]> git.proxmox.com Git - mirror_ubuntu-eoan-kernel.git/log
mirror_ubuntu-eoan-kernel.git
16 years ago[GFS2] Fix write alloc required shortcut calculation
Steven Whitehouse [Wed, 16 Jan 2008 14:24:05 +0000 (14:24 +0000)]
[GFS2] Fix write alloc required shortcut calculation

The comparison was being made against the wrong quantity.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] gfs2_alloc_required performance
Bob Peterson [Fri, 11 Jan 2008 19:44:50 +0000 (13:44 -0600)]
[GFS2] gfs2_alloc_required performance

This is a small I/O performance enhancement to gfs2.  (Actually, it is a rework of
an earlier version I got wrong).  The idea here is to check if the write extends
past the last block in the file.  If so, the function can save itself a lot of
time and trouble because it knows an allocate will be required.  Benchmarks like
iozone should see better performance.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove unneeded i_spin
Bob Peterson [Fri, 11 Jan 2008 19:31:12 +0000 (13:31 -0600)]
[GFS2] Remove unneeded i_spin

This patch removes a vestigial variable "i_spin" from the gfs2_inode
structure.  This not only saves us memory (>300000 of these in memory
for the oom test) it also saves us time because we don't have to
spend time initializing it (i.e. slightly better performance).

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Reduce inode size by moving i_alloc out of line
Steven Whitehouse [Thu, 10 Jan 2008 15:18:55 +0000 (15:18 +0000)]
[GFS2] Reduce inode size by moving i_alloc out of line

It is possible to reduce the size of GFS2 inodes by taking the i_alloc
structure out of the gfs2_inode. This patch allocates the i_alloc
structure whenever its needed, and frees it afterward. This decreases
the amount of low memory we use at the expense of requiring a memory
allocation for each page or partial page that we write. A quick test
with postmark shows that the overhead is not measurable and I also note
that OCFS2 use the same approach.

In the future I'd like to solve the problem by shrinking down the size
of the members of the i_alloc structure, but for now, this reduces the
immediate problem of using too much low-memory on x86 and doesn't add
too much overhead.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Fix assert in log code
Steven Whitehouse [Thu, 10 Jan 2008 14:49:43 +0000 (14:49 +0000)]
[GFS2] Fix assert in log code

Although the values were all being calculated correctly, there was a
race in the assert due to the way it was using atomic variables. This
changes the value we assert on so that we get the same effect by testing
a different variable. This prevents the assert triggering when it shouldn't.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Fix problems relating to execution of files on GFS2
Steven Whitehouse [Tue, 8 Jan 2008 08:14:30 +0000 (08:14 +0000)]
[GFS2] Fix problems relating to execution of files on GFS2

This patch fixes a couple of problems which affected the execution of files
on GFS2. The first is that there was a corner case where inodes were not
always uptodate at the point at which permissions checks were being carried
out, this was resulting in refusal of execute permission, but only on the
first lookup, subsequent requests worked correctly. The second was a problem
relating to incorrect updating of file sizes which was introduced with the
write_begin/end code for GFS2 a little while back.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Cc: Abhijith Das <adas@redhat.com>
16 years ago[GFS2] Initialize extent_list earlier
Bob Peterson [Thu, 3 Jan 2008 15:24:53 +0000 (09:24 -0600)]
[GFS2] Initialize extent_list earlier

Here is a patch for the latest upstream GFS2 code:
The journal extent map needs to be initialized sooner than it
currently is.  Otherwise failed mount attempts (e.g. not enough
journals, etc.) may panic trying to access the uninitialized list.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Allow page migration for writeback and ordered pages
Steven Whitehouse [Thu, 3 Jan 2008 11:31:38 +0000 (11:31 +0000)]
[GFS2] Allow page migration for writeback and ordered pages

To improve performance on NUMA, we use the VM's standard page
migration for writeback and ordered pages. Probably we could
also do the same for journaled data, but that would need a
careful audit of the code, so will be the subject of a later
patch.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove unused variable
Steven Whitehouse [Wed, 2 Jan 2008 10:16:56 +0000 (10:16 +0000)]
[GFS2] Remove unused variable

The go_drop_th function is never called or referenced.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Fix log block mapper
Steven Whitehouse [Fri, 14 Dec 2007 14:04:34 +0000 (14:04 +0000)]
[GFS2] Fix log block mapper

A missing offset in the calculation.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Minor correction
Bob Peterson [Wed, 12 Dec 2007 23:52:13 +0000 (17:52 -0600)]
[GFS2] Minor correction

This is a small correction to my previously posted patch1.
It just changes a divide to a shift.  It's faster and doesn't
introduce odd dependencies on 32-bit compiles.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Eliminate the no longer needed sd_statfs_mutex
Bob Peterson [Wed, 12 Dec 2007 17:44:41 +0000 (11:44 -0600)]
[GFS2] Eliminate the no longer needed sd_statfs_mutex

This patch eliminates the unneeded sd_statfs_mutex mutex but preserves
the ordering as discussed.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Incremental patch to fix compiler warning
Bob Peterson [Wed, 12 Dec 2007 15:24:08 +0000 (09:24 -0600)]
[GFS2] Incremental patch to fix compiler warning

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Function meta_read optimization
Bob Peterson [Wed, 12 Dec 2007 01:29:17 +0000 (19:29 -0600)]
[GFS2] Function meta_read optimization

This patch optimizes function gfs2_meta_read.  Basically, gfs2_meta_wait
was being called regardless of whether a disk read was requested.
This just pulls that wait into the if that triggers the read.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Only fetch the dinode once in block_map
Bob Peterson [Wed, 12 Dec 2007 01:16:09 +0000 (19:16 -0600)]
[GFS2] Only fetch the dinode once in block_map

Function gfs2_block_map was often looking up the disk inode twice.
This optimizes it so that only does it once.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Reorganize function gfs2_glmutex_lock
Bob Peterson [Wed, 12 Dec 2007 01:13:54 +0000 (19:13 -0600)]
[GFS2] Reorganize function gfs2_glmutex_lock

This patch optimizes the function gfs2_glmutex_lock.
The basic theory is: Why bother initializing a holder, setting up
wait bits and then waiting on them, if you know the glock can be
yours.  So the holder stuff is placed inside the if checking if the
glock is locked.  This one needs careful scrutiny because changing
anything to do with locking should strike terror into one's heart.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Run through full bitmaps quicker in gfs2_bitfit
Bob Peterson [Wed, 12 Dec 2007 01:00:16 +0000 (19:00 -0600)]
[GFS2] Run through full bitmaps quicker in gfs2_bitfit

I eliminated the passing of an unused parameter into gfs2_bitfit called rgd.

