openwrt/staging/blogic.git
14 years agoblock: fix bad use of min() on different types
Jens Axboe [Thu, 29 Apr 2010 07:28:21 +0000 (09:28 +0200)]
block: fix bad use of min() on different types

Just cast the page size to sector_t, that will always fit.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agocfq-iosched: fix broken cfq_ref_get_cfqf() for CONFIG_BLK_CGROUP=y && CFQ_GROUP_IOSCHED=n
Dmitry Monakhov [Wed, 28 Apr 2010 17:50:33 +0000 (19:50 +0200)]
cfq-iosched: fix broken cfq_ref_get_cfqf() for CONFIG_BLK_CGROUP=y && CFQ_GROUP_IOSCHED=n

We should return the cfq_group for this case, not NULL.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkdev: add blkdev_issue_zeroout helper function
Dmitry Monakhov [Wed, 28 Apr 2010 13:55:09 +0000 (17:55 +0400)]
blkdev: add blkdev_issue_zeroout helper function

- Add bio_batch helper primitive. This is rather generic primitive
  for submitting/waiting a complex request which consists of several
  bios.
- blkdev_issue_zeroout() generate number of zero filed write bios.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkdev: move blkdev_issue helper functions to separate file
Dmitry Monakhov [Wed, 28 Apr 2010 13:55:08 +0000 (17:55 +0400)]
blkdev: move blkdev_issue helper functions to separate file

Move blkdev_issue_discard from blk-barrier.c because it is
not barrier related.
Later the file will be populated by other helpers.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkdev: allow async blkdev_issue_flush requests
Dmitry Monakhov [Wed, 28 Apr 2010 13:55:07 +0000 (17:55 +0400)]
blkdev: allow async blkdev_issue_flush requests

In some places caller don't want to wait a request to complete.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkdev: generalize flags for blkdev_issue_fn functions
Dmitry Monakhov [Wed, 28 Apr 2010 13:55:06 +0000 (17:55 +0400)]
blkdev: generalize flags for blkdev_issue_fn functions

The patch just convert all blkdev_issue_xxx function to common
set of flags. Wait/allocation semantics preserved.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblock: implement bd_claiming and claiming block
Tejun Heo [Wed, 7 Apr 2010 09:53:59 +0000 (18:53 +0900)]
block: implement bd_claiming and claiming block

Currently, device claiming for exclusive open is done after low level
open - disk->fops->open() - has completed successfully.  This means
that exclusive open attempts while a device is already exclusively
open will fail only after disk->fops->open() is called.

cdrom driver issues commands during open() which means that O_EXCL
open attempt can unintentionally inject commands to in-progress
command stream for burning thus disturbing burning process.  In most
cases, this doesn't cause problems because the first command to be
issued is TUR which most devices can process in the middle of burning.
However, depending on how a device replies to TUR during burning,
cdrom driver may end up issuing further commands.

This can't be resolved trivially by moving bd_claim() before doing
actual open() because that means an open attempt which will end up
failing could interfere other legit O_EXCL open attempts.
ie. unconfirmed open attempts can fail others.

This patch resolves the problem by introducing claiming block which is
started by bd_start_claiming() and terminated either by bd_claim() or
bd_abort_claiming().  bd_claim() from inside a claiming block is
guaranteed to succeed and once a claiming block is started, other
bd_start_claiming() or bd_claim() attempts block till the current
claiming block is terminated.

bd_claim() can still be used standalone although now it always
synchronizes against claiming blocks, so the existing users will keep
working without any change.

blkdev_open() and open_bdev_exclusive() are converted to use claiming
blocks so that exclusive open attempts from these functions don't
interfere with the existing exclusive open.

This problem was discovered while investigating bko#15403.

  https://bugzilla.kernel.org/show_bug.cgi?id=15403

The burning problem itself can be resolved by updating userspace
probing tools to always open w/ O_EXCL.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Matthias-Christian Ott <ott@mirix.org>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblock: factor out bd_may_claim()
Tejun Heo [Wed, 7 Apr 2010 09:52:29 +0000 (18:52 +0900)]
block: factor out bd_may_claim()

Factor out bd_may_claim() from bd_claim(), add comments and apply a
couple of cosmetic edits.  This is to prepare for further updates to
claim path.

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblk-cgroup: config options re-arrangement
Vivek Goyal [Mon, 26 Apr 2010 17:27:56 +0000 (19:27 +0200)]
blk-cgroup: config options re-arrangement

This patch fixes few usability and configurability issues.

o All the cgroup based controller options are configurable from
  "Genral Setup/Control Group Support/" menu. blkio is the only exception.
  Hence make this option visible in above menu and make it configurable from
  there to bring it inline with rest of the cgroup based controllers.

o Get rid of CONFIG_DEBUG_CFQ_IOSCHED.

  This option currently does two things.

  - Enable printing of cgroup paths in blktrace
  - Enables CONFIG_DEBUG_BLK_CGROUP, which in turn displays additional stat
    files in cgroup.

  If we are using group scheduling, blktrace data is of not really much use
  if cgroup information is not present. To get this data, currently one has to
  also enable CONFIG_DEBUG_CFQ_IOSCHED, which in turn brings the overhead of
  all the additional debug stat files which is not desired.

  Hence, this patch moves printing of cgroup paths under
  CONFIG_CFQ_GROUP_IOSCHED.

  This allows us to get rid of CONFIG_DEBUG_CFQ_IOSCHED completely. Now all
  the debug stat files are controlled only by CONFIG_DEBUG_BLK_CGROUP which
  can be enabled through config menu.

Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Acked-by: Divyesh Shah <dpshah@google.com>
Reviewed-by: Gui Jianfeng <guijianfeng@cn.fujitsu.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: Fix another BUG_ON() crash due to cfqq movement across groups
Vivek Goyal [Mon, 26 Apr 2010 17:25:11 +0000 (19:25 +0200)]
blkio: Fix another BUG_ON() crash due to cfqq movement across groups

o Once in a while, I was hitting a BUG_ON() in blkio code. empty_time was
  assuming that upon slice expiry, group can't be marked empty already (except
  forced dispatch).

  But this assumption is broken if cfqq can move (group_isolation=0) across
  groups after receiving a request.

