]> git.neil.brown.name Git - history.git/log
history.git
23 years ago[PATCH] ia64: fsys-version of gettimeofday()
Louis Yu-Kiu Kwan [Thu, 27 Feb 2003 02:02:41 +0000 (18:02 -0800)]
[PATCH] ia64: fsys-version of gettimeofday()

This version executes in around 300 cycles on Itanium I (down from 900
or so for the original version), and so can be said to have
microsecond precision.

23 years agoia64: Don't output backspaces in palinfo output.
David Mosberger [Mon, 24 Feb 2003 04:01:19 +0000 (20:01 -0800)]
ia64: Don't output backspaces in palinfo output.

23 years agoia64: Fix do_gettimeoffset() to not update last_nsec_offset with (potentially)
David Mosberger [Wed, 12 Feb 2003 02:00:12 +0000 (18:00 -0800)]
ia64: Fix do_gettimeoffset() to not update last_nsec_offset with (potentially)
invalid values.

23 years agoia64: Sync up with 2.5.60.
David Mosberger [Tue, 11 Feb 2003 16:11:41 +0000 (08:11 -0800)]
ia64: Sync up with 2.5.60.

23 years ago[PATCH] ia64: perfmon patch for 2.5.59
Stéphane Eranian [Tue, 11 Feb 2003 08:37:26 +0000 (00:37 -0800)]
[PATCH] ia64: perfmon patch for 2.5.59

Here is the patch that matches the changes I made for RHAS w.r.t
perfmon and the O(1) scheduler. This fixes the deadlocks we were
seing in RHAS and therefore on 2.5 as well.

The key change is that SIGPROF notifications are now delivered
in the perfmon kernel exit handler (called from entry.S).
At that point, we know none of the runqueue locks are held.

I have run the same test suite I used on RHAS overnight and no
sign of deadlocks.

23 years agoia64: Fix typo in #error message of page-fault handler.
David Mosberger [Tue, 11 Feb 2003 03:00:52 +0000 (19:00 -0800)]
ia64: Fix typo in #error message of page-fault handler.

23 years agoMerge tiger.hpl.hp.com:/data1/bk/vanilla/linux-2.5
David Mosberger [Tue, 11 Feb 2003 02:46:59 +0000 (18:46 -0800)]
Merge tiger.hpl.hp.com:/data1/bk/vanilla/linux-2.5
into tiger.hpl.hp.com:/data1/bk/lia64/to-linus-2.5

23 years agoia64: Make signal deliver work when the current register frame is
David Mosberger [Mon, 10 Feb 2003 10:28:07 +0000 (02:28 -0800)]
ia64: Make signal deliver work when the current register frame is
incomplete (as a result of a faulting mandatory RSE load).

23 years agoia64: Minor cleanups.
David Mosberger [Mon, 10 Feb 2003 10:22:30 +0000 (02:22 -0800)]
ia64: Minor cleanups.

23 years ago[PATCH] checker bounds/limits fixes
Randy Dunlap [Mon, 10 Feb 2003 08:39:58 +0000 (00:39 -0800)]
[PATCH] checker bounds/limits fixes

Part of the bounds checking bugs found by the Stanford checker.

The 4 fixes below have been acked by their maintainers.

23 years ago[PATCH] de4x5 compile fix
Andrew Morton [Mon, 10 Feb 2003 08:29:58 +0000 (00:29 -0800)]
[PATCH] de4x5 compile fix

Move the definition of version[] down to where __initdata has been defined.

23 years ago[PATCH] ncpfs compile fix
Andrew Morton [Mon, 10 Feb 2003 08:29:49 +0000 (00:29 -0800)]
[PATCH] ncpfs compile fix

Patch from Joel Becker <Joel.Becker@oracle.com>

The task_struct->sig -> task_struct->signal bits.

23 years ago[PATCH] Fix synchronous writers to wait properly for the result
Andrew Morton [Mon, 10 Feb 2003 08:29:41 +0000 (00:29 -0800)]
[PATCH] Fix synchronous writers to wait properly for the result

Mikulas Patocka <mikulas@artax.karlin.mff.cuni.cz> points out a bug in
ll_rw_block() usage.

Typical usage is:

mark_buffer_dirty(bh);
ll_rw_block(WRITE, 1, &bh);
wait_on_buffer(bh);

the problem is that if the buffer was locked on entry to this code sequence
(due to in-progress I/O), ll_rw_block() will not wait, and start new I/O.  So
this code will wait on the _old_ I/O, and will then continue execution,
leaving the buffer dirty.