This also changes the gfs2_bitfit code that searches for free (or used) blocks.
Before, the code was trying to check for bytes that indicated 4 blocks in
the undesired state.  The problem is, it was spending more time trying to
do this than it actually was saving.  This version only optimizes the case
where we're looking for free blocks, and it checks a machine word at a time.
So on 32-bit machines, it will check 32-bits (16 blocks) and on 64-bit
machines, it will check 64-bits (32 blocks) at a time.  The compiler
optimizes that quite well and we save some time, especially when running
through full bitmaps (like the bitmaps allocated for the journals).

There's probably a more elegant or optimized way to do this, but I haven't
thought of it yet.  I'm open to suggestions.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Get rid of useless "found" variable in quota.c
Bob Peterson [Wed, 12 Dec 2007 00:51:25 +0000 (18:51 -0600)]
[GFS2] Get rid of useless "found" variable in quota.c

This just eliminates an unused variable from the quota code.
Not likely to be a time saver.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Journal extent mapping
Bob Peterson [Wed, 12 Dec 2007 00:49:21 +0000 (18:49 -0600)]
[GFS2] Journal extent mapping

This patch saves a little time when gfs2 writes to the journals by
keeping a mapping between logical and physical blocks on disk.
That's better than constantly looking up indirect pointers in
buffers, when the journals are several levels of indirection
(which they typically are).

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove function gfs2_get_block
Bob Peterson [Mon, 10 Dec 2007 20:13:27 +0000 (14:13 -0600)]
[GFS2] Remove function gfs2_get_block

This patch is just a cleanup.  Function gfs2_get_block() just calls
function gfs2_block_map reversing the last two parameters.  By
reversing the parameters, gfs2_block_map() may be called directly
and function gfs2_get_block may be eliminated altogether.
Since this function is done for every block operation,
this streamlines the code and makes it a little bit more efficient.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] use pid for plock owner for nfs clients
David Teigland [Thu, 6 Dec 2007 15:35:25 +0000 (09:35 -0600)]
[GFS2] use pid for plock owner for nfs clients

The fl_owner is that of lockd when posix locks arrive from nfs
clients, so it can't be used to distinguish between lock holders.
Use fl_pid as owner instead; it's the pid of the process on the
nfs client.

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove unused variable
Steven Whitehouse [Fri, 30 Nov 2007 08:17:15 +0000 (08:17 +0000)]
[GFS2] Remove unused variable

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] patch to check for recursive lock requests in gfs2_rename code path
Abhijith Das [Thu, 29 Nov 2007 20:13:54 +0000 (14:13 -0600)]
[GFS2] patch to check for recursive lock requests in gfs2_rename code path

A certain scenario in the rename code path triggers a kernel BUG()
because it accidentally does recursive locking The first lock is
requested to unlink an already existing inode (replacing a file) and the
second lock is requested when the destination directory needs to alloc
some space. It is rare that these two
events happen during the same rename call, and even more rare that these
two instances try to lock the same rgrp. It is, however, possible.
https://bugzilla.redhat.com/show_bug.cgi?id=404711

Signed-off-by: Abhijith Das <adas@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove lock methods for lock_nolock protocol
Wendy Cheng [Thu, 29 Nov 2007 22:56:51 +0000 (17:56 -0500)]
[GFS2] Remove lock methods for lock_nolock protocol

GFS2 supports two modes of locking - lock_nolock for single node filesystem
and lock_dlm for cluster mode locking. The gfs2 lock methods are removed from
file operation table for lock_nolock protocol. This would allow VFS to handle
posix lock and flock logics just like other in-tree filesystems without
duplication.

Signed-off-by: S. Wendy Cheng <wcheng@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove unrequired code
Fabio M. Di Nitto [Wed, 28 Nov 2007 15:22:09 +0000 (16:22 +0100)]
[GFS2] Remove unrequired code

Signed-off-by: Fabio M. Di Nitto <fabbione@ubuntu.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Fix build warnings
Fabio Massimo Di Nitto [Tue, 27 Nov 2007 05:16:42 +0000 (06:16 +0100)]
[GFS2] Fix build warnings

Hi Steven,

Steven Whitehouse wrote:
> Hi,
>
> Now in the -nmw git tree. Thanks,
>
> Steve.
>
> On Wed, 2007-11-21 at 11:54 -0600, Ryan O'Hara wrote:

this patch introduces a bunch of build warnings by leaving around

struct inode *inode = &ip->i_inode;

The patch in attachment cleans them up. Please apply.

Signed-off-by: Fabio Massimo Di Nitto <fabbione@ubuntu.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] remove unnecessary permission checks
Ryan O'Hara [Wed, 21 Nov 2007 17:54:54 +0000 (11:54 -0600)]
[GFS2] remove unnecessary permission checks

Remove read/write permission() checks from xattr operations.
VFS layer is already handling permission for xattrs via the
xattr_permission() call, so there is no need for gfs2 to
check permissions. Futhermore, using permission() for SELinux
xattrs ops is incorrect.

Signed-off-by: Ryan O'Hara <rohara@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Fix runtime issue with UP kernels
Fabio Massimo Di Nitto [Fri, 16 Nov 2007 09:50:40 +0000 (09:50 +0000)]
[GFS2] Fix runtime issue with UP kernels

The issue is indeed UP vs SMP and it is totally random.

spin_is_locked() is a bad assertion because there is no correct answer on UP.
on UP spin_is_locked() has to return either one value or another, always.

