]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoClean up some leftovers from the root mount cleanup that was done some
peter [Sun, 4 Feb 2001 15:35:10 +0000 (15:35 +0000)]
Clean up some leftovers from the root mount cleanup that was done some
time ago.  FFS_ROOT and CD9660_ROOT are obsolete.

23 years agofix duplicate rcsid
darrenr [Sun, 4 Feb 2001 15:25:15 +0000 (15:25 +0000)]
fix duplicate rcsid

23 years agoThis seems to stop syslog console messages looping. The problem
dwmalone [Sun, 4 Feb 2001 15:16:08 +0000 (15:16 +0000)]
This seems to stop syslog console messages looping. The problem
is that LOG_FAC doesn't do quite what you think it does.

PR: 24704

23 years agofix conflicts
darrenr [Sun, 4 Feb 2001 14:26:56 +0000 (14:26 +0000)]
fix conflicts

23 years agoThis commit was generated by cvs2svn to compensate for changes in r72003,
darrenr [Sun, 4 Feb 2001 14:25:38 +0000 (14:25 +0000)]
This commit was generated by cvs2svn to compensate for changes in r72003,
which included commits to RCS files with non-trunk default branches.

23 years agoImport IP Filter update
darrenr [Sun, 4 Feb 2001 14:25:38 +0000 (14:25 +0000)]
Import IP Filter update

23 years agoRemove the need to list each and every cpu platform. Config will now
peter [Sun, 4 Feb 2001 13:17:38 +0000 (13:17 +0000)]
Remove the need to list each and every cpu platform.  Config will now
take your word for the 'machine' switch.

23 years agoMechanical change to use <sys/queue.h> macro API instead of
phk [Sun, 4 Feb 2001 13:13:25 +0000 (13:13 +0000)]
Mechanical change to use <sys/queue.h> macro API instead of
fondling implementation details.

Created with: sed(1)
Reviewed by: md5(1)

23 years agoUse <sys/queue.h> macro API.
phk [Sun, 4 Feb 2001 12:37:48 +0000 (12:37 +0000)]
Use <sys/queue.h> macro API.

23 years ago'device agp' was missing
peter [Sun, 4 Feb 2001 12:36:50 +0000 (12:36 +0000)]
'device agp' was missing

23 years agoRemove some stray traces of KMODDEPS
peter [Sun, 4 Feb 2001 12:35:22 +0000 (12:35 +0000)]
Remove some stray traces of KMODDEPS

23 years agoDocument API CS20
wilko [Sun, 4 Feb 2001 12:11:09 +0000 (12:11 +0000)]
Document API CS20

23 years agoRemove a DIAGNOSTIC check which belongs in <sys/queue.h> if anyplace at all.
phk [Sun, 4 Feb 2001 11:53:51 +0000 (11:53 +0000)]
Remove a DIAGNOSTIC check which belongs in <sys/queue.h> if anyplace at all.

23 years agoWarn if people include vn(4) in their config.
phk [Sun, 4 Feb 2001 11:47:16 +0000 (11:47 +0000)]
Warn if people include vn(4) in their config.

23 years agoPull the rug from under the 'LKM Compatability' macro - PSEUDO_SET().
peter [Sun, 4 Feb 2001 11:46:17 +0000 (11:46 +0000)]
Pull the rug from under the 'LKM Compatability' macro - PSEUDO_SET().
There are two 3rd party code chunks using this still - the IPv6 stuff and
i4b.  Give them a private copy as an alternative to changing them too much.

XXX sys/kernel.h still has a #include <sys/module.h> in it.  I will be
taking this out shortly - this affects a number of drivers.

23 years agoRemove the LABPC driver.
phk [Sun, 4 Feb 2001 11:18:15 +0000 (11:18 +0000)]
Remove the LABPC driver.

Doesn't work, no maintainer, more promising code exists elsewhere.

23 years agoZap obsolete (died with LKM) EXPORT_SYMS variable
peter [Sun, 4 Feb 2001 10:52:25 +0000 (10:52 +0000)]
Zap obsolete (died with LKM) EXPORT_SYMS variable

23 years agoUse macro API to <sys/queue.h>
phk [Sun, 4 Feb 2001 10:34:45 +0000 (10:34 +0000)]
Use macro API to <sys/queue.h>

23 years agoRemove a sizeof(void *) == sizeof(int) assumption.
phk [Sun, 4 Feb 2001 10:06:24 +0000 (10:06 +0000)]
Remove a sizeof(void *) == sizeof(int) assumption.

