]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years ago- In kern_ndis.c, implement ndis_unsched(), the complement to ndis_sched(),
wpaul [Thu, 25 Mar 2004 08:23:08 +0000 (08:23 +0000)]
- In kern_ndis.c, implement ndis_unsched(), the complement to ndis_sched(),
  which pulls a job off a thread work queue (assuming it hasn't run yet).
  This is needed for KeRemoveQueueDpc().

- In subr_ntoskrnl.c, implement KeInsertQueueDpc() and KeRemoveQueueDpc(),
  to go with KeInitializeDpc() to round out the API. Also change the
  KeTimer implementation to use this API instead of the private
  timer callout scheduler. Functionality of the timer API remains
  unchanged, but we get a couple new Windows kernel API routines and
  more closely imitate the way thing works in Windows. (As of yet
  I haven't encountered any drivers that use KeInsertQueueDpc() or
  KeRemoveQueueDpc(), but it doesn't hurt to have them.)

20 years agoMFi386: correctly calculate the top-of-stack when a kthread is created
peter [Thu, 25 Mar 2004 03:39:38 +0000 (03:39 +0000)]
MFi386: correctly calculate the top-of-stack when a kthread is created
with a larger kernel stack.

20 years agoRun print_AMD_features() for both AuthenticAMD and GenuineIntel cpus.
peter [Thu, 25 Mar 2004 03:38:31 +0000 (03:38 +0000)]
Run print_AMD_features() for both AuthenticAMD and GenuineIntel cpus.
Report the %ecx bits in cpuid function 1.  This is a hack.
When reporting AMD Features, only mask off the common bits.  Otherwise
the SEP bit masks off SYSCALL etc in the report.

20 years agoAdd NTFS since many may want to dual-boot MS-Win64 w/FreeBSD.
obrien [Thu, 25 Mar 2004 03:19:28 +0000 (03:19 +0000)]
Add NTFS since many may want to dual-boot MS-Win64 w/FreeBSD.

20 years agoRestore part of rev 1.92; share/libdata/gcc should not have been deleted.
obrien [Thu, 25 Mar 2004 00:41:06 +0000 (00:41 +0000)]
Restore part of rev 1.92; share/libdata/gcc should not have been deleted.

20 years agoForce gdb6 into disc1 for amd64, sparc64, and ia64
obrien [Thu, 25 Mar 2004 00:06:03 +0000 (00:06 +0000)]
Force gdb6 into disc1 for amd64, sparc64, and ia64
(until they have /usr/src gdb support)

20 years agoLets give -O2 kernels a try on sparc64.
obrien [Thu, 25 Mar 2004 00:02:49 +0000 (00:02 +0000)]
Lets give -O2 kernels a try on sparc64.

Reviewed by: kensmith

20 years agoUse uiomove_fromphys() instead of pmap_qenter() and pmap_qremove() in
alc [Wed, 24 Mar 2004 23:35:04 +0000 (23:35 +0000)]
Use uiomove_fromphys() instead of pmap_qenter() and pmap_qremove() in
proc_rwmem().

20 years agoInstall the pf rc.d-script (missed Makefile update in original commit)
mlaier [Wed, 24 Mar 2004 21:54:44 +0000 (21:54 +0000)]
Install the pf rc.d-script (missed Makefile update in original commit)

Approved by: bms(mentor)

20 years agoRemove mention of doscmd.
des [Wed, 24 Mar 2004 20:19:57 +0000 (20:19 +0000)]
Remove mention of doscmd.

20 years agoBack out revision 1.295. share/libdata/gcc/ is valid and used.
obrien [Wed, 24 Mar 2004 19:37:53 +0000 (19:37 +0000)]
Back out revision 1.295.   share/libdata/gcc/ is valid and used.

20 years agoMove ct.4 to ctau.4. Step two: add ct.4 as ctau.4.
rik [Wed, 24 Mar 2004 19:28:50 +0000 (19:28 +0000)]
Move ct.4 to ctau.4. Step two: add ct.4 as ctau.4.
Do not bother about too small history for ct.4.

Approved by:    imp(mentor)

20 years agoMove ct.4 to ctau.4. Step one: remove ctau.4 link.
rik [Wed, 24 Mar 2004 19:23:44 +0000 (19:23 +0000)]
Move ct.4 to ctau.4. Step one: remove ctau.4 link.