This means that in my setup I am lucky enough to trigger the issue and your you
are lucky enough not to.

the patch in attachment removes the bogus calls to BUG_ON and according to David
(in CC and thanks for the long explanation on the problem) we can rely upon
things like lockdep to find problem that might be trying to catch.

Signed-off-by: Fabio M. Di Nitto <fabbione@ubuntu.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] tidy up error message
David Teigland [Thu, 15 Nov 2007 15:01:13 +0000 (09:01 -0600)]
[GFS2] tidy up error message

Print error with log_error() to be consistent with others.

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Fabio M. Di Nitto <fabbione@ubuntu.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Check for installation of mount helpers for DLM mounts
Fabio Massimo Di Nitto [Thu, 15 Nov 2007 13:48:52 +0000 (13:48 +0000)]
[GFS2] Check for installation of mount helpers for DLM mounts

The patch is a fix to abort mount if the mount.gfs* and possible
umount.* are missing from /sbin.

While we do what we can to guarantee that they are installed properly in
userland (CVS HEAD), we want to make sure that mount still aborts properly.

The only sign of missing helpers is that lock_dlm will receive no mount options
at all. According to David the problem does not exist for lock_nolock as the
helpers are not required.

The patch has been tested for both gfs and gfs2 and it works as expected. The
lack of mount.gfs* will generate an error that is propagated to mount:

oot@node1:~# mount -t  gfs2 /dev/nbd2 /mnt/
mount: wrong fs type, bad option, bad superblock on /dev/nbd2,
       missing codepage or helper program, or other error
       In some cases useful info is found in syslog - try
       dmesg | tail  or so

[ 3513.303346] GFS2: fsid=: Trying to join cluster "lock_dlm", "gutsy:gfs2"
[ 3513.304546] DLM/GFS2/GFS ERROR: (u)mount helpers are not installed properly!
[ 3513.306290] GFS2: fsid=: can't mount proto=lock_dlm, table=gutsy:gfs2, hostdata=

You might want to notice that it will also avoid mount to hang or fail silently
or with strange errors that will require the cluster to reboot/restart before
you can actually mount the filesystem again.

Signed-off-by: Fabio M. Di Nitto <fabbione@ubuntu.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Don't periodically update the jindex
Steven Whitehouse [Fri, 9 Nov 2007 10:07:21 +0000 (10:07 +0000)]
[GFS2] Don't periodically update the jindex

We only care about the content of the jindex in two cases,
one is when we mount the fs and the other is when we need
to recover another journal. In both cases we have to update
the jindex anyway, so there is no point in updating it
periodically between times, so this removes it to simplify
gfs2_logd.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Move gfs2_logd into log.c
Steven Whitehouse [Fri, 9 Nov 2007 10:01:41 +0000 (10:01 +0000)]
[GFS2] Move gfs2_logd into log.c

This means that we can mark gfs2_ail1_empty static and prepares
the way for further changes.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Use atomic_t for journal free blocks counter
Steven Whitehouse [Thu, 8 Nov 2007 14:55:03 +0000 (14:55 +0000)]
[GFS2] Use atomic_t for journal free blocks counter

This patch changes the counter which keeps track of the free
blocks in the journal to an atomic_t in preparation for the
following patch which will update the log reservation code.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Don't add glocks to the journal
Steven Whitehouse [Thu, 8 Nov 2007 14:25:12 +0000 (14:25 +0000)]
[GFS2] Don't add glocks to the journal

The only reason for adding glocks to the journal was to keep track
of which locks required a log flush prior to release. We add a
flag to the glock to allow this check to be made in a simpler way.

This reduces the size of a glock (by 12 bytes on i386, 24 on x86_64)
and means that we can avoid extra work during the journal flush.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] check kthread_should_stop when waiting
David Teigland [Wed, 7 Nov 2007 15:03:56 +0000 (09:03 -0600)]
[GFS2] check kthread_should_stop when waiting

Use wait_event_interruptible() in the lock_dlm thread instead
of an open coded equivalent, and include a kthread_should_stop()
check in the wait test so we don't miss a kthread_stop().

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Given device ID rather than s_id in "id" sysfs file
Bob Peterson [Fri, 2 Nov 2007 14:37:15 +0000 (09:37 -0500)]
[GFS2] Given device ID rather than s_id in "id" sysfs file

This patch changes the /sys/fs/gfs2/<s_id>/id file to give the device
id "major:minor" rather than the s_id.  That enables gfs2_tool to
match devices properly (by id, not name) when locating the tuning files.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove flags no longer required
Steven Whitehouse [Fri, 2 Nov 2007 09:14:31 +0000 (09:14 +0000)]
[GFS2] Remove flags no longer required

The HIF_MUTEX and HIF_PROMOTE flags were set on the glock holders
depending upon which of the two waiters lists they were going to
be queued upon. They were then tested when the holders were taken
off the lists to ensure that the right type of holder was being
dequeued.

Since we are already using separate lists, there doesn't seem a
lot of point having these flags as well, and since setting them
and testing them is in the fast path for locking and unlocking
glock, this patch removes them.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Reorder writeback for glock sync
Steven Whitehouse [Fri, 2 Nov 2007 08:39:34 +0000 (08:39 +0000)]
[GFS2] Reorder writeback for glock sync

Previously we were doing (write data, wait for data, write metadata, wait
for metadata). After this patch we so (write metadata, write data, wait for
data, wait for metadata) which should be more efficient.