Submitted by: Loren James Rittle <rittle@latour.rsch.comm.mot.com>

23 years agoZap some bad examples:
peter [Sun, 4 Feb 2001 08:23:14 +0000 (08:23 +0000)]
Zap some bad examples:
opt_foo.h:
touch opt_foo.h
.. is unnecessary - kmod.mk does this for us.

23 years agoAll the world is not an i386. Merge rev 1.438 of i386/i386/machdep.c.
peter [Sun, 4 Feb 2001 07:00:47 +0000 (07:00 +0000)]
All the world is not an i386.  Merge rev 1.438 of i386/i386/machdep.c.
Make buffer_map a system map.

23 years agoThis commit represents work mainly submitted by Tor and slightly modified
dillon [Sun, 4 Feb 2001 06:19:28 +0000 (06:19 +0000)]
This commit represents work mainly submitted by Tor and slightly modified
by myself.  It solves a serious vm_map corruption problem that can occur
with the buffer cache when block sizes > 64K are used.  This code has been
heavily tested in -stable but only tested somewhat on -current.  An MFC
will occur in a few days.  My additions include the vm_map_simplify_entry()
and minor buffer cache boundry case fix.

Make the buffer cache use a system map for buffer cache KVM rather then a
normal map.

Ensure that VM objects are not allocated for system maps.  There were cases
where a buffer map could wind up with a backing VM object -- normally
harmless, but this could also result in the buffer cache blocking in places
where it assumes no blocking will occur, possibly resulting in corrupted
maps.

Fix a minor boundry case in the buffer cache size limit is reached that
could result in non-optimal code.

Add vm_map_simplify_entry() calls to prevent 'creeping proliferation'
of vm_map_entry's in the buffer cache's vm_map.  Previously only a simple
linear optimization was made.  (The buffer vm_map typically has only a
handful of vm_map_entry's.  This stabilizes it at that level permanently).

PR: 20609
Submitted by: (Tor Egge) tegge

23 years agoGrumble, I broke this file with a vi accident before commit. :-(
peter [Sun, 4 Feb 2001 04:13:12 +0000 (04:13 +0000)]
Grumble, I broke this file with a vi accident before commit. :-(

Submitted by: Christian Weisgerber <naddy@mips.inka.de>

23 years agoSpelling.
mckay [Sun, 4 Feb 2001 02:05:16 +0000 (02:05 +0000)]
Spelling.

23 years agoIn the hope of saving others from hours of tedious recovery work,
mckay [Sun, 4 Feb 2001 02:02:38 +0000 (02:02 +0000)]
In the hope of saving others from hours of tedious recovery work,
document that cp still isn't very useful for recursive copies even
with the -R flag.  This is because hard links are broken by cp.

23 years agoExtend the sanity checks in ufs_lookup to ensure that each directory
iedowse [Sun, 4 Feb 2001 01:52:11 +0000 (01:52 +0000)]
Extend the sanity checks in ufs_lookup to ensure that each directory
entry fits within its DIRBLKSIZ block. The surrounding code is
extremely fragile with respect to corruption of the directory entry
'd_reclen' field; if directory corruption occurs, it can blindly
scan forward beyond the end of the filesystem block. Usually this
results in a 'fault on nofault entry' panic.

Directory corruption is now much more likely to be detected, resulting
in a 'ufs_dirbad' panic. If the filesystem is read-only, it will
simply print a warning message, and skip the corrupted block.

Reviewed by: mckusick

23 years agoMPPE_MasterKeyValid is only there if HAVE_DES is defined.
brian [Sun, 4 Feb 2001 01:20:33 +0000 (01:20 +0000)]
MPPE_MasterKeyValid is only there if HAVE_DES is defined.

23 years agoDon't send an authentication failure response if we fail sending
brian [Sun, 4 Feb 2001 01:08:26 +0000 (01:08 +0000)]
Don't send an authentication failure response if we fail sending
a radius accounting packet.

23 years agoMake the MPPE MasterKey Invalid messages a bit clearer (it now
brian [Sun, 4 Feb 2001 01:08:24 +0000 (01:08 +0000)]
Make the MPPE MasterKey Invalid messages a bit clearer (it now
complains that you can't do MPPE without CHAP81).

Reset MasterKeyValid to zero when we hit phase DEAD.

