x86_64: vsyscall time() fix

The vsyscall time() function basically returns the second portion of
xtime directly.  This however means that there is about a ticks worth of
time each second where time() will return a second value less then what
gettimeofday() does.

Additionally, this window where vtime() is behind vgettimeofday() grows
when dynticks is enabled, so its probably good to get this in before
dynticks lands.

Big thanks to Sripathi for noticing this issue and creating a test case
to work with!

This patch changes the vtime() implemenation to call vgettimeofday(),
much as syscall time() implementation calls gettimeofday().

2.6.21 stable candidate too

Signed-off-by: John Stultz <johnstul@us.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

x86_64: early_print kernel console should send CRLF not LFCR

In
commit d358788f3f30113e49882187d794832905e42592
Author: Russell King <rmk@dyn-67.arm.linux.org.uk>
Date:   Mon Mar 20 20:00:09 2006 +0000

Glen Turner reported that writing LFCR rather than the more
traditional CRLF causes issues with some terminals.

Since this afflicts many serial drivers, extract the common code to a
library function (uart_console_write) and arrange for each driver to
supply a "putchar" function.

but early_printk is left out.

Signed-off-by: Yinghai Lu <yinghai.lu@sun.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

i386: Enable CX8/PGE CPUID bits early on VIA C3

Fix boot failures with the early CPUID checking on VIA C3

Includes fixes from Christian Volkmann

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

i386: Fix wrong CPU error message in early boot path

- boot/setup.S did not print "PANIC: CPU too old for this kernel"
  ( not visible, also the message did not match )
- I add "# missed before: set ds"
  => somebody should check if I am right with the way to set.
  => seems to be a generic error in setup.S not to set "ds" for error messages.

AK: extracted patch out of other changes
AK: also couldn't find any other case where ds is wrong
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

i386: Clear MCE flag on AMD K6

It reports machine check capability in CPUID, but doesn't actually
implement all the necessary MSRs of the standard Intel machine
check architecture.

This fixes a boot failure on K6s recently introduced.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

x86_64: Support gcc 5 properly

The ifdef tests were broken.  Assume it acts like gcc 4

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

i386: Fix K8/core2 oprofile on multiple CPUs

Only try to allocate MSRs once instead of for every CPU.

This assumes the MSRs are the same on all CPUs which is currently
true. P4-HT is a special case for different SMT threads, but the code
always saves/restores all MSRs so it works identical.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

x86_64: Support x86_64 in make buildtar

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

i386: Update defconfig

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

x86_64: Update defconfig

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin SPI: cleanup according to David Brownell's review

a) platorm_driver_probe(...) instead of platform_driver_register(&driver);
b) set bfin_spi_enable and bfin_spi_disable static
c) Why is the width flag a u32?
d) maybe use dev_dbg() instead of pr_debug()

Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin serial driver: implement support for ignoring parity/break errors

properly setting up and respecting the read_status_mask / ignore_status_mask fields of the serial core

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin serial driver: fix overhead issue

Signed-off-by: Aubrey Li <aubrey.li@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: update pm.c according to power management API change.

Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: update defconfig files

Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: add board default configs to blackfin arch

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: Change NO_ACCESS_CHECK to ACCESS_CHECK

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: fix signal handling bug

There's a forum thread at
https://blackfin.uclinux.org/gf/project/uclinux-dist/forum/?action=ForumBrowse&_forum_action=MessageReply&message_id=24741

which has a testcase involving signal handling that crashes quite readily.
Inspecting the code I believe what happens is that signal handling can become
confused when it is invoked on return from an interrupt, if the contents of
P0 and R0 at the time of the interrupt happen to be such that P0 is larger
than zero (indicating to the signal code that we're in a syscall), and R0
happens to have a value of something like -EINTR or -ERESTARTSYS.
Fixed by setting orig_p0 to -1 if we're returning from an interrupt.  The
testcase now seems to run without problems.

