]> git.hungrycats.org Git - linux/log
linux
22 years ago[PATCH] drivers/atm/ambassador.c delousing
Alexander Viro [Thu, 3 Jun 2004 13:38:14 +0000 (06:38 -0700)]
[PATCH] drivers/atm/ambassador.c delousing

Translated to C.  Original was anything but.

BTW, even if they were writing in Pascal, their use of nested functions
would be gratuitous for such a situation.

22 years ago[PATCH] sparse: rest of drivers/block annotation
Alexander Viro [Thu, 3 Jun 2004 13:28:37 +0000 (06:28 -0700)]
[PATCH] sparse: rest of drivers/block annotation

22 years ago[PATCH] sparse: paride annotation
Alexander Viro [Thu, 3 Jun 2004 13:28:26 +0000 (06:28 -0700)]
[PATCH] sparse: paride annotation

22 years ago[PATCH] sparse: cpqarray annotation
Alexander Viro [Thu, 3 Jun 2004 13:28:16 +0000 (06:28 -0700)]
[PATCH] sparse: cpqarray annotation

22 years ago[PATCH] sparse: cciss annotation
Alexander Viro [Thu, 3 Jun 2004 13:28:05 +0000 (06:28 -0700)]
[PATCH] sparse: cciss annotation

cciss annotated, definition of inlined helpers moved up - before their
first use.

22 years ago[PATCH] sparse: DAC960 annotation
Alexander Viro [Thu, 3 Jun 2004 13:27:54 +0000 (06:27 -0700)]
[PATCH] sparse: DAC960 annotation

22 years ago[PATCH] sparse: drivers/pnp annotation
Alexander Viro [Thu, 3 Jun 2004 13:27:43 +0000 (06:27 -0700)]
[PATCH] sparse: drivers/pnp annotation

22 years ago[PATCH] sparse: drivers/md annotation
Alexander Viro [Thu, 3 Jun 2004 13:27:32 +0000 (06:27 -0700)]
[PATCH] sparse: drivers/md annotation

22 years ago[PATCH] sparse: drivers/cdrom annotation
Alexander Viro [Thu, 3 Jun 2004 13:27:21 +0000 (06:27 -0700)]
[PATCH] sparse: drivers/cdrom annotation

22 years ago[PATCH] sparse: drivers/cpufreq annotation
Alexander Viro [Thu, 3 Jun 2004 13:27:10 +0000 (06:27 -0700)]
[PATCH] sparse: drivers/cpufreq annotation

drivers/cpufreq annotated

22 years agosparse: annotate (and comment on) kmod.c user pointer usage
Linus Torvalds [Thu, 3 Jun 2004 12:54:49 +0000 (05:54 -0700)]
sparse: annotate (and comment on) kmod.c user pointer usage

Big comment, because it wasn't clear why this cast was valid.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] sparse: forte annotation
Alexander Viro [Thu, 3 Jun 2004 12:41:31 +0000 (05:41 -0700)]
[PATCH] sparse: forte annotation

forte annotated

22 years ago[PATCH] sparse: msnd annotation
Alexander Viro [Thu, 3 Jun 2004 12:41:21 +0000 (05:41 -0700)]
[PATCH] sparse: msnd annotation

msnd annotated

22 years ago[PATCH] sparse: ymfpci annotation
Alexander Viro [Thu, 3 Jun 2004 12:41:10 +0000 (05:41 -0700)]
[PATCH] sparse: ymfpci annotation

ymfpci annotated

22 years ago[PATCH] sparse: maestro annotation
Alexander Viro [Thu, 3 Jun 2004 12:40:59 +0000 (05:40 -0700)]
[PATCH] sparse: maestro annotation

maestro annotated

22 years ago[PATCH] sparse: btaudio annotation
Alexander Viro [Thu, 3 Jun 2004 12:40:48 +0000 (05:40 -0700)]
[PATCH] sparse: btaudio annotation

btaudio annotated

22 years ago[PATCH] sparse: ad1889 annotation
Alexander Viro [Thu, 3 Jun 2004 12:40:37 +0000 (05:40 -0700)]
[PATCH] sparse: ad1889 annotation

ad1889 annotated

22 years ago[PATCH] sparse: the rest of sound/* annotation
Alexander Viro [Thu, 3 Jun 2004 12:40:27 +0000 (05:40 -0700)]
[PATCH] sparse: the rest of sound/* annotation

Remaining bits and pieces in sound/oss annotated.

At that point sound/* is done - we have no noise warnings left there.

