]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoAs the default MAXDSIZ and DFLDSIZ is 512MB, bump the example values
tanimura [Wed, 31 Jan 2001 04:07:34 +0000 (04:07 +0000)]
As the default MAXDSIZ and DFLDSIZ is 512MB, bump the example values
to 1GB. A box of mine is running with MAXDSIZ and DFLDSIZ increased
up to 1.5GB.

Wishlist: It would be nice to warn if MAXTSIZ + MAXDSIZ + MAXSSIZ
exceeds VM_MAXUSER_ADDRESS - VM_MINUSER_ADDRESS.

23 years agos/_thread_sys_write/__sys_write/
deischen [Wed, 31 Jan 2001 02:16:57 +0000 (02:16 +0000)]
s/_thread_sys_write/__sys_write/

Submitted by: Mike Heffner <mheffner@vt.edu>

23 years agoFix documentation: "proxy" is not a keyword for the arp(8) command line.
archie [Wed, 31 Jan 2001 00:03:31 +0000 (00:03 +0000)]
Fix documentation: "proxy" is not a keyword for the arp(8) command line.

23 years agoRevert mutex initialization check to look at mtx_description.
jasone [Tue, 30 Jan 2001 22:28:03 +0000 (22:28 +0000)]
Revert mutex initialization check to look at mtx_description.

Pointed out by: jlemon, jhb

23 years agoImplement direct support for semipersistant nodes.
julian [Tue, 30 Jan 2001 20:51:52 +0000 (20:51 +0000)]
Implement direct support for semipersistant nodes.
(e.g. ethernet nodes are persistent until you rip out the hardware)
Use this support in the ethernet and sample nodes.
Add some more abstraction on the 'item's so that  node and
hook reference counting can be checked easier.
Slight man page correction.
Make pppoe type dependent on ethernet type.
Clean up node shutdown a little.
Move a mutex from MTX_SPIN to MTX_DEF (oops)
Fix small ref-counting bug.
remove warning on one2many type.

23 years agoArgh, fix a nit that snuck in while trying to resolve conflicts.
jhb [Tue, 30 Jan 2001 20:29:11 +0000 (20:29 +0000)]
Argh, fix a nit that snuck in while trying to resolve conflicts.

23 years ago- Fix TUNABLE_STR_FETCH() to actually be a code fragment rather than
jhb [Tue, 30 Jan 2001 20:13:02 +0000 (20:13 +0000)]
- Fix TUNABLE_STR_FETCH() to actually be a code fragment rather than
  declaring a static function.
- Modify TUNABLE_*_DECL() to use TUNABLE_*_FETCH() to avoid code
  duplication.

Reviewed by: peter

23 years agoAdded used include of <sys/mutex.h>. The SMP case was broken by
bde [Tue, 30 Jan 2001 17:05:58 +0000 (17:05 +0000)]
Added used include of <sys/mutex.h>.  The SMP case was broken by
incompletely converting simplelocks to mutexes (COM_LOCK() is supposed
to hide the SMP locking internals, but it now depends on mutex interfaces
being visible).

23 years agoNew release notes: bridge(4)/dummynet(4) updates, find(1) -empty,
bmah [Tue, 30 Jan 2001 16:18:56 +0000 (16:18 +0000)]
New release notes:  bridge(4)/dummynet(4) updates, find(1) -empty,
tftpd(8) -c/-C, vidcontrol(1) -g, ldconfig(8) permission/ownership checks.

23 years agoUpdate port_after.h to reflect FreeBSD values.
asmodai [Tue, 30 Jan 2001 12:28:21 +0000 (12:28 +0000)]
Update port_after.h to reflect FreeBSD values.

23 years agoAllow the output of /etc/security to be logged or mailed to different
brian [Tue, 30 Jan 2001 10:24:18 +0000 (10:24 +0000)]
Allow the output of /etc/security to be logged or mailed to different
users in line with ${daily,weekly,monthly}_output using a new
$daily_status_security_output variable.

