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

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

Westwood code cleanup;
	* use const.
	* avoid needless paren's and returns
	* inline acked_count (called once)
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Enhancement to tcp diagnostics used by ss command.
* use jiffies_to_usecs/msec instead of hardcode math
* report bandwidth for Westwood
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The choice of congestion algorithm needs to be made when connection
is setup to avoid problems when the sysctl values change later and the
necessary data hasn't been collected.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>'

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

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

be before the end of the vmdb, just make sure it is not out of bounds.
If it is abort.  If it is not just continue happily.
Signed-off-by: Anton Altaparmakov <aia21@cantab.net>

The reordering of the drivers in drivers/video/Makefile introduced a
stray 0 into the pxafb.
Signed-off-by: Ian Campbell <icampbell@arcom.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The "DART" iommu used on Apple U3 chipset will appear into 3rd party
products soon, thus the code shouldn't be named "pmac_*" anymore.

Also, the explicit config option is no longer needed, there is no
reason to build a PowerMac kernel without the iommu support, and
it can always be disabled at runtime with a cmdline option for
testing or debugging.

This patch do the appropriate renaming and makes the config option
implicit & selected when pmac support is included.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

include/asm-ppc64/systemcfg.h defines a structure that currently gets
exposed to userland via /proc, and which I intend to deprecate in the
near future once I have better alternatives available.

In the meantime, this patch removes a bunch of stuff from this file that
were only defined for non-__KERNEL__, like an inline function for
getting to that struture via /proc, and some CPU & platform type
definitions that were duplicates of the ones in asm-ppc64/processor.h.
These things have nothing to do in a kernel header.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

A small spelling error slipped into a new ppc_md. platform call I added
in my last patch. Not very problematic but worth fixing before any code
using the new call gets pushed.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

There appears to be one case missing from the wait_event() family - the
uninterruptible timeout wait.  The following patch adds this.

This wait is particularly useful when (eg) you wish to pass work off to an
interrupt handler to perform, but also want to know if the hardware has
decided to go gaga under you.  You don't want to sit around waiting for
something that'll never happen - you want to go and wack the gremlin which
caused the failure and retry.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This should allow sys_rtas to work again on PPC64 pSeries.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This fixes the bug that made the card DMA lots of data to random memory
locations, causing lockups and instant reboots. 

The problem was that the yoffset variable got modified inside the loop,
but the logic in the switch statement was meant to work on the initial
value of the yoffset variable.

(Bug fix extracted from
http://marc.theaimsgroup.com/?l=linux-kernel&m=109532814823565&w=2)
Signed-off-by: Peter Osterlund <petero2@telia.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Agh !

I had two patches, a broken one and a good one and ... of course I sent
the broken one, sorry :(

This fixes the broken fix.
Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Prevent accidently referencing f->fn_alias list
head as a real fib_alias object.
Signed-off-by: David S. Miller <davem@davemloft.net>

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

to be linked unconditionally.

The current 32 bits translation of the SI_TIMER is wrong on ppc64, causing
the tst-timer4 testcase of glibc to fail in 32 bits. This patch fixes it.
Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The "blocked_list" used for POSIX deadlock detection should only contain
POSIX locks, or posix_locks_deadlock() can get confused.

As I've noted before (but never posted a patch), we should probably stop
checking for deadlock.  POSIX says it's optional for an implementation
to check for deadlock, and since the change to match on tgid, we can
report a deadlock when none exists.

[ Further explanation for Linus: POSIX locks and BSD locks don't block
  each other, so checking that the blocker is a POSIX lock also implies
  that the waiter is a POSIX lock. ]

We need to convert from bytes to blocks, and we also need to
handle the fact that old smbd's will round up the disk usage
bytecount to even multiples of 1MB, and gracefully fall back 
on an estimate based on the file size instead in that case.

Acked by Samuel Thibault who noted the problem in the first
place, and Jeremy Allison wrt smbd behavior.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

into kanga.twiddle.home:/home/rth/work/linux/axp-2.6

Signed-off-by: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>

Signed-off-by: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>

Because in6_dev_get() fails during device deletion,
we failed to leave solicited-node multicast address.
Signed-off-by: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>

Use __ipv6_dev_mc_dec() instead where the caller knows idev.
Signed-off-by: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>

dev is unused in the function.
Signed-off-by: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>

Case 1: if net.ipv6.conf.IFACE.forwarding has been set to 1,
we multiply join routers' multicast/anycast addresses 
by setting net.ipv6.conf.all.forwarding to 1.

Noticed by Andre Tomt <andre@tomt.net>.

Case 2: if we change net.ipv6.conf.FOO.forwarding from 1 to 2,
we multiply join routers' multicast/anycast addresses.
Signed-off-by: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>

We failed to leave all-routers multicast address.
Signed-off-by: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>

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

The recent addition of linux/bitops.h inside linux/kernel.h has
created a circular dependency:

asm/system.h -> linux/kernel.h -> linux/bitops.h -> asm/bitops.h
 -> asm/system.h

which results in an unbuildable kernel.  We appear not to need
linux/kernel.h in asm/system.h, but do need linux/linkage.h, so
make this the case.

Also, make neigh_hash_alloc() use GFP_ATOMIC.
Signed-off-by: David S. Miller <davem@davemloft.net>

Based almost entirely upon work by Tim Gardner
(timg@tpi.com) and Harald Welte (laforge@gnumonks.org)
Signed-off-by: David S. Miller <davem@davemloft.net>

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

into kernel.bkbits.net:/home/davem/sparc-2.6

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>