12 years ago[PATCH] ufs: truncate negative to unsigned fix
Evgeniy Dushistov [Mon, 29 Jan 2007 21:19:55 +0000]
[PATCH] ufs: truncate negative to unsigned fix

During ufs_trunc_direct which is subroutine of ufs::truncate, we try the first
of all free parts of block and then whole blocks.  But we calculate size of
block's part to free in the wrong way.

This may cause bad update of used blocks and fragments statistic, and you can
got report that you have free 32T on 1Gb partition.

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] ufs: alloc metadata null page fix
Evgeniy Dushistov [Mon, 29 Jan 2007 21:19:54 +0000]
[PATCH] ufs: alloc metadata null page fix

These series of patches result of UFS1 write support stress testing, like
running fsx-linux, untar and build linux kernel etc

We pass from ufs::get_block_t to levels below: pointer to the current page, to
make possible things like reallocation of blocks on the fly, and we also uses
this pointer for indication, what actually we allocate data block or meta data
block, but currently we make decision about what we allocate on the wrong
level, this may and cause oops if we allocate blocks in some special order.

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] fuse: fix bug in control filesystem mount
Miklos Szeredi [Mon, 29 Jan 2007 21:19:54 +0000]
[PATCH] fuse: fix bug in control filesystem mount

The BUG in fuse_ctl_add_dentry() could be triggered if the control
filesystem was unmounted and mounted again while one or more fuse
filesystems were present.

The fix is to reset the dentry counter in fuse_ctl_kill_sb().

Bug reported by Florent Mertens.

Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] use __u8 rather than u8 in userspace SIZE defines in hdreg.h
Mike Frysinger [Mon, 29 Jan 2007 21:19:53 +0000]
[PATCH] use __u8 rather than u8 in userspace SIZE defines in hdreg.h

Use __u8 rather than u8 in SIZE defines exported to userspace.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Cc: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] knfsd: ratelimit some nfsd messages that are triggered by external events
NeilBrown [Mon, 29 Jan 2007 21:19:52 +0000]
[PATCH] knfsd: ratelimit some nfsd messages that are triggered by external events

Also remove {NFSD,RPC}_PARANOIA as having the defines doesn't really add
anything.

The printks covered by RPC_PARANOIA were triggered by badly formatted
packets and so should be ratelimited.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] fs/lockd/clntlock.c: add missing newlines to dprintk's
Adrian Bunk [Mon, 29 Jan 2007 21:19:51 +0000]
[PATCH] fs/lockd/clntlock.c: add missing newlines to dprintk's

This patch adds missing newlines to dprintk's.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] m68k: uaccess.h needs sched.h
Andrew Morton [Mon, 29 Jan 2007 21:19:50 +0000]
[PATCH] m68k: uaccess.h needs sched.h

In file included from include/linux/crypto.h:26,
                 from crypto/cipher.c:17:
include/linux/uaccess.h: In function 'pagefault_disable':
include/linux/uaccess.h:18: error: dereferencing pointer to incomplete type
include/linux/uaccess.h: In function 'pagefault_enable':
include/linux/uaccess.h:33: error: dereferencing pointer to incomplete type

video_buf need PCI.

Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] Fix "CONFIG_X86_64_" typo in drivers/kvm/svm.c
Robert P. J. Day [Mon, 29 Jan 2007 21:19:50 +0000]
[PATCH] Fix "CONFIG_X86_64_" typo in drivers/kvm/svm.c

Fix what looks like an obvious typo in the file drivers/kvm/svm.c.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Acked-by: Avi Kivity <avi@qumranet.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] use __u8/__u32 in userspace ioctl defines for I2O
Mike Frysinger [Mon, 29 Jan 2007 21:19:49 +0000]
[PATCH] use __u8/__u32 in userspace ioctl defines for I2O

Make sure exported I2O ioctls utilize userspace safe types.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Cc: Markus Lidel <Markus.Lidel@shadowconnect.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] uml: fix mknod
Johannes Stezenbach [Mon, 29 Jan 2007 21:19:44 +0000]
[PATCH] uml: fix mknod

Fix UML hostfs mknod(): userspace has differernt dev_t size and encoding
than kernel, so extract major/minor and reencode using glibc makedev()
macro.

Signed-off-by: Johannes Stezenbach <js@linuxtv.org>
Acked-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] namespaces: fix exit race by splitting exit
Serge E. Hallyn [Mon, 29 Jan 2007 21:19:40 +0000]
[PATCH] namespaces: fix exit race by splitting exit

Fix exit race by splitting the nsproxy putting into two pieces.  First
piece reduces the nsproxy refcount.  If we dropped the last reference, then
it puts the mnt_ns, and returns the nsproxy as a hint to the caller.  Else
it returns NULL.  The second piece of exiting task namespaces sets
tsk->nsproxy to NULL, and drops the references to other namespaces and
frees the nsproxy only if an nsproxy was passed in.

A little awkward and should probably be reworked, but hopefully it fixes
the NFS oops.

Signed-off-by: Serge E. Hallyn <serue@us.ibm.com>
Cc: Herbert Poetzl <herbert@13thfloor.at>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Cedric Le Goater <clg@fr.ibm.com>
Cc: Daniel Hokka Zakrisson <daniel@hozac.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agoBroadcom 4400 resume small fix
Dmitriy Monakhov [Sat, 27 Jan 2007 08:00:03 +0000]
Broadcom 4400 resume small fix

Some issues in b44_resume().

- Return value of pci_enable_device() was ignored.

- If request_irq() has failed we have to just disable device and exit.

Signed-off-by: Dmitriy Monakhov <dmonakhov@openvz.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agob44: src_desc->addr is little-endian
Al Viro [Tue, 30 Jan 2007 13:23:50 +0000]
b44: src_desc->addr is little-endian

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoe100: fix irq leak on suspend/resume
Auke Kok [Mon, 29 Jan 2007 22:31:16 +0000]
e100: fix irq leak on suspend/resume

e100: fix irq leak on suspend/resume

From: Frederik Deweerdt <frederik.deweerdt@gmail.com>

The e100_resume() function should be calling netif_device_detach and
free_irq. This fixes multiple irq's being allocated after resume.