Also I noticed that the drop_bh and xmote_bh functions were almost
identical. In fact the only difference was a single test, and that
test is such that in the drop_bh case, it would always evaluate to
the correct result. As such we can use the xmote_bh functions in
all the places where we were using the drop_bh function and remove
the drop_bh functions.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Add sync_page to metadata address space operations
Steven Whitehouse [Thu, 1 Nov 2007 09:34:14 +0000 (09:34 +0000)]
[GFS2] Add sync_page to metadata address space operations

This set of address space operations was missing a sync_page
operation.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove "reclaim limit"
Steven Whitehouse [Thu, 1 Nov 2007 09:26:54 +0000 (09:26 +0000)]
[GFS2] Remove "reclaim limit"

This call to reclaim glocks is not needed, and in particular we don't want it
in the fast path for locking glocks. The limit was entirely arbitrary anyway
and we can't expect users to adjust things like this, the remaining code will
do the right thing on its own.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove unused variables
Steven Whitehouse [Wed, 31 Oct 2007 14:24:33 +0000 (14:24 +0000)]
[GFS2] Remove unused variables

These haven't been used for some time, remove them.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Use correct include file in ops_address.c
Steven Whitehouse [Thu, 18 Oct 2007 10:15:50 +0000 (11:15 +0100)]
[GFS2] Use correct include file in ops_address.c

Something changed in the upstream kernel, and it needs this
one-liner to allow ops_address.c to build.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Don't hold page lock when starting transaction
Steven Whitehouse [Wed, 17 Oct 2007 13:05:41 +0000 (14:05 +0100)]
[GFS2] Don't hold page lock when starting transaction

This is an addendum to the new AOPs work which moves the point
at which we take the page lock so that we don't get it until
the last possible moment. This resolves a conflict between
starting transactions and the page lock.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Add writepages for GFS2 jdata
Steven Whitehouse [Wed, 17 Oct 2007 08:04:24 +0000 (09:04 +0100)]
[GFS2] Add writepages for GFS2 jdata

This patch resolves a lock ordering issue where we had been getting
a transaction lock in the wrong order with respect to the page lock.
By using writepages rather than just writepage, it is then possible
to start a transaction before locking the page, and thus matching the
locking order elsewhere in the code.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Split gfs2_writepage into three cases
Steven Whitehouse [Fri, 28 Sep 2007 12:49:05 +0000 (13:49 +0100)]
[GFS2] Split gfs2_writepage into three cases

This patch splits gfs2_writepage into separate functions for each of
the three cases: writeback, ordered and journalled. As a result
it becomes a lot easier to see what each one is doing. The common
code is moved into gfs2_writepage_common.

This fixes a performance bug where we were doing more work than
strictly required in the ordered write case.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Introduce gfs2_set_aops()
Steven Whitehouse [Wed, 17 Oct 2007 07:47:38 +0000 (08:47 +0100)]
[GFS2] Introduce gfs2_set_aops()

Just like ext3 we now have three sets of address space operations
to cover the cases of writeback, ordered and journalled data
writes. This means that the individual operations can now become
less complicated as we are able to remove some of the tests for
file data mode from the code.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Add gfs2_is_writeback()
Steven Whitehouse [Wed, 17 Oct 2007 07:35:19 +0000 (08:35 +0100)]
[GFS2] Add gfs2_is_writeback()

This adds a function "gfs2_is_writeback()" along the lines of the
existing "gfs2_is_jdata()" in order to clean up the code and make
the various tests for the inode mode more obvious. It also fixes
the PageChecked() logic where we were resetting the flag too early
in the case of an error path.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove unused field in struct gfs2_inode
Steven Whitehouse [Tue, 16 Oct 2007 10:47:04 +0000 (11:47 +0100)]
[GFS2] Remove unused field in struct gfs2_inode

Removes a field that is not used.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Remove useless i_cache from inodes
Steven Whitehouse [Mon, 15 Oct 2007 15:29:05 +0000 (16:29 +0100)]
[GFS2] Remove useless i_cache from inodes

The i_cache was designed to keep references to the indirect blocks
used during block mapping so that they didn't have to be looked
up continually. The idea failed because there are too many places
where the i_cache needs to be freed, and this has in the past been
the cause of many bugs.

In addition there was no performance benefit being gained since the
disk blocks in question were cached anyway. So this patch removes
it in order to simplify the code to prepare for other changes which
would otherwise have had to add further support for this feature.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Use ->page_mkwrite() for mmap()
Steven Whitehouse [Mon, 15 Oct 2007 14:40:33 +0000 (15:40 +0100)]
[GFS2] Use ->page_mkwrite() for mmap()

This cleans up the mmap() code path for GFS2 by implementing the
page_mkwrite function for GFS2. We are thus able to use the
generic filemap_fault function for our ->fault() implementation.

This now means that shared writable mappings will be much more
efficiently shared across the cluster if there is a reasonable
proportion of read activity (the greater proportion, the better).

As a side effect, it also reduces the size of the code, removes
special cases from readpage and readpages, and makes the code
path easier to follow.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[GFS2] Clean up internal read function
Steven Whitehouse [Mon, 15 Oct 2007 13:42:35 +0000 (14:42 +0100)]
[GFS2] Clean up internal read function

As requested by Christoph, this patch cleans up GFS2's internal
read function so that it no longer uses the do_generic_mapping_read
function. This function is obsolete and GFS2 is the last user of it.

As a side effect the internal read code gets smaller and easier
to read and gfs2_readpage is split into two. One function has the locking
and the other function has the rest of the logic.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Cc: Christoph Hellwig <hch@infradead.org>
16 years ago[GFS2] Handle multiple glock demote requests
Wendy Cheng [Fri, 5 Oct 2007 04:27:58 +0000 (00:27 -0400)]
[GFS2] Handle multiple glock demote requests

Fix a race condition where multiple glock demote requests are sent to
a node back-to-back. This patch does a check inside handle_callback()
to see whether a demote request is in progress. If true, it sets a flag
to make sure run_queue() will loop again to handle the new request,
instead of erronously setting gl_demote_state to a different state.

Signed-off-by: S. Wendy Cheng <wcheng@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years agoLinux 2.6.24
Linus Torvalds [Thu, 24 Jan 2008 22:58:37 +0000 (14:58 -0800)]
Linux 2.6.24

16 years agospi: omap2_mcspi PIO RX fix
Kalle Valo [Thu, 24 Jan 2008 22:00:40 +0000 (14:00 -0800)]
spi: omap2_mcspi PIO RX fix

Before transmission of the last word in PIO RX_ONLY mode rx+tx mode
is enabled:

/* prevent last RX_ONLY read from triggering
 * more word i/o: switch to rx+tx
 */
if (c == 0 && tx == NULL)
mcspi_write_cs_reg(spi,
OMAP2_MCSPI_CHCONF0, l);

But because c is decremented after the test, c will never be zero and
rx+tx will not be enabled. This breaks RX_ONLY mode PIO transfers.