Signed-off-by: Bernd Schmidt <bernd.schmidt@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: cache SWRST value at bootup so other things like watchdog can non-destructively query it

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: update blackfin header files to latest one in VDSP.

a) add new processor BF52x/BF54x header files
b) update blackfin BF533/BF537/BF561 header files to latest one in VDSP.
c) scrub watchdog/rtc masks from headers as we dont need/want them (too generic and the drivers dont use them)

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Roy Huang <roy.huang@analog.com>
Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: make sure we use local labels

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: dont clear the bit that tells coreb to start booting

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: dma_memcpy borken for > 64K

Signed-off-by: Aubrey Li <aubrey.li@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: document why we have to touch the UART peripheral in our boot up code

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: issue reset via SWRST so we dont clobber the watchdog state

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: move board specific setup out of common init code and into the board specific init code

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: finish removing p* volatile defines for MMRs

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: dont clear status register bits in SWRST so we can actually use it

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: make sure we declare the revid functions as pure (since they are)

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: initial tepla-bf561 board support

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: Fix bug using usb keyboard crashes kernel

Without conswitchp preset, we have the following situation:

 - During initcalls: con_init is called, and returns because of
   !display_desc.

 - At this point there is no memory allocated for vc_cons[].d
   A bit later vty_init calls kbd_init.

 - From now on events are passed to kbd_event which will then call
   kbd_keycode.

 - kbd_keycode will oops on a NULL pointer dereference on vc->vc_tty

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
[ Added commit description based on email thread. - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: fix trace output for FLAT binaries

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: GPIO fix some defines

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: DMA operation cleanup

1) Disable Interrupts during DMA memcpy to avoid raise conditions.
2) Mark MDMA channel 0 as reserved, since were using it internally.
3) Add DMA based equivalents for insX and outsX.
4) Our insX and outsX only handles len <= 2^16.

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: Move write to VR_CTL closer to IDLE

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: Fix reserved map after we changed PORT_H definition

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: mark a bunch of local functions as static

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: Add configuration data for ISP176x on BF561

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: fix a few random warnings

Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: ISP1761 doesn't work for USB flash disk

ZONE_DMA is required for some drivers subsystem, such as USB/SCSI.

Signed-off-by: Aubrey Li <aubrey.li@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: add SPI MMC driver support on bf533-stamp, tested on STAMP-BF533

Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Blackfin arch: Add Workaround for ANOMALY 05000257

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Detach sched.h from mm.h

First thing mm.h does is including sched.h solely for can_do_mlock() inline
function which has "current" dereference inside. By dealing with can_do_mlock()
mm.h can be detached from sched.h which is good. See below, why.

This patch
a) removes unconditional inclusion of sched.h from mm.h
b) makes can_do_mlock() normal function in mm/mlock.c
c) exports can_do_mlock() to not break compilation
d) adds sched.h inclusions back to files that were getting it indirectly.
e) adds less bloated headers to some files (asm/signal.h, jiffies.h) that were
   getting them indirectly

Net result is:
a) mm.h users would get less code to open, read, preprocess, parse, ... if
   they don't need sched.h
b) sched.h stops being dependency for significant number of files:
   on x86_64 allmodconfig touching sched.h results in recompile of 4083 files,
   after patch it's only 3744 (-8.3%).

Cross-compile tested on

all arm defconfigs, all mips defconfigs, all powerpc defconfigs,
alpha alpha-up
arm
i386 i386-up i386-defconfig i386-allnoconfig
ia64 ia64-up
m68k
mips
parisc parisc-up
powerpc powerpc-up
s390 s390-up
sparc sparc-up
sparc64 sparc64-up
um-x86_64
x86_64 x86_64-up x86_64-defconfig x86_64-allnoconfig

as well as my two usual configs.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Fix "fs: convert core functions to zero_user_page"

The bug was introduced by 01f2705daf5a36208e69d7cf95db9c330f843af6.
It misses to convert the first argument, it should be "new_page".