Signed-off-by: Frederik Deweerdt <frederik.deweerdt@gmail.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agobonding: ARP monitoring broken on x86_64
Andy Gospodarek [Mon, 29 Jan 2007 20:08:38 +0000]
bonding: ARP monitoring broken on x86_64

While working with the latest bonding code I noticed a nasty problem that
will prevent arp monitoring from always functioning correctly on x86_64
systems.  Comparing ints to longs and expecting reliable results on x86_64
is a bad idea.  With this patch, arp monitoring works correctly again.

Signed-off-by: Andy Gospodarek <andy@greyhouse.net>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Stephen Hemminger <shemminger@osdl.org>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoehea: Fixed missing tasklet_kill() call
Thomas Klein [Mon, 29 Jan 2007 17:44:41 +0000]
ehea: Fixed missing tasklet_kill() call

NEQ-Tasklet wasn't killed when module is removed.

Signed-off-by: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoehea: Fixed wrong jumbo frames status query
Thomas Klein [Mon, 29 Jan 2007 17:44:01 +0000]
ehea: Fixed wrong jumbo frames status query

This patch fixes the wrong query and logging of the per interface jumbo frames
enabled/disabled status.

Signed-off-by: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years ago82596 warning fixes
Andrew Morton [Sat, 27 Jan 2007 08:00:04 +0000]
82596 warning fixes

drivers/net/82596.c: In function 'i596_start_xmit':
drivers/net/82596.c:1069: warning: cast from pointer to integer of different size
drivers/net/82596.c: In function 'i82596_probe':
drivers/net/82596.c:1249: warning: format '%d' expects type 'int', but argument 4 has type 'long unsigned int'

Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoFS_ENET: OF-related fixup for FEC and SCC MAC's
Vitaly Bordug [Sat, 27 Jan 2007 08:00:04 +0000]
FS_ENET: OF-related fixup for FEC and SCC MAC's

Updated direct resource pass with ioremap call, make it grant proper IRQ
mapping, stuff incompatible with the new approach were respectively put
under #ifndef CONFIG_PPC_MERGE.  It is required so that both ppc and
powerpc could utilize fs_enet effectively.

Signed-off-by: Vitaly Bordug <vbordug@ru.mvista.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agonet: ifb error path loop fix
Mariusz Kozlowski [Sat, 27 Jan 2007 08:00:01 +0000]
net: ifb error path loop fix

On error we should start freeing resources at [i-1] not [i-2].

Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agob44: Fix frequent link changes
Michael Chan [Sat, 27 Jan 2007 07:59:57 +0000]
b44: Fix frequent link changes

This fixes the issue of frequent link changes under heavy traffic reported
below:

http://bugzilla.kernel.org/show_bug.cgi?id=7696
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=216338

The b44 chip occasionally needs to be reset when ISTAT_ERRORS are
encountered.  The reset sequence includes a PHY reset that will take many
seconds to complete and cause the link to go down and up.  By skipping the
PHY reset, it will greatly reduce the interruption when ISTAT_ERRORS are
encountered.

Change the full_reset parameter to reset_kind parameter in b44_init_hw().
This will allow PHY reset to be skipped when ISTAT_ERRORS are encountered.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agolibata: fix translation for START STOP UNIT
Robert Hancock [Tue, 30 Jan 2007 08:59:18 +0000]
libata: fix translation for START STOP UNIT

libata's SCSI translation for the SCSI START STOP UNIT command with the
START bit clear (i.e.  stopping the drive) appears to be incorrect.  It
sends an ATA STANDBY command with the time period set to 0, which the code
comment says means "now", but the ATA standard says this means disable the
standby timer, which effectively does nothing.  Change this to issue a
STANDBY IMMEDIATE command which will actually spin the drive down.  The SAT
(SCSI/ATA Translation) standard revision 9 concurs with this choice.

Signed-off-by: Robert Hancock <hancockr@shaw.ca>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agolibata-scsi: ata_task_ioctl should return ATA registers from sense data
David Milburn [Tue, 30 Jan 2007 08:59:15 +0000]
libata-scsi: ata_task_ioctl should return ATA registers from sense data

User applications using the HDIO_DRIVE_TASK ioctl through libata expect
specific ATA registers to be returned to userspace.  Verified that
ata_task_ioctl correctly returns register values to the smartctl
application.

Signed-off-by: David Milburn <dmilburn@redhat.com>
Acked-by: Tejun Heo <htejun@gmail.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agopata_platform: set_mode fix
Andrew Morton [Tue, 30 Jan 2007 08:59:14 +0000]
pata_platform: set_mode fix

drivers/ata/pata_platform.c:85: warning: initialization from incompatible pointer type

Cc: Jeff Garzik <jeff@garzik.org>
Cc: Tejun Heo <htejun@gmail.com>
Acked-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoata_if_xfermask() word 51 fix
Lennert Buytenhek [Mon, 29 Jan 2007 12:28:47 +0000]
ata_if_xfermask() word 51 fix

If word 53 bit 1 isn't set, the maximum PIO mode is indicated by
the upper 8 bits of word 51, not the lower 8 bits.  Fixes PIO mode
detection on old Compact Flash cards.

Signed-off-by: Lennert Buytenhek <buytenh@wantstofly.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agopata_sil680: PIO1 taskfile transfers overclocking fix (repost)
Sergei Shtylyov [Sun, 28 Jan 2007 18:33:44 +0000]
pata_sil680: PIO1 taskfile transfers overclocking fix (repost)

Fix PIO mode 1 overclocked taskfile transfers -- probably a typo carried over
from drivers/ide/pci/siimage.c where I've found it by documentation check...

Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoMerge branch 'master' into upstream-fixes
Jeff Garzik [Tue, 30 Jan 2007 14:14:40 +0000]
Merge branch 'master' into upstream-fixes

12 years agoHID: fix pb_fnmode and move it to generic HID
Jiri Kosina [Mon, 29 Jan 2007 11:44:41 +0000]
HID: fix pb_fnmode and move it to generic HID

The apple powerbook people are used to switch the pb_fnmode
setting at runtime through writing to sysfs, altering the
module parameter value. This was broken for them in 2.6.20-rc1
when generic HID layer was introduced, as the pb_fnmode flag
was made per-hiddevice, instead of global variable.

This patch moves the pb_fnmode module parameter from usbhid module
to hid module, but apart from that retains backward compatibility
with respect to changing the mode through sysfs.