22 years ago[PATCH] sparse: maestro3 annotation
Alexander Viro [Thu, 3 Jun 2004 12:40:16 +0000 (05:40 -0700)]
[PATCH] sparse: maestro3 annotation

maestro3 annotated

22 years ago[PATCH] sparse: sonicvibes annotation
Alexander Viro [Thu, 3 Jun 2004 12:40:06 +0000 (05:40 -0700)]
[PATCH] sparse: sonicvibes annotation

sonicvibes annotated

22 years ago[PATCH] sparse: ali5455 annotation
Alexander Viro [Thu, 3 Jun 2004 12:39:55 +0000 (05:39 -0700)]
[PATCH] sparse: ali5455 annotation

ali5455 annotated

22 years ago[PATCH] sparse: cs46xx annotation
Alexander Viro [Thu, 3 Jun 2004 12:39:44 +0000 (05:39 -0700)]
[PATCH] sparse: cs46xx annotation

cs46xx annotated

22 years ago[PATCH] sparse: i810 annotation
Alexander Viro [Thu, 3 Jun 2004 12:39:33 +0000 (05:39 -0700)]
[PATCH] sparse: i810 annotation

i810 annotated

22 years ago[PATCH] sparse: rme96xx annotation
Alexander Viro [Thu, 3 Jun 2004 12:39:22 +0000 (05:39 -0700)]
[PATCH] sparse: rme96xx annotation

rme96xx annotated

22 years ago[PATCH] sparse: esssolo annotation
Alexander Viro [Thu, 3 Jun 2004 12:39:11 +0000 (05:39 -0700)]
[PATCH] sparse: esssolo annotation

esssolo annotated

22 years ago[PATCH] sparse: es1370 annotation
Alexander Viro [Thu, 3 Jun 2004 12:39:00 +0000 (05:39 -0700)]
[PATCH] sparse: es1370 annotation

es1370 annotated

22 years ago[PATCH] sparse: es1371 annotation
Alexander Viro [Thu, 3 Jun 2004 12:38:50 +0000 (05:38 -0700)]
[PATCH] sparse: es1371 annotation

es1371 annotated

22 years ago[PATCH] sparse: cmpci annotation
Alexander Viro [Thu, 3 Jun 2004 12:38:39 +0000 (05:38 -0700)]
[PATCH] sparse: cmpci annotation

cmpci annotated

22 years ago[PATCH] sparse: cs4281 annotation
Alexander Viro [Thu, 3 Jun 2004 12:38:28 +0000 (05:38 -0700)]
[PATCH] sparse: cs4281 annotation

cs4281 annotated

22 years ago[PATCH] sparse: rest of ALSA annotation
Alexander Viro [Thu, 3 Jun 2004 12:38:17 +0000 (05:38 -0700)]
[PATCH] sparse: rest of ALSA annotation

The rest of annotation for ALSA drivers; only OSS stuff left to do in
sound/*

22 years ago[PATCH] sparse: emu10k1 annotation
Alexander Viro [Thu, 3 Jun 2004 12:38:05 +0000 (05:38 -0700)]
[PATCH] sparse: emu10k1 annotation

emu10k1 annotated.

22 years ago[PATCH] sparse: rme9652 annotation
Alexander Viro [Thu, 3 Jun 2004 12:37:54 +0000 (05:37 -0700)]
[PATCH] sparse: rme9652 annotation

Annotated driver and ioctl structure used by it.

22 years ago[PATCH] sparse: trident annotation
Alexander Viro [Thu, 3 Jun 2004 12:37:43 +0000 (05:37 -0700)]
[PATCH] sparse: trident annotation

Both ALSA and OSS variants

22 years ago[PATCH] sparse: vx annotation
Alexander Viro [Thu, 3 Jun 2004 12:37:32 +0000 (05:37 -0700)]
[PATCH] sparse: vx annotation

Both PCI and PCMCIA variants of driver annotated

22 years ago[PATCH] sparse: sb annotation
Alexander Viro [Thu, 3 Jun 2004 12:37:22 +0000 (05:37 -0700)]
[PATCH] sparse: sb annotation

sb annotation (both ALSA and OSS drivers)

22 years ago[PATCH] sparse: wavefront annotation
Alexander Viro [Thu, 3 Jun 2004 11:19:22 +0000 (04:19 -0700)]
[PATCH] sparse: wavefront annotation

Both ALSA and OSS drivers + wavefront ioctl structure annotated.  NB:
both should be switched to generic firmware loading - as it is, they are
using a homegrown and rather ugly variant

22 years ago[PATCH] sparse: sscape annotation
Alexander Viro [Thu, 3 Jun 2004 11:19:11 +0000 (04:19 -0700)]
[PATCH] sparse: sscape annotation

22 years ago[PATCH] sparse: sound/isa/gus annotation
Alexander Viro [Thu, 3 Jun 2004 11:19:00 +0000 (04:19 -0700)]
[PATCH] sparse: sound/isa/gus annotation

GUS annotated.

22 years ago[PATCH] sparse: rest of sound/core
Alexander Viro [Thu, 3 Jun 2004 11:18:49 +0000 (04:18 -0700)]
[PATCH] sparse: rest of sound/core

The rest of sond/core annotated; reverted bogus addition of __user in
snd_seq_kernel_client_ctl() - I should've guessed from the name alone
;-)

22 years ago[PATCH] sparse: opl annatation
Alexander Viro [Thu, 3 Jun 2004 11:18:40 +0000 (04:18 -0700)]
[PATCH] sparse: opl annatation

Annotated sound/drivers/opl*

22 years ago[PATCH] sparse: sound/core/pcm* annotation
Alexander Viro [Thu, 3 Jun 2004 11:18:29 +0000 (04:18 -0700)]
[PATCH] sparse: sound/core/pcm* annotation

The tricky part here was an iterator that used to take a callback and
argument for that callback as parameters.  Iterator itself didn't care
what type that argument had been; it's entirely up to callback.  The
thing is, two callbacks expect (and get) char __user * while other two
expect (and also get) char __user **.