This became a cause of fatfs corruption.

Cc: Nate Diller <nate.diller@gmail.com>
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge /pub/scm/linux/kernel/git/herbert/crypto-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  [CRYPTO] api: Read module pointer before freeing algorithm

[CRYPTO] api: Read module pointer before freeing algorithm

The function crypto_mod_put first frees the algorithm and then drops
the reference to its module.  Unfortunately we read the module pointer
which after freeing the algorithm and that pointer sits inside the
object that we just freed.

So this patch reads the module pointer out before we free the object.

Thanks to Luca Tettamanti for reporting this.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Linux v2.6.22-rc2

cciss: Fix pci_driver.shutdown while device is still active

Fix an Oops in the cciss driver caused by system shutdown while a filesystem
on a cciss device is still active.  The cciss_remove_one function only
properly removes the device if the device has been cleanly released by its
users, which is not the case when the pci_driver.shutdown method is called.

This patch adds a new cciss_shutdown function to better match the pattern
used by various SCSI drivers: deactivate device interrupts and flush caches.
It also alters the cciss_remove_one function to match and readds the
__devexit annotation that was removed when cciss_remove_one was serving as
the pci_driver.shutdown method.

Signed-off-by: Gerald Britton <gbritton@alum.mit.edu>
Acked-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Further update of the i386 boot documentation

A number of items in the i386 boot documentation have been either
vague, outdated or hard to read.  This text revision adds several more
examples, including a memory map for a modern kernel load.  It also
adds a field-by-field detailed description of the setup header, and a
bootloader ID for Qemu.

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

Merge git://git./linux/kernel/git/herbert/crypto-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  [CRYPTO] tcrypt: Add missing error check
  [CRYPTO] padlock: Make CRYPTO_DEV_PADLOCK a tristate again

Fix roundup_pow_of_two(1)

1 is a power of two, therefore roundup_pow_of_two(1) should return 1. It does
in case the argument is a variable but in case it's a constant it behaves
wrong and returns 0. Probably nobody ever did it so this was never noticed.

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

timerfd use waitqueue lock ...

The timerfd was using the unlocked waitqueue operations, but it was
using a different lock, so poll_wait() would race with it.

This makes timerfd directly use the waitqueue lock.

Signed-off-by: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

eventfd use waitqueue lock ...

The eventfd was using the unlocked waitqueue operations, but it was
using a different lock, so poll_wait() would race with it.

This makes eventfd directly use the waitqueue lock.

Signed-off-by: Davide Libenzi <davidel@xmailserver.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge branch 'master' of git://git./linux/kernel/git/paulus/powerpc

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc: (32 commits)
  [POWERPC] Remove build warnings in windfarm_core
  [POWERPC] Pass per-file CFLAGs for platform specific op codes
  [POWERPC] Correct #endif comment
  [POWERPC] Fix ppc_rtas_progress_show()
  [POWERPC] Fix sed command lines for zlib source construction
  [POWERPC] Specify GNUTARGET on $(AR) invocations
  [POWERPC] Make sure device node type/name is not NULL on hot-added nodes
  [POWERPC] Small fixes for the Ebony device tree
  [POWERPC] Fix warning on UP
  [POWERPC] cell_defconfig: Disable cpufreq and pmi
  [POWERPC] Fix IO space on PCI buses created from of_platform
  [POWERPC] Add spinlock to request_phb_iospace()
  [POWERPC] Fix make rules for treeImage.initrd
  [POWERPC] Remove warning in mpic.c
  [POWERPC] Update pasemi_defconfig
  [POWERPC] pasemi: CONFIG_GENERIC_TBSYNC no longer needed
  [POWERPC] Update iseries_defconfig
  [POWERPC] Wire up some more syscalls
  [POWERPC] Fix bug adding properties with flatdevtree.c's ft_set_prop()
  [POWERPC] Remove fixup_bigphys_addr() for arch/powerpc to avoid link error
  ...

Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Fix sched_clock() et al.

Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [IPV4]: Remove IPVS icmp hack from route.c for now.
  [IPV4]: Correct rp_filter help text.
  [TCP]: TCP_CONG_YEAH requires TCP_CONG_VEGAS
  [TCP] slow start: Make comments and code logic clearer.
  [BLUETOOTH]: Fix locking in hci_sock_dev_event().
  [NET]: Fix BMSR_100{HALF,FULL}2 defines in linux/mii.h
  [NET]: lockdep classes in register_netdevice

slub: another slabinfo fix

The slab manipulation functions should not be triggered by slabs that
are unresovable in the subset of slabs selected on the command line.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

revert "cancel_delayed_work: use del_timer() instead of del_timer_sync()"

As pointed out by Jarek Poplawski, the patch

[WORKQUEUE]: cancel_delayed_work: use del_timer() instead of del_timer_sync()
commit: 071b638689464c6b39407025eedd810d5b5e6f5d

was wrong, it was merged by mistake after that.

From the changelog:

after this patch:
...
delayed_work_timer_fn->__queue_work() in progress.

The latter doesn't differ from the caller's POV,

it does make a difference if the caller calls flush_workqueue() after
cancel_delayed_work(), in that case flush_workqueue() can miss this
work_struct.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Jarek Poplawski <jarkao2@o2.pl>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[IPV4]: Remove IPVS icmp hack from route.c for now.

Revert: 2d771cd86d4c3af26f34a7bcdc1b87696824cad9

This is dangerous if enabled and a better solution to the
problem is being worked on.

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

[CRYPTO] tcrypt: Add missing error check

The return value of crypto_hash_final isn't checked in test_hash_cycles.
This patch corrects this.  Thanks to Eric Sesterhenn for reporting this.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[SPARC64]: Fix sched_clock() et al.

SPARC64_NSEC_PER_CYC_SHIFT was set too high.

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

Revert "[PATCH] x86: Drop cc-options call for all options supported in gcc 3.2+"

This reverts commit c8fdd247255a3a027cd9f66dcf93e6847d1d2f85.

It turns out the kernel was correct, and the gcc complaint was a gcc
bug.  The preferred stack boundary is expressed not in bytes, but in the
the log2() of the preferred boundary, so "-mpreferred-stack-boundary=2"
is in fact exactly what we want, but a gcc that is compiled for x86-64
will consider it an error (because the 64-bit calling sequence says that
the stack should be 16-byte aligned) even if we are then using "-m32" to
generate 32-bit code.

Noted-by: Mikulas Patocka <mikulas@artax.karlin.mff.cuni.cz>
Cc: Jan Hubicka <jh@suse.cz>
Acked-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

[CRYPTO] padlock: Make CRYPTO_DEV_PADLOCK a tristate again

Turning it into a boolean was unnecessary and caused ALGAPI to be
pinned down as a boolean to.  This patch makes it a tristate again.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Merge git://git.linux-nfs.org/pub/linux/nfs-2.6

* git://git.linux-nfs.org/pub/linux/nfs-2.6:
  SUNRPC: Fix sparse warnings
  NLM: Fix sparse warnings
  NFS: Fix more sparse warnings
  NFS: Fix some 'sparse' warnings...
  SUNRPC: remove dead variable 'rpciod_running'
  NFS4: Fix incorrect use of sizeof() in fs/nfs/nfs4xdr.c
  NFS: use zero_user_page
  NLM: don't use CLONE_SIGHAND in nlmclnt_recovery
  NLM: Fix locking client timeouts...

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:
  sata_via: pcim_iomap_regions() conversion missed BAR5
  libata: remove libata.spindown_compat
  sata_nv: fix fallout of devres conversion
  drivers/ata: remove the wildcard from sata_nv driver

sata_via: pcim_iomap_regions() conversion missed BAR5

pcim_iomap_regions() conversion missed BAR5.  Fix it.

