]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoGive wider types to sscanf to fix two warnings (u_short cannot be > 0xffff)
peter [Sun, 26 Oct 2003 04:36:47 +0000 (04:36 +0000)]
Give wider types to sscanf to fix two warnings (u_short cannot be > 0xffff)
and to make sure that we catch oversized arguments rather than silently
truncate them.  I dont know if sscanf will reject an integer if it will
not fit in the short return variable or not, but this way it should be
detected.

20 years agoTidy up some xdrproc_t related warnings.
peter [Sun, 26 Oct 2003 04:32:53 +0000 (04:32 +0000)]
Tidy up some xdrproc_t related warnings.

20 years agoPacify gcc warning with a Douglas Adams reference.
peter [Sun, 26 Oct 2003 04:30:05 +0000 (04:30 +0000)]
Pacify gcc warning with a Douglas Adams reference.

20 years agoFix some warnings by defining has_sa_sigaction to 1. When the configure
peter [Sun, 26 Oct 2003 04:26:52 +0000 (04:26 +0000)]
Fix some warnings by defining has_sa_sigaction to 1.  When the configure
script was run (years and years ago), we didn't have it.

20 years agoSince I'm having so much fun with const, beat my head against the wall
peter [Sun, 26 Oct 2003 04:20:19 +0000 (04:20 +0000)]
Since I'm having so much fun with const, beat my head against the wall
some more.  Make this compile cleanly.

20 years agoPointers dont fit in an int on 64 bit platforms. Fix a gcc warning.
peter [Sun, 26 Oct 2003 04:12:02 +0000 (04:12 +0000)]
Pointers dont fit in an int on 64 bit platforms.  Fix a gcc warning.

20 years agoThe third arg to strncmp() is size_t, not int. This causes a warning on
peter [Sun, 26 Oct 2003 04:10:50 +0000 (04:10 +0000)]
The third arg to strncmp() is size_t, not int.  This causes a warning on
systems where size_t is long, not int.

20 years agoPreserve the constness of the value argument passed to env_setenv() as
peter [Sun, 26 Oct 2003 04:04:12 +0000 (04:04 +0000)]
Preserve the constness of the value argument passed to env_setenv() as
it gets passed through the filter functions.

20 years agoPacify gcc about casting pointers to integers (for the lowest few bits).
peter [Sun, 26 Oct 2003 03:55:58 +0000 (03:55 +0000)]
Pacify gcc about casting pointers to integers (for the lowest few bits).

20 years ago(mostly) Clean up some const warnings here. The code takes some liberties
peter [Sun, 26 Oct 2003 03:51:47 +0000 (03:51 +0000)]
(mostly) Clean up some const warnings here.  The code takes some liberties
because it is the originator of various const strings and knows that they
came from malloc.

20 years agoclnt_call takes args of type xdrproc_t.
peter [Sun, 26 Oct 2003 03:43:35 +0000 (03:43 +0000)]
clnt_call takes args of type xdrproc_t.

20 years agoMove the -Wtraditional warning from the global cpp flags to when the -p
peter [Sun, 26 Oct 2003 03:34:13 +0000 (03:34 +0000)]
Move the -Wtraditional warning from the global cpp flags to when the -p
(portability warnings) switch is used.  Add -Wno-system-headers after it so
that we dont get 500 screenfulls of warnings about #elif in /usr/include.
I'm not entirely happy with this.  Maybe cdefs.h shouldn't use #elif and
instead nest #else clauses?

20 years agoDisable the log() builtin (the math function), because it will not do much
peter [Sun, 26 Oct 2003 03:15:54 +0000 (03:15 +0000)]
Disable the log() builtin (the math function), because it will not do much
logging of information like the program expects.

20 years agoWhen we pass a string as auxillary data (type long), be sure to convert
peter [Sun, 26 Oct 2003 03:12:47 +0000 (03:12 +0000)]
When we pass a string as auxillary data (type long), be sure to convert
it to a suitable type for the initialization.

