]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoRestore FD_DEBUG for pc98.
nyan [Mon, 23 Aug 2004 11:59:32 +0000 (11:59 +0000)]
Restore FD_DEBUG for pc98.

20 years agos/the on/on the/
blackend [Mon, 23 Aug 2004 11:00:20 +0000 (11:00 +0000)]
s/the on/on the/

PR: misc/70842
Submitted by: Sean C. Farley <sean-freebsd@farley.org>

20 years agoo Fix whitespace bug introduced in the previous commit.
sobomax [Mon, 23 Aug 2004 10:09:29 +0000 (10:09 +0000)]
o Fix whitespace bug introduced in the previous commit.

Submitted by: ru

o Simplify p4tcc_power_profile().

Submitted by: maxim

20 years agoo Extend boot output: print out mimimum/maximum performance value and number
sobomax [Mon, 23 Aug 2004 09:47:56 +0000 (09:47 +0000)]
o Extend boot output: print out mimimum/maximum performance value and number
  of performance steps available;

o similarly to Enhanced SpeedStep driver, export list of all available steps
  via hw.p4tcc.cpuperf_levels sysctl.

20 years agoAdd missing GEOM classes, which are aware of geom(8).
pjd [Mon, 23 Aug 2004 06:23:17 +0000 (06:23 +0000)]
Add missing GEOM classes, which are aware of geom(8).

Submitted by: kuriyama

20 years agoMake this compile again in the standalone and the MODULES_WITH_WORLD
imp [Mon, 23 Aug 2004 03:38:21 +0000 (03:38 +0000)]
Make this compile again in the standalone and the MODULES_WITH_WORLD
environments.  Chances are good that this doesn't produce a good
module, but I leave the proper defaults to the dummy opt_* files to
the author.

20 years agoRemove in6_prefix.[ch] and the contained router renumbering capability.
rwatson [Mon, 23 Aug 2004 03:00:27 +0000 (03:00 +0000)]
Remove in6_prefix.[ch] and the contained router renumbering capability.
The prefix management code currently resides in nd6, leaving only the
unused router renumbering capability in the in6_prefix files.  Removing
it will make it easier for us to provide locking for the remainder of
IPv6 by reducing the number of objects requiring synchronized access.

This functionality has also been removed from NetBSD and OpenBSD.

Submitted by: George Neville-Neil <gnn at neville-neil.com>
Discussed with/approved by: suz, keiichi at kame.net, core at kame.net

20 years agoTemporarily back out r1.74 as it seems to cause a number of regressions
kan [Mon, 23 Aug 2004 02:39:45 +0000 (02:39 +0000)]
Temporarily back out r1.74 as it seems to cause a number of regressions
accordimg to numerous reports. It  might get reintroduced some time later
when an exact failure mode is understood better.

20 years agoPass a correct lowaddr to bus_dma_tag_create(), lnc(4) cards can only
mux [Sun, 22 Aug 2004 23:01:13 +0000 (23:01 +0000)]
Pass a correct lowaddr to bus_dma_tag_create(), lnc(4) cards can only
deal with 24-bit addresses.  While the two other attachments, namely
isa and cbus, do it properly, the PCI attachment was passing
BUS_SPACE_MAXADDR instead of BUS_SPACE_MAXADDR_24BIT.  This bug
became apparent with the new contigmalloc() code.

This fixes the problem reported with lnc(4) interfaces inside VMWare,
and should theoritically also fix any user of a PCI lnc(4) card.  It
is a RELENG_5 MFC candidate.

Tested by: Florian Le Goff <madflo@beertech.org>

20 years agoStyle update: use newer style function prototypes in if_sl.c in
rwatson [Sun, 22 Aug 2004 21:32:52 +0000 (21:32 +0000)]
Style update: use newer style function prototypes in if_sl.c in
prep for merging locking.

20 years agoMove the cow field between wire_count and hold_count. This is the
marcel [Sun, 22 Aug 2004 20:52:23 +0000 (20:52 +0000)]
Move the cow field between wire_count and hold_count. This is the
position that is 64-bit aligned and makes sure that the valid and
dirty fields are also 64-bit aligned. This means that if PAGE_SIZE
is 32K, the size of the vm_page structure is only increased by 8
bytes instead of 16 bytes. More importantly, the vm_page structure
is either 120 or 128 bytes on ia64. These are "interesting" sizes.

20 years agoWhen creating a new md, wait for geom's event queue to become empty
cperciva [Sun, 22 Aug 2004 19:44:24 +0000 (19:44 +0000)]
When creating a new md, wait for geom's event queue to become empty
before returning.  Device nodes are created via the "taste" mechanism,
so this is necessary in order to make sure that devfs entries are
created before mdconfig(8) returns.

This may be a MFC candidate for 5.3.