PR: 24643

23 years agoMake mount_mfs annoy users for 15 seconds and point them at mdconfig(8).
phk [Tue, 30 Jan 2001 10:21:20 +0000 (10:21 +0000)]
Make mount_mfs annoy users for 15 seconds and point them at mdconfig(8).

23 years agoMake klds for the ar and sr devices and hook them up.
jhay [Tue, 30 Jan 2001 10:20:43 +0000 (10:20 +0000)]
Make klds for the ar and sr devices and hook them up.

The normal versions are called if_ar.ko and if_sr.ko and the netgraph
versions ng_sync_ar.ko and ng_sync_sr.ko.

23 years agoMake vnconfig annoy users for 15 seconds to make them switch to mdconfig.
phk [Tue, 30 Jan 2001 10:12:51 +0000 (10:12 +0000)]
Make vnconfig annoy users for 15 seconds to make them switch to mdconfig.

23 years agoRun purgedir (a local function) on /var/run instead of rm /var/run/*
brian [Tue, 30 Jan 2001 10:07:03 +0000 (10:07 +0000)]
Run purgedir (a local function) on /var/run instead of rm /var/run/*

PR: 24612
Submitted by: David Drum <david@mu.org>

23 years agoAdd a BUF_KERNPROC() in the BIO_DELETE path.
phk [Tue, 30 Jan 2001 10:06:08 +0000 (10:06 +0000)]
Add a BUF_KERNPROC() in the BIO_DELETE path.

This seems to fix the problem which md(4) backed filesystems exposed.

23 years agoThe ar and sr devices have moved to sys/dev/.
jhay [Tue, 30 Jan 2001 10:05:44 +0000 (10:05 +0000)]
The ar and sr devices have moved to sys/dev/.

23 years agoThis include file has found a new home in sys/dev/ic/.
jhay [Tue, 30 Jan 2001 10:04:25 +0000 (10:04 +0000)]
This include file has found a new home in sys/dev/ic/.

23 years agoNewbusify the sr device and move it to its new location.
jhay [Tue, 30 Jan 2001 10:02:10 +0000 (10:02 +0000)]
Newbusify the sr device and move it to its new location.

23 years agoUpdate include files to reflect the new location of the ar device.
jhay [Tue, 30 Jan 2001 10:01:15 +0000 (10:01 +0000)]
Update include files to reflect the new location of the ar device.

23 years agoReflect the new location of the ar and sr devices.
jhay [Tue, 30 Jan 2001 10:00:04 +0000 (10:00 +0000)]
Reflect the new location of the ar and sr devices.

23 years agoStyle (whitespace) nit.
peter [Tue, 30 Jan 2001 09:44:37 +0000 (09:44 +0000)]
Style (whitespace) nit.

23 years agoFix two minor nits.
phk [Tue, 30 Jan 2001 08:39:52 +0000 (08:39 +0000)]
Fix two minor nits.

Existences revealed, but no details offered by: bp

23 years agoFix cut and paste error in a comment.
julian [Tue, 30 Jan 2001 07:58:30 +0000 (07:58 +0000)]
Fix cut and paste error in a comment.
Submitted by: Peter Wemm <peter@freebsd.org>

23 years agoFix a race between the syncer and umount. When you umount a softupdates
dillon [Tue, 30 Jan 2001 06:31:59 +0000 (06:31 +0000)]
Fix a race between the syncer and umount.  When you umount a softupdates
filesystem softdep_process_worklist() is called in a loop until it indicates
that no dependancies remain, but the determination of that fact depends on
there only being one softdep_process_worklist() instance running.  It was
possible for the syncer to also be running softdep_process_worklist()
and the pre-existing checks in the code to prevent this were not sufficient
to prevent the race.  This patch solves the problem.