Iterator used to use void * as "opaque data"; I've switched it to
unsigned long.  Note that there was nothing that said "it's a pointer" -
use of callback that would take e.g.  int is also perfectly legitimate.

The rest is triviali annotation.

22 years ago[PATCH] sparse: annotation of include/sound (generic parts)
Alexander Viro [Thu, 3 Jun 2004 09:48:01 +0000 (02:48 -0700)]
[PATCH] sparse: annotation of include/sound (generic parts)

Annotated method prototypes, commonly used ioctl-only structures and
generic helpers.

Instances of methods in drivers are _not_ annotated at that point - they
will go with driver patches

22 years ago[PATCH] sparse: sound/core annotation - trivial part
Alexander Viro [Thu, 3 Jun 2004 09:47:51 +0000 (02:47 -0700)]
[PATCH] sparse: sound/core annotation - trivial part

Trivial parts of sound/core annotation.

22 years ago[PATCH] sparse: sound/core/timer.c fix
Alexander Viro [Thu, 3 Jun 2004 09:47:39 +0000 (02:47 -0700)]
[PATCH] sparse: sound/core/timer.c fix

copy_from_user() where copy_to_user() should've been.  Unfortunately, on
x86 they are almost identical, so it went unnoticed for quite a while.

22 years ago[PATCH] sparse: msnd sound fix
Alexander Viro [Thu, 3 Jun 2004 09:47:28 +0000 (02:47 -0700)]
[PATCH] sparse: msnd sound fix

msnd_pinnacle/msnd_classic do copy_{to,from}_user under a spinlock.

Taken out of spinlock (into a temp. buffer). Calls of msnd_fifo_{read,write}
always go from kernel buffer now, so we can drop the 'int user' argument in
them _and_ simplify error handling - all errors were from copy_..._user() and
now these are called directly by dsp_read()/dsp_write().

22 years ago[PATCH] sparse: sound sb fix
Alexander Viro [Thu, 3 Jun 2004 09:47:17 +0000 (02:47 -0700)]
[PATCH] sparse: sound sb fix

In some cases snd_sb_csp_load() did kmalloc() and copy_from_user()
under a spinlock.  Split into snd_sb_csp_load() and snd_sb_csp_load_user() -
ther former always from kernel pointer, the latter - from userland.
snd_sb_csp_load_user() doesn't take any locks itself, it just
does kmalloc, copy_from_user and calls snd_sb_csp_load() to do the rest.

22 years ago[PATCH] Add the sixth arg to the sys_futex() prototype.
Andrew Morton [Thu, 3 Jun 2004 09:12:53 +0000 (02:12 -0700)]
[PATCH] Add the sixth arg to the sys_futex() prototype.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years agoMerge bk://bk.arm.linux.org.uk/linux-2.6-serial
Linus Torvalds [Thu, 3 Jun 2004 09:11:33 +0000 (02:11 -0700)]
Merge bk://bk.arm.linux.org.uk/linux-2.6-serial
into ppc970.osdl.org:/home/torvalds/v2.6/linux

22 years ago[ARM PATCH] 1906/1: S3C2410: serial minor number fix
Ben Dooks [Thu, 3 Jun 2004 16:34:49 +0000 (17:34 +0100)]
[ARM PATCH] 1906/1: S3C2410: serial minor number fix

Patch from Ben Dooks

Fix the minor number of the serial device to be the same
as the configuration in the 2.4.x series kernels.

22 years agoMerge bk://bk.arm.linux.org.uk/linux-2.6-pcmcia
Linus Torvalds [Thu, 3 Jun 2004 09:09:47 +0000 (02:09 -0700)]
Merge bk://bk.arm.linux.org.uk/linux-2.6-pcmcia
into ppc970.osdl.org:/home/torvalds/v2.6/linux

22 years ago[ARM PATCH] 1890/2: Consolidate CPUFREQ handling in SOC PCMCIA driver
Ian Campbell [Thu, 3 Jun 2004 22:13:07 +0000 (23:13 +0100)]
[ARM PATCH] 1890/2:  Consolidate CPUFREQ handling in SOC PCMCIA driver

Patch from Ian Campbell

Rediffed against latest BK. Compiles on Assabet and my PXA255

platform. Tested on my 255 platform.