Suggested by: phk

20 years agoThe new contigmalloc code is exposing a lot of misuses of busdma memory
green [Sun, 22 Aug 2004 18:57:40 +0000 (18:57 +0000)]
The new contigmalloc code is exposing a lot of misuses of busdma memory
allocation. Notably, in this case, the driver tries to allocate several
pieces of memory and then fails if the pieces allocated after the first
do not come after it physically, and within a specific range (8MB I
believe).  Of course, this could just as easily fail for any number of
reasons, but it almost always fails now that contiguous allocations start
at the end of possible specified memory locations rather than the beginning.

Allocate all the possibly-needed memory up front, even though it's a waste,
to get around this.  The least bogus solution would be to take the physical
address from the first allocation and create a new tag that specified that
further allocations must follow it within that 8MB window, then use that
when allocating new channels, but that's left for anyone else that really
feels like doing it.

Tested by: Erwin Lansing <erwin@lansing.dk>

20 years agoImplement the possibility to remove drives.
le [Sun, 22 Aug 2004 17:07:55 +0000 (17:07 +0000)]
Implement the possibility to remove drives.

20 years agoLoopback a fix from Cedric Berger:
mlaier [Sun, 22 Aug 2004 16:58:06 +0000 (16:58 +0000)]
Loopback a fix from Cedric Berger:
Fix table add/replace commands with securelevel=2.
Reported by James J. Lippard.

Discussed with: yongari
MFC after: 5 days

20 years agoAllow early drop for non-ALTQ enabled queues in an ALTQ-enabled kernel.
mlaier [Sun, 22 Aug 2004 16:42:28 +0000 (16:42 +0000)]
Allow early drop for non-ALTQ enabled queues in an ALTQ-enabled kernel.
Previously the early drop was disabled unconditionally for ALTQ-enabled
kernels.

This should give some benefit for the normal gateway + LAN-server case with
a busy LAN leg and an ALTQ managed uplink.

Reviewed and style help from: cperciva, pjd

20 years agoRegression tests for 'verify reading' algorithm.
pjd [Sun, 22 Aug 2004 16:26:09 +0000 (16:26 +0000)]
Regression tests for 'verify reading' algorithm.

20 years agoFix sysctl name.
pjd [Sun, 22 Aug 2004 16:22:20 +0000 (16:22 +0000)]
Fix sysctl name.

20 years agoImplementation of 'verify reading' algorithm, which uses parity data for
pjd [Sun, 22 Aug 2004 16:21:12 +0000 (16:21 +0000)]
Implementation of 'verify reading' algorithm, which uses parity data for
verification of regular data when device is in complete state.
On verification error, EIO error is returned for the bio and sysctl
kern.geom.raid3.stat.parity_mismatch is increased.

Suggested by: phk

20 years agoWorkaround devices that responds with registers as *both* master & slave,
sos [Sun, 22 Aug 2004 15:54:08 +0000 (15:54 +0000)]
Workaround devices that responds with registers as *both* master & slave,
but fail utterly when we try to talk to the "fake" device.

20 years agoUse securelevel_gt instead of reading global securelevel unprotected.
mlaier [Sun, 22 Aug 2004 15:23:48 +0000 (15:23 +0000)]
Use securelevel_gt instead of reading global securelevel unprotected.

Submitted by: yongari
MFC after: 3 days

20 years agoopt_agp.h is not required anymore.
nyan [Sun, 22 Aug 2004 15:14:44 +0000 (15:14 +0000)]
opt_agp.h is not required anymore.

20 years agoMFi386: revision 1.215.
nyan [Sun, 22 Aug 2004 15:13:08 +0000 (15:13 +0000)]
MFi386: revision 1.215.

20 years agoMerged from sys/dev/fdc/fdc.c revision 1.283.
nyan [Sun, 22 Aug 2004 15:10:53 +0000 (15:10 +0000)]
Merged from sys/dev/fdc/fdc.c revision 1.283.

20 years agoMake debug.kdb.stop_cpus also a TUNABLE() so it can be set prior to boot
rwatson [Sun, 22 Aug 2004 15:10:52 +0000 (15:10 +0000)]
Make debug.kdb.stop_cpus also a TUNABLE() so it can be set prior to boot
to help debug early nasty hangs.

20 years agoDon't need to assert Giant in fw_output(), only in the firewire start
rwatson [Sun, 22 Aug 2004 14:48:55 +0000 (14:48 +0000)]
Don't need to assert Giant in fw_output(), only in the firewire start
routine.

20 years agoIn the PCI error interrupt handler, specify the width of the PCI configuration
gibbs [Sun, 22 Aug 2004 14:02:43 +0000 (14:02 +0000)]
In the PCI error interrupt handler, specify the width of the PCI configuration
cycle using the correct argument.  The location and width were reversed.