Approved-by: mckusick
23 years agoUpdate release notes: devfs(5) now in GENERIC, libc_r changes.
bmah [Tue, 30 Jan 2001 05:02:15 +0000 (05:02 +0000)]
Update release notes:  devfs(5) now in GENERIC, libc_r changes.

23 years agoRemove some leftovers from the CMAP* stuff in globaldata and the
peter [Tue, 30 Jan 2001 04:02:28 +0000 (04:02 +0000)]
Remove some leftovers from the CMAP* stuff in globaldata and the
BSP and AP startup.

23 years agoRemove unused GD_CPU_LOCKID, GD_OTHER_CPUS, PS_IDLESTACK and
peter [Tue, 30 Jan 2001 04:00:11 +0000 (04:00 +0000)]
Remove unused GD_CPU_LOCKID, GD_OTHER_CPUS, PS_IDLESTACK and
PS_IDLESTACK_TOP

23 years agoRemove unnecessary locking to protect the p_upages_obj and p_addr
jhb [Tue, 30 Jan 2001 00:35:35 +0000 (00:35 +0000)]
Remove unnecessary locking to protect the p_upages_obj and p_addr
pointers.

23 years ago- Use the right name for the proctree lock in the locking key.
jhb [Tue, 30 Jan 2001 00:10:37 +0000 (00:10 +0000)]
- Use the right name for the proctree lock in the locking key.
- Add a note about the special locking semantics used for members such as
  p_cred that are read by multiple processes but only written to by the
  current process.
- Change p_upages_obj's locking key to note that it is created at fork
  and left alone afterwards (the actual pointer, not what it points to.)
- Mark p_intr_nesting_level as being implicitly locked since only curproc
  accesses it.

Reviewed by: jake

23 years agoUpdate for security advisories: 01:11, 01:12, 01:13, BIND fix/update.
bmah [Mon, 29 Jan 2001 23:54:48 +0000 (23:54 +0000)]
Update for security advisories:  01:11, 01:12, 01:13, BIND fix/update.

23 years agoPut picobsd build scripts in sync with the ones in -STABLE.
luigi [Mon, 29 Jan 2001 22:47:46 +0000 (22:47 +0000)]
Put picobsd build scripts in sync with the ones in -STABLE.

With these scripts i was able to build a _working_ image of a
bridge-like floppy image with a reasonable set of utilities
in it, despite the code size increase we have in 5.0

The scripts are slightly different from the previous ones especially
on the place where the kernel and binaries are built. Hopefully
this will not cause too much trouble to people (is there any???)
using the old scripts.

23 years agoUnbreak world by correctly specifying the prototype for __sys_aio_suspend.
deischen [Mon, 29 Jan 2001 18:59:53 +0000 (18:59 +0000)]
Unbreak world by correctly specifying the prototype for __sys_aio_suspend.
A make buildworld was done but not with the committed pthread_private.h.

Reported by: Manfred Antar <null@pozo.com>

23 years agoAdd option ATA_ENABLE_WC for enabling write caching (now off by default).
sos [Mon, 29 Jan 2001 18:00:35 +0000 (18:00 +0000)]
Add option ATA_ENABLE_WC for enabling write caching (now off by default).

23 years agoAdd text for option ATA_ENABLE_WC.
sos [Mon, 29 Jan 2001 17:58:16 +0000 (17:58 +0000)]
Add text for option ATA_ENABLE_WC.

23 years agoAARGH! This does not mix too well with the -o noatime mount option.
peter [Mon, 29 Jan 2001 15:26:35 +0000 (15:26 +0000)]
AARGH!  This does not mix too well with the -o noatime  mount option.
:-(

23 years agoFlesh out EFI support somewhat.
dfr [Mon, 29 Jan 2001 13:31:19 +0000 (13:31 +0000)]
Flesh out EFI support somewhat.

