]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoAdd cross-reference to loader man page, plus various minor
rnordier [Tue, 16 Mar 1999 13:16:11 +0000 (13:16 +0000)]
Add cross-reference to loader man page, plus various minor
improvements.

25 years agoAdd majors for new ATA/ATAPI driver.
sos [Tue, 16 Mar 1999 12:50:02 +0000 (12:50 +0000)]
Add majors for new ATA/ATAPI driver.

25 years agoFix a dummynet bug caused by passing a bad next hop address (the
luigi [Tue, 16 Mar 1999 12:06:11 +0000 (12:06 +0000)]
Fix a dummynet bug caused by passing a bad next hop address (the
symptom was the msg "arp failure -- host is not on local network" that
some user have seen on multihomed machines.
Bug tracked down by Emmanuel Duros

25 years agoMention changes to the default authentication behaviour.
brian [Tue, 16 Mar 1999 11:54:02 +0000 (11:54 +0000)]
Mention changes to the default authentication behaviour.

25 years agoMinor adjustments to make LINT compile again (thanks to phk for reminding me).
hm [Tue, 16 Mar 1999 11:07:04 +0000 (11:07 +0000)]
Minor adjustments to make LINT compile again (thanks to phk for reminding me).
Because the Fritz!Card PCMCIA stuff is also affected, this includes the fix
from PR kern/10278 submitted by Udo Schweigert (udo.schweigert@mchp.siemens.de).
PR: kern/10278

25 years agoDon't run FLL fodder through the median-filter.
phk [Tue, 16 Mar 1999 08:39:37 +0000 (08:39 +0000)]
Don't run FLL fodder through the median-filter.

Reduce max integration time to 128sec and use 50% exponential decay rather
than 256sec/25%.

25 years agoCorrect a problem in kmem_malloc: A kmem_malloc allowing "wait" may
alc [Tue, 16 Mar 1999 07:39:07 +0000 (07:39 +0000)]
Correct a problem in kmem_malloc: A kmem_malloc allowing "wait" may
block (VM_WAIT) holding the map lock.  This is bad.  For example, a subsequent
kmem_malloc by an interrupt handler on the same map may find the lock held
and panic in the lockmgr.

25 years agoDetect pred1 packets where the length != the packet length
brian [Tue, 16 Mar 1999 01:24:23 +0000 (01:24 +0000)]
Detect pred1 packets where the length != the packet length
Use a maximum of MAX_MRU + 2 bytes for incoming packets, not
MAX_MTU + 2.
Tidy up some diagnostics.

25 years agoHost names are case-insensitive.
brian [Tue, 16 Mar 1999 01:23:09 +0000 (01:23 +0000)]
Host names are case-insensitive.

25 years agoAdd "-fkeep-inline-functions" to CFLAGS so that higher optimization
jdp [Mon, 15 Mar 1999 21:56:54 +0000 (21:56 +0000)]
Add "-fkeep-inline-functions" to CFLAGS so that higher optimization
levels (-O3 and above) won't remove essential code.  Many thanks
to Dmitrij Tejblum <dima@tejblum.dnttm.rssi.ru> for pointing out
that it was the optimizer's removal of this code that caused make
world with -O3 to break.  With this change, make buildworld now
completes.

25 years agoSpelling and grammar fixes.
ghelmer [Mon, 15 Mar 1999 16:01:22 +0000 (16:01 +0000)]
Spelling and grammar fixes.

PR: docs/10401

25 years agoFix reference (FreeBSD 3.0.1 -> FreeBSD 3.1), remove apparent typo,
ghelmer [Mon, 15 Mar 1999 15:43:10 +0000 (15:43 +0000)]
Fix reference (FreeBSD 3.0.1 -> FreeBSD 3.1), remove apparent typo,
and fix reference to sysctl(8).

PR: docs/10428 docs/10482

25 years agoReference the sysctl variables related to various limits.
ghelmer [Mon, 15 Mar 1999 15:36:56 +0000 (15:36 +0000)]
Reference the sysctl variables related to various limits.