Approved by: imp(mentor)

20 years agoNew release note:
hrs [Wed, 24 Mar 2004 18:03:55 +0000 (18:03 +0000)]
New release note:
isp(4) fix on sparc64,
ipcs(1) -u option,
rmdir(1) -v flag,
rc.d/diskless split,
rc.d/pf.

Update release note:
MFC of umass(4) timeout/ATAPI MMC support,
items related to the rc.d scripts moved into a separate section.

20 years ago- Update for the new interrupt code on i386 meaning that SMP kernels can
jhb [Wed, 24 Mar 2004 17:51:54 +0000 (17:51 +0000)]
- Update for the new interrupt code on i386 meaning that SMP kernels can
  now run on UP machines, options APIC_IO is now spelled device apic, and
  HT CPUs are now enabled by default.
- Add a 'I386 NOTES' section header to the i386-specific paragraphs.
  Someone with more mdoc fu can suggest a better name if needed.

Requested by: kris (1)

20 years agoProperly reprogram the hardware when IFF_ALLMULTI flag is set.
ru [Wed, 24 Mar 2004 17:43:45 +0000 (17:43 +0000)]
Properly reprogram the hardware when IFF_ALLMULTI flag is set.

20 years agosppp (4) to netgraph (4) node. As always: I'l connect it to the
rik [Wed, 24 Mar 2004 17:24:01 +0000 (17:24 +0000)]
sppp (4) to netgraph (4) node. As always: I'l connect it to the
system after extra check.

Approved by: imp (mentor)
Approved by: julian (in general)

20 years agoConform to local file sytle and prefer (a && (b & flag)).
imp [Wed, 24 Mar 2004 16:49:37 +0000 (16:49 +0000)]
Conform to local file sytle and prefer (a && (b & flag)).

20 years agoThe topology lock is not needed for g_error_provider() function.
pjd [Wed, 24 Mar 2004 15:12:17 +0000 (15:12 +0000)]
The topology lock is not needed for g_error_provider() function.

Discussed with: phk

20 years agoAdd locking so that arc4random(3) functions are all reentrant for
green [Wed, 24 Mar 2004 14:44:57 +0000 (14:44 +0000)]
Add locking so that arc4random(3) functions are all reentrant for
pthreads.

Submitted by: Christian S.J. Peron <maneo@bsdpro.com>

20 years agoAdded a note, that g_access() function cannot failed when dc[rwe] <= 0.
pjd [Wed, 24 Mar 2004 14:42:26 +0000 (14:42 +0000)]
Added a note, that g_access() function cannot failed when dc[rwe] <= 0.

20 years agoFunctions:
pjd [Wed, 24 Mar 2004 14:25:46 +0000 (14:25 +0000)]
Functions:

- g_new_geomf(),
- g_new_providerf(),
- g_new_consumer()

cannot failed (by returning NULL), so don't scare programmers.

20 years agoAdded 'nojail' keyword for rc.d/mixer script, while mixer(8) is not
pjd [Wed, 24 Mar 2004 12:49:34 +0000 (12:49 +0000)]
Added 'nojail' keyword for rc.d/mixer script, while mixer(8) is not
usable in jail by default (no /dev/mixer).

20 years agoBring man page back in shape. Many option descriptions lacked the
se [Wed, 24 Mar 2004 10:25:24 +0000 (10:25 +0000)]
Bring man page back in shape. Many option descriptions lacked the
option name and paramaters they apply to and there was a mix of
formatting methods and styles.

I'm not sure that the description of "nfs_retransmit_counter" is
correct, but it does now at least match what's in -STABLE.

20 years agoDon't duplicate a packet, just to throw it away when we discover the
julian [Wed, 24 Mar 2004 08:53:45 +0000 (08:53 +0000)]
Don't duplicate a packet, just to throw it away when we discover the
duplicat hook is not connected.

20 years agoAdded the `toolchain' top-level target, which builds enough of
ru [Wed, 24 Mar 2004 08:26:22 +0000 (08:26 +0000)]
Added the `toolchain' top-level target, which builds enough of
buildworld, up to and including libraries, except for actually
building the world.

