openwrt/staging/blogic.git
17 years agoIB/iser: Don't defer connection failure notification to workqueue
Erez Zilber [Sun, 1 Apr 2007 10:53:43 +0000 (12:53 +0200)]
IB/iser: Don't defer connection failure notification to workqueue

When a connection is terminated asynchronously from the iSCSI layer's
perspective, iSER needs to notify the iSCSI layer that the connection
has failed.  This is done using a workqueue (switched to from the iSER
tasklet context).  Meanwhile, the connection object (that holds the
work struct) is released.  If the workqueue function wasn't called
yet, it will be called later with a NULL pointer, which will crash the
kernel.

The context switch (tasklet to workqueue) is not required, and
everything can be done from the iSER tasklet. This eliminates the NULL
work struct bug (and simplifies the code).

Signed-off-by: Erez Zilber <erezz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
17 years ago[PATCH] ia64: desc_empty thinko/typo fix
Maciej Zenczykowski [Thu, 5 Apr 2007 02:08:27 +0000 (19:08 -0700)]
[PATCH] ia64: desc_empty thinko/typo fix

Just a one-byter for an ia64 thinko/typo - already fixed for i386 and x86_64.

Acked-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] Fix build error on zs serial driver
Yoichi Yuasa [Thu, 5 Apr 2007 02:08:25 +0000 (19:08 -0700)]
[PATCH] Fix build error on zs serial driver

drivers/tc/zs.c:73:24: error: asm/dec/tc.h: No such file or directory

Signed-off-by: Yoichi Yuasa <yoichi_yuasa@tripeaks.co.jp>
Acked-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] fix jiffies clocksource inittime
john stultz [Thu, 5 Apr 2007 02:08:24 +0000 (19:08 -0700)]
[PATCH] fix jiffies clocksource inittime

In debugging a problem w/ the -rt tree, I noticed that on systems that mark
the tsc as unstable before it is registered, the TSC would still be
selected and used for a short period of time.  Digging in it looks to be a
result of the mix of the clocksource list changes and my clocksource
initialization changes.

With the -rt tree, using a bad TSC, even for a short period of time can
results in a hang at boot.  I was not able to reproduce this hang w/
mainline, but I'm not completely certain that someone won't trip on it.

This patch resolves the issue by initializing the jiffies clocksource
earlier so a bad TSC won't get selected just because nothing else is yet
registered.

Signed-off-by: John Stultz <johnstul@us.ibm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] cciss: add init of drv->cylinders back to cciss_geometry_inquiry
Mike Miller (OS Dev) [Thu, 5 Apr 2007 02:08:23 +0000 (19:08 -0700)]
[PATCH] cciss: add init of drv->cylinders back to cciss_geometry_inquiry

This patch adds initialization of drv->cylinders back into the failing case in
cciss_geometry_inquiry. I inadvertently removed it in one my 2TB updates.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] remove protection of LANANA-reserved majors
Andrew Morton [Thu, 5 Apr 2007 02:08:22 +0000 (19:08 -0700)]
[PATCH] remove protection of LANANA-reserved majors

Revert all this.  It can cause device-mapper to receive a different major from
earlier kernels and it turns out that the Amanda backup program (via GNU tar,
apparently) checks major numbers on files when performing incremental backups.

Which is a bit broken of Amanda (or tar), but this feature isn't important
enough to justify the churn.

Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] swsusp: fix memory shrinker
Rafael J. Wysocki [Thu, 5 Apr 2007 02:08:21 +0000 (19:08 -0700)]
[PATCH] swsusp: fix memory shrinker

Fix a bug in the swsusp's memory shrinker that causes some systems using
highmem to refuse to suspend to disk if image_size is set above 1/2 of
available RAM.

Special thanks to Jiri Slaby for reporting the problem and assistance in
debugging it.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Jiri Slaby <jirislaby@gmail.com>
Cc: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] net/sunrpc/svcsock.c: fix a check
Adrian Bunk [Thu, 5 Apr 2007 02:08:21 +0000 (19:08 -0700)]
[PATCH] net/sunrpc/svcsock.c: fix a check

The return value of kernel_recvmsg() should be assigned to "err", not
compared with the random value of a never initialized "err" (and the "< 0"
check wrongly always returned false since == comparisons never have a
result < 0).

Spotted by the Coverity checker.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] md: avoid a deadlock when removing a device from an md array via sysfs
NeilBrown [Thu, 5 Apr 2007 02:08:18 +0000 (19:08 -0700)]
[PATCH] md: avoid a deadlock when removing a device from an md array via sysfs

A device can be removed from an md array via e.g.
  echo remove > /sys/block/md3/md/dev-sde/state

This will try to remove the 'dev-sde' subtree which will deadlock
since
  commit e7b0d26a86943370c04d6833c6edba2a72a6e240

With this patch we run the kobject_del via schedule_work so as to
avoid the deadlock.

Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] kernel/time.c: add missing symbol exports
Thomas Bittermann [Wed, 4 Apr 2007 20:20:54 +0000 (22:20 +0200)]
[PATCH] kernel/time.c: add missing symbol exports

This patch adds 2 missing symbol exports: jiffies_to_timeval() and
timeval_to_jiffies().  The (not yet merged) dm-raid4-5 module will need
them, and they used to be indirectly exported by virtue of being inline
functions.

Commit 8b9365d753d9870bb6451504c13570b81923228f ("[PATCH] Uninline
jiffies.h functions") uninlined them, and thus modules now need them
explicitly exported to use them.

Signed-off-by: Thomas Bittermann <t.bittermann@online.de>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: john stultz <johnstul@us.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Wed, 4 Apr 2007 17:11:16 +0000 (10:11 -0700)]
Merge branch 'for-linus' of git://git390.osdl.marist.edu/linux-2.6

* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
  [S390] cio: Fix handling of interrupt for csch().
  [S390] page_mkclean data corruption.

17 years ago[PATCH] net: Ignore sysfs network device rename bugs.
Eric W. Biederman [Tue, 3 Apr 2007 06:07:30 +0000 (00:07 -0600)]
[PATCH] net: Ignore sysfs network device rename bugs.

The generic networking code ensures that no two networking devices
have the same name, so  there is no time except when sysfs has
implementation bugs that device_rename when called from
dev_change_name will fail.

The current error handling for errors from device_rename in
dev_change_name is wrong and results in an unusable and unrecoverable
network device if device_rename is happens to return an error.

This patch removes the buggy error handling.  Which confines the mess
when device_rename hits a problem to sysfs, instead of propagating it
the rest of the network stack.  Making linux a little more robust.