Fix it by decrementing c in the beginning of the various I/O loops.

Signed-off-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoRevert "mac80211: warn when receiving frames with unaligned data"
Linus Torvalds [Thu, 24 Jan 2008 21:35:10 +0000 (13:35 -0800)]
Revert "mac80211: warn when receiving frames with unaligned data"

This reverts commit 81100eb80add328c4d2a377326f15aa0e7236398 for the
release, to avoid the unnecessary warning noise that is only really
relevant to wireless driver developers.

The warning will probably go right back in after I cut the release, but
at least we won't unnecessarily worry users.

Acked-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Thu, 24 Jan 2008 16:12:38 +0000 (08:12 -0800)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Partially revert "Constify function pointer tables."

16 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Thu, 24 Jan 2008 16:09:50 +0000 (08:09 -0800)]
Merge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  Revert "ACPI: Fan: Drop force_power_state acpi_device option"
  ACPI: EC: "DEBUG" needs to be defined earlier
  ACPI: EC: add leading zeros to debug messages
  ACPI: EC: fix dmesg spam regression
  ACPI: DMI blacklist to reduce console warnings on OSI(Linux) systems.
  ACPI: Add ThinkPad R61, ThinkPad T61 to OSI(Linux) white-list
  ACPI: make _OSI(Linux) console messages smarter
  ACPI: Delete Intel Customer Reference Board (CRB) from OSI(Linux) DMI list
  ACPI: on OSI(Linux), print needed DMI rather than requesting dmidecode output
  ACPI: create acpi_dmi_dump()
  DMI: create dmi_get_slot()
  DMI: move dmi_available declaration to linux/dmi.h
  ACPI: processor: Fix null pointer dereference in throttling

16 years agoslab: partially revert list3 changes
Mel Gorman [Thu, 24 Jan 2008 13:49:54 +0000 (05:49 -0800)]
slab: partially revert list3 changes

Partial revert the changes made by 04231b3002ac53f8a64a7bd142fde3fa4b6808c6
to the kmem_list3 management. On a machine with a memoryless node, this
BUG_ON was triggering

static void *____cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid)
{
struct list_head *entry;
struct slab *slabp;
struct kmem_list3 *l3;
void *obj;
int x;

l3 = cachep->nodelists[nodeid];
BUG_ON(!l3);

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Acked-by: Christoph Lameter <clameter@sgi.com>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Nishanth Aravamudan <nacc@us.ibm.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofix hugepages leak due to pagetable page sharing
Larry Woodman [Thu, 24 Jan 2008 13:49:25 +0000 (05:49 -0800)]
fix hugepages leak due to pagetable page sharing

The shared page table code for hugetlb memory on x86 and x86_64
is causing a leak.  When a user of hugepages exits using this code
the system leaks some of the hugepages.

-------------------------------------------------------
Part of /proc/meminfo just before database startup:
HugePages_Total:  5500
HugePages_Free:   5500
HugePages_Rsvd:      0
Hugepagesize:     2048 kB

Just before shutdown:
HugePages_Total:  5500
HugePages_Free:   4475
HugePages_Rsvd:      0
Hugepagesize:     2048 kB

After shutdown:
HugePages_Total:  5500
HugePages_Free:   4988
HugePages_Rsvd:
0 Hugepagesize:     2048 kB
----------------------------------------------------------

The problem occurs durring a fork, in copy_hugetlb_page_range().  It
locates the dst_pte using huge_pte_alloc().  Since huge_pte_alloc() calls
huge_pmd_share() it will share the pmd page if can, yet the main loop in
copy_hugetlb_page_range() does a get_page() on every hugepage.  This is a
violation of the shared hugepmd pagetable protocol and creates additional
referenced to the hugepages causing a leak when the unmap of the VMA
occurs.  We can skip the entire replication of the ptes when the hugepage
pagetables are shared.  The attached patch skips copying the ptes and the
get_page() calls if the hugetlbpage pagetable is shared.

[akpm@linux-foundation.org: coding-style cleanups]
Signed-off-by: Larry Woodman <lwoodman@redhat.com>
Signed-off-by: Adam Litke <agl@us.ibm.com>
Cc: Badari Pulavarty <pbadari@us.ibm.com>
Cc: Ken Chen <kenchen@google.com>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: William Lee Irwin III <wli@holomorphy.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agosysctl: kill binary sysctl KERN_PPC_L2CR
Eric W. Biederman [Thu, 24 Jan 2008 13:52:13 +0000 (05:52 -0800)]
sysctl: kill binary sysctl KERN_PPC_L2CR

: Stefan Roese <sr@denx.de> said:
> ppc: 4xx: sysctl table check failed: /kernel/l2cr .1.31 Missing strategy
>
> I'm seeing this error message when booting an recent arch/ppc kernel on
> 4xx platforms (tested on Ocotea and other 4xx platforms). Booting NFS
> rootfs still works fine, but this message kind of makes me "nervous".
> This is not seen on 4xx arch/powerpc platforms. Here the bootlog:

Because the data field was never filled and a binary sysctl handler was
never written this sysctl has never been usable through the sys_sysctl
interface.  So just remove the binary sysctl number.  Making the kernel
sanity checks happy.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Reported-by: Stefan Roese <sr@denx.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agolockdep: fix kernel crash on module unload
Arjan van de Ven [Thu, 24 Jan 2008 06:00:45 +0000 (07:00 +0100)]
lockdep: fix kernel crash on module unload

Michael Wu noticed in his lkml post at

    http://marc.info/?l=linux-kernel&m=119396182726091&w=2

that certain wireless drivers ended up having their name in module
memory, which would then crash the kernel on module unload.