MFC in: 2 days

20 years agoSet AHD_BUSFREEREV_BUG in the bug field for Rev B chips, not the
gibbs [Sun, 22 Aug 2004 13:54:27 +0000 (13:54 +0000)]
Set AHD_BUSFREEREV_BUG in the bug field for Rev B chips, not the
feature field.

Reported by: Ken Westerback <krw@openbsd.org>
MFC in: 2 days

20 years agoAdd forgotten format specifier in a KASSERT and shut up the compiler.
le [Sun, 22 Aug 2004 13:34:24 +0000 (13:34 +0000)]
Add forgotten format specifier in a KASSERT and shut up the compiler.

Submitted by: Gavin Atkinson <gavin.atkinson@ury.york.ac.uk>

20 years agoOnly compile fbsd-thread.c on amd64 and i386, other architectures
davidxu [Sun, 22 Aug 2004 10:59:33 +0000 (10:59 +0000)]
Only compile fbsd-thread.c on amd64 and i386, other architectures
have not been tested yet.

20 years agoUnbreak the build. Translators, please let me know
hrs [Sun, 22 Aug 2004 09:53:24 +0000 (09:53 +0000)]
Unbreak the build.  Translators, please let me know
if I break something in your localized docs.

20 years agoFix comments, most are style fix.
davidxu [Sun, 22 Aug 2004 09:05:57 +0000 (09:05 +0000)]
Fix comments, most are style fix.

20 years agoTrim EOL white spaces.
maxim [Sun, 22 Aug 2004 08:57:05 +0000 (08:57 +0000)]
Trim EOL white spaces.

20 years agoCheck a return code from read(2) correctly.
maxim [Sun, 22 Aug 2004 08:56:01 +0000 (08:56 +0000)]
Check a return code from read(2) correctly.

PR: kern/70815
Submitted by: info [at] haakh.de

20 years agoFix some text messages.
davidxu [Sun, 22 Aug 2004 08:48:56 +0000 (08:48 +0000)]
Fix some text messages.

20 years agoPart 2 of fixing the boot code: gcc 3.4 fixes.
marcel [Sun, 22 Aug 2004 06:24:59 +0000 (06:24 +0000)]
Part 2 of fixing the boot code: gcc 3.4 fixes.

The whole problem seems to be size. Which is odd, because it is said
that size doesn't matter. Anyway... Add -Os to strategic places in the
makefile to have the final loader be as mall as possible. This seems
to be enough to make it work. For now... I think something is more
fundamentally wrong; or something more fundamental is wrong. Potato,
potaato.

20 years agoAdd code to support statically linked binary and .core file of threaded
davidxu [Sun, 22 Aug 2004 05:53:38 +0000 (05:53 +0000)]
Add code to support statically linked binary and .core file of threaded
program.

20 years agoFound another one. Why does mdconfig hate me? Add a "sleep 5" to
kensmith [Sun, 22 Aug 2004 05:34:07 +0000 (05:34 +0000)]
Found another one.  Why does mdconfig hate me?  Add a "sleep 5" to
this script, without it sparc64 ISO building was consistently failing
because the /dev/md0 device name was not present when the commands
following mdconfig ran.  Apparently there is the possibility of a delay
between when mdconfig finishes and the names become visible in /dev.
Yes, we could code this better than an unconditional call to "sleep 5"
but IMHO we should fix the underlying problem instead.

MFC after: 3 days

20 years agodiff reduction for upcoming patch. Use a macro that masks
julian [Sun, 22 Aug 2004 05:21:41 +0000 (05:21 +0000)]
diff reduction for upcoming patch. Use a macro that masks
some of the odd goings on with sub-structures, because they will
go away anyhow.

20 years agoAdd support for Intel E7205 AGP.
anholt [Sun, 22 Aug 2004 03:55:04 +0000 (03:55 +0000)]
Add support for Intel E7205 AGP.

PR: kern/69858
Submitted by: Jacobo Arvelo <unix4all at gulic dot org>

20 years agoCurrently, if the secure level is low enough, system flags can
csjp [Sun, 22 Aug 2004 02:03:41 +0000 (02:03 +0000)]
Currently, if the secure level is low enough, system flags can
be manipulated by prison root. In 4.x prison root can not manipulate
system flags, regardless of the security level. This behavior
should remain consistent to avoid any surprises which could lead
to security problems for system administrators which give out
privileged access to jails.

This commit changes suser_cred's flag argument from SUSER_ALLOWJAIL
to 0. This will prevent prison root from being able to manipulate
system flags on files.

This may be a MFC candidate for RELENG_5.