PR: docs/6764

25 years agoAdd import of ISC's DHCP client & the upgrade of Groff.
obrien [Mon, 15 Mar 1999 09:41:22 +0000 (09:41 +0000)]
Add import of ISC's DHCP client & the upgrade of Groff.

25 years agoFix conflicts.
obrien [Mon, 15 Mar 1999 09:18:28 +0000 (09:18 +0000)]
Fix conflicts.

25 years agoThe man page, version 0.1.
dcs [Mon, 15 Mar 1999 08:52:23 +0000 (08:52 +0000)]
The man page, version 0.1.

25 years agoThis commit was generated by cvs2svn to compensate for changes in r44777,
obrien [Mon, 15 Mar 1999 08:36:16 +0000 (08:36 +0000)]
This commit was generated by cvs2svn to compensate for changes in r44777,
which included commits to RCS files with non-trunk default branches.

25 years agoVirgin import of FSF groff v1.11
obrien [Mon, 15 Mar 1999 08:36:16 +0000 (08:36 +0000)]
Virgin import of FSF groff v1.11

25 years agoAllow !suser() R/O access to ntp_adjtime()
phk [Mon, 15 Mar 1999 08:35:40 +0000 (08:35 +0000)]
Allow !suser() R/O access to ntp_adjtime()

Noticed by: Reg Clemens <reg@dwf.com>

25 years agoFix date parsing to allow '0' (none) date value.
davidn [Mon, 15 Mar 1999 08:16:01 +0000 (08:16 +0000)]
Fix date parsing to allow '0' (none) date value.

25 years agoRemove redundant code.
dcs [Mon, 15 Mar 1999 08:08:20 +0000 (08:08 +0000)]
Remove redundant code.

25 years agoTwo changes:
alc [Mon, 15 Mar 1999 06:24:52 +0000 (06:24 +0000)]
Two changes:

In general, vm_map_simplify_entry should be performed INSIDE
the loop that traverses the map, not outside.  (Changed:
vm_map_inherit, vm_map_pageable.)

vm_fault_unwire doesn't acquire the map lock (or block holding
it).  Thus, vm_map_set/clear_recursive shouldn't be called.
(Changed: vm_map_user_pageable, vm_map_pageable.)

25 years agofix breakage for alphas.
julian [Mon, 15 Mar 1999 05:11:27 +0000 (05:11 +0000)]
fix breakage for alphas.
Submitted by: Andrew Gallatin <gallatin@cs.duke.edu>

25 years agoFix breakage in last commit
julian [Mon, 15 Mar 1999 05:09:48 +0000 (05:09 +0000)]
Fix breakage in last commit
Submitted by: Brian Feldman <green@unixhelp.org>

25 years agoGrammar nits, double negatives
billf [Mon, 15 Mar 1999 02:57:29 +0000 (02:57 +0000)]
Grammar nits, double negatives

PR: docs/10491
Submitted By: Tom Hukins <tom@eborcom.com>

25 years ago1) line 76: "fine" should be "file"
billf [Mon, 15 Mar 1999 02:49:13 +0000 (02:49 +0000)]
1) line 76: "fine" should be "file"
2) line 80: "abso" should be "also"

PR: docs/10496
Submitted by: Osamu Mihara <mihara@prd.fc.nec.co.jp>

25 years agoTypo - latter -> letter
billf [Mon, 15 Mar 1999 02:34:40 +0000 (02:34 +0000)]
Typo - latter -> letter

PR: docs/10591
Submitted by: Kanenda <vanitas@ma3.seikyou.ne.jp>

25 years agoGrrr... botched remote commit. Let's try this again: vlan updates,
wpaul [Mon, 15 Mar 1999 01:22:01 +0000 (01:22 +0000)]
Grrr... botched remote commit. Let's try this again: vlan updates,
take two.

25 years agoUpdates for vlan stuff:
wpaul [Mon, 15 Mar 1999 01:17:26 +0000 (01:17 +0000)]
Updates for vlan stuff:

- add support for devices that do vlan tag insertion/deletion in firmware
- add multicast support
- add vlan_unconfig() to complement vlan_config()
- update ifconfig(8) to configure vlan interfaces (vlan tag and
  parent device)

Also fix a small bug in ifconfig; sometimes sa_family is overwritten
by ioctls.