Signed-off-by: Jiri Kosina <jkosina@suse.cz>

12 years ago[MIPS] Fix typo of "CONFIG_MT_SMP".
Robert P. J. Day [Thu, 25 Jan 2007 23:41:35 +0000]
[MIPS] Fix typo of "CONFIG_MT_SMP".

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

12 years ago[MIPS] Ocelot G: Fix a few misspellings of CONFIG_GALILEO_GT64240_ETH
Ralf Baechle [Thu, 25 Jan 2007 23:55:17 +0000]
[MIPS] Ocelot G: Fix a few misspellings of CONFIG_GALILEO_GT64240_ETH

Reported by Robert P. J. Day <rpjday@mindspring.com>.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

12 years ago[PATCH] Malta: Fix build if CONFIG_MTD is diabled.
Jan Altenberg [Thu, 25 Jan 2007 15:25:56 +0000]
[PATCH] Malta: Fix build if CONFIG_MTD is diabled.

Signed-off-by: Jan Altenberg <jan@linutronix.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

12 years ago[PATCH] Fix SG_IO timeout jiffy conversion
Mike Christie [Tue, 30 Jan 2007 02:18:38 +0000]
[PATCH] Fix SG_IO timeout jiffy conversion

Commit 85e04e371b5a321b5df2bc3f8e0099a64fb087d7 cleaned up the timeout
conversion, but did it exactly the wrong way.  We get msecs from user
space, and should convert them into jiffies. Not the other way around.

Here is a fix with the overflow check sg.c has added in.  This fixes DVD
burnign with Nero.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
[ "you'll be wanting a comma there" - Andrew ]
Cc: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] Fix try_to_free_buffer() locking
Nick Piggin [Tue, 30 Jan 2007 03:36:27 +0000]
[PATCH] Fix try_to_free_buffer() locking

Fix commit ecdfc9787fe527491baefc22dce8b2dbd5b2908d

Not to put too fine a point on it, but in a nutshell...

__set_page_dirty_buffers() | try_to_free_buffers()
---------------------------+---------------------------
                           | spin_lock(private_lock);
                           | drop_bufers()
                           | spin_unlock(private_lock);
spin_lock(private_lock)    |
!page_has_buffers()        |
spin_unlock(private_lock)  |
SetPageDirty()             |
                           | cancel_dirty_page()

                          oops!

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] sky2: revert IRQ dance on suspend/resume
Stephen Hemminger [Tue, 30 Jan 2007 00:38:07 +0000]
[PATCH] sky2: revert IRQ dance on suspend/resume

Let's just backout the IRQ hack, and for those crap machines (like some
Sony VAIO's) can just disable MSI with the module parameter.

This reverts 44ade178249fe53d055fd92113eaa271e06acddd.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Frédéric Riss <frederic.riss@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agoFix balance_dirty_page() calculations with CONFIG_HIGHMEM
Linus Torvalds [Tue, 30 Jan 2007 00:37:38 +0000]
Fix balance_dirty_page() calculations with CONFIG_HIGHMEM

This makes balance_dirty_page() always base its calculations on the
amount of non-highmem memory in the machine, rather than try to base it
on total memory and then falling back on non-highmem memory if the
mapping it was writing wasn't highmem capable.

This not only fixes a situation where two different writers can have
wildly different notions about what is a "balanced" dirty state, but it
also means that people with highmem machines don't run into an OOM
situation when regular memory fills up with dirty pages.

We used to try to handle the latter case by scaling down the dirty_ratio
if the machine had a lot of highmem pages in page_writeback_init(), but
it wasn't aggressive enough for some situations, and since basing the
dirty ratio on highmem memory was broken in the first place, let's just
stop doing so.

(A variation of this theme fixed Justin Piszcz's OOM problem when
copying an 18GB file on a RAID setup).

Acked-by: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Justin Piszcz <jpiszcz@lucidpixels.com>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Neil Brown <neilb@suse.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Randy Dunlap <rdunlap@xenotime.net>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] ALSA: Fix sysfs breakage
Takashi Iwai [Fri, 26 Jan 2007 11:40:31 +0000]
[PATCH] ALSA: Fix sysfs breakage

The recent change for a new sysfs tree with card* object breaks the
/sys/class/sound tree if CONFIG_SYSFS_DEPRECATED is enabled.
The device in each entry doesn't point the correct device object:

  /sys/class/sound
  ...
  |-- pcmC0D0c
  |   |-- dev
  |   |-- device -> ../../../class/sound/card0
  |   |-- pcm_class
  |   |-- power
  |   |   `-- wakeup
  |   |-- subsystem -> ../../../class/sound
  |   `-- uevent

Also, this change breaks some drivers (like sound/arm/*) referring
card->dev directly to obtain the device object for memory handling.

This patch reverts the semantics of card->dev to the former version,
which points to a real device object.  The card* object is stored in a
new card->card_dev field, instead.  The device parent is chosen either
card->dev or card->card_dev according to CONFIG_SYSFS_DEPRECATED to
keep the tree compatibility.
Also, card* isn't created if CONFIG_SYSFS_DEPRECATED is enabled.  The
reason of card* object is a root of all beloing devices, and it makes
little sense if each sound device points to the real device object
directly.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Acked-by: Monty Montgomery <xiphmont@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

12 years ago[ARM] 4117/1: S3C2412: Fix writel() usage in selection code
Ben Dooks [Sun, 28 Jan 2007 23:11:29 +0000]
[ARM] 4117/1: S3C2412: Fix writel() usage in selection code

The S3C2412 DMA selection code has the
arguments to writel() the wrong way around.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

12 years ago[CPUFREQ] Remove unneeded errata workaround from p4-clockmod.
Dave Jones [Mon, 29 Jan 2007 05:07:04 +0000]
[CPUFREQ] Remove unneeded errata workaround from p4-clockmod.

This workaround unnecessarily cripples functionality to work
around an errata that doesn't seem possible to hit due to
us using the automatic clock throttling in the p4 mcheck code.

See http://lkml.org/lkml/2006/10/28/148 for complete reasoning
and lack of disconsent.

