09:18 +0000 (09:09 +0000)]
Remove an 'optimization' I hope to never see again.

The pipe code could not handle running out of kva, it would panic
if that happened.  Instead return ENFILE to the application which
is an acceptable error return from pipe(2).

There was some slightly tricky things that needed to be worked on,
namely that the pipe code can 'realloc' the size of the buffer if
it detects that the pipe could use a bit more room.  However if it
failed the reallocation it could not cope and would panic.  Fix
this by attempting to grow the pipe while holding onto our old
resources.  If all goes well free the old resources and use the
new ones, otherwise continue to use the smaller buffer already
allocated.

While I'm here add a few blank lines for style(9) and remove
'register'.

23 years agoBe a little clearer all characters after a # are ignored unless the
brian [Tue, 8 May 2001 08:58:57 +0000 (08:58 +0000)]
Be a little clearer all characters after a # are ignored unless the
# is escaped or quoted.  Add an example of # characters as part of
a phone number.

PR: 26605

23 years agoExploit recent improvements in the disk minilayer to simplify error
phk [Tue, 8 May 2001 08:30:48 +0000 (08:30 +0000)]
Exploit recent improvements in the disk minilayer to simplify error
handling a bit.

Dogmatic lingupurists can celebrate that a number of gotos got removed.

Reviewed by: mjacob, ken

23 years agoAlways initialize bio_resid from bio_bcount in the disk mini-layer so
phk [Tue, 8 May 2001 08:24:54 +0000 (08:24 +0000)]
Always initialize bio_resid from bio_bcount in the disk mini-layer so
that the drivers don't have to do it umpteen times.

23 years agomdoc(7) police: fix markup, rename and reorder some sections.
ru [Tue, 8 May 2001 08:12:53 +0000 (08:12 +0000)]
mdoc(7) police: fix markup, rename and reorder some sections.

23 years agoRemove all the mutex stuff - suggested by jhb
brian [Tue, 8 May 2001 07:55:33 +0000 (07:55 +0000)]
Remove all the mutex stuff - suggested by jhb

Tidy up includes, credit Slawa Olhovchenkov, John Prince and Eric Hernes
for their efforts and add a couple of missing parenthesis around return
expressions.

23 years agoWhen running with soft updates, track the number of blocks and files
mckusick [Tue, 8 May 2001 07:42:20 +0000 (07:42 +0000)]
When running with soft updates, track the number of blocks and files
that are committed to being freed and reflect these blocks in the
counts returned by statfs (and thus also by the `df' command). This
change allows programs such as those that do news expiration to
know when to stop if they are trying to create a certain percentage
of free space. Note that this change does not solve the much harder
problem of making this to-be-freed space available to applications
that want it (thus on a nearly full filesystem, you may still
encounter out-of-space conditions even though the free space will
show up eventually). Hopefully this harder problem will be the
subject of a future enhancement.

23 years agoSeveral fixes for units errors:
mckusick [Tue, 8 May 2001 07:29:03 +0000 (07:29 +0000)]
Several fixes for units errors:
1) Do not assume that the superblock will be of size fs->fs_bsize.
   This fixes a panic when taking a snapshot on a filesystem with
   a block size bigger than 8K.
2) Properly calculate the number of fragments that follow the
   superblock summary information. This fixes a bug with inconsistent
   snapshots.
3) When cleaning up a snapshot that is about to be removed, properly
   calculate the number of blocks that need to be checked. This fixes
   a bug that created partially allocated inodes.
4) When moving blocks from a snapshot that is about to be removed
   to another snapshot, properly account for the reduced number of
   blocks in the snapshot from which they are taken. This fixes a
   bug in which the number of blocks released from a snapshot did not
   match the number that it claimed to have.

23 years agomdoc(7) police: sort xrefs.
ru [Tue, 8 May 2001 07:19:19 +0000 (07:19 +0000)]
mdoc(7) police: sort xrefs.

23 years agoWhen syncing out snapshot metadata, we must temporarily allow recursive
mckusick [Tue, 8 May 2001 07:13:00 +0000 (07:13 +0000)]
When syncing out snapshot metadata, we must temporarily allow recursive
buffer locking so as to avoid locking against ourselves if we need to
write filesystem metadata.

23 years agoGC some dead code relating to running df on unmounted block devices,
kris [Tue, 8 May 2001 06:58:25 +0000 (06:58 +0000)]
GC some dead code relating to running df on unmounted block devices,
and remove the setgid operator bit from the installed binary: if you want
to view free disk space on an unmounted device, you should have read
permissions to access it.