It turns out that all callers were only writing one buffer, and they were all
waiting on that writeout.  So I added a new sync_dirty_buffer() function:

void sync_dirty_buffer(struct buffer_head *bh)
{
lock_buffer(bh);
if (test_clear_buffer_dirty(bh)) {
get_bh(bh);
bh->b_end_io = end_buffer_io_sync;
submit_bh(WRITE, bh);
} else {
unlock_buffer(bh);
}
}

which allowed a fair amount of code to be removed, while adding the desired
data-integrity guarantees.

UFS has its own wrappers around ll_rw_block() which got in the way, so this
operation was open-coded in that case.

23 years agoMerge tetrachloride.(none):/mnt/stuff/kernel/2.5/bk-linus
Dave Jones [Mon, 10 Feb 2003 18:43:40 +0000 (17:43 -0100)]
Merge tetrachloride.(none):/mnt/stuff/kernel/2.5/bk-linus
into tetrachloride.(none):/mnt/stuff/kernel/2.5/agpgart-respin

23 years ago[AGPGART] Enable support for VIA PLE133 chipset
Dave Jones [Mon, 10 Feb 2003 17:49:26 +0000 (16:49 -0100)]
[AGPGART] Enable support for VIA PLE133 chipset

23 years ago[AGPGART] Remove pointless enums from VIA GART driver.
Dave Jones [Mon, 10 Feb 2003 17:49:13 +0000 (16:49 -0100)]
[AGPGART] Remove pointless enums from VIA GART driver.

23 years ago[AGPGART] Merge VIA KT400 AGP3 support into main via-agp module.
Dave Jones [Mon, 10 Feb 2003 17:45:11 +0000 (16:45 -0100)]
[AGPGART] Merge VIA KT400 AGP3 support into main via-agp module.

This removes lots of annoying problems trying to prevent both modules
from being loaded, and also shares quite a bit of code.
CONFIG_AGP3 will disable AGP3 mode operation of KT400s.

23 years ago[AGPGART] Fix missed agp_bridge conversion that caused oops.
Dave Jones [Mon, 10 Feb 2003 16:53:44 +0000 (15:53 -0100)]
[AGPGART] Fix missed agp_bridge conversion that caused oops.

23 years ago[AGPGART] Additional VIA ids.
Dave Jones [Mon, 10 Feb 2003 16:53:37 +0000 (15:53 -0100)]
[AGPGART] Additional VIA ids.

23 years ago[AGPGART] Add extra VIA GART IDs.
Dave Jones [Mon, 10 Feb 2003 16:53:30 +0000 (15:53 -0100)]
[AGPGART] Add extra VIA GART IDs.

Based upon information from VIA, this also adds a bunch of placeholder
entries that will get filled in over time when they have been proven
to work with the code with no extra modification.

23 years ago[AGPGART] First step towards multiple AGP buses.
Dave Jones [Mon, 10 Feb 2003 16:53:24 +0000 (15:53 -0100)]
[AGPGART] First step towards multiple AGP buses.

The AGP3 spec allows for >1 AGP bus. This is the first of several patches
from Jeff Hartmann towards a context-using agp_bridge, by replacing
agp_bridge.foo accesses with agp_bridge->foo accesses. For now, there
should be no functional differences, as there is still only a single
agp_bridge_data struct defined.

23 years ago[AGPGART] alpha agp infrastructure
Dave Jones [Mon, 10 Feb 2003 16:53:17 +0000 (15:53 -0100)]
[AGPGART] alpha agp infrastructure

23 years ago[AGPGART] kt400's enable routine can't be __init
Dave Jones [Mon, 10 Feb 2003 16:53:09 +0000 (15:53 -0100)]
[AGPGART] kt400's enable routine can't be __init

23 years ago[AGPGART] VIA KT400 Aperture size is 12 bit in AGP3 mode.
Dave Jones [Mon, 10 Feb 2003 16:53:02 +0000 (15:53 -0100)]
[AGPGART] VIA KT400 Aperture size is 12 bit in AGP3 mode.

23 years ago[AGPGART] More failure path sanity checking.
Dave Jones [Mon, 10 Feb 2003 16:52:55 +0000 (15:52 -0100)]
[AGPGART] More failure path sanity checking.

23 years ago[AGPGART] Add ident for VIA KT400 in disguise as a KT266
Dave Jones [Mon, 10 Feb 2003 16:52:48 +0000 (15:52 -0100)]
[AGPGART] Add ident for VIA KT400 in disguise as a KT266

23 years ago[AGPGART] Handle failure during initialisation more gracefully.
Dave Jones [Mon, 10 Feb 2003 16:52:41 +0000 (15:52 -0100)]
[AGPGART] Handle failure during initialisation more gracefully.

23 years ago[AGPGART] Handle the "KT400 in disguise as a KT266" case.
Dave Jones [Mon, 10 Feb 2003 16:52:34 +0000 (15:52 -0100)]
[AGPGART] Handle the "KT400 in disguise as a KT266" case.

