]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoSync with sys/i386/conf/Makefile.i386 revision 1.129.
kato [Mon, 16 Nov 1998 08:09:16 +0000 (08:09 +0000)]
Sync with sys/i386/conf/Makefile.i386 revision 1.129.

25 years agovmunix -> kernel
nectar [Mon, 16 Nov 1998 04:25:03 +0000 (04:25 +0000)]
vmunix -> kernel
PR: port/8543
Submitted by: jkb@best.com

25 years agoUpdate to reflect reality.
nectar [Mon, 16 Nov 1998 03:49:39 +0000 (03:49 +0000)]
Update to reflect reality.
PR: kern/8629

25 years agoput hosts before bind.
jkh [Mon, 16 Nov 1998 02:02:30 +0000 (02:02 +0000)]
put hosts before bind.

25 years agoUpdate tcopy for CAM. (/dev/rst0 -> /dev/rsa0)
ken [Mon, 16 Nov 1998 00:05:46 +0000 (00:05 +0000)]
Update tcopy for CAM.  (/dev/rst0 -> /dev/rsa0)

PR: i386/8688
Submitted by: Bill Trost <trost@cloud.rain.com>

25 years agoNow that the wd driver is fixed (Thanks Bruce!), re-enable the
ken [Sun, 15 Nov 1998 23:57:22 +0000 (23:57 +0000)]
Now that the wd driver is fixed (Thanks Bruce!), re-enable the
devstat_end_transaction error message that gets printed whenever the
busy count is < 0.

This will help catch drivers that improperly implement devstat(9) support.

25 years agoUpdate ssh version number. It's only informational, anyway.
abial [Sun, 15 Nov 1998 23:30:26 +0000 (23:30 +0000)]
Update ssh version number. It's only informational, anyway.

Submitted by: jkh

25 years agoUse console instead of ttyv0 - it didn't work with serial console...
abial [Sun, 15 Nov 1998 23:26:32 +0000 (23:26 +0000)]
Use console instead of ttyv0 - it didn't work with serial console...

25 years agoUse snprintf to make sure we don't overflow a buffer.
ken [Sun, 15 Nov 1998 23:17:39 +0000 (23:17 +0000)]
Use snprintf to make sure we don't overflow a buffer.

25 years agoFix an error message. (it was using an uninitialized variable)
ken [Sun, 15 Nov 1998 23:12:42 +0000 (23:12 +0000)]
Fix an error message.  (it was using an uninitialized variable)

Reported by: dan@math.berkeley.edu (Dan Strick)

25 years agoThe below patch helps to reduce the leakage of internal socket information
guido [Sun, 15 Nov 1998 21:35:09 +0000 (21:35 +0000)]
The below patch helps to reduce the leakage of internal socket information
when a TCP "stealth" scan is directed at a *BSD box by ensuring the window
is 0 for all RST packets generated through tcp_respond()
Reviewed by: Don Lewis <Don.Lewis@tsc.tdk.com>
Obtained from: Bugtraq (from: Darren Reed <avalon@COOMBS.ANU.EDU.AU>)

25 years agoReimplement the NFS ACCESS RPC cache as an "accelerator" rather than a true
msmith [Sun, 15 Nov 1998 20:36:18 +0000 (20:36 +0000)]
Reimplement the NFS ACCESS RPC cache as an "accelerator" rather than a true
cache.  If the cached result lets us say "yes", then go with that.  If
we're not sure, or we think the answer might be "no", go to the wire to be
certain.    This avoids all of the possible false veto cases, and allows us
to key the cached value with just the UID for which the cached value holds,
reducing the bloat of the nfsnode structure from 104 bytes to just 12 bytes.

Since the "yes" case is by far the most common, this should still provide
a substantial performance improvement.  Also default the cache to on, with
a conservative timeout (2 seconds).  This improves performance if NFS is
loaded as a KLD module, as there's not (yet) code to parse an option out
of the module arguments to set it, and sysctl doesn't work (yet) for OIDs
in modules.

The 'accelerator' mode was suggested by Bjoern Groenvall (bg@sics.se)

Feedback on this would be appreciated as testing has been necessarily
limited by Comdex, and it would be valuable to have this in 2.2.8.

