This is the fix for the following problem:

https://bugzilla.redhat.com/show_bug.cgi?id=227657

The bnx2 device 5706 complains about MSI not working behind a
ServerWorks HT1000 PCIX bridge. An earlier commit to fix the problem:

e3008ded:

"PCI: disable MSI by default on systems with Serverworks HT1000 chips"

was not entirely correct, and has been reverted.

MSI does not work on the PCIX bus because the BIOS did not set the
HT_MSI_FLAGS_ENABLE bit in the HyperTransport MSI capability on the
bridge.  We use the existing quirk_msi_ht_cap() to detect the problem
and disable MSI in all buses behind it.
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Cc: Anantha Subramanyam <ananth@broadcom.com>
Cc: Naren Sankar <nsankar@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NJeff Garzik <jgarzik@redhat.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This reverts commit e3008ded.

The real bug was an INTX issue in the tg3 ethernet chip, and
cured by commit c129d962a66c76964954a98b38586ada82cf9381
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NJeff Garzik <jgarzik@redhat.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
  [S390] tod clock: announce clocksource as perfect
  [S390] Rename "idle_time" attribute to "idle_time_us".
  [S390] Fix priority mistakes in drivers/s390/cio/cmf.c
  [S390] Fix memory detection.
  [S390] Fix compile on !CONFIG_SMP.
  [S390] device_schedule_callback() for dcssblk.
  [S390] Fix smsgiucv init on no iucv machines
  [S390] cio: use INIT_WORK to initialize struct work.

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-lguest:
  lguest: tidy up documentation
  kernel/futex.c: make 3 functions static
  unexport access_process_vm
  lguest: make async_hcall() static

Commit f2a0bd37 defines the function
with "void cpm_load_patch(cpm8xx_t *cp)" prtotype and is declared as
"extern void cpm_load_patch(volatile immap_t *immr)" in the header file.
Signed-off-by: NKamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Fix the memory leak that may occur when we attempt to reuse a cpu_slab
that was allocated while we reenabled interrupts in order to be able to
grow a slab cache.

The per cpu freelist may contain objects and in that situation we may
overwrite the per cpu freelist pointer loosing objects.  This only
occurs if we find that the concurrently allocated slab fits our
allocation needs.

If we simply always deactivate the slab then the freelist will be
properly reintegrated and the memory leak will go away.
Signed-off-by: NChristoph Lameter <clameter@sgi.com>
Acked-by: NHugh Dickins <hugh@veritas.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

* master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] ibmvscsi: Prevent IO during partner login
  [SCSI] lpfc : Correct queue tag handling
  [SCSI] Update MAINTAINER email address and trees
  [SCSI] osst: fix if (...) \n #if... cases missing semicolons when false

After Adrian Bunk's "make async_hcall static" moved things around, update
comments to match (aka "make Guest").
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

The following functions can now become static again:
- get_futex_key()
- get_futex_key_refs()
- drop_futex_key_refs()
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

This patch removes the no longer used EXPORT_SYMBOL_GPL(access_process_vm).
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

async_hcall() can become static.
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

The Time of Day clock is the standard time source for s390. It is
- monotonic
- allows very fast reading
- architecture guarantees at least microsecond stepping
- available as part of the architecture

We should announce the rate of tod as 400 to be in sync with the
description found in clocksource.h:
"400-499:Perfect The ideal clocksource. A must-use where available."

This change will prefer tod over less reliable clock sources.
Signed-off-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Seems that people prefer to have the unit encoded in the attribute
name. Also makes parsing easier.

Now we have:

# cat /sys/devices/system/cpu/cpu0/idle_time_us
131473592

instead of

# cat /sys/devices/system/cpu/cpu0/idle_time
131473592 us

Cc: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Fixes priority mistakes similar to '!x & y'
Signed-off-by: NRoel Kluin <12o3l@tiscali.nl>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Yet another patch in the countless series of memory detection fixes:
if the last area of the reported storage size is a hole the detection
loop will loop forever.
Just break chunk detection loop if its end is going to be larger than
reported storage size.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Commit fae8b22d
"[S390] Add per-cpu idle time / idle count sysfs attributes" causes
a link error on !CONFIG_SMP.
Fix this by adding some #ifdef's. Real fix would be to cleanup the
code since we don't register a cpu on !CONFIG_SMP. But that would
be quite a big patch. For the time being this is good enough.

arch/s390/kernel/built-in.o: In function `do_monitor_call':
(.text+0x50d4): undefined reference to `per_cpu__s390_idle'
arch/s390/kernel/built-in.o: In function `cpu_idle':
(.text+0x518c): undefined reference to `per_cpu__s390_idle'
make: *** [.tmp_vmlinux1] Error 1
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Unregistering a device from within a device attribute handler leads to
a deadlock. Need to use device_schedule_callback() to unregister device
in error path.
Signed-off-by: NGerald Schaefer <geraldsc@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

smsgiucv is a driver that relies on iucv to work properly. If
iucv ans smsgiucv are compiled into the kernel and run on an
lpar the following scenario happens:
iucv is initialized early as a subsystem. It checks for z/VM and
returns with EPROTONOTSUPPORT. Later smsgiucv tries to run
driver_register with iucv_bus as bus. As this bus is not
initialized the driver core and list debugging issue several
warnings and oopses.