Requested by: many

20 years agoPrevious commit =~ s,share/,,
des [Wed, 24 Mar 2004 08:23:52 +0000 (08:23 +0000)]
Previous commit =~ s,share/,,

20 years agoRemove mention of share/libdata/doscmd/{,fonts/} and share/libdata/gcc.
des [Wed, 24 Mar 2004 08:22:35 +0000 (08:22 +0000)]
Remove mention of share/libdata/doscmd/{,fonts/} and share/libdata/gcc.

20 years agoPrevious commit should have read: remove share/libdata/doscmd/{,fonts/}.
des [Wed, 24 Mar 2004 08:21:30 +0000 (08:21 +0000)]
Previous commit should have read: remove share/libdata/doscmd/{,fonts/}.
This time around, remove share/libdata/gcc/ which seems unused.

20 years agoRemove share/doscmd/{,fonts/}
des [Wed, 24 Mar 2004 08:19:46 +0000 (08:19 +0000)]
Remove share/doscmd/{,fonts/}

20 years agoremove stale documentation
luigi [Wed, 24 Mar 2004 08:01:15 +0000 (08:01 +0000)]
remove stale documentation

20 years agoremove floppy image type
luigi [Wed, 24 Mar 2004 07:55:08 +0000 (07:55 +0000)]
remove floppy image type

20 years agoremove this floppy image too.
luigi [Wed, 24 Mar 2004 07:54:00 +0000 (07:54 +0000)]
remove this floppy image too.

20 years agoremove this floppy image.
luigi [Wed, 24 Mar 2004 07:53:12 +0000 (07:53 +0000)]
remove this floppy image.

20 years agoRemove this picobsd floppy type.
luigi [Wed, 24 Mar 2004 07:51:40 +0000 (07:51 +0000)]
Remove this picobsd floppy type.
We only want to support one in -current, and let people take care
of their own images.

20 years agofix building of bridge floppies
luigi [Wed, 24 Mar 2004 07:30:54 +0000 (07:30 +0000)]
fix building of bridge floppies

20 years agoThe ndis_wlan_bssid_ex structure we retrieve in ndis_get_assoc() is
wpaul [Wed, 24 Mar 2004 05:35:03 +0000 (05:35 +0000)]
The ndis_wlan_bssid_ex structure we retrieve in ndis_get_assoc() is
variable length, so we should not be trying to copy it into a fixed
length buffer, especially one on the stack. malloc() a buffer of the
right size and return a pointer to that instead.

Fixes a crash I discovered when testing whe a Cisco AP in infrastructure
mode, which returns several information elements that make the
ndis_wlan_bssid_ex structure larger than expected.

20 years agoAdd -u option to ipcs(1) and document it in ipcs.1
bmilekic [Wed, 24 Mar 2004 04:01:43 +0000 (04:01 +0000)]
Add -u option to ipcs(1) and document it in ipcs.1

It allows you to list IPC facilities owned by username/uid.

Submitted by: Christian S.J. Peron <maneo@bsdpro.com>
PR: bin/63463

20 years agoFix a bug introduced in rev 1.33(mega API change).
simokawa [Wed, 24 Mar 2004 01:29:08 +0000 (01:29 +0000)]
Fix a bug introduced in rev 1.33(mega API change).
Because xfer->send.payload is a pointer to the buffer, '&' shouldn't be there.

Submitted by: John Weisgerber <weisgerberj@gsilumonics.com>
PR: misc/64623

20 years agoCorrect the boundary parameter to the bus_dma_tag_create() calls (it was
tmm [Tue, 23 Mar 2004 23:41:39 +0000 (23:41 +0000)]
Correct the boundary parameter to the bus_dma_tag_create() calls (it was
(1 << 24) - 2 instead of 1 << 24, which it was obviously intended to
be). This fixes SBus isp(4)s on sparc64 machines.

Report and testing:  Marius Strobl <marius@alchemy.franken.de>

20 years agoOverhaul the /etc/rc.d/diskless script by splitting it out into
brooks [Tue, 23 Mar 2004 23:22:35 +0000 (23:22 +0000)]
Overhaul the /etc/rc.d/diskless script by splitting it out into
hostname, resolve, tmp, and var scripts.  The latter three are new and
were repo copied.  These scripts no longer depend on being booted with
and NFS root instead attempt to automaticly create mfs /tmp and /var
volumes if the they are not writable.  This behavior can be overridden
in /etc/rc.conf.

