This makes balance_dirty_page() always base its calculations on the
amount of non-highmem memory in the machine, rather than try to base it
on total memory and then falling back on non-highmem memory if the
mapping it was writing wasn't highmem capable.

This not only fixes a situation where two different writers can have
wildly different notions about what is a "balanced" dirty state, but it
also means that people with highmem machines don't run into an OOM
situation when regular memory fills up with dirty pages.

We used to try to handle the latter case by scaling down the dirty_ratio
if the machine had a lot of highmem pages in page_writeback_init(), but
it wasn't aggressive enough for some situations, and since basing the
dirty ratio on highmem memory was broken in the first place, let's just
stop doing so.

(A variation of this theme fixed Justin Piszcz's OOM problem when
copying an 18GB file on a RAID setup).
Acked-by: NNick Piggin <nickpiggin@yahoo.com.au>
Cc: Justin Piszcz <jpiszcz@lucidpixels.com>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Neil Brown <neilb@suse.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Randy Dunlap <rdunlap@xenotime.net>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Adrian Bunk <bunk@stusta.de>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
  [POWERPC] Fix sys_pciconfig_iobase bus matching
  [POWERPC] PS3: add not complete comment to kconfig
  [POWERPC] ps3_free_io_irq: Fix inverted error check
  [POWERPC] PS3: Fix uniprocessor kernel build

A stupid bug has been plaguing the sys_pciconfig_iobase on ppc64. It wasn't
noticed until recently as it seems to not affect G5s but it's been causing
problems running X servers on some other machines recently. The bus number
matching was bogus.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Add a comment to the PS3 config option to inform users that the current
implementation is not yet complete.
Signed-off-by: NGeoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [IPV4]: Fix single-entry /proc/net/fib_trie output.
  [SELINUX]: Fix 2.6.20-rc6 build when no xfrm

 
When main table is just a single leaf this gets printed as belonging to the 
local table in /proc/net/fib_trie. A fix is below.
Signed-off-by: NRobert Olsson <robert.olsson@its.uu.se>
Acked-by: NEric W. Biederman <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch is an incremental fix to the flow_cache_genid
patch for selinux that breaks the build of 2.6.20-rc6 when
xfrm is not configured.
Signed-off-by: NVenkat Yekkirala <vyekkirala@TrustedCS.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Mirror the logic in the sun4u handler, we have to update
both registers even when we branch out to window fault
fixup handling.

The way it works is that if we are in etrap processing a
fault already, g4/g5 holds the original fault information.
If we take a window spill fault while doing etrap, then
we put the window spill fault info into g4/g5 and this is
what the top-level fault handler ends up processing first.

Then we retry the originally faulting instruction, and
process the original fault at that time.

This is all necessary because of how constrained the trap
registers are in these code paths.  These cases trigger
very rarely, so even if there is some performance implication
it's doesn't happen very often.  In fact the rarity is why
it took so long to trigger and find this particular bug.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add an official boot loader ID for Gujin.
Signed-off-by: NH. Peter Anvin <hpa@zytor.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Fix a bug which was introduced when I synced up ocfs2_fs.h with ocfs2-tools.
We can't do u64/u32 in kernel.
Signed-off-by: NMark Fasheh <mark.fasheh@oracle.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

* 'for-linus' of git://www.atmel.no/~hskinnemoen/linux/kernel/avr32:
  [AVR32] Update ATSTK1000 defconfig: Enable macb by default
  [AVR32] Export clear_page symbol

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [NETFILTER]: nf_conntrack_pptp: fix NAT setup of expected GRE connections
  [NETFILTER]: nf_nat_pptp: fix expectation removal
  [NETFILTER]: nf_nat: fix ICMP translation with statically linked conntrack
  [TCP]: Restore SKB socket owner setting in tcp_transmit_skb().
  [AF_PACKET]: Check device down state before hard header callbacks.
  [DECNET]: Handle a failure in neigh_parms_alloc (take 2)
  [BNX2]: Fix 2nd port's MAC address.
  [TCP]: Fix sorting of SACK blocks.
  [AF_PACKET]: Fix BPF handling.
  [IPV4]: Fix the fib trie iterator to work with a single entry routing tables

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  Fix Maple PATA IRQ assignment.
  ahci: use 0x80 as wait stat value instead of 0xff
  sata_via: style clean up, no indirect method call in LLD
  ahci: fix endianness in spurious interrupt message
  libata-sff: Don't call bmdma_stop on non DMA capable controllers
  libata: implement ATA_FLAG_IGN_SIMPLEX and use it in sata_uli
  ahci: improve and limit spurious interrupt messages, take#3
  sata_via: don't diddle with ATA_NIEN in ->freeze
  libata: set_mode, Fix the FIXME
  libata hpt3xn: Hopefully sort out the DPLL logic versus the vendor code
  libata cmd64x: whack into a shape that looks like the documentation

On the Maple board, the AMD8111 IDE is in legacy mode... except that it
appears on IRQ 20 instead of IRQ 15. For drivers/ide this was handled by
the architecture's "pci_get_legacy_ide_irq()" function, but in libata we
just hard-code the numbers 14 and 15.