Signed-off-by: Dave Jones <davej@redhat.com>

12 years ago[CPUFREQ] check sysfs_create_link return value
Ahmed S. Darwish [Fri, 5 Jan 2007 03:44:54 +0000]
[CPUFREQ] check sysfs_create_link return value

Trivial patch to check sysfs_create_link return values.
Fail gracefully if needed.

Signed-off-by: Ahmed Darwish <darwish.07@gmail.com>
Signed-off-by: Dave Jones <davej@redhat.com>

12 years ago[AGPGART] Add new IDs to VIA AGP.
Dave Jones [Sun, 28 Jan 2007 22:58:33 +0000]
[AGPGART] Add new IDs to VIA AGP.

Culled from the VIA codedrop.
Also fixes up one ID used in amd64-agp to use the
VIA part number instead of the board name in its ID.

Signed-off-by: Dave Jones <davej@redhat.com>

12 years ago[AGPGART] Remove pointless assignment.
Dave Jones [Sun, 28 Jan 2007 22:50:17 +0000]
[AGPGART] Remove pointless assignment.

No point in clearing local pointers then returning.
Also fix up some CodingStyle nits.

Signed-off-by: Dave Jones <davej@redhat.com>

12 years ago[AGPGART] Remove pointless typedef in ati-agp
Dave Jones [Sun, 28 Jan 2007 22:41:37 +0000]
[AGPGART] Remove pointless typedef in ati-agp

This seems to exist just to save people typing 'struct' a few times,
and doesn't provide any additional value.

Signed-off-by: Dave Jones <davej@redhat.com>

12 years ago[AGPGART] Prevent (unlikely) memory leak in amd_create_gatt_pages()
Dave Jones [Sun, 28 Jan 2007 22:39:19 +0000]
[AGPGART] Prevent (unlikely) memory leak in amd_create_gatt_pages()

If we fail an alloc, unwind the previous allocs that succeeded.

Spotted-by: Alan Grimes <agrimes@speakeasy.net>
Signed-off-by: Dave Jones <davej@redhat.com>

12 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Sun, 28 Jan 2007 20:45:22 +0000]
Merge branch 'merge' of git://git./linux/kernel/git/paulus/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
  [POWERPC] Fix sys_pciconfig_iobase bus matching
  [POWERPC] PS3: add not complete comment to kconfig
  [POWERPC] ps3_free_io_irq: Fix inverted error check
  [POWERPC] PS3: Fix uniprocessor kernel build

12 years ago[POWERPC] Fix sys_pciconfig_iobase bus matching
Benjamin Herrenschmidt [Sat, 27 Jan 2007 20:45:53 +0000]
[POWERPC] Fix sys_pciconfig_iobase bus matching

A stupid bug has been plaguing the sys_pciconfig_iobase on ppc64. It wasn't
noticed until recently as it seems to not affect G5s but it's been causing
problems running X servers on some other machines recently. The bus number
matching was bogus.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

12 years ago[POWERPC] PS3: add not complete comment to kconfig
Geoff Levand [Fri, 26 Jan 2007 02:07:14 +0000]
[POWERPC] PS3: add not complete comment to kconfig

Add a comment to the PS3 config option to inform users that the current
implementation is not yet complete.

Signed-off-by: Geoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

12 years ago[SCSI] st: A MTIOCTOP/MTWEOF within the early warning will cause the file number...
Kai Makisara [Thu, 25 Jan 2007 22:38:39 +0000]
[SCSI] st: A MTIOCTOP/MTWEOF within the early warning will cause the file number to be incorrect

On Wed, 24 Jan 2007, Andrew Morton wrote:

> On Mon, 22 Jan 2007 13:07:20 -0800
> bugme-daemon@bugzilla.kernel.org wrote:
>
> > http://bugzilla.kernel.org/show_bug.cgi?id=7864
> >
> >            Summary: A MTIOCTOP/MTWEOF within the early warning will cause
> >                     the file number to be incorrect
> >     Kernel Version: 2.6.19.2
> >             Status: NEW
> >           Severity: low
> >              Owner: io_scsi@kernel-bugs.osdl.org
> >          Submitter: ce_reisinger@yahoo.com
> >
> >
> > Write records to a SCSI tape until a write fails with a ENOSPC (you have reached
> > early warning.
> > Now perform a:
> >    struct mtget before, after;
> >    ioctl(fd, MTIOCGET, &before);
> >    struct mtop mtop = { MTWEOF, 1 };
> >    ioctl(fd, MTIOCTOP, &mtop);
> >    ioctl(fd, MTIOCGET, &after);
> >
> > Check the value of mt_fileno in the before and after structures. Notice the
> > after is 2 greater then the before.
> >
> > The problem appears to be in the block of code starting at line 2817 in st.c.
> > This block is entered because the drive did return a CHECK CONDITION with NO
> > SENSE and the SENSE_EOM bit set. At lines 2824/5 the fileno is incremented. But
> > it has already been increased by the number of filemarks requested by the
> > MTIOCTOP. I believe that the residue count in the sense data should be
> > subtracted from fileno, not a increment as is done.
> >
>
> Thanks.  Could you please send us a tested patch to fix these things, as
> per http://www.zip.com.au/~akpm/linux/patches/stuff/tpp.txt ?
>
The analysis is basically correct and explains the bug. According to the
SCSI standards, the sense code is NO SENSE or RECOVERED ERROR in case
writing filemark(s) succeeds. If it fails (partly or completely) the sense
code is VOLUME OVERFLOW. The patch below is tested to fix the case when
one filemark is successfully written after the EOM early warning. It
should also fix the case at real EOM but this has not been tested.

Carl, thanks for reporting the bug and providing the analysis for the fix.