Reviewed by: phk

23 years agoJust notify us once when encountering a partially allocated inode.
mckusick [Tue, 8 May 2001 06:41:56 +0000 (06:41 +0000)]
Just notify us once when encountering a partially allocated inode.

23 years agoSync up with OpenBSD. Too many changes to note, but the major features
kris [Tue, 8 May 2001 06:19:06 +0000 (06:19 +0000)]
Sync up with OpenBSD.  Too many changes to note, but the major features
are:
* Implement cpio compatibility mode when pax is invoked as cpio
* Extend tar compatibility mode to cover many of the GNU tar single-letter
  options (bzip2 mode, aka -y/-j is not present in OpenBSD).  When
  invoked as tar, pax is now full-featured enough for use by the ports
  collection to extract distfiles and create packages.
* Many bug fixes to the operation of pax and the tar compatibility modes
* Code fixes for things like correct string buffer termination.

I tried to preserve existing FreeBSD fixes to this utility; please let me
know if I have inadvertently spammed something.

23 years agoRicoh RL5C46x cardbus bridges have the bits for 3E0 and 3E2. The
imp [Tue, 8 May 2001 02:28:41 +0000 (02:28 +0000)]
Ricoh RL5C46x cardbus bridges have the bits for 3E0 and 3E2.  The
RL5C47x cards do not.  Only set them for that set of bridges.

Submitted by: shiba (Takeshi Shibagaki-san)

23 years agoAdd some additional register definitions for some work I have in progress.
imp [Tue, 8 May 2001 02:06:03 +0000 (02:06 +0000)]
Add some additional register definitions for some work I have in progress.

23 years agosys/mutex.h requires sys/lock.h for LINT
brian [Mon, 7 May 2001 23:52:08 +0000 (23:52 +0000)]
sys/mutex.h requires sys/lock.h for LINT

Re-spotted by: phk

23 years agoCorrect prototype (entry_p -> *entry_p)
jedgar [Mon, 7 May 2001 23:16:25 +0000 (23:16 +0000)]
Correct prototype (entry_p -> *entry_p)

Submitted by: Alex Zepeda <jazepeda@pacbell.net>

23 years agoSet the slot pointer in the pc98 case. Correct the name of the bridge
imp [Mon, 7 May 2001 22:28:05 +0000 (22:28 +0000)]
Set the slot pointer in the pc98 case.  Correct the name of the bridge
chip to the one that the Japanese use.  Now we get insert/remove
events on my PC-9821Ne.  More work in bus space is needed to make
drivers work.

MFC after: 3 days

23 years agoFix typos.
horikawa [Mon, 7 May 2001 22:15:18 +0000 (22:15 +0000)]
Fix typos.
o replace `of possible' with `if possible'
o VOP_SETACL(9) is modified to say about `ACL' instead of `extended
  attributed'
o EOPNOTSUPP of VOP_SETEXTATTR(9) is modified to say about
  VOP_SETEXTATTR(9) instead of VOP_GETEXTATTR(9)

Reviewed by: Robert Watson <rwatson@FreeBSD.org>,
 Chris Costello <chris@calldei.com>

23 years agoUse constants in <pccard/cis.h> for scannign the memory window with.
dmlb [Mon, 7 May 2001 21:58:31 +0000 (21:58 +0000)]
Use constants in <pccard/cis.h> for scannign the memory window with.

Approved by: imp

23 years agoEnsure that pccardd sets up memory windows correctly for drivers other
dmlb [Mon, 7 May 2001 21:56:57 +0000 (21:56 +0000)]
Ensure that pccardd sets up memory windows correctly for drivers other
than if_ed. The code for if_ed to set the offset and memory width
remains.

Approved by: imp

23 years agoFix a minor printing bug that prints incorrect information for memory
dmlb [Mon, 7 May 2001 21:51:54 +0000 (21:51 +0000)]
Fix a minor printing bug that prints incorrect information for memory
block sizes.

This orginally worked in PAO-3 and worked on their r330 branch but got
broken in PAO-3 around December 1998!

Approved by: imp
Obtained from: PAO-3

23 years agoMinor updates:
msmith [Mon, 7 May 2001 21:46:44 +0000 (21:46 +0000)]
Minor updates:

 - Rework of twe_report_request to use the command status value rather
   than the flags register. (Joel Jacobson @ 3ware)
 - Update to match some changes in -current vs. stable.