This patch provides asm-powerpc/libata-portmap.h which maps the IRQ as
appropriate, having added a pci_dev argument to the
ATA_{PRIM,SECOND}ARY_IRQ macros.

There's probably a better way to do this -- especially if we observe
that the _only_ case in which this seemingly-generic
"pci_get_legacy_ide_irq()" function returns anything other than 14 and
15 for primary and secondary respectively is the case of the AMD8111 on
the Maple board -- couldn't we handle that with a special case in the
pata_amd driver, or perhaps with a PCI quirk for Maple to switch it into
native mode during early boot and assign resources properly?
Signed-off-by: NDavid Woodhouse <dwmw2@infradead.org>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Before hardreset, ahci initialized stat part of received FIS area to
0xff to wait for the first D2H Reg FIS which would change the value to
device ready state.  This used to work but now libata considers status
value of 0xff as device not present making this wait prone to failure.

This patch makes ahci use 0x80 for the wait stat value instead of
0xff to fix the above problem.
Signed-off-by: NTejun Heo <htejun@gmail.com>

 drivers/ata/ahci.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Call ata_bmdma_irq_clear() directly instead of through
ap->ops->irq_clear() according to libata style guideline.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Fix endianness in spurious interrupt message.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJeff Garzik <jeff@garzik.org>

Fix race when deleting an EFI variable and issuing another EFI command on
the same variable.  The removal of the variable from the efivars_list
should be done in efivar_delete and not delayed until the kobject release.

Furthermore, remove the item from the list at module unload time, and use
list_for_each_entry_safe() rather than list_for_each_safe() for
readability.

Tested on ia64.
Signed-off-by: NPrarit Bhargava <prarit@redhat.com>
Signed-off-by: NMatt Domsch <Matt_Domsch@dell.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NFrancois Romieu <romieu@fr.zoreil.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Proposed patch to fix #5 in
http://www.isec.pl/vulnerabilities/isec-0017-binfmt_elf.txt
aka
http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2004-1073

To reproduce, do
* grab poc at the end of advisory.
* add line "eph.p_memsz = 4096;" after "eph.p_filesz = 4096;"
  where first "4096" is something equal to or greater than 4096.
* ./poc /usr/bin/sudo && ls -l

Here I get with 2.6.20-rc5:

 -rw------- 1 ad   ad   102400 2007-01-15 19:17 core
 ---s--x--x 2 root root 101820 2007-01-15 19:15 /usr/bin/sudo

Check for MAY_READ like binfmt_misc.c does.
Signed-off-by: NAlexey Dobriyan <adobriyan@openvz.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

raid5_mergeable_bvec tries to ensure that raid5 never sees a read request
that does not fit within just one chunk.  However as we must always accept
a single-page read, that is not always possible.

So when "in_chunk_boundary" fails, it might be unusual, but it is not a
problem and printing a message every time is a bad idea.
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This fixes UML on hosts with non-standard VM splits.  We had changed the
config variable that controls UML behavior on such hosts, but not
propogated the change everywhere.  In particular, the values of STUB_CODE
and STUB_DATA relied on the old variable.

I also reformatted the HOST_VMSPLIT_3G help to make it more standard.

Spotted by uml@flonatel.org.
Signed-off-by: NJeff Dike <jdike@addtoit.com>
Cc: Blaisorblade <blaisorblade@yahoo.it>
Cc: Pravin <shindepravin@gmail.com>
Cc: <uml@flonatel.org>
Cc: <stable@kernel.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

NFS can handle the case where invalidate_inode_pages2_range() fails, so the
premise behind commit 8258d4a5 is now gone.

Remove the WARN_ON_ONCE() which is causing users grief as we can see from
http://bugzilla.kernel.org/show_bug.cgi?id=7826Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

If a GFP_KERNEL allocation is attempted in md while the mddev_lock is held,
it is possible for a deadlock to eventuate.

This happens if the array was marked 'clean', and the memalloc triggers a
write-out to the md device.

For the writeout to succeed, the array must be marked 'dirty', and that
requires getting the mddev_lock.

So, before attempting a GFP_KERNEL allocation while holding the lock, make
sure the array is marked 'dirty' (unless it is currently read-only).
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

nfsd defines a type 'encode_dent_fn' which is much like 'filldir_t' except
that the first pointer is 'struct readdir_cd *' rather than 'void *'.  It
then casts encode_dent_fn points to 'filldir_t' as needed.  This hides any
other type mismatches between the two such as the fact that the 'ino' arg
recently changed from ino_t to u64.

So: get rid of 'encode_dent_fn', get rid of the cast of the function type,
change the first arg of various functions from 'struct readdir_cd *' to
'void *', and live with the fact that we have a little less type checking
on the calling of these functions now.  Less internal (to nfsd) checking
offset by more external checking, which is more important.

