[PATCH] Make __always_inline actually force always inlining

This patch is the first in a series that tries to optimize the kernel in terms
of size (and thus cache behavior, both cpu and pagecache).

This first patch changes __always_inline to be a forced inline instead of the
"regular" inline it was on everything except alpha.  This forced inline
matches the intention of the define better as a matter of documentation.
There is no change in behavior by this patch, since "inline" currently is
mapped to a forced inline anyway.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] fbdev: Update maintainers list

Update maintainers list.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] fbdev: Sanitize ->fb_mmap prototype

No need for a file argument.  If we'd really need it it's in vma->vm_file
already.  gbefb and sgivwfb used to set vma->vm_file to the file argument, but
the kernel alrady did that.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] fbdev: Sanitize ->fb_ioctl prototype

The ioctl and file arguments to ->fb_mmap are totally unused and there's not
reason a driver should need them.

Also update the ->fb_compat_ioctl prototype to be the same as ->fb_mmap.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] gx1fb: (try to) play nicer with various BIOSes

Seems that the CS5530A chip used in Geode GX1 systems has some crazy feature
that causes SMI traps when accessing the PCI configuration space of the video
device.  Various GX1 BIOSes seem to use this 'feature' to hide the real BARs
of the device.  This patch disables these traps (in an early PCI fixup) so
that Linux sees the real, physical BARs and not the virtual ones provided by
the BIOS.

This should allow the GX1 framebuffer driver to work on more systems that have
different BIOSes as the driver no longer guesses at what the virtual BARs
mean.

I'm not entirely sure it the correct solution as I can neither test regular
VGA console nor the X's 'cyrix' video driver so there might be some breakage
there -- probably best to get some more testers before applying it.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] neofb: take existing display configuration as default

On a Dell Latitude CPi-A I noticed a strangeness wrt.  the handling of an
external monitor by the neomagic framebuffer driver, namely when the laptop is
docked in a C/Dock II with the lid shut.

A cold boot would result in the BIOS configuring the video chip to use the
"external monitor only" mode, yet neofb would default to "internal LCD only".
An attempt for a quick fix by using the Fn-F8 keystroke to toggle the display
combination modes resulted in a reproductible hard lock, powering down being
the only solution.

The attached patch makes neofb probe the register for the current display
mode, using that value as a default if nothing was specified as kernel/module
parameter.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Update kernel-parameters.txt IOSCHED to spell out 'anticipatory'

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nlm kernel-parameters update

Add 2 lockd kernel parameters and spell 2 others correctly.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: Neil Brown <neilb@cse.unsw.edu.au>
Cc: Olaf Kirch <okir@suse.de>
Cc: <buraphalinuxserver@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] cs89x0: add ixdp2351 support

This patch adds support for the Intel IXDP2351 to the CS89x0 driver.

Signed-off-by: Deepak Saxena <dsaxena@plexity.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] cs89x0: use u16 for device register data

cs89x0 inconsistently used 'int' and 'u32' for device register data.
As the cs89x0 is a 16-bit chip, change the I/O accessors over to 'u16'.
(Spotted by Deepak Saxena.)

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

[PATCH] cs89x0: use #elif instead of #else/#if/#endif

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

[PATCH] cs89x0: credit Dmitry Pervushin

Credit Dmitry Pervushin for the PNX010X platform support.

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

[PATCH] dell_rbu: fix Bug 5854

This fixes http://bugzilla.kernel.org/show_bug.cgi?id=5854

Root cause:

The dell_rbu driver creates entries in /sys/class/firmware/dell_rbu/ by
calling request_firmware_nowait (without hotplug ) this function inturn
starts a kernel thread which creates the entries in
/sys/class/firmware/dell_rbu/loading , data and the thread waits on the
user action to return control back to the callback fucntion of dell_rbu.
The thread calls wait_on_completion which puts it in a D state until the
user action happens.  If there is no user action happening the load average
goes up as the thread D state is taken in to account.  Also after
downloading the BIOS image the enrties go away momentarily but they are
recreated from the callback function in dell_rbu.  This causes the thread
to get recreated causing the load average to permenently stay around 1.