23 years ago[AGPGART] Don't oops when deregistering failed to init agp modules.
Dave Jones [Mon, 10 Feb 2003 16:52:27 +0000 (15:52 -0100)]
[AGPGART] Don't oops when deregistering failed to init agp modules.

23 years ago[AGPGART] Fix same logic bug in KT400 mode determination.
Dave Jones [Mon, 10 Feb 2003 16:52:20 +0000 (15:52 -0100)]
[AGPGART] Fix same logic bug in KT400 mode determination.

23 years ago[AGPGART] Fix up lots of 'comparison between signed and unsigned' warnings
Dave Jones [Mon, 10 Feb 2003 16:52:09 +0000 (15:52 -0100)]
[AGPGART] Fix up lots of 'comparison between signed and unsigned' warnings

23 years agoLinux v2.5.60 v2.5.60
Linus Torvalds [Mon, 10 Feb 2003 02:32:47 +0000 (18:32 -0800)]
Linux v2.5.60

23 years ago[PATCH] Finish job of trimming ".o" module extension in Kconfig files
Steven Cole [Mon, 10 Feb 2003 01:40:33 +0000 (17:40 -0800)]
[PATCH] Finish job of trimming ".o" module extension in Kconfig files

Most of the instances of <module>.o in Kconfig files have had the ".o"
extension trimmed.  This change came from GertJan Spoelman through Rusty
"Trivial" Russell.

However, there are a few files that didn't get trimmed.  This brings
them line with the rest of the tree.

23 years agoWake up a stopped task _after_ having marked the SIGCONT pending,
Linus Torvalds [Sun, 9 Feb 2003 23:58:04 +0000 (15:58 -0800)]
Wake up a stopped task _after_ having marked the SIGCONT pending,
so that there isn't any window for running before the signal
handler has been invoced.

23 years agoCreate "wake_up_state()" macro that selectively wakes up processes only
Linus Torvalds [Sun, 9 Feb 2003 13:32:37 +0000 (05:32 -0800)]
Create "wake_up_state()" macro that selectively wakes up processes only
from certain states.

This simplifies "default_wake_function()", and makes it possible for
signal handling to wake up only the processes it _should_ wake up
without races.

23 years ago[PATCH] Change all <module>.o to .ko in Kconfig files
Rusty Russell [Sun, 9 Feb 2003 11:03:12 +0000 (03:03 -0800)]
[PATCH] Change all <module>.o to .ko in Kconfig files

From:  GertJan Spoelman <kl@gjs.cc>

  OK, here is a new patch, I edited the old patch and took out the .ko's
  so now the extension is trimmed instead.

23 years ago[PATCH] Write with buffer>2GB returns broken errno (2)
Rusty Russell [Sun, 9 Feb 2003 11:01:17 +0000 (03:01 -0800)]
[PATCH] Write with buffer>2GB returns broken errno (2)

[ Acked by AKPM --RR ]
From:  Kazuto MIYOSHI <miyoshi@hpc.bs1.fc.nec.co.jp>

  On 64-bit platforms, issuing write(2) with buffer larger than
  2GB will return -1 and broken errno (such as 2147483640)
  Requested data itself is written correctly.

  That is because generic_file_write() and other relating functions
  store 'ssize_t written' into 'int err'. Written byte is trimmed to
  int and then sign-extended to a negative ssize_t value, which
  wrongly indicates an error.

  (On 64bit platform, current glibc defines SSIZE_MAX as 'LONG_MAX')

23 years ago[PATCH] add two help texts to drivers_media_video_Kconfig
Rusty Russell [Sun, 9 Feb 2003 11:01:09 +0000 (03:01 -0800)]
[PATCH] add two help texts to drivers_media_video_Kconfig

From:  Steven Cole <elenstev@mesatop.com>

  Here are some help texts from 2.4.21-pre3 Configure.help which are
  needed in 2.5.59 drivers/media/video/Kconfig.

23 years ago[PATCH] Change "char version" to initdata in drivers_net_tulip_de4x5.c
Rusty Russell [Sun, 9 Feb 2003 11:00:57 +0000 (03:00 -0800)]
[PATCH] Change "char version" to initdata in drivers_net_tulip_de4x5.c

From:  Pablo Menichini <pablo@menichini.com.ar>

23 years ago[PATCH] add four help texts to drivers_char_watchdog_Kconfig
Rusty Russell [Sun, 9 Feb 2003 11:00:51 +0000 (03:00 -0800)]
[PATCH] add four help texts to drivers_char_watchdog_Kconfig

From:  Steven Cole <elenstev@mesatop.com>

  Here are some help texts from 2.4.21-pre3 Configure.help which are
  needed in 2.5.59 drivers/char/watchdog/Kconfig.