Reviewed by: wollman

25 years agoThis is a temporary workaround for kernel panic occurring when
yokota [Mon, 15 Mar 1999 00:20:56 +0000 (00:20 +0000)]
This is a temporary workaround for kernel panic occurring when
switching video modes.  People typically see the panic when the screen
saver has been running and is being stopped.

WARNING: this is not a real fix.  Something really funny must be
happening in the page table entries for the physical memory
0x0-0xa0000, which is mapped to the beginning of the kernel region.

25 years agoEACESS -> EACCES
alex [Mon, 15 Mar 1999 00:14:57 +0000 (00:14 +0000)]
EACESS -> EACCES

25 years agoEACESS -> EACCES
alex [Mon, 15 Mar 1999 00:14:16 +0000 (00:14 +0000)]
EACESS -> EACCES

Submitted by: garath@code.ridgefield.org via OpenBSD

25 years agoReplace /kernel.config with /boot/kernel.conf.
dcs [Sun, 14 Mar 1999 21:57:49 +0000 (21:57 +0000)]
Replace /kernel.config with /boot/kernel.conf.

25 years agoNew names for the list; the DES crypt and libwrap for tcp_wrappers.
markm [Sun, 14 Mar 1999 20:45:05 +0000 (20:45 +0000)]
New names for the list; the DES crypt and libwrap for tcp_wrappers.

25 years agoquick BDE premption
julian [Sun, 14 Mar 1999 20:43:09 +0000 (20:43 +0000)]
quick BDE premption

25 years agoBitrot. Remove objlink as it is not pleasant to be downwind.
markm [Sun, 14 Mar 1999 20:41:14 +0000 (20:41 +0000)]
Bitrot. Remove objlink as it is not pleasant to be downwind.

PR: 8071
Reviewed by: Sheldon Hearn <sheldonh@iafrica.com>

25 years ago A bit of a hack, but allows the vn device to be a module again.
julian [Sun, 14 Mar 1999 20:40:15 +0000 (20:40 +0000)]
 A bit of a hack, but allows the vn device to be a module again.

Submitted by: Matt Dillon <dillon@freebsd.org>

25 years agoFix bogons in previous commit.
des [Sun, 14 Mar 1999 20:26:39 +0000 (20:26 +0000)]
Fix bogons in previous commit.

25 years agoFix syntax for boot command
msmith [Sun, 14 Mar 1999 19:55:01 +0000 (19:55 +0000)]
Fix syntax for boot command

Noticed by: Ruslan Ermilov <ru@ucb.crimea.ua>

25 years agoBring in some MASTER_SITE_{AFTERSTEP,WINDOWMAKER} from bsd.port.mk
billf [Sun, 14 Mar 1999 19:05:16 +0000 (19:05 +0000)]
Bring in some MASTER_SITE_{AFTERSTEP,WINDOWMAKER} from bsd.port.mk

25 years agoBuild tcp_wrappers' userland. I am not building tcpd, because in a day
markm [Sun, 14 Mar 1999 18:02:14 +0000 (18:02 +0000)]
Build tcp_wrappers' userland. I am not building tcpd, because in a day
or two, inetd will gain the necessary functionality. At that stage,
I'll make wrapping the default for sendmail and portmapper as well.

25 years agoBuild libwrap - the core to tcp_wrappers.
markm [Sun, 14 Mar 1999 17:56:11 +0000 (17:56 +0000)]
Build libwrap - the core to tcp_wrappers.

25 years agoThis commit was generated by cvs2svn to compensate for changes in r44743,
markm [Sun, 14 Mar 1999 17:13:19 +0000 (17:13 +0000)]
This commit was generated by cvs2svn to compensate for changes in r44743,
which included commits to RCS files with non-trunk default branches.

25 years agoClean import of TCP-wrappers by Wietse Venema.
markm [Sun, 14 Mar 1999 17:13:19 +0000 (17:13 +0000)]
Clean import of TCP-wrappers by Wietse Venema.
Rest of build to follow.