23 years agoSupply a stub bpf_validate() (always returning false - the script is not
peter [Mon, 29 Jan 2001 13:26:14 +0000 (13:26 +0000)]
Supply a stub bpf_validate() (always returning false - the script is not
valid) if BPF is missing.
The netgraph_bpf node forced bpf to be present, reflect that in the
options.
Stop doing a 'count bpf' - we provide stubs.
Since a handful of drivers still refer to "bpf.h", provide a more accurate
indication that the API is present always. (eg: netinet6)

23 years agoUse M_PANIC instead of if (sc == NULL) panic();
peter [Mon, 29 Jan 2001 13:21:34 +0000 (13:21 +0000)]
Use M_PANIC instead of if (sc == NULL) panic();

23 years agoRemove unused variable 'int n;'
peter [Mon, 29 Jan 2001 13:05:21 +0000 (13:05 +0000)]
Remove unused variable 'int n;'

23 years agoAdd M_PANIC flag to the list of available flags passed to malloc().
bp [Mon, 29 Jan 2001 12:48:37 +0000 (12:48 +0000)]
Add M_PANIC flag to the list of available flags passed to malloc().
With this flag set malloc() will panic if memory allocation failed.
This usable only in critical places where failed allocation is fatal.

Reviewed by: peter

23 years agoStop counting sppp interfaces, we were just testing its presence to give
peter [Mon, 29 Jan 2001 12:27:46 +0000 (12:27 +0000)]
Stop counting sppp interfaces, we were just testing its presence to give
a warning if it was missing.

23 years agoConvert mca (microchannel bus support) from something that we count
peter [Mon, 29 Jan 2001 11:57:27 +0000 (11:57 +0000)]
Convert mca (microchannel bus support) from something that we count
(bogus) to something that we test for the presence of.

23 years agoAdd a few ``const''s to silence some -Wwrite-strings warnings
brian [Mon, 29 Jan 2001 11:44:13 +0000 (11:44 +0000)]
Add a few ``const''s to silence some -Wwrite-strings warnings

23 years agoZap silly #if NPCI > 0 and the hoops that we jump through for the module
peter [Mon, 29 Jan 2001 11:38:39 +0000 (11:38 +0000)]
Zap silly #if NPCI > 0 and the hoops that we jump through for the module
case.  Use an 'and' case in conf/files so that it only gets compiled if
pci is present.

23 years agoYikes, these files bogusly #include "loop.h" but didn't use the value.
peter [Mon, 29 Jan 2001 11:28:20 +0000 (11:28 +0000)]
Yikes, these files bogusly #include "loop.h" but didn't use the value.
My searching for NLOOP missed them. :-(

23 years agoConvert 'count en', 'count lnc', 'count fpa', 'count loop' and 'count ar'
peter [Mon, 29 Jan 2001 11:21:00 +0000 (11:21 +0000)]
Convert 'count en', 'count lnc', 'count fpa', 'count loop' and 'count ar'
back to standard 'optional'.

23 years agoRemove unused #include "lnc.h"
peter [Mon, 29 Jan 2001 11:19:30 +0000 (11:19 +0000)]
Remove unused #include "lnc.h"

23 years agoMake the number of loopback interfaces dynamically tunable. Why one
peter [Mon, 29 Jan 2001 11:06:26 +0000 (11:06 +0000)]
Make the number of loopback interfaces dynamically tunable.  Why one
would *want* to is a different story, but it used to be able to be done
statically.  Get rid of #include "loop.h" and struct ifnet loif[NLOOP];
This could be used as an example of how to do this in other drivers,
for example: ccd.

23 years agoFix 'tunefs -p'
ben [Mon, 29 Jan 2001 11:00:16 +0000 (11:00 +0000)]
Fix 'tunefs -p'

Reviewed by: sheldonh

23 years agoRemove unused #include "snp.h"
peter [Mon, 29 Jan 2001 10:06:22 +0000 (10:06 +0000)]
Remove unused #include "snp.h"

23 years agoConvert ata and atapi #if NATA* > 0 to options instead. Stop config
peter [Mon, 29 Jan 2001 10:01:11 +0000 (10:01 +0000)]
Convert ata and atapi #if NATA* > 0 to options instead.  Stop config
trying to count the number of ata* devs since they were not used anyway.