20 years agoFix a warning about mismatched pointers. A pointer to "void *" is not the
peter [Sun, 26 Oct 2003 03:01:24 +0000 (03:01 +0000)]
Fix a warning about mismatched pointers.  A pointer to "void *" is not the
same as a pointer to "char *".  Tell the compiler this is ok.

20 years agoconstify bintime_add, bintime_sub, bintime2timespec, timespec2bintime,
alfred [Sun, 26 Oct 2003 02:38:34 +0000 (02:38 +0000)]
constify bintime_add, bintime_sub, bintime2timespec, timespec2bintime,
bintime2timeval and timeval2bintime.

20 years agoconstify the second args to timevaladd() and timevalsub().
alfred [Sun, 26 Oct 2003 02:19:00 +0000 (02:19 +0000)]
constify the second args to timevaladd() and timevalsub().

20 years agoAdd coup d'état in Chile, 1973.
grog [Sun, 26 Oct 2003 01:18:08 +0000 (01:18 +0000)]
Add coup d'état in Chile, 1973.

20 years agoConst poison string accessor functions.
imp [Sun, 26 Oct 2003 00:51:40 +0000 (00:51 +0000)]
Const poison string accessor functions.

20 years agostyle.Makefile: Add a '?' before '=' in WARNS.
trhodes [Sun, 26 Oct 2003 00:35:05 +0000 (00:35 +0000)]
style.Makefile: Add a '?' before '=' in WARNS.

20 years agoAdd end and start dates for DST.
grog [Sun, 26 Oct 2003 00:03:49 +0000 (00:03 +0000)]
Add end and start dates for DST.

20 years agoConsistently cast to (u_char *) when filling with junk.
phk [Sat, 25 Oct 2003 23:47:33 +0000 (23:47 +0000)]
Consistently cast to (u_char *) when filling with junk.

20 years ago - Add some of the required vm object locking, including assertions where
alc [Sat, 25 Oct 2003 23:42:17 +0000 (23:42 +0000)]
 - Add some of the required vm object locking, including assertions where
   the vm object lock is required and already held.

20 years agofix spelling of ATTACH
jmg [Sat, 25 Oct 2003 22:03:10 +0000 (22:03 +0000)]
fix spelling of ATTACH

Submitted by: Johny Mattsson
MFC after: 2 week
(do to code freeze)

20 years agoMinor style nits suggested by sam and mdodd:
imp [Sat, 25 Oct 2003 21:28:40 +0000 (21:28 +0000)]
Minor style nits suggested by sam and mdodd:

o give an argument to EP_BUSY_WAIT
o use foo_locked rather than foo_body
o Add locking assertions for extra safety.

20 years agoConvert to bus_space.
imp [Sat, 25 Oct 2003 19:56:19 +0000 (19:56 +0000)]
Convert to bus_space.
Make the pccard attachment work with NEWCARD
Start locking of the driver, but only the macros are defined right now

Tested on: Megahertz CC10BT/2
# (These cards are very popular on ebay these days, and run < $10 including
#  shipping from some sellers).

20 years agoBetter safe than clever.
des [Sat, 25 Oct 2003 19:53:28 +0000 (19:53 +0000)]
Better safe than clever.

Submitted by: das

20 years agoFor the SMP case, flush the TLB at the beginning of the page zero/copy
peter [Sat, 25 Oct 2003 18:51:41 +0000 (18:51 +0000)]
For the SMP case, flush the TLB at the beginning of the page zero/copy
routines.  Otherwise we run into trouble with speculative tlb preloads
on SMP systems.  This effectively defeats Jeff's revision 1.438
optimization (for his pentium4-M laptop) in the SMP case.  It breaks
other systems, particularly athlon-MP's.