Solution is to let smsgiucv also check for z/VM and return
EPROTONOTSUPPORT as well.
Signed-off-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Use INIT_WORK to initialize struct work and don't initialize a
struct work partial by explicitly initializing its private structures.
Fixes the following lockdep bug because no key was assigned:

INFO: trying to register non-static key.
the code is fine but needs lockdep annotation.
turning off the locking correctness validator.
0000000001f07bb8 0000000001f07bf8 0000000000000002 0000000000000000
       0000000001f07c98 0000000001f07c10 0000000001f07c10 0000000000015406
       0000000000000000 0000000000000002 0000000000000000 0000000000000000
       0000000001f07bf8 000000000000000c 0000000001f07bf8 0000000001f07c68
       000000000039ae60 0000000000015406 0000000001f07bf8 0000000001f07c48
Call Trace:
([<0000000000015376>] show_trace+0xda/0x104)
 [<0000000000015460>] show_stack+0xc0/0xf8
 [<00000000000154c6>] dump_stack+0x2e/0x3c
 [<000000000006a71e>] __lock_acquire+0x47e/0x11a0
 [<000000000006b4f0>] lock_acquire+0xb0/0xd8
 [<00000000000555a6>] run_workqueue+0x1aa/0x24c
 [<00000000000556de>] worker_thread+0x96/0xf4
 [<000000000005c210>] kthread+0x90/0xb4
 [<000000000001947a>] kernel_thread_starter+0x6/0xc
 [<0000000000019474>] kernel_thread_starter+0x0/0xc
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

In accordance with the newly formalized 32-bit boot protocol, set
%ebx == %ebp == %edi == 0 in order to support future extensions to the
protocol.
Signed-off-by: NH. Peter Anvin <hpa@zytor.com>

The 386 and 486 needs a jump immediately after setting %cr0 in order
to serialize the pipeline.
Signed-off-by: NH. Peter Anvin <hpa@zytor.com>

* 'drm-patches' of master.kernel.org:/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/sis: missing mutex unlock in error path.
  radeon: set the address to access the GART table on the CPU side correctly

* 'an' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  [libata] Utilize new SCSI event infrastructure
  SCSI: add asynchronous event notification API

airlied: separated this out from a patch on lkml.
Signed-off-by: NRoel Kluin <12o3l@tiscali.nl>
Signed-off-by: NDave Airlie <airlied@linux.ie>

This code relied on the CPU and GPU address for the aperture being the same,
On some r5xx hardware I was playing with I noticed that this isn't always true.
This fixes issues seen on some r400 cards. (bugs.freedesktop.org 9957)
Signed-off-by: NDave Airlie <airlied@redhat.com>

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6:
  ieee1394: iso and async streams: s/g list fix

I have a system here that actively relies upon RTC wake alarms, and it
has been failing (again) for a few days when attempting to use the
/sys/class/rtc/rtc?/wakealarm interface.

The old (fixed by Linus) /proc/ interface still works, but I'd like to
get it using the new one.

This patch fixes rtc-cmos to ignore the two upper bits when reading the
BCD mday (day of month) register from CMOS.  Some systems (eg.  mine)
seem to have the top bit set to "1" for some reason.

The older /proc/ interface ignores the upper bits, and so we should too.
Signed-off-by: NMark Lord <mlord@pobox.com>
Acked-by: NDavid Brownell <david-b@pacbell.net>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Torsten Kaiser wrote:
> Looking that calltrace upwards, it seems replacing the
> memset(dma->sglist,...) with sg_init_table(...) would fix the BUG_ON()
> as that inits the SG_MAGIC.
Tested-by: NTorsten Kaiser <just.for.lkml@googlemail.com>
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

An end to CD-ROM polling (if you have a device that supports AN)...
hooray!
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

Originally based on a patch by Kristen Carlson Accardi @ Intel.
Copious input from James Bottomley.
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  [BLOCK] Don't allow empty barriers to be passed down to queues that don't grok them
  dm: bounce_pfn limit added
  Deadline iosched: Fix batching fairness
  Deadline iosched: Reset batch for ordered requests
  Deadline iosched: Factor out finding latter reques