Discussed with: cperciva
Reviewed by: rwatson
Approved by: bmilekic (mentor)
PR: kern/70298

20 years agoWhen sliding the m_data pointer forward, update m_pktrhdr.len as well
rwatson [Sun, 22 Aug 2004 01:32:48 +0000 (01:32 +0000)]
When sliding the m_data pointer forward, update m_pktrhdr.len as well
as m_len, or the pkthdr length will be inconsistent with the actual
length of data in the mbuf chain.  The symptom of this occuring was
"out of data" warnings from in_cksum_skip() on large UDP packets sent
via the loopback interface.

Foot shot: green

20 years agoPart 1 of fixing the boot code: binutils 2.15 fixes.
marcel [Sun, 22 Aug 2004 00:26:01 +0000 (00:26 +0000)]
Part 1 of fixing the boot code: binutils 2.15 fixes.

The binutils 2.15 assembler now automaticly and non-optionally adds
the .eh_frame section for unwind information. This section appears
to wreck havoc to the final boot code. Fix this by using a special
linker script that discards the .eh_frame sections, but is otherwise
identical to the linker internal script used for -N.

Compiler used: gcc 3.3.5
Verified with: binutils 2.14 & binutils 2.15 (stock and in-tree)
Tested with: /boot/loader & /boot/netboot

20 years agoIn the previous revision, I failed to condition an early release of Giant
alc [Sun, 22 Aug 2004 00:08:43 +0000 (00:08 +0000)]
In the previous revision, I failed to condition an early release of Giant
in vm_fault() on debug_mpsafevm.  If debug_mpsafevm was not set, the result
was an assertion failure early in the boot process.

Reported by: green@

20 years agoIf a tunable for the routing socket netisr queue max is defined, allow it
rwatson [Sat, 21 Aug 2004 21:45:40 +0000 (21:45 +0000)]
If a tunable for the routing socket netisr queue max is defined, allow it
to override the default value, rather than the default value overriding
the tunable.

20 years agoAllow the size of the routing socket netisr queue to be configured using
rwatson [Sat, 21 Aug 2004 21:20:06 +0000 (21:20 +0000)]
Allow the size of the routing socket netisr queue to be configured using
the tunable or sysctl 'net.route.netisr_maxqlen'.  Default the maximum
depth to 256 rather than IFQ_MAXLEN due to the downsides of dropping
routing messages.

MT5 candidate.

Discussed with: mdodd, mlaier, Vincent Jardin <jardin at 6wind.com>

20 years agoAdd version history.
pjd [Sat, 21 Aug 2004 21:15:03 +0000 (21:15 +0000)]
Add version history.

20 years agoProtect the period! And add a new line.
trhodes [Sat, 21 Aug 2004 20:40:51 +0000 (20:40 +0000)]
Protect the period!  And add a new line.

Guilty of bad merge: trhodes

20 years agoDocument recently added features and bump the doc date.
trhodes [Sat, 21 Aug 2004 20:26:03 +0000 (20:26 +0000)]
Document recently added features and bump the doc date.

20 years agoFix positional parameter reference to ${10}.
thomas [Sat, 21 Aug 2004 20:21:49 +0000 (20:21 +0000)]
Fix positional parameter reference to ${10}.
Fix target location for generation of fstab.

PR: bin/70795

20 years agoAllow mac_bsdextended(4) to log failed attempts to syslog's AUTHPRIV
trhodes [Sat, 21 Aug 2004 20:19:19 +0000 (20:19 +0000)]
Allow mac_bsdextended(4) to log failed attempts to syslog's AUTHPRIV
facility.  This is disabled by default but may be turned on by using
the mac_bsdextended_logging sysctl.

Reviewed by: re (jhb)
Approved by: re (jhb)

20 years agoGive the mac_bsdextended(4) policy the ability to match and apply on a first
trhodes [Sat, 21 Aug 2004 20:15:08 +0000 (20:15 +0000)]
Give the mac_bsdextended(4) policy the ability to match and apply on a first
rule only in place of all rules match.  This is similar to how ipfw(8) works.

Provide a sysctl, mac_bsdextended_firstmatch_enabled, to enable this
feature.

Reviewed by: re (jhb)
Aprroved by: re (jhb)

20 years agoClarify __FreeBSD_version scheme.
obrien [Sat, 21 Aug 2004 20:05:33 +0000 (20:05 +0000)]
Clarify __FreeBSD_version scheme.

Submitted by: ru

20 years agoHit people over the head so they realize run-time errors of the form
obrien [Sat, 21 Aug 2004 19:44:43 +0000 (19:44 +0000)]
Hit people over the head so they realize run-time errors of the form
/libexec/ld-elf.so.1: Undefined symbol "_ZNSs20_S_empty_rep_storageE"
does mean they are hitting the GCC 3.4 ABI change issue.