20 years ago - Align a comment within struct vm_page.
alc [Sat, 25 Oct 2003 18:33:04 +0000 (18:33 +0000)]
 - Align a comment within struct vm_page.
 - Annotate the vm_page's valid field as synchronized by the containing
   vm object's lock.

20 years agoCheck (locked) before performing an advisory unlock following a failure
rwatson [Sat, 25 Oct 2003 16:43:50 +0000 (16:43 +0000)]
Check (locked) before performing an advisory unlock following a failure
of vn_start_write().  Otherwise, we may inconsistently attempt to release
the advisory lock.

Pointed out by: teggej

20 years agoWhen generate a core dump, use advisory locking in an advisory way:
rwatson [Sat, 25 Oct 2003 16:14:09 +0000 (16:14 +0000)]
When generate a core dump, use advisory locking in an advisory way:
if we do acquire an advisory lock, great!  We'll release it later.
However, if we fail to acquire a lock, we perform the coredump
anyway.  This problem became particularly visible with NFS after
the introduction of rpc.lockd: if the lock manager isn't running,
then locking calls will fail, aborting the core dump (resulting in
a zero-byte dump file).

Reported by: Yogeshwar Shenoy <ynshenoy@alumni.cs.ucsb.edu>

20 years agoAllow MAC policies to block/revoke kern_alq write access to a file.
rwatson [Sat, 25 Oct 2003 16:10:41 +0000 (16:10 +0000)]
Allow MAC policies to block/revoke kern_alq write access to a file.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
Reviewed by: jeff

20 years agoMake MAC_EXTERNALIZE() and MAC_INTERNALIZE() simply take the object
rwatson [Sat, 25 Oct 2003 15:28:20 +0000 (15:28 +0000)]
Make MAC_EXTERNALIZE() and MAC_INTERNALIZE() simply take the object
type, rather than "object_label" as the first argument.  This reduces
complexity a little for the consumer, and also makes it easier for
use to rename the underlying entry points in struct mac_policy_obj.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

20 years agoFix fwmem_strategy() race in 4-stable.
simokawa [Sat, 25 Oct 2003 15:05:59 +0000 (15:05 +0000)]
Fix fwmem_strategy() race in 4-stable.

20 years agoMove validity check of 'xfer->fc != NULL' to right place.
simokawa [Sat, 25 Oct 2003 15:04:49 +0000 (15:04 +0000)]
Move validity check of 'xfer->fc != NULL' to right place.

20 years agoupdate for conserver-8.0.4.
simokawa [Sat, 25 Oct 2003 14:53:53 +0000 (14:53 +0000)]
update for conserver-8.0.4.

20 years agoremove debug message.
simokawa [Sat, 25 Oct 2003 14:51:36 +0000 (14:51 +0000)]
remove debug message.

20 years agoDocument fabsl(3).
des [Sat, 25 Oct 2003 13:45:11 +0000 (13:45 +0000)]
Document fabsl(3).

Submitted by: Stefan Farfeleder <stefan@fafoe.narf.at>

20 years agodcons_crom.c needs dcons and firewire.
simokawa [Sat, 25 Oct 2003 12:59:22 +0000 (12:59 +0000)]
dcons_crom.c needs dcons and firewire.

20 years agoStyle changes. Inching closer to convergence with OpenBSD.
phk [Sat, 25 Oct 2003 12:56:51 +0000 (12:56 +0000)]
Style changes.  Inching closer to convergence with OpenBSD.

20 years agorevert following unwanted changes:
ume [Sat, 25 Oct 2003 10:57:08 +0000 (10:57 +0000)]
revert following unwanted changes:
  - __packed to __attribute__((__packed__)
  -  uintN_t back to u_intN_t

Reported by: bde

20 years agocorrect namespace pollution.
ume [Sat, 25 Oct 2003 09:37:10 +0000 (09:37 +0000)]
correct namespace pollution.