23 years agoUse <sys/queue.h> instead of home-rolled list.
phk [Mon, 29 Jan 2001 09:45:51 +0000 (09:45 +0000)]
Use <sys/queue.h> instead of home-rolled list.

Submitted by: "Jason Smethers" <jsmethers@pdq.net>

23 years agoTurn '#if NSNP > 0' into an option.
peter [Mon, 29 Jan 2001 09:43:36 +0000 (09:43 +0000)]
Turn '#if NSNP > 0' into an option.

23 years agoSend "#if NISA > 0" to the bit-bucket and replace it with an option.
peter [Mon, 29 Jan 2001 09:38:39 +0000 (09:38 +0000)]
Send "#if NISA > 0" to the bit-bucket and replace it with an option.
These were compile-time "is the isa code present?" tests and not
'how many isa busses' tests.

23 years agoI confess. This is a truely nasty hack. I am so ashamed!
peter [Mon, 29 Jan 2001 09:34:43 +0000 (09:34 +0000)]
I confess.  This is a truely nasty hack.  I am so ashamed!
(I think config(8) source does bad things to your brain :-)

Clean up likely stray *.h files in the build directory.
Eg: if isa.h ceases being generated, zap it.
The heuristics to figure out a 'likely' file are pretty revolting.

23 years agoStyle police.
des [Mon, 29 Jan 2001 08:59:30 +0000 (08:59 +0000)]
Style police.

Submitted by: ru

23 years agoFix some log_Printf() casting
brian [Mon, 29 Jan 2001 08:46:58 +0000 (08:46 +0000)]
Fix some log_Printf() casting

Obtained from: NetBSD (pkgsrc)

23 years agoGag. These compiled because I had a stray "eisa.h" in my config dir.
peter [Mon, 29 Jan 2001 08:40:16 +0000 (08:40 +0000)]
Gag. These compiled because I had a stray "eisa.h" in my config dir.

23 years agoRemove stray #include "isa.h"
peter [Mon, 29 Jan 2001 08:33:55 +0000 (08:33 +0000)]
Remove stray #include "isa.h"

23 years agoReplace the full path to sysinstall with a standard manual page
sheldonh [Mon, 29 Jan 2001 08:22:21 +0000 (08:22 +0000)]
Replace the full path to sysinstall with a standard manual page
reference.  The sysinstall binary is now in root's standard PATH,
so there's no need for explicit pathing, and there's some value
in a manual page reference.

23 years agoThis patch reestablishes the spec_fsync() guarentee that synchronous
dillon [Mon, 29 Jan 2001 08:19:28 +0000 (08:19 +0000)]
This patch reestablishes the spec_fsync() guarentee that synchronous
fsyncs, which typically occur during unmounting, will drain all dirty
buffers even if it takes multiple passes to do so.  The guarentee was
mangled by the last patch which solved a problem due to -current disabling
interrupts while holding giant (which caused an infinite spin loop waiting for
I/O to complete).  -stable does not have either patch, but has a similar
bug in the original spec_fsync() code which is triggered by a bug in the
softupdates umount code, a fix for which will be committed to -current
as soon as Kirk stamps it.  Then both solutions will be MFC'd to -stable.

-stable currently suffers from a combination of the softupdates bug and
a small window of opportunity in the original spec_fsync() code, and -stable
also suffers from the spin-loop bug but since interrupts are enabled the
spin resolves itself in a few milliseconds.

23 years agochange 'count eisa' to 'optional eisa' and update the only consumer
peter [Mon, 29 Jan 2001 08:19:02 +0000 (08:19 +0000)]
change 'count eisa' to 'optional eisa' and update the only consumer
of 'NEISA' - userconfig.c.
While there, send some defunct code to the file history.