25 years agoImplement the nfs_access_cache variable, allowing us to set the timeout for
msmith [Sun, 15 Nov 1998 20:30:04 +0000 (20:30 +0000)]
Implement the nfs_access_cache variable, allowing us to set the timeout for
the NFS client's ACCESS cache.

25 years agoDocument the new nfs_access_cache variable.
msmith [Sun, 15 Nov 1998 20:29:27 +0000 (20:29 +0000)]
Document the new nfs_access_cache variable.

25 years agoMake it possible to adjust the IDE probe delay from kernel config files.
eivind [Sun, 15 Nov 1998 20:08:50 +0000 (20:08 +0000)]
Make it possible to adjust the IDE probe delay from kernel config files.

25 years agoDepend on __FreeBSD_version instead of __FreeBSD__
eivind [Sun, 15 Nov 1998 19:30:48 +0000 (19:30 +0000)]
Depend on __FreeBSD_version instead of __FreeBSD__

25 years ago* Add hooks to allow the X server to access I/O ports and memory.
dfr [Sun, 15 Nov 1998 18:25:17 +0000 (18:25 +0000)]
* Add hooks to allow the X server to access I/O ports and memory.
* Update drivers to the latest version of the bus interface.

The ISA drivers' use of the new resource api is minimal.  Garrett has
some much cleaner drivers which should be more easily shared between
i386 and alpha.  This has only been tested on cia based machines.  It
should work on lca and apecs but I might have broken something.

25 years agoBuild subr_rman.c by default.
dfr [Sun, 15 Nov 1998 18:16:23 +0000 (18:16 +0000)]
Build subr_rman.c by default.

25 years agoDon't include link_elf.c twice in the link.
dfr [Sun, 15 Nov 1998 18:15:06 +0000 (18:15 +0000)]
Don't include link_elf.c twice in the link.

25 years agoA couple of minor sanity checks.
dfr [Sun, 15 Nov 1998 18:11:21 +0000 (18:11 +0000)]
A couple of minor sanity checks.

Submitted by: Nick Hibma <nick.hibma@jrc.it>

25 years agoFix printf format errors on alpha.
dfr [Sun, 15 Nov 1998 18:10:14 +0000 (18:10 +0000)]
Fix printf format errors on alpha.

25 years ago* Change 'struct resource' to 'struct config_resource'.
dfr [Sun, 15 Nov 1998 18:07:35 +0000 (18:07 +0000)]
* Change 'struct resource' to 'struct config_resource'.
* Bump config version.

25 years agoIn sendfile(2), check against sb_lowat when filling the socket buffer,
dg [Sun, 15 Nov 1998 16:55:09 +0000 (16:55 +0000)]
In sendfile(2), check against sb_lowat when filling the socket buffer,
rather than 0.

25 years agoFixed a misspelling of boolean true as MNT_WAIT.
bde [Sun, 15 Nov 1998 15:46:33 +0000 (15:46 +0000)]
Fixed a misspelling of boolean true as MNT_WAIT.

25 years agoFinished updating module event handlers to be compatible with
bde [Sun, 15 Nov 1998 15:33:52 +0000 (15:33 +0000)]
Finished updating module event handlers to be compatible with
modeventhand_t.

25 years agoPrevent MakeMaker to produce /usr/local/local/bin path for
ache [Sun, 15 Nov 1998 15:26:12 +0000 (15:26 +0000)]
Prevent MakeMaker to produce /usr/local/local/bin path for
INSTALLSCRIPT and INSTALLBIN: it broke ports

25 years agoDon't generate module event handlers of the wrong (old) type.
bde [Sun, 15 Nov 1998 15:25:56 +0000 (15:25 +0000)]
Don't generate module event handlers of the wrong (old) type.

Fixed some pedantic syntax errors (an extra semicolon in each
SYSUNINIT() expansion).

25 years agoFixed the type of vfs_modevent (a little late, to match the corresponding
bde [Sun, 15 Nov 1998 15:18:30 +0000 (15:18 +0000)]
Fixed the type of vfs_modevent (a little late, to match the corresponding
change in sys/mount.h).

25 years agoFixed the type and order of vfs_modevent. This fixes part of a spew of
bde [Sun, 15 Nov 1998 15:12:58 +0000 (15:12 +0000)]
Fixed the type and order of vfs_modevent.  This fixes part of a spew of
warnings for the recent change of the type of a module event handler.