23 years agoUntangle some cunfusion between the CLOSE_STAYDOWN, CLOSE_LCP and
brian [Sun, 4 Feb 2001 01:08:22 +0000 (01:08 +0000)]
Untangle some cunfusion between the CLOSE_STAYDOWN, CLOSE_LCP and
CLOSE_NORMAL meanings.  CLOSE_NORMAL doesn't change the currently
required state, the others do.  This should stop ppp from entering
DATALINK_READY when LCP shutdown doesn't end up happening cleanly.

Bump our version number to reflect this change.

23 years agoDon't specify root:wheel for tun*, it's the default.
brian [Sat, 3 Feb 2001 23:31:31 +0000 (23:31 +0000)]
Don't specify root:wheel for tun*, it's the default.

Pointed out by: bde

23 years agoUse the correct flags field when checking for a read-only filesystem
iedowse [Sat, 3 Feb 2001 21:25:32 +0000 (21:25 +0000)]
Use the correct flags field when checking for a read-only filesystem
in ufs_dirbad(). The mnt_stat.f_flags field is only updated by the
syscalls *statfs and getfsstat, so mnt_flag should be used instead.

This only affects whether or not a panic is generated on detection of
certain types of directory corruption.

Reviewed by: mckusick

23 years agoAdd a -g flag for verifying an installed package against its recorded
jkh [Sat, 3 Feb 2001 20:56:32 +0000 (20:56 +0000)]
Add a -g flag for verifying an installed package against its recorded
checksums (to see if it's been modified post-installation).  Naturally,
this mechanism is only as secure as the contents of /var/db/pkg if you're
using it for auditing purposes.

Submitted by: Roman Shterenzon <roman@xpert.com>

23 years agoDocument API CS20 support.
wilko [Sat, 3 Feb 2001 18:38:24 +0000 (18:38 +0000)]
Document API CS20 support.

HARDWARE.TXT update to follow.

23 years agoMake the code act the same in the case of BRIDGE being defined, but not
julian [Sat, 3 Feb 2001 17:25:21 +0000 (17:25 +0000)]
Make the code act the same in the case of BRIDGE being defined, but not
turned on, and the case of it not being defined at all.
i.e. Disabling bridging re-enables some of the checks it disables.

Submitted by: "Rogier R. Mulhuijzen" <drwilco@drwilco.net>

23 years agoUse LIST_FOREACH() to traverse ifp->if_multiaddrs list, instead of
phk [Sat, 3 Feb 2001 16:29:10 +0000 (16:29 +0000)]
Use LIST_FOREACH() to traverse ifp->if_multiaddrs list, instead of
<sys/queue.h> implementation details.

Created with:   /usr/sbin/sed
Reviewed with:  /sbin/md5

23 years agoUse <sys/queue.h> macro api rather than fondle its implementation detals.
phk [Sat, 3 Feb 2001 11:46:35 +0000 (11:46 +0000)]
Use <sys/queue.h> macro api rather than fondle its implementation detals.

Created with: /usr/bin/sed
Reviewed by: /sbin/md5

23 years agoMove wicontrol from i386 to MI section. (yes, it compiles on alpha, where
peter [Sat, 3 Feb 2001 08:56:52 +0000 (08:56 +0000)]
Move wicontrol from i386 to MI section. (yes, it compiles on alpha, where
theoretically NEWCARD might work one day).  if_wi.c is in the MI config
section already.

23 years agoConditionalize the alpha interrupt preemption for now to buy us some
peter [Sat, 3 Feb 2001 03:26:39 +0000 (03:26 +0000)]
Conditionalize the alpha interrupt preemption for now to buy us some
time to sort out the quirks.  Add 'options PREEMPTION' to test it on
the Alpha.

Reviewed by: jhb

23 years agoo Reinstate Kerberos IV support for sshd when MAKE_KERBEROS4 is
rwatson [Sat, 3 Feb 2001 02:09:20 +0000 (02:09 +0000)]
o Reinstate Kerberos IV support for sshd when MAKE_KERBEROS4 is
  compiled in.  This involves a commented out sshd line to match the
  remainder of the commented out pam_kerberosIV.so entries.  This
  doesn't quite restore the correct behavior, as ticket files are
  not managed properly, but it's an improvement.

Forgotten by: green

23 years agoPick up all messages* files less than two days old rather than
brian [Sat, 3 Feb 2001 01:28:46 +0000 (01:28 +0000)]
Pick up all messages* files less than two days old rather than
just messages{,.0*} when looking for login failures and refused
connections.