22 years agoMerge bk://bk.arm.linux.org.uk/linux-2.6-rmk
Linus Torvalds [Thu, 3 Jun 2004 09:07:03 +0000 (02:07 -0700)]
Merge bk://bk.arm.linux.org.uk/linux-2.6-rmk
into ppc970.osdl.org:/home/torvalds/v2.6/linux

22 years ago[ARM PATCH] 1910/1: rework Mainstone and IDP include files
Nicolas Pitre [Thu, 3 Jun 2004 21:04:33 +0000 (22:04 +0100)]
[ARM PATCH] 1910/1: rework Mainstone and IDP include files

Patch from Nicolas Pitre

Don't include them from hardware.h

22 years ago[ARM] Don't include lubbock.h in asm/arch/hardware.h
Russell King [Thu, 3 Jun 2004 18:01:25 +0000 (19:01 +0100)]
[ARM] Don't include lubbock.h in asm/arch/hardware.h

Since asm/arch/hardware.h is included (indirectly) by most kernel
files, we don't want all these files depending on the individual
machine support files, especially as only five files really require
the header.

Instead, explicitly include lubbock.h into files as necessary.

22 years ago[ARM PATCH] 1902/1: OMAP update 1/3: entry-armv.S
Tony Lindgren [Thu, 3 Jun 2004 16:21:54 +0000 (17:21 +0100)]
[ARM PATCH] 1902/1: OMAP update 1/3: entry-armv.S

Patch from Tony Lindgren

Update the IO register names in entry-armv.S

22 years ago[PATCH] ppc32: Fix locks.c properly this time
Paul Mackerras [Thu, 3 Jun 2004 08:06:05 +0000 (01:06 -0700)]
[PATCH] ppc32: Fix locks.c properly this time

When I moved the exports into arch/ppc/lib/locks.c, I forgot to
include module.h, so it doesn't compile (with CONFIG_SMP +
CONFIG_SPINLOCK_DEBUG).  This patch fixes it.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] DMA mode setup fixes for piix.c/ata_piix.c
Bartlomiej Zolnierkiewicz [Thu, 3 Jun 2004 06:16:15 +0000 (23:16 -0700)]
[PATCH] DMA mode setup fixes for piix.c/ata_piix.c

From: Go Taniguchi <go@turbolinux.co.jp> (Turbolinx Inc.)

Change reg48 and reg54 in 8bit access and remove unused reg44.

Jeff acked ata_piix.c part.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@elka.pw.edu.pl>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] Export swapper_space
Russell King [Thu, 3 Jun 2004 01:26:30 +0000 (18:26 -0700)]
[PATCH] Export swapper_space

swapper_space is needed by at least loop/st/sg these days.

22 years ago[PATCH] ppc32: Reduce WARN_ON(0) to nothing
Paul Mackerras [Thu, 3 Jun 2004 01:22:18 +0000 (18:22 -0700)]
[PATCH] ppc32: Reduce WARN_ON(0) to nothing

The last patch I sent means that we have WARN_ON(0) in a couple of
places when CONFIG_PREEMPT=n.  This patch makes that reduce to
nothing (rather than a conditional trap on a 0 value), and also makes
BUG_ON(0) reduce to nothing for completeness.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc32: Fix preemptible check
Paul Mackerras [Thu, 3 Jun 2004 01:22:08 +0000 (18:22 -0700)]
[PATCH] ppc32: Fix preemptible check

Ben H added a check in a couple of places to make sure that we had
preemption disabled when we call enable_kernel_{fp,altivec}.
Unfortunately the check he used trips in the case when
CONFIG_PREEMPT=n.  This patch fixes it by defining a preemptible()
macro (which reduces to 0 when CONFIG_PREEMPT=n) and doing
WARN_ON(preemptible()).

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc32: Make ppc32 PCI code more robust
Paul Mackerras [Thu, 3 Jun 2004 01:21:57 +0000 (18:21 -0700)]
[PATCH] ppc32: Make ppc32 PCI code more robust

The main thrust of this patch is to make the ppc32 PCI code more
robust by checking for bus->resource[] being NULL before using it.  We
can legitimately get elements of bus->resource being NULL and I have
actually hit that on some machines.  This patch also allows resources
starting at 0 to be accepted as assigned (we can and do get PCI
resources starting at 0 in I/O space on PPC machines) and provides a
sensible default for the case where Open Firmware doesn't give us a
bus-range property for a PCI bridge.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc32: Use -fPIC instead of -mrelocatable-lib
Paul Mackerras [Thu, 3 Jun 2004 01:21:47 +0000 (18:21 -0700)]
[PATCH] ppc32: Use -fPIC instead of -mrelocatable-lib

The ppc32 boot code has a couple of files that are executed very early
on before the kernel is mapped at the address it is linked at.  We
have been using -mrelocatable-lib to compile these files, but
apparently -mrelocatable-lib is deprecated and the gcc developers are
threatening to remove it.  In fact the -fPIC flag does what we need.
This patch changes -mrelocatable-lib to -fPIC.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc32: Suppress bogus info in /proc/ppc_htab on 64-bit cpus
Paul Mackerras [Thu, 3 Jun 2004 01:21:36 +0000 (18:21 -0700)]
[PATCH] ppc32: Suppress bogus info in /proc/ppc_htab on 64-bit cpus

In the ppc32 kernel, we have a /proc/ppc_htab file that trawls through
the MMU hash table and prints various statistics on it such as percent
occupancy.  However, the hash table entry format is different on
64-bit cpus (POWER3, G5) which the ppc32 kernel does support (in
32-bit mode).