23 years ago[PATCH] add six help texts to drivers_ide_Kconfig
Rusty Russell [Sun, 9 Feb 2003 11:00:43 +0000 (03:00 -0800)]
[PATCH] add six help texts to drivers_ide_Kconfig

From:  Steven Cole <elenstev@mesatop.com>

  Here are some help texts from 2.4.21-pre3 Configure.help which are
  needed in 2.5.59 drivers/ide/Kconfig.

23 years ago[PATCH] swsusp: do not panic on bad signature with noresume
Rusty Russell [Sun, 9 Feb 2003 11:00:36 +0000 (03:00 -0800)]
[PATCH] swsusp: do not panic on bad signature with noresume

From:  Pavel Machek <pavel@ucw.cz>

  This patch makes kernel ignore bad signature on suspend device when
  "noresume" is given, and cleans things up a little bit. Please apply,

23 years ago[PATCH] fix linewrap in Documentation_arm_SA1100_CERF
Rusty Russell [Sun, 9 Feb 2003 11:00:30 +0000 (03:00 -0800)]
[PATCH] fix linewrap in Documentation_arm_SA1100_CERF

[ Verified that no text changed with tr and cmp --RR ]
From:  ookhoi@humilis.net

  With this patch I tried to make Documentation/arm/SA1100/CERF more
  readible by fixing the linewrap.

23 years ago[PATCH] fix spelling of kernel in arch_v850_kernel_mach.h
Rusty Russell [Sun, 9 Feb 2003 11:00:22 +0000 (03:00 -0800)]
[PATCH] fix spelling of kernel in arch_v850_kernel_mach.h

From:  Steven Cole <elenstev@mesatop.com>

  This fixes the only instance of "kernal" in 2.5.59.

23 years ago[PATCH] Squash unused function in fs_nfs_mount_clnt.c
Rusty Russell [Sun, 9 Feb 2003 11:00:16 +0000 (03:00 -0800)]
[PATCH] Squash unused function in fs_nfs_mount_clnt.c

From:  David Gibson <david@gibson.dropbear.id.au>

  is never used, so this patch removes it.

23 years ago[PATCH] nfs_write.c warning
Rusty Russell [Sun, 9 Feb 2003 11:00:08 +0000 (03:00 -0800)]
[PATCH] nfs_write.c warning

From:  William Lee Irwin III <wli@holomorphy.com>

  This trivially corrects an unused variable warning in nfs/write.c:

23 years ago[PATCH] remove LinuxVersionCode from de4x5.h
Rusty Russell [Sun, 9 Feb 2003 11:00:00 +0000 (03:00 -0800)]
[PATCH] remove LinuxVersionCode from de4x5.h

From:  Adrian Bunk <bunk@fs.tum.de>

  drivers/net/tulip/de4x5.h in 2.5.54 contains a definition of
  LinuxVersionCode. LinuxVersionCode isn't used and it's anyway obsoleted
  by KERNEL_VERSION in version.h.

23 years ago[PATCH] Kill unused code
Rusty Russell [Sun, 9 Feb 2003 10:59:52 +0000 (02:59 -0800)]
[PATCH] Kill unused code

From:  Pavel Machek <pavel@ucw.cz>

  Second part of this patch never got in (and I was told it was not bug
  in ASUS but in linux), so it is useless junk... Please apply,

23 years ago[PATCH] Fix return code of init_module in drivers_net_arlan.c (2)
Rusty Russell [Sun, 9 Feb 2003 10:59:46 +0000 (02:59 -0800)]
[PATCH] Fix return code of init_module in drivers_net_arlan.c (2)

From:  Pablo Menichini <pablo@menichini.com.ar>

   This patch returns correct error codes if init_modules fail.
  Because of this, we can take the printks indicating the error as these
  corrected error codes return miningfull information.

23 years ago[PATCH] fix typo of members name in drivers_mtd_ftl.c
Rusty Russell [Sun, 9 Feb 2003 10:59:38 +0000 (02:59 -0800)]
[PATCH] fix typo of members name in drivers_mtd_ftl.c

From:  Pablo Menichini <pablo@menichini.com.ar>

23 years ago[PATCH] RTC alarm and wildcards
Rusty Russell [Sun, 9 Feb 2003 10:59:32 +0000 (02:59 -0800)]
[PATCH] RTC alarm and wildcards

(Included in 2.4)
From:  Paul Gortmaker <p_gortmaker@yahoo.com>

  Summary: Wildcards in RTC alarm settings failed to work

  Description:
   The RTC has provision for wildcards when setting the alarm; to
   use them you have to write a value higher than 0xc0 to the
   appropriate hr/min/sec entry.  The driver used 0xff, which is
   fine, but it mistakenly fed the 0xff through BIN_TO_BCD before
   writing them (which is < 0xc0) and so wildcards didn't work.
   (Thanks to Gerhard Kurz for reporting the bug.)