PR: 23415
Mostly submitted by: phk

Convert a few "  "s to tabs while I'm here - for consistency.

23 years agoChange the permissions on /dev/tun* to 0600 root:wheel
brian [Sat, 3 Feb 2001 00:32:17 +0000 (00:32 +0000)]
Change the permissions on /dev/tun* to 0600 root:wheel

23 years agoo Allow non-root users to open /dev/tun* (remove suser()
brian [Sat, 3 Feb 2001 00:31:39 +0000 (00:31 +0000)]
o Allow non-root users to open /dev/tun* (remove suser()
  in tunopen())
o Change the default device permissions to 0600 root:wheel
  (were uucp:dialer)
o Only let root (suser()) change the MTU

This makes it possible for an administrator to open up the
permissions on /dev/tun*, letting non-root programs service
a tun interface.  Co-operation is still required with a
priviledged program that will configure the interface side
of things.

23 years agoAt the point in time where most devices are created, we don't know what
phk [Fri, 2 Feb 2001 22:54:41 +0000 (22:54 +0000)]
At the point in time where most devices are created, we don't know what
time it is because boottime is not yet initialized.  Finagle the relevant
fields when we get the chance.

23 years agoUnfortunately low-altitude overflights is not enough to make penguins crash.
phk [Fri, 2 Feb 2001 22:35:13 +0000 (22:35 +0000)]
Unfortunately low-altitude overflights is not enough to make penguins crash.

23 years agoWhen turning off TCP_NOPUSH, call tcp_output to immediately flush
jlemon [Fri, 2 Feb 2001 18:48:25 +0000 (18:48 +0000)]
When turning off TCP_NOPUSH, call tcp_output to immediately flush
out any data pending in the buffer.

Submitted by: Tony Finch <dot@dotat.at>

23 years agoOnly superuser can create symlinks.
phk [Fri, 2 Feb 2001 18:35:29 +0000 (18:35 +0000)]
Only superuser can create symlinks.
Give symlinks mode 755 by default to avoid triggering alert eyes.
(the mode isn't use on symlinks)

23 years agoBIND update now has security advisory number.
bmah [Fri, 2 Feb 2001 17:02:05 +0000 (17:02 +0000)]
BIND update now has security advisory number.

23 years agoquieten the esscontrol device
cg [Fri, 2 Feb 2001 16:41:06 +0000 (16:41 +0000)]
quieten the esscontrol device

23 years agostrcpy -> strlcpy paranoia
nectar [Fri, 2 Feb 2001 13:22:43 +0000 (13:22 +0000)]
strcpy -> strlcpy paranoia

Submitted by: Mike Heffner <mheffner@vt.edu>
Reviewed by: freebsd-audit

23 years agoRemoved vestiges of vop_mmap.
bde [Fri, 2 Feb 2001 12:51:22 +0000 (12:51 +0000)]
Removed vestiges of vop_mmap.

23 years agoFix tftpd and tftp to support file transfers of over 65535 blocks
asmodai [Fri, 2 Feb 2001 10:53:02 +0000 (10:53 +0000)]
Fix tftpd and tftp to support file transfers of over 65535 blocks
(about 31 MB - 32 MB).

Submitted (partially)
by: Pascal Hofstee <daeron@wit401305.student.utwente.nl>

23 years ago* Move exception_return to exception.s which is a more logical home for it.
dfr [Fri, 2 Feb 2001 10:20:30 +0000 (10:20 +0000)]
* Move exception_return to exception.s which is a more logical home for it.
* Optimise the return path for syscalls so that they only restore a minimal
  set of registers instead of performing a full exception_return.

A new flag in the trapframe indicates that the frame only holds partial
state. When it is necessary to perform a full state restore (e.g. after an
execve or signal), the flag is cleared to force a full restore.

23 years agoPatch the reintroduced (~0 << 32) == 1 bug with a cryptic but functional
msmith [Fri, 2 Feb 2001 08:31:50 +0000 (08:31 +0000)]
Patch the reintroduced (~0 << 32) == 1 bug with a cryptic but functional
version.  This should fix the read-before-write problem again.

Obtained from: "Adam J. Richter" <adam@yggdrasil.com>

23 years agoAllocate lock table and mutex not only for parity plexes, but also for
grog [Fri, 2 Feb 2001 07:14:13 +0000 (07:14 +0000)]
Allocate lock table and mutex not only for parity plexes, but also for
striped plexes.  This prevents various panics introduced in the last
rewrite of the locking code.