25 years agoSubmitted by: Matt Dillon <dillon@freebsd.org>
julian [Sun, 14 Mar 1999 09:20:01 +0000 (09:20 +0000)]
Submitted by: Matt Dillon <dillon@freebsd.org>
The old VN device broke in -4.x when the definition of B_PAGING
changed. This patch fixes this plus implements additional capabilities.
The new VN device can be backed by a file ( as per normal ), or it can
be directly backed by swap.

Due to dependencies in VM include files  (on opt_xxx options) the new
vn device cannot be a module yet. This will be fixed in a later commit.
This commit delimitted by tags {PRE,POST}_MATT_VNDEV

25 years agoMerge NetBSD 1.80->1.82 changes from vendor branch into mainline.
peter [Sun, 14 Mar 1999 08:32:52 +0000 (08:32 +0000)]
Merge NetBSD 1.80->1.82 changes from vendor branch into mainline.

25 years agoImplemented workaround for EPIC's Application Note 7-15 (concerning
semenu [Sun, 14 Mar 1999 08:30:23 +0000 (08:30 +0000)]
Implemented workaround for EPIC's Application Note 7-15 (concerning
chip int. and ext. clock synchronisation). Fixed workaround for
transmit threshold underrun. Added volatile keyword to CSR_READ_* and
CSR_WRITE_* macroses. Added DELAYs to eliminate randomness caused
by processor speed. Fixed all TXCON and RXCON registers to be accessed
only when chip is idle, as manual told. Changed epic_init_phy to
drop link by isolating and going loopback, should should force link
partner to restart autonegotiation.

PR: kern/10535, kern/9742, kern/10575
Submitted by: Peter Jeremy, David Greenman

25 years agoCorrect two optimization errors in vm_object_page_remove:
alc [Sun, 14 Mar 1999 06:36:00 +0000 (06:36 +0000)]
Correct two optimization errors in vm_object_page_remove:

1. The size of vm_object::memq is vm_object::resident_page_count,
not vm_object::size.

2. The "size > 4" test sometimes results in the traversal of a ~1000 page
memq in order to locate ~10 pages.

25 years agoDisable tagged queueing for the IBM DCAS drives. These drives have poor
ken [Sun, 14 Mar 1999 05:15:38 +0000 (05:15 +0000)]
Disable tagged queueing for the IBM DCAS drives.  These drives have poor
write performance when tagged queueing is enabled.

Although the PR was submitted for the 4 gig version of this drive, the
assumption is that the 2 gig version has the same problem.  Therefore
tagged queueing is disabled for both.

Also, update the comment for the Western Digital Enterprise drives to note
that the best performance for those drives is achieved when tagged queueing
is disabled and write caching is enabled.

PR: kern/10398
Submitted by: Hideaki Okada <hokada@isl.melco.co.jp>

25 years agoClarify that the stop command only works in interactive mode.
grog [Sun, 14 Mar 1999 04:06:49 +0000 (04:06 +0000)]
Clarify that the stop command only works in interactive mode.

25 years agocreate command: get the message right if an error occurs with a
grog [Sun, 14 Mar 1999 04:06:00 +0000 (04:06 +0000)]
create command: get the message right if an error occurs with a
non-verbose create.

25 years agodeallocrqg: deallocate the correct object, also check that the object
grog [Sun, 14 Mar 1999 04:04:23 +0000 (04:04 +0000)]
deallocrqg: deallocate the correct object, also check that the object
is really in the chain.

25 years agoChown the pty device nodes back to root:tty.
des [Sun, 14 Mar 1999 03:20:22 +0000 (03:20 +0000)]
Chown the pty device nodes back to root:tty.

25 years agoEnforce monotonicity of apparent process user, system and interrupt times.
bde [Sat, 13 Mar 1999 19:46:13 +0000 (19:46 +0000)]
Enforce monotonicity of apparent process user, system and interrupt times.