Submitted by: bde

20 years ago - fabsl.c should be named s_fabsl.c for consistency with libmsun's
des [Sat, 25 Oct 2003 09:32:18 +0000 (09:32 +0000)]
 - fabsl.c should be named s_fabsl.c for consistency with libmsun's
   documented naming scheme (unfortunately the documentation isn't in the
   tree as far as I can tell); no repocopy is required as there is no
   history to preserve.

 - replace simple and almost-correct implementation with slightly hackish
   but definitely correct implementation (tested on i386, alpha, sparc64)
   which requires pulling in fpmath.h and the MD _fpmath.h from libc.

 - try not to make a mess of the Makefile in the process.

 - enterprising minds are encouraged to implement more C99 long double
   functions.

20 years agoaccording to RFC3542 10.5, the 5th argment of inet6_opt_next()
ume [Sat, 25 Oct 2003 06:51:23 +0000 (06:51 +0000)]
according to RFC3542 10.5, the 5th argment of inet6_opt_next()
is not size_t but socklen_t.

Reported by: tinderbox

20 years ago - Call vnode_pager_input_old() with the vm object locked.
alc [Sat, 25 Oct 2003 05:21:16 +0000 (05:21 +0000)]
 - Call vnode_pager_input_old() with the vm object locked.

20 years agoGC workaround code for detecting pentium4's and disabling PSE and PG_G.
peter [Sat, 25 Oct 2003 05:14:38 +0000 (05:14 +0000)]
GC workaround code for detecting pentium4's and disabling PSE and PG_G.
It's been ifdef'ed out for ages.

20 years agoAdd devctl(4) notify support to ACPI. Various subsystems now notify
njl [Sat, 25 Oct 2003 05:03:25 +0000 (05:03 +0000)]
Add devctl(4) notify support to ACPI.  Various subsystems now notify
userland whenever events occur.  See the example in devd.conf below
to see how to use it.

20 years agoWhole grab-bag of changes:
imp [Sat, 25 Oct 2003 04:09:49 +0000 (04:09 +0000)]
Whole grab-bag of changes:
o Make the driver MPSAFE
o Some changes due to diff reduction effort with vx.
o Removed some obsolete junk.

Reviewed by: sam, modd

20 years agoStart to minimize diffs between vx and ep. These latter is based on a
imp [Sat, 25 Oct 2003 04:05:33 +0000 (04:05 +0000)]
Start to minimize diffs between vx and ep.  These latter is based on a
more advanced version of the chips supported by the former.  Matt Dodd
and I are working towards merging them, and this a step on that path.

20 years agoSort type declarations together.
rwatson [Sat, 25 Oct 2003 03:50:44 +0000 (03:50 +0000)]
Sort type declarations together.
Remove an excess carriage return.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

20 years agoConvenience functions to generate notifications from the kernel. The ACPI
imp [Fri, 24 Oct 2003 22:41:54 +0000 (22:41 +0000)]
Convenience functions to generate notifications from the kernel.  The ACPI
code will start using these shortly.

Reviewed by: njl

20 years agoAdd support for another multiple serial port card based on OX16PCI954 device
ambrisko [Fri, 24 Oct 2003 22:34:56 +0000 (22:34 +0000)]
Add support for another multiple serial port card based on OX16PCI954 device
id 0x950a.

MFC: after 4.9 release.

20 years agoParse the ! lines that will soon be coming from the kernel. These are
imp [Fri, 24 Oct 2003 22:02:29 +0000 (22:02 +0000)]
Parse the ! lines that will soon be coming from the kernel.  These are
a generalized notification mechanism for subsystems wishing to report
events.

Revieded by: njl

# The kernel side seems like it might be causing panics for me, but should
# be forthcoming shortly.

20 years agoDon't try to use dev->dma_lock unless dma is initialized (dev->dma != NULL)
anholt [Fri, 24 Oct 2003 21:45:21 +0000 (21:45 +0000)]
Don't try to use dev->dma_lock unless dma is initialized (dev->dma != NULL)
in bufs_info sysctl handler.  dev->dma and dev->dma_lock existence are
protected by DRM_LOCK().  Fixes panic on sysctl hw.dri when the device is
uninitialied (when you aren't in X).

