Based on comments from hch and jejb, an update to the ibmvscsi
driver patch.  Primarily style issues, though the removal of
gratuitous casting lead to some significant refactoring
of functions.

- Moved #include <asm/...> after #include <linux/...>
- removed the map_direct_data and map_indirect_data and moved
   the function inline
- Fixed function declaration style issues (static on a line by
   itself)
- Removed bogus cast from dma_map_single
- Moved setting of data transfer direction to a single helper
   function
- Fixed up some gratuitous casting of types
- Collapsed a couple of other functions inline
- Removed bogus check for NULL from remove
- Removed the "evt_struct_for" routine as being needlessly obfuscatory
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

Original Patch From: Adrian Bunk <bunk@fs.tum.de>

I looked at those 3 functions. dc390_EnableMsgOut_Abort is called 5 times
in the code and is 0x30 bytes long (as compiled with 3.3.2), so,
uninlining it, probably, makes most sense. The other 2 functions are
called only ones each and from the interrupt, so, I applied the
"alternative approach" to them - moved above the calling functions.
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

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

Trying to compile drivers/scsi/nsp32.c in 2.6.8-rc1-mm1 using gcc 3.4
results in compile errors starting with the following:

<--  snip  -->

...
  CC      drivers/scsi/nsp32.o
drivers/scsi/nsp32.c: In function `nsp32_prom_start':
drivers/scsi/nsp32.c:270: sorry, unimplemented: inlining failed in call
to 'nsp32_prom_set': function body not available
drivers/scsi/nsp32.c:3348: sorry, unimplemented: called from here
drivers/scsi/nsp32.c:270: sorry, unimplemented: inlining failed in call
to 'nsp32_prom_set': function body not available
drivers/scsi/nsp32.c:3349: sorry, unimplemented: called from here
drivers/scsi/nsp32.c:270: sorry, unimplemented: inlining failed in call
to 'nsp32_prom_set': function body not available
drivers/scsi/nsp32.c:3350: sorry, unimplemented: called from here
drivers/scsi/nsp32.c:270: sorry, unimplemented: inlining failed in call
to 'nsp32_prom_set': function body not available
drivers/scsi/nsp32.c:3351: sorry, unimplemented: called from here
drivers/scsi/nsp32.c:270: sorry, unimplemented: inlining failed in call
to 'nsp32_prom_set': function body not available
drivers/scsi/nsp32.c:3353: sorry, unimplemented: called from here
make[2]: *** [drivers/scsi/nsp32.o] Error 1


<--  snip  -->


The patch below removes the inlines from nsp32_prom_{get,set}.

An alternative approach would be to move the functions above the place
where they are called the first time.
Signed-off-by: Adrian Bunk <bunk@fs.tum.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

Trying to compile drivers/scsi/dc395x.c in 2.6.8-rc1-mm1 using gcc 3.4
results in compile errors starting with the following:

<--  snip  -->

...
  CC      drivers/scsi/dc395x.o
drivers/scsi/dc395x.c: In function `dc395x_handle_interrupt':
drivers/scsi/dc395x.c:388: sorry, unimplemented: inlining failed in call
to 'enable_msgout_abort': function body not available
drivers/scsi/dc395x.c:1740: sorry, unimplemented: called from here
make[2]: *** [drivers/scsi/dc395x.o] Error 1

<--  snip  -->


The patch below moves enable_msgout_abort above the place where it is
called the first time and removes the bogus inline from the prototype of
set_xfer_rate (the function itself isn't marked as inline).

An alternative approach would be to remove the inline from
enable_msgout_abort.
Signed-off-by: Adrian Bunk <bunk@fs.tum.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

[patch] kill local sg_ms_to_jif/sg_jif_to_ms functions and use
        msecs_to_jiffies/jiffies_to_msecs instead
Signed-off-by: Adrian Bunk <bunk@fs.tum.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

  Update version number to 8.00.00b15-k.
Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

  Correct usage of tag-command-queueing methods:

  o Properly call scsi_activate_tcq() rather than
    scsi_adjust_queue_depth().
  o Properly retrieve tag message from command via
    scsi_populate_tag_msg().
Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

  Fix issue where firmware options were not properly being
  set before firmware initialization.
Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

  Fix issue where host-reset would incorrectly return a
  FAILED status.
Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