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

libata: remove libata.spindown_compat

With STANDBYDOWN tracking added, libata.spindown_compat isn't
necessary anymore.  If userspace shutdown(8) issues STANDBYNOW, libata
warns.  If userspace shutdown(8) doesn't issue STANDBYNOW, libata does
the right thing.  Userspace can tell whether kernel supports spindown
by testing whether sysfs node manage_start_stop exists as before.

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

sata_nv: fix fallout of devres conversion

As with all other drivers, sata_nv's hpriv is allocated with
devm_kzalloc() and there's no need to free it explicitly.  Kill
nv_remove_one() which incorrectly used kfree() instead of devm_kfree()
and use ata_pci_remove_one() directly.

Original fix is from Peer Chen.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: Peer Chen <pchen@nvidia.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

drivers/ata: remove the wildcard from sata_nv driver

Because nvidia SATA controllers onward base on AHCI, so wildcard in sata_nv
driver is unnecessary.  Also the wildcard sometimes cause sata_nv driver to
be loaded for AHCI controllers,which is not as expected.

Signed-off-by: Peer Chen <pchen@nvidia.com>
Cc: Tejun Heo <htejun@gmail.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

e1000: Fix msi enable leak on error, don't print error message, cleanup

pci_enable_msi failure is a normal event so we should not print any error.
Going over the code I spotted a missing pci_disable_msi() leak when irq
allocation fails. The whole code also needed a cleanup, so I combined the
two different calls to pci_request_irq into a single call making this
look a lot better. All #ifdef CONFIG_PCI_MSI's have been removed.

Compile tested with both CONFIG_PCI_MSI enabled and disabled.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ixgb: don't print error if pci_enable_msi() fails, cleanup minor leak

pci_enable_msi calls can fail for normal operational reasons. Driver
should not print an error message in that case. Fix a leak that leaves
msi enabled if pci_request_irq fails. We can remove CONFIG_PCI_MSI
ifdefs alltogether

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

NetXen: Fix NetXen driver ping on system-p

NetXen: Fix for driver on System-p
This patch will fix a ping issue on system-p

Signed-off by: Milan Bag <mbag@netxen.com>
Signed-off by: Adhiraj Joshi <adhiraj@netxen.com>
Signed-by: Mithlesh Thukral <mithlesh@netxen.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

spidernet: node-aware skbuff allocation

Spidernet was the driver I original did all the node-aware netdevice
allocation for, but after a year it still hasn't hit mainline.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Linas Vepstas <linas@austin.ibm.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

gianfar: Add I/O barriers when touching buffer descriptor ownership.

The hardware must not see that is given ownership of a buffer until it is
completely written, and when the driver receives ownership of a buffer,
it must ensure that any other reads to the buffer reflect its final
state.  Thus, I/O barriers are added where required.

Without this patch, I have observed GCC reordering the setting of
bdp->length and bdp->status in gfar_new_skb.  Hardware reordering
was also theoretically possible.

Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ibm_emac: fix link speed detection change

Fix link speed detection change.
Thanks to Stefan Roese <sr@denx.de> for finding this bug.

CC: Stefan Roese <sr@denx.de>
Signed-off-by: Eugene Surovegin <ebs@ebshome.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ibm_emac: improved PHY support

Original patch is from Jeff Haran  <jharan@brocade.com> with my minor style
fixes. His comments follow:

The first problem was in the function that configures the PHY for
autonegotiation, genmii_setup_aneg(). The original code does a
read/modify/write of the autonegotiation advertizement register (reg 4),
followed by a read/modify/write of the control register (reg 0). While
the original code follows the proper procedure as per reading the IEEE
specs, what I found is that on at least one PHY model (National DP83843)
the read of the control register comes back with the soft reset bit set
(bit 15). Because of the read/modify/write operation, this causes the
write to write a 1 back to the reset bit, which initiates a software
reset of the PHY. This software reset causes the PHY to return to its
power up state which advertizes all modes of operation, thus negating
the write to the autoneg advertizement register. The modification is to
spin reading the control register until the soft reset bit is clear
before doing the modify/write.
The second problem was in the function that configures the PHY for
forced operation, genmii_setup_forced(). The original code initiates a
software reset operation via a write of a 1 to bit 15 of the control
register (reg 0), but then proceeds to do a second write to that same
register without waiting until that reset bit is cleared by the PHY
itself (which according to the IEEE specs indicates that the PHY reset
is complete). This is a violation of how one is supposed to use this
software reset feature of these PHYs and I believe was the cause of
mysterious, difficult to reproduce link failures that we've observed on
some of our systems that use this driver. The fix is to modify the
function so that it spins waiting for the reset bit to clear after doing
the soft reset and before doing the subsequent write.

Signed-off-by: Jeff Haran <jharan@brocade.com>
CC: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Eugene Surovegin <ebs@ebshome.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ibm_emac: fix section mismatch warnings

Fix "Section mismatch" warnings

Signed-off-by: Eugene Surovegin <ebs@ebshome.net>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

small netdevices.txt fix

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

sky2: memory barriers change

Do some memory barrier changes for safety/perfomance:
Don't need read after update to index, mmiowb() followed by read at end
of irq is sufficient.

Signed-off-by: Stephn Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

sky2: remove dual port workaround

This workaround was added to deal with NAPI core and how
it affected dual port shared polling. It turned out not to
be necessary. Stopping device 0 only doesn't stop NAPI from
working completely after that.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

sky2: MIB counter overflow handling

Make sure that if we ever get a MIB counter overflow interrupt (normally
masked off), that the IRQ is cleared.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

sky2: keep track of receive alloc failures

When driver can't allocate receive buffer it drops incoming
packet. Keep a counter.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

sky2: PHY register settings

Align the PHY setup of the sky2 driver with the vendor sk98lin (10.0.4.3)
driver. The PHY register settings are mostly black magic, even with access
to the documentation it isn't clear what the right values are. The changes
are mostly comments, the code change only affects the Yukon FE (100 mbit only)
version.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

sky2: remove Gigabyte 88e8056 restriction

The problems with Gigabyte motherboards are system configuration dependent.
Since it works fine for some users, it doesn't make sense to deprive
them.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

[IPV4]: Correct rp_filter help text.

As mentioned in http://bugzilla.kernel.org/show_bug.cgi?id=5015
The helptext implies that this is on by default.
This may be true on some distros (Fedora/RHEL have it enabled
in /etc/sysctl.conf), but the kernel defaults to it off.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[TCP]: TCP_CONG_YEAH requires TCP_CONG_VEGAS

These two congestion control modules share code.

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

[TCP] slow start: Make comments and code logic clearer.

Add more comments to describe our version of tcp_slow_start().

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[BLUETOOTH]: Fix locking in hci_sock_dev_event().

We presently use lock_sock() to acquire a lock on a socket in
hci_sock_dev_event(), but this goes BUG because lock_sock()
can sleep and we're already holding a read-write spinlock at
that point. So, we must use the non-sleeping BH version,
bh_lock_sock().

However, hci_sock_dev_event() is called from user context and
hence using simply bh_lock_sock() will deadlock against a
concurrent softirq that tries to acquire a lock on the same
socket. Hence, disabling BH's before acquiring the socket lock
and enable them afterwards, is the proper solution to fix
socket locking in hci_sock_dev_event().

Signed-off-by: Satyam Sharma <ssatyam@cse.iitk.ac.in>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Fix BMSR_100{HALF,FULL}2 defines in linux/mii.h

Noticed by Matvejchikov Ilya.

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

[NET]: lockdep classes in register_netdevice

After initializing dev->_xmit_lock register_netdevice()
sets lockdep class according to dev->type.

Idea of this patch - by David Miller.