Suffered by:   "Niels Chr. Bank-Pedersen" <ncbp@bank-pedersen.dk>

23 years agoPass the minor number rather than the unit number to make_dev()
brian [Fri, 2 Feb 2001 03:32:25 +0000 (03:32 +0000)]
Pass the minor number rather than the unit number to make_dev()
from the clone handler.

23 years agoKASSERT that the minor number passed to make_dev() is valid.
brian [Fri, 2 Feb 2001 03:32:11 +0000 (03:32 +0000)]
KASSERT that the minor number passed to make_dev() is valid.

23 years agoFix typo.
nik [Fri, 2 Feb 2001 03:32:03 +0000 (03:32 +0000)]
Fix typo.

PR:             docs/23936
Submitted by:   Garret Rooney <rooneg@rpi.edu>

23 years agoStructure the options listing to be more standard.
nik [Fri, 2 Feb 2001 03:29:09 +0000 (03:29 +0000)]
Structure the options listing to be more standard.

The PR also included documentation for other options, but upon
inspection of the source these options aren't used.

PR:             docs/24149
Submitted by:   Jesse Monroy, Jr. <opentrax@email.com>

23 years agoDeclaring functions inside functions was deprecated twice. Keep the
nik [Fri, 2 Feb 2001 03:11:22 +0000 (03:11 +0000)]
Declaring functions inside functions was deprecated twice.  Keep the
second recommendation, which includes more rationale, and nix the first.

PR:             docs/24690
Submitted by:   Alex Kapranoff <alex@kapran.bitmcnit.bryansk.su

23 years agoAdd an xref for mount_ntfs.
nik [Fri, 2 Feb 2001 03:08:48 +0000 (03:08 +0000)]
Add an xref for mount_ntfs.

PR:             docs/24693
Submitted by:   Etienne Vidal <hallik@libertysurf.fr>

23 years agoEnable use of DHCP extensions.
tegge [Fri, 2 Feb 2001 02:35:40 +0000 (02:35 +0000)]
Enable use of DHCP extensions.
Reviewed by: Per Kristian Hove <Per.Hove@math.ntnu.no>

23 years agoArgh, I missed some #include "sio.h". I was looking primarily for NSIO
peter [Fri, 2 Feb 2001 01:48:40 +0000 (01:48 +0000)]
Argh, I missed some #include "sio.h".  I was looking primarily for NSIO
when I did my sweeps.

Submitted by: mjacob

23 years agoTry to kldload if_tun if we get ENOENT from opening /dev/tunN -
brian [Fri, 2 Feb 2001 01:41:26 +0000 (01:41 +0000)]
Try to kldload if_tun if we get ENOENT from opening /dev/tunN -
not just if we get ENXIO.

This makes ppp work with DEVFS when if_tun isn't built into
the kernel (without needing to manually kldload it).

23 years agomtx_try_enter() returns an int, not void.
jhb [Fri, 2 Feb 2001 00:49:17 +0000 (00:49 +0000)]
mtx_try_enter() returns an int, not void.

23 years agoMFS: bridge/ipfw/dummynet fixes
luigi [Fri, 2 Feb 2001 00:19:25 +0000 (00:19 +0000)]
MFS: bridge/ipfw/dummynet fixes

23 years agoMFS: bridge/ipfw/dummynet fixes (bridge.c will be committed separately)
luigi [Fri, 2 Feb 2001 00:18:00 +0000 (00:18 +0000)]
MFS: bridge/ipfw/dummynet fixes (bridge.c will be committed separately)

23 years agoRemove inclusion of now vanished sio.h.
mjacob [Thu, 1 Feb 2001 21:59:00 +0000 (21:59 +0000)]
Remove inclusion of now vanished sio.h.

23 years agoGuess that this is what Doug *intended* to commit....
mjacob [Thu, 1 Feb 2001 21:58:34 +0000 (21:58 +0000)]
Guess that this is what Doug *intended* to commit....

23 years agoChange the kernel internal ABI number as the HOOK structure has changed.
julian [Thu, 1 Feb 2001 21:25:06 +0000 (21:25 +0000)]
Change the kernel internal ABI number as the HOOK structure has changed.
Forgotten by: me

23 years agoRevert previous checkin ("proxy" is required in some instances).
archie [Thu, 1 Feb 2001 21:23:16 +0000 (21:23 +0000)]
Revert previous checkin ("proxy" is required in some instances).
Add a note to that effect to the man page.