Signed-off-by: Kai Makisara <kai.makisara@kolumbus.fi>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

12 years ago[SCSI] qla4xxx: bug fixes
David C Somayajulu [Mon, 22 Jan 2007 20:26:11 +0000]
[SCSI] qla4xxx: bug fixes

The included patch fixes the following issues:

1. qla3xxx/qla4xxx co-existence issue which can result in a lockup
when qla3xxx driver is unloaded, or when ifdown; ifup is performed on
one of the interfaces correponding to qla3xxx. This is because qla4xxx
HBA supports one ethernet and iscsi interfaces per port. Both iscsi
and ethernet interfaces share the same state machine. The problem has
to do with synchronizing access to the state machine in the event of a
reset

2. mutex_lock() is sometimes not followed by mutex_unlock() prior to
invoking a msleep() in qla4xxx_mailbox_command()

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

12 years ago[SCSI] Fix scsi_add_device() for async scanning
Matthew Wilcox [Tue, 16 Jan 2007 01:07:09 +0000]
[SCSI] Fix scsi_add_device() for async scanning

I had thought that all drivers which didn't call scsi_scan_host()
called scsi_scan_target().  Some, such as sbp2, mptsas and libata-scsi,
call scsi_add_device() or __scsi_add_device().  We just need to wait
for the currently executing async scans to complete first.  This is the
same code that's in scsi_scan_target(), except that we have to return
an error instead of void when we're declining to scan at all.

Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

12 years agovia82cxxx/pata_via: correct PCI_DEVICE_ID_VIA_SATA_EIDE ID and add support for CX700...
Josepch Chan [Sat, 27 Jan 2007 12:47:08 +0000]
via82cxxx/pata_via: correct PCI_DEVICE_ID_VIA_SATA_EIDE ID and add support for CX700 and 8237S

This patch:
* Corrects the wrong device ID of PCI_DEVICE_ID_VIA_SATA_EIDE
  from 0x0581 to 0x5324.
* Adds VIA CX700 and VT8237S support in drivers/ide/pci/via82cxxx.c
* Adds VIA VT8237S support in drivers/ata/pata_via.c

Signed-off-by: Josepch Chan <josephchan@via.com.tw>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agoide: unregister idepnp driver on unload
Tejun Heo [Sat, 27 Jan 2007 12:47:02 +0000]
ide: unregister idepnp driver on unload

idepnp driver is registered as a pnp driver on ide init but doesn't
get unregistered on ide unload causing driver list corruption and
eventually oops.  Fix it.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agoide: add missing __init tags to IDE PCI host drivers
Bartlomiej Zolnierkiewicz [Sat, 27 Jan 2007 12:46:56 +0000]
ide: add missing __init tags to IDE PCI host drivers

also change __devinit tag for sgiioc4.c:ioc4_ide_init() to __init

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agoia64: add pci_get_legacy_ide_irq()
Bartlomiej Zolnierkiewicz [Sat, 27 Jan 2007 12:46:54 +0000]
ia64: add pci_get_legacy_ide_irq()

Add pci_get_legacy_ide_irq() identical to the one used by i386/x86_64.
Fixes amd74xx driver build on ia64 (bugzilla bug #6644).

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agoide/generic: Jmicron has its own drivers now
Alan Cox [Sat, 27 Jan 2007 12:46:45 +0000]
ide/generic: Jmicron has its own drivers now

Drop ide-generic support for Jmicron identifiers as we now trust Jmicron.c for
this with drivers/ide.  The code check remains for the all-generic-ide case.

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agoatiixp.c: add cable detection support for ATI IDE
Conke Hu [Sat, 27 Jan 2007 12:46:40 +0000]
atiixp.c: add cable detection support for ATI IDE

IDE HDD does not work if it uses a 40-pin PATA cable on ATI chipset.
This patch fixes the bug.

Signed-off-by: Conke Hu <conke.hu@amd.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agoatiixp.c: sb600 ide only has one channel
Conke Hu [Sat, 27 Jan 2007 12:46:30 +0000]
atiixp.c: sb600 ide only has one channel

AMD/ATI SB600 IDE/PATA controller only has one channel.

Signed-off-by: Conke Hu <conke.hu@amd.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agoatiixp.c: remove unused code
Conke Hu [Sat, 27 Jan 2007 12:46:28 +0000]
atiixp.c: remove unused code

A previous patch to atiixp.c was removed but some code has not been
cleaned. Now we remove these code sine they are no use any longer.

Signed-off-by: Conke Hu <conke.hu@amd.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agojmicron: fix warning
Andrew Morton [Sat, 27 Jan 2007 12:46:21 +0000]
jmicron: fix warning

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agoide: update MAINTAINERS entry
Bartlomiej Zolnierkiewicz [Sat, 27 Jan 2007 12:46:14 +0000]
ide: update MAINTAINERS entry

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

12 years agolibata: fix ata_eh_suspend() return value
Tejun Heo [Fri, 26 Jan 2007 11:10:25 +0000]
libata: fix ata_eh_suspend() return value

ata_eh_suspend() was returning 0 regardless of failure.  This bug has
potential to lose data on suspend.  Fix it.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoahci: port_no should be used when clearing IRQ in ahci_thaw()
Tejun Heo [Sat, 27 Jan 2007 02:04:26 +0000]
ahci: port_no should be used when clearing IRQ in ahci_thaw()

ap->id is logcial port ID which is unique among all ATA ports and
doesn't have anything to do with hardware port index.  ap->port_no is
the hardware port index and thus should be used when clearing IRQ mask
in ahci_thaw().

This problem has been spotted by Jeff Garzik <jgarzik@pobox.com>.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 27 Jan 2007 03:39:40 +0000]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [IPV4]: Fix single-entry /proc/net/fib_trie output.
  [SELINUX]: Fix 2.6.20-rc6 build when no xfrm

12 years ago[IPV4]: Fix single-entry /proc/net/fib_trie output.
Robert Olsson [Sat, 27 Jan 2007 03:06:01 +0000]
[IPV4]: Fix single-entry /proc/net/fib_trie output.

When main table is just a single leaf this gets printed as belonging to the
local table in /proc/net/fib_trie. A fix is below.

Signed-off-by: Robert Olsson <robert.olsson@its.uu.se>
Acked-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years ago[SELINUX]: Fix 2.6.20-rc6 build when no xfrm
Venkat Yekkirala [Sat, 27 Jan 2007 03:03:48 +0000]
[SELINUX]: Fix 2.6.20-rc6 build when no xfrm

This patch is an incremental fix to the flow_cache_genid
patch for selinux that breaks the build of 2.6.20-rc6 when
xfrm is not configured.

Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years ago[SPARC64]: Set g4/g5 properly in sun4v dtlb-prot handling.
David S. Miller [Sat, 27 Jan 2007 02:48:16 +0000]
[SPARC64]: Set g4/g5 properly in sun4v dtlb-prot handling.

Mirror the logic in the sun4u handler, we have to update
both registers even when we branch out to window fault
fixup handling.