Fix:

The dell_rbu also creates the entry
/sys/devices/platform/dell_rbu/image_type at driver load time.  The image
type by default is mono if required the user can echo packet to image_type
to make the BIOS update mechanism using packets.  Also by echoing init in
to image_type the /sys/class/firmware/dell_rbu entries can be created.

The driver code was changed to not create /sys/class/firmware/dell_rbu
entries during load time, and also to not create the above entries from the
callback function.  The entries are only created by echoing init to
/sys/devices/platform/dell_rbu/image_type The user now needs to create the
entries to download the image monolithic or packet.  This fixes the issue
since the kernel thread only is created when ever the user is ready to
download the BIOS image; this minimizes the life span of the kernel thread
and the load average goes back to normal.

Signed off by Abhay Salunke <abhay_salunke@dell.com>

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

[PATCH] smbfs: remove kmalloc wrapper

Remove the remaining kmalloc() wrapper bits from fs/smbfs/.

Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ncpfs: remove kmalloc wrapper

Remove remaining kmalloc wrapper bits from fs/ncpfs/.

Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] abandon gcc 295x main.c tidy

After abandon-gcc-295x.patch, this relocates the error-out-early comment.

Signed-off-by: Coywolf Qi Hunt <qiyong@fc-cn.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] quota: make useless quota error message informative

fs/quota_v2.c can, under some conditions, issue a kernel message that says,
in totality, 'failed read'.  This patch does the following:

1) Gives a hint who issued the error message, so people reading the logs
   don't have to go grepping the entire kernel tree (with 11 false
   positives).

2) Say what amount of data we expected, and actually got.

Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Cc: Jan Kara <jack@ucw.cz>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] reiserfs: remove d_splice_alias NULL check from reiserfs_lookup

Remove redundant NULL check in reiserfs_lookup() as d_splice_alias() can take
NULL inode as input.

Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] isofs: remove d_splice_alias NULL check from isofs_lookup

Remove redundant NULL check in isofs_lookup() as d_splice_alias() can take
NULL inode as input.

Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ext3: remove d_splice_alias NULL check from ext3_lookup

Remove redundant NULL check in ext3_lookup() as d_splice_alias() can take NULL
inode as input.

Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ext2: remove d_splice_alias NULL check from ext2_lookup

Remove redundant NULL check in ext2_lookup() as d_splice_alias() can take NULL
inode as input.

Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] cpuset oom lock fix

The problem, reported in:

  http://bugzilla.kernel.org/show_bug.cgi?id=5859

and by various other email messages and lkml posts is that the cpuset hook
in the oom (out of memory) code can try to take a cpuset semaphore while
holding the tasklist_lock (a spinlock).

One must not sleep while holding a spinlock.

The fix seems easy enough - move the cpuset semaphore region outside the
tasklist_lock region.

This required a few lines of mechanism to implement.  The oom code where
the locking needs to be changed does not have access to the cpuset locks,
which are internal to kernel/cpuset.c only.  So I provided a couple more
cpuset interface routines, available to the rest of the kernel, which
simple take and drop the lock needed here (cpusets callback_sem).

Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: fix blk_queue_ordered call in dasd.c fixup

The QUEUE_ORDERED_* numbers got renumbered and by accident the dasd driver
was changed to use QUEUE_ORDERED_DRAIN instead of QUEUE_ORDERED_TAG.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: email-address change

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: chps[] array too short

The chps[] array in struct channel_subsystem is one too short; therefore the
code doesn't realize the chpid ff is already known.  When several devices on
chpid ff become available, the message "new_channel_path: could not register
ff" is displayed for every device but the first one.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: cputime misaccounting

finish_arch_switch needs to update the user cpu time as well, not just the
system cpu time.  Otherwise the partial user cpu time of a process that is
stored in the lowcore will be (mis-)accounted to the next process.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: fix blk_queue_ordered call in dasdc

Add the missing third argument to the blk_queue_ordered call and use the
constant QUEUE_ORDERED_DRAIN instead of "1".