  I think most likely in this case we got a request in a cfqq and accounted
  the rq in one group, later while adding the cfqq to tree, we moved the queue
  to a different group which was already marked empty and after dispatch from
  slice we found group already marked empty and raised alarm.

  This patch does not error out if group is already marked empty. This can
  introduce some empty_time stat error only in case of group_isolation=0. This
  is better than crashing. In case of group_isolation=1 we should still get
  same stats as before this patch.

[  222.308546] ------------[ cut here ]------------
[  222.309311] kernel BUG at block/blk-cgroup.c:236!
[  222.309311] invalid opcode: 0000 [#1] SMP
[  222.309311] last sysfs file: /sys/devices/virtual/block/dm-3/queue/scheduler
[  222.309311] CPU 1
[  222.309311] Modules linked in: dm_round_robin dm_multipath qla2xxx scsi_transport_fc dm_zero dm_mirror dm_region_hash dm_log dm_mod [last unloaded: scsi_wait_scan]
[  222.309311]
[  222.309311] Pid: 4780, comm: fio Not tainted 2.6.34-rc4-blkio-config #68 0A98h/HP xw8600 Workstation
[  222.309311] RIP: 0010:[<ffffffff8121ad88>]  [<ffffffff8121ad88>] blkiocg_set_start_empty_time+0x50/0x83
[  222.309311] RSP: 0018:ffff8800ba6e79f8  EFLAGS: 00010002
[  222.309311] RAX: 0000000000000082 RBX: ffff8800a13b7990 RCX: ffff8800a13b7808
[  222.309311] RDX: 0000000000002121 RSI: 0000000000000082 RDI: ffff8800a13b7a30
[  222.309311] RBP: ffff8800ba6e7a18 R08: 0000000000000000 R09: 0000000000000001
[  222.309311] R10: 000000000002f8c8 R11: ffff8800ba6e7ad8 R12: ffff8800a13b78ff
[  222.309311] R13: ffff8800a13b7990 R14: 0000000000000001 R15: ffff8800a13b7808
[  222.309311] FS:  00007f3beec476f0(0000) GS:ffff880001e40000(0000) knlGS:0000000000000000
[  222.309311] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  222.309311] CR2: 000000000040e7f0 CR3: 00000000a12d5000 CR4: 00000000000006e0
[  222.309311] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  222.309311] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[  222.309311] Process fio (pid: 4780, threadinfo ffff8800ba6e6000, task ffff8800b3d6bf00)
[  222.309311] Stack:
[  222.309311]  0000000000000001 ffff8800bab17a48 ffff8800bab17a48 ffff8800a13b7800
[  222.309311] <0> ffff8800ba6e7a68 ffffffff8121da35 ffff880000000001 00ff8800ba5c5698
[  222.309311] <0> ffff8800ba6e7a68 ffff8800a13b7800 0000000000000000 ffff8800bab17a48
[  222.309311] Call Trace:
[  222.309311]  [<ffffffff8121da35>] __cfq_slice_expired+0x2af/0x3ec
[  222.309311]  [<ffffffff8121fd7b>] cfq_dispatch_requests+0x2c8/0x8e8
[  222.309311]  [<ffffffff8120f1cd>] ? spin_unlock_irqrestore+0xe/0x10
[  222.309311]  [<ffffffff8120fb1a>] ? blk_insert_cloned_request+0x70/0x7b
[  222.309311]  [<ffffffff81210461>] blk_peek_request+0x191/0x1a7
[  222.309311]  [<ffffffffa0002799>] dm_request_fn+0x38/0x14c [dm_mod]
[  222.309311]  [<ffffffff810ae61f>] ? sync_page_killable+0x0/0x35
[  222.309311]  [<ffffffff81210fd4>] __generic_unplug_device+0x32/0x37
[  222.309311]  [<ffffffff81211274>] generic_unplug_device+0x2e/0x3c
[  222.309311]  [<ffffffffa00011a6>] dm_unplug_all+0x42/0x5b [dm_mod]
[  222.309311]  [<ffffffff8120ca37>] blk_unplug+0x29/0x2d
[  222.309311]  [<ffffffff8120ca4d>] blk_backing_dev_unplug+0x12/0x14
[  222.309311]  [<ffffffff81109a7a>] block_sync_page+0x35/0x39
[  222.309311]  [<ffffffff810ae616>] sync_page+0x41/0x4a
[  222.309311]  [<ffffffff810ae62d>] sync_page_killable+0xe/0x35
[  222.309311]  [<ffffffff8158aa59>] __wait_on_bit_lock+0x46/0x8f
[  222.309311]  [<ffffffff810ae4f5>] __lock_page_killable+0x66/0x6d
[  222.309311]  [<ffffffff81056f9c>] ? wake_bit_function+0x0/0x33
[  222.309311]  [<ffffffff810ae528>] lock_page_killable+0x2c/0x2e
[  222.309311]  [<ffffffff810afbc5>] generic_file_aio_read+0x361/0x4f0
[  222.309311]  [<ffffffff810ea044>] do_sync_read+0xcb/0x108
[  222.309311]  [<ffffffff811e42f7>] ? security_file_permission+0x16/0x18
[  222.309311]  [<ffffffff810ea6ab>] vfs_read+0xab/0x108
[  222.309311]  [<ffffffff810ea7c8>] sys_read+0x4a/0x6e
[  222.309311]  [<ffffffff81002b5b>] system_call_fastpath+0x16/0x1b
[  222.309311] Code: 58 01 00 00 00 48 89 c6 75 0a 48 83 bb 60 01 00 00 00 74 09 48 8d bb a0 00 00 00 eb 35 41 fe cc 74 0d f6 83 c0 01 00 00 04 74 04 <0f> 0b eb fe 48 89 75 e8 e8 be e0 de ff 66 83 8b c0 01 00 00 04
[  222.309311] RIP  [<ffffffff8121ad88>] blkiocg_set_start_empty_time+0x50/0x83
[  222.309311]  RSP <ffff8800ba6e79f8>
[  222.309311] ---[ end trace 32b4f71dffc15712 ]---

Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Acked-by: Divyesh Shah <dpshah@google.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: Fix blkio crash during rq stat update
Vivek Goyal [Wed, 21 Apr 2010 15:44:16 +0000 (17:44 +0200)]
blkio: Fix blkio crash during rq stat update

blkio + cfq was crashing even when two sequential readers were put in two
separate cgroups (group_isolation=0).