Fixed a rotted comment (numeric types of filesystems are not listed here).

Made the function protototype in VFS_SET() more like the corresponding
function definition (don't use extern for prototypes).

Enforce a semicolon after the LKM case of VFS_SET().

25 years agoFixed a missing include. <sys/kernel.h> is needed by the new
bde [Sun, 15 Nov 1998 14:11:06 +0000 (14:11 +0000)]
Fixed a missing include.  <sys/kernel.h> is needed by the new
MALLOC_DEFINE() and MALLOC_DEFINE() is needed by the recently
reenabled "reallocblks" code, but <sys/kernel.h> was only included
if CLUSTERDEBUG was defined.  This was too harmless.  gcc only
warns about garbage like `SYSINIT(blech);' at file scope ...

25 years agoFixed the bug that caused negative busycounts in devstat. The
bde [Sun, 15 Nov 1998 13:54:31 +0000 (13:54 +0000)]
Fixed the bug that caused negative busycounts in devstat.  The
devstart_start_transaction() call is misplaced - it is after the
wdustart() call that queues the transaction on the controller queue.
Normally this doesn't matter because we're running at splbio() so
nothing will look at the controller queue.  However, obsolescent
code for syncing labels sometimes slept after the transaction was
started, so the transaction sometimes completed before it was
[recorded as] started.  This code was misplaced even for syncing
labels.  Move it to the right place.  It should go away, but
something may depend on its side effects.

25 years agoRemove share/info from build tools because it breaks the build when
jb [Sun, 15 Nov 1998 10:03:12 +0000 (10:03 +0000)]
Remove share/info from build tools because it breaks the build when
INFODIR is set.

25 years agoInterrupt threads waiting in select etc.
jb [Sun, 15 Nov 1998 10:01:34 +0000 (10:01 +0000)]
Interrupt threads waiting in select etc.

Submitted by: Alec Wolman <wolman@cs.washington.edu>

25 years agoClose a window between unlocking a spinlock and changing the thread state.
jb [Sun, 15 Nov 1998 09:58:26 +0000 (09:58 +0000)]
Close a window between unlocking a spinlock and changing the thread state.

25 years agoo Break fixups into binary fixups and XFree86 based fixups.
jkh [Sun, 15 Nov 1998 09:06:20 +0000 (09:06 +0000)]
o Break fixups into binary fixups and XFree86 based fixups.
o Move fixups into extraction routine so all consumers don't have to duplicate
  the right behavior.
o Make some things more orthogonal (just for asthetics sake)
o Add option to go back and do it again if XF86Setup fails (possibly with
  a different setup - this one has always annoyed me).

25 years agoFixed missing 'v' in usage message.
bde [Sun, 15 Nov 1998 05:55:58 +0000 (05:55 +0000)]
Fixed missing 'v' in usage message.

25 years agoFixed bugs in previous commit:
bde [Sun, 15 Nov 1998 05:51:55 +0000 (05:51 +0000)]
Fixed bugs in previous commit:
- missing `v' in synopsis.
- 'v' disordered relative to 'V' in description of options.
- line too long in description of 'v'.

25 years agoNew file formats.
obrien [Sun, 15 Nov 1998 03:35:10 +0000 (03:35 +0000)]
New file formats.

Obtained from: NetBSD-current/src/usr.bin/file/magdir

25 years agoAdditional file formats.
obrien [Sun, 15 Nov 1998 03:34:31 +0000 (03:34 +0000)]
Additional file formats.
Obtained from: NetBSD-current/src/usr.bin/file/magdir/msdos

25 years agoindent
dima [Sun, 15 Nov 1998 00:50:59 +0000 (00:50 +0000)]
indent

25 years agoUse mmap for file coping. 5-10% faster than the read system call.
wosch [Sat, 14 Nov 1998 23:43:11 +0000 (23:43 +0000)]
Use mmap for file coping. 5-10% faster than the read system call.

25 years agoUse mmap for reading files. This is faster than read().
wosch [Sat, 14 Nov 1998 23:37:32 +0000 (23:37 +0000)]
Use mmap for reading files. This is faster than read().

25 years agoFixed a couple of nits in sendfile(2): clear PG_ZERO before unbusying
dg [Sat, 14 Nov 1998 23:36:17 +0000 (23:36 +0000)]
Fixed a couple of nits in sendfile(2): clear PG_ZERO before unbusying
the page, and use passed-in "p" rather than curproc in uio struct.