20 years agoFurther reduce the use of Giant by vm_fault(): Giant is held only when
alc [Sat, 21 Aug 2004 19:20:21 +0000 (19:20 +0000)]
Further reduce the use of Giant by vm_fault(): Giant is held only when
manipulating a vnode, e.g., calling vput().  This reduces contention for
Giant during many copy-on-write faults, resulting in some additional
speedup on SMPs.

Note: debug_mpsafevm must be enabled for this optimization to take effect.

20 years agoProperly free the temporary sf_buf in uiomove_fromphys() if a copyin or
alc [Sat, 21 Aug 2004 18:50:34 +0000 (18:50 +0000)]
Properly free the temporary sf_buf in uiomove_fromphys() if a copyin or
copyout fails.

Obtained from: DragonFlyBSD

20 years agoFix PCI link irq programming on resume. A logic bug prevented a device
njl [Sat, 21 Aug 2004 18:18:32 +0000 (18:18 +0000)]
Fix PCI link irq programming on resume.  A logic bug prevented a device
match, inverting which links actually got resumed.

Submitted by: Hiroyuki Aizu <eyes_at_navi.org>
MFC after: 3 days

20 years agoAdd regression test for the new "round-robin reading" feature.
pjd [Sat, 21 Aug 2004 18:14:55 +0000 (18:14 +0000)]
Add regression test for the new "round-robin reading" feature.

20 years agoImplement new reading algorithm, which will use parity component for reading
pjd [Sat, 21 Aug 2004 18:11:46 +0000 (18:11 +0000)]
Implement new reading algorithm, which will use parity component for reading
as well, even if device is in complete state.
I observe 40% of speed-up with this option for random read operations,
but slowdown for sequential reads.
Basically, without this option reading from a RAID3 device built from 5
components (c0-c4) looks like this:

Request no. Used components
1 c0+c1+c2+c3
2 c0+c1+c2+c3
3 c0+c1+c2+c3

With the new feature:

Request no. Used components
1 c0+c1+c2+c3
2 (c1^c2^c3^c4)+c1+c2+c3
3 c0+(c0^c2^c3^c4)+c2+c3
4 c0+c1+(c0^c1^c3^c4)+c3
5 c0+c1+c2+(c0^c1^c2^c4)
6 c0+c1+c2+c3
[...]

20 years agoFix where my automated script blew the SCM ID format conversion.
obrien [Sat, 21 Aug 2004 17:44:57 +0000 (17:44 +0000)]
Fix where my automated script blew the SCM ID format conversion.

20 years agoWhen a prison is given the ability to create raw sockets (when the
csjp [Sat, 21 Aug 2004 17:38:57 +0000 (17:38 +0000)]
When a prison is given the ability to create raw sockets (when the
security.jail.allow_raw_sockets sysctl MIB is set to 1) where privileged
access to jails is given out, it is possible for prison root to manipulate
various network parameters which effect the host environment. This commit
plugs a number of security holes associated with the use of raw sockets
and prisons.

This commit makes the following changes:

- Add a comment to rtioctl warning developers that if they add
  any ioctl commands, they should use super-user checks where necessary,
  as it is possible for PRISON root to make it this far in execution.
- Add super-user checks for the execution of the SIOCGETVIFCNT
  and SIOCGETSGCNT IP multicast ioctl commands.
- Add a super-user check to rip_ctloutput(). If the calling cred
  is PRISON root, make sure the socket option name is IP_HDRINCL,
  otherwise deny the request.

Although this patch corrects a number of security problems associated
with raw sockets and prisons, the warning in jail(8) should still
apply, and by default we should keep the default value of
security.jail.allow_raw_sockets MIB to 0 (or disabled) until
we are certain that we have tracked down all the problems.

Looking forward, we will probably want to eliminate the
references to curthread.

This may be a MFC candidate for RELENG_5.

Reviewed by: rwatson
Approved by: bmilekic (mentor)

20 years agoWhen notifying protocol components of an event on an in6pcb, use the
rwatson [Sat, 21 Aug 2004 17:38:48 +0000 (17:38 +0000)]
When notifying protocol components of an event on an in6pcb, use the
result of the notify() function to decide if we need to unlock the
in6pcb or not, rather than always unlocking.  Otherwise, we may unlock
and already unlocked in6pcb.

Reported by: kuriyama, Gordon Bergling <gbergling at 0xfce3.net>
Tested by: kuriyama, Gordon Bergling <gbergling at 0xfce3.net>
Discussed with: mdodd

20 years agoWhen prepending space onto outgoing UDP datagram payloads to hold the
rwatson [Sat, 21 Aug 2004 16:14:04 +0000 (16:14 +0000)]
When prepending space onto outgoing UDP datagram payloads to hold the
UDP/IP header, make sure that space is also allocated for the link
layer header.  If an mbuf must be allocated to hold the UDP/IP header
(very likely), then this will avoid an additional mbuf allocation at
the link layer.  This trick is also used by TCP and other protocols to
avoid extra calls to the mbuf allocator in the ethernet (and related)
output routines.