The way it works is that if we are in etrap processing a
fault already, g4/g5 holds the original fault information.
If we take a window spill fault while doing etrap, then
we put the window spill fault info into g4/g5 and this is
what the top-level fault handler ends up processing first.

Then we retry the originally faulting instruction, and
process the original fault at that time.

This is all necessary because of how constrained the trap
registers are in these code paths.  These cases trigger
very rarely, so even if there is some performance implication
it's doesn't happen very often.  In fact the rarity is why
it took so long to trigger and find this particular bug.

Signed-off-by: David S. Miller <davem@davemloft.net>

12 years ago[PATCH] Boot loader ID for Gujin
H. Peter Anvin [Fri, 26 Jan 2007 17:15:47 +0000]
[PATCH] Boot loader ID for Gujin

Add an official boot loader ID for Gujin.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] ocfs2: fix thinko in ocfs2_backup_super_blkno()
Mark Fasheh [Fri, 26 Jan 2007 18:46:59 +0000]
[PATCH] ocfs2: fix thinko in ocfs2_backup_super_blkno()

Fix a bug which was introduced when I synced up ocfs2_fs.h with ocfs2-tools.
We can't do u64/u32 in kernel.

Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agoMerge branch 'for-linus' of git://www.atmel.no/~hskinnemoen/linux/kernel/avr32
Linus Torvalds [Fri, 26 Jan 2007 22:48:53 +0000]
Merge branch 'for-linus' of git://atmel.no/~hskinnemoen/linux/kernel/avr32

* 'for-linus' of git://www.atmel.no/~hskinnemoen/linux/kernel/avr32:
  [AVR32] Update ATSTK1000 defconfig: Enable macb by default
  [AVR32] Export clear_page symbol

12 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Fri, 26 Jan 2007 22:47:05 +0000]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [NETFILTER]: nf_conntrack_pptp: fix NAT setup of expected GRE connections
  [NETFILTER]: nf_nat_pptp: fix expectation removal
  [NETFILTER]: nf_nat: fix ICMP translation with statically linked conntrack
  [TCP]: Restore SKB socket owner setting in tcp_transmit_skb().
  [AF_PACKET]: Check device down state before hard header callbacks.
  [DECNET]: Handle a failure in neigh_parms_alloc (take 2)
  [BNX2]: Fix 2nd port's MAC address.
  [TCP]: Fix sorting of SACK blocks.
  [AF_PACKET]: Fix BPF handling.
  [IPV4]: Fix the fib trie iterator to work with a single entry routing tables

12 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Fri, 26 Jan 2007 22:45:18 +0000]
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  Fix Maple PATA IRQ assignment.
  ahci: use 0x80 as wait stat value instead of 0xff
  sata_via: style clean up, no indirect method call in LLD
  ahci: fix endianness in spurious interrupt message
  libata-sff: Don't call bmdma_stop on non DMA capable controllers
  libata: implement ATA_FLAG_IGN_SIMPLEX and use it in sata_uli
  ahci: improve and limit spurious interrupt messages, take#3
  sata_via: don't diddle with ATA_NIEN in ->freeze
  libata: set_mode, Fix the FIXME
  libata hpt3xn: Hopefully sort out the DPLL logic versus the vendor code
  libata cmd64x: whack into a shape that looks like the documentation

12 years agoFix Maple PATA IRQ assignment.
David Woodhouse [Mon, 1 Jan 2007 19:31:15 +0000]
Fix Maple PATA IRQ assignment.

On the Maple board, the AMD8111 IDE is in legacy mode... except that it
appears on IRQ 20 instead of IRQ 15. For drivers/ide this was handled by
the architecture's "pci_get_legacy_ide_irq()" function, but in libata we
just hard-code the numbers 14 and 15.

This patch provides asm-powerpc/libata-portmap.h which maps the IRQ as
appropriate, having added a pci_dev argument to the
ATA_{PRIM,SECOND}ARY_IRQ macros.

There's probably a better way to do this -- especially if we observe
that the _only_ case in which this seemingly-generic
"pci_get_legacy_ide_irq()" function returns anything other than 14 and
15 for primary and secondary respectively is the case of the AMD8111 on
the Maple board -- couldn't we handle that with a special case in the
pata_amd driver, or perhaps with a PCI quirk for Maple to switch it into
native mode during early boot and assign resources properly?

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoahci: use 0x80 as wait stat value instead of 0xff
Tejun Heo [Fri, 26 Jan 2007 06:37:20 +0000]
ahci: use 0x80 as wait stat value instead of 0xff

Before hardreset, ahci initialized stat part of received FIS area to
0xff to wait for the first D2H Reg FIS which would change the value to
device ready state.  This used to work but now libata considers status
value of 0xff as device not present making this wait prone to failure.

This patch makes ahci use 0x80 for the wait stat value instead of
0xff to fix the above problem.

Signed-off-by: Tejun Heo <htejun@gmail.com>

 drivers/ata/ahci.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agosata_via: style clean up, no indirect method call in LLD
Tejun Heo [Fri, 26 Jan 2007 05:57:31 +0000]
sata_via: style clean up, no indirect method call in LLD

Call ata_bmdma_irq_clear() directly instead of through
ap->ops->irq_clear() according to libata style guideline.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years agoahci: fix endianness in spurious interrupt message
Tejun Heo [Fri, 26 Jan 2007 05:47:38 +0000]
ahci: fix endianness in spurious interrupt message

Fix endianness in spurious interrupt message.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

12 years ago[PATCH] Fix race in efi variable delete code
Matt Domsch [Fri, 26 Jan 2007 08:57:18 +0000]
[PATCH] Fix race in efi variable delete code

Fix race when deleting an EFI variable and issuing another EFI command on
the same variable.  The removal of the variable from the efivars_list
should be done in efivar_delete and not delayed until the kobject release.

Furthermore, remove the item from the list at module unload time, and use
list_for_each_entry_safe() rather than list_for_each_safe() for
readability.

Tested on ia64.

Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Matt Domsch <Matt_Domsch@dell.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] netdev: add a MAINTAINERS entry for via-velocity and update my address
Francois Romieu [Fri, 26 Jan 2007 08:57:17 +0000]
[PATCH] netdev: add a MAINTAINERS entry for via-velocity and update my address

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] core-dumping unreadable binaries via PT_INTERP
Alexey Dobriyan [Fri, 26 Jan 2007 08:57:16 +0000]
[PATCH] core-dumping unreadable binaries via PT_INTERP

Proposed patch to fix #5 in
http://www.isec.pl/vulnerabilities/isec-0017-binfmt_elf.txt
aka
http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2004-1073

To reproduce, do
* grab poc at the end of advisory.
* add line "eph.p_memsz = 4096;" after "eph.p_filesz = 4096;"
  where first "4096" is something equal to or greater than 4096.
* ./poc /usr/bin/sudo && ls -l

Here I get with 2.6.20-rc5:

 -rw------- 1 ad   ad   102400 2007-01-15 19:17 core
 ---s--x--x 2 root root 101820 2007-01-15 19:15 /usr/bin/sudo

Check for MAY_READ like binfmt_misc.c does.

Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] md: remove unnecessary printk when raid5 gets an unaligned read.
NeilBrown [Fri, 26 Jan 2007 08:57:14 +0000]
[PATCH] md: remove unnecessary printk when raid5 gets an unaligned read.

raid5_mergeable_bvec tries to ensure that raid5 never sees a read request
that does not fit within just one chunk.  However as we must always accept
a single-page read, that is not always possible.

So when "in_chunk_boundary" fails, it might be unusual, but it is not a
problem and printing a message every time is a bad idea.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] Fix UML on non-standard VM split hosts
Jeff Dike [Fri, 26 Jan 2007 08:57:12 +0000]
[PATCH] Fix UML on non-standard VM split hosts

This fixes UML on hosts with non-standard VM splits.  We had changed the
config variable that controls UML behavior on such hosts, but not
propogated the change everywhere.  In particular, the values of STUB_CODE
and STUB_DATA relied on the old variable.

I also reformatted the HOST_VMSPLIT_3G help to make it more standard.

Spotted by uml@flonatel.org.

Signed-off-by: Jeff Dike <jdike@addtoit.com>
Cc: Blaisorblade <blaisorblade@yahoo.it>
Cc: Pravin <shindepravin@gmail.com>
Cc: <uml@flonatel.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] MM: Remove [PATCH] invalidate_inode_pages2_range() debug
Trond Myklebust [Fri, 26 Jan 2007 08:57:11 +0000]
[PATCH] MM: Remove [PATCH] invalidate_inode_pages2_range() debug

NFS can handle the case where invalidate_inode_pages2_range() fails, so the
premise behind commit 8258d4a574d3a8c01f0ef68aa26b969398a0e140 is now gone.

Remove the WARN_ON_ONCE() which is causing users grief as we can see from
http://bugzilla.kernel.org/show_bug.cgi?id=7826

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] md: fix potential memalloc deadlock in md
NeilBrown [Fri, 26 Jan 2007 08:57:11 +0000]
[PATCH] md: fix potential memalloc deadlock in md

If a GFP_KERNEL allocation is attempted in md while the mddev_lock is held,
it is possible for a deadlock to eventuate.

This happens if the array was marked 'clean', and the memalloc triggers a
write-out to the md device.

For the writeout to succeed, the array must be marked 'dirty', and that
requires getting the mddev_lock.

So, before attempting a GFP_KERNEL allocation while holding the lock, make
sure the array is marked 'dirty' (unless it is currently read-only).

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] knfsd: Fix type mismatch with filldir_t used by nfsd
NeilBrown [Fri, 26 Jan 2007 08:57:10 +0000]
[PATCH] knfsd: Fix type mismatch with filldir_t used by nfsd

nfsd defines a type 'encode_dent_fn' which is much like 'filldir_t' except
that the first pointer is 'struct readdir_cd *' rather than 'void *'.  It
then casts encode_dent_fn points to 'filldir_t' as needed.  This hides any
other type mismatches between the two such as the fact that the 'ino' arg
recently changed from ino_t to u64.

So: get rid of 'encode_dent_fn', get rid of the cast of the function type,
change the first arg of various functions from 'struct readdir_cd *' to
'void *', and live with the fact that we have a little less type checking
on the calling of these functions now.  Less internal (to nfsd) checking
offset by more external checking, which is more important.

Thanks to Gabriel Paubert <paubert@iram.es> for discovering this and
providing an initial patch.

Signed-off-by: Gabriel Paubert <paubert@iram.es>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] fix various kernel-doc in header files
Robert P. J. Day [Fri, 26 Jan 2007 08:57:09 +0000]
[PATCH] fix various kernel-doc in header files

Fix a number of kernel-doc entries for header files in include/linux by
making sure they begin with the appropriate '/**' notation and use @var
notation.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] remove __devinit markings from rtc_sysfs_add_device()
Mike Frysinger [Fri, 26 Jan 2007 08:57:08 +0000]
[PATCH] remove __devinit markings from rtc_sysfs_add_device()

rtc_sysfs_add_device is needed even after dev initialization, so drop __devinit.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] dm-multipath: fix stall on noflush suspend/resume
Jun'ichi Nomura [Fri, 26 Jan 2007 08:57:07 +0000]
[PATCH] dm-multipath: fix stall on noflush suspend/resume

Allow noflush suspend/resume of device-mapper device only for the case
where the device size is unchanged.

Otherwise, dm-multipath devices can stall when resumed if noflush was used
when suspending them, all paths have failed and queue_if_no_path is set.

Explanation:
 1. Something is doing fsync() on the block dev,
    holding inode->i_sem
 2. The fsync write is blocked by all-paths-down and queue_if_no_path
 3. Someone requests to suspend the dm device with noflush.
    Pending writes are left in queue.
 4. In the middle of dm_resume(), __bind() tries to get
    inode->i_sem to do __set_size() and waits forever.

'noflush suspend' is a new device-mapper feature introduced in
early 2.6.20. So I hope the fix being included before 2.6.20 is
released.

Example of reproducer:
 1. Create a multipath device by dmsetup
 2. Fail all paths during mkfs
 3. Do dmsetup suspend --noflush and load new map with healthy paths
 4. Do dmsetup resume

Signed-off-by: Jun'ichi Nomura <j-nomura@ce.jp.nec.com>
Acked-by: Alasdair G Kergon <agk@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] 9p: null terminate error strings for debug print
Eric Van Hensbergen [Fri, 26 Jan 2007 08:57:06 +0000]
[PATCH] 9p: null terminate error strings for debug print

We weren't properly NULL terminating protocol error strings for our debug
printk resulting in garbage being included in the output when debug was
enabled.

Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] 9p: fix segfault caused by race condition in meta-data operations
Eric Van Hensbergen [Fri, 26 Jan 2007 08:57:06 +0000]
[PATCH] 9p: fix segfault caused by race condition in meta-data operations

Running dbench multithreaded exposed a race condition where fid structures
were removed while in use.  This patch adds semaphores to meta-data operations
to protect the fid structure.  Some cleanup of error-case handling in the
inode operations is also included.

Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] 9p: update documentation regarding server applications
Eric Van Hensbergen [Fri, 26 Jan 2007 08:57:05 +0000]
[PATCH] 9p: update documentation regarding server applications

Update the documentation to cover using Inferno as a server for 9p and to
include information about spfs (a stable single-threaded stand-alone 9p
server).

Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] 9p: fix rename return code
Eric Van Hensbergen [Fri, 26 Jan 2007 08:57:04 +0000]
[PATCH] 9p: fix rename return code

9p doesn't handle renames between directories -- however, we were returning
EPERM instead of EXDEV when we detected this case.

Signed-off-by: Eric Van Hensbergren <ericvh@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] 9p: fix bogus return code checks during initialization
Eric Van Hensbergen [Fri, 26 Jan 2007 08:57:04 +0000]
[PATCH] 9p: fix bogus return code checks during initialization

There is a simple logic error in init_v9fs - the return code checks are
reversed.  This patch fixes the return code and adds some messages to prevent
module initialization from failing silently.

Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] md: avoid reading past the end of a bitmap file
NeilBrown [Fri, 26 Jan 2007 08:57:03 +0000]
[PATCH] md: avoid reading past the end of a bitmap file

In most cases we check the size of the bitmap file before reading data from
it.  However when reading the superblock, we always read the first PAGE_SIZE
bytes, which might not always be appropriate.  So limit that read to the size
of the file if appropriate.

Also, we get the count of available bytes wrong in one place, so that too can
read past the end of the file.

Cc: "yang yin" <yinyang801120@gmail.com>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] md: make sure the events count in an md array never returns to zero
NeilBrown [Fri, 26 Jan 2007 08:57:02 +0000]
[PATCH] md: make sure the events count in an md array never returns to zero

Now that we sometimes step the array events count backwards (when
transitioning dirty->clean where nothing else interesting has happened - so
that we don't need to write to spares all the time), it is possible for the
event count to return to zero, which is potentially confusing and triggers and
MD_BUG.

We could possibly remove the MD_BUG, but is just as easy, and probably safer,
to make sure we never return to zero.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] md: make 'repair' actually work for raid1
NeilBrown [Fri, 26 Jan 2007 08:57:01 +0000]
[PATCH] md: make 'repair' actually work for raid1

When 'repair' finds a block that is different one the various parts of the
mirror.  it is meant to write a chosen good version to the others.  However it
currently writes out the original data to each.  The memcpy to make all the
data the same is missing.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] md: update email address and status for MD in MAINTAINERS
NeilBrown [Fri, 26 Jan 2007 08:57:01 +0000]
[PATCH] md: update email address and status for MD in MAINTAINERS

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] knfsd: Don't mess with the 'mode' when storing a exclusive-create cookie
Peter Staubach [Fri, 26 Jan 2007 08:57:00 +0000]
[PATCH] knfsd: Don't mess with the 'mode' when storing a exclusive-create cookie

NFS V3 (and V4) support exclusive create by passing a 'cookie' which can get
stored with the file.  If the file exists but has exactly the right cookie
stored, then we assume this is a retransmit and the exclusive create was
successful.

The cookie is 64bits and is traditionally stored in the mtime and atime
fields.  This causes a problem with Solaris7 as negative mtime or atime
confuse it.  So we moved two bits into the mode word instead.

But inherited ACLs sometimes overwrite the mode word on create, so this is a
problem.

So we give up and just store 62 of the 64 bits and assume that is close
enough.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] knfsd: replace some warning ins nfsfh.h with BUG_ON or WARN_ON
NeilBrown [Fri, 26 Jan 2007 08:56:59 +0000]
[PATCH] knfsd: replace some warning ins nfsfh.h with BUG_ON or WARN_ON

A couple of the warnings will be followed by an Oops if they ever fire, so may
as well be BUG_ON.  Another isn't obviously fatal but has never been known to
fire, so make it a WARN_ON.

Cc: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] knfsd: fix an NFSD bug with full sized, non-page-aligned reads
NeilBrown [Fri, 26 Jan 2007 08:56:59 +0000]
[PATCH] knfsd: fix an NFSD bug with full sized, non-page-aligned reads

NFSd assumes that largest number of pages that will be needed for a
request+response is 2+N where N pages is the size of the largest permitted
read/write request.  The '2' are 1 for the non-data part of the request, and 1
for the non-data part of the reply.

However, when a read request is not page-aligned, and we choose to use
->sendfile to send it directly from the page cache, we may need N+1 pages to
hold the whole reply.  This can overflow and array and cause an Oops.

This patch increases size of the array for holding pages by one and makes sure
that entry is NULL when it is not in use.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] knfsd: fix setting of ACL server versions
NeilBrown [Fri, 26 Jan 2007 08:56:58 +0000]
[PATCH] knfsd: fix setting of ACL server versions

Due to silly typos, if the nfs versions are explicitly set, no NFSACL versions
get enabled.

Also improve an error message that would have made this bug a little easier to
find.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] knfsd: update email address and status for NFSD in MAINTAINERS
NeilBrown [Fri, 26 Jan 2007 08:56:57 +0000]
[PATCH] knfsd: update email address and status for NFSD in MAINTAINERS

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years ago[PATCH] Gigaset ISDN driver error handling fixes
Tilman Schmidt [Fri, 26 Jan 2007 08:56:56 +0000]
[PATCH] Gigaset ISDN driver error handling fixes

Fix several flaws in the error handling of the Siemens Gigaset ISDN driver,
including one that would cause an Oops when connecting more than one device
of the same type.

Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>