25 years agoMy changes to the new device interface:
wollman [Sat, 14 Nov 1998 21:58:51 +0000 (21:58 +0000)]
My changes to the new device interface:

- Interface wth the new resource manager.
- Allow for multiple drivers implementing a single devclass.
- Remove ordering dependencies between header files.
- Style cleanup.
- Add DEVICE_SUSPEND and DEVICE_RESUME methods.
- Move to a single-phase interrupt setup scheme.

Kernel builds on the Alpha are brken until Doug gets a chance to incorporate
these changes on that side.

Agreed to in principle by: dfr

25 years agoChange USERCONFIG_BOOT -> INTRO_USERCONFIG
abial [Sat, 14 Nov 1998 20:10:23 +0000 (20:10 +0000)]
Change USERCONFIG_BOOT -> INTRO_USERCONFIG

25 years agoAdded a -v (verbose) option and hid multi-job file banner generation
dg [Sat, 14 Nov 1998 16:15:04 +0000 (16:15 +0000)]
Added a -v (verbose) option and hid multi-job file banner generation
behind it.

25 years agoAdd a missing newline
luigi [Sat, 14 Nov 1998 13:07:59 +0000 (13:07 +0000)]
Add a missing newline
Submitted by: Roger Hardiman

25 years agoAdd EXTRA_PKG_FLAGS to PKG_FLAGS. You can now add arguments to pkg_create
asami [Sat, 14 Nov 1998 09:45:09 +0000 (09:45 +0000)]
Add EXTRA_PKG_FLAGS to PKG_FLAGS.  You can now add arguments to pkg_create
without redefining PKG_FLAGS after bsd.port.mk.

25 years agoupdate for latest vendor import
obrien [Sat, 14 Nov 1998 03:37:59 +0000 (03:37 +0000)]
update for latest vendor import

25 years agoClean up conflicts.
obrien [Sat, 14 Nov 1998 03:13:33 +0000 (03:13 +0000)]
Clean up conflicts.

25 years agoThis commit was generated by cvs2svn to compensate for changes in r41142,
obrien [Sat, 14 Nov 1998 03:11:37 +0000 (03:11 +0000)]
This commit was generated by cvs2svn to compensate for changes in r41142,
which included commits to RCS files with non-trunk default branches.

25 years agoVirgin import of AMD (am-utils) v6.0b1
obrien [Sat, 14 Nov 1998 03:11:37 +0000 (03:11 +0000)]
Virgin import of AMD (am-utils) v6.0b1

25 years agoAdd documentation for $rootdev.
msmith [Fri, 13 Nov 1998 23:53:02 +0000 (23:53 +0000)]
Add documentation for $rootdev.

25 years agoAdd a new variable $num_ide_disks which is used to offset the unit number
msmith [Fri, 13 Nov 1998 23:40:02 +0000 (23:40 +0000)]
Add a new variable $num_ide_disks which is used to offset the unit number
for SCSI disks when converting from BIOS unit numbers to da unit numbers.

Prompted by Kevin Street <street@iname.com>

25 years agoAvoid a null pointer reference if the target of an NFS rename has been
msmith [Fri, 13 Nov 1998 22:58:48 +0000 (22:58 +0000)]
Avoid a null pointer reference if the target of an NFS rename has been
sillrenamed, or if the source vnode doesn't have an associated nfsnode.

Bug report from Andrew Gallatin <gallatin@cs.duke.edu>

25 years agoDon't count non-local dirty buffers as outstanding when shutting down.
msmith [Fri, 13 Nov 1998 22:40:37 +0000 (22:40 +0000)]
Don't count non-local dirty buffers as outstanding when shutting down.
This avoids the fsck-on-reboot symptoms if you're shutting down with a
hung or unreachable NFS server mounted.  Also remove non-local
filesystems from the mount list to prevent the system hanging when it tries
to unmount them (for the same reason).

Drew points out that there's a good argument for forcibly removing all
"non syncable" filesystems from the mount list (eg. NFS mounts, disks
that aren't responding, etc.) as this then allows you to sync and
cleanly unmount their parents.  No such change is included in this
patch.