20 years agoAdd &release.type.current; to unbreak the build.
hrs [Sat, 21 Aug 2004 14:27:21 +0000 (14:27 +0000)]
Add &release.type.current; to unbreak the build.

20 years agomdoc(7) janitor:
simon [Sat, 21 Aug 2004 12:58:09 +0000 (12:58 +0000)]
mdoc(7) janitor:
- Fix hard sentence breaks.

MFC after: 3 days

20 years agoAdd &release.type.current; to unbreak the build.
hrs [Sat, 21 Aug 2004 12:34:42 +0000 (12:34 +0000)]
Add &release.type.current; to unbreak the build.

20 years agoPull debug symbols in for statically linked binary.
davidxu [Sat, 21 Aug 2004 11:49:19 +0000 (11:49 +0000)]
Pull debug symbols in for statically linked binary.

Reviewed by: desichen

20 years agoBump version numbers for 6.0-CURRENT and fix some sentences.
hrs [Sat, 21 Aug 2004 08:59:53 +0000 (08:59 +0000)]
Bump version numbers for 6.0-CURRENT and fix some sentences.

20 years agoAttempt to make the probe for hardware more specific.
phk [Sat, 21 Aug 2004 08:40:00 +0000 (08:40 +0000)]
Attempt to make the probe for hardware more specific.

20 years agoBack out the call to _init_tls() - something is broken there and it
dfr [Sat, 21 Aug 2004 08:22:00 +0000 (08:22 +0000)]
Back out the call to _init_tls() - something is broken there and it
prevents all static binaries from running.

20 years agoThe UTF2 encoding and the INVALID keyword are now obsolete.
tjr [Sat, 21 Aug 2004 08:16:36 +0000 (08:16 +0000)]
The UTF2 encoding and the INVALID keyword are now obsolete.

20 years agoRe-word warning about the UTF2 encoding, taking care to use the word
tjr [Sat, 21 Aug 2004 08:08:29 +0000 (08:08 +0000)]
Re-word warning about the UTF2 encoding, taking care to use the word
"obsolete" instead of "deprecated".

20 years agoBump document date for previous.
tjr [Sat, 21 Aug 2004 08:03:18 +0000 (08:03 +0000)]
Bump document date for previous.

20 years agoRe-word warning about the rune interface, taking care to use the word
tjr [Sat, 21 Aug 2004 08:00:31 +0000 (08:00 +0000)]
Re-word warning about the rune interface, taking care to use the word
"obsolete" instead of "deprecated".

20 years agoChange "deprecated" in link-time warnings about various rune functions
tjr [Sat, 21 Aug 2004 07:48:06 +0000 (07:48 +0000)]
Change "deprecated" in link-time warnings about various rune functions
to "obsolete".

20 years agoRe-word compatibility section, taking care to use the word "obsolete" to
tjr [Sat, 21 Aug 2004 07:37:08 +0000 (07:37 +0000)]
Re-word compatibility section, taking care to use the word "obsolete" to
describe the 4.4BSD extension of accepting characters (runes) outside of
the range of unsigned char.

20 years agoAvoid accidental use of ANSI C trigraphs.
maxim [Sat, 21 Aug 2004 07:23:41 +0000 (07:23 +0000)]
Avoid accidental use of ANSI C trigraphs.

Spotted by: Pawel Worach
Obtained from: NetBSD (rev. 1.11, 1.13)

20 years agoLet GCC know that ___runetype(), ___tolower() and ___toupper() are pure
tjr [Sat, 21 Aug 2004 07:00:40 +0000 (07:00 +0000)]
Let GCC know that ___runetype(), ___tolower() and ___toupper() are pure
functions, allowing it to generate better code for the <ctype.h> and
<wctype.h> functions. For example, it can now keep _CurrentRuneLocale
in a register across calls to these functions, and can delete calls to
___runetype() if the result is already known or not used.

20 years agoFix aperture size detection on some ALi chipsets by only using the lowest 4 bits
anholt [Sat, 21 Aug 2004 06:24:21 +0000 (06:24 +0000)]
Fix aperture size detection on some ALi chipsets by only using the lowest 4 bits
to check aperture size, avoiding hangs.  Maintain the rest of the bits when
setting/unsetting ATTBASE.  This essentially matches Linux's AGP driver as well.

PR: kern/70037
Submitted by: Mark Tinguely <tinguely at casselton dot net>
Obtained from: NetBSD

20 years agoApply some stylistic changes based off of kern/70037 (content changes to
anholt [Sat, 21 Aug 2004 06:18:11 +0000 (06:18 +0000)]
Apply some stylistic changes based off of kern/70037 (content changes to
follow).