Reviewed by:    luigi, pjd

20 years agoCorrect the termination condition of the DVMA pruning loop in
tmm [Tue, 23 Mar 2004 23:20:34 +0000 (23:20 +0000)]
Correct the termination condition of the DVMA pruning loop in
iommu_dvma_vallocseg(), which I botched in r1.32. This bug could
cause an endless loop when a map was loaded and DVMA was scarce,
or that map had a stringent alignment or boundary.

Report and additional testing: Marius Strobl <marius@alchemy.franken.de>

20 years agoIntitialize the frame pointer and return pc of a new process created
tmm [Tue, 23 Mar 2004 23:12:02 +0000 (23:12 +0000)]
Intitialize the frame pointer and return pc of a new process created
in cpu_fork(). This prevents the stack tracer from running past the
end of the stack (only the pc is checked in that case), which became
fatal when db_print_backtrace() was introduced and called outside
of ddb.

Additional testing: kris

20 years agoAdd rc.d script for pf(4) (more to come once pflogd(8) works as well).
mlaier [Tue, 23 Mar 2004 22:30:15 +0000 (22:30 +0000)]
Add rc.d script for pf(4) (more to come once pflogd(8) works as well).
Update defaults and write some lines for rc.conf(5) also.
Mostly dup'ed from ipf

Reviewed by: -current
Approved by: bms(mentor)

20 years agoEnable interrupt preemption on Alpha. Previously, flipping this switch on
jhb [Tue, 23 Mar 2004 22:28:16 +0000 (22:28 +0000)]
Enable interrupt preemption on Alpha.  Previously, flipping this switch on
caused hangs on SMP systems under load.  My theory was that an interrupted
thread was migrating and returning to PAL on a different CPU and that that
caused the hangs.  To prevent this, I used the recently added sched_pin()
API to pin the interrupted thread to the CPU that received the interrupt
across ithread_schedule() to prevent migration.  This seems to have fixed
the hangs based on tests by several folks on the alpha@ list.

Tested by: wilko, tisco, several others on alpha@

20 years agoRemove doscmd from the base system now that it lives in the ports tree.
des [Tue, 23 Mar 2004 22:27:24 +0000 (22:27 +0000)]
Remove doscmd from the base system now that it lives in the ports tree.

20 years agoDrop the -d argument to mtree since some of the spec files now list
des [Tue, 23 Mar 2004 22:17:34 +0000 (22:17 +0000)]
Drop the -d argument to mtree since some of the spec files now list
symlinks.

20 years agoAdd share/nls/{POSIX,en_US.US-ASCII} as symlinks to C.
des [Tue, 23 Mar 2004 22:16:38 +0000 (22:16 +0000)]
Add share/nls/{POSIX,en_US.US-ASCII} as symlinks to C.

20 years agoAdd myself as maintainer for contrib/pf
mlaier [Tue, 23 Mar 2004 21:22:24 +0000 (21:22 +0000)]
Add myself as maintainer for contrib/pf

Approved by: bms(mentor)

20 years agoRecently I realized that the ADMtek 8211 driver wasn't working correctly
wpaul [Tue, 23 Mar 2004 19:51:17 +0000 (19:51 +0000)]
Recently I realized that the ADMtek 8211 driver wasn't working correctly
(NIC would claim to establish a link with an ad-hoc net but it couldn't
send/receive packets). It turns out that every time the checkforhang
handler was called by ndis_ticktask(), the driver would generate a new
media connect event. The NDIS spec says the checkforhang handler is
called "approximately every 2 seconds" but using exactly 2 seconds seems
too fast. Using 3 seconds makes it happy again, so we'll go with that
for now.

20 years agoAdd an implementation of uiomove_fromphys() for PowerPC. This
alc [Tue, 23 Mar 2004 18:26:03 +0000 (18:26 +0000)]
Add an implementation of uiomove_fromphys() for PowerPC.  This
implementation uses the direct virtual-to-physical mapping.

Discussed with: grehan

20 years agoAdd support for decoding Timespec, Timeval, Itimerval, Pollfd,
dwmalone [Tue, 23 Mar 2004 12:37:02 +0000 (12:37 +0000)]
Add support for decoding Timespec, Timeval, Itimerval, Pollfd,
Fd_set and Sigaction structures. Use these for printing the arguments
to sigaction(), nanosleep(), select(), poll(), gettimeofday(),
clock_gettime(), recvfrom(), getitimer() and setitimer().