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: add dummy pm_power_off

Define a dummy pm_power_off pointer to make sys_reboot happy.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: spinlock fixes

Remove useless spin_retry_counter and fix compilation for UP kernels.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: fix cpcmd calls on UP

Add missing fourth argument to cpcmd calls under !CONFIG_SMP.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: sigcontext.h vs __user

Add an include of linux/compiler.h in sigcontext.h to avoid compiler errors in
user space apps because of a missing definition for __user.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: show_task oops

The show_task function walks the kernel stack backchain of processes assuming
that the processes are not running.  Since this assumption is not correct
walking the backchain can lead to an addressing exception and therefore to a
kernel hang.  So prevent the kernel hang (you still get incorrect results)
verity that all read accesses are within the bounds of the kernel stack before
performing them.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: sha256 crypto code fix

Fix processing of messages larger than 2 * SHA256_BLOCK_SIZE.

Signed-off-by: Jan Glauber <jan.glauber@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: aes crypto code fixes

Call KM[C] only with a multiple of block size.  Check return value of KM[C]
instructions and complain about erros

Signed-off-by: Jan Glauber <jan.glauber@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: des crypto code speedup

Provide ECB and CBC encrypt / decrypt functions to crypto API to speed up our
hardware accelerated DES implementation.  This new functions allow the crypto
API to call ECB / CBC directly with large blocks in difference to the old
functions that were calles with algorithm block size (8 bytes for DES).

This is up to factor 10 faster than our old hardware implementation :)

Signed-off-by: Jan Glauber <jan.glauber@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] s390: des crypto code cleanup

Beautify the s390 in-kernel-crypto des code.

Signed-off-by: Jan Glauber <jan.glauber@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] i386: fix stack dump loglevel

Recent changes caused part of stack traces from SysRq-T to print at
KERN_EMERG loglevel.  Also, parts of stack dump during oops were failing to
print at that level when they should.

Signed-off-by: Chuck Ebbert <76306.1226@compuserve.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] i386: put HOTPLUG_CPU under Processor type, not Bus options

Move the HOTPLUG_CPU option under "Processor type" instead of under "Bus
options".  This makes it the same for i386 as most other processor types
(arm, ia64, parisc, ppc, s390, & x86_64; but not for powerpc).  Besides, it
takes me too long to find it under Bus options.  I can't be the only person
who has trouble finding it.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] powerpc: Add support for the MPC83xx watchdog

Add support for the PowerPC MPC83xx watchdog.  The MPC83xx has a simple
watchdog that once enabled it can not be stopped, has some simple timeout
range selection, and the ability to either reset the processor or take a
machine check.

Signed-off-by: Dave Updegraff <dave@cray.org>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Cc: Wim Van Sebroeck <wim@iguana.be>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Add tmpfs options for memory placement policies

Anything that writes into a tmpfs filesystem is liable to disproportionately
decrease the available memory on a particular node.  Since there's no telling
what sort of application (e.g.  dd/cp/cat) might be dropping large files
there, this lets the admin choose the appropriate default behavior for their
site's situation.

Introduce a tmpfs mount option which allows specifying a memory policy and
a second option to specify the nodelist for that policy.  With the default
policy, tmpfs will behave as it does today.  This patch adds support for
preferred, bind, and interleave policies.

The default policy will cause pages to be added to tmpfs files on the node
which is doing the writing.  Some jobs expect a single process to create
and manage the tmpfs files.  This results in a node which has a
significantly reduced number of free pages.

With this patch, the administrator can specify the policy and nodes for
that policy where they would prefer allocations.

This patch was originally written by Brent Casavant and Hugh Dickins.  I
added support for the bind and preferred policies and the mpol_nodelist
mount option.

Signed-off-by: Brent Casavant <bcasavan@sgi.com>
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Robin Holt <holt@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Fix for CONFIG_NUMA without CONFIG_SWAP

Some people apparently run CONFIG_NUMA without CONFIG_SWAP.  The migration
code currently depends on swap.  This patch provides a set of inline
fallback functions so that the kernel properly compiles.  However, calls to
migration functions will fail.