PR: 975, 10402

25 years agoInstall uuxqt sgid uucp. Since "rnews" should not be executable for
torstenb [Sat, 13 Mar 1999 17:35:18 +0000 (17:35 +0000)]
Install uuxqt sgid uucp. Since "rnews" should not be executable for
everyone (to prevent fake-postings) and modern Inn installs
rnews as news.uucp and mode 4550 this is the only save way to allow
uuxqt to process rnews batches.

25 years agoReduce Bruce-filter verbosity.
phk [Sat, 13 Mar 1999 16:22:18 +0000 (16:22 +0000)]
Reduce Bruce-filter verbosity.

25 years agoMake NDGBPORTS an official option.
joerg [Sat, 13 Mar 1999 13:20:59 +0000 (13:20 +0000)]
Make NDGBPORTS an official option.

25 years agoSync with sys/i386/conf/Makefile.i386 revision 1.141.
kato [Sat, 13 Mar 1999 13:18:00 +0000 (13:18 +0000)]
Sync with sys/i386/conf/Makefile.i386 revision 1.141.

25 years agoset if_snd.ifq_maxlen to something reasonable - note that if_de.c has
peter [Sat, 13 Mar 1999 09:21:27 +0000 (09:21 +0000)]
set if_snd.ifq_maxlen to something reasonable - note that if_de.c has
a wierd double-queue arrangement..  It always empties the if_snd queue
then puts the transmit packets into a different queue that is limited
by the number of TX descriptors and does it's own discards...
This should stop the boot-time XXX warning anyway.

25 years agoClarify the undesirability of static configuration.
grog [Sat, 13 Mar 1999 07:49:52 +0000 (07:49 +0000)]
Clarify the undesirability of static configuration.
Describe the problems of synchronization between userland and driver.

Tripped-over-by: Chris Foote <chris@senet.com.au>
25 years agoChange prototype of vinum_inactive.
grog [Sat, 13 Mar 1999 07:46:34 +0000 (07:46 +0000)]
Change prototype of vinum_inactive.

25 years agoDescribe the new debug flags DEBUG_EOFINFO and DEBUG_MEMFREE.
grog [Sat, 13 Mar 1999 07:45:52 +0000 (07:45 +0000)]
Describe the new debug flags DEBUG_EOFINFO and DEBUG_MEMFREE.

Fix a confusion between slices and partitions.

Clarify the problems of small stripe sizes (< 128 kB).

25 years agoinfo command: Remove opencount parameter, print value of open flag.
grog [Sat, 13 Mar 1999 07:44:17 +0000 (07:44 +0000)]
info command: Remove opencount parameter, print value of open flag.

25 years agocreate command: if an error occurs, print the offending line even if
grog [Sat, 13 Mar 1999 07:43:28 +0000 (07:43 +0000)]
create command: if an error occurs, print the offending line even if
the -v flag has not been set.

25 years agoImprove the manner in which the location of the kld is found. It
grog [Sat, 13 Mar 1999 07:42:15 +0000 (07:42 +0000)]
Improve the manner in which the location of the kld is found.  It
should now be completely automatic.

25 years agoRemove opencount from vinum_conf and struct vol, and disk index from
grog [Sat, 13 Mar 1999 07:40:52 +0000 (07:40 +0000)]
Remove opencount from vinum_conf and struct vol, and disk index from
struct vol.

Add time to struct mc, remove KVA crud.

Add debug flags DEBUG_EOFINFO and DEBUG_MEMFREE.

25 years agoPart of removing vol->opencount: now we just use a flag.
grog [Sat, 13 Mar 1999 07:39:17 +0000 (07:39 +0000)]
Part of removing vol->opencount: now we just use a flag.

25 years agoAdd DEBUG_EOFINFO log messages.
grog [Sat, 13 Mar 1999 07:38:27 +0000 (07:38 +0000)]
Add DEBUG_EOFINFO log messages.

25 years agoRemove antique crud.
grog [Sat, 13 Mar 1999 07:37:44 +0000 (07:37 +0000)]
Remove antique crud.

MMalloc: save the time at which the request was granted, remove more
crud.

FFree: add a circular buffer of the last 64 Free requests if
DEBUG_MEMFREE is set.