The patch below does the following small cleanups for the advansys
driver:
- remove obsolete maintainer information
- remove kernel 2.2 code from advansys.h

diffstat output:
 MAINTAINERS             |    7
 drivers/scsi/advansys.c |  342 +++++++---------------------------------
 drivers/scsi/advansys.h |   26 ---
 3 files changed, 60 insertions(+), 315 deletions(-)
Signed-off-by: Adrian Bunk <bunk@fs.tum.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

Trying to compile drivers/scsi/ips.c with gcc 3.4 and
  # define inline         __inline__ __attribute__((always_inline))
results in the following error:

<--  snip  -->

...
  CC      drivers/scsi/ips.o
drivers/scsi/ips.c: In function `ips_eh_abort':
drivers/scsi/ips.c:490: sorry, unimplemented: inlining failed in call to
'ips_removeq_copp': function body not available
drivers/scsi/ips.c:843: sorry, unimplemented: called from here
drivers/scsi/ips.c:488: sorry, unimplemented: inlining failed in call to
'ips_removeq_wait': function body not available
drivers/scsi/ips.c:847: sorry, unimplemented: called from here
make[2]: *** [drivers/scsi/ips.o] Error 1

<--  snip  -->


The patch below removes all inlines from ips.c. As a side effect, this
showed that 3 formerly inlined functions are completely unused which are
also removed in the patch.

An alternative approach to removing the inlines would be to keep all
inlines that are _really_ required and reorder the functions in the file
accordingly.


diffstat output:
 drivers/scsi/ips.c |  130 ++++++---------------------------------------
 1 files changed, 19 insertions(+), 111 deletions(-)
Signed-off-by: Adrian Bunk <bunk@fs.tum.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

From: Mika Kukkonen <mika@osdl.org>

  CC [M]  drivers/block/cciss.o
In file included from drivers/scsi/scsi.h:27,
                 from drivers/block/cciss_scsi.c:31,
                 from drivers/block/cciss.c:188:
include/scsi/scsi_device.h:191: warning: `inline' is not at beginning of declaration

Another case of reordering the 'inline' and function return value.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

If you send an email to linux-megaraid-devel@dell.com, you get an
automated response to send the mail to linux-scsi@vger.kernel.org
instead.

The patch below updates megaraid.c accordingly.
Signed-off-by: Adrian Bunk <bunk@fs.tum.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

> Short: The attached patch fixes the bug-report #2139 from kernel bugzilla.

I wasn't quite convenient with the previous version of the patch, but I
just wanted to get it working at first and get some comments as to whether
the approach (put a workaround vs. blacklisting) is at all valid. Since no
objections arrived, I guess, nobody minds this small trick. So, I am
enclosing a reworked patch, which does essentially the same, but has some
more careful checking, and overall looks saner, than the previous one.
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

drivers/message/fusion/mptbase.c: In function `GetIocFacts':
drivers/message/fusion/mptbase.c:2420: warning: int format, different type arg (arg 4)
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

From: "Moore, Eric Dean" <Emoore@lsil.com>

This driver incorporates the patches from Christoph Hellwig
and Masao Fukuchi.

Changelog of this release:

* Patch provided by  Masao Fukuchi [fukuchi.masao@jp.fujitsu.com]
 - mptctl updates for hot plug

* Patch provided by Christoph Hellwig [hch@infradead.org]
 - mptlan updates for hot plug
 - remove CONFIG_LBD ifdefs from fusion { mptscsih.c}
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

The following patch fixes an oops I was seeing on a machine with
misconfigured scsi cables, but could feasibly happen in other paths.
The oops was occurring because scsi_remove_device was getting
called for a device before scsi_add_device had fully completed.
This resulted in sysfs_remove_link being called with a NULL dentry.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

This restores a fix to the aacraid reset handler.  It makes sure that
there are no active commands on any attached device before exiting with
SUCCESS.
Signed-off-by: Mark Haverkamp <markh@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

Use correct config symbol name for MACH_JAZZ in JAZZ_ESP driver config.
Acked by Ralf.
Signed-off-by: Randy Dunlap <rddunlap@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

