I think I got the cause for the Oops observed in
http://www.mail-archive.com/dccp@vger.kernel.org/msg00578.html

The problem is always with applications listening on PF_INET6 sockets. Apart
from the mentioned oops, I observed another one one, triggered at irregular
intervals via timer interrupt:

    run_timer_softirq -> dccp_keepalive_timer
                      -> inet_csk_reqsk_queue_prune
                      -> reqsk_free
                      -> dccp_v6_reqsk_destructor

The latter function is the problem and is also the last function to be called
in said kernel panic.

In any case, there is a real problem with allocating the right request_sock
which is what this patch tackles.

It fixes the following problem:
 - application listens on PF_INET6
 - DCCPv4 packet comes in, is handed over to dccp_v4_do_rcv, from there
   to dccp_v4_conn_request

Now: socket is PF_INET6, packet is IPv4. The following code then furnishes the
connection with IPv6 - request_sock operations:

   req = reqsk_alloc(sk->sk_prot->rsk_prot);

The first problem is that all further incoming packets will get a Reset since
the connection can not be looked up.

The second problem is worse:
 --> reqsk_alloc is called instead of inet6_reqsk_alloc
 --> consequently inet6_rsk_offset is never set (dangling pointer)
 --> the request_sock_ops are nevertheless still dccp6_request_ops
 --> destructor is called via reqsk_free
 --> dccp_v6_reqsk_destructor tries to free random memory location (inet6_rsk_offset not set)
 --> panic

I have tested this for a while, DCCP sockets are now handled correctly in all
three scenarios (v4/v6 only/v4-mapped).

Commiter note: I've added the dccp_request_sock_ops forward declaration to keep
               the tree building and to reduce the size of the patch for 2.6.19,
               later I'll move the functions to the top of the affected source
               code to match what we have in the TCP counterpart, where this
               problem hasn't existed in the first place, dumb me not to have
               done the same thing on DCCP land 8)
Signed-off-by: NGerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>

* 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6:
  [PATCH] x86-64: increase PHB1 split transaction timeout
  [PATCH] x86-64: Fix C3 timer test

This patch increases the timeout for PCI split transactions on PHB1 on
the first Calgary to work around an issue with the aic94xx
adapter. Fixes kernel.org bugzilla #7180
(http://bugzilla.kernel.org/show_bug.cgi?id=7180)

Based on excellent debugging and a patch by Darrick J. Wong
<djwong@us.ibm.com>
Signed-off-by: NMuli Ben-Yehuda <muli@il.ibm.com>
Signed-off-by: NJon Mason <jdmason@kudzu.us>
Signed-off-by: NAndi Kleen <ak@suse.de>
Acked-by: NDarrick J. Wong <djwong@us.ibm.com>

There was a typo in the C3 latency test to decide of the TSC
should be used or not. It used the C2 latency threshold, not the
C3 one. Fix that.

This should fix the time on various dual core laptops.
Signed-off-by: NAndi Kleen <ak@suse.de>

The current implementation uses a sequence of a cacheflush and a copy.
This is racy in case of a multithreaded debuggee and renders GDB
virtually unusable.

Aside this fixes a performance hog rendering access to /proc/cmdline very
slow and resulting in a enough cache stalls for the 34K AP/SP programming
model to make the bare metal code on the non-Linux VPE miss RT deadlines.

The main part of this patch was originally written by Ralf Baechle;
Atushi Nemoto did the the debugging.
Signed-off-by: NAtsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

Noticed by Samium Gromoff but his patch got stale in flight ...
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  [PATCH] libata-sff: Allow for wacky systems
  [PATCH] ahci: readability tweak
  [PATCH] libata: typo fix
  [PATCH] ATA must depend on BLOCK
  [PATCH] libata: use correct map_db values for ICH8