Without this patch you can observe what happens when sysfs has a bug
when CONFIG_SYSFS_DEPRECATED is not set and you attempt to rename
a real network device to a name like (broken_parity_status, device,
modalias, power, resource2, subsystem_vendor, class,  driver, irq,
msi_bus, resource, subsystem, uevent, config, enable, local_cpus,
numa_node, resource0, subsystem_device, vendor)

Greg has a patch that fixes the sysfs bugs but he doesn't trust it
for a 2.6.21 timeframe.  This patch which just ignores errors should
be safe and it keeps the system from going completely wacky.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] SLAB: Mention slab name when listing corrupt objects
David Howells [Mon, 2 Apr 2007 22:46:28 +0000 (23:46 +0100)]
[PATCH] SLAB: Mention slab name when listing corrupt objects

Mention the slab name when listing corrupt objects.  Although the function
that released the memory is mentioned, that is frequently ambiguous as such
functions often release several pieces of memory.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Wed, 4 Apr 2007 15:44:40 +0000 (08:44 -0700)]
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:
  libata: Limit ATAPI DMA to R/W commands only for TORiSAN DVD drives (take 3)
  libata: Limit max sector to 128 for TORiSAN DVD drives (take 3)
  libata: Clear tf before doing request sense (take 3)
  libata: reorder HSM_ST_FIRST for easier decoding (take 3)
  libata bugfix: preserve LBA bit for HDIO_DRIVE_TASK
  2.6.21 fix lba48 bug in libata fill_result_tf()

17 years ago[PATCH] libata: add NCQ blacklist entries from Silicon Image Windows driver (v2)
Robert Hancock [Tue, 3 Apr 2007 04:05:29 +0000 (22:05 -0600)]
[PATCH] libata: add NCQ blacklist entries from Silicon Image Windows driver (v2)

This adds some NCQ blacklist entries taken from the Silicon Image 3124/3132
Windows driver .inf files. There are some confirming reports of problems
with these drives under Linux (for example http://lkml.org/lkml/2007/3/4/178)
so let's disable NCQ on these drives.

 [ I'm personally starting to wonder whether we shouldn't disable NCQ by
   default, and perhaps have a white-list. There seems to be a *lot* of
   drives that do this wrong..   - Linus ]

Signed-off-by: Robert Hancock <hancockr@shaw.ca>
Acked-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Wed, 4 Apr 2007 15:36:01 +0000 (08:36 -0700)]
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6:
  r8169: fix suspend/resume for down interface
  r8169: issue request_irq after the private data are completely initialized
  b44: fix IFF_ALLMULTI handling of CAM slots
  cxgb3 - Firwmare update
  cxgb3 - Tighten xgmac workaround
  cxgb3 - detect NIC only adapters
  cxgb3 - Safeguard TCAM size usage

17 years ago[S390] cio: Fix handling of interrupt for csch().
Cornelia Huck [Wed, 4 Apr 2007 12:37:11 +0000 (14:37 +0200)]
[S390] cio: Fix handling of interrupt for csch().

Wipe internal irb if the clear function bit is set before accumulating
bits from the irb in order to follow hardware behaviour.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
17 years ago[S390] page_mkclean data corruption.
Martin Schwidefsky [Wed, 4 Apr 2007 12:37:10 +0000 (14:37 +0200)]
[S390] page_mkclean data corruption.

The git commit c2fda5fed81eea077363b285b66eafce20dfd45a which
added the page_test_and_clear_dirty call to page_mkclean and the
git commit 7658cc289288b8ae7dd2c2224549a048431222b3 which fixes
the "nasty and subtle race in shared mmap'ed page writeback"
problem in clear_page_dirty_for_io cause data corruption on s390.

The effect of the two changes is that for every call to
clear_page_dirty_for_io a page_test_and_clear_dirty is done. If
the per page dirty bit is set set_page_dirty is called. Strangly
clear_page_dirty_for_io is called for not-uptodate pages, e.g.
over this call-chain:

 [<000000000007c0f2>] clear_page_dirty_for_io+0x12a/0x130
 [<000000000007c494>] generic_writepages+0x258/0x3e0
 [<000000000007c692>] do_writepages+0x76/0x7c
 [<00000000000c7a26>] __writeback_single_inode+0xba/0x3e4
 [<00000000000c831a>] sync_sb_inodes+0x23e/0x398
 [<00000000000c8802>] writeback_inodes+0x12e/0x140
 [<000000000007b9ee>] wb_kupdate+0xd2/0x178
 [<000000000007cca2>] pdflush+0x162/0x23c

The bad news now is that page_test_and_clear_dirty might claim
that a not-uptodate page is dirty since SetPageUptodate which
resets the per page dirty bit has not yet been called. The page
writeback that follows clobbers the data on disk.

The simplest solution to this problem is to move the call to
page_test_and_clear_dirty under the "if (page_mapped(page))".
If a file backed page is mapped it is uptodate.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
17 years agolibata: Limit ATAPI DMA to R/W commands only for TORiSAN DVD drives (take 3)
Albert Lee [Mon, 2 Apr 2007 03:39:25 +0000 (11:39 +0800)]
libata: Limit ATAPI DMA to R/W commands only for TORiSAN DVD drives (take 3)

patch 4/4:

  Limit ATAPI DMA to R/W commands only for TORiSAN DRD-N216 DVD-ROM drives
  (http://bugzilla.kernel.org/show_bug.cgi?id=6710)

Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agolibata: Limit max sector to 128 for TORiSAN DVD drives (take 3)
Albert Lee [Mon, 2 Apr 2007 03:34:15 +0000 (11:34 +0800)]
libata: Limit max sector to 128 for TORiSAN DVD drives (take 3)

patch 3/4:
  The TORiSAN drive locks up when max sector == 256.
  Limit max sector to 128 for the TORiSAN DRD-N216 drives.
  (http://bugzilla.kernel.org/show_bug.cgi?id=6710)

Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agolibata: Clear tf before doing request sense (take 3)
Albert Lee [Mon, 2 Apr 2007 03:30:46 +0000 (11:30 +0800)]
libata: Clear tf before doing request sense (take 3)

patch 2/4:
  Clear tf before doing request sense.

This fixes the AOpen 56X/AKH timeout problem.
(http://bugzilla.kernel.org/show_bug.cgi?id=8244)

Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agolibata: reorder HSM_ST_FIRST for easier decoding (take 3)
Albert Lee [Mon, 2 Apr 2007 03:28:52 +0000 (11:28 +0800)]
libata: reorder HSM_ST_FIRST for easier decoding (take 3)

patch 1/4:
  Reorder HSM_ST_FIRST, such that the task state transition is easier decoded with human eyes.

Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agolibata bugfix: preserve LBA bit for HDIO_DRIVE_TASK
Mark Lord [Fri, 30 Mar 2007 21:45:52 +0000 (17:45 -0400)]
libata bugfix: preserve LBA bit for HDIO_DRIVE_TASK

Preserve the LBA bit in the DevSel/Head register for HDIO_DRIVE_TASK.