The reason being that cfqq can migrate across groups based on its being
sync-noidle or not, it can happen that at request insertion time, cfqq
belonged to one cfqg and at request dispatch time, it belonged to root
group. In this case request stats per cgroup can go wrong and it also runs
into BUG_ON().

This patch implements rq stashing away a cfq group pointer and not relying
on cfqq->cfqg pointer alone for rq stat accounting.

[   65.163523] ------------[ cut here ]------------
[   65.164301] kernel BUG at block/blk-cgroup.c:117!
[   65.164301] invalid opcode: 0000 [#1] SMP
[   65.164301] last sysfs file: /sys/devices/pci0000:00/0000:00:05.0/0000:60:00.1/host9/rport-9:0-0/target9:0:0/9:0:0:2/block/sde/stat
[   65.164301] CPU 1
[   65.164301] Modules linked in: dm_round_robin dm_multipath qla2xxx scsi_transport_fc dm_zero dm_mirror dm_region_hash dm_log dm_mod [last unloaded: scsi_wait_scan]
[   65.164301]
[   65.164301] Pid: 4505, comm: fio Not tainted 2.6.34-rc4-blk-for-35 #34 0A98h/HP xw8600 Workstation
[   65.164301] RIP: 0010:[<ffffffff8121924f>]  [<ffffffff8121924f>] blkiocg_update_io_remove_stats+0x5b/0xaf
[   65.164301] RSP: 0018:ffff8800ba5a79e8  EFLAGS: 00010046
[   65.164301] RAX: 0000000000000096 RBX: ffff8800bb268d60 RCX: 0000000000000000
[   65.164301] RDX: ffff8800bb268eb8 RSI: 0000000000000000 RDI: ffff8800bb268e00
[   65.164301] RBP: ffff8800ba5a7a08 R08: 0000000000000064 R09: 0000000000000001
[   65.164301] R10: 0000000000079640 R11: ffff8800a0bd5bf0 R12: ffff8800bab4af01
[   65.164301] R13: ffff8800bab4af00 R14: ffff8800bb1d8928 R15: 0000000000000000
[   65.164301] FS:  00007f18f75056f0(0000) GS:ffff880001e40000(0000) knlGS:0000000000000000
[   65.164301] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   65.164301] CR2: 000000000040e7f0 CR3: 00000000ba52b000 CR4: 00000000000006e0
[   65.164301] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   65.164301] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[   65.164301] Process fio (pid: 4505, threadinfo ffff8800ba5a6000, task ffff8800ba45ae80)
[   65.164301] Stack:
[   65.164301]  ffff8800ba5a7a08 ffff8800ba722540 ffff8800bab4af68 ffff8800bab4af68
[   65.164301] <0> ffff8800ba5a7a38 ffffffff8121d814 ffff8800ba722540 ffff8800bab4af68
[   65.164301] <0> ffff8800ba722540 ffff8800a08f6800 ffff8800ba5a7a68 ffffffff8121d8ca
[   65.164301] Call Trace:
[   65.164301]  [<ffffffff8121d814>] cfq_remove_request+0xe4/0x116
[   65.164301]  [<ffffffff8121d8ca>] cfq_dispatch_insert+0x84/0xe1
[   65.164301]  [<ffffffff8121e833>] cfq_dispatch_requests+0x767/0x8e8
[   65.164301]  [<ffffffff8120e524>] ? submit_bio+0xc3/0xcc
[   65.164301]  [<ffffffff810ad657>] ? sync_page_killable+0x0/0x35
[   65.164301]  [<ffffffff8120ea8d>] blk_peek_request+0x191/0x1a7
[   65.164301]  [<ffffffffa000109c>] ? dm_get_live_table+0x44/0x4f [dm_mod]
[   65.164301]  [<ffffffffa0002799>] dm_request_fn+0x38/0x14c [dm_mod]
[   65.164301]  [<ffffffff810ad657>] ? sync_page_killable+0x0/0x35
[   65.164301]  [<ffffffff8120f600>] __generic_unplug_device+0x32/0x37
[   65.164301]  [<ffffffff8120f8a0>] generic_unplug_device+0x2e/0x3c
[   65.164301]  [<ffffffffa00011a6>] dm_unplug_all+0x42/0x5b [dm_mod]
[   65.164301]  [<ffffffff8120b063>] blk_unplug+0x29/0x2d
[   65.164301]  [<ffffffff8120b079>] blk_backing_dev_unplug+0x12/0x14
[   65.164301]  [<ffffffff81108a82>] block_sync_page+0x35/0x39
[   65.164301]  [<ffffffff810ad64e>] sync_page+0x41/0x4a
[   65.164301]  [<ffffffff810ad665>] sync_page_killable+0xe/0x35
[   65.164301]  [<ffffffff81589027>] __wait_on_bit_lock+0x46/0x8f
[   65.164301]  [<ffffffff810ad52d>] __lock_page_killable+0x66/0x6d
[   65.164301]  [<ffffffff81055fd4>] ? wake_bit_function+0x0/0x33
[   65.164301]  [<ffffffff810ad560>] lock_page_killable+0x2c/0x2e
[   65.164301]  [<ffffffff810aebfd>] generic_file_aio_read+0x361/0x4f0
[   65.164301]  [<ffffffff810e906c>] do_sync_read+0xcb/0x108
[   65.164301]  [<ffffffff811e32a3>] ? security_file_permission+0x16/0x18
[   65.164301]  [<ffffffff810e96d3>] vfs_read+0xab/0x108
[   65.164301]  [<ffffffff810e97f0>] sys_read+0x4a/0x6e
[   65.164301]  [<ffffffff81002b5b>] system_call_fastpath+0x16/0x1b
[   65.164301] Code: 00 74 1c 48 8b 8b 60 01 00 00 48 85 c9 75 04 0f 0b eb fe 48 ff c9 48 89 8b 60 01 00 00 eb 1a 48 8b 8b 58 01 00 00 48 85 c9 75 04 <0f> 0b eb fe 48 ff c9 48 89 8b 58 01 00 00 45 84 e4 74 16 48 8b
[   65.164301] RIP  [<ffffffff8121924f>] blkiocg_update_io_remove_stats+0x5b/0xaf
[   65.164301]  RSP <ffff8800ba5a79e8>
[   65.164301] ---[ end trace 1b2b828753032e68 ]---

Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: Initialize blkg->stats_lock for the root cfqg too
Divyesh Shah [Fri, 16 Apr 2010 06:10:51 +0000 (08:10 +0200)]
blkio: Initialize blkg->stats_lock for the root cfqg too

This fixes the lockdep warning reported by Gui Jianfeng.