23 years agoAdd the MD_ROOT_SIZE option.
luigi [Thu, 1 Feb 2001 21:05:52 +0000 (21:05 +0000)]
Add the MD_ROOT_SIZE option.
These files might need further work before producing a compilable kernel.

23 years agoAdd a required option, MD_ROOT_SIZE
luigi [Thu, 1 Feb 2001 21:04:32 +0000 (21:04 +0000)]
Add a required option, MD_ROOT_SIZE

also put in a better set of devices and options.

23 years agoClean up reference counting with relation to queued packets and the worklist,
julian [Thu, 1 Feb 2001 20:51:23 +0000 (20:51 +0000)]
Clean up reference counting with relation to queued packets and the worklist,
and while I'm there, clean up the worklist insertion and removal.

Inspired by: Harti Brandt <brandt@fokus.gmd.de>

23 years agoDriver for the ESS Maestro3 and Allegro sound chips. Note that due to the
scottl [Thu, 1 Feb 2001 20:29:16 +0000 (20:29 +0000)]
Driver for the ESS Maestro3 and Allegro sound chips.  Note that due to the
amount of GPL'd firmware in the driver, it will only be built as a module.

Approved by: cg

23 years agomdoc(7) police: split punctuation characters + misc fixes.
ru [Thu, 1 Feb 2001 17:12:45 +0000 (17:12 +0000)]
mdoc(7) police: split punctuation characters + misc fixes.

23 years agomdoc(7) police: split punctuation characters + misc fixes.
ru [Thu, 1 Feb 2001 16:44:04 +0000 (16:44 +0000)]
mdoc(7) police: split punctuation characters + misc fixes.

23 years agoAdd atspeaker_load.
des [Thu, 1 Feb 2001 16:41:58 +0000 (16:41 +0000)]
Add atspeaker_load.

23 years agoKLDify the "speaker" device (which calls itself atspeaker internally, and
des [Thu, 1 Feb 2001 16:39:34 +0000 (16:39 +0000)]
KLDify the "speaker" device (which calls itself atspeaker internally, and
is i386-specific, so name the module atspeaker rather than speaker).

23 years agomdoc(7) police: split punctuation characters + misc fixes.
ru [Thu, 1 Feb 2001 16:38:02 +0000 (16:38 +0000)]
mdoc(7) police: split punctuation characters + misc fixes.

23 years agoMake the exclamation mark (`!') a recognizable punctuation character.
ru [Thu, 1 Feb 2001 16:17:06 +0000 (16:17 +0000)]
Make the exclamation mark (`!') a recognizable punctuation character.

Obtained from: mdocNG

23 years agoUnbreak test coverage of cy driver.
bde [Thu, 1 Feb 2001 09:57:59 +0000 (09:57 +0000)]
Unbreak test coverage of cy driver.

23 years agoConverted to new-bus.
bde [Thu, 1 Feb 2001 09:49:57 +0000 (09:49 +0000)]
Converted to new-bus.

Reviewed by: imp

23 years agoFix breakage caused by incomplete transition to IF_HANDOFF().
bp [Thu, 1 Feb 2001 08:34:38 +0000 (08:34 +0000)]
Fix breakage caused by incomplete transition to IF_HANDOFF().
Remove unused variable.

23 years agoImplement preemptive scheduling of hardware interrupt threads.
jake [Thu, 1 Feb 2001 03:34:20 +0000 (03:34 +0000)]
Implement preemptive scheduling of hardware interrupt threads.

- If possible, context switch to the thread directly in sched_ithd(),
  rather than triggering a delayed ast reschedule.

- Disable interrupts while restoring fpu state in the trap handler,
  in order to ensure that we are not preempted in the middle, which
  could cause migration to another cpu.

Reviewed by: peter
Tested by: peter (alpha)

23 years agoUse mdconfig/md instead of vn.
phk [Wed, 31 Jan 2001 22:58:39 +0000 (22:58 +0000)]
Use mdconfig/md instead of vn.

23 years agoUpdate the module Makefile to reflect the state of things from the last
scottl [Wed, 31 Jan 2001 22:40:01 +0000 (22:40 +0000)]
Update the module Makefile to reflect the state of things from the last
acpi-ca import.