23 years ago[PATCH] Memory leak in drivers_net_arlan.c (1)
Rusty Russell [Sun, 9 Feb 2003 10:59:23 +0000 (02:59 -0800)]
[PATCH] Memory leak in drivers_net_arlan.c (1)

From:  Pablo Menichini <pablo@menichini.com.ar>

23 years ago[PATCH] make i2c-core driver_lock static
Rusty Russell [Sun, 9 Feb 2003 10:59:12 +0000 (02:59 -0800)]
[PATCH] make i2c-core driver_lock static

From:  Muli Ben-Yehuda <mulix@mulix.org>

  The i2c driver_lock is needlessly exported.
  This makes it static.

23 years ago[PATCH] Remove compile warning from fs_xfs_support_move.c
Rusty Russell [Sun, 9 Feb 2003 10:59:05 +0000 (02:59 -0800)]
[PATCH] Remove compile warning from fs_xfs_support_move.c

From:  Bob Miller <rem@osdl.org>

  Include string.h to remove a compiler warning.

23 years ago[PATCH] add two help texts to drivers_i2c_Kconfig
Rusty Russell [Sun, 9 Feb 2003 10:58:57 +0000 (02:58 -0800)]
[PATCH] add two help texts to drivers_i2c_Kconfig

From:  Steven Cole <elenstev@mesatop.com>

  Here are some help texts from 2.4.21-pre3 Configure.help which are
  needed in 2.5.59 drivers/i2c/Kconfig.

23 years ago[PATCH] Change "char _version" to "char in drivers_net_mac8390.c
Rusty Russell [Sun, 9 Feb 2003 10:58:49 +0000 (02:58 -0800)]
[PATCH] Change "char _version" to "char in drivers_net_mac8390.c

From:  Pablo Menichini <pablo@menichini.com.ar>

23 years ago[PATCH] scripts_ver_linux
Rusty Russell [Sun, 9 Feb 2003 10:58:40 +0000 (02:58 -0800)]
[PATCH] scripts_ver_linux

From:  Frank Davis <fdavis@si.rr.com>

     The ver_linux script is still using rmmod to determine
  module-init-tools version. The following patch uses depmod,
  which produces the appropriate result.

23 years ago[PATCH] add one help text to drivers_atm_Kconfig
Rusty Russell [Sun, 9 Feb 2003 10:58:33 +0000 (02:58 -0800)]
[PATCH] add one help text to drivers_atm_Kconfig

From:  Steven Cole <elenstev@mesatop.com>

  Here is a help text from 2.4.21-pre4 Configure.help which is
  needed in 2.5.59 drivers/atm/Kconfig.

23 years ago[PATCH] Change "char _version" to "char in drivers_lcs.c
Rusty Russell [Sun, 9 Feb 2003 10:58:26 +0000 (02:58 -0800)]
[PATCH] Change "char _version" to "char in drivers_lcs.c

From:  Pablo Menichini <pablo@menichini.com.ar>

23 years ago[PATCH] parport_pc and !CONFIG_PNP
Rusty Russell [Sun, 9 Feb 2003 10:58:19 +0000 (02:58 -0800)]
[PATCH] parport_pc and !CONFIG_PNP

From:  Geert Uytterhoeven <geert@linux-m68k.org>

  parport_pc_pnp_driver is const if !CONFIG_PNP, while pnp_register_driver()
  takes a non-const pointer as parameter.

  An alternative fix is to change the prototype of the dummy
  pnp_register_driver(), but this may affect other drivers.

23 years ago[PATCH] remove check_region from drivers_net_irda_irport.c
Rusty Russell [Sun, 9 Feb 2003 10:58:12 +0000 (02:58 -0800)]
[PATCH] remove check_region from drivers_net_irda_irport.c

From:  william stinson <wstinson@wanadoo.fr>

  this patch for drivers/net/irda/irport.c IRDA driver removes one call
  to check_region using request_region instead.  The patch also moves
  the call to request_region to before the allocation of the driver
  instance.

23 years ago[PATCH] fix comment in module.c
Rusty Russell [Sun, 9 Feb 2003 10:58:06 +0000 (02:58 -0800)]
[PATCH] fix comment in module.c

From:  John Levon <levon@movementarian.org>

23 years ago[PATCH] Remove superflous 'either'
Rusty Russell [Sun, 9 Feb 2003 10:57:59 +0000 (02:57 -0800)]
[PATCH] Remove superflous 'either'

From:  John Bradford <john@grabjohn.com>

23 years ago[PATCH] Documentation_Changes
Rusty Russell [Sun, 9 Feb 2003 10:57:52 +0000 (02:57 -0800)]
[PATCH] Documentation_Changes

From:  Frank Davis <fdavis@si.rr.com>

  this was already mentioned on l-k by ramune@net-ronin.org, but isn't in
  2.5.59. Placing on the trivial queue for inclusion.