MFC in: 1 week

23 years agoPointy hat fix -- reapply the SRA PAM patch. To -current this time.
nsayer [Mon, 7 May 2001 20:42:02 +0000 (20:42 +0000)]
Pointy hat fix -- reapply the SRA PAM patch. To -current this time.

23 years agoAdd PAM support to SRA authentication. Cribbed mostly from ftpd. This
nsayer [Mon, 7 May 2001 20:38:39 +0000 (20:38 +0000)]
Add PAM support to SRA authentication. Cribbed mostly from ftpd. This
doesn't solve the problem of root being allowed to log in, but that sort
of thing is something PAM should be doing anyway.

23 years agoProperly copy the P_ALTSTACK flag in struct proc::p_flag to the child
knu [Mon, 7 May 2001 18:07:29 +0000 (18:07 +0000)]
Properly copy the P_ALTSTACK flag in struct proc::p_flag to the child
process on fork(2).

It is the supposed behavior stated in the manpage of sigaction(2), and
Solaris, NetBSD and FreeBSD 3-STABLE correctly do so.

The previous fix against libc_r/uthread/uthread_fork.c fixed the
problem only for the programs linked with libc_r, so back it out and
fix fork(2) itself to help those not linked with libc_r as well.

PR: kern/26705
Submitted by: KUROSAWA Takahiro <fwkg7679@mb.infoweb.ne.jp>
Tested by: knu, GOTOU Yuuzou <gotoyuzo@notwork.org>,
and some other people
Not objected by: hackers
MFC in: 3 days

23 years agoDisable the card after sending the removed event up to the pccard
imp [Mon, 7 May 2001 16:50:34 +0000 (16:50 +0000)]
Disable the card after sending the removed event up to the pccard
layer.  This fixes an ordering problem that would cause the ISR for
the device to run with now power applied to the device.  Most cards
failed to deal with this gracefully, and thus would hang on card
eject.

The power down event, for those keeping score, is what causes the
interrupt for the card.

Many folks in the Japanese nomads list have reported this, so I'll be
MFCing quickly for their benefit.

Submitted by: Masayuki FUKUI
MFC after: 2 days

23 years agoOnly load KOI8-R-specific bits if document language is set
ru [Mon, 7 May 2001 16:04:49 +0000 (16:04 +0000)]
Only load KOI8-R-specific bits if document language is set
via the \*[lang] register, not if just device is `koi8-r'.

Reviewed by: ache

23 years agoUse foo () instead of foo ( ) for function definition,
schweikh [Mon, 7 May 2001 14:15:02 +0000 (14:15 +0000)]
Use foo () instead of foo ( ) for function definition,
so zsh can parse this file as well.

23 years agoRemove if_ray_oldcard.h because pccard support multiple windows now.
dmlb [Mon, 7 May 2001 13:11:08 +0000 (13:11 +0000)]
Remove if_ray_oldcard.h because pccard support multiple windows now.

Setup attribute memory resource in ray_probe so that it is added to
the print out of the resource list on card insertion.

23 years agoA bit of markup and spelling fixes.
ru [Mon, 7 May 2001 12:48:38 +0000 (12:48 +0000)]
A bit of markup and spelling fixes.

23 years agoRestore printing of routes cloned from a RTF_CLONING parent
ru [Mon, 7 May 2001 12:29:36 +0000 (12:29 +0000)]
Restore printing of routes cloned from a RTF_CLONING parent
(e.g., ARP table entries) with the default routing display.

PR: bin/26970
Reviewed by: wollman
MFC after: 3 days

23 years agoChange COM_LOCK/COM_UNLOCK to a regular mutex - still conditional on
brian [Mon, 7 May 2001 11:13:11 +0000 (11:13 +0000)]
Change COM_LOCK/COM_UNLOCK to a regular mutex - still conditional on
SMP being defined.

23 years agoCosmetics: .Dl -> .Li.
ru [Mon, 7 May 2001 07:42:56 +0000 (07:42 +0000)]
Cosmetics: .Dl -> .Li.

23 years agos/mtx_try_enter/mtx_trylock/
dd [Sun, 6 May 2001 22:04:17 +0000 (22:04 +0000)]
s/mtx_try_enter/mtx_trylock/

23 years agoMake the disk mini-layer check for and handle zero-length transfers
phk [Sun, 6 May 2001 21:55:22 +0000 (21:55 +0000)]
Make the disk mini-layer check for and handle zero-length transfers
instead of the underlying drivers.