23 years agoAdd static dependency to libisc to get isc_movefile().
asmodai [Mon, 29 Jan 2001 07:56:34 +0000 (07:56 +0000)]
Add static dependency to libisc to get isc_movefile().

Submitted by: Munehiro Matsuda <haro@kubota.co.jp>

23 years agoFix a braino in ccd's clone routine.
phk [Mon, 29 Jan 2001 06:18:14 +0000 (06:18 +0000)]
Fix a braino in ccd's clone routine.

Submitted by: tegge

23 years agoBack out rev 1.102. The IBM DNES drives work fine for me, jedgar and other
ken [Mon, 29 Jan 2001 05:33:14 +0000 (05:33 +0000)]
Back out rev 1.102.  The IBM DNES drives work fine for me, jedgar and other
folks.

My guess is that reducing the number of tags is just masking the real
problem for the PR submitter.  I'll re-open the PR and see if I can work
with the submitter to diagnose the problem.

PR: 21139

23 years ago_exit in libc is now __sys_exit not __sys__exit.
deischen [Mon, 29 Jan 2001 03:24:23 +0000 (03:24 +0000)]
_exit in libc is now __sys_exit not __sys__exit.

Add another check for thread library initialization (jdp, we
really need a way to get _thread_init called at program start
before any constructors are run).

23 years agoClean up syscall generation in libc by removing HIDDEN_SYSCALLS
deischen [Mon, 29 Jan 2001 03:23:46 +0000 (03:23 +0000)]
Clean up syscall generation in libc by removing HIDDEN_SYSCALLS
and treating (almost) all system calls the same way:

__sys_foo - actual syscall
foo, _foo - weak definitions to __sys_foo

Change PSEUDO syscalls (currently only _exit and _getlogin) to
be __sys_foo (T) and _foo (W).

Add $FreeBSD$ to a few files to satisfy commitprep.

Suggested by: bde

23 years agoDon't hard-code alignment and data declarations valid for 64-bit
marcel [Mon, 29 Jan 2001 01:55:54 +0000 (01:55 +0000)]
Don't hard-code alignment and data declarations valid for 64-bit
machines (duh!). This was one reason why this script broke on
i386. The other being that on i386 sections did not have the
proper alignment. This has been fixed in sys/sys/linker_set.h.

23 years agoo Add explicit alignment to linker set definitions. On i386, data
marcel [Mon, 29 Jan 2001 01:46:50 +0000 (01:46 +0000)]
o  Add explicit alignment to linker set definitions. On i386, data
   declarations will not be aligned by default.
o  Remove the alignment work-around for alpha. Our current alpha
   as(1) does not assume alignment after section switching, nor
   does the ia64 as(1).

23 years agoOnly remove socket files with ``set server open''.
brian [Mon, 29 Jan 2001 01:35:06 +0000 (01:35 +0000)]
Only remove socket files with ``set server open''.
Only show the mask in ``show bundle'' when it's been specified.
Complain about unexpected arguments after ``set server {none,open,closed}''
Log re-open failures as warnings rather than phase messages.
Fix some markup for the ``set server'' man page description.

23 years agoIgnore leading witespace in the string given to PacketAliasProxyRule().
brian [Mon, 29 Jan 2001 00:30:01 +0000 (00:30 +0000)]
Ignore leading witespace in the string given to PacketAliasProxyRule().

23 years agoAdd static dependency on libisc.a to get isc_movefile() on which named
asmodai [Sun, 28 Jan 2001 23:21:00 +0000 (23:21 +0000)]
Add static dependency on libisc.a to get isc_movefile() on which named
now depends.  This keeps named the same as before the import, that is: only
linking against libc dynamically, at a little space increase, which might
be due to the source code changes anyway.  Very neglectable space
difference.

Some people might dub it a hack.  It will do for now at least.

23 years agoUpdate Makefile for the new sources.
asmodai [Sun, 28 Jan 2001 23:14:16 +0000 (23:14 +0000)]
Update Makefile for the new sources.