23 years agoMerge http://linux-isdn.bkbits.net/linux-2.5.make
Linus Torvalds [Sun, 9 Feb 2003 10:47:26 +0000 (02:47 -0800)]
Merge http://linux-isdn.bkbits.net/linux-2.5.make
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoFix missing break, causing sigprocmask(SIG_SETMASK ...) to always
Linus Torvalds [Sun, 9 Feb 2003 10:38:19 +0000 (02:38 -0800)]
Fix missing break, causing sigprocmask(SIG_SETMASK ...) to always
return an error.

Interestingly, nobody much seems to care. Apparently few programs
check the error value.

23 years agoMake sigprocmask() available to kernel threads too, since a lot of
Linus Torvalds [Sun, 9 Feb 2003 10:17:26 +0000 (02:17 -0800)]
Make sigprocmask() available to kernel threads too, since a lot of
them do want to temporarily block signals.

Kernel users can also block signals that are normally unblockable
to user space, ie SIGKILL and SIGSTOP.

Make nfsd and autofs use the new interface, as an example to others.

23 years ago[PATCH] simple EXT2 patch
Mikulas Patocka [Sun, 9 Feb 2003 09:13:18 +0000 (01:13 -0800)]
[PATCH] simple EXT2 patch

Do not crash on null pointer dereference, if cannot reread superblock.

23 years ago[PATCH] zap_other_threads() needs tasklist_lock held
Roland McGrath [Sun, 9 Feb 2003 09:11:21 +0000 (01:11 -0800)]
[PATCH] zap_other_threads() needs tasklist_lock held

23 years agoMerge bk://kernel.bkbits.net/davem/net-2.5
Linus Torvalds [Sun, 9 Feb 2003 09:01:59 +0000 (01:01 -0800)]
Merge bk://kernel.bkbits.net/davem/net-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years ago[PATCH] lock group_send_sig_info() properly
Ingo Molnar [Sun, 9 Feb 2003 09:00:12 +0000 (01:00 -0800)]
[PATCH] lock group_send_sig_info() properly

  - a read_lock(&tasklist_lock) is missing around the group_send_sig_info()
    in send_sig_info().

23 years ago[PATCH] Lock session and group ID setting
Ingo Molnar [Sun, 9 Feb 2003 08:59:53 +0000 (00:59 -0800)]
[PATCH] Lock session and group ID setting

 - session-IDs and group-IDs are set outside the tasklist lock. This
   causes breakage in the USB code. The correct fix is to do this:

I introduced the bug with the new pidhash.

23 years agoMerge http://linux-scsi.bkbits.net/scsi-for-linus-2.5
Linus Torvalds [Sun, 9 Feb 2003 08:53:26 +0000 (00:53 -0800)]
Merge http://linux-scsi.bkbits.net/scsi-for-linus-2.5
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoMerge http://linux-sound.bkbits.net/linux-sound
Linus Torvalds [Sun, 9 Feb 2003 08:52:49 +0000 (00:52 -0800)]
Merge http://linux-sound.bkbits.net/linux-sound
into home.transmeta.com:/home/torvalds/v2.5/linux

23 years agoppc64: Fix nasty bug in cmpxchg where we would sign extend the old value.
Anton Blanchard [Mon, 10 Feb 2003 21:45:43 +0000 (08:45 +1100)]
ppc64: Fix nasty bug in cmpxchg where we would sign extend the old value.

23 years agoppc64: update for signal changes
Anton Blanchard [Mon, 10 Feb 2003 21:02:47 +0000 (08:02 +1100)]
ppc64: update for signal changes

23 years agoMerge samba.org:/scratch/anton/linux-2.5
Anton Blanchard [Mon, 10 Feb 2003 20:56:04 +0000 (07:56 +1100)]
Merge samba.org:/scratch/anton/linux-2.5
into samba.org:/scratch/anton/sfr

23 years agoMerge samba.org:/scratch/anton/linux-2.5
Anton Blanchard [Mon, 10 Feb 2003 07:17:06 +0000 (18:17 +1100)]
Merge samba.org:/scratch/anton/linux-2.5
into samba.org:/scratch/anton/sfr