Submitted by: Andrew Gallatin <gallatin@cs.duke.edu>

25 years agoIncreased PID_MAX to 99999. The main reason for doing this is to make the
dg [Fri, 13 Nov 1998 17:53:55 +0000 (17:53 +0000)]
Increased PID_MAX to 99999. The main reason for doing this is to make the
pid space somewhat more sparse which improves the performance of finding
an unused pid on systems with large numbers of processes. The new value
was chosen so that it doesn't overflow the 5 digit pid fields in various
programs.

25 years agoMove a diagnostic behind "if (bootverbose)".
gibbs [Fri, 13 Nov 1998 13:23:36 +0000 (13:23 +0000)]
Move a diagnostic behind "if (bootverbose)".

25 years agoFix a panic in nfsrv_dorec() where a NULL pointer could be passed to
dfr [Fri, 13 Nov 1998 09:44:12 +0000 (09:44 +0000)]
Fix a panic in nfsrv_dorec() where a NULL pointer could be passed to
free() sometimes.

Reviewed by: Eric Haug <ejh@eas.slu.edu>

25 years agoA couple of fixes to device_delete_child() to delete all children of
dfr [Fri, 13 Nov 1998 09:39:37 +0000 (09:39 +0000)]
A couple of fixes to device_delete_child() to delete all children of
the device and to free the device which is being deleted (not its
parent - oops).

Submitted by: Nick Hibma <nick.hibma@jrc.it>

25 years agoSync with sys/i386/conf/GENERIC revision 1.131.
kato [Fri, 13 Nov 1998 08:18:44 +0000 (08:18 +0000)]
Sync with sys/i386/conf/GENERIC revision 1.131.

25 years agoSync with sys/i386/boot/kzipboot/malloc.c revision 1.7.
kato [Fri, 13 Nov 1998 08:18:14 +0000 (08:18 +0000)]
Sync with sys/i386/boot/kzipboot/malloc.c revision 1.7.

25 years agoImplement NFS ACCESS RPC result caching.
msmith [Fri, 13 Nov 1998 02:39:09 +0000 (02:39 +0000)]
Implement NFS ACCESS RPC result caching.

This yields startling performance increases for NFS clients for many
access profiles, due to the fact that ACCESS results are persistently
cached in the namecache in many cases.

Note that the code is somewhat conservative in that it requires an
exact credential match for a cache hit.  This bloats the nfsnode
structure by sizeof(struct ucred) (96 bytes).  Any less conservative
approach opens the possibility for a false veto in eg. setuid
applications.  Alternative suggestions would be welcomed.

The cache is normally disabled, to activate set the sysctl variable
vfs.nfs.access_cache_timeout to a nonzero value.  This is the time in
seconds that a cached entry will be considered valid; useful values appear
to be 2-10 seconds.  Performance of the cache can be monitored with the
vfs.nfs.access_cache_hits and vfs.nfs.access_cache_hits variables.

25 years agoRestored the "reallocblks" code to its former glory. What this does is
dg [Fri, 13 Nov 1998 01:01:44 +0000 (01:01 +0000)]
Restored the "reallocblks" code to its former glory. What this does is
basically do a on-the-fly defragmentation of the FFS filesystem, changing
file block allocations to make them contiguous. Thanks to Kirk McKusick
for providing hints on what needed to be done to get this working.

25 years agoEnable libradius and libtacplus.
jdp [Fri, 13 Nov 1998 01:01:37 +0000 (01:01 +0000)]
Enable libradius and libtacplus.

25 years agoAdd libradius and libtacplus to the includes target. Note their
jdp [Fri, 13 Nov 1998 00:59:00 +0000 (00:59 +0000)]
Add libradius and libtacplus to the includes target.  Note their
dependence on libmd.

25 years agoThis commit was generated by cvs2svn to compensate for changes in r41120,
jdp [Fri, 13 Nov 1998 00:54:26 +0000 (00:54 +0000)]
This commit was generated by cvs2svn to compensate for changes in r41120,
which included commits to RCS files with non-trunk default branches.

25 years agoInitial import of TACACS+ client library donated by Juniper Networks, Inc.
jdp [Fri, 13 Nov 1998 00:54:26 +0000 (00:54 +0000)]
Initial import of TACACS+ client library donated by Juniper Networks, Inc.