This is based on Dan's patch from the PR but I've hacked it for
style and some other issues. While Dan has checked this patch, any
goofs are probably my fault.

(The PR also contains support for the dual return values of pipe().
These will follow once I've ported that support to platforms other
than i386.)

PR: 52190
Submitted by: Dan Nelson <dnelson@allantgroup.com>

20 years agoUse pread to implement pread, rather than taking a detour throug stdio.
dwmalone [Tue, 23 Mar 2004 12:29:17 +0000 (12:29 +0000)]
Use pread to implement pread, rather than taking a detour throug stdio.

PR: 52190
Submitted by: Dan Nelson <dnelson@allantgroup.com>

20 years agodecode mprotect args while i'm here.
alfred [Tue, 23 Mar 2004 09:16:18 +0000 (09:16 +0000)]
decode mprotect args while i'm here.

20 years agodecode fcntl and mmap arguments.
alfred [Tue, 23 Mar 2004 09:04:06 +0000 (09:04 +0000)]
decode fcntl and mmap arguments.

20 years agoDo not copy vm_exitingcnt to the new vmspace in vmspace_exec(). Copying
tjr [Tue, 23 Mar 2004 08:37:34 +0000 (08:37 +0000)]
Do not copy vm_exitingcnt to the new vmspace in vmspace_exec(). Copying
it led to impossibly high values in the new vmspace, causing it to never
drop to 0 and be freed.

20 years agoRe-add the hand-optimised assembler versions of some of the ciphers
markm [Tue, 23 Mar 2004 08:32:29 +0000 (08:32 +0000)]
Re-add the hand-optimised assembler versions of some of the ciphers
to the build.

Should have done this ages ago: markm
Reminded above to do this: peter

20 years agomdoc fixes
kientzle [Tue, 23 Mar 2004 03:14:01 +0000 (03:14 +0000)]
mdoc fixes

20 years agoChange the !MPSAFE boot string to something that doesn't potentially
obrien [Tue, 23 Mar 2004 01:58:09 +0000 (01:58 +0000)]
Change the !MPSAFE boot string to something that doesn't potentially
scare users that the kernel won't run on MP systems.

20 years agoEmit a traceback when witness_trace is set and witness_warn() is
alfred [Tue, 23 Mar 2004 00:32:27 +0000 (00:32 +0000)]
Emit a traceback when witness_trace is set and witness_warn() is
called and triggers (typically caused by sleeping with a non-sleepable
lock).

Reviewed by: jhb

20 years agoRemove another case of grabbing Giant before doing a kthread_exit()
wpaul [Mon, 22 Mar 2004 22:46:22 +0000 (22:46 +0000)]
Remove another case of grabbing Giant before doing a kthread_exit()
which is now no longer needed.

20 years agoRather than display which interrupts are MPSAFE, display those that aren't.
obrien [Mon, 22 Mar 2004 22:36:11 +0000 (22:36 +0000)]
Rather than display which interrupts are MPSAFE, display those that aren't.
This way we can take stock of the work to be done.  boot -v will note those
interrupts that are MPSAFE.

20 years agoClarify the description of the "established" option.
ceri [Mon, 22 Mar 2004 21:24:38 +0000 (21:24 +0000)]
Clarify the description of the "established" option.

PR: docs/50391
Submitted by: root@edcsm.jussieu.fr
MFC after: 1 week

20 years agoNote that only one webnfs share is allowed per NFS server.
ceri [Mon, 22 Mar 2004 21:05:57 +0000 (21:05 +0000)]
Note that only one webnfs share is allowed per NFS server.

PR: docs/45371
Submitted by: Mattias Pantzare <pantzer@ludd.luth.se>,
Matthew D. Fuller <fullermd@over-yonder.net>
MFC after: 2 days

20 years agoUse the correct length for appending an extended irq resource. This may
njl [Mon, 22 Mar 2004 20:39:20 +0000 (20:39 +0000)]
Use the correct length for appending an extended irq resource.  This may
have broken APIC routing.  This bug has been present since rev 1.33.

20 years agoShorten some printfs to fit better. No other functional changes.
njl [Mon, 22 Mar 2004 20:36:33 +0000 (20:36 +0000)]
Shorten some printfs to fit better.  No other functional changes.

20 years agoWhitespace and comment changes. No MD5 change to the object file.
njl [Mon, 22 Mar 2004 20:32:27 +0000 (20:32 +0000)]
Whitespace and comment changes.  No MD5 change to the object file.