23 years agoALSA update
Jaroslav Kysela [Sun, 9 Feb 2003 20:47:59 +0000 (21:47 +0100)]
ALSA update
  - moved inclusion of <linux/interrupt.h> from <sound/asound.h> to <sound/timer.h>
  - pmac driver - removed beep stuff for 2.5 kernels
  - USB driver - fixed compilation

23 years agoresolve conflicts
Anton Blanchard [Sun, 9 Feb 2003 10:51:36 +0000 (21:51 +1100)]
resolve conflicts

23 years ago[SCSI] Add missing list head init of cmd_list
James Bottomley [Sun, 9 Feb 2003 06:09:43 +0000 (00:09 -0600)]
[SCSI] Add missing list head init of cmd_list

23 years ago[SCSI] Remove 53c7,8xx since we have plenty of alternatives.
James Bottomley [Sun, 9 Feb 2003 05:01:41 +0000 (23:01 -0600)]
[SCSI] Remove 53c7,8xx since we have plenty of alternatives.

We have 53c700.c and 53c7xx for the 7xx series and
ncr53c8xx for the 720.  The sym53c8xx_2 covers all the 8xx chips.

23 years ago[SCSI] add commands at the tail of cmd_list
Luben Tuikov [Sun, 9 Feb 2003 04:56:35 +0000 (22:56 -0600)]
[SCSI] add commands at the tail of cmd_list

It's probably going to be a fifo, so it should be more efficient
for taking them off again

23 years ago[SCSI] Migrate sim710 to 53c700 chip driver
James Bottomley [Sun, 9 Feb 2003 04:07:51 +0000 (22:07 -0600)]
[SCSI] Migrate sim710 to 53c700 chip driver

This should add synchronous support and Tagged Command Queueing.
At the moment, it cuts down on the number of command line options,
but we can add those back in later.

This patch also migrates the driver to the new device model for
both MCA and EISA.

23 years agoRestore device command queue functionality
James Bottomley [Sun, 9 Feb 2003 03:36:18 +0000 (21:36 -0600)]
Restore device command queue functionality

The recent slab allocation changes mean that we no longer keep a
permanent list of commands on the device_queue list.  However,
certain pieces of SCSI code relied on being able to traverse this
list to find details of all outstanding commands (the error handler
being the prime example).  This code adds back a new dynamic cmd_list
which keeps the list of commands currently allocated to the device.
Since the list is dynamic, it is protected by a lock (list_lock).

23 years agoMerge davem@nuts.ninka.net:/home/davem/src/BK/net-2.5
David S. Miller [Sat, 8 Feb 2003 18:48:18 +0000 (10:48 -0800)]
Merge davem@nuts.ninka.net:/home/davem/src/BK/net-2.5
into kernel.bkbits.net:/home/davem/net-2.5

23 years ago[PATCH] CONFIG_PREEMPT fix of do_debug()
Jamie Lokier [Sat, 8 Feb 2003 18:41:34 +0000 (10:41 -0800)]
[PATCH] CONFIG_PREEMPT fix of do_debug()

If CONFIG_PREEMPT is enabled, and the kernel is preempted just before
do_debug() has a chance to save the debug register values, DR6 could be
read from the wrong CPU.

It is exactly the same problem as reading %cr2 in the page fault
handler.  Same fix: make the handler a interrupt gate, and enable
interrupts only once safe.

23 years ago[PATCH] Fix compile warning for 'sys_exit_group'
Andrew Morton [Sat, 8 Feb 2003 18:37:45 +0000 (10:37 -0800)]
[PATCH] Fix compile warning for 'sys_exit_group'

sys_exit_group() doesn't return any value, and obviously cannot.

So don't make the compiler unhappy about it by claiming it does.

23 years ago[PATCH] Fix Alt-SysRQ-T status, and comment
Russell King [Sat, 8 Feb 2003 18:37:37 +0000 (10:37 -0800)]
[PATCH] Fix Alt-SysRQ-T status, and comment

Fix wrong order of process status. It's

#define TASK_RUNNING            0
#define TASK_INTERRUPTIBLE      1
#define TASK_UNINTERRUPTIBLE    2
#define TASK_STOPPED            4
#define TASK_ZOMBIE             8
#define TASK_DEAD               16

but SysRQ printout routines switch stopped and zombie around.

So, for one more time, here's another mailing of the same patch to fix
this brokenness.  In addition, fix the wrong comment in fs/proc/array.c

23 years agoMore signal handling fixups for the threaded signal fix upheavals.
Linus Torvalds [Sat, 8 Feb 2003 14:02:27 +0000 (06:02 -0800)]
More signal handling fixups for the threaded signal fix upheavals.

This fixes the signal code to not wake up threads with blocked signals,
especially noticeable with kernel threads that may not be able to handle
signals at all.

We also don't unnecessarily wake processes in TASK_UNINTERRUPTIBLE.