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years ago2.6.21 fix lba48 bug in libata fill_result_tf()
Mark Lord [Mon, 2 Apr 2007 20:20:35 +0000 (16:20 -0400)]
2.6.21 fix lba48 bug in libata fill_result_tf()

Current 2.6.21 libata does the following:

void ata_tf_read(struct ata_port *ap, struct ata_taskfile *tf)
{
        struct ata_ioports *ioaddr = &ap->ioaddr;

        tf->command = ata_check_status(ap);
...
        if (tf->flags & ATA_TFLAG_LBA48) {
                iowrite8(tf->ctl | ATA_HOB, ioaddr->ctl_addr);
                tf->hob_feature = ioread8(ioaddr->error_addr);
                ...
        }
}
...
static void fill_result_tf(struct ata_queued_cmd *qc)
{
        struct ata_port *ap = qc->ap;

        ap->ops->tf_read(ap, &qc->result_tf);
        qc->result_tf.flags = qc->tf.flags;
}

Based on this, those last two statements fill_result_tf()
appear to me to be in the wrong order, in that the tf->flags
are uninitialized at the point where tf_read() is invoked.
So for lba48 commands, tf_read() won't be reading back the
full lba48 register contents..

Correct?

This patch corrects fill_result_tf() so that the flags
get copied to result_tf before they are used by tf_read().

Signed-off-by: Mark Lord <mlord@pobox.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agor8169: fix suspend/resume for down interface
Francois Romieu [Mon, 2 Apr 2007 21:01:11 +0000 (23:01 +0200)]
r8169: fix suspend/resume for down interface

The PM hooks are no-op if the r8169 interface is down (i.e. !IFF_UP).
However, as the chipset is enabled, the device will not work after a
suspend/resume cycle. The patch always issue the required PCI suspend
sequence and removes the module unload/reload workaround.

Signed-off-by: Arnaud Patard <apatard@mandriva.com>
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agor8169: issue request_irq after the private data are completely initialized
Francois Romieu [Mon, 2 Apr 2007 20:59:59 +0000 (22:59 +0200)]
r8169: issue request_irq after the private data are completely initialized

The irq handler schedules a NAPI poll request unconditionally as soon as
the status register is not clean. It has been there - and wrong - for
ages but a recent timing change made it apparently easier to trigger.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: Jay Cliburn <jacliburn@bellsouth.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agob44: fix IFF_ALLMULTI handling of CAM slots
Bill Helfinstine [Sun, 1 Apr 2007 17:10:28 +0000 (13:10 -0400)]
b44: fix IFF_ALLMULTI handling of CAM slots

If you set the IFF_ALLMULTI flag on a b44 device, or if you join more than
B44_MCAST_TABLE_SIZE multicast groups, the device will stop receiving unicast
messages.  This is because the __b44_set_mac_addr call sets the zeroth CAM
entry to the MAC address of the device, and then the loop at line 1722
proceeds to overwrite it unless the value of i is set by the __b44_load_mcast
call.  However, when IFF_ALLMULTI is set, that call is bypassed, leaving i set
to zero.

Fixed by starting the loop at 1 to make it skip the CAM entry for the MAC
address.

Signed-off-by: Bill Helfinstine <bhelf@flitterfly.whirpon.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agocxgb3 - Firwmare update
Divy Le Ray [Sat, 31 Mar 2007 07:23:30 +0000 (00:23 -0700)]
cxgb3 - Firwmare update

Introduce FW micro version.
Bump up FW version to 3.3.0

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agocxgb3 - Tighten xgmac workaround
Divy Le Ray [Sat, 31 Mar 2007 07:23:24 +0000 (00:23 -0700)]
cxgb3 - Tighten xgmac workaround

Run the watchdog task when the link is up.
Flush the XGMAC Tx FIFO when the link drops.

Also remove a statistics update that should have gone
in the previous modification of xgmac.c.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agocxgb3 - detect NIC only adapters
Divy Le Ray [Sat, 31 Mar 2007 07:23:19 +0000 (00:23 -0700)]
cxgb3 - detect NIC only adapters

Differentiate NIC only adapters from RNICs.
Initialize offload capabilities for RNICs only.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agocxgb3 - Safeguard TCAM size usage
Divy Le Ray [Sat, 31 Mar 2007 07:23:13 +0000 (00:23 -0700)]
cxgb3 - Safeguard TCAM size usage

Ensure that the TCAM active region size is at least 16.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years ago[PATCH] msi: synchronously mask and unmask msi-x irqs.
Eric W. Biederman [Tue, 3 Apr 2007 07:41:49 +0000 (01:41 -0600)]
[PATCH] msi: synchronously mask and unmask msi-x irqs.

This is a simplified and actually more comprehensive form of a bug
fix from Mitch Williams <mitch.a.williams@intel.com>.

When we mask or unmask a msi-x irqs the writes may be posted because
we are writing to memory mapped region.  This means the mask and
unmask don't happen immediately but at some unspecified time in the
future.  Which is out of sync with how the mask/unmask logic work
for ioapic irqs.

The practical result is that we get very subtle and hard to track down
irq migration bugs.

This patch performs a read flush after writes to the MSI-X table for mask
and unmask operations.  Since the SMP affinity is set while the interrupt
is masked, and since it's unmasked immediately after, no additional flushes
are required in the various affinity setting routines.

The testing by Mitch Williams on his especially problematic system should
still be valid as I have only simplified the code, not changed the
functionality.

We currently have 7 drivers: cciss, mthca, cxgb3, forceth, s2io,
pcie/portdrv_core, and qla2xxx in 2.6.21 that are affected by this
problem when the hardware they driver is plugged into the right slot.

Given the difficulty of reproducing this bug and tracing it down to
anything that even remotely resembles a cause, even if people are
being affected we aren't likely to see many meaningful bug reports, and
the people who see this bug aren't likely to be able to reproduce this
bug in a timely fashion.  So it is best to get this problem fixed
as soon as we can so people don't have problems.

Then if people do have a kernel message stating "No irq for vector" we
will know it is yet another novel cause that needs a complete new
investigation.

Cc: Greg KH <greg@kroah.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Mitch Williams <mitch.a.williams@intel.com>
Acked-by: "Siddha, Suresh B" <suresh.b.siddha@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Mon, 2 Apr 2007 22:23:08 +0000 (15:23 -0700)]
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SCSI]: Fix scsi_send_eh_cmnd scatterlist handling
  [SPARC]: Add unsigned to unused bit field in a.out.h

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 2 Apr 2007 22:22:20 +0000 (15:22 -0700)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [TCP]: Do receiver-side SWS avoidance for rcvbuf < MSS.
  [BNX2]: Fix nvram write logic.
  [IPv6]: Fix incorrect length check in rawv6_sendmsg()
  [NET_SCHED]: cls_basic: fix memory leak in basic_destroy
  [NET]: Change "not found" return value for rule lookup

17 years ago[SCSI]: Fix scsi_send_eh_cmnd scatterlist handling
David S. Miller [Mon, 2 Apr 2007 21:21:55 +0000 (14:21 -0700)]
[SCSI]: Fix scsi_send_eh_cmnd scatterlist handling

This fixes a regression caused by commit:

2dc611de5a3fd955cd0298c50691d4c05046db97

The sense buffer code in scsi_send_eh_cmnd was changed to use
alloc_page() and a scatter list, but the sense data copy was not
updated to match so what we actually get in the sense buffer is total
grabage starting with the kernel address of the struct page we got.
Basically the stack frame of scsi_send_eh_cmd() is what ends up
in the sense buffer.