20 years agoEnter the long awaited start of FreeBSD 6.0!
obrien [Sat, 21 Aug 2004 03:58:39 +0000 (03:58 +0000)]
Enter the long awaited start of FreeBSD 6.0!

20 years agoAdd some details about our wonderful system.
obrien [Sat, 21 Aug 2004 03:36:47 +0000 (03:36 +0000)]
Add some details about our wonderful system.

20 years agoDocument when this function came into FreeBSD.
obrien [Sat, 21 Aug 2004 03:31:17 +0000 (03:31 +0000)]
Document when this function came into FreeBSD.

20 years agoDon't bother calling the module event handlers from module_shutdown()
truckman [Fri, 20 Aug 2004 21:47:48 +0000 (21:47 +0000)]
Don't bother calling the module event handlers from module_shutdown()
in the shutdown_final state if the RB_NOSYNC flag is set.

The specific motivation in this case is that a system panic in an
interrupt context results in a call to module_shutdown(), which
calls g_modevent(), which calls g_malloc(..., M_WAITOK), which
results in a second panic.   While g_modevent() could be fixed to
not call malloc() for MOD_SHUTDOWN events (which it doesn't handle
in any case), it is probably also a good idea to entirely skip the
execution of the module shutdown handlers after a panic.

This may be a MFC candidate for RELENG_5.

20 years agoDon't attempt to trigger the syncer thread final sync code in the
truckman [Fri, 20 Aug 2004 19:21:47 +0000 (19:21 +0000)]
Don't attempt to trigger the syncer thread final sync code in the
shutdown_pre_sync state if the RB_NOSYNC flag is set.  This is the
likely cause of hangs after a system panic that are keeping crash
dumps from being done.

This is a MFC candidate for RELENG_5.

MFC after: 3 days

20 years agoCorrect the args to busdma, mostly cosmetic.
sos [Fri, 20 Aug 2004 19:05:22 +0000 (19:05 +0000)]
Correct the args to busdma, mostly cosmetic.

20 years agoRemove some dead code under a straggling APIC_IO #ifdef that I missed
jhb [Fri, 20 Aug 2004 17:24:52 +0000 (17:24 +0000)]
Remove some dead code under a straggling APIC_IO #ifdef that I missed
back before 5.2.

20 years agoUse the new start for the offset, not the old end.
njl [Fri, 20 Aug 2004 17:04:49 +0000 (17:04 +0000)]
Use the new start for the offset, not the old end.

20 years agoAdd description of raidtest tool.
pjd [Fri, 20 Aug 2004 16:56:43 +0000 (16:56 +0000)]
Add description of raidtest tool.

20 years agoCorrectly handle BIOS resources that are duplicated (!). There are many
njl [Fri, 20 Aug 2004 16:52:44 +0000 (16:52 +0000)]
Correctly handle BIOS resources that are duplicated (!).  There are many
systems that have overlapping regions specified in their sysresource
objects.  This patch fixes ATA DMA and acpi_timer allocation for such
sysctems.  It should eventually be moved to resource_list_add() if it is
a valid generalized approach.  The minimal approach for 5.3 is:

"Loop through all current resources to see if the new one overlaps
any existing ones.  If so, the old one always takes precedence and
the new one is adjusted (or rejected).  We check for three cases:

1. Tail of new resource overlaps head of old resource:  truncate the
   new resource so it is contiguous with the start of the old.
2. New resource wholly contained within the old resource:  error.
3. Head of new resource overlaps tail of old resource:  truncate the
   new resource so it is contiguous, following the old."

Tested by: Radek Kozlowski <radek_at_raadradd.com>
Discussed with: imp
MFC after: 4 days

20 years agoRemove a check that is too strict. With BIOSen that specify an IO/ctl port
njl [Fri, 20 Aug 2004 16:34:30 +0000 (16:34 +0000)]
Remove a check that is too strict.  With BIOSen that specify an IO/ctl port
of 0x3f2-0x3f5,0x3f7 the ports are not 7 bytes apart.  This should fix
floppy probing on such systems.  (We handle the case of adjusting for
a start of 0x3f2 -> 0x3f0 separately, although that code should still be
checked if there are still floppy problems for others.)

Tested by: Sarunas Vancevicius <vsarunas_at_eircom.net>
MFC after: 3 days

20 years agoBack out uipc_socket.c:1.208, as it incorrectly assumes that all
rwatson [Fri, 20 Aug 2004 16:24:23 +0000 (16:24 +0000)]
Back out uipc_socket.c:1.208, as it incorrectly assumes that all
sockets are connection-oriented for the purposes of kqueue
registration.  Since UDP sockets aren't connection-oriented, this
appeared to break a great many things, such as RPC-based
applications and services (i.e., NFS).  Since jmg isn't around I'm
backing this out before too many more feet are shot, but intend to
investigate the right solution with him once he's available.