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6: (22 commits)
  [PATCH] ibmveth: Fix index increment calculation
  [PATCH] Fix timer race
  [PATCH] Remove useless comment from sb1250
  [PATCH] ucc_geth: changes to ucc_geth driver as a result of qe_lib changes and bugfixes
  [PATCH] sky2: 88E803X transmit lockup
  [PATCH] e1000: Reset all functions after a PCI error
  [PATCH] WAN/pc300: handle, propagate minor errors
  [PATCH] Update smc91x driver with ARM Versatile board info
  [PATCH] wireless: WE-20 compatibility for ESSID and NICKN ioctls
  [PATCH] zd1211rw: fix build-break caused by association race fix
  [PATCH] sotftmac: fix a slab corruption in WEP restricted key association
  [PATCH] airo: check if need to freeze
  [PATCH] wireless: More WE-21 potential overflows...
  [PATCH] zd1201: Possible NULL dereference
  [PATCH] orinoco: fix WE-21 buffer overflow
  [PATCH] airo.c: check returned values
  [PATCH] bcm43xx-softmac: Fix system hang for x86-64 with >1GB RAM
  [PATCH] bcm43xx-softmac: check returned value from pci_enable_device
  [PATCH] softmac: Fix WX and association related races
  [PATCH] bcm43xx: fix race condition in periodic work handler
  ...

* master.kernel.org:/pub/scm/linux/kernel/git/davej/agpgart:
  [AGPGART] uninorth: Add module param 'aperture' for aperture size

* 'for-linus' of git://one.firstfloor.org/home/andi/git/linux-2.6:
  [PATCH] x86-64: Revert timer routing behaviour back to 2.6.16 state
  [PATCH] x86-64: Overlapping program headers in physical addr space fix
  [PATCH] x86-64: Put more than one cpu in TARGET_CPUS
  [PATCH] x86: Revert new unwind kernel stack termination
  [PATCH] x86-64: Use irq_domain in ioapic_retrigger_irq
  [PATCH] i386: Disable nmi watchdog on all ThinkPads
  [PATCH] x86-64: Revert interrupt backlink changes
  [PATCH] x86-64: Fix ENOSYS in system call tracing
  [PATCH] i386: Fix fake return address
  [PATCH] x86-64: x86_64 add NX mask for PTE entry
  [PATCH] x86-64: Speed up dwarf2 unwinder
  [PATCH] x86: Use -maccumulate-outgoing-args
  [PATCH] x86-64: fix page align in e820 allocator
  [PATCH] x86-64: Fix for arch/x86_64/pci/Makefile CFLAGS
  [PATCH] i386: fix .cfi_signal_frame copy-n-paste error
  [PATCH] x86-64: typo in __assign_irq_vector when updating pos for vector and offset
  [PATCH] x86-64: x86_64 hot-add memory srat.c fix
  [PATCH] i386: Update defconfig
  [PATCH] x86-64: Update defconfig

If someone has renamed a directory on the server, triggering the d_move
code in d_materialise_unique(), then we need to invalidate the cached
directory information in the source parent directory.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Maneesh Soni <maneesh@in.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: Neil Brown <neilb@cse.unsw.edu.au>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

If the caller tries to instantiate a directory using an inode that already
has a dentry alias, then we attempt to rename the existing dentry instead
of instantiating a new one.  Fail with an ELOOP error if the rename would
affect one of our parent directories.

This behaviour is needed in order to avoid issues such as

  http://bugzilla.kernel.org/show_bug.cgi?id=7178Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Maneesh Soni <maneesh@in.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: Neil Brown <neilb@cse.unsw.edu.au>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

CCISS was producing warnings about shifts being greater than the size of
the type and pointers being of incompatible type.  Turns out this is
because it's calling do_div on a 32-bit quantity.  Upon further
investigation, the sector_t total_size is being assigned to an int, and
then we're calling do_div on that int.  Obviously, sector_div is called for
here, and I took the chance to refactor the code a little.
Signed-off-by: NMatthew Wilcox <matthew@wil.cx>
Acked-by: NMike Miller <mike.miller@hp.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Just like everyone else.
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Cc: Markus Lidel <Markus.Lidel@shadowconnect.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Mistyped an ifdef CONFIG_CPUSETS - fixed.

I doubt that anyone ever noticed.  The impact of this typo was
that if someone:
 1) was using MPOL_BIND to force off node allocations
 2) while using cpusets to constrain memory placement
 3) when that cpuset was migrating that jobs memory
 4) while the tasks in that job were actively forking
then there was a rare chance that future allocations using
that MPOL_BIND policy would be node local, not off node.
Signed-off-by: NPaul Jackson <pj@sgi.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The zonelist may contain zones of nodes that have not been bootstrapped and
we will oops if we try to allocate from those zones.  So check if the node
information for the slab and the node have been setup before attempting an
allocation.  If it has not been setup then skip that zone.