Signed-off-by: Divyesh Shah <dpshah@google.com>
Reviewed-by: Gui Jianfeng <guijianfeng@cn.fujitsu.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: fix for modular blk-cgroup build
Divyesh Shah [Thu, 15 Apr 2010 06:54:59 +0000 (08:54 +0200)]
blkio: fix for modular blk-cgroup build

After merging the block tree, 20100414's linux-next build (x86_64
allmodconfig) failed like this:

ERROR: "get_gendisk" [block/blk-cgroup.ko] undefined!
ERROR: "sched_clock" [block/blk-cgroup.ko] undefined!

This happens because the two symbols aren't exported and hence not available
when blk-cgroup code is built as a module. I've tried to stay consistent with
the use of EXPORT_SYMBOL or EXPORT_SYMBOL_GPL with the other symbols in the
respective files.

Signed-off-by: Divyesh Shah <dpshah@google.com>
Acked-by: Gui Jianfeng <guijianfeng@cn.fujitsu.cn>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblock: ensure jiffies wrap is handled correctly in blk_rq_timed_out_timer
Richard Kennedy [Wed, 14 Apr 2010 18:53:37 +0000 (20:53 +0200)]
block: ensure jiffies wrap is handled correctly in blk_rq_timed_out_timer

blk_rq_timed_out_timer() relied on blk_add_timer() never returning a
timer value of zero, but commit 7838c15b8dd18e78a523513749e5b54bda07b0cb
removed the code that bumped this value when it was zero.
Therefore when jiffies is near wrap we could get unlucky & not set the
timeout value correctly.

This patch uses a flag to indicate that the timeout value was set and so
handles jiffies wrap correctly, and it keeps all the logic in one
function so should be easier to maintain in the future.

Signed-off-by: Richard Kennedy <richard@rsk.demon.co.uk>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: Fix compile errors
Divyesh Shah [Wed, 14 Apr 2010 09:22:38 +0000 (11:22 +0200)]
blkio: Fix compile errors

Fixes compile errors in blk-cgroup code for empty_time stat and a merge fix in
CFQ. The first error was when CONFIG_DEBUG_CFQ_IOSCHED is not set.

Signed-off-by: Divyesh Shah <dpshah@google.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoMerge branch 'master' into for-2.6.35
Jens Axboe [Tue, 13 Apr 2010 18:03:21 +0000 (20:03 +0200)]
Merge branch 'master' into for-2.6.35

Conflicts:
block/blk-cgroup.c
block/cfq-iosched.c

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblock: Update to io-controller stats
Divyesh Shah [Tue, 13 Apr 2010 17:59:17 +0000 (19:59 +0200)]
block: Update to io-controller stats

Changelog from v1:
o Call blkiocg_update_idle_time_stats() at cfq_rq_enqueued() instead of at
  dispatch time.

Changelog from original patchset: (in response to Vivek Goyal's comments)
o group blkiocg_update_blkio_group_dequeue_stats() with other DEBUG functions
o rename blkiocg_update_set_active_queue_stats() to
  blkiocg_update_avg_queue_size_stats()
o s/request/io/ in blkiocg_update_request_add_stats() and
  blkiocg_update_request_remove_stats()
o Call cfq_del_timer() at request dispatch() instead of
  blkiocg_update_idle_time_stats()

Signed-off-by: Divyesh Shah<dpshah@google.com>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoio-controller: Document for blkio.weight_device
Gui Jianfeng [Tue, 13 Apr 2010 08:07:50 +0000 (16:07 +0800)]
io-controller: Document for blkio.weight_device

Here is the document for blkio.weight_device

Signed-off-by: Gui Jianfeng <guijianfeng@cn.fujitsu.com>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoio-controller: Add a new interface "weight_device" for IO-Controller
Gui Jianfeng [Tue, 13 Apr 2010 08:05:49 +0000 (16:05 +0800)]
io-controller: Add a new interface "weight_device" for IO-Controller

Currently, IO Controller makes use of blkio.weight to assign weight for
all devices. Here a new user interface "blkio.weight_device" is introduced to
assign different weights for different devices. blkio.weight becomes the
default value for devices which are not configured by "blkio.weight_device"

You can use the following format to assigned specific weight for a given
device:
#echo "major:minor weight" > blkio.weight_device

major:minor represents device number.

And you can remove weight for a given device as following:
#echo "major:minor 0" > blkio.weight_device

V1->V2 changes:
- use user interface "weight_device" instead of "policy" suggested by Vivek
- rename some struct suggested by Vivek
- rebase to 2.6-block "for-linus" branch
- remove an useless list_empty check pointed out by Li Zefan
- some trivial typo fix

V2->V3 changes:
- Move policy_*_node() functions up to get rid of forward declarations
- rename related functions by adding prefix "blkio_"

Signed-off-by: Gui Jianfeng <guijianfeng@cn.fujitsu.com>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoLinux 2.6.34-rc4
Linus Torvalds [Tue, 13 Apr 2010 01:41:35 +0000 (18:41 -0700)]
Linux 2.6.34-rc4

14 years agoMerge branch 'anonvma'
Linus Torvalds [Tue, 13 Apr 2010 01:39:58 +0000 (18:39 -0700)]
Merge branch 'anonvma'

* anonvma:
  anonvma: when setting up page->mapping, we need to pick the _oldest_ anonvma
  anon_vma: clone the anon_vma chain in the right order
  vma_adjust: fix the copying of anon_vma chains
  Simplify and comment on anon_vma re-use for anon_vma_prepare()