23 years agoResolve conflicts after import.
asmodai [Sun, 28 Jan 2001 23:06:48 +0000 (23:06 +0000)]
Resolve conflicts after import.

23 years agoThis commit was generated by cvs2svn to compensate for changes in r71756,
asmodai [Sun, 28 Jan 2001 23:00:22 +0000 (23:00 +0000)]
This commit was generated by cvs2svn to compensate for changes in r71756,
which included commits to RCS files with non-trunk default branches.

23 years agoVirgin import of BIND v8.2.3-REL
asmodai [Sun, 28 Jan 2001 23:00:22 +0000 (23:00 +0000)]
Virgin import of BIND v8.2.3-REL

23 years agoDon't pass PacketAliasProxyRule() a buffer with leading whitespace as it
brian [Sun, 28 Jan 2001 21:56:27 +0000 (21:56 +0000)]
Don't pass PacketAliasProxyRule() a buffer with leading whitespace as it
can't handle it.

23 years agoCall trimdomain properly for ip4 addresses.
brian [Sun, 28 Jan 2001 21:51:25 +0000 (21:51 +0000)]
Call trimdomain properly for ip4 addresses.

PR: 24659
realhostname_sa() stuff submitted by: Jim.Pirzyk@disney.com

23 years agoFix region code reporting.
ken [Sun, 28 Jan 2001 21:43:17 +0000 (21:43 +0000)]
Fix region code reporting.

Reported by: Andrew Gordon <arg@arg1.demon.co.uk>

23 years agoRevert previous commit. I messed up my testing.
marcel [Sun, 28 Jan 2001 21:24:05 +0000 (21:24 +0000)]
Revert previous commit. I messed up my testing.

23 years agoUse \a instead of \007 for making noise.
phk [Sun, 28 Jan 2001 21:21:37 +0000 (21:21 +0000)]
Use \a instead of \007 for making noise.

Submitted by: "Jason Smethers" <jsmethers@pdq.net>

23 years agoRemove devstat entries in mddelete()
phk [Sun, 28 Jan 2001 20:55:55 +0000 (20:55 +0000)]
Remove devstat entries in mddelete()

Spotted: tegge

23 years agoDuh, the version of mdconfig.c I committed came from the wrong machine.
phk [Sun, 28 Jan 2001 20:17:46 +0000 (20:17 +0000)]
Duh, the version of mdconfig.c I committed came from the wrong machine.

Commit the right version, but without the -b option which is too evil
for the present.

Remove -b and preload from the manpage as well.

23 years agoRevert part of previous commit and initialize locale_nroff to " -Tascii"
des [Sun, 28 Jan 2001 20:17:14 +0000 (20:17 +0000)]
Revert part of previous commit and initialize locale_nroff to " -Tascii"
so man(1) works properly when no locale is set.

Spotted by: bde

23 years agoDon't specify fragment size, it autosizes from the sectorsize.
phk [Sun, 28 Jan 2001 20:03:02 +0000 (20:03 +0000)]
Don't specify fragment size, it autosizes from the sectorsize.

Remember to set mode 1777 on /tmp

Submitted by: mjacob

23 years agoA couple of chip errata work-arounds refined:
groudier [Sun, 28 Jan 2001 19:58:21 +0000 (19:58 +0000)]
A couple of chip errata work-arounds refined:
- When used on a 33MHz PCI BUS, the 53C1010-66 revision 0
  requires extra clocks to be inserted in data out phase.
  Revision 1 is fixed.
- The 53C1010-33 revision 1 requires internal cycles to be
  disabled due to possible contentions on IO registers.
  Revision 2 is fixed.
Fix:
- The probing of HVD from GPIO3 bit by the driver was reversed.
  The driver could misprobe the bus mode of a 825 or 875 chip
  that was not previously initialized (no BIOS for example).