25 years agolock_drive: report on and recover from an obscure situation where the
grog [Sat, 13 Mar 1999 07:35:40 +0000 (07:35 +0000)]
lock_drive: report on and recover from an obscure situation where the
daemon can deadlock against itself.

25 years agoVINUM_RESETCONFIG: do the reset even if the config is open (as it has
grog [Sat, 13 Mar 1999 07:34:56 +0000 (07:34 +0000)]
VINUM_RESETCONFIG: do the reset even if the config is open (as it has
to be for us to issue the ioctl :-).  Also remove the opencount
parameter of the config, which just never worked.

25 years agoDon't frob vp->v_usecount, it scares the hell out of the driver under
grog [Sat, 13 Mar 1999 07:33:37 +0000 (07:33 +0000)]
Don't frob vp->v_usecount, it scares the hell out of the driver under
certain circumstances.

25 years agoChange the way of recognizing whether a plex was given to a volume
grog [Sat, 13 Mar 1999 07:32:40 +0000 (07:32 +0000)]
Change the way of recognizing whether a plex was given to a volume
after the volume had been fully operational; involves a change in the
use of the VF_NEWBORN flag.  Now if you add a plex to a volume which
is up, the plex will be down and the subdisks stale.  You need to
explicitly start the subdisks, which copies data from the good
subdisks to the uninitialized ones.

Stumbled-over-by: Ludwig Pummer <ludwigp@bigfoot.com>
give_sd_to_drive:
   correct method to give the entire largest chunk of drive to the
   subdisk.  Now it's enough to specify a length, and vinum will give
   you as much as it can.  Not to be recommended except for empty
   drives.

   Correct a bogon which made vinum refuse to give the last sector of
   a drive to a subdisk.

Last-reported-by: Ludwig Pummer <ludwigp@bigfoot.com>
Change %q formats to %ll before the former go away.  This doesn't make
much difference, since kernel kvprintf currently doesn't support
either, and the messages in question are just error messages.

25 years agopmap_qenter/pmap_qremove:
alc [Sat, 13 Mar 1999 07:31:29 +0000 (07:31 +0000)]
pmap_qenter/pmap_qremove:
Use the pmap_kenter/pmap_kremove inline functions
instead of duplicating them.

pmap_remove_all:
Eliminate an unused (but initialized) variable.

pmap_ts_reference:
Change the implementation.  The new implementation is much smaller
and simpler, but functionally identical.  (Reviewed by
"John S. Dyson" <dyson@iquest.net>.)

25 years agovinum_inactive: distinguish whether we care whether the config
grog [Sat, 13 Mar 1999 07:22:00 +0000 (07:22 +0000)]
vinum_inactive: distinguish whether we care whether the config
is open or not.

25 years agoAdd a dire warning about the folly of configuring vinum in
grog [Sat, 13 Mar 1999 06:32:06 +0000 (06:32 +0000)]
Add a dire warning about the folly of configuring vinum in
the kernel.

25 years agoWhen printing out V1 info, make sure that there is additional info
imp [Sat, 13 Mar 1999 04:41:35 +0000 (04:41 +0000)]
When printing out V1 info, make sure that there is additional info
before printing it.  Terminate when we come to a 0xff byte.  This
allows there to be zero or more additional info fields printed
correctly.  Before, the old code would print bogons or dump core when
presented with this case.

I don't know what the spec says about this, exactly, but this allows
me to do a dumpcis of my non-ATA AMP 4M FLASH cards w/o pccardc
dumping core.

25 years agoNote that a standard MBR supports booting only from hard drive 0
rnordier [Sat, 13 Mar 1999 01:40:57 +0000 (01:40 +0000)]
Note that a standard MBR supports booting only from hard drive 0

25 years agoReverse the order of processing object files within an executable or
jdp [Sat, 13 Mar 1999 01:35:44 +0000 (01:35 +0000)]
Reverse the order of processing object files within an executable or
shared library when invoking global constructors and destructors.
For constructors, the object files used to be processed from first
to last; now they're done from last to first.  (Destructors are done
in the opposite order, as required by the C++ standard.)  This makes
us consistent with standard gcc and egcs compilers.  It also
eliminates ordering differences between dynamic and static
executables.