14 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Tue, 13 Apr 2010 01:37:34 +0000 (18:37 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm: (21 commits)
  ARM: Fix ioremap_cached()/ioremap_wc() for SMP platforms
  ARM: 6043/1: AT91 slow-clock resume: Don't wait for a disabled PLL to lock
  ARM: 6031/1: fix Thumb-2 decompressor
  ARM: 6029/1: ep93xx: gpio.c: local functions should be static
  ARM: 6028/1: ARM: add MAINTAINERS for U300
  ARM: 6024/1: bcmring: fix missing down on semaphore in dma.c
  MXC: mach_armadillo5x0: Add USB Host support.
  ARM mach-mx3: duplicated include
  ARM mach-mx3: duplicated include
  imx31: add watchdog device on litekit board.
  imx3: Add watchdog platform device support
  MXC: mach-mx31_3ds: add support for freescale mc13783 power management device.
  MXC: mach-mx31_3ds: Add SPI1 device support.
  MXC: mach-mx31_3ds: Add support for on board NAND Flash.
  MXC: mach-mx31_3ds: Update variable names over recent mach name modification.
  imx31: fix parent clock for rtc
  i.MX51: remove NFC AXI static mapping
  i.MX51: determine silicon revision dynamically
  i.MX51: map TZIC dynamically
  i.MX51: Use correct clock for gpt
  ...

14 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs...
Linus Torvalds [Tue, 13 Apr 2010 01:37:04 +0000 (18:37 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/mason/btrfs-unstable

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
  Btrfs: make sure the chunk allocator doesn't create zero length chunks
  Btrfs: fix data enospc check overflow

14 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux...
Linus Torvalds [Tue, 13 Apr 2010 01:36:49 +0000 (18:36 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs-2.6:
  quota: Fix possible dq_flags corruption
  quota: Hide warnings about writes to the filesystem before quota was turned on
  ext3: symlink must be handled via filesystem specific operation
  ext2: symlink must be handled via filesystem specific operation

14 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux...
Linus Torvalds [Tue, 13 Apr 2010 01:36:34 +0000 (18:36 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-udf-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-udf-2.6:
  udf: add speciffic ->setattr callback
  udf: potential integer overflow

14 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Tue, 13 Apr 2010 01:36:11 +0000 (18:36 -0700)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus: (36 commits)
  MIPS: Calculate proper ebase value for 64-bit kernels
  MIPS: Alchemy: DB1200: Remove custom wait implementation
  MIPS: Big Sur: Make defconfig more useful.
  MIPS: Fix __vmalloc() etc. on MIPS for non-GPL modules
  MIPS: Sibyte: Fix M3 TLB exception handler workaround.
  MIPS: BCM63xx: Fix build failure in board_bcm963xx.c
  MIPS: uasm: Add OR instruction.
  MIPS: Sibyte: Apply M3 workaround only on affected chip types and versions.
  MIPS: BCM63xx: Initialize gpio_out_low & out_high to current value at boot.
  MIPS: BCM63xx: Register SSB SPROM fallback in board's first stage callback
  MIPS: BCM63xx: Fix typo in cpu-feature-overrides file.
  MIPS: BCM63xx: Add support for second uart.
  MIPS: BCM63xx: Fix double gpio registration.
  MIPS: BCM63xx: Add DWVS0 board
  MIPS: BCM63xx: Add the RTA1025W-16 BCM6348-based board to suppported boards.
  MIPS: BCM63xx: Fix BCM6338 and BCM6345 gpio count
  MIPS: libgcc.h: Checkpatch cleanup
  MIPS: Loongson-2F: Flush the branch target history in BTB and RAS
  MIPS: Move signal trampolines off of the stack.
  MIPS: Preliminary VDSO
  ...

14 years agoMerge branch 'for-2.6.34' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Tue, 13 Apr 2010 01:34:56 +0000 (18:34 -0700)]
Merge branch 'for-2.6.34' of git://linux-nfs.org/~bfields/linux

* 'for-2.6.34' of git://linux-nfs.org/~bfields/linux:
  svcrdma: RDMA support not yet compatible with RPC6

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke...
Linus Torvalds [Tue, 13 Apr 2010 01:34:25 +0000 (18:34 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ryusuke/nilfs2

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke/nilfs2:
  nilfs2: fix typo "numer" -> "number" in alloc.c
  nilfs2: Remove an uninitialization warning in nilfs_btree_propagate_v()
  nilfs2: fix a wrong type conversion in nilfs_ioctl()

14 years agoanonvma: when setting up page->mapping, we need to pick the _oldest_ anonvma
Linus Torvalds [Mon, 12 Apr 2010 19:44:29 +0000 (12:44 -0700)]
anonvma: when setting up page->mapping, we need to pick the _oldest_ anonvma

Otherwise we might be mapping in a page in a new mapping, but that page
(through the swapcache) would later be mapped into an old mapping too.
The page->mapping must be the case that works for everybody, not just
the mapping that happened to page it in first.

Here's the scenario:

 - page gets allocated/mapped by process A. Let's call the anon_vma we
   associate the page with 'A' to keep it easy to track.

 - Process A forks, creating process B. The anon_vma in B is 'B', and has
   a chain that looks like 'B' -> 'A'. Everything is fine.

 - Swapping happens. The page (with mapping pointing to 'A') gets swapped
   out (perhaps not to disk - it's enough to assume that it's just not
   mapped any more, and lives entirely in the swap-cache)

 - Process B pages it in, which goes like this:

        do_swap_page ->
          page = lookup_swap_cache(entry);
         ...
          set_pte_at(mm, address, page_table, pte);
          page_add_anon_rmap(page, vma, address);

   And think about what happens here!

   In particular, what happens is that this will now be the "first"
   mapping of that page, so page_add_anon_rmap() used to do

        if (first)
                __page_set_anon_rmap(page, vma, address);

   and notice what anon_vma it will use? It will use the anon_vma for
   process B!

   What happens then? Trivial: process 'A' also pages it in (nothing
   happens, it's not the first mapping), and then process 'B' execve's
   or exits or unmaps, making anon_vma B go away.