The patch he proposed was a bit clumsy in that it increased the size of
a lockdep entry significantly; the patch below tries another approach,
it checks, on module teardown, if the name of a class is in module space
and then zaps the class.  This is very similar to what we already do
with keys that are in module space.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years ago[SPARC64]: Partially revert "Constify function pointer tables."
David S. Miller [Thu, 24 Jan 2008 05:32:04 +0000 (21:32 -0800)]
[SPARC64]: Partially revert "Constify function pointer tables."

This partially reverts 872e2be7c4056496c2871bd9b0f2fae6c374fe47
(Constify function pointer tables.)

The solaris/socksys.c transformation wasn't valid:

arch/sparc64/solaris/socksys.c:192: error: assignment of read-only variable ‘socksys_file_ops’
arch/sparc64/solaris/socksys.c:195: error: assignment of read-only variable ‘socksys_file_ops’
arch/sparc64/solaris/socksys.c:196: error: assignment of read-only variable ‘socksys_file_ops’

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoPull dmi-2.6.24 into release branch
Len Brown [Thu, 24 Jan 2008 04:50:01 +0000 (23:50 -0500)]
Pull dmi-2.6.24 into release branch

16 years agoPull bugzilla-9798 into release branch
Len Brown [Thu, 24 Jan 2008 04:48:46 +0000 (23:48 -0500)]
Pull bugzilla-9798 into release branch

16 years agoPull bugzilla-8459 into release branch
Len Brown [Thu, 24 Jan 2008 04:48:33 +0000 (23:48 -0500)]
Pull bugzilla-8459 into release branch

16 years agoPull bugzilla-9747 into release branch
Len Brown [Thu, 24 Jan 2008 04:48:19 +0000 (23:48 -0500)]
Pull bugzilla-9747 into release branch

16 years agoRevert "ACPI: Fan: Drop force_power_state acpi_device option"
Len Brown [Thu, 24 Jan 2008 03:41:20 +0000 (22:41 -0500)]
Revert "ACPI: Fan: Drop force_power_state acpi_device option"

This reverts commit 93ad7c07ad487b036add8760dabcc35666a550ef.

http://bugzilla.kernel.org/show_bug.cgi?id=9798

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: EC: "DEBUG" needs to be defined earlier
Márton Németh [Thu, 24 Jan 2008 03:34:09 +0000 (22:34 -0500)]
ACPI: EC: "DEBUG" needs to be defined earlier

The "DEBUG" symbol needs to be defined before #including <linux/kernel.h> to
get the pr_debug() working.

Signed-off-by: Márton Németh <nm127@freemail.hu>
Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: EC: add leading zeros to debug messages
Márton Németh [Thu, 24 Jan 2008 03:33:06 +0000 (22:33 -0500)]
ACPI: EC: add leading zeros to debug messages

Add leading zeros to pr_debug() calls. For example if x=0x0a, the format
"0x%2x" will result the string "0x a", the format "0x%2.2x" will result "0x0a".

Signed-off-by: Márton Németh <nm127@freemail.hu>
Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: EC: fix dmesg spam regression
Alexey Starikovskiy [Thu, 24 Jan 2008 03:28:34 +0000 (22:28 -0500)]
ACPI: EC: fix dmesg spam regression

Return OBF_1 optimization workaround

http://bugzilla.kernel.org/show_bug.cgi?id=8459

Signed-off-by: Alexey Starikovskiy <astarikovskiy@suse.de>
Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Thu, 24 Jan 2008 02:46:25 +0000 (18:46 -0800)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC]: Constify function pointer tables.
  [SPARC64]: Fix section error in sparcspkr
  [SPARC64]: Fix of section mismatch warnings.

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Thu, 24 Jan 2008 02:41:51 +0000 (18:41 -0800)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  sis190: scheduling while atomic error
  sis190: mdio operation failure is not correctly detected
  sis190: remove duplicate INIT_WORK
  sis190: add cmos ram access code for the SiS19x/968 chipset pair
  [INET]: Fix truesize setting in ip_append_data
  [NETNS]: Re-export init_net via EXPORT_SYMBOL.
  iwlwifi: fix possible read attempt on ucode that is not available
  [IPV4]: Add missing skb->truesize increment in ip_append_page().
  [TULIP] DMFE: Fix SROM parsing regression.
  [BLUETOOTH]: Move children of connection device to NULL before connection down.

16 years agoACPI: DMI blacklist to reduce console warnings on OSI(Linux) systems.
Len Brown [Thu, 24 Jan 2008 02:19:27 +0000 (21:19 -0500)]
ACPI: DMI blacklist to reduce console warnings on OSI(Linux) systems.

This DMI blacklist reduces the console messages
on systems which have a BIOS that invokes OSI(Linux).

As the DMI blacklist already knows about these systems,
the request for DMI info itself is disabled.

Further, if OSI(Linux) has already been determined
to have no beneift, we disable the console message
requesting acpi_osi=Linux test results.

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: Add ThinkPad R61, ThinkPad T61 to OSI(Linux) white-list
Len Brown [Thu, 24 Jan 2008 01:56:18 +0000 (20:56 -0500)]
ACPI: Add ThinkPad R61, ThinkPad T61 to OSI(Linux) white-list

acpi_osi=Linux helps sound on these systems.

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: make _OSI(Linux) console messages smarter
Len Brown [Thu, 24 Jan 2008 01:50:56 +0000 (20:50 -0500)]
ACPI: make _OSI(Linux) console messages smarter

If BIOS invokes _OSI(Linux), the kernel response
depends on what the ACPI DMI list knows about the system,
and that is reflectd in dmesg:

1) System unknown to DMI:

ACPI: BIOS _OSI(Linux) query ignored
ACPI: DMI System Vendor: LENOVO
ACPI: DMI Product Name: 7661W1P
ACPI: DMI Product Version: ThinkPad T61
ACPI: DMI Board Name: 7661W1P
ACPI: DMI BIOS Vendor: LENOVO
ACPI: DMI BIOS Date: 10/18/2007
ACPI: Please send DMI info above to linux-acpi@vger.kernel.org
ACPI: If "acpi_osi=Linux" works better, please notify linux-acpi@vger.kernel.org

2) System known to DMI, but effect of OSI(Linux) unknown:

ACPI: DMI detected: Lenovo ThinkPad T61
...
ACPI: BIOS _OSI(Linux) query ignored via DMI
ACPI: If "acpi_osi=Linux" works better, please notify linux-acpi@vger.kernel.org

3) System known to DMI, which disables _OSI(Linux):

ACPI: DMI detected: Lenovo ThinkPad T61
...
ACPI: BIOS _OSI(Linux) query ignored via DMI

4) System known to DMI, which enable _OSI(Linux):

ACPI: DMI detected: Lenovo ThinkPad T61
ACPI: Added _OSI(Linux)
...
ACPI: BIOS _OSI(Linux) query honored via DMI

cmdline overrides take precidence over the built-in
default and the DMI prescribed default.
cmdline "acpi_osi=Linux" results in:

ACPI: BIOS _OSI(Linux) query honored via cmdline

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: Delete Intel Customer Reference Board (CRB) from OSI(Linux) DMI list
Len Brown [Thu, 24 Jan 2008 01:06:41 +0000 (20:06 -0500)]
ACPI: Delete Intel Customer Reference Board (CRB) from OSI(Linux) DMI list

Linux does not want BIOS writers to invoke _OSI(Linux) -
for in the field it causes more Windows incompatibility problems
than it solves.

So when it is seen in the BIOS for an Intel Customer Reference Board,
Linux should ignore its effect by default, and should complain loudly.
Otherwise, the reference BIOS will go unfixed, and the bad BIOS
will spread to the field.

Users of this board can get the old behavior with "acpi_osi=Linux"

As this was the only entry, delete acpi_osl_dmi_table[].

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: on OSI(Linux), print needed DMI rather than requesting dmidecode output
Len Brown [Thu, 24 Jan 2008 01:04:28 +0000 (20:04 -0500)]
ACPI: on OSI(Linux), print needed DMI rather than requesting dmidecode output

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoACPI: create acpi_dmi_dump()
Len Brown [Thu, 24 Jan 2008 01:01:22 +0000 (20:01 -0500)]
ACPI: create acpi_dmi_dump()

A utility routine to print common entries used
for ACPI-related DMI blacklist entries.

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoDMI: create dmi_get_slot()
Len Brown [Wed, 23 Jan 2008 21:36:45 +0000 (16:36 -0500)]
DMI: create dmi_get_slot()

This simply allows other sub-systems (such as ACPI)
to access and print out slots in static dmi_ident[].

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoDMI: move dmi_available declaration to linux/dmi.h
Len Brown [Wed, 16 Jan 2008 22:20:37 +0000 (17:20 -0500)]
DMI: move dmi_available declaration to linux/dmi.h

Signed-off-by: Len Brown <len.brown@intel.com>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Thu, 24 Jan 2008 02:05:28 +0000 (18:05 -0800)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] initio: fix module hangs on loading

16 years agodrm/i915: add support for E7221 chipset
Carlos Martín [Thu, 24 Jan 2008 00:34:10 +0000 (10:34 +1000)]
drm/i915: add support for E7221 chipset

E7221 chipset is a server version of the i915.

Signed-off-by: Dave Airlie <airlied@linux.ie>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoagp/intel: add support for E7221 chipset
Carlos Martín [Thu, 24 Jan 2008 00:34:09 +0000 (10:34 +1000)]
agp/intel: add support for E7221 chipset

The E7221 chipset is a 915 rebadged for the Intel server line.

Signed-off-by: Dave Airlie <airlied@linux.ie>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoxen: disable vcpu_info placement for now
Jeremy Fitzhardinge [Thu, 24 Jan 2008 00:07:17 +0000 (16:07 -0800)]
xen: disable vcpu_info placement for now

There have been several reports of Xen guest domains locking up when
using vcpu_info structure placement.  Disable it for now.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years ago[SCSI] initio: fix module hangs on loading
Stuart Swales [Wed, 23 Jan 2008 20:00:48 +0000 (20:00 +0000)]
[SCSI] initio: fix module hangs on loading

I've verified (on my Initio 9100 with a DAT drive) that the
2.6.24-rc8-git6 initio module still hangs on loading.

These fixes (other than the printk) are needed to get the module to load
ok (and work correctly) with my adapter & tape drive.

a) printk cosmetic fix

b) cblk->sglen needs setting for later DMA I/O routines to use

c) host->bios_addr needs setting for debug output correctness

d) semaph & semaph_lock initialisation had got lost since 2.6.22

e) since 2.6.22 the bios data address was truncated to 16 bits (needs 20
when shifted left)

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
16 years agoUpdate ctime and mtime for memory-mapped files
Anton Salikhmetov [Tue, 22 Jan 2008 23:21:18 +0000 (02:21 +0300)]
Update ctime and mtime for memory-mapped files

Update ctime and mtime for memory-mapped files at a write access on
a present, read-only PTE, as well as at a write on a non-present PTE.

Signed-off-by: Anton Salikhmetov <salikhmetov@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agosis190: scheduling while atomic error
Francois Romieu [Sat, 17 Nov 2007 20:29:47 +0000 (21:29 +0100)]
sis190: scheduling while atomic error

sis190_tx_timeout
-> sis190_hw_start
   -> sis190_soft_reset
      -> msleep *splat*

PCI transactions are correctly flushed here.
The msleep() is probably useless.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: K.M. Liu <kmliu@sis.com.tw>
16 years agosis190: mdio operation failure is not correctly detected
Francois Romieu [Sat, 17 Nov 2007 15:56:43 +0000 (16:56 +0100)]
sis190: mdio operation failure is not correctly detected

i ranges from 0 to 100 in the 'for' loop a few lines above.

Reported by davem.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: K.M. Liu <kmliu@sis.com.tw>
16 years agosis190: remove duplicate INIT_WORK
Francois Romieu [Sat, 17 Nov 2007 14:55:10 +0000 (15:55 +0100)]
sis190: remove duplicate INIT_WORK

It is already done in sis190_init_one.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: K.M. Liu <kmliu@sis.com.tw>
16 years agosis190: add cmos ram access code for the SiS19x/968 chipset pair
Francois Romieu [Tue, 4 Dec 2007 21:58:41 +0000 (22:58 +0100)]
sis190: add cmos ram access code for the SiS19x/968 chipset pair

More work is needed to handle correctly the PHY of the new devices
when connected to a 10Mb link but this change already helps some
users as is.