Reported & tested by: "Yuriy N. Shkandybin" <jura@netams.com>
Signed-off-by: Jarek Poplawski <jarkao2@o2.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' of /home/trondmy/repositories/git/linux-2.6/

Fix incorrect prototype for ipxrtr_route_packet()

The function ipxrtr_route_packet() takes a 'len' argument of type
size_t. However, its prototype in af_ipx.c incorrectly suggests that the
corresponding argument is of type 'int' instead.

Discovered by building with --combine and letting the compiler see it
all at once.

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

NS16550A: Restore HS settings in EXCR2 on resume

After a suspend/resume cycle, the UART may have been reset into
low-speed mode -- either because it's actually been reset, or because
the firmware pokes at the old-style divisor registers. If we detected it
as a NS16550A SuperIO chip in the first place and set baud_base to
921600, then we should do so again in the resume path.

This patch adds that code to serial8250_resume_port(), and also makes
serial8250_resume() actually call serial8250_resume_port() for each port
instead of just calling uart_resume_port() directly. And thus fixes
serial port operation after suspend/resume.

It also fixes a bogus comment where we write the EXCR2 register with a
comment saying /* EXCR1 */

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Acked-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: more rmap checking

Re-introduce rmap verification patches that Hugh removed when he removed
PG_map_lock. PG_map_lock actually isn't needed to synchronise access to
anonymous pages, because PG_locked and PTL together already do.

These checks were important in discovering and fixing a rare rmap corruption
in SLES9.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Fix page allocation flags in grow_dev_page()

grow_dev_page() simply passes GFP_NOFS to find_or_create_page.  This means
the allocation of radix tree nodes is done with GFP_NOFS and the allocation
of a new page is done using GFP_NOFS.

The mapping has a flags field that contains the necessary allocation flags
for the page cache allocation.  These need to be consulted in order to get
DMA and HIGHMEM allocations etc right.  And yes a blockdev could be
allowing Highmem allocations if its a ramdisk.

Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

swsusp: fix sysfs interface

The sysfs files /sys/power/disk and /sys/power/state do not work as
documented, since they allow the user to write only a few initial
characters of the input string to trigger the option (eg.  'echo pl >
/sys/power/disk' activates the platform mode of hibernation).  Fix it.

Special thanks to Peter Moulder <Peter.Moulder@infotech.monash.edu.au> for
pointing out the problem.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

circular locking dependency found in QUOTA OFF

i_mutex on quota files is special.  Unlike i_mutexes for other inodes it is
acquired under dqonoff_mutex.  Tell lockdep about this lock ranking.  Also
comment and code in quota_sync_sb() seem to be bogus (as i_mutex for quota
file can be acquired under dqonoff_mutex).  Move truncate_inode_pages()
call under dqonoff_mutex and save some problems with races...

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

i386: don't check_pgt_cache in flush_tlb_mm

No other architecture calls check_pgt_cache() from within flush_tlb_mm(),
and i386 is already calling check_pgt_cache() from the usual places,
tlb_finish_mmu() and cpu_idle() (the latter being odd, but not unusual).
flush_tlb_mm() has no business to be freeing pages: remove that line, which
sneaked in with slub's i386 support.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Cc: Andi Kleen <ak@suse.de>
Acked-by: Christoph Lameter <clameter@sgi.com>
Acked-by: William Lee Irwin III <wli@holomorphy.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ecryptfs: use zero_user_page

Use zero_user_page() instead of open-coding it.

Signed-off-by: Nate Diller <nate.diller@gmail.com>
Cc: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

make sysctl/kernel/core_pattern and fs/exec.c agree on maximum core filename size

Make sysctl/kernel/core_pattern and fs/exec.c agree on maximum core
filename size and change it to 128, so that extensive patterns such as
'/local/cores/%e-%h-%s-%t-%p.core' won't result in truncated filename
generation.

Signed-off-by: Dan Aloni <da-x@monatomic.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>