   End result: process A has a page that points to anon_vma B, but
   anon_vma B does not exist any more.  This can go on forever.  Forget
   about RCU grace periods, forget about locking, forget anything like
   that.  The bug is simply that page->mapping points to an anon_vma
   that was correct at one point, but was _not_ the one that was shared
   by all users of that possible mapping.

Changing it to always use the deepest anon_vma in the anonvma chain gets
us to the safest model.

This can be improved in certain cases: if we know the page is private to
just this particular mapping (for example, it's a new page, or it is the
only swapcache entry), we could pick the top (most specific) anon_vma.

But that's a future optimization. Make it _work_ reliably first.

Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Tested-by: Borislav Petkov <bp@alien8.de> [ "What do you know, I think you fixed it!" ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoanon_vma: clone the anon_vma chain in the right order
Linus Torvalds [Mon, 12 Apr 2010 00:15:03 +0000 (17:15 -0700)]
anon_vma: clone the anon_vma chain in the right order

We want to walk the chain in reverse order when cloning it, so that the
order of the result chain will be the same as the order in the source
chain.  When we add entries to the chain, they go at the head of the
chain, so we want to add the source head last.

Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Tested-by: Borislav Petkov <bp@alien8.de> [ "No, it still oopses" ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agovma_adjust: fix the copying of anon_vma chains
Linus Torvalds [Sat, 10 Apr 2010 22:22:30 +0000 (15:22 -0700)]
vma_adjust: fix the copying of anon_vma chains

When we move the boundaries between two vma's due to things like
mprotect, we need to make sure that the anon_vma of the pages that got
moved from one vma to another gets properly copied around.  And that was
not always the case, in this rather hard-to-follow code sequence.

Clarify the code, and fix it so that it copies the anon_vma from the
right source.

Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Tested-by: Borislav Petkov <bp@alien8.de> [ "Yeah, not so much this one either" ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoSimplify and comment on anon_vma re-use for anon_vma_prepare()
Linus Torvalds [Sat, 10 Apr 2010 17:36:19 +0000 (10:36 -0700)]
Simplify and comment on anon_vma re-use for anon_vma_prepare()

This changes the anon_vma reuse case to require that we only reuse
simple anon_vma's - ie the case when the vma only has a single anon_vma
associated with it.

This means that a reuse of an anon_vma from an adjacent vma will always
guarantee that both vma's are associated not only with the same
anon_vma, they will also have the same anon_vma chain (of just a single
entry in this case).

And since anon_vma re-use was the only case where the same anon_vma
might be associated with different chains of anon_vma's, we now have the
case that every vma that shares the same anon_vma will always also have
the same chain.  That makes it much easier to think about merging vma's
that share the same anon_vma's: you can always just drop the other
anon_vma chain in anon_vma_merge() since you know that they are always
identical.

This also splits up the function to validate the anon_vma re-use, and
adds a lot of commentary about the possible races.

Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Tested-by: Borislav Petkov <bp@alien8.de> [ "That didn't fix it" ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoquota: Fix possible dq_flags corruption
Andrew Perepechko [Mon, 12 Apr 2010 18:16:50 +0000 (22:16 +0400)]
quota: Fix possible dq_flags corruption

dq_flags are modified non-atomically in do_set_dqblk via __set_bit calls and
atomically for example in mark_dquot_dirty or clear_dquot_dirty.  Hence a
change done by an atomic operation can be overwritten by a change done by a
non-atomic one. Fix the problem by using atomic bitops even in do_set_dqblk.

Signed-off-by: Andrew Perepechko <andrew.perepechko@sun.com>
Signed-off-by: Jan Kara <jack@suse.cz>
14 years agoquota: Hide warnings about writes to the filesystem before quota was turned on
Jan Kara [Tue, 6 Apr 2010 16:52:47 +0000 (18:52 +0200)]
quota: Hide warnings about writes to the filesystem before quota was turned on

For a root filesystem write to the filesystem before quota is turned on happens
regularly and there's no way around it because of writes to syslog, /etc/mtab,
and similar. So the warning is rather pointless for ordinary users. It's
still useful during development so we just hide the warning behind
__DQUOT_PARANOIA config option.

Signed-off-by: Jan Kara <jack@suse.cz>
14 years agoext3: symlink must be handled via filesystem specific operation
Dmitry Monakhov [Fri, 26 Mar 2010 16:29:55 +0000 (19:29 +0300)]
ext3: symlink must be handled via filesystem specific operation

generic setattr implementation is no longer responsible for
quota transfer so synlinks must be handled via ext3_setattr.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jan Kara <jack@suse.cz>
14 years agoext2: symlink must be handled via filesystem specific operation
Dmitry Monakhov [Fri, 26 Mar 2010 16:29:54 +0000 (19:29 +0300)]
ext2: symlink must be handled via filesystem specific operation

generic setattr implementation is no longer responsible for
quota transfer so synlinks must be handled via ext2_setattr.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jan Kara <jack@suse.cz>
14 years agoMIPS: Calculate proper ebase value for 64-bit kernels
David Daney [Tue, 6 Apr 2010 20:29:50 +0000 (13:29 -0700)]
MIPS: Calculate proper ebase value for 64-bit kernels

The ebase is relative to CKSEG0 not CAC_BASE.  On a 32-bit kernel they
are the same thing, for a 64-bit kernel they are not.

It happens to kind of work on a 64-bit kernel as they both reference
the same physical memory.  However since the CPU uses the CKSEG0 base,
determining if a J instruction will reach always gives the wrong result
unless we use the same number the CPU uses.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1093/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Alchemy: DB1200: Remove custom wait implementation
Manuel Lauss [Sat, 3 Apr 2010 15:07:03 +0000 (17:07 +0200)]
MIPS: Alchemy: DB1200: Remove custom wait implementation

While playing with the out-of-tree MAE driver module, the system would
panic after a while in the db1200 custom wait code after wakeup due to
a clobbered k0 register being used as target address of a store op.

Remove the custom wait implementation and revert back to the Alchemy-
recommended implementation already set as default.

Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
To: Linux-MIPS <linux-mips@linux-mips.org>
Patchwork: http://patchwork.linux-mips.org/patch/1092/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Big Sur: Make defconfig more useful.
Ralf Baechle [Sat, 3 Apr 2010 09:59:52 +0000 (10:59 +0100)]
MIPS: Big Sur: Make defconfig more useful.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Fix __vmalloc() etc. on MIPS for non-GPL modules
Anton Altaparmakov [Thu, 25 Mar 2010 20:48:12 +0000 (20:48 +0000)]
MIPS: Fix __vmalloc() etc. on MIPS for non-GPL modules

Commit b3594a089f1c17ff919f8f78505c3f20e1f6f8ce (lmo) rsp.
351336929ccf222ae38ff0cb7a8dd5fd5c6236a0 (kernel.org) break non-GPL modules
that use __vmalloc() or any of the vmap(), vm_map_ram(), etc functions on
MIPS.

All those functions are EXPORT_SYMBOL() so are meant to be allowed to be
used by non-GPL kernel modules.  These calls all take page protection as
an argument which is normally a constant like PAGE_KERNEL.

This commit causes all protection constants like PAGE_KERNEL to not be
constants and instead to contain the GPL-only symbol _page_cachable_default.

This means that all calls to __vmalloc(), vmap(), etc, cause non-GPL
modules to fail to link with the complaint that they are trying to use the
GPL-only symbol _page_cachable_default...

Change EXPORT_SYMBOL_GPL(_page_cachable_default) to EXPORT_SYMBOL() for
non-GPL modules that call __vmalloc(), vmap(), vm_map_ram() etc.

Signed-off-by: Anton Altaparmakov <aia21@cantab.net>
Cc: Chris Dearman <chris@mips.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: http://patchwork.linux-mips.org/patch/1084/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Sibyte: Fix M3 TLB exception handler workaround.
Ralf Baechle [Tue, 23 Mar 2010 16:56:38 +0000 (17:56 +0100)]
MIPS: Sibyte: Fix M3 TLB exception handler workaround.

The M3 workaround needs to cmpare the region and VPN2 fields only.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Fix build failure in board_bcm963xx.c
Florian Fainelli [Tue, 23 Mar 2010 09:30:08 +0000 (10:30 +0100)]
MIPS: BCM63xx: Fix build failure in board_bcm963xx.c

Since 2083e8327aeeaf818b0e4522a9d2539835c60423, the SPROM is now registered
in the board_prom_init callback, but it references variables and functions
which are declared below.  Move the variables and functions above
board_prom_init.

Signed-off-by: Florian Fainelli <ffainelli@freebox.fr>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1077/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: uasm: Add OR instruction.
Ralf Baechle [Tue, 23 Mar 2010 14:54:50 +0000 (15:54 +0100)]
MIPS: uasm: Add OR instruction.

This is needed for the fix of the M3 workaround.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Sibyte: Apply M3 workaround only on affected chip types and versions.
Ralf Baechle [Mon, 22 Mar 2010 23:02:43 +0000 (00:02 +0100)]
MIPS: Sibyte: Apply M3 workaround only on affected chip types and versions.

Previously it was unconditionally used on all Sibyte family SOCs.  The
M3 bug has to be handled in the TLB exception handler which is extremly
performance sensitive, so this modification is expected to deliver around
2-3% performance improvment.  This is important as required changes to the
M3 workaround will make it more costly.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Initialize gpio_out_low & out_high to current value at boot.
Maxime Bizon [Sat, 30 Jan 2010 17:34:58 +0000 (18:34 +0100)]
MIPS: BCM63xx: Initialize gpio_out_low & out_high to current value at boot.

To avoid a glitch during GPIO initialisation read GPIO output register
values left by the firmware.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/903/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Register SSB SPROM fallback in board's first stage callback
Florian Fainelli [Tue, 2 Mar 2010 13:38:47 +0000 (14:38 +0100)]
MIPS: BCM63xx: Register SSB SPROM fallback in board's first stage callback

Signed-off-by: Florian Fainelli <ffainelli@freebox.fr>
To: Maxime Bizon <mbizon@freebox.fr>
Cc: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1017/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Fix typo in cpu-feature-overrides file.
Maxime Bizon [Sat, 30 Jan 2010 17:34:56 +0000 (18:34 +0100)]
MIPS: BCM63xx: Fix typo in cpu-feature-overrides file.

Fix typo: CONFIG_BCMCPU_IS_63xx does not exist;
CONFIG_BCM63XX_CPU_63xx is the valid config option.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
To: linux-mips@linux-mips.org
Cc: Maxime Bizon <mbizon@freebox.fr>
Patchwork: http://patchwork.linux-mips.org/patch/901/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Add support for second uart.
Maxime Bizon [Sat, 30 Jan 2010 17:34:55 +0000 (18:34 +0100)]
MIPS: BCM63xx: Add support for second uart.

The BCm63xx SOC has two uarts.  Some boards use the second one for
bluetooth.  This patch changes platform device registration code to
handle this.  Changes to the UART driver were already merged in
6a2c7eabfd09ca7986bf96b8958a87ca041a19d8.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
To: linux-mips@linux-mips.org
Cc: Maxime Bizon <mbizon@freebox.fr>
Patchwork: http://patchwork.linux-mips.org/patch/900/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Fix double gpio registration.
Maxime Bizon [Sat, 30 Jan 2010 17:34:54 +0000 (18:34 +0100)]
MIPS: BCM63xx: Fix double gpio registration.

bcm63xx_gpio_init is already called from prom_init to allow board to use
them early, so we can remove the unneeded arch_initcall.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
To: linux-mips@linux-mips.org
Cc: Maxime Bizon <mbizon@freebox.fr>
Patchwork: http://patchwork.linux-mips.org/patch/899/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Add DWVS0 board
Florian Fainelli [Mon, 1 Mar 2010 22:36:27 +0000 (23:36 +0100)]
MIPS: BCM63xx: Add DWVS0 board

The DWVS0 board is a BCM6358-based board with an on-board OHCI controler.

Signed-off-by: Florian Fainelli <florian@openwrt.org>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1015/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Add the RTA1025W-16 BCM6348-based board to suppported boards.
Florian Fainelli [Mon, 1 Mar 2010 22:36:22 +0000 (23:36 +0100)]
MIPS: BCM63xx: Add the RTA1025W-16 BCM6348-based board to suppported boards.