Apologies to: jmg
Discussed with: imp, scottl

20 years agoRemove unused file.
phk [Fri, 20 Aug 2004 15:15:48 +0000 (15:15 +0000)]
Remove unused file.

20 years agoRewrite of the floppy driver to make it MPsafe & GEOM friendly:
phk [Fri, 20 Aug 2004 15:14:25 +0000 (15:14 +0000)]
Rewrite of the floppy driver to make it MPsafe & GEOM friendly:

Centralize the fdctl_wr() function by adding the offset in
the resource to the softc structure.

Bugfix: Read the drive-change signal from the correct place:
same place as the ctl register.

Remove the cdevsw{} related code and implement a GEOM class.

Ditch the state-engine and park a thread on each controller
to service the queue.

Make the interrupt FAST & MPSAFE since it is just a simple
wakeup(9) call.

Rely on a per controller mutex to protect the bioqueues.
Grab GEOMs topology lock when we have to and Giant when
ISADMA needs it.  Since all access to the hardware is
isolated in the per controller thread, the rest of the
driver is lock & Giant free.

Create a per-drive queue where requests are parked while
the motor spins up.  When the motor is running the requests
are purged to the per controller queue.  This allows
requests to other drives to be serviced during spin-up.

Only setup the motor-off timeout when we finish the last
request on the queue and cancel it when a new request
arrives.  This fixes the bug in the old code where the motor
turned off while we were still retrying a request.

Make the "drive-change" work reliably.  Probe the drive on
first opens.  Probe with a recal and a seek to cyl=1 to
reset the drive change line and check again to see if we
have a media.

When we see the media disappear we destroy the geom provider,
create a new one, and flag that autodetection should happen
next time we see a media (unless a specific format is configured).

Add sysctl tunables for a lot of drive related parameters.
If you spend a lot of time waiting for floppies you can
grab the i82078 pdf from Intels web-page and try tuning
these.

Add sysctl debug.fdc.debugflags which will enable various
kinds of debugging printfs.

Add central definitions of our well known floppy formats.

Simplify datastructures for autoselection of format and
call the code at the right times.

Bugfix: Remove at least one piece of code which would have
made 2.88M floppies not work.

Use implied seeks on enhanced controllers.

Use multisector transfers on all controllers.  Increase
ISADMA bounce buffers accordingly.

Fall back to single sector when retrying.  Reset retry count
on every successful transaction.

Sort functions in a more sensible order and generally tidy
up a fair bit here and there.

Assorted related fixes and adjustments in userland utilities.

WORKAROUNDS:
Do allow r/w opens of r/o media but refuse actual write
operations.  This is necessary until the p4::phk_bufwork
branch gets integrated (This problem relates to remounting
not reopening devices, see sys/*/*/${fs}_vfsops.c for details).

Keep PC98's private copy of the old floppy driver compiling
and presumably working (see below).

TODO (planned)

Move probing of drives until after interrupts/timeouts work
(like for ATA/SCSI drives).

TODO (unplanned)

This driver should be made to work on PC98 as well.

Test on YE-DATA PCMCIA floppy drive.

Fix 2.88M media.

This is a MT5 candidate (depends on the bioq_takefirst() addition).

20 years agoImmortalize the words of a -current reader.
des [Fri, 20 Aug 2004 13:39:28 +0000 (13:39 +0000)]
Immortalize the words of a -current reader.

20 years ago- fflush() standard output before fork()ing.
pjd [Fri, 20 Aug 2004 12:49:09 +0000 (12:49 +0000)]
- fflush() standard output before fork()ing.
- Remove redundant 'break'.

20 years agoAdd the raidtest tool, which can be used for performance tests of storage devices.
pjd [Fri, 20 Aug 2004 12:02:34 +0000 (12:02 +0000)]
Add the raidtest tool, which can be used for performance tests of storage devices.
It uses random offsets, random requests size and random operation type (READ or
WRITE). It also allows to run many processes to send I/O requests in parallel.

20 years agoAdd my birthday.
lesi [Fri, 20 Aug 2004 12:01:19 +0000 (12:01 +0000)]
Add my birthday.

Approved by: anholt (mentor)

20 years agoRemove a couple of warnings and tweak an error message.
des [Fri, 20 Aug 2004 10:23:22 +0000 (10:23 +0000)]
Remove a couple of warnings and tweak an error message.

20 years agoCopy open_locked() from tinderbox.pl and use it to optionally acquire a
des [Fri, 20 Aug 2004 08:10:30 +0000 (08:10 +0000)]
Copy open_locked() from tinderbox.pl and use it to optionally acquire a
lock file upon startup.  If this fails, tbmaster will simply terminate.