This patch disables the scanning of the MMU hash table and printing of
the statistics that we get from it on 64-bit cpus.  Since the
statistics are only for interest, and the ppc32 kernel is being used
less and less on 64-bit cpus now that the ppc64 kernel is in
reasonable shape, I didn't think it worth while to add code to deal
with the 64-bit HPTE format.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc32: Don't synchronize in disable_irq() if no handler
Paul Mackerras [Thu, 3 Jun 2004 01:21:26 +0000 (18:21 -0700)]
[PATCH] ppc32: Don't synchronize in disable_irq() if no handler

This patch is the ppc32 counterpart to a fix that went into
arch/i386/kernel/irq.c last October.  The bug was noted by Al Viro: if
no handler exists, and we have IRQ_INPROGRESS set because of an
earlier irq that got through, synchronize_irq() will end up waiting
forever.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc32: Add _raw_write_trylock
Paul Mackerras [Thu, 3 Jun 2004 01:21:15 +0000 (18:21 -0700)]
[PATCH] ppc32: Add _raw_write_trylock

I tried compiling a PPC32 kernel with PREEMPT + SMP and it failed
because we didn't have a _raw_write_trylock.  This patch adds
_raw_write_trylock, moves the exports of _raw_*lock from
arch/ppc/kernel/ppc_ksyms.c to arch/ppc/lib/locks.c, and makes
__spin_trylock static since it is only used in locks.c.

Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years agosparse: fix up futex address space warning
Linus Torvalds [Thu, 3 Jun 2004 01:20:09 +0000 (18:20 -0700)]
sparse: fix up futex address space warning

22 years ago[PATCH] direct-io invalidation fix
Andrew Morton [Thu, 3 Jun 2004 01:05:20 +0000 (18:05 -0700)]
[PATCH] direct-io invalidation fix

clean_blockdev_aliases() is using the wrong thing to work out how many
filesystem blocks should be invalidated.  It invalidates too many, which can
cause live fs metadata buffers to be invalidated when they are pending
writeout.  It's a filesystem-wrecker, although seems very hard to hit.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] bug in sys_io_setup
Andrew Morton [Thu, 3 Jun 2004 01:05:09 +0000 (18:05 -0700)]
[PATCH] bug in sys_io_setup

From: Jerzy Szczepkowski <js189202@zodiac.mimuw.edu.pl>

There is a bug in sys_io_setup().

If ioctx_alloc() succeeds and put_user() fails io_destroy() is called.
io_destroy() assumes that ioctx->users >= 2 (if context is alive) and calls
put_ioctx twice, while in this sequence ioctx->users == 1.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] Use decimal instead of hex for EDD values
Andrew Morton [Thu, 3 Jun 2004 01:04:58 +0000 (18:04 -0700)]
[PATCH] Use decimal instead of hex for EDD values

From: "Patrick J. LoPresti" <patl@users.sourceforge.net>

This patch changes default_cylinders, default_heads,
default_sectors_per_track, legacy_max_cylinder, legacy_max_head,
legacy_sectors_per_track, and sectors to decimal.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] Better names for EDD legacy_* fields
Andrew Morton [Thu, 3 Jun 2004 01:04:48 +0000 (18:04 -0700)]
[PATCH] Better names for EDD legacy_* fields

From: "Patrick J. LoPresti" <patl@users.sourceforge.net>

Renames legacy_heads to legacy_max_head legacy_sectors to
legacy_sectors_per_track.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] use c99 struct initializer in hotcpu_notifier
Andrew Morton [Thu, 3 Jun 2004 01:04:36 +0000 (18:04 -0700)]
[PATCH] use c99 struct initializer in hotcpu_notifier

From: Nathan Lynch <nathanl@austin.ibm.com>

The hotcpu_notifier macro does not properly record the given priority in
the notifier block.  This causes trouble only for callers which specify a
non-zero priority, of which there are none (yet).

Signed-off-by: Nathan Lynch <nathanl@austin.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ext3_orphan_del may double-decrement bh->b_count
Andrew Morton [Thu, 3 Jun 2004 01:04:25 +0000 (18:04 -0700)]
[PATCH] ext3_orphan_del may double-decrement bh->b_count

From: Jeff Mahoney <jeffm@suse.com>

Chris Mason and I ran across this one while hunting down another bug.