20 years agoUse 'k' as suffix for Kilo
phk [Fri, 24 Oct 2003 21:23:47 +0000 (21:23 +0000)]
Use 'k' as suffix for Kilo

Pointed out by: several.

20 years ago - fix description of what processes SIGCONT can be sent to
kensmith [Fri, 24 Oct 2003 21:20:26 +0000 (21:20 +0000)]
- fix description of what processes SIGCONT can be sent to

PR: docs/58413
Reviewed by: rwatson
Approved by: blackend (mentor)

20 years agodon't allow reading from files that haven't been open'd for reading.
jmg [Fri, 24 Oct 2003 21:07:53 +0000 (21:07 +0000)]
don't allow reading from files that haven't been open'd for reading.

20 years ago- Add a DDB command 'show intrcnt' to show the non-zero interrupt counts.
jhb [Fri, 24 Oct 2003 21:05:30 +0000 (21:05 +0000)]
- Add a DDB command 'show intrcnt' to show the non-zero interrupt counts.
- Add a DDB function to dump the contents of an ithread and optionally
  details about each handler in that ithread.  This function can be used
  by MD code to implement DDB commands that display information about
  interrupt sources and their registered handlers.

20 years agoA few whitespace and comment tweaks.
jhb [Fri, 24 Oct 2003 21:02:26 +0000 (21:02 +0000)]
A few whitespace and comment tweaks.

20 years ago- Fail to probe if acpi0 probed ok as this driver basically tries to probe
jhb [Fri, 24 Oct 2003 21:01:31 +0000 (21:01 +0000)]
- Fail to probe if acpi0 probed ok as this driver basically tries to probe
  the ACPI timer and we shouldn't do that if ACPI is already around to do
  that for us.
- Set a description and tweak the order of checks in the probe function
  to more closely match other PCI drivers.

This should probably be moved to sys/dev/piix/piix.c at some point and
turned on for all i386 kernels rather than just SMP ones.

20 years ago- Renumber the skerberos4 dist to fill in the gaps left when krb4 was axed.
jhb [Fri, 24 Oct 2003 20:55:15 +0000 (20:55 +0000)]
- Renumber the skerberos4 dist to fill in the gaps left when krb4 was axed.
- Update the DIST_CRYPTO_ALL value to match reality.

20 years agoremove the ip6r0_addr and ip6r0_slmap members from ip6_rthdr0{}
ume [Fri, 24 Oct 2003 20:37:05 +0000 (20:37 +0000)]
remove the ip6r0_addr and ip6r0_slmap members from ip6_rthdr0{}
according to rfc2292bis.

Obtained from: KAME

20 years agocorrect tab and order.
ume [Fri, 24 Oct 2003 19:51:49 +0000 (19:51 +0000)]
correct tab and order.

20 years agoClose the right consumers if we run into trouble opening them all.
phk [Fri, 24 Oct 2003 18:47:31 +0000 (18:47 +0000)]
Close the right consumers if we run into trouble opening them all.

Submitted by: Pawel Jakub Dawidek <nick@garage.freebsd.pl>

20 years agoFix two old/new consumer confusions.
phk [Fri, 24 Oct 2003 18:46:23 +0000 (18:46 +0000)]
Fix two old/new consumer confusions.

Submitted by:    Pawel Jakub Dawidek <nick@garage.freebsd.pl>

20 years agooops, revert previous change to getaddrinfo.c. This is not related
ume [Fri, 24 Oct 2003 18:43:24 +0000 (18:43 +0000)]
oops, revert previous change to getaddrinfo.c.  This is not related
to RFC3493.  The previous change was related to RFC3484 (Default
Address Selection for IPv6), and it will come later.