20 years agoRevert 1.22. Go back to "FreeBSD" as our _OS value. Since we now support
njl [Mon, 22 Mar 2004 20:11:49 +0000 (20:11 +0000)]
Revert 1.22.  Go back to "FreeBSD" as our _OS value.  Since we now support
_OSI, we can claim compatibility with MS products without claiming to BE
one.  :-)

20 years agoBump FreeBSD version to indicate new bus_alloc_resource_any API.
njl [Mon, 22 Mar 2004 19:57:34 +0000 (19:57 +0000)]
Bump FreeBSD version to indicate new bus_alloc_resource_any API.

20 years agoI'm a dumbass: the test in the MOD_SHUTDOWN case in ndis_modevent()
wpaul [Mon, 22 Mar 2004 18:34:37 +0000 (18:34 +0000)]
I'm a dumbass: the test in the MOD_SHUTDOWN case in ndis_modevent()
that checks to see if any devices are still in the devlist was reversed.

20 years agoif_ndis.c no longer depends on either pci or pccard. Also, add an
wpaul [Mon, 22 Mar 2004 18:00:39 +0000 (18:00 +0000)]
if_ndis.c no longer depends on either pci or pccard. Also, add an
extra entry for if_ndis_pci.c that depends on cardbus, just to cover
all the bases. (I don't think you can have cardbus without PCI, but
just in case...)

20 years agoCommit a message to UPDATING indicating that during network stack locking
rwatson [Mon, 22 Mar 2004 16:37:32 +0000 (16:37 +0000)]
Commit a message to UPDATING indicating that during network stack locking
merges, debug.mpsafenet will no longer be safe to use, and that it should
be turned off (which is the default).  I will commit a follow-up message
to UPDATING (and post to freebsd-current) when it's safe to turn it on,
with a description of under what circumstances it is safe to use.

20 years agoSet start_cmd and stop_cmd correctly so the code that extracts the names
des [Mon, 22 Mar 2004 16:35:35 +0000 (16:35 +0000)]
Set start_cmd and stop_cmd correctly so the code that extracts the names
of the ntp servers from ntp.conf is actually used.  Remove pidfile since
ntpdate is not a daemon.

20 years agoLock down global variables in if_gre:
rwatson [Mon, 22 Mar 2004 16:04:43 +0000 (16:04 +0000)]
Lock down global variables in if_gre:

- Add gre_mtx to protect global softc list.
- Hold gre_mtx over various list operations (insert, delete).
- Centralize if_gre interface teardown in gre_destroy(), and call this
  from modevent unload and gre_clone_destroy().
- Export gre_mtx to ip_gre.c, which walks the gre list to look up gre
  interfaces during encapsulation.  Add a wonking comment on how we need
  some sort of drain/reference count mechanism to keep gre references
  alive while in use and simultaneous destroy.

This commit does not lockdown softc data, which follows in a future
commit.

20 years agoUse ' rather than ".
obrien [Mon, 22 Mar 2004 15:45:17 +0000 (15:45 +0000)]
Use ' rather than ".

Submitted by: ru

20 years agoLock down global variables in if_gif:
rwatson [Mon, 22 Mar 2004 15:43:14 +0000 (15:43 +0000)]
Lock down global variables in if_gif:

- Add gif_mtx, which protects globals.
- Hold gif_mtx around manipulation of gif_softc_list.
- Abstract gif destruction code into gif_destroy(), which tears down
  a softc after it's been removed from the global list by either module
  unload or clone destroy.
- Lock gif_called, even though we know gif_called is broken with reentrant
  network processing.
- Document an event ordering problem in gif_set_tunnel() that will need
  to be fixed.

gif_softc fields not locked down in this commit.

20 years agoForced commit note to explain why NO_POSIX_COMPAT was removed with upgrade.
ache [Mon, 22 Mar 2004 14:44:00 +0000 (14:44 +0000)]
Forced commit note to explain why NO_POSIX_COMPAT was removed with upgrade.

NO_POSIX_COMPAT was added in 1996 as attempt to resolve POSIX regex
functions conflict for fixit floppy (they are both in libc and libgnuregex)
The way it done (simple excluding POSIX functions from compile) breaks
legitimate case when, say, some program from ports calls both POSIX
and GNU regex functions (it will call POSIX ones from libc and fails,
because header <regex.h> is from GNU).