Depending upon how pointers look on a given platform, you can
end up getting sr_ioctl.c errors when you mount a cdrom.  If
the CDROM gives a check condition for GPCMD_GET_CONFIGURATION issued
by drivers/cdrom/cdrom.c:cdrom_mmc_profile(), sr_ioctl will
spit out this error message in sr_do_ioctl() with the way pointers
are on sparc64:

default:
printk(KERN_ERR "%s: CDROM (ioctl) error, command: ", cd->cdi.name);
__scsi_print_command(cgc->cmd);
scsi_print_sense_hdr("sr", &sshdr);
err = -EIO;

This is the error Tom Callaway reported in:

http://marc.info/?l=linux-sparc&m=117407453208101&w=2

Anyways, fix this by using page_address(sgl.page) which is OK
because we know this is low-mem due to GFP_ATOMIC.

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Christoph Hellwig <hch@lst.de>
17 years ago[SPARC]: Add unsigned to unused bit field in a.out.h
Robert Reif [Wed, 28 Mar 2007 21:21:08 +0000 (14:21 -0700)]
[SPARC]: Add unsigned to unused bit field in a.out.h

Add unsigned to unused bit field in a.out.h to make sparse happy.

[ I took care of the sparc64 side as well -DaveM ]

Signed-off-by: Robert Reif <reif@earthlink.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Do receiver-side SWS avoidance for rcvbuf < MSS.
John Heffner [Mon, 2 Apr 2007 20:56:32 +0000 (13:56 -0700)]
[TCP]: Do receiver-side SWS avoidance for rcvbuf < MSS.

Signed-off-by: John Heffner <jheffner@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoMerge branch 'master' of ssh://master.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Mon, 2 Apr 2007 20:52:10 +0000 (13:52 -0700)]
Merge branch 'master' of ssh:///linux/kernel/git/mchehab/v4l-dvb

* 'master' of ssh://master.kernel.org/pub/scm/linux/kernel/git/mchehab/v4l-dvb:
  V4L/DVB (5496): Pluto2: fix incorrect TSCR register setting
  V4L/DVB (5495): Tda10086: fix DiSEqC message length

17 years ago[BNX2]: Fix nvram write logic.
Michael Chan [Fri, 30 Mar 2007 21:53:06 +0000 (14:53 -0700)]
[BNX2]: Fix nvram write logic.

The nvram dword alignment logic was broken when writing less than 4
bytes on a non-aligned offset.  It was missing logic to round the
length to 4 bytes.

The page erase code is also moved so that it is only called when
using non-buffered flash for better code clarity.

Update version to 1.5.7.

Based on initial patch from Tony Cureington <tony.cureington@hp.com>.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPv6]: Fix incorrect length check in rawv6_sendmsg()
YOSHIFUJI Hideaki [Fri, 30 Mar 2007 21:45:35 +0000 (14:45 -0700)]
[IPv6]: Fix incorrect length check in rawv6_sendmsg()

In article <20070329.142644.70222545.davem@davemloft.net> (at Thu, 29 Mar 2007 14:26:44 -0700 (PDT)), David Miller <davem@davemloft.net> says:

> From: Sridhar Samudrala <sri@us.ibm.com>
> Date: Thu, 29 Mar 2007 14:17:28 -0700
>
> > The check for length in rawv6_sendmsg() is incorrect.
> > As len is an unsigned int, (len < 0) will never be TRUE.
> > I think checking for IPV6_MAXPLEN(65535) is better.
> >
> > Is it possible to send ipv6 jumbo packets using raw
> > sockets? If so, we can remove this check.
>
> I don't see why such a limitation against jumbo would exist,
> does anyone else?
>
> Thanks for catching this Sridhar.  A good compiler should simply
> fail to compile "if (x < 0)" when 'x' is an unsigned type, don't
> you think :-)

Dave, we use "int" for returning value,
so we should fix this anyway, IMHO;
we should not allow len > INT_MAX.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Acked-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET_SCHED]: cls_basic: fix memory leak in basic_destroy
Patrick McHardy [Fri, 30 Mar 2007 20:36:23 +0000 (13:36 -0700)]
[NET_SCHED]: cls_basic: fix memory leak in basic_destroy

tp->root is not freed on destruction.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: Change "not found" return value for rule lookup
Steven Whitehouse [Fri, 30 Mar 2007 20:34:27 +0000 (13:34 -0700)]
[NET]: Change "not found" return value for rule lookup

This changes the "not found" error return for the lookup
function to -ESRCH so that it can be distinguished from
the case where a rule or route resulting in -ENETUNREACH
has been found during the search.

It fixes a bug where if DECnet was compiled with routing
support, but no routes were added to the routing table,
it was failing to fall back to endnode routing.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoMerge branch 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6
Linus Torvalds [Mon, 2 Apr 2007 18:41:55 +0000 (11:41 -0700)]
Merge branch 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6

* 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6:
  [PATCH] x86: Don't probe for DDC on VBE1.2
  [PATCH] x86-64: Increase NMI watchdog probing timeout
  [PATCH] x86-64: Let oprofile reserve MSR on all CPUs
  [PATCH] x86-64: Disable local APIC timer use on AMD systems with C1E

17 years ago[PATCH] i386: fix file_read_actor() and pipe_read() for original i386 systems
Thomas Gleixner [Mon, 2 Apr 2007 12:25:31 +0000 (14:25 +0200)]
[PATCH] i386: fix file_read_actor() and pipe_read() for original i386 systems

The __copy_to_user_inatomic() calls in file_read_actor() and pipe_read()
are broken on original i386 machines, where WP-works-ok == false, as
__copy_to_user_inatomic() on such systems calls functions which might
sleep and/or contain cond_resched() calls inside of a kmap_atomic()
region.

The original check for WP-works-ok was in access_ok(), but got moved
during the 2.5 series to fix a race vs. swap.

Return the number of bytes to copy in the case where we are in an atomic
region, so the non atomic code pathes in file_read_actor() and
pipe_read() are taken.

This could be optimized to avoid the kmap_atomicby moving the check for
WP-works-ok into fault_in_pages_writeable(), but this is more intrusive
and can be done later.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] vt: fix potential race in VT_WAITACTIVE handler
Michal Januszewski [Mon, 2 Apr 2007 06:49:51 +0000 (23:49 -0700)]
[PATCH] vt: fix potential race in VT_WAITACTIVE handler

On a multiprocessor machine the VT_WAITACTIVE ioctl call may return 0 if
fg_console has already been updated in redraw_screen() but the console
switch itself hasn't been completed.  Fix this by checking fg_console in
vt_waitactive() with the console sem held.