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

[PATCH] Watchdog: Winsystems EPX-C3 SBC

This is a 2.6 patch that adds support for the watchdog timer built into the
EPX-C3 single board computer manufactured by Winsystems, Inc.

Driver details:

This is for x86 only.  This watchdog is pretty basic and simple.  It is
only configurable via jumpers on the SBC, and it only has either a 1.5s or
200s interval.  The watchdog can either be auto-configured to start as soon
as the machine powers up (bad idea for the 1.5s interval!) or it can be
enabled and disabled by writing to io port 0x1ee.  Petting the watchdog
involves writing any value to io port 0x1ef.

The only unfortunate thing about this watchdog (and it is not at all
uncommmon in watchdogs that linux supports) is that it is not a PCI or
ISA-PNP device and as such it isn't at all probeable.  Either the watchdog
exists as 2 bytes at 0x1ee, or it doesn't.  Thus, using this driver on a
machine that doesn't have that watchdog can potentially hang/crash the
system, etc.  So only use this driver if you in fact are on a Winsystems
EPX-C3 SBC.

Anyway this driver fits into the already-existing watchdog framework quite
nicely and I already tested it on my EPX-C3 and it works like a charm.

Signed-off-by: Calin A. Culianu <calin@ajvar.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] uml: fix symbol for mktime

  LD      .tmp_vmlinux1
/usr/lib/gcc-lib/i486-linux/3.3.4/../../../libc.a(mktime.o): In function `timelocal':
: multiple definition of `mktime'
kernel/built-in.o:kernel/time.c:604: first defined here
/usr/bin/ld: Warning: size of symbol `mktime' changed from 134 in kernel/built-in.o to 44 in /usr/lib/gcc-lib/i486-linux/3.3.4/../../../libc.a(mktime.o)
collect2: ld returned 1 exit status

Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Unlinline a bunch of other functions

Remove the "inline" keyword from a bunch of big functions in the kernel with
the goal of shrinking it by 30kb to 40kb

Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] sched: add new SCHED_BATCH policy

Add a new SCHED_BATCH (3) scheduling policy: such tasks are presumed
CPU-intensive, and will acquire a constant +5 priority level penalty.  Such
policy is nice for workloads that are non-interactive, but which do not
want to give up their nice levels.  The policy is also useful for workloads
that want a deterministic scheduling policy without interactivity causing
extra preemptions (between that workload's tasks).

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Altix: ioc3 serial support

Add driver support for a 2 port PCI IOC3-based serial card on Altix boxes:

This is a re-submission.  On the original submission I was asked to
organize the code so that the MIPS ioc3 ethernet and serial parts could be
used with this driver.  Stanislaw Skowronek was kind enough to provide the
shim layer for this - thanks Stanislaw.  This patch includes the shim layer
and the Altix PCI ioc3 serial driver.  The MIPS merged ioc3 ethernet and
serial support is forthcoming.

Signed-off-by: Patrick Gefre <pfg@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] convert /proc/devices to use seq_file interface

A Christoph suggested that the /proc/devices file be converted to use the
seq_file interface.  This patch does that.

I've obxerved one or two installation that had sufficiently large sans that
they overran the 4k limit on /proc/devices.

Signed-off-by: Neil Horman <nhorman@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] autofs4 oops fix

We forgot to initialise a couple of nameidata fields.

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

[PATCH] Fix double decrement of mqueue_mnt->mnt_count in sys_mq_open

Fixed the refcounting on failure exits in sys_mq_open() and
cleaned the logics up.  Rules are actually pretty simple - dentry_open()
expects vfsmount and dentry to be pinned down and it either transfers
them into created struct file or drops them.  Old code had been very
confused in that area - if dentry_open() had failed either in do_open()
or do_create(), we ended up dentry and mqueue_mnt dropped twice, once
by dentry_open() cleanup and then by sys_mq_open().