Now our fixit floppy (and other crunched floppies, as I check) not
use libgnuregex, so the hack can be removed.

20 years agoMove "called", a static function variable used to detect recursive
rwatson [Mon, 22 Mar 2004 14:24:26 +0000 (14:24 +0000)]
Move "called", a static function variable used to detect recursive
processing with gif interfaces, to a global variable named "gif_called".
Add an annotation that this approach will not work with a reentrant
network stack, and that we should instead use packet tags to detect
excessive recursive processing.

20 years agoMFi386: revision 1.480.
nyan [Mon, 22 Mar 2004 13:37:11 +0000 (13:37 +0000)]
MFi386: revision 1.480.

20 years agoOn startup, warn if inetd's config file doesn't exist. This isn't
dwmalone [Mon, 22 Mar 2004 12:01:23 +0000 (12:01 +0000)]
On startup, warn if inetd's config file doesn't exist. This isn't
exactly the same as patch from the PR, which also exited if the
config file was missing. I didn't use Jeff's patch because I was
worried that some people might start inetd, create the config file
and then HUP inetd.

PR: 60806
Submitted by: Jeff Ito <jeffi@rcn.com>
MFC after: 2 weeks

20 years agoUse fseeko() to properly support large files.
tjr [Mon, 22 Mar 2004 11:15:03 +0000 (11:15 +0000)]
Use fseeko() to properly support large files.

20 years agoRemove some netbsd debug code that crept into rev 1.116
ps [Mon, 22 Mar 2004 10:17:40 +0000 (10:17 +0000)]
Remove some netbsd debug code that crept into rev 1.116

20 years agoUse errc().
tjr [Mon, 22 Mar 2004 09:33:10 +0000 (09:33 +0000)]
Use errc().

20 years agoAdd an implementation of uiomove_fromphys() to sparc64. This
alc [Mon, 22 Mar 2004 08:08:25 +0000 (08:08 +0000)]
Add an implementation of uiomove_fromphys() to sparc64.  This
implementation could be characterized as a hybrid of the amd64 and i386
implementations.  Specifically, the direct virtual-to-physical mapping is
used if possible and sf_buf_alloc() is used if the direct map cannot.

20 years agoJust use autoconf-compatible feature macros; don't try to be clever.
kientzle [Mon, 22 Mar 2004 05:17:23 +0000 (05:17 +0000)]
Just use autoconf-compatible feature macros; don't try to be clever.

20 years agoAdd an observation by Dijkstra software engineering.
grog [Mon, 22 Mar 2004 05:02:01 +0000 (05:02 +0000)]
Add an observation by Dijkstra software engineering.

20 years agoRename 'ddpcb' variable to 'ddpcb_list' to better distinguish it from
rwatson [Mon, 22 Mar 2004 04:54:36 +0000 (04:54 +0000)]
Rename 'ddpcb' variable to 'ddpcb_list' to better distinguish it from
'struct ddpcb'.

20 years agoRename 'at_ifaddr' list to 'at_ifaddr_list' so that the variable is
rwatson [Mon, 22 Mar 2004 04:50:36 +0000 (04:50 +0000)]
Rename 'at_ifaddr' list to 'at_ifaddr_list' so that the variable is
more easily mechanically distinguished from 'struct at_ifaddr'.

20 years agoCompare pointers with NULL rather than 0, or treating them as boolans in
rwatson [Mon, 22 Mar 2004 03:57:01 +0000 (03:57 +0000)]
Compare pointers with NULL rather than 0, or treating them as boolans in
if statements.

at_rmx gets a $FreeBSD$ out of the deal also (this code appears to be
unused).

20 years agoMAC addresses are 8 bits in ARCNET. Adjust bcopy().
mdodd [Mon, 22 Mar 2004 03:52:51 +0000 (03:52 +0000)]
MAC addresses are 8 bits in ARCNET.  Adjust bcopy().