Signed-off-by: Michal Januszewski <spock@gentoo.org>
Acked-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] Fix microcode-related suspend problem
Rafael J. Wysocki [Mon, 2 Apr 2007 06:49:49 +0000 (23:49 -0700)]
[PATCH] Fix microcode-related suspend problem

Fix the regression resulting from the recent change of suspend code
ordering that causes systems based on Intel x86 CPUs using the microcode
driver to hang during the resume.

The problem occurs since the microcode driver uses request_firmware() in
its CPU hotplug notifier, which is called after tasks has been frozen and
hangs.  It can be fixed by telling the microcode driver to use the
microcode stored in memory during the resume instead of trying to load it
from disk.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Adrian Bunk <bunk@stusta.de>
Cc: Tigran Aivazian <tigran@aivazian.fsnet.co.uk>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Maxim <maximlevitsky@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] driver core: fix built-in drivers sysfs links
Kay Sievers [Mon, 2 Apr 2007 06:49:48 +0000 (23:49 -0700)]
[PATCH] driver core: fix built-in drivers sysfs links

built-in drivers had broken sysfs links that caused bootup hangs for
certain driver unregistry sequences.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] rtc-cmos lockdep fix, irq updates
David Brownell [Mon, 2 Apr 2007 06:49:47 +0000 (23:49 -0700)]
[PATCH] rtc-cmos lockdep fix, irq updates

Lockdep reported cmos_suspend() and cmos_resume() calling rtc_update_irq()
with IRQs enabled; not allowed.

Also fix problems seen on some hardware, whereby false alarm IRQs could be
reported (primarily to userspace); and update two comments to match changes
in ACPI.  Those make up most of this patch, by volume.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] Correctly report PnP 64bit resources
Petr Vandrovec [Mon, 2 Apr 2007 06:49:46 +0000 (23:49 -0700)]
[PATCH] Correctly report PnP 64bit resources

Change PnP resource handling code to use proper type for resource start and
length.  Fixes bogus regions reported in /proc/iomem.

I've also made some pointer constant, as they are constant...

Signed-off-by: Petr Vandrovec <petr@vandrovec.name>
Cc: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: Adam Belay <ambx1@neo.rr.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] revert "retries in ext4_prepare_write() violate ordering requirements"
Andrew Morton [Mon, 2 Apr 2007 06:49:44 +0000 (23:49 -0700)]
[PATCH] revert "retries in ext4_prepare_write() violate ordering requirements"

Revert b46be05004abb419e303e66e143eed9f8a6e9f3f.  Same reasoning as for ext3.

Cc: Kirill Korotaev <dev@openvz.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Ken Chen <kenneth.w.chen@intel.com>
Cc: Andrey Savochkin <saw@sw.ru>
Cc: <linux-ext4@vger.kernel.org>
Cc: Dmitriy Monakhov <dmonakhov@openvz.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] revert "retries in ext3_prepare_write() violate ordering requirements"
Andrew Morton [Mon, 2 Apr 2007 06:49:43 +0000 (23:49 -0700)]
[PATCH] revert "retries in ext3_prepare_write() violate ordering requirements"

Revert e92a4d595b464c4aae64be39ca61a9ffe9c8b278.

Dmitry points out

"When we block_prepare_write() failed while ext3_prepare_write() we jump to
 "failure" label and call ext3_prepare_failure() witch search last mapped bh
 and invoke commit_write untill it.  This is wrong!!  because some bh from
 begining to the last mapped bh may be not uptodate.  As a result we commit to
 disk not uptodate page content witch contains garbage from previous usage."

and

"Unexpected file size increasing."

   Call trace the same as it was in first issue but result is different.
   For example we have file with i_size is zero.  we want write two blocks ,
   but fs has only one free block.

   ->ext3_prepare_write(...from == 0, to == 2048)
     retry:
     ->block_prepare_write() == -ENOSPC# we failed but allocated one block here.
     ->ext3_prepare_failure()
       ->commit_write( from == 0, to == 1024) # after this i_size becomes 1024 :)
     if (ret == -ENOSPC && ext3_should_retry_alloc(inode->i_sb, &retries))
        goto retry;

   Finally when all retries will be spended ext3_prepare_failure return
   -ENOSPC, but i_size was increased and later block trimm procedures can't
   help here.

We don't appear to have the horsepower to fix these issues, so let's put
things back the way they were for now.

Cc: Kirill Korotaev <dev@openvz.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Ken Chen <kenneth.w.chen@intel.com>
Cc: Andrey Savochkin <saw@sw.ru>
Cc: <linux-ext4@vger.kernel.org>
Cc: Dmitriy Monakhov <dmonakhov@openvz.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] fix page leak during core dump
Brian Pomerantz [Mon, 2 Apr 2007 06:49:41 +0000 (23:49 -0700)]
[PATCH] fix page leak during core dump

When the dump cannot occur most likely because of a full file system and
the page to be written is the zero page, the call to page_cache_release()
is missed.

Signed-off-by: Brian Pomerantz <bapper@mvista.com>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: David Howells <dhowells@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] CPUSETS: add mems to basic usage documentation
Simon Horman [Mon, 2 Apr 2007 06:49:40 +0000 (23:49 -0700)]
[PATCH] CPUSETS: add mems to basic usage documentation

It seems that there must be at least one node in mems and at least one CPU
in cpus in order to be able to assign tasks to a cpuset.  This makes sense.
 And I think it would also make sense to include a mems setting in the
basic usage section of the documentation.

I also wonder if something logged to dmsg, explaining why a write failed,
would be a good enhancement.  I ended up having rummage arround in cpuset.c
in order to work out why my configuration was failing.

Signed-off-by: Simon Horman <horms@verge.net.au>
Acked-by: Paul Jackson <pj@sgi.com>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] drivers/mfd/sm501.c: fix an off-by-one
Adrian Bunk [Mon, 2 Apr 2007 06:49:38 +0000 (23:49 -0700)]
[PATCH] drivers/mfd/sm501.c: fix an off-by-one

Fix an off-by-one spotted by the Coverity checker.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: Vincent Sanders <vince@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] uml: fix unreasonably long udelay
Paolo 'Blaisorblade' Giarrusso [Mon, 2 Apr 2007 06:49:37 +0000 (23:49 -0700)]
[PATCH] uml: fix unreasonably long udelay

Currently we have a confused udelay implementation.

* __const_udelay does not accept usecs but xloops in i386 and x86_64
* our implementation requires usecs as arg
* it gets a xloops count when called by asm/arch/delay.h

Bugs related to this (extremely long shutdown times) where reported by some
x86_64 users, especially using Device Mapper.