Bump the value of __FreeBSD_version to 400002 to reflect this
change.

25 years agoUpdate to post-branch state (yeesh!)
jkh [Fri, 12 Mar 1999 22:45:42 +0000 (22:45 +0000)]
Update to post-branch state (yeesh!)

25 years agoFix an old cut&paste bogon.
phk [Fri, 12 Mar 1999 21:58:54 +0000 (21:58 +0000)]
Fix an old cut&paste bogon.

Noticed by: bde

25 years agoDon't allow anyone except root to mount file systems that map uids.
imp [Fri, 12 Mar 1999 19:40:39 +0000 (19:40 +0000)]
Don't allow anyone except root to mount file systems that map uids.
This can have bad security implications, but the impact on FreeBSD
systems is minimal because this fs isn't in the default kernels and it
is unknown if it even works.

Submitted by: Manuel Bouyer <bouyer@antioche.eu.org> and
Artur Grabowski <art@stacken.kth.se>

25 years agoRemove static delcarations from {,un}register_pcic_intr declarations
imp [Fri, 12 Mar 1999 18:45:47 +0000 (18:45 +0000)]
Remove static delcarations from {,un}register_pcic_intr declarations
in the not APIC_IO case.  This should silence the warnings when
building pcic.c as well as the undefined functions when the kernel
links.

Noticed by: several people in -current and me building the kernel

25 years agoMove the code for the ".init" and ".fini" sections outside of a
jdp [Fri, 12 Mar 1999 17:33:28 +0000 (17:33 +0000)]
Move the code for the ".init" and ".fini" sections outside of a
C function so the compiler won't try to emit line numbers for it
with "-g", breaking the build.  This has the nice side-effect of
making crtbegin.o and crtbeginS.o a little bit smaller.

Remove "-Wno-unused" from the Makefile.  Replace it with "__unused"
on particular function and variable declarations.

25 years agosrc/usr.sbin/natd -> src/sbin/natd (after a repo-copy by jdp)
brian [Fri, 12 Mar 1999 15:38:55 +0000 (15:38 +0000)]
src/usr.sbin/natd -> src/sbin/natd (after a repo-copy by jdp)

25 years agomake libkvm capable of dealing with a crashdump on alphas
gallatin [Fri, 12 Mar 1999 14:47:33 +0000 (14:47 +0000)]
make libkvm capable of dealing with a crashdump on alphas
Reviewed by: Doug Rabson <dfr@nlsystems.com>

25 years agoFix savecore so that it works on alphas
gallatin [Fri, 12 Mar 1999 14:46:00 +0000 (14:46 +0000)]
Fix savecore so that it works on alphas
Reviewed by: Doug Rabson <dfr@nlsystems.com>

25 years agoEnable crashdumps on alphas.
gallatin [Fri, 12 Mar 1999 14:44:46 +0000 (14:44 +0000)]
Enable crashdumps on alphas.
Reviewed by: Doug Rabson <dfr@nlsystems.com>

25 years agoRemove duplicate include.
phk [Fri, 12 Mar 1999 11:09:50 +0000 (11:09 +0000)]
Remove duplicate include.

Noticed by: bde

25 years agoFix Typo.
phk [Fri, 12 Mar 1999 11:09:18 +0000 (11:09 +0000)]
Fix Typo.

25 years agoRemove vm_page_frees from kmem_malloc that are performed
alc [Fri, 12 Mar 1999 08:05:49 +0000 (08:05 +0000)]
Remove vm_page_frees from kmem_malloc that are performed
by vm_map_delete/vm_object_page_remove anyway.