23 years agoRemove an outdated DEVFS non-description.
phk [Sun, 28 Jan 2001 19:32:23 +0000 (19:32 +0000)]
Remove an outdated DEVFS non-description.

23 years agoDocument which options to use for API UP1[01]00
wilko [Sun, 28 Jan 2001 19:21:46 +0000 (19:21 +0000)]
Document which options to use for API UP1[01]00

23 years agoClear intr_nesting_level when an interrupt thread has no more
jake [Sun, 28 Jan 2001 17:20:11 +0000 (17:20 +0000)]
Clear intr_nesting_level when an interrupt thread has no more
handlers and wants to exit, so it doesn't panic in exit1()
which malloc()s with M_WAITOK.

Reported by: Bob Bishop <rb@gid.co.uk>

23 years agoAdd a new distribution algorythm to the 'one2many' node type.
julian [Sun, 28 Jan 2001 15:37:06 +0000 (15:37 +0000)]
Add a new distribution algorythm to the 'one2many' node type.
The new method is 'flood' (in addition to the old round-robin)
in which incoming packets are sent to more than one outgoing hook.
(I'm not sure what Rogier is using this for but it seems generally useful
and isn't much extra)

Submitted by:   Rogier R. Mulhuijzen (drwilco@drwilco.net )

23 years agoMerged from sys/i386/isa/npx.c revisions 1.88 and 1.89.
kato [Sun, 28 Jan 2001 12:31:03 +0000 (12:31 +0000)]
Merged from sys/i386/isa/npx.c revisions 1.88 and 1.89.

23 years agoSynced with sys/i386/i386/machdep.c revision 1.436.
kato [Sun, 28 Jan 2001 11:06:28 +0000 (11:06 +0000)]
Synced with sys/i386/i386/machdep.c revision 1.436.

23 years agoAPI UP1100 mainboard has been verified to work.
wilko [Sun, 28 Jan 2001 10:09:17 +0000 (10:09 +0000)]
API UP1100 mainboard has been verified to work.

23 years agoFinish cleaning up the error reporting (standardize on err*()/warn*())
kris [Sun, 28 Jan 2001 09:20:47 +0000 (09:20 +0000)]
Finish cleaning up the error reporting (standardize on err*()/warn*())
and fix some overflows.

Submitted by:   Mike Heffner <mheffner@vt.edu>

23 years agoAdd gd_witness_spin_check.
marcel [Sun, 28 Jan 2001 08:06:50 +0000 (08:06 +0000)]
Add gd_witness_spin_check.

23 years agoFix typo.
marcel [Sun, 28 Jan 2001 08:05:55 +0000 (08:05 +0000)]
Fix typo.

23 years agoImprove kernel bootstrapping:
marcel [Sun, 28 Jan 2001 06:39:56 +0000 (06:39 +0000)]
Improve kernel bootstrapping:
o  Use objdump instead of gensetdefs(1) to build the linker sets.
o  Allow overriding of nm and objdump in resp. genassym.sh and
   gensetdefs.pl for non-native toolchains.

Reviewed by: arch
Perl improvements: Jos Backus <josb@cncdsl.com>, benno

23 years agoMove the setting of curproc to idleproc up earlier in ap_init(). The
bmilekic [Sun, 28 Jan 2001 03:41:01 +0000 (03:41 +0000)]
Move the setting of curproc to idleproc up earlier in ap_init(). The
problem is that a mutex lock, prior to this change, is acquired before
the curproc is set to idleproc, so we mess ourselves up by calling
the mutex lock routine with curproc == NULL.

Moving it up after the aps_ready spin-wait has us hopefully setting it
after idleproc is setup.

Solved by: jake (the allmighty) :-)

23 years agoDefer assignment of low level interrupt handlers for PCI interrupts
tegge [Sun, 28 Jan 2001 01:07:54 +0000 (01:07 +0000)]
Defer assignment of low level interrupt handlers for PCI interrupts
described in the MP table until something asks for the interrupt number
later on.