Thanks to Gabriel Paubert <paubert@iram.es> for discovering this and
providing an initial patch.
Signed-off-by: NGabriel Paubert <paubert@iram.es>
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Fix a number of kernel-doc entries for header files in include/linux by
making sure they begin with the appropriate '/**' notation and use @var
notation.
Signed-off-by: NRobert P. J. Day <rpjday@mindspring.com>
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

rtc_sysfs_add_device is needed even after dev initialization, so drop __devinit.
Signed-off-by: NMike Frysinger <vapier@gentoo.org>
Acked-by: NAlessandro Zummo <a.zummo@towertech.it>
Cc: <stable@kernel.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Allow noflush suspend/resume of device-mapper device only for the case
where the device size is unchanged.

Otherwise, dm-multipath devices can stall when resumed if noflush was used
when suspending them, all paths have failed and queue_if_no_path is set.

Explanation:
 1. Something is doing fsync() on the block dev,
    holding inode->i_sem
 2. The fsync write is blocked by all-paths-down and queue_if_no_path
 3. Someone requests to suspend the dm device with noflush.
    Pending writes are left in queue.
 4. In the middle of dm_resume(), __bind() tries to get
    inode->i_sem to do __set_size() and waits forever.

'noflush suspend' is a new device-mapper feature introduced in
early 2.6.20. So I hope the fix being included before 2.6.20 is
released.

Example of reproducer:
 1. Create a multipath device by dmsetup
 2. Fail all paths during mkfs
 3. Do dmsetup suspend --noflush and load new map with healthy paths
 4. Do dmsetup resume
Signed-off-by: NJun'ichi Nomura <j-nomura@ce.jp.nec.com>
Acked-by: NAlasdair G Kergon <agk@redhat.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

We weren't properly NULL terminating protocol error strings for our debug
printk resulting in garbage being included in the output when debug was
enabled.
Signed-off-by: NEric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Running dbench multithreaded exposed a race condition where fid structures
were removed while in use.  This patch adds semaphores to meta-data operations
to protect the fid structure.  Some cleanup of error-case handling in the
inode operations is also included.
Signed-off-by: NEric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Update the documentation to cover using Inferno as a server for 9p and to
include information about spfs (a stable single-threaded stand-alone 9p
server).
Signed-off-by: NEric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

9p doesn't handle renames between directories -- however, we were returning
EPERM instead of EXDEV when we detected this case.
Signed-off-by: NEric Van Hensbergren <ericvh@gmail.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

There is a simple logic error in init_v9fs - the return code checks are
reversed.  This patch fixes the return code and adds some messages to prevent
module initialization from failing silently.
Signed-off-by: NEric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

In most cases we check the size of the bitmap file before reading data from
it.  However when reading the superblock, we always read the first PAGE_SIZE
bytes, which might not always be appropriate.  So limit that read to the size
of the file if appropriate.

Also, we get the count of available bytes wrong in one place, so that too can
read past the end of the file.

Cc: "yang yin" <yinyang801120@gmail.com>
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Now that we sometimes step the array events count backwards (when
transitioning dirty->clean where nothing else interesting has happened - so
that we don't need to write to spares all the time), it is possible for the
event count to return to zero, which is potentially confusing and triggers and
MD_BUG.

We could possibly remove the MD_BUG, but is just as easy, and probably safer,
to make sure we never return to zero.
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

When 'repair' finds a block that is different one the various parts of the
mirror.  it is meant to write a chosen good version to the others.  However it
currently writes out the original data to each.  The memcpy to make all the
data the same is missing.
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

NFS V3 (and V4) support exclusive create by passing a 'cookie' which can get
stored with the file.  If the file exists but has exactly the right cookie
stored, then we assume this is a retransmit and the exclusive create was
successful.

The cookie is 64bits and is traditionally stored in the mtime and atime
fields.  This causes a problem with Solaris7 as negative mtime or atime
confuse it.  So we moved two bits into the mode word instead.

But inherited ACLs sometimes overwrite the mode word on create, so this is a
problem.

So we give up and just store 62 of the 64 bits and assume that is close
enough.
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

A couple of the warnings will be followed by an Oops if they ever fire, so may
as well be BUG_ON.  Another isn't obviously fatal but has never been known to
fire, so make it a WARN_ON.

Cc: Adrian Bunk <bunk@stusta.de>
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

NFSd assumes that largest number of pages that will be needed for a
request+response is 2+N where N pages is the size of the largest permitted
read/write request.  The '2' are 1 for the non-data part of the request, and 1
for the non-data part of the reply.

However, when a read request is not page-aligned, and we choose to use
->sendfile to send it directly from the page cache, we may need N+1 pages to
hold the whole reply.  This can overflow and array and cause an Oops.

This patch increases size of the array for holding pages by one and makes sure
that entry is NULL when it is not in use.
Signed-off-by: NNeil Brown <neilb@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>