From: Christoph Hellwig <hch@lst.de>

Reduce ifdef clutter in arch/ppc64/kernel/sysfs.c
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Steve Munroe points out that ppc64 glibc builds stubs for a number of 32bit
only syscalls.  While none of them exist in the kernel syscall table, their
existence in unistd.h means glibc still tries to use them then falls back
onto the 64bit safe versions.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

PCI_GET_DN() doesnt check to see if ->sysdata has been initialised
correctly - we should instead use pci_device_to_OF_node.  Leave
PCI_GET_DN() in the one performance critical case (iommu table lookup in
pci DMA functions).  In this case ->sysdata is guaranteed to have been
initialised by the iommu setup code.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Use printk_ratelimit() in rtc code to avoid flooding the kernel log buffer
with errors.  Also use rtas_get_error_log_max() instead of duplicating it
in __fetch_rtas_last_error.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove KRANGE_{START,END} macros from ppc64 code.  These were not used
anywhere.  Further KRANGE_END was misleading, since it implied a limit on
the linear mapping range based on the pagetable structure, whereas in fact
the linear mapping does not use a (Linux) pagetable at all.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Here's a patch to extend the current Linux on Power support for PURR to
legacy IBM iSeries servers (pre-Power5 processor models).  This patch
enables the reporting of timebase metrics to reflect physical processor
utilization in a system running multiple logical partitions which share the
same physical processors.

The patch simply uses existing user interfaces for Linux IBM Power5 based
servers to report data already collected by the hypervisor.  The values
reported with each call are running values in units of the system timebase.
 The calculation of physical processor utilization results from two samples
(purr1 and purr2) differing by a know interval (time) such that:

    physical utilization = (purr2 - purr1) / (time * number of procs *
timebase)

where the number of procs and timebase can be obtained from /proc/cpuinfo.

Applications have been written to the interface already defined and these
applications have value back on the legacy iSeries models.

Signed-off by: Jeff Scheel (scheel at vnet.ibm.com)
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>

Motorola Sandpoint builds broke recently, as part of the pci_find_device
-> pci_get_device change.  The following is the trivial fix.
Signed-off-by: Tom Rini <trini@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

When new_inode failed, shmem_get_inode forgot to restore free_inodes.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

In the case of i386 CONFIG_DISCONTIGMEM CONFIG_HIGHMEM without highmem,
highmem_start_page was wrongly initialized (from a NULL zone_mem_map),
causing __change_page_attr to BUG on boot.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

without patch:
      pages_min   pages_low   pages_high
dma        4          8          12
normal   234        468         702
high     128        256         384

with patch:
      pages_min   pages_low   pages_high
dma       17         21          25
normal   939       1173        1408
high     128        160         192

without patch:
                             | GFP_KERNEL        | GFP_ATOMIC
allocate immediately         |   9 dma, 469 norm |  9 dma, 469 norm
allocate after waking kswapd |   5 dma, 234 norm |  3 dma,  88 norm
allocate after synch reclaim |   5 dma, 234 norm |  n/a

with patch:
                             | GFP_KERNEL         | GFP_ATOMIC
allocate immediately         |  22 dma, 1174 norm | 22 dma, 1174 norm
allocate after waking kswapd |  18 dma,  940 norm |  6 dma,  440 norm
allocate after synch reclaim |  18 dma,  940 norm |  n/a

So the buffer between GFP_KERNEL and GFP_ATOMIC allocations is:

2.6.8      | 465 dma, 117 norm, 582 tot = 2328K
2.6.10-rc  |   2 dma, 146 norm, 148 tot =  592K
patch      |  12 dma, 500 norm, 512 tot = 2048K

Which is getting pretty good.

kswap starts at:
2.6.8     477 dma, 496 norm, 973 total
2.6.10-rc   8 dma, 468 norm, 476 total
patched    17 dma, 939 norm, 956 total

So in terms of total pages, that's looking similar to 2.6.8.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The current kernel oopses on x86-64 when gdb steps into the vsyscall page. 
This patch fixes it.