23 years agoMake LINT compile again.
brian [Sun, 6 May 2001 21:03:37 +0000 (21:03 +0000)]
Make LINT compile again.

Spotted by: phk

23 years agoActually biofinish(struct bio *, struct devstat *, int error) is more general
phk [Sun, 6 May 2001 20:00:03 +0000 (20:00 +0000)]
Actually biofinish(struct bio *, struct devstat *, int error) is more general
than the bioerror().

Most of this patch is generated by scripts.

23 years agoIntroduce bioerror(struct bio*, int err, int complete);
phk [Sun, 6 May 2001 18:57:08 +0000 (18:57 +0000)]
Introduce bioerror(struct bio*, int err, int complete);

23 years agoRemove unneeded devfs_badop()
phk [Sun, 6 May 2001 17:40:34 +0000 (17:40 +0000)]
Remove unneeded devfs_badop()

Noticed by: rwatson

23 years agoFix return type of vop_stdputpages()
phk [Sun, 6 May 2001 17:40:22 +0000 (17:40 +0000)]
Fix return type of vop_stdputpages()

Noticed by: rwatson

23 years agoFix a panic if MD devices were left half-created.
phk [Sun, 6 May 2001 17:17:23 +0000 (17:17 +0000)]
Fix a panic if MD devices were left half-created.

XXX: the real bug is that devstat isn't part of the disk minilayer.

PR: 27158
Submitted by: Anders Nordby <anders@fix.no>

23 years agoAdd elf.h header back, its existance is an SVR4-ELF tradition.
obrien [Sun, 6 May 2001 16:39:57 +0000 (16:39 +0000)]
Add elf.h header back, its existance is an SVR4-ELF tradition.
Our ELF hints bits are still a seperate file.

Requested by: jdp

23 years agoo First step in cleaning up authorization code for the posix4
rwatson [Sun, 6 May 2001 16:15:42 +0000 (16:15 +0000)]
o First step in cleaning up authorization code for the posix4
  implementation.  Move from direct uid 0 comparision to using suser_xxx()
  call with the same semantics.  Simplify CAN_AFFECT() macro as passed
  pcred was redundant.  The checks here still aren't "right", but they
  are probably "better".

Obtained from: TrustedBSD Project

23 years agoSGMLify now completed. Comments are invited. I hope not to have introduced
wilko [Sun, 6 May 2001 13:43:24 +0000 (13:43 +0000)]
SGMLify now completed. Comments are invited. I hope not to have introduced
too much typos / errors during the conversion.

23 years agoUpdate the DiskOnChip firmware to OSAK version 4.1.
phk [Sun, 6 May 2001 12:37:42 +0000 (12:37 +0000)]
Update the DiskOnChip firmware to OSAK version 4.1.

Sponsored by: Redfern Broadband Networks (www.RedFernNetworks.com)

23 years agoRestore I/O port resources to the condition before adv_isa_probe() is
nyan [Sun, 6 May 2001 09:28:18 +0000 (09:28 +0000)]
Restore I/O port resources to the condition before adv_isa_probe() is
called.

Submitted by: yokota

23 years agoMerged from sys/i386/i386/machdep.c revision 1.448.
kato [Sun, 6 May 2001 09:18:32 +0000 (09:18 +0000)]
Merged from sys/i386/i386/machdep.c revision 1.448.

23 years agoUse correct resource id for bus_release_resource().
nyan [Sun, 6 May 2001 08:33:29 +0000 (08:33 +0000)]
Use correct resource id for bus_release_resource().

23 years agoMove unused functions into #if 0 ... #endif.
nyan [Sun, 6 May 2001 08:07:10 +0000 (08:07 +0000)]
Move unused functions into #if 0 ... #endif.

23 years agoMerged from sys/i386/conf/GENERIC revision 1.304.
nyan [Sun, 6 May 2001 05:57:46 +0000 (05:57 +0000)]
Merged from sys/i386/conf/GENERIC revision 1.304.

23 years agoMerged from sys/isa/fd.c revision 1.193.
nyan [Sun, 6 May 2001 05:49:11 +0000 (05:49 +0000)]
Merged from sys/isa/fd.c revision 1.193.

23 years agoCreate temporary files securely using mkstemp() instead of mktemp()
kris [Sun, 6 May 2001 03:07:12 +0000 (03:07 +0000)]
Create temporary files securely using mkstemp() instead of mktemp()

Reviewed by: mikeh, audit@
MFC after: 1 week