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

no function changes, just some more gotos and less nested ifs to make
the code readable.
Signed-off-by: Jes Sorensen <jes@wildopensource.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

qla1280_nvram_config is a huge mess.  Split it up into managable
subroutines and add suport for the ISP1040 to it.  Add missing call
to set the age limit also on 1280/1x160.
Signed-off-by: Jes Sorensen <jes@wildopensource.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

The ISP1040 needs to load firmware by PIO, and while we're at it clean
the convoluted mess of firmware loading up by splitting it into
managable subroutines.
Signed-off-by: Jes Sorensen <jes@wildopensource.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

The code is already messy due to the explicit pci id checks for
1280 vs 1x160, so add some nice helpers for that.
Signed-off-by: Jes Sorensen <jes@wildopensource.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

The old 1020/1040 has some registers where the newer controlers only
have reserved space, add that to qla1280.h
Signed-off-by: Jes Sorensen <jes@wildopensource.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

CAP_DAC_OVERRIDE fails to give search permission in a directory with
mode 0.

Patch acked by Chris Wright <chrisw@osdl.org>
Signed-off-by: Andries Brouwer <aeb@cwi.nl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

There's no reason to keep files that
a) nobody #include's
b) produce #error 
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

For each node there are a defined list of MAX_NR_ZONES zones.  These are
selected as a result of the __GFP_DMA and __GFP_HIGHMEM zone modifier flags
being passed to the memory allocator as part of the GFP mask.  Each node
has a set of zone lists, node_zonelists, which defines the list and order
of zones to scan for each flag combination.  When initialising these lists
we iterate over modifier combinations 0 ..  MAX_NR_ZONES.  However, this is
only correct when there are at most ZONES_SHIFT flags.  If another flag is
introduced zonelists for it would not be initialised.

This patch introduces GFP_ZONETYPES (based on GFP_ZONEMASK) as a bound for
the number of modifier combinations.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Like a few other drivers the vc driver is doing unlocked careless
references to file offsets.  Its got kind of 2.2 locking that hasnt been
updated in the lseek function so someone at least tried. 

Fortunately we have a real lock for this and can just make lseek use that
too.

The 'assume 64bit load is atomic' bug seems to be very widely spread

akpm@osdl.org:

   Move declarations to vt_kern.h
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Here's the accel capabilities patch for radeonfb.  It updates radeonfb to
advertise its acceleration capabilities via fbinfo.flags.  I've tested this
on my box, and it gives me a nice fast console.

defect: "$ fbset -accel 0" doesn't work for radeonfb -- disabling accel
will only work from the kernel command line :-/

Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: David Eger <eger@havoc.gtf.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This has been reported a couple of times and is consistently causing some
folks grief, so Urban, would you mind terribly if i send this patch to at
least clear current bug reports.  If there is additional stuff you want
ontop of this let me know and i can send a follow up patch.

The bug is that at times we haven't completed setting up the smb_ops so we
have a temporary 'null' ops in place until the connection is completely up.
 With this setup it's possible to hit ->readdir() whilst the null ops are
still in place, so we put the process to sleep until the connection setup
is complete and then call the real ->readdir().

This patch addresses the bugzilla report at
http://bugzilla.kernel.org/show_bug.cgi?id=1671Signed-off-by: Zwane Mwaikambo <zwane@linuxpower.ca>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

- Fix the !CONFIG_FRAME_POINTER && !CONFIG_KALLSYMS output formatting.

- Make print_context_stack() static
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

If a fault in the kernel leads to an unexpected protection fault whilst in
a code path which holds mmap_sem we will deadlock in do_page_fault() while
trying to classify the fault.  By carefully testing the source of the fault
we can detect and OOPS on the vast majority of these, greatly enhancing
diagnosis of such bugs.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

We need setup.h for COMMAND_LINE_SIZE.
Signed-off-by: Joseph Fannin <jhf@rivenstone.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

into ppc970.osdl.org:/home/torvalds/v2.6/linux

A microsecond is 1 millionth of a second not
1 thousandth of a second.

Noticed by Markus Mottl <markus@oefai.at>
Signed-off-by: David S. Miller <davem@redhat.com>

Signed-off-by: Adrian Bunk <bunk@fs.tum.de>
Signed-off-by: David S. Miller <davem@redhat.com>

into nuts.davemloft.net:/disk1/BK/net-2.6

into nuts.davemloft.net:/disk1/BK/sparc-2.6