Approved by: msmith

23 years agoAdd the ability to declare ore-ride methods on a per-hook basis
julian [Wed, 31 Jan 2001 20:46:00 +0000 (20:46 +0000)]
Add the ability to declare ore-ride methods on a per-hook basis
for the rcvdata() and rcvmsg() methods.

Also bring the man page up to sync with my last commit. (and this one)

23 years agoFsck_ffs did not properly range-check the inode 'di_size'
iedowse [Wed, 31 Jan 2001 15:16:56 +0000 (15:16 +0000)]
Fsck_ffs did not properly range-check the inode 'di_size'
field, so it was possible for a filesystem marked clean by fsck_ffs
to cause kernel crashes later when mounted. This could occur when
fsck_ffs was used to repair a badly corrupted filesystem.

As pointed out by bde, it is not sufficient to restrict di_size to
just the superblock fs_maxfilesize limit. The use of 32-bit logical
block numbers (both in fsck and the kernel) induces another file
size limit which is usually lower than fs_maxfilesize. Also, the
old 4.3BSD filesystem does not have fs_maxfilesize initialised.

Following this change, fsck_ffs will enforce exactly the same
file size limits as are used by the kernel.

PR: kern/15065
Discussed with: bde
Reviewed by: bde, mckusick

23 years agoDocument the mpsafe argument to callout_init().
ben [Wed, 31 Jan 2001 12:17:45 +0000 (12:17 +0000)]
Document the mpsafe argument to callout_init().

PR: 24529
Submitted by: harti brandt <brandt@fokus.gmd.de>
Reviewed by: jlemon

23 years agoTry and make it clearer that the static units thing is a FYI only.
peter [Wed, 31 Jan 2001 11:18:49 +0000 (11:18 +0000)]
Try and make it clearer that the static units thing is a FYI only.

23 years ago* Move exception_return to exception.s which is a more logical home for it.
dfr [Wed, 31 Jan 2001 11:17:00 +0000 (11:17 +0000)]
* Move exception_return to exception.s which is a more logical home for it.
* Optimise the return path for syscalls so that they only restore a minimal
  set of registers instead of performing a full exception_return.

A new flag in the trapframe indicates that the frame only holds partial
state. When it is necessary to perform a full state restore (e.g. after an
execve or signal), the flag is cleared to force a full restore.

23 years agoRemove count for NSIO. The only places it was used it were incorrect.
peter [Wed, 31 Jan 2001 10:54:45 +0000 (10:54 +0000)]
Remove count for NSIO.  The only places it was used it were incorrect.
(alpha-gdbstub.c got sync'ed up a bit with the i386 version)

23 years agoUnwind a bit more cruft - we only have one type of device now.
peter [Wed, 31 Jan 2001 10:30:30 +0000 (10:30 +0000)]
Unwind a bit more cruft - we only have one type of device now.

23 years agoOops. An old version of a local change leaked in with the last commit.
peter [Wed, 31 Jan 2001 10:06:06 +0000 (10:06 +0000)]
Oops. An old version of a local change leaked in with the last commit.
Since it is here, clean it up a bit.

23 years agoMerge conflicts from the ACPI CA 20010125 import.
msmith [Wed, 31 Jan 2001 09:44:45 +0000 (09:44 +0000)]
Merge conflicts from the ACPI CA 20010125 import.

23 years agoAdd some debugging.
msmith [Wed, 31 Jan 2001 09:35:50 +0000 (09:35 +0000)]
Add some debugging.

Turn off semaphores.  Nobody else implements them, and there is lots of
AML out there which does totally absurd things with them, meaning that
if we try to do the right thing we are guaranteed to fail.

23 years agoAdd some debugging statements.
msmith [Wed, 31 Jan 2001 09:34:54 +0000 (09:34 +0000)]
Add some debugging statements.

23 years agoTidy up.
msmith [Wed, 31 Jan 2001 09:33:51 +0000 (09:33 +0000)]
Tidy up.

Don't print temperatures at attach time - they're usually wrong.

Use acpi_EvaluateInteger instead of doing things the hard way.

23 years agoAdd some debugging.
msmith [Wed, 31 Jan 2001 09:32:44 +0000 (09:32 +0000)]
Add some debugging.

Use acpi_EvaluateInteger where possible.

Use FuncName rather than &FuncName when passing function addresses.

Don't evaluate the _REG method when we attach to an address space -
AcpiInstallAddressSpaceHandler does it for us.