Fix consists of making the rules for do_create() and do_open()
same as for dentry_open() and updating the sys_mq_open() accordingly;
that actually leads to more straightforward code and less work on
normal path.

Signed-off-by: Al Viro <aviro@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Merge /pub/scm/linux/kernel/git/jejb/scsi-misc-2.6

Merge /pub/scm/linux/kernel/git/gregkh/spi-2.6

Merge /pub/scm/linux/kernel/git/gregkh/driver-2.6

[PATCH] remove unused tmp_buf_sem's

tmp_buf_sem sems to be a common name for something completely unused...

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Greg Kroah-Hartman <gregkh@suse.de> ("usb portion")
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] docs: update some updated code docs

Based on comments from Randy Dunlap on my previous commit
5b0ed2c64d8fdafb5fcfb3baabdd288628b1ff9b

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

[PATCH] ufs cleanup

Here is update of ufs cleanup patch, brought on by the recently fixed
ubh_get_usb_second() bug that made some ugly code rather painfully
obvious.  It also includes

 - fix compilation warnings which appears if debug mode turn on
 - remove unnecessary duplication of code to support UFS2

I tested it on ufs1 and ufs2 file-systems.

Signed-off-by: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

Merge /pub/scm/linux/kernel/git/dtor/input

[SCSI] qla2xxx: Update version number to 8.01.04-k.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Disable port-type RSCN handling via driver state-machine.

Given the semantic changes in both the device-model and
fc-transport APIs, the driver's handling of port-type RSCNs
via a series of ADISCs and PLOGIs can cause series of
badness ranging from unexpectedly device loss to devices not
being discovered.

In the interim, disable (via a module-parameter) this
feature and allow RSCN management to continue to occur
within the driver's DPC thread.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Kconfig update: Add URL to download firmware images.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Correct setting of FDMI supported/current port speed.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Correct execution-throttle setting for ISP24xx.

Similarly to other ISPs, set execution throttle to maximum
allowed value since 'throttling' is done on a per-lun basis
via queue-depth.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Collapse load RISC RAM implementations.

Simplify essentially duplicate load RISC RAM implementation
in qla2x00_load_ram_ext() and qla2x00_load_ram().

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Correct swing/emphasis settings for ISP24XX.

Swing/emphasis settings in NVRAM were not being honoured due
to the driver not converting the serial-link options from LE
to host-endian format.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Correct issue where portstate does not transition during loop-resync.

If the Get Port Database call fails during local-loop
update, then schedule the DPC routine to perform a rescan as
the firmware would have updated the Get ID List port-entries
of their new state.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Update firmware-dump procedure for ISP24xx.

Small changes to register retrieval and order as per latest
firmware specification.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Re-enable flash-part write protection on ISP24xx boards.

Driver would not correctly re-enable the write-protection
bits of the flash part after updates.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Correct excessive delay during LOAD-RISC-RAM mailbox command.

Problem report (against 2.4.x driver) from Jeff Layton
<jlayton@redhat.com>:

An OEM noticed that the U6 qla2200 driver would hang for
around 2 minutes at boot time and then proceed normally. I
found that the delay was occurring when loading the new
firmware into the card, and was due to a
schedule_timeout(10) added to the bottom of the polling
loop.

Some testing showed that the load ram operation on the card
was very quick (on the order of a couple of jiffies), but
the sleep in the polling loop was making each operation take
around 25-30.

The attached patch corrects this by making it skip sleeping
during the load ram operation, since I believe we only do
that when the module is plugged in. It also skips sleeping
if the mbox_int flag got set during the current loop.

This corrected the hang on my test setup, and OEM also
confirmed that it corrected the problem for them.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Use msleep() as delay during ISP polling.

Mailbox commands are polled for completion during ISP
initialization.  During potentially 'long' mailbox commands
(i.e. fabric login), we really don't want a busy-wait delay
to potentially trigger a (benign) soft-lockup BUG().

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Drop noisy 'UNDERRUN' status message.

There's no point in displaying the message during a valid
underrun case.  Limit the message to potentially problematic
cases.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Correct FC4 feature assignment during RFF_ID.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] qla2xxx: Reference proper node/port names in fc_host class.