To hit this bug, a compile-time constant time parameter must be passed -
that's why UML seems to work most times.  Fix this with a simple udelay
implementation.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Acked-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] proc: fix linkage with CONFIG_SYSCTL=y, CONFIG_PROC_SYSCTL=n
Andrew Morton [Mon, 2 Apr 2007 06:49:35 +0000 (23:49 -0700)]
[PATCH] proc: fix linkage with CONFIG_SYSCTL=y, CONFIG_PROC_SYSCTL=n

We're using #ifdef CONFIG_SYSCTL, but we should be using CONFIG_PROC_SYSCTL,
so we get

 fs/built-in.o: In function `proc_root_init':
 /usr/src/linux/fs/proc/root.c:83: undefined reference to `proc_sys_init'

Fix that up and remove an ifdef-in-C.

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Helge Hafting <helgehaf@aitel.hist.no>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoV4L/DVB (5496): Pluto2: fix incorrect TSCR register setting
Andreas Oberritter [Mon, 2 Apr 2007 13:44:35 +0000 (10:44 -0300)]
V4L/DVB (5496): Pluto2: fix incorrect TSCR register setting

The ADEF bits in the TSCR register have different meanings in read and
write mode. For this reason ADEF has to be reset on every
read-modify-write operation.
This patch introduces a special write function for this register, which
takes care of it.

Thanks to Holger Magnussen for pointing my nose at this problem.

Signed-off-by: Andreas Oberritter <obi@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
17 years agoV4L/DVB (5495): Tda10086: fix DiSEqC message length
Andreas Oberritter [Sun, 1 Apr 2007 22:29:16 +0000 (19:29 -0300)]
V4L/DVB (5495): Tda10086: fix DiSEqC message length

Setting the message length to zero means to send one byte, so you need a
subtraction instead of an addition.

Signed-off-by: Andreas Oberritter <obi@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
17 years ago[PATCH] x86: Don't probe for DDC on VBE1.2
Zwane Mwaikambo [Mon, 2 Apr 2007 10:14:12 +0000 (12:14 +0200)]
[PATCH] x86: Don't probe for DDC on VBE1.2

VBE1.2 doesn't support function 15h (DDC) resulting in a 'hang' whilst
uncompressing kernel with some video cards. Make sure we check VBE version
before fiddling around with DDC.

http://bugzilla.kernel.org/show_bug.cgi?id=1458

Opened: 2003-10-30 09:12 Last update: 2007-02-13 22:03

Much thanks to Tobias Hain for help in testing and investigating the bug.
Tested on;

i386, Chips & Technologies 65548 VESA VBE 1.2
CONFIG_VIDEO_SELECT=Y
CONFIG_FIRMWARE_EDID=Y

Untested on x86_64.

Signed-off-by: Zwane Mwaikambo <zwane@infradead.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Increase NMI watchdog probing timeout
Andi Kleen [Mon, 2 Apr 2007 10:14:12 +0000 (12:14 +0200)]
[PATCH] x86-64: Increase NMI watchdog probing timeout

A 4 core Opteron needs longer than 10 ticks for this.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Let oprofile reserve MSR on all CPUs
Andi Kleen [Mon, 2 Apr 2007 10:14:12 +0000 (12:14 +0200)]
[PATCH] x86-64: Let oprofile reserve MSR on all CPUs

The MSR reservation is per CPU and oprofile would only allocate them
on the CPU it was initialized on. Change this to handle all CPUs.

This also fixes a warning about unprotected use of smp_processor_id()
in preemptible kernels.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Disable local APIC timer use on AMD systems with C1E
Andi Kleen [Mon, 2 Apr 2007 10:14:12 +0000 (12:14 +0200)]
[PATCH] x86-64: Disable local APIC timer use on AMD systems with C1E

AMD dual core laptops with C1E do not run the APIC timer correctly
when they go idle. Previously the code assumed this only happened
on C2 or deeper.  But not all of these systems report support C2.

Use a AMD supplied snippet to detect C1E being enabled and then disable
local apic timer use.

This supercedes an earlier workaround using DMI detection of specific systems.

Thanks to Mark Langsdorf for the detection snippet.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Sun, 1 Apr 2007 21:43:57 +0000 (14:43 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 4298/1: fix memory barriers for DMA coherent and SMP platforms
  [ARM] 4295/2: Fix error-handling in pxaficp_ir.c (version 2)
  [ARM] Fix __NR_kexec_load
  [ARM] Export dma_channel_active()
  [ARM] 4296/1: ixp4xx: compile fix
  [ARM] 4289/1: AT91: SAM9260 NAND flash timing

17 years ago[ARM] 4298/1: fix memory barriers for DMA coherent and SMP platforms
Lennert Buytenhek [Sat, 31 Mar 2007 11:03:20 +0000 (12:03 +0100)]
[ARM] 4298/1: fix memory barriers for DMA coherent and SMP platforms

This patch:
- Switches mb/rmb/wmb back to being full-blown DMBs on ARM SMP systems,
  since mb/rmb/wmb are required to order Normal memory accesses as well.
- Enables the use of DMB and ISB on XSC3 (which is an ARMv5TE ISA core
  but conforms to the ARMv6 memory ordering model and supports the
  various ARMv6 barriers.)
- Makes DMA coherent platforms (only ixp23xx at the moment) map
  mb/rmb/wmb to dmb(), as on DMA coherent platforms, DMA consistent
  mappings are done as Normal mappings, which are weakly ordered.

Signed-off-by: Lennert Buytenhek <buytenh@wantstofly.org>
Acked-by: David Howells <dhowells@redhat.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
17 years ago[ARM] 4295/2: Fix error-handling in pxaficp_ir.c (version 2)
Guennadi Liakhovetski [Fri, 30 Mar 2007 07:49:55 +0000 (08:49 +0100)]
[ARM] 4295/2: Fix error-handling in pxaficp_ir.c (version 2)

This patch addresses the following issues with the pxa2xx FIr driver:

1. increment overrun error counter and not frame error counter on ICSR1_ROR bit set in ICSR1.
2. drop frames reported with the frame error from the IC.
3. when resetting the receiver and preparing it for the next DMA in pxa_irda_fir_irq() actually clear the Rx FIFO. See description in Table 11-2 in PXA270 Developer's Manual of the RXE bit.

Correction added in version 2: clearing the IC Rx FIFO also has to be done in pxa_irda_fir_dma_tx_irq()

Signed-off-by: G. Liakhovetski <gl@dsa-ac.de>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
17 years ago[ARM] Fix __NR_kexec_load
Russell King [Sun, 1 Apr 2007 21:35:01 +0000 (22:35 +0100)]
[ARM] Fix __NR_kexec_load

It's __NR_kexec_load, not __NR_sys_kexec_load

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
17 years ago[PATCH] kbuild: fix dependency generation
Jan Beulich [Thu, 29 Mar 2007 09:27:14 +0000 (10:27 +0100)]
[PATCH] kbuild: fix dependency generation

Commit 2e3646e51b2d6415549b310655df63e7e0d7a080 changed the way the
split config tree is built, but failed to also adjust fixdep accordingly
- if changing a config option from or to m, files referencing the
respective CONFIG_..._MODULE (but not the corresponding CONFIG_...)
didn't get rebuilt.