If ext3_mark_iloc_dirty() fails in ext3_orphan_del() on the outer buffer,
bh->b_count will be decremented twice.  ext3_mark_iloc_dirty() will brelse
the buffer, even on error.  ext3_orphan_del() is explicity brelse'ing the
buffer on error.  Prior to calling ext3_mark_iloc_dirty(), this is the
correct behavior.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] update elilo loader location in Kconfig
Andrew Morton [Thu, 3 Jun 2004 01:04:14 +0000 (18:04 -0700)]
[PATCH] update elilo loader location in Kconfig

From: Matt Tolentino <metolent@snoqualmie.dp.intel.com>

The elilo EFI boot loader has been moved to sourceforge.  So, update the
location of where one might look for it.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] kill off efi_dir in efi.h
Andrew Morton [Thu, 3 Jun 2004 01:04:03 +0000 (18:04 -0700)]
[PATCH] kill off efi_dir in efi.h

From: Matt Tolentino <metolent@snoqualmie.dp.intel.com>

The /proc support for efi 'stuff' isn't used/needed anymore, yet the
efi_dir declaration remains.  This removes it.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] quota: fix for old quota format
Andrew Morton [Thu, 3 Jun 2004 01:03:52 +0000 (18:03 -0700)]
[PATCH] quota: fix for old quota format

From: Jan Kara <jack@ucw.cz>

Fix a problem in the old quota format when we tried to read quota
information after the end of quota file (that is correct as it might a user
with sufficiently large UID which has no limits or usage).

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] quota: fix writing of quota info
Andrew Morton [Thu, 3 Jun 2004 01:03:41 +0000 (18:03 -0700)]
[PATCH] quota: fix writing of quota info

From: Jan Kara <jack@ucw.cz>

Fixes a problem with some quota operations not writing the quota info they
changed which could later cause that some transaction to use more buffers
than it had reserved or it could cause corrupted quota files when the
system was rebooted at the right time.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] s390: network device driver
Andrew Morton [Thu, 3 Jun 2004 01:03:27 +0000 (18:03 -0700)]
[PATCH] s390: network device driver

From: Martin Schwidefsky <schwidefsky@de.ibm.com>

Network driver changes:
 - iucv: Fix special case of a "Connection Pending" interrupt within
   iucv_do_int.
 - netiucv: Revoke broken iucvMagic change for more than one connection.
 - qeth: Fix string parsing in notifier_register attribute function.
 - qeth: Add code for socket ioctl SIOC_QETH_GET_CARD_TYPE.
 - qeth: Fix debug log entry and buffer copy in qeth_snmp_command_cb.
 - qeth: Fix race on qeth_dbf_txt_buf.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] s390: block device driver
Andrew Morton [Thu, 3 Jun 2004 01:03:07 +0000 (18:03 -0700)]
[PATCH] s390: block device driver

From: Martin Schwidefsky <schwidefsky@de.ibm.com>

block device driver changes:
 - dasd: Fix diag discipline if it is loaded as a module.
 - dcssblk: Replace r/w lock with r/w semaphore to be able to call
   device_register inside a critical section.
 - dcssblk: Fix error handling in write function for dcss "add" attribute.
 - xpram & dcssblk: Fix sanity check for sector number.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] s390: common i/o layer
Andrew Morton [Thu, 3 Jun 2004 01:02:51 +0000 (18:02 -0700)]
[PATCH] s390: common i/o layer

From: Martin Schwidefsky <schwidefsky@de.ibm.com>

Common i/o layer changes:
 - qdio: Lose the adapter lock for thin interrupts to improve performance
   and do unregister of the adapter interrupt handler with rcu.
 - ccwgroup: Fix error handling when creating a ccwgroup device.
 - Convert the slow crw kernel thread to a single threaded workqueue.
 - Use the slow crw workqueue to unregister a subchannel after it was
   found not operational to serialize it with other possible unregister/
   register events coming in via machine checks.
 - Trigger a rescan of the css via the slow path if a missing channel path
   is found in __recover_lost_chpids.
 - Use saner default levels for the debug feature, add some debugging code.
 - Remove request_irq and free_irq stubs.
 - Remove bogus inlines.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] s390: core
Andrew Morton [Thu, 3 Jun 2004 01:02:35 +0000 (18:02 -0700)]
[PATCH] s390: core

From: Martin Schwidefsky <schwidefsky@de.ibm.com>

s390 core changes:
 - Make use of the ipte instruction for ptep_set_access_flags
 - Fix atomic64_inc_and_test primitive as well.
 - Fix return type handler for __copy_in_user.
 - New default configuration.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] hugetlbpage: reinitialise compound page destructor
Andrew Morton [Thu, 3 Jun 2004 01:02:18 +0000 (18:02 -0700)]
[PATCH] hugetlbpage: reinitialise compound page destructor

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

Currently the hugepage code stores the hugepage destructor in the mapping
field of the second of the compound pages.  However, this field is never
cleared again, which causes tracebacks from free_pages_check() if the
hugepage is later destroyed by reducing the number in
/proc/sys/vm/nr_hugepages.  This patch fixes the bug by clearing the
mapping field when the hugepage is freed.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] direct-io hole fix
Andrew Morton [Thu, 3 Jun 2004 01:02:02 +0000 (18:02 -0700)]
[PATCH] direct-io hole fix

From: Chris Mason <mason@suse.com>

When filling holes via DIRECT_IO, we fall back to normal buffered io.  For
this to work properly, the direct io funcs have to return a value of zero to
the file write functions, so the file write functions know where to start
writing.