* 'sg' of git://git.kernel.dk/linux-2.6-block:
  [SG] Get rid of __sg_mark_end()
  cleanup asm/scatterlist.h includes
  SG: Make sg_init_one() use general table init functions

Merge branch 'v2.6.24-rc1-lockdep' of git://git.kernel.org/pub/scm/linux/kernel/git/peterz/linux-2.6-lockdep

* 'v2.6.24-rc1-lockdep' of git://git.kernel.org/pub/scm/linux/kernel/git/peterz/linux-2.6-lockdep:
  lockdep: fix a typo in the __lock_acquire comment
  sched: fix unconditional irq lock
  lockdep: fixup irq tracing

* master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh64-2.6:
  sh64: Update defconfigs.
  sh64: fix dma_cache_sync() compilation
  sh64: Move DMA macros from pci.h to scatterlist.h.

* master.kernel.org:/pub/scm/linux/kernel/git/lethal/sh-2.6: (24 commits)
  sh: Update r7785rp defconfig.
  sh: mach-type updates.
  sh: Fix up r7780rp highlander CF access size.
  sh: Terminate .eh_frame in VDSO with a 4-byte 0.
  sh: Correct SUBARCH matching.
  sh: Decouple 4k and soft/hardirq stacks.
  sh: Fix optimized __copy_user() movca.l usage.
  sh: Clean up SR.RB Kconfig mess.
  sh: Kill off dead ipr_irq_demux().
  sh: Make SH7750 oprofile compile again.
  sh: Provide a __read_mostly section wrapper.
  sh: linker script tidying.
  sh: Move zero page param defs somewhere sensible.
  sh: Use generic SMP_CACHE_BYTES/L1_CACHE_ALIGN.
  sh: Kill off legacy embedded ramdisk section.
  sh: Fix up early mem cmdline parsing.
  sh: Enable USBF on MS7722SE.
  sh: Add resource of USBF for SH7722.
  maple: Fix maple bus compiler warning
  sh: fix zImage build with >=binutils-2.18
  ...

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev: (23 commits)
  libata: don't configure downstream links faster than the upstream link
  libata: request PHY speed configuration on SControl access failure
  libata: consider errors not associated with commands for speed down
  libata: more robust reset failure handling
  libata: cosmetic clean up / reorganization of ata_eh_reset()
  libata: fix timing computation in ata_eh_reset()
  libata: increase 128 KB / cmd limit for ATAPI tape drives
  sata_promise: fix endianess bug in ASIC PRD bug workaround
  libata: fix docbook
  make ata_scsi_lpm_get() static
  libata: suppress two warnings
  ata/sata_fsl: Remove ata_scsi_suspend/resume callbacks
  ata/sata_fsl: Remove sending LOG EXT command in sata_fsl_softreset()
  ata/sata_fsl: Move MPC8315DS link speed limit workaround to specific ifdef
  ata/sata_fsl: cleanup style problem
  ata/sata_fsl: remove unneeded sata_fsl_hardreset()
  ata/sata_fsl: remove unneeded on-stack copy of FIS
  ata/sata_fsl: cleanup needless casts to/from void __iomem *
  ata/sata_fsl: Remove unnecessary SCR cases
  ata/sata_fsl: Kill ata_sg_is_last()
  ...

The regression was caused by:
        commit[a32ea1e1] Fix read/truncate race

This causes ntfs_readpage() to be called for a zero i_size inode, which
failed when the file was compressed and non-resident.

Thanks a lot to Mike Galbraith for reporting the issue and tracking down
the commit that caused the regression.

Looking into it I found three bugs which the patch fixes.
Signed-off-by: NAnton Altaparmakov <aia21@cantab.net>
Tested-by: NMike Galbraith <efault@gmx.de>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

By setting the request_limit in send_srp_login to 1 we allowed login
requests to be sent to the server adapter.  If this was not an initial
login, but was a login after a disconnect with the server, other I/O
requests could attempt to be processed before the login occured.  These
I/O requests would fail, sometimes resulting in filesystems getting
marked read-only.

To address this we can set the request_limit to 0 while doing the login
and add an exception where login requests, along with task management
events, are always passed to the server.

There is a case where the request_limit had already reached 0 would result
in all events being sent rather than returning SCSI_MLQUEUE_HOST_BUSY; this
has also been fixed by this patch.
Signed-off-by: NRobert Jennings <rcj@linux.vnet.ibm.com>
Signed-off-by: NBrian King <brking@linux.vnet.ibm.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

This patch corrects the lpfc tag handling issue identified by Hannes Reinecke
http://marc.info/?l=linux-scsi@m=119270235628850&w=2

The basis for this patch originated from Hajime Kai. Thank You Hajime.

Signed-off-by: hajime-kai@soft.fujitsu.com
Signed-off-by: NJames Smart <James.Smart@emulex.com>
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>

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