The problem is that trisate symbol are represent with three different
symbols:
    SYMBOL=n => no symbol defined
    SYMBOL=y => CONFIG_SYMBOL defined to '1'
    SYMBOL=m => CONFIG_SYMBOL_MODULE defined to '1'

But conf_split_config do not distingush between the =y and =m case, so
only the =y case is honoured.

This is fixed in fixdep so when a CONFIG symbol with _MODULE is found we
skip that part and only look for the CONFIG_SYMBOL version.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agodriver core: do not wait unnecessarily in driver_unregister()
Linus Torvalds [Sun, 1 Apr 2007 17:54:13 +0000 (10:54 -0700)]
driver core: do not wait unnecessarily in driver_unregister()

Ingo reported that built-in drivers suffered bootup hangs with certain
driver unregistry sequences, due to sysfs breakage.

Do the minimal fix for v2.6.21: only wait if the driver is a module.

Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[ARM] Export dma_channel_active()
Russell King [Sat, 31 Mar 2007 20:36:53 +0000 (21:36 +0100)]
[ARM] Export dma_channel_active()

dma_channel_active() is used by some modules and is part of our
DMA API, so export it.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Sat, 31 Mar 2007 03:40:56 +0000 (20:40 -0700)]
Merge /pub/scm/linux/kernel/git/herbert/crypto-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  [CRYPTO] api: Flush the current page right than the next
  [CRYPTO] api: Use the right value when advancing scatterwalk_copychunks

17 years ago[CRYPTO] api: Flush the current page right than the next
Herbert Xu [Sat, 31 Mar 2007 02:58:20 +0000 (12:58 +1000)]
[CRYPTO] api: Flush the current page right than the next

On platforms where flush_dcache_page is needed we're currently flushing
the next page right than the one we've just processed.  This patch fixes
the off-by-one error.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
17 years ago[PATCH] crypto api: Use the right value when advancing scatterwalk_copychunks
Herbert Xu [Sat, 31 Mar 2007 02:14:37 +0000 (12:14 +1000)]
[PATCH] crypto api: Use the right value when advancing scatterwalk_copychunks

In the scatterwalk_copychunks loop, We should be advancing by
len_this_page and not nbytes.  The latter is the total length.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[PATCH] uml: fix static linking for real
Paolo 'Blaisorblade' Giarrusso [Sat, 31 Mar 2007 01:20:27 +0000 (03:20 +0200)]
[PATCH] uml: fix static linking for real

There was a typo in commit 7632fc8f809a97f9d82ce125e8e3e579390ce2e5,
preventing it from working - 32bit binaries crashed hopelessly before
the below fix and work perfectly now.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[CRYPTO] api: Use the right value when advancing scatterwalk_copychunks
Herbert Xu [Sat, 31 Mar 2007 02:16:20 +0000 (12:16 +1000)]
[CRYPTO] api: Use the right value when advancing scatterwalk_copychunks

In the scatterwalk_copychunks loop, We should be advancing by
len_this_page and not nbytes.  The latter is the total length.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
17 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Fri, 30 Mar 2007 18:49:37 +0000 (11:49 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  Revert "ACPI: parse 2nd MADT by default"

17 years agoRevert "ACPI: parse 2nd MADT by default"
Len Brown [Fri, 30 Mar 2007 18:16:10 +0000 (14:16 -0400)]
Revert "ACPI: parse 2nd MADT by default"

This reverts commit 09fe58356d148ff66901ddf639e725ca1a48a0af.
http://bugzilla.kernel.org/show_bug.cgi?id=8283

Signed-off-by: Len Brown <len.brown@intel.com>
17 years ago[PATCH] Maxtor 6B250S0/BANC1B70 hangs with NCQ
Jens Axboe [Fri, 30 Mar 2007 07:27:58 +0000 (09:27 +0200)]
[PATCH] Maxtor 6B250S0/BANC1B70 hangs with NCQ

I've seen this several times on this drive, completely reproducible.

Once it has hung, power needs to be cut from the drive to recover it, a
simple reboot is not enough.  So I'd suggest disabling NCQ on this
drive.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Acked-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[ARM] 4296/1: ixp4xx: compile fix
Vladimir Barinov [Thu, 29 Mar 2007 15:29:20 +0000 (16:29 +0100)]
[ARM] 4296/1: ixp4xx: compile fix

Fix compilation fail for ixp4xx platforms for the case when CONFIG_IXP4XX_INDIRECT_PCI is set. That is due to the check_signature() is appeared in include/linux/io.h.

Signed-off-by: Vladimir Barinov <vbarinov@ru.mvista.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
17 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Fri, 30 Mar 2007 03:30:12 +0000 (20:30 -0700)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  [MIPS] SMTC: Fix recursion in instant IPI replay code.
  [MIPS] BCM1480: Fix setting of irq affinity.
  [MIPS] do_page_fault() needs to use raw_smp_processor_id().
  [MIPS] SMTC: Fix false trigger of debug code on single VPE.
  [MIPS] SMTC: irq_{enter,leave} and kstats keeping for relayed timer ints.
  [MIPS] lockdep: Deal with interrupt disable hazard in TRACE_IRQFLAGS
  [MIPS] lockdep: Handle interrupts in R3000 style c0_status register.
  [MIPS] MV64340: Add missing prototype for mv64340_irq_init().
  [MIPS] MT: MIPS_MT_SMTC_INSTANT_REPLAY currently conflicts with PREEMPT.
  [MIPS] EV64120: Include <asm/irq.h> to fix warning.
  [MIPS] Ocelot: Fix warning.
  [MIPS] Ocelot: Give PMON_v1_setup a proper prototype.

17 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Fri, 30 Mar 2007 03:25:03 +0000 (20:25 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] Fix arch/ia64/pci/pci.c:571: warning: `return' with a value
  [IA64] Speed up boot - skip unnecessary clock calibration
  [IA64] bugfix stack layout upside-down
  [IA64] Fix possible invalid memory access in ia64_setup_msi_irq()

17 years ago[MIPS] SMTC: Fix recursion in instant IPI replay code.
Ralf Baechle [Thu, 29 Mar 2007 22:42:42 +0000 (23:42 +0100)]
[MIPS] SMTC: Fix recursion in instant IPI replay code.

local_irq_restore -> raw_local_irq_restore -> irq_restore_epilog ->
smtc_ipi_replay -> smtc_ipi_dq -> spin_unlock_irqrestore ->
_spin_unlock_irqrestore -> local_irq_restore

The recursion does abort when there is no more IPI queued for a CPU, so
this isn't usually fatal which is why we got away with this for so long
until this was discovered by code inspection.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] BCM1480: Fix setting of irq affinity.
Mark Mason [Mon, 26 Mar 2007 20:28:26 +0000 (13:28 -0700)]
[MIPS] BCM1480: Fix setting of irq affinity.