The initial-control-block references are not always correct
as the use-node-name qualifier during NVRAM configuration
will cause the firmware to use the portname as a base for
the nodename.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] iscsi: use pageslab

From: FUJITA Tomonori <tomof@acm.org> and zhenyu.z.wang@intel.com:

We cannot handle filesystems like XFS becuase of the pages they
are sending us. We had thought page_count could be used to
work around this, but the correct test is for PageSlab.

The proper solution is to figure out what type of pages
filesystems can use so we do not have to add tests like
this or handle it in the block layer for all network block drivers
but the issue still has not been resolved on fs-devel
so we are sending this patch as a temporary fix.

This is last patch just in case it is Nakd with the explanation
that we need to push the correct fix through fs-devel, mm
or the block layer. The rest of the patchset can live without
the patch, but the driver will not work with filesystems like
XFS.

Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] iscsi: fix 4k stack iscsi setups

When we run the xmit code from queuecomand the stack trace
gets too deep. The patch runs the xmit code from the scsi_host
work queue. This fixes 4k stack and xfs support and should
fix the st and sg stack usage bugs.

Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] iscsi: seperate iscsi interface from setup functions

This is the second version of the patch to address Christoph's comments.
Instead of doing the lib, I just kept everything in scsi_trnapsort_iscsi.c
like the FC and SPI class. This was becuase the driver model and sysfs
class is tied to the session and connection setup so separating did not
buy very much at this time.

The reason for this patch was becuase HW iscsi LLDs like qla4xxx cannot
use the iscsi class becuase the scsi_host was tied to the interface and
class code. This patch just seperates the session from scsi host so
that LLDs that allocate the host per some resource like pci device
can still use the class.

This is also fixes a couple refcount bugs that can be triggered
when users have a sysfs file open, close the session, then
read or write to the file.

Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] iscsi: add high mem support

From Mike Christie <michaelc@cs.wisc.edu> and FUJITA Tomonori <tomof@acm.org>:

We cannot use page_address becuase some pages could be highmem.
Instead, we can use sock_no_sendpage which does kmap for us.

Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] iscsi: data digest page cache usage fix

Users can write to a page while we are sending it and making
digest calculations. This ends up causing us to retry the command
when a digest error is later reported. By using sock_no_sendpage
when data digests are calculated we can avoid a lot of (not all but it
helps) the retries becuase sock_no_sendpage is not zero copy.

Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] iscsi: host locking fix

We should be taking the host_lock instead of the conn lock when
checking host_busy.

Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] iscsi: data under/over flow fix

We need to check the ISCSI_FLAG_DATA_* flags.

Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] iscsi: whitespace cleanup

Remove extra whitespaces.

Signed-off-by: Alex Aizman <itn780@yahoo.com>
Signed-off-by: Dmitry Yusupov <dmitry_yus@yahoo.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] sem2mutex 3w-[x9]xxx

Convert a the 3w-9xxx.c and 3w-xxxx.c drivers to use mutexes instead
of semaphores. Untested, but compiles and looks obviously correct.

Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] mptfc: need to select transport attrs

Now that mptfc actually uses the transport class, it can't be built
without it.

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

[SCSI] fusion - fix pci express bug

The fix is to write  'MPI_HIM_DIM' to the Host Interrupt Mask
register, when enabling interrupts.  Instead of the
tilde of MPI_HIM_RIM.

Apparently writing '1's to some of the reserved bits was causing
all the bits to go to `1`, which effectly disabled all interrupts.

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

[SCSI] scsi_transport_sas: mapping the rphy channel equal to the port identifier

We will be mapping the RAID volumes in mptsas to a reserved
channel that
is one larger than the anticapated number of ports on the direct
attached host
adapter.

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

[SCSI] fusion - mpi header udpate

This updates mpi headers in fusion drivers to version 1.5.12.

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

[SCSI] fusion - adding raid support in mptsas

The SAS RAID volumes are reported beyond the expected number of phys.