20 years agoWhen grabbing vnodes to service NFS requests, make sure to call
phk [Fri, 24 Oct 2003 18:36:49 +0000 (18:36 +0000)]
When grabbing vnodes to service NFS requests, make sure to call
vn_start_write() early to avoid snapshot deadlocks.

By: mckusick

20 years agoSwitch Advanced Sockets API for IPv6 from RFC2292 to RFC3542
ume [Fri, 24 Oct 2003 18:26:30 +0000 (18:26 +0000)]
Switch Advanced Sockets API for IPv6 from RFC2292 to RFC3542
(aka RFC2292bis).  Though I believe this commit doesn't break
backward compatibility againt existing binaries, it breaks
backward compatibility of API.
Now, the applications which use Advanced Sockets API such as
telnet, ping6, mld6query and traceroute6 use RFC3542 API.

Obtained from: KAME

20 years agoshow maxmtu.
ume [Fri, 24 Oct 2003 18:00:17 +0000 (18:00 +0000)]
show maxmtu.

Obtained from: KAME

20 years agoSince dp->dom_ifattach calls malloc() with M_WAITOK, we cannot
ume [Fri, 24 Oct 2003 16:57:59 +0000 (16:57 +0000)]
Since dp->dom_ifattach calls malloc() with M_WAITOK, we cannot
use mutex lock directly here.  Protect ifp->if_afdata instead.

Reported by: grehan

20 years agoSysctl names should not contain dots. Convert them to underlines.
harti [Fri, 24 Oct 2003 16:44:27 +0000 (16:44 +0000)]
Sysctl names should not contain dots. Convert them to underlines.

20 years agoAdd dumb console driver and related bits.
simokawa [Fri, 24 Oct 2003 15:44:10 +0000 (15:44 +0000)]
Add dumb console driver and related bits.

dcons(4): very simple console and gdb port driver
dcons_crom(4): FireWire attachment
dconschat(8): User interface to dcons

Tested with: i386, i386-PAE, and sparc64.

20 years agoMdoc Janitor:
hmp [Fri, 24 Oct 2003 15:42:38 +0000 (15:42 +0000)]
Mdoc Janitor:

  * Use uppercase for .Dt macro arguments.

  * Use 'manual' instead of 'man' for consistency.

20 years agoFix for FW_ASYREQ.
simokawa [Fri, 24 Oct 2003 13:55:51 +0000 (13:55 +0000)]
Fix for FW_ASYREQ.
- set send.pay_len correctly.
- copy response only if needed.
- remove unnecessary 'err = 0'.

20 years agoMention the fact that our crontab is not fully POSIX.2 conform, because
harti [Fri, 24 Oct 2003 13:02:39 +0000 (13:02 +0000)]
Mention the fact that our crontab is not fully POSIX.2 conform, because
it doesn't allow the dangerous variant of calling it without any
argument.

20 years agoAllow nghook to execute a program with the data socket connected to
harti [Fri, 24 Oct 2003 10:01:36 +0000 (10:01 +0000)]
Allow nghook to execute a program with the data socket connected to
stdin and stdout instead of relaying the data. Now it is possible
to say:

nghook -e path: hook /usr/local/bin/foo arg1 arg2

and foo will have the hook to path: at file descriptors 0 and 1.

Add an option to specify control messages to be send to the node before
either executing the program or entering the data relay loop.

20 years agoDon't check timeout just after booted.
simokawa [Fri, 24 Oct 2003 07:42:21 +0000 (07:42 +0000)]
Don't check timeout just after booted.
Some transactions could be considered wrongly to be timeout
bacause interrupts are disabled during boot process.

20 years agoModule build infrastructure for the NgATM SAA layer.
harti [Fri, 24 Oct 2003 07:42:08 +0000 (07:42 +0000)]
Module build infrastructure for the NgATM SAA layer.