25 years agoThis commit was generated by cvs2svn to compensate for changes in r41118,
jdp [Fri, 13 Nov 1998 00:53:01 +0000 (00:53 +0000)]
This commit was generated by cvs2svn to compensate for changes in r41118,
which included commits to RCS files with non-trunk default branches.

25 years agoInitial import of RADIUS client library donated by Juniper Networks, Inc.
jdp [Fri, 13 Nov 1998 00:53:01 +0000 (00:53 +0000)]
Initial import of RADIUS client library donated by Juniper Networks, Inc.

25 years agoArrg, ppi*) corrected
nsouch [Thu, 12 Nov 1998 22:48:16 +0000 (22:48 +0000)]
Arrg, ppi*) corrected

25 years agoAdd ppi*) iic*) and smb*)
nsouch [Thu, 12 Nov 1998 22:45:24 +0000 (22:45 +0000)]
Add ppi*) iic*) and smb*)

25 years agoSimplify call to dokern.sh - now the no MFS case is the default.
jkh [Thu, 12 Nov 1998 20:40:45 +0000 (20:40 +0000)]
Simplify call to dokern.sh - now the no MFS case is the default.

25 years agoBoy, two typos in one day. Time to go back to bed.
jkh [Thu, 12 Nov 1998 20:37:20 +0000 (20:37 +0000)]
Boy, two typos in one day.  Time to go back to bed.

25 years agoElimintate gratuitous backslash
jkh [Thu, 12 Nov 1998 20:12:52 +0000 (20:12 +0000)]
Elimintate gratuitous backslash

25 years agoFix a typo in a comment.
ken [Thu, 12 Nov 1998 17:47:24 +0000 (17:47 +0000)]
Fix a typo in a comment.

25 years agoRemove `amd', `nca' SCSI devices to match Mike's LINT commit.
obrien [Thu, 12 Nov 1998 11:29:28 +0000 (11:29 +0000)]
Remove `amd', `nca' SCSI devices to match Mike's LINT commit.

25 years ago(fixing bum commit log)
obrien [Thu, 12 Nov 1998 11:29:17 +0000 (11:29 +0000)]
(fixing bum commit log)
Remove `amd', `nca', and `sea' SCSI devices to match Mikes LINT commit.
(and reality).

25 years agoRemove
obrien [Thu, 12 Nov 1998 11:27:54 +0000 (11:27 +0000)]
Remove

25 years agoFill-in the manpages a little.
obrien [Thu, 12 Nov 1998 11:10:26 +0000 (11:10 +0000)]
Fill-in the manpages a little.

25 years agobootforth seems to be working well enough for me and others to turn
jkh [Thu, 12 Nov 1998 07:40:51 +0000 (07:40 +0000)]
bootforth seems to be working well enough for me and others to turn
on by default.  If it causes trouble, it's very easy to switch off again.

25 years agoBuild libmd shared for a.out too. Required for some PAM modules.
jdp [Thu, 12 Nov 1998 04:08:10 +0000 (04:08 +0000)]
Build libmd shared for a.out too.  Required for some PAM modules.

25 years agoMF22: Bring in some linux sound ioctl support which I committed to 2.2
jkh [Thu, 12 Nov 1998 00:42:08 +0000 (00:42 +0000)]
MF22: Bring in some linux sound ioctl support which I committed to 2.2
for PR 7792 but did not bring forward.

Submitted by: Avatar Liang <avatar@www.mmlab.cse.yzu.edu.tw>
PR: 8656

25 years agoInstall scripts without .sh suffix.
jkh [Thu, 12 Nov 1998 00:37:39 +0000 (00:37 +0000)]
Install scripts without .sh suffix.

25 years agoModify pciconf(8) so that it will print out PCI devices that have no driver
ken [Thu, 12 Nov 1998 00:22:30 +0000 (00:22 +0000)]
Modify pciconf(8) so that it will print out PCI devices that have no driver
configured in the kernel.  It gives them a device name of "none" and
monotonically incrementing unit numbers.  (starting at 0)  Before, pciconf
would just skip over unconfigured devices.  (unconfigured devices can be
detected because they have a null string for a device name)

Update the man page to reflect the new pciconf output.  Unfortunately, this
causes the sample 'pciconf -l' output lines to wrap, but I'm not sure what
to do about that really.