Signed-off-by: Mark Mason <mason@broadcom.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] do_page_fault() needs to use raw_smp_processor_id().
Ralf Baechle [Thu, 29 Mar 2007 21:30:01 +0000 (22:30 +0100)]
[MIPS] do_page_fault() needs to use raw_smp_processor_id().

Original patch posted by Deepak Saxena <dsaxena@plexity.net>.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] SMTC: Fix false trigger of debug code on single VPE.
Ralf Baechle [Tue, 27 Mar 2007 14:19:58 +0000 (15:19 +0100)]
[MIPS] SMTC: Fix false trigger of debug code on single VPE.

Make smtc_setup_irq() update the list of interrupts which need to be
watched by the debug code itself.  Also there is no need to initialize the
IPI swint when running with a single VPE, so don't initialize it.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] SMTC: irq_{enter,leave} and kstats keeping for relayed timer ints.
Ralf Baechle [Tue, 27 Mar 2007 14:11:54 +0000 (15:11 +0100)]
[MIPS] SMTC: irq_{enter,leave} and kstats keeping for relayed timer ints.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] lockdep: Deal with interrupt disable hazard in TRACE_IRQFLAGS
Chris Dearman [Mon, 26 Mar 2007 13:48:50 +0000 (14:48 +0100)]
[MIPS] lockdep: Deal with interrupt disable hazard in TRACE_IRQFLAGS

Between the mtc0 or di instruction that disables interrupts and the
following hazard barrier a processor may still take interrupts.  If an
interrupt is taken after interrupts are disabled but before the state
is updated it will appear to restore_all that it is incorrectly returning
with interrupts disabled.

Signed-off-by: Chris Dearman <chris@mips.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] lockdep: Handle interrupts in R3000 style c0_status register.
Chris Dearman [Mon, 26 Mar 2007 13:47:06 +0000 (14:47 +0100)]
[MIPS] lockdep: Handle interrupts in R3000 style c0_status register.

Check the IEP bit for R3000 style processors when checking to see if
interrupts will be reenabled in restore_all.

Signed-off-by: Chris Dearman <chris@mips.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] MV64340: Add missing prototype for mv64340_irq_init().
Ralf Baechle [Mon, 26 Mar 2007 08:30:32 +0000 (09:30 +0100)]
[MIPS] MV64340: Add missing prototype for mv64340_irq_init().

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] MT: MIPS_MT_SMTC_INSTANT_REPLAY currently conflicts with PREEMPT.
Ralf Baechle [Mon, 26 Mar 2007 14:13:57 +0000 (15:13 +0100)]
[MIPS] MT: MIPS_MT_SMTC_INSTANT_REPLAY currently conflicts with PREEMPT.

So until MIPS_MT_SMTC_INSTANT_REPLAY has been rewritten to solve this
issue, don't allow selecting it with PREEMPT.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] EV64120: Include <asm/irq.h> to fix warning.
Ralf Baechle [Mon, 26 Mar 2007 08:22:00 +0000 (09:22 +0100)]
[MIPS] EV64120: Include <asm/irq.h> to fix warning.

arch/mips/pci/pci-ev64120.c:10: warning: implicit declaration of function 'allocate_irqno'

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] Ocelot: Fix warning.
Ralf Baechle [Mon, 26 Mar 2007 08:14:24 +0000 (09:14 +0100)]
[MIPS] Ocelot: Fix warning.

Remove unused variable.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[MIPS] Ocelot: Give PMON_v1_setup a proper prototype.
Ralf Baechle [Mon, 26 Mar 2007 07:59:52 +0000 (08:59 +0100)]
[MIPS] Ocelot: Give PMON_v1_setup a proper prototype.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
17 years ago[IA64] Fix arch/ia64/pci/pci.c:571: warning: `return' with a value
Tony Luck [Thu, 29 Mar 2007 22:41:37 +0000 (15:41 -0700)]
[IA64] Fix arch/ia64/pci/pci.c:571: warning: `return' with a value

Typo/thinko in bba6f6fc68e74d4572028646f61dd3505a68747e

Signed-off-by: Tony Luck <tony.luck@intel.com>
17 years ago[IA64] Speed up boot - skip unnecessary clock calibration
Jack Steiner [Tue, 27 Mar 2007 19:30:19 +0000 (14:30 -0500)]
[IA64] Speed up boot - skip unnecessary clock calibration

Skip clock calibration if cpu being brought online is exactly the same
speed, stepping, etc., as the previous cpu. This significantly reduces
the time to boot very large systems.

Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
17 years ago[IA64] bugfix stack layout upside-down
KAMEZAWA Hiroyuki [Fri, 23 Mar 2007 03:17:46 +0000 (12:17 +0900)]
[IA64] bugfix stack layout upside-down

ia64 expects following vm layout:

== low memory
[register-stack grows up]
[memory-stack grows down]
== high memory

But the code assigns the base of the register stack at the
maximum stack size offset from the fixed address where the
stack *might* start.  Stack randomization will result in the
memory stack starting at a lower address than this, and if the
user has set a low stack limit with "ulimit -s", then you can
end up with the register stack above the memory stack (or if
you were very unlucky right on top of it!).

Fix: Calculate the base address for the register stack starting
from the actual address of the memory stack.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
17 years ago[IA64] Fix possible invalid memory access in ia64_setup_msi_irq()
Kenji Kaneshige [Mon, 26 Mar 2007 00:38:42 +0000 (09:38 +0900)]
[IA64] Fix possible invalid memory access in ia64_setup_msi_irq()

The following 'if' statement in ia64_setup_msi_irq() always fails even
if create_irq() returns <0 value, because variable 'irq' is defined as
unsigned int. It would cause invalid memory access.

        irq = create_irq();
        if (irq < 0)
                return irq;

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Thu, 29 Mar 2007 20:22:07 +0000 (13:22 -0700)]
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6:
  NetXen: Fix hardware access for ppc architecture.
  sis190: new PHY support
  atl1: save mac address on remove

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Thu, 29 Mar 2007 20:15:13 +0000 (13:15 -0700)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [IFB]: Fix crash on input device removal
  [BNX2]: Fix link interrupt problem.

17 years ago[POWERPC] qe: ucc_slow.guemr is in the wrong place
Timur Tabi [Mon, 26 Mar 2007 19:25:42 +0000 (14:25 -0500)]
[POWERPC] qe: ucc_slow.guemr is in the wrong place

The definition of struct ucc_slow puts the guemr register immediately after the
utpt register, when it should be at offset 0x90.  This patch adds the missing
0x52-byte padding.

Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
17 years ago[POWERPC] qe: Fix QUICC Engine SDMA setup errors
Chuck Meade [Tue, 27 Mar 2007 14:46:10 +0000 (10:46 -0400)]
[POWERPC] qe: Fix QUICC Engine SDMA setup errors

Correct the alignment of the internal buffer used by the QUICC Engine
SDMA controller to 4Kbytes.  Correct the shift direction in the logic
that sets up the SDMR register for the QUICC Engine SDMA controller.

Signed-off-by: Chuck Meade <chuckmeade@mindspring.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>