23 years ago[PATCH] missing sound include file
Chris Wedgwood [Sat, 8 Feb 2003 13:58:32 +0000 (05:58 -0800)]
[PATCH] missing sound include file

Sound drivers need <linus/interrupt.h> for tasklets

23 years agoALSA update
Jaroslav Kysela [Sat, 8 Feb 2003 22:33:57 +0000 (23:33 +0100)]
ALSA update
  - emu10k1 - fixed makefile to not build synth module when emu10k1 is not selected

23 years agoALSA update
Jaroslav Kysela [Sat, 8 Feb 2003 22:11:54 +0000 (23:11 +0100)]
ALSA update
  - cmipci driver cleanups (ac3 & surround)
  - replaced snd_dma_residue() with snd_dma_pointer()
  - GCC 3.3 warnings removal
  - timer interface
    - recoded using tasklet
    - improved slave timer locking (should be much faster)
    - added async support
  - improved ioctl32 wrapper functions
  - fixed Makefile problems (synth modules were build for not selected driver)
  - AC97 codec
    - improved SPSA control
    - moved reset function outside the main init code
    - improved ALC650 initialization
  - USB driver
    - added quirk for Roland XV-2020

23 years agokbuild: Add a bug trap for people playing with SUBDIRS too much
Kai Germaschewski [Sat, 8 Feb 2003 08:33:21 +0000 (02:33 -0600)]
kbuild: Add a bug trap for people playing with SUBDIRS too much

If SUBDIRS is set manually on the command line, the contents of
.tmp_versions are not deleted before descending and can accumulate
stale entries. Print a warning if that case is detected, but deal with
it gracefully.

23 years agokbuild: Modversions fix
Kai Germaschewski [Sat, 8 Feb 2003 08:30:37 +0000 (02:30 -0600)]
kbuild: Modversions fix

We're still using the old genksyms binary, that's why we have to
postprocess the output to convert it into a linker script - that
postprocessing got confused by "__verify_write".

Kick out the grep, do it all and correctly within sed.

Bug reported by Thomas Molina.

23 years agokbuild: Warn on obsolete export-objs use
Kai Germaschewski [Sat, 8 Feb 2003 08:00:46 +0000 (02:00 -0600)]
kbuild: Warn on obsolete export-objs use

Setting export-objs is not necessary anymore, so warn on encountering it
to prevent it from creeping back in ;)

Also, make the error when we find someone still using O_TARGET non-fatal,
so that people sharing stuff between 2.4 and 2.5 don't have more hassle
than necessary.

23 years agokbuild: Handle external SUBDIRS with modversions
Kai Germaschewski [Sat, 8 Feb 2003 07:59:13 +0000 (01:59 -0600)]
kbuild: Handle external SUBDIRS with modversions

We need to collect a list of all modules during the recursive build. I used
a "touch .tmp_versions/<path/to/module.ko>" to do so, which however doesn't
work so well, when path/to isn't inside the kernel tree.

The best way to build external modules is currently using kbuild by saying
"make SUBDIRS=/some/external/dir modules", which was thus broken. While this
way is not all that optimal and I hope to come up with something better
before 2.6, it works and should keep working, so this patch fixes the usage
above.

Instead of touching files with the entire path added, we just create a
<module>.mod file in $(MODVERDIR) now, and save the path to the module.ko
in it. Since module names are unique, a flat hierarchy is actually fine here.

23 years ago[PATCH] TASK_STOPPED wakeup cleanup
Roland McGrath [Fri, 7 Feb 2003 08:31:37 +0000 (00:31 -0800)]
[PATCH] TASK_STOPPED wakeup cleanup

For handle_stop_signal to do the special case for SIGKILL and have it
work right in all SMP cases (without changing all the existing ptrace
stops), it needs to at least set TIF_SIGPENDING on each thread before
resuming it.

handle_stop_signal addresses a related race for SIGCONT by setting
TIF_SIGPENDING already, so having SIGKILL handled the same way makes
sense.

Now it seems pretty clean to have handle_stop_signal resume threads for
SIGKILL, and have on SIGKILL special case in group_send_sig_info.

There is also an SMP race issue with cases like do_syscall_trace, i.e.
TASK_STOPPED state set without holding the siglock.  So I think
handle_stop_signal should call wake_up_process unconditionally.

23 years ago[PATCH] signal locking update
Chris Wedgwood [Fri, 7 Feb 2003 08:24:40 +0000 (00:24 -0800)]
[PATCH] signal locking update

Accomodate the signal locking moving from "tsk->sig" to "tsk->sighand".

23 years agoHand merge
Daniel Jacobowitz [Fri, 7 Feb 2003 11:39:09 +0000 (06:39 -0500)]
Hand merge