20 years agoNetgraph part of the NgATM signalling AA layer. These nodes can
harti [Fri, 24 Oct 2003 07:39:11 +0000 (07:39 +0000)]
Netgraph part of the NgATM signalling AA layer. These nodes can
also be used as a general-purpose transport protocol above any
packet layer (IP, UDP).

20 years agoBackswards is as backwards does: If we're MPSAFE, then we don't need
imp [Fri, 24 Oct 2003 07:20:13 +0000 (07:20 +0000)]
Backswards is as backwards does: If we're MPSAFE, then we don't need
giant, which implies that we need to take out giant it we're NOT
MPSAFE.

# I can't believe the number of people that looked at this failed to
# detect this.

20 years agoRemove unused header. See also ia64/disasm/disasm.h.
marcel [Fri, 24 Oct 2003 06:53:43 +0000 (06:53 +0000)]
Remove unused header. See also ia64/disasm/disasm.h.

20 years agostyle.
ume [Fri, 24 Oct 2003 06:53:12 +0000 (06:53 +0000)]
style.

Reported by: bde

20 years agoRemove ia64_pack_bundle() and ia64_unpack_bundle(). They are not
marcel [Fri, 24 Oct 2003 06:52:21 +0000 (06:52 +0000)]
Remove ia64_pack_bundle() and ia64_unpack_bundle(). They are not
used anymore.

20 years agoRemove unused file. db_disasm() has been implemented in db_interface.c
marcel [Fri, 24 Oct 2003 06:48:41 +0000 (06:48 +0000)]
Remove unused file. db_disasm() has been implemented in db_interface.c
now.

20 years ago - Push down Giant from vm_pageout() to vm_pageout_scan(), freeing
alc [Fri, 24 Oct 2003 06:43:04 +0000 (06:43 +0000)]
 - Push down Giant from vm_pageout() to vm_pageout_scan(), freeing
   vm_pageout_page_stats() from Giant.
 - Modify vm_pager_put_pages() and vm_pager_page_unswapped() to expect the
   vm object to be locked on entry.  (All of the pager routines now expect
   this.)

20 years agoImplement db_disasm() by using the new disassembler. Temporarily
marcel [Fri, 24 Oct 2003 06:42:03 +0000 (06:42 +0000)]
Implement db_disasm() by using the new disassembler. Temporarily
unimplement db_write_breakpoint() and db_clear_breakpoint().

20 years agoReduce the number of tcp time_wait structs to maxsockets / 5; this ensures
silby [Fri, 24 Oct 2003 05:44:14 +0000 (05:44 +0000)]
Reduce the number of tcp time_wait structs to maxsockets / 5; this ensures
that at most 20% of sockets can be in time_wait at one time, ensuring
that time_wait sockets do not starve real connections from inpcb
structures.

No implementation change is needed, jlemon already implemented a nice
LRU-ish algorithm for tcp_tw structure recycling.

This should reduce the need for sysadmins to lower the default msl on
busy servers.

20 years agoUse a TR of size 1 << IA64_ID_PAGE_SHIFT instead of 16M to avoid
arun [Fri, 24 Oct 2003 04:56:58 +0000 (04:56 +0000)]
Use a TR of size 1 << IA64_ID_PAGE_SHIFT instead of 16M to avoid
overlapping TR/TC entries (which results in a machine check). Note
that we don't look at the size of the memory descriptor, because
it doesn't guarantee non-overlap.

With this change, a UP kernel could boot on a Intel Tiger4 machine
with the following options:

options         LOG2_ID_PAGE_SIZE=26 # 64M
options         LOG2_PAGE_SIZE=14               # 16K

Approved by: marcel

20 years agooops, EAI_NONAME is not EAINONAME.
ume [Fri, 24 Oct 2003 03:49:38 +0000 (03:49 +0000)]
oops, EAI_NONAME is not EAINONAME.