In some cases, dio->result was getting returned by direct_io_worker, and that
wasn't always zero, causing some data not to be written.

From: <akpm@osdl.org>:

- Simplify things by setting `ret' later on, fix up subsequent damage to the
  dio_complete() args.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] hugetlbpage msync() fix
Andrew Morton [Thu, 3 Jun 2004 01:01:46 +0000 (18:01 -0700)]
[PATCH] hugetlbpage msync() fix

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

Currently, calling msync() on a hugepage area will cause the kernel to blow
up with a bad_page() (at least on ppc64, but I think the problem will exist
on other archs too).  The msync path attempts to walk pagetables which may
not be there, or may have an unusual layout for hugepages.

Lucikly we shouldn't need to do anything for an msync on hugetlbfs beyond
flushing the cache, so this patch should be sufficient to fix the problem.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] move #endif to correct place
Andrew Morton [Thu, 3 Jun 2004 01:01:30 +0000 (18:01 -0700)]
[PATCH] move #endif to correct place

From: David Mosberger <davidm@napali.hpl.hp.com>

Darrene Williams <dsw@gelato.unsw.edu.au> noticed that the #endif for
__ARCH_WANT_SYS_SIGPROCMASK was off by one routine.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] aio: fix io_getevents() timer expiry setting
Andrew Morton [Thu, 3 Jun 2004 01:01:14 +0000 (18:01 -0700)]
[PATCH] aio: fix io_getevents() timer expiry setting

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

start_jiffies was not respected by set_timeout(), which reread jiffies
instead of respecting what read_events() passed it.  This difference can be
significant, particularly if the calling process slept during the
copy_to_user() operation in read_events().  To correct this, this patch
teaches it to respect its argument, with the additional bonus of converting
it to use timespec_to_jiffies() instead of open-coding it.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] use const in time.h unit conversion functions
Andrew Morton [Thu, 3 Jun 2004 01:01:02 +0000 (18:01 -0700)]
[PATCH] use const in time.h unit conversion functions

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

The time conversion functions may have const args, which is in fact useful
for when they are passed const variables as arguments so as to avoid
discarding qualifiers from pointer types warnings.  This is a preparatory
cleanup for a minor aio bugfix.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] mm/oom_kill.c trivial cleanup
Andrew Morton [Thu, 3 Jun 2004 01:00:46 +0000 (18:00 -0700)]
[PATCH] mm/oom_kill.c trivial cleanup

From: "Luiz Fernando N. Capitulino" <lcapitulino@prefeitura.sp.gov.br>

Remove duplicated assignment.

Signed-off by: Luiz Capitulino <lcapitulino@prefeitura.sp.gov.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] dm: add DM_REMOVE_ALL_32 compat ioctl
Andrew Morton [Thu, 3 Jun 2004 01:00:30 +0000 (18:00 -0700)]
[PATCH] dm: add DM_REMOVE_ALL_32 compat ioctl

From: Marcus Meissner <meissner@suse.de>

Add DM_REMOVE_ALL_32 to the list of compatible ioctls.

Signed-off-By: Marcus Meissner <meissner@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] fix possible NULL pointer in fs/ext3/super.c.
Andrew Morton [Thu, 3 Jun 2004 01:00:14 +0000 (18:00 -0700)]
[PATCH] fix possible NULL pointer in fs/ext3/super.c.

From: "Luiz Fernando N. Capitulino" <lcapitulino@prefeitura.sp.gov.br>

In fs/ext3/super.c::ext3_get_journal() at line 1675 `journal' can be NULL,
but it is not handled right (detect by Coverity's checker).

Signed-off by: Luiz Capitulino <lcapitulino@prefeitura.sp.gov.br>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] Fix loop device cache handling
Andrew Morton [Thu, 3 Jun 2004 00:59:59 +0000 (17:59 -0700)]
[PATCH] Fix loop device cache handling

From: Russell King <rmk+lkml@arm.linux.org.uk>

It appears the loop driver has had one flush_dcache_page() call added for
the case where it writes to the backing device page cache pages.

However, it seems to be missing the call where it writes to its own page
cache pages.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] sched: balance-on-exec fix
Andrew Morton [Thu, 3 Jun 2004 00:59:43 +0000 (17:59 -0700)]
[PATCH] sched: balance-on-exec fix

From: Jack Steiner <steiner@sgi.com>

It looks like the call to sched_balance_exec() from do_execve() is in the
wrong spot.  The code calls sched_balance_exec() before determining whether
"filename" actually exists.

In many cases, users have several entries in $PATH.  If a full path name is
not specified on the 'exec" call, the library code iterates thru the files
in the PATH list until it finds the program.  This can result is numerous
migrations of the parent process before the program is actually found.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] floppy minor number fix
Andrew Morton [Thu, 3 Jun 2004 00:59:32 +0000 (17:59 -0700)]
[PATCH] floppy minor number fix

From: Paul Slootman <paul+nospam@wurtel.net>

Correctly set the floppy disk's first minor.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] selinux: check processed security context length
Andrew Morton [Thu, 3 Jun 2004 00:59:21 +0000 (17:59 -0700)]
[PATCH] selinux: check processed security context length

From: Stephen Smalley <sds@epoch.ncsc.mil>

This patch changes security_context_to_sid to check the length of the
processed security context against the full length of the provided context,
rejecting any further data.