Fix for:
http://bugzilla.kernel.org/show_bug.cgi?id=9467

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: K.M. Liu <kmliu@sis.com.tw>
Cc: J. Gleacher <jgleacher@yahoo.com>
Cc: Alexandre Penasso Teixeira <alexandre@keepsoftware.com>
Cc: Arliton Rocha <arliton@gmail.com>
Cc: Juan Jose Pablos <juanjo@apertus.es>
Cc: Wipat Srutiprom <wipat.s@psu.ac.th>
16 years ago[INET]: Fix truesize setting in ip_append_data
Herbert Xu [Wed, 23 Jan 2008 06:39:26 +0000 (22:39 -0800)]
[INET]: Fix truesize setting in ip_append_data

As it is ip_append_data only counts page fragments to the skb that
allocated it.  As such it means that the first skb gets hit with a
4K charge even though it might have only used a fraction of it while
all subsequent skb's that use the same page gets away with no charge
at all.

This bug was exposed by the UDP accounting patch.

[ The wmem_alloc bumping needs to be moved with the truesize,
  noticed by Takahiro Yasui.  -DaveM ]

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[NETNS]: Re-export init_net via EXPORT_SYMBOL.
Denis V. Lunev [Wed, 23 Jan 2008 06:05:33 +0000 (22:05 -0800)]
[NETNS]: Re-export init_net via EXPORT_SYMBOL.

init_net is used added as a parameter to a lot of old API calls, f.e.
ip_dev_find. These calls were exported as EXPORT_SYMBOL. So, export init_net
as EXPORT_SYMBOL to keep networking API consistent.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoiwlwifi: fix possible read attempt on ucode that is not available
Reinette Chatre [Mon, 21 Jan 2008 18:08:31 +0000 (10:08 -0800)]
iwlwifi: fix possible read attempt on ucode that is not available

This fixes a NULL pointer dereference that can occur when the
ucode is not loaded at the time __iwl_up is called.

The problem was reported at http://kerneloops.org/raw.php?rawid=2765&msgid=

Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years ago[IPV4]: Add missing skb->truesize increment in ip_append_page().
David S. Miller [Wed, 23 Jan 2008 07:44:31 +0000 (23:44 -0800)]
[IPV4]: Add missing skb->truesize increment in ip_append_page().

And as noted by Takahiro Yasui, we thus need to bump the
sk->sk_wmem_alloc at this spot as well.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[TULIP] DMFE: Fix SROM parsing regression.
David S. Miller [Tue, 22 Jan 2008 07:20:58 +0000 (23:20 -0800)]
[TULIP] DMFE: Fix SROM parsing regression.

Changeset 16b110c3fd760620b4a787db6ed512fe531ab1b5 (dmfe warning fix)
bothed up the offsets read from the SROM so that it doesn't read the
same datums it used to.

The change made transformations like turning:

"srom + 34"

into

"(__le32 *)srom + 34/4"

which doesn't work because 4 does not divide evenly
into 34 so we're using a different pointer offset
than in the original code.

I've changed theses cases in dmfe_parse_srom() to
consistently use "(type *)(srom + offset)" preserving
the offsets from the original code.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[BLUETOOTH]: Move children of connection device to NULL before connection down.
Dave Young [Tue, 22 Jan 2008 06:35:21 +0000 (22:35 -0800)]
[BLUETOOTH]: Move children of connection device to NULL before connection down.

The rfcomm tty device will possibly retain even when conn is down, and
sysfs doesn't support zombie device moving, so this patch move the tty
device before conn device is destroyed.

For the bug refered please see :
http://lkml.org/lkml/2007/12/28/87

Signed-off-by: Dave Young <hidave.darkstar@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[SPARC]: Constify function pointer tables.
Jan Engelhardt [Wed, 23 Jan 2008 02:29:20 +0000 (18:29 -0800)]
[SPARC]: Constify function pointer tables.

Signed-off-by: Jan Engelhardt <jengelh@computergmbh.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agox86: GEODE fix a race condition in the MFGPT timer tick
Jordan Crouse [Tue, 22 Jan 2008 22:30:16 +0000 (23:30 +0100)]
x86: GEODE fix a race condition in the MFGPT timer tick

When we set the MFGPT timer tick, there is a chance that we'll
immediately assert an event.  If for some reason the IRQ routing
for this clock has been setup for some other purpose, then we
could end up firing an interrupt into the SMM handler or worse.

This rearranges the timer tick init function to initalize the handler
before we set up the MFGPT clock to make sure that even if we get
an event, it will go to the handler.

Furthermore, in the handler we need to make sure that we clear the
event, even if the timer isn't running.

Signed-off-by: Jordan Crouse <jordan.crouse@amd.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ingo Molnar <mingo@elte.hu>
Tested-by: Arnd Hannemann <hannemann@i4.informatik.rwth-aachen.de>
16 years agoFix file references in documentation and Kconfig
Johann Felix Soden [Sun, 20 Jan 2008 13:41:18 +0000 (14:41 +0100)]
Fix file references in documentation and Kconfig

Fix typo in arch/powerpc/boot/flatdevtree_env.h.
There is no Documentation/networking/ixgbe.txt.

README.cycladesZ is now in Documentation/.
wavelan.p.h is now in drivers/net/wireless/.
HFS.txt is now Documentation/filesystems/hfs.txt.
OSS-files are now in sound/oss/.

Signed-off-by: Johann Felix Soden <johfel@users.sourceforge.net>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'release' of git://lm-sensors.org/kernel/mhoffman/hwmon-2.6
Linus Torvalds [Tue, 22 Jan 2008 17:25:55 +0000 (09:25 -0800)]
Merge branch 'release' of git://lm-sensors.org/kernel/mhoffman/hwmon-2.6

* 'release' of git://lm-sensors.org/kernel/mhoffman/hwmon-2.6:
  hwmon: (it87) request only Environment Controller ports