20 years agoAdd __va_copy and make it always visible, in spite of the __ISO_C_VISIBLE
peter [Fri, 24 Oct 2003 02:50:39 +0000 (02:50 +0000)]
Add __va_copy and make it always visible, in spite of the __ISO_C_VISIBLE
setting.  Make va_copy be an alias if __ISO_C_VISIBLE >= 1999.

Why?  more than a few ports have an autoconf that looks for __va_copy
because it is available on glibc.  It is critical that we use it if
at all possible on amd64.  It generally isn't a problem for i386 and its
ilk because autoconf driven code tends to fall back to an assignment.

20 years agoDon't use fuword() or suword() unconditionally. They explicitly
marcel [Fri, 24 Oct 2003 02:33:26 +0000 (02:33 +0000)]
Don't use fuword() or suword() unconditionally. They explicitly
disallow reading or writing.

20 years agoRemove two unused fields in the operand structure (o_read & o_write).
marcel [Fri, 24 Oct 2003 02:05:53 +0000 (02:05 +0000)]
Remove two unused fields in the operand structure (o_read & o_write).

20 years agoUpdate to latest from DRI CVS. Primary new feature is mostly-complete smpng
anholt [Fri, 24 Oct 2003 01:48:17 +0000 (01:48 +0000)]
Update to latest from DRI CVS.  Primary new feature is mostly-complete smpng
locking, and the apparently unnecessary locking for -stable has been removed.
This may fix issues with missed interrupts since April, which manifested
themselves as slowdowns or hangs in radeon, in particular. Many cleanups also
took place.  In the shared code, there are improvements to r128 driver
stability.

20 years agoo restructure initialization code so data structures are setup
sam [Fri, 24 Oct 2003 00:09:18 +0000 (00:09 +0000)]
o restructure initialization code so data structures are setup
  when loaded as a module
o cleanup data structures on module unload when no application has
  been started (i.e. kldload, kldunload w/o mrtd)
o remove extraneous unlocks immediately prior to destroying them

Supported by: FreeBSD Foundation

20 years agocheck return result from rtalloc1 before invoking RTUNLOCK
sam [Thu, 23 Oct 2003 21:41:00 +0000 (21:41 +0000)]
check return result from rtalloc1 before invoking RTUNLOCK

20 years agoWrites to p_flag in __setugid() no longer need Giant.
jhb [Thu, 23 Oct 2003 21:20:34 +0000 (21:20 +0000)]
Writes to p_flag in __setugid() no longer need Giant.

20 years agoMove the P_COWINPROGRESS flag from being a per-process p_flag to being a
jhb [Thu, 23 Oct 2003 21:14:08 +0000 (21:14 +0000)]
Move the P_COWINPROGRESS flag from being a per-process p_flag to being a
per-thread td_pflag which doesn't require any locks to read or write as it
is only read or written by curthread on itself.

Glanced at by: mckusick

20 years agoAdd appropriate const poisoning to the assert_*locked() family so that I can
wollman [Thu, 23 Oct 2003 18:17:36 +0000 (18:17 +0000)]
Add appropriate const poisoning to the assert_*locked() family so that I can
call ASSERT_VOP_LOCKED(vp, __func__) without a diagnostic.

Inspired by: the evil and rude OpenAFS cache manager code

20 years agoAdd simple support for AGP 3.0 including enabling 8x mode. The simple
jhb [Thu, 23 Oct 2003 18:08:56 +0000 (18:08 +0000)]
Add simple support for AGP 3.0 including enabling 8x mode.  The simple
part of the support is that it still assumes one master and one target
where as AGP 3.0 actually supports multiple devices on the bus.

Submitted by: Keith Whitwell <keith@tungstengraphics.com>
Sponsored by: The Weather Channel

20 years agoworkaround to have backward compatibility for EAI_NODATA.
ume [Thu, 23 Oct 2003 17:54:17 +0000 (17:54 +0000)]
workaround to have backward compatibility for EAI_NODATA.
it will be removed on 23 Apr 2004.

Submitted by: terry