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

[SCSI] remove target parent limitiation

When James Smart fixed the issue of the userspace scan atributes
crashing the system with the FC transport class he added a patch to
let the transport class check if the parent is valid for a given
transport class.

When adding support for the integrated raid of fusion sas devices
we ran into a problem with that, as it didn't allow adding virtual
raid volumes without the transport class knowing about it.

So this patch adds a user_scan attribute instead, that takes over from
scsi_scan_host_selected if the transport class sets it and thus lets
the transport class control the user-initiated scanning.  As this
plugs the hole about user-initiated scanning the target_parent hook
goes away and we rely on callers of the scanning routines to do
something sensible.

For SAS this meant I had to switch from a spinlock to a mutex to
synchronize the topology linked lists, in FC they were completely
unsynchronized which seems wrong.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] fusion - adding support for FC949ES

Add software recognition for the new LSI Logic Fibre Channel controller.

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

[SCSI] mptfusion - fc transport attributes

Signed-off-by: Michael Reed <mdr@sgi.com>
Signed-off-by: Eric Moore <Eric.Moore@lsil.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] sem2mutex: scsi_transport_spi.c

Convert the SCSI transport class code to use a mutex rather than a
semaphore.

Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] fix up message/i2o/pci.c

There was a use before initialisation of c->name

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

[SCSI] fusion: kzalloc / kcalloc conversion

Convert kmalloc + memset to kzalloc or kcalloc in fusion.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] fusion: convert semaphores to mutexes

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] mptsas: support basic hotplug

Adds hotplug support for SAS end devices.  Unfortunately the fusion
firmware doesn't generate similar events for expanders addition/removal
so we can't support them yet.  Eric has an idea about a clever scheme to
find out about expander changes so that'll be added later on.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] I2O: move pci_request_regions() just behind pci_enable_device()

The problem in dpt_i2o could be the pci config space accesses it
triggers as it loads, dangerous to do if there is any I/O activity going
on in the other driver (probable if a boot driver I guess).

I approve this patch to dpt_i2o.c, and am applying it to the Adaptec
branch of the driver.

Thanks for the investigation Ryoji.

---

In linux 2.6.15, data transfer does hang when both dpt_i2o
and i2o_block drivers are loaded.
It seems that location of pci_request_regions() are wrong.
I moved it just behind pci_enable_device() like other drivers,
and it becomes fine.

Signed-off-by: Ryoji Kamei <kamei@miraclelinux.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] ips: Mode Sense (Caching Page ) fix

To avoid the "sda: got wrong page" message, the ServeRAID driver
should be setting flags indicating that the Mode Sense commands are
not supported.

Signed-off-by: Jack Hammer <jack_hammer@adaptec.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] zfcp: transport class adaptations II

Replaced zfcp adapter attributes with fc_host attributes:
fc_topology by port_type, physical_wwpn by permanent_port_name.
Make use of fc_host attribute supported_speeds.
Removed zfcp adapter attribute physical_s_id.

Signed-off-by: Andreas Herrmann <aherrman@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] zfcp: transport class adaptations

Added host stats, removed superfluous get_starget_ functions,
removed some attributes from zfcp specific sysfs tree (e.g.
scsi_host_no, scsi_lun, wwnn and d_id).
Host stats are given for the physical adapter port not for the
virtual adapter. Reset stats is implemented in the device driver.

Signed-off-by: Andreas Herrmann <aherrman@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] zfcp: handle unsolicited status notification lost

Handle unsolicited adapter status that informs about loss of
previous unsolicited status notification(s).

Signed-off-by: Maxim Shchetynin <maxim@de.ibm.com>
Signed-off-by: Andreas Herrmann <aherrman@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[SCSI] fc transport: add permanent_port_name fc_host attribute

Add fc_host attribute permanent_port_name which is
used to show the port name of the primary port -
the port that initially logged into the fabric.

For a virtual port (registered via the primary port with
FDISC command) it is useful to know not only its (virtual)
port name but also the permanent port name.

Signed-off-by: Andreas Herrmann <aherrman@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>