Usually we will not encounter this situation since the slab bootstrap code
avoids falling back before we have setup the respective nodes but we seem
to have a special needs for pppc.
Signed-off-by: NChristoph Lameter <clameter@sgi.com>
Acked-by: NAndy Whitcroft <apw@shadowen.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Mike Kravetz <kravetz@us.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: NMel Gorman <mel@csn.ul.ie>
Acked-by: NWill Schmidt <will_schmidt@vnet.ibm.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Reintroduce NODES_SPAN_OTHER_NODES for powerpc

Revert "[PATCH] Remove SPAN_OTHER_NODES config definition"
    This reverts commit f62859bb.
Revert "[PATCH] mm: remove arch independent NODES_SPAN_OTHER_NODES"
    This reverts commit a94b3ab7.

Also update the comments to indicate that this is still required
and where its used.
Signed-off-by: NAndy Whitcroft <apw@shadowen.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Mike Kravetz <kravetz@us.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: NMel Gorman <mel@csn.ul.ie>
Acked-by: NWill Schmidt <will_schmidt@vnet.ibm.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: <stable@kernel.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The recent commit 751ae21c introduced a bug
in the producer/consumer index calculation in the ibmveth driver -
incautious use of the post-increment ++ operator resulted in an increment
being immediately reverted.  This patch corrects the logic.

Without this patch, the driver oopses almost immediately after activation
on at least some machines.
Signed-off-by: NDavid Gibson <dwg@au1.ibm.com>
Acked-by: NSantiago Leon <santil@us.ibm.com>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

We seem to have lost the declaration of pci_get_device_reverse(), if we ever
had one.

Add a CONFIG_PCI=0 stub too.
Acked-by: NAlan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

I have an acpi_pm that goes backwards, but it's not intel.  I tested the
verified read and my acpi_pm started to function properly.  So I added it
to the greylist.  I'm assuming that's the right spot.

I also added an unlikely() to the while, cause it seems appropriate.
Signed-off-by: NDaniel Walker <dwalker@mvista.com>
Acked-by: NJohn Stultz <johnstul@us.ibm.com>
Acked-by: NOGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And a couple of bug fixes found by sparse.
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Includes a couple of bugfixes found by sparse.
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

..  so that you can use bitmaps with 32bit userspace on a 64 bit kernel.
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Two less-used md personalities have bugs in the calculation of ->degraded (the
extent to which the array is degraded).
Signed-off-by: NNeil Brown <neilb@suse.de>
Cc: <stable@kernel.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

The change from __setup() to module_param_named() requires users to prefix
the option with "generic.".

This patch re-adds the __setup() additionally to the module_param_named().

Usually it would make sense getting rid of such an obsolete __setup() at
some time, but considering that drivers/ide/ is slowly approaching a RIP
status it's already implicitely scheduled for removal.

This patch fixes kernel Bugzilla #7353.
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Acked-by: NAlan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Commits 881a8c12 and
efe1ec27 corrects pci device matching in
only one way; it no longer oopses/crashes, despite hotplug is not solved
in these changes.

Whenever pci_find_device -> pci_get_device change is performed, also
pci_dev_get and pci_dev_put should be in most cases called to properly
handle hotplug.  This patch does exactly this thing -- increase refcount
to let kernel know, that we are using this piece of HW just now.

It affects moxa and rio char drivers.

Cc: <R.E.Wolff@BitWizard.nl>
Acked-by: NAmit Gud <gud@eth.net>
Acked-by: NGreg Kroah-Hartman <gregkh@suse.de>
Acked-by: NAlan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

There are some Linux supported platforms that simply cannot hit the low
I/O addresses used by ATA legacy mode PCI mappings. These platforms have
a window for PCI space that is fixed by the board logic and doesn't
include the neccessary locations.

Provide a config option so that such platforms faced with a controller
that they cannot support simply error it and punt
Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Typo fix in commment.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Fix the following compile error with CONFIG_ATA=y, CONFIG_BLOCK=n:

...
  CC      drivers/ata/libata-scsi.o
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c: In function ‘ata_scsi_dev_config’:
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c:791: warning: implicit declaration of function ‘blk_queue_max_sectors’
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c:799: error: ‘request_queue_t’ undeclared (first use in this function)
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c:799: error: (Each undeclared identifier is reported only once
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c:799: error: for each function it appears in.)
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c:799: error: ‘q’ undeclared (first use in this function)
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c:800: warning: implicit declaration of function ‘blk_queue_max_hw_segments’
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c: In function ‘ata_scsi_slave_config’:
/home/bunk/linux/kernel-2.6/git/linux-2.6/drivers/ata/libata-scsi.c:831:
warning: implicit declaration of function ‘blk_queue_max_phys_segments’
make[3]: *** [drivers/ata/libata-scsi.o] Error 1