Signed-off-by: Stephen Smalley <sds@epoch.ncsc.mil>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] Add reference_init.pl to `make buildcheck' target
Andrew Morton [Thu, 3 Jun 2004 00:59:10 +0000 (17:59 -0700)]
[PATCH] Add reference_init.pl to `make buildcheck' target

`make buildcheck' only checks for calls to linker discarded sections,

reference_init checks for calls to sections discarded at run time, init was
cloned from discarded.  They are separate because the linker detects the
discarded case and I did not want to confuse users with messages about init
text/data while they were fixing the linker errors.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] partition table validity checking
Andrew Morton [Thu, 3 Jun 2004 00:58:58 +0000 (17:58 -0700)]
[PATCH] partition table validity checking

From: Andries Brouwer <Andries.Brouwer@cwi.nl>

The patch examines a putative partition table, and if that doesnt look like a
valid partition table it goes away again.

Some devices have partition tables (and there are many styles of such), some
don't.  Traditionally fixed disks have one, floppies don't.  Nobody knows what
happens with ZIP disks, USB sticks and other such things.  Both the DOS-type
partition table, and the "big floppy" whole disk FAT filesystem are common.
It is undesirable for the kernel to detect partitions where there are none.
This leads to great confusion, sometimes to kernel crashes.

In the particular case of DOS-type partition tables a partition entry has a
1-byte field boot_ind that traditionally is 0x80 for the boot partition and 0
for the other three primary partitions.  Linux does not use this field, and
one sometimes sees tables with all four entries zero.

The patch tells the kernel not to think that something is a valid DOS-type
partition table when a value other than 0 or 0x80 is encountered.  I think it
is a fairly safe change: I do not know of any fdisk-type program that will
write other values there.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc64 gives up too quickly on hotplugged cpu
Andrew Morton [Thu, 3 Jun 2004 00:58:47 +0000 (17:58 -0700)]
[PATCH] ppc64 gives up too quickly on hotplugged cpu

From: Nathan Lynch <nathanl@austin.ibm.com>

On some systems it can take a hotplugged cpu much longer to come up than it
would at boot.  If the cpu comes up after we've given up on it, it tends to
die in its first attempt to kmem_cache_alloc (uninitialized percpu data, I
imagine).

In my experimentation I haven't seen a processor take more than one second
to become available; the patch waits five seconds just to be safe.

Signed-off-by: Nathan Lynch <nathanl@austin.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc64: update info about available iseries_veth interfaces
Andrew Morton [Thu, 3 Jun 2004 00:58:36 +0000 (17:58 -0700)]
[PATCH] ppc64: update info about available iseries_veth interfaces

From: Olaf Hering <olh@suse.de>

/proc/iSeries/config contains now the number of configured virtual ethernet
adapters.  AVAILABLE_VETH should only indicate if there is at least one
interface available, iseries_veth must be loaded in this case.

Printing the entire map will give installers some hints about what
interface numbers will appear and how the MAC address may look like.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc64: kernel Makefile options for $(AS)
Andrew Morton [Thu, 3 Jun 2004 00:58:25 +0000 (17:58 -0700)]
[PATCH] ppc64: kernel Makefile options for $(AS)

From: Olaf Hering <olh@suse.de>

`-64' is no longer a valid option for as.  Pass it `-a64' instead.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] ppc32: add "indirect" DCR access, pass 2
Andrew Morton [Thu, 3 Jun 2004 00:58:14 +0000 (17:58 -0700)]
[PATCH] ppc32: add "indirect" DCR access, pass 2

From: Matt Porter <mporter@kernel.crashing.org>

DCR number is encoded in mfdcr/mtdcr command itself and this prevents easy
DCR access when register number is not known on compile time.  This patch
adds __mfdcr & __mtdcr helpers which use pre-generated mfdcr/mtdcr
sequences for all possible DCR numbers.  We also use GCC extension
__builtin_constant_p to optimize cases when DCR number is in fact known
during compilation.

Signed-off-by: Eugene Surovegin <ebs@ebshome.net>
Signed-off-by: Matt Porter <mporter@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
22 years ago[PATCH] shrink_all_memory() fixes
Andrew Morton [Thu, 3 Jun 2004 00:58:03 +0000 (17:58 -0700)]
[PATCH] shrink_all_memory() fixes

- Off-by-one in balance_pgdat means that we're not scanning the zones all
  the way down to priority=0.

- Always set zone->temp_priority in shrink_caches().  I'm not sure why I had
  the `if (zone->free_pages < zone->pages_high)' test in there, but it's
  preventing us from setting ->prev_priority correctly on the
  try_to_free_pages() path.

- Set zone->prev_priority to the current priority if it's currently a
  "lower" priority.  This allows us to build up the pressure on mapped pages
  on the first scanning pass rather than only on successive passes.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>