25 years agoThis solves a deadlock that can occur when read()ing into a file-mmap()
julian [Fri, 12 Mar 1999 03:09:29 +0000 (03:09 +0000)]
This solves a deadlock that can occur when read()ing into a file-mmap()
space. When doing this, it is possible to for another process to attempt
to get an exclusive lock on the vnode and deadlock the mmap/read
combination when the uiomove() call tries to obtain a second
shared lock on the vnode. There is still a potential deadlock
situation with write()/mmap().
Submitted by: Matt Dillon <dillon@freebsd.org>
Reviewed by: Luoqi Chen <luoqi@freebsd.org>
Delimmitted by tag PRE_MATT_MMAP_LOCK and POST_MATT_MMAP_LOCK
in kern/kern_lock.c kern/kern_subr.c

25 years agoReviewed by: Many at differnt times in differnt parts,
julian [Fri, 12 Mar 1999 02:24:58 +0000 (02:24 +0000)]
Reviewed by: Many at differnt times in differnt parts,
including alan, john, me, luoqi, and kirk
Submitted by: Matt Dillon <dillon@frebsd.org>

This change implements a relatively sophisticated fix to getnewbuf().
There were two problems with getnewbuf(). First, the writerecursion
can lead to a system stack overflow when you have NFS and/or VN
devices in the system. Second, the free/dirty buffer accounting was
completely broken. Not only did the nfs routines blow it trying to
manually account for the buffer state, but the accounting that was
done did not work well with the purpose of their existance: figuring
out when getnewbuf() needs to sleep.

The meat of the change is to kern/vfs_bio.c. The remaining diffs are
all minor except for NFS, which includes both the fixes for bp
interaction AND fixes for a 'biodone(): buffer already done' lockup.
Sys/buf.h also contains a chaining structure which is not used by
this patchset but is used by other patches that are coming soon.
This patch deliniated by tags PRE_MAT_GETBUF and POST_MAT_GETBUF.
(sorry for the missing T matt)

25 years agoFix the 'fwd' option to ipfw when asked to divert to another machine.
julian [Fri, 12 Mar 1999 01:15:57 +0000 (01:15 +0000)]
Fix the 'fwd' option to ipfw when asked to divert to another machine.
also rely less on other modules clearing static values, and clear them
in a few cases we missed before.
Submitted by: Matthew Reimer <mreimer@vpop.net>

25 years agoStop the mfs from trying to swap out crucial bits of the mfs
julian [Fri, 12 Mar 1999 00:44:03 +0000 (00:44 +0000)]
Stop the mfs from trying to swap out crucial bits of the mfs
as this can lead to deadlock.
Submitted by: Mat dillon <dillon@freebsd.org>

25 years agoAdd support for long long modifier (e.g. %llx, %lld).
dfr [Thu, 11 Mar 1999 22:44:02 +0000 (22:44 +0000)]
Add support for long long modifier (e.g. %llx, %lld).

Reviewed by: bde

25 years agoFixed runtime accounting. The time since the previous context switch
bde [Thu, 11 Mar 1999 21:53:12 +0000 (21:53 +0000)]
Fixed runtime accounting.  The time since the previous context switch
was discarded on every call to calcru().  Hacking on the `switchtime'
global for a related fix in rev.1.38 of kern_resource.c was too fragile
and broke when p_switchtime went away.

PR: 10402

25 years agoFixed variable overflow problem.
dg [Thu, 11 Mar 1999 18:50:39 +0000 (18:50 +0000)]
Fixed variable overflow problem.

Obtained from: NetBSD via Mark J. Taylor <mtaylor@cybernet.com>

25 years agoIncreased kernel virtual address space to 1GB. NOTE: You MUST have fixed
dg [Thu, 11 Mar 1999 18:28:46 +0000 (18:28 +0000)]
Increased kernel virtual address space to 1GB. NOTE: You MUST have fixed
bootblocks in order to boot the kernel after this! Also note that this
change breaks BSDI BSD/OS compatibility.
Also increased default NKPT to 17 so that FreeBSD can boot on machines
with >=2GB of RAM. Booting on machines with exactly 4GB requires other
patches, not included.

25 years agoChange permissions on /etc/ppp/ppp.conf to 0640.
brian [Thu, 11 Mar 1999 18:22:23 +0000 (18:22 +0000)]
Change permissions on /etc/ppp/ppp.conf to 0640.