Bug report by Jesper Juhl.
Signed-off-by: NAdrian Bunk <bunk@stusta.de>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Use valid values for ICH8 map_db.  With the old values, when the
controller was in Native mode, and SCC was 1 (drives configured for
IDE), any drive plugged into a slave port was not recognized.  For
Combined Mode (and SCC is still 1), 2 is a value value for MAP.map_value,
and needs to be recognized.
Signed-off-by: NKristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

On Thu, Oct 12, 2006 at 06:22:14PM +1000, David Gibson wrote:
> Your recent ibmveth commit, 751ae21c
> ("fix int rollover panic"), causes a rapid oops on my test machine
> (POWER5 LPAR).
>
> I've bisected it down to that commit, but am still investigating the
> cause of the crash itself.

Found the problem, I believe: an object lesson in the need for great
caution using ++.

[...]
@@ -213,6 +213,7 @@ static void ibmveth_replenish_buffer_poo
 		}

 		free_index = pool->consumer_index++ % pool->size;
+		pool->consumer_index = free_index;
 		index = pool->free_map[free_index];

 		ibmveth_assert(index != IBM_VETH_INVALID_MAP);

Since the ++ is used as post-increment, the increment is not included
in free_index, and so the added line effectively reverts the
increment.  The produced_index side has an analagous bug.

The following change corrects this:

The recent commit 751ae21c introduced
a bug in the producer/consumer index calculation in the ibmveth driver
- incautious use of the post-increment ++ operator resulted in an
increment being immediately reverted.  This patch corrects the logic.

Without this patch, the driver oopses almost immediately after
activation on at least some machines.
Signed-off-by: NDavid Gibson <dwg@au1.ibm.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Merge branch 'upstream-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-2.6 into upstream-fixes

When closing the driver or reinitializing the hardware there is the
usual del_timer() race condition that exists when timers re-add
themselves.  Fix by conversion to del_timer_sync().
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Signed-off-by: NDave Jones <davej@redhat.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Merge branch 'we21-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-2.6 into tmp

changes due to qe_lib changes include:

o removed inclusion of platform header file
o removed platform_device code, replaced with of_device
o removed typedefs
o uint -> u32 conversions
o removed following defines:
  QE_SIZEOF_BD, BD_BUFFER_ARG, BD_BUFFER_CLEAR, BD_BUFFER,
  BD_STATUS_AND_LENGTH_SET, BD_STATUS_AND_LENGTH, and BD_BUFFER_SET
  because they hid sizeof/in_be32/out_be32 operations from the reader.
o removed irrelevant comments, added others to resemble removed BD_ defines
o const'd and uncasted all get_property() assignments

bugfixes, courtesy of Scott Wood, include:

- Read phy_address as a u32, not u8.
- Match on type == "network" as well as compatible == "ucc_geth", as
  device_is_compatible() will only compare up to the length of the
  test string, allowing "ucc_geth_phy" to match as well.
- fixes the MAC setting code in ucc_geth.c.  The old code was overwriting and dereferencing random stack contents.
Signed-off-by: NLi Yang <leoli@freescale.com>
Signed-off-by: NKim Phillips <kim.phillips@freescale.com>
Signed-off-by: NScott Wood <scottwood@freescale.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

The reason sky2 driver was locking up on transmit on the Yukon-FE chipset
is that it was misconfiguring the internal RAM buffer so the transmitter
and receiver were sharing the same space.

The code assumed there was 16K of RAM on Yukon-FE (taken from vendor driver
sk98lin which is even more f*cked up on this). Then it assigned based on that.
The giveaway was that the registers would only hold 9bits so both RX/TX
had 0..1ff for space. It is a wonder it worked at all!

This patch addresses this, and fixes an easily reproducible hang on Transmit.
Only the Yukon-FE chip is Marvell 88E803X (10/100 only) are affected.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

During the handling of the PCI error recovery sequence, the current e1000
driver erroneously blocks a device reset for any but the first PCI
function.  It shouldn't -- this is a cut-n-paste error from a different
driver (which tolerated only one hardware reset per hardware card).
Signed-off-by: NLinas Vepstas <linas@austin.ibm.com>
Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
Acked-by: NAuke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>