I also removed the bogus NULL checks of *_offset and replaced them with
proper _none checks.  I made them BUGs because vsyscall pages should be
always mapped.
Signed-off-by: Andi Kleen <ak@suse.de>
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

OK, let's make it the same as IPv4.  This is still broken
as the calculation is wrong.  However, to fix it properly
we need to store the MTUs inside the dst's.  I'll get
around to that sooner or later.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

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

Patch from Ben Dooks

S3C2410/S3C2440 new serial driver

this patch includes the following

- clock selection using clock core
- handling for both 2410 and 2440 uarts
- uart suspend/resume support
- Dimitry Andric's fix for un-initialised spinlocks
- Herbert Poetzl's fixes for the following
-> break character recognition
-> Magic-SYSRQ handling
-> Uart software flow control (IrDA console)

Updated to include rmk's comments from Patch #2234/1
and fix a few checks for serial clocks, as well as
merging fixes from rmk that had taken time to go
through the biteepr to release.

Signed-off-by: Ben Dooks
Signed-off-by: Russell King

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

Patch from Ben Dooks

Move the configuration for the UART to use for the low-level
messages generated by the uncompressor out of the kernel debug
section, which means the system builds properly without having
kernel debug enabled.

Also remove the use of including <config/xxx.h> headers,
and the default configuration updates

Signed-off-by: Ben Dooks
Signed-off-by: Russell King

Patch from Ben Dooks

Ensure the i2c platform device has the correct name if the
system is an s3c2440.

Signed-off-by: Ben Dooks
Signed-off-by: Russell King

Patch from Ben Dooks

Move the pm code to use the machine's init_machine
code, instead of an late_initcall(). This change is
as-per the documentation, and means we do not need to
check for machine type in the init code.

Thanks to Dimitry Andric for pointing this out

Signed-off-by: Ben Dooks
Signed-off-by: Russell King

Patch from Ben Dooks

Updates for the Simtec BAST (EB2410ITX) default configuration
to fix the following:

- add mtd support for NAND and NOR
- remove PC style parallel port (does not build)
- use S3C2410 core RTC for time
- remove non-bast S3C2410 machines

Signed-off-by: Ben Dooks
Signed-off-by: Russell King

Patch from Ben Dooks

updates to the default configuration to change the
following:

- use S3C2410 internal rtc instead of PC style rtc
- add mtd support for nand
- add mtd support for nor (and bast nor mapping)

Signed-off-by: Ben Dooks
Signed-off-by: Russell King

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

Add support for 5753 chips which is mostly just adding
in the appropriate PCI ids and recognizing that these
chips do not use GPIO2 for Vaux switching.

Also do not set DMA read water mark on PCI Express.
In such configurations these bits are reserved.
Signed-off-by: David S. Miller <davem@davemloft.net>

Just tried the new toy. It works.

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

Signed-off-by: Phil Oester <kernel@linuxace.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

into coreworks.de:/home/kaber/src/nf/nf-2.6

into coreworks.de:/home/kaber/src/nf/nf-2.6

into coreworks.de:/home/kaber/src/nf/nf-2.6

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

Found running netfilter code under valgrind: we return some
uninitialized stack to userspace (needs root).  Almost certainly
harmless, but this suppressed the valgrind complaint, and
doesn't hurt.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

The fix to find_appropriate_src left a stupid test, which has no
effect but is incoherent at best.  It covers a corner case: if a
previous connection from this source was mapped onto a different IP
address (because it was explicitly told to), should the next one be
mapped that was as well, if no explicit rule says to remap the src IP?

I think the answer is yes: unless the user explicitly tells us to map
into a particular range, we should follow the Kegel draft.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

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

While doing the 2.4 version of the netlink patch, I noticed a bug
in my mc_list code.  First of all it wasn't holding the table lock
when adding nodes to the list.  It also didn't take the node off
the list when multicast is switched off.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

Make sure loopback_dev, if up, has the ipv6 bits
for it setup before the addrconf netdev notifier
is registered.
Signed-off-by: David S. Miller <davem@davemloft.net>

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

The patch to change traps and interrupts to the fastcall convention
missed the machine check handlers.
Signed-off-by: Brian Gerst <bgerst@didntduck.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

asked for it.