If anyone presents a reasonable case for printing out something other than
"none1" for unconfigured devices, I'm willing to listen.

25 years agoAdd MATH_EMULATE back to kern.flp. Now at least the FPU-less have *some*
jkh [Wed, 11 Nov 1998 23:08:24 +0000 (23:08 +0000)]
Add MATH_EMULATE back to kern.flp.  Now at least the FPU-less have *some*
way to install 3.0.x.

25 years agoConsign the 'amd' driver to the Attic.
msmith [Wed, 11 Nov 1998 21:32:56 +0000 (21:32 +0000)]
Consign the 'amd' driver to the Attic.

25 years agoRemove the 'amd' driver.
msmith [Wed, 11 Nov 1998 21:30:59 +0000 (21:30 +0000)]
Remove the 'amd' driver.

25 years agoRemove all mention of the 'amd' driver. It can come back if we grow
msmith [Wed, 11 Nov 1998 21:29:09 +0000 (21:29 +0000)]
Remove all mention of the 'amd' driver.  It can come back if we grow
support for it again.

25 years agoBe sure to pullup entire IP header when dealing with fragment packets.
dg [Wed, 11 Nov 1998 21:17:59 +0000 (21:17 +0000)]
Be sure to pullup entire IP header when dealing with fragment packets.

25 years agocoda_lookup now passes up an extra flag. But old veni will
rvb [Wed, 11 Nov 1998 20:32:20 +0000 (20:32 +0000)]
coda_lookup now passes up an extra flag.  But old veni will
be ok; new veni will check /dev/cfs0 to make sure that a new
kernel is running.
Also, a bug in vc_nb_close iff CODA_SIGNAL's were seen has been
fixed.

25 years agoClosed a small race condition between wiring/unwiring pages that involved
dg [Wed, 11 Nov 1998 15:07:57 +0000 (15:07 +0000)]
Closed a small race condition between wiring/unwiring pages that involved
the page's wire_count.

25 years agoArgh, I don't know how I managed this, but it would help to actually
peter [Wed, 11 Nov 1998 14:28:08 +0000 (14:28 +0000)]
Argh, I don't know how I managed this, but it would help to actually
provide the routines that kern/inflate.c expects.

25 years agoregenerate (+kldsym)
peter [Wed, 11 Nov 1998 13:05:29 +0000 (13:05 +0000)]
regenerate (+kldsym)

25 years agokldsym(2) prototype implementation
peter [Wed, 11 Nov 1998 13:04:40 +0000 (13:04 +0000)]
kldsym(2) prototype implementation

25 years agoregenerate (+kldsym)
peter [Wed, 11 Nov 1998 12:57:05 +0000 (12:57 +0000)]
regenerate (+kldsym)

25 years agoA kldsym(2) syscall prototype for extracting information from the in-kernel
peter [Wed, 11 Nov 1998 12:45:14 +0000 (12:45 +0000)]
A kldsym(2) syscall prototype for extracting information from the in-kernel
linker.  This is intended to replace kvm_mkdb etc.  The first version
only does name->value lookups, but it's open ended.  value->name lookups
would probably be a good thing to do too.

It's been suggested to try and connect the symbol tables to sysctl (which
is probably a more flexible way of doing it if it's done right), but that
is far more complex and difficult than I was ready to have a shot at.

25 years agoI got another batch of suggestions for cosmetic changes from bde.
truckman [Wed, 11 Nov 1998 10:56:07 +0000 (10:56 +0000)]
I got another batch of suggestions for cosmetic changes from bde.

25 years agoInstalled the second patch attached to kern/7899 with some changes suggested
truckman [Wed, 11 Nov 1998 10:04:13 +0000 (10:04 +0000)]
Installed the second patch attached to kern/7899 with some changes suggested
by bde, a few other tweaks to get the patch to apply cleanly again and
some improvements to the comments.

This change closes some fairly minor security holes associated with
F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN
had on tty devices.  For more details, see the description on the PR.

Because this patch increases the size of the proc and pgrp structures,
it is necessary to re-install the includes and recompile libkvm,
the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w.

PR: kern/7899
Reviewed by: bde, elvind

25 years agoRevise error messages.
rnordier [Wed, 11 Nov 1998 08:56:17 +0000 (08:56 +0000)]
Revise error messages.