20 years agoAlso modify ddp_input.c with the following changes previously applied
rwatson [Mon, 22 Mar 2004 03:48:31 +0000 (03:48 +0000)]
Also modify ddp_input.c with the following changes previously applied
to other files in netatalk:

  Log:
  Since I have my hands all over netatalk adding locking and restructuring
  it, cinch the file's style closer to style(9) with regard to parenthesis:

    s/( /(/g
    s/ )/)/g
    s/return(/return (/g
    s/return 0/return (0)/
    s/return 1/return (1)/

20 years agoSince I have my hands all over netatalk adding locking and restructuring
rwatson [Mon, 22 Mar 2004 03:24:10 +0000 (03:24 +0000)]
Since I have my hands all over netatalk adding locking and restructuring
it, cinch the file's style closer to style(9) with regard to parenthesis:

  s/( /(/g
  s/ )/)/g
  s/return(/return (/g
  s/return 0/return (0)/
  s/return 1/return (1)/

20 years agoRemove WFORMAT, style.Makefile(5).
obrien [Mon, 22 Mar 2004 03:17:36 +0000 (03:17 +0000)]
Remove WFORMAT, style.Makefile(5).

20 years agoDont count bytes being sent to a disconnected hook.
julian [Mon, 22 Mar 2004 01:15:28 +0000 (01:15 +0000)]
Dont count bytes being sent to a disconnected hook.

MFC after: 3 days

20 years agoShare the i386 boot manual page.
obrien [Mon, 22 Mar 2004 00:52:27 +0000 (00:52 +0000)]
Share the i386 boot manual page.

20 years agoThe Intel 2200BG NDIS driver does an alloca() of about 5000 bytes
wpaul [Mon, 22 Mar 2004 00:41:41 +0000 (00:41 +0000)]
The Intel 2200BG NDIS driver does an alloca() of about 5000 bytes
when it associates with a net. Because FreeBSD's kstack size is only
2 pages by default, this blows the stack and causes a double fault.

To deal with this, we now create all our kthreads with 8 stack pages.
Also, we now run all timer callouts in the ndis swi thread (since
they would otherwise run in the clock ithread, whose stack is too
small). It happens that the alloca() in this case was occuring within
the interrupt handler, which was already running in the ndis swi
thread, but I want to deal with the callouts too just to be extra
safe.

NOTE: this will only work if you update vm_machdep.c with the change
I just committed. If you don't include this fix, setting the number
of stack pages with kthread_create() has essentially no effect.

20 years agoThe kthread_create() API is supposed to allow you to create threads
wpaul [Mon, 22 Mar 2004 00:28:38 +0000 (00:28 +0000)]
The kthread_create() API is supposed to allow you to create threads
with more than the normal amount of stack pages, however the stack
pointer always wound up being initialized using KSTACK_PAGES. It
should be using td->td_kstack_pages instead. This means that although
the vm subsystem would give you all the stack pages you asked for,
%esp would always be initialized as if you had just 2 pages, and
the rest would go to waste.

I wanted to use the 'give me more stack pages' feature of kthread_create()
because the Intel 2200BG NDIS driver does an alloca() of about 5000 bytes,
which wrecks the stack with the default 2 page size, and I was baffled
that no matter how much code I shoved into thread contexts with
allegedly larger stacks, the thing would still crash unless I changed
KSTACK_PAGES.

Note: this bug is present in _ALL_ arches at this point. Peter has
promised to merge this fix into all of them.

20 years agoAdd an implementation of uiomove_fromphys() for i386. This implementation
alc [Sun, 21 Mar 2004 20:28:36 +0000 (20:28 +0000)]
Add an implementation of uiomove_fromphys() for i386.  This implementation
uses sf_buf_alloc() and sf_buf_free() to create and destroy the necessary
ephemeral mappings.

20 years agoMake if_ndis_pci.c and if_ndis_pccard.c use bus_alloc_resource() again
wpaul [Sun, 21 Mar 2004 19:56:41 +0000 (19:56 +0000)]
Make if_ndis_pci.c and if_ndis_pccard.c use bus_alloc_resource() again
instead of bus_alloc_resource_any() to restore source compatibility
with 5.2-REL and 5.2.1-REL systems. bus_alloc_resource_any() doesn't
really do anything besides hide some of bus_alloc_resource()'s arguments
from us, and in my opinion this isn't worth breaking backwards
compatibility for people who want to use the NDISulator code on 5.2.x.

20 years agoLogin classes are not "unused."
dannyboy [Sun, 21 Mar 2004 19:36:16 +0000 (19:36 +0000)]
Login classes are not "unused."
Reference login.conf(5).

PR: 62830
Submitted by: Matthew D. Fuller <fullermd@over-yonder.net>