Signed-off-by: Florian Fainelli <florian@openwrt.org>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1014/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: BCM63xx: Fix BCM6338 and BCM6345 gpio count
Florian Fainelli [Mon, 1 Mar 2010 22:36:32 +0000 (23:36 +0100)]
MIPS: BCM63xx: Fix BCM6338 and BCM6345 gpio count

The number of GPIOs on BCM6338 is 8, while BCM6345 has only 16 GPIOs
available.

Signed-off-by: Florian Fainelli <florian@openwrt.org>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1016/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: libgcc.h: Checkpatch cleanup
Andrea Gelmini [Sat, 27 Feb 2010 16:51:23 +0000 (17:51 +0100)]
MIPS: libgcc.h: Checkpatch cleanup

arch/mips/lib/libgcc.h:21: ERROR: open brace '{' following union go on the same line

Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
To: linux-kernel@vger.kernel.org
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: linux-mips@linux-mips.org
Cc: linux-sh@vger.kernel.org
Patchwork: http://patchwork.linux-mips.org/patch/1007/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Loongson-2F: Flush the branch target history in BTB and RAS
Wu Zhangjin [Sat, 13 Mar 2010 04:34:15 +0000 (12:34 +0800)]
MIPS: Loongson-2F: Flush the branch target history in BTB and RAS

As per chapter 15 "Errata: Issue of Out-of-order in loongson"[1] to work
around the Loongson 2F erratum we need to do:

"When switching from user mode to kernel mode, you should flush the
branch target history such as BTB and RAS."

[1] Chinese version: http://www.loongson.cn/uploadfile/file/200808211
[2] English version of chapter 15:
    http://groups.google.com.hk/group/loongson-dev/msg/e0d2e220958f10a6?dmode=source

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips@linux-mips.org
Cc: Shinya Kuribayashi <shinya.kuribayashi@necel.com>
Patchwork: http://patchwork.linux-mips.org/patch/1066/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Move signal trampolines off of the stack.
David Daney [Fri, 19 Feb 2010 00:13:05 +0000 (16:13 -0800)]
MIPS: Move signal trampolines off of the stack.

This is a follow on to the vdso patch.

Since all processes now have signal trampolines permanently mapped, we
can use those instead of putting the trampoline on the stack and
invalidating the corresponding icache across all CPUs.  We also get rid
of a bunch of ICACHE_REFILLS_WORKAROUND_WAR code.

[Ralf: GDB 7.1 which has the necessary modifications to allow backtracing
over signal frames will supposedly be released tomorrow.  The old signal
frame format obsoleted by this patch exists in two variations, for sane
processors and for those requiring ICACHE_REFILLS_WORKAROUND_WAR.  So
there was never a GDB which did support backtracing over signal frames
on all MIPS systems.  This convinved me this series should be applied and
pushed upstream as soon as possible.]

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/974/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Preliminary VDSO
David Daney [Fri, 19 Feb 2010 00:13:04 +0000 (16:13 -0800)]
MIPS: Preliminary VDSO

This is a preliminary patch to add a vdso to all user processes.  Still
missing are ELF headers and .eh_frame information.  But it is enough to
allow us to move signal trampolines off of the stack.  Note that emulation
of branch delay slots in the FPU emulator still requires the stack.

We allocate a single page (the vdso) and write all possible signal
trampolines into it.  The stack is moved down by one page and the vdso is
mapped into this space.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/975/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Add SYSCALL to uasm.
David Daney [Fri, 19 Feb 2010 00:13:03 +0000 (16:13 -0800)]
MIPS: Add SYSCALL to uasm.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/976/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Fix SSB PCIcore IO resource management
Andreas Ferber [Tue, 16 Mar 2010 11:35:51 +0000 (12:35 +0100)]
MIPS: Fix SSB PCIcore IO resource management

The SSB PCIcore code reused the IO resource fixup code from the original
2.4.x Broadcom patch for BCM47xx based devices, which was a quick hack
for doing PCI IO resource configuration back then (the boot loader
doesn't configure PCI devices on this platform).

However, this code is no longer necessary since the kernel now can do
PCI resource management fine all by itself, so remove the old code.

When removing the code, it becomes obvious that the mem_offset setting
in the PCIcore driver was wrong, however this was masked by the fixup
code before, except in a few cases involving yenta_socket. For BCM47xx,
the correct offset is 0, and since this is the only device using PCIcore
in host mode, the offset can simply be removed unconditionally.

Signed-off-by: Andreas Ferber <af@chaos-agency.de>
Signed-off-by: Michael Buesch <mb@bu3sch.de>
Cc: Markus Wigge <markus@cultcom.de>
Cc: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1070/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: make CAC_ADDR and UNCAC_ADDR account for PHYS_OFFSET
Florian Fainelli [Wed, 10 Mar 2010 08:51:09 +0000 (09:51 +0100)]
MIPS: make CAC_ADDR and UNCAC_ADDR account for PHYS_OFFSET

On AR7, we already redefine PHYS_OFFSET to match the system specifities, it
is however not sufficient when unsing dma_{map,unmap}_single, specifically
in the ethernet driver, we must also adjust CAC_ADDR and UNCAC_ADDR for DMA
to work correctly. This patch fixes the following issue, seen in cpmac_open:

ops[#1]:
Cpu 0
$ 0   : 00000000 10008400 a0f5b120 00000000
$ 4   : 94c59000 94270f64 00000020 00000010
$ 8   : 00000010 94103ce0 0000000a 94c03400
$12   : ffffffff 94c03408 94c03410 00000001
$16   : a0f5ba20 00000041 94c592c0 94c59200
$20   : 94c59000 000005ee 00002000 9438c8f0
$24   : 00000010 00000000
$28   : 94fac000 94fadd58 94390000 942724a8
Hi    : 00000000
Lo    : 00000001
epc   : 94272518 cpmac_open+0x208/0x3f8
    Not tainted
ra    : 942724a8 cpmac_open+0x198/0x3f8
Status: 10008403    KERNEL EXL IE
Cause : 3080000c
BadVA : 00000000
PrId  : 00018448 (MIPS 4KEc)
Modules linked in:
Process ifconfig (pid: 278, threadinfo=94fac000, task=94e79590, tls=00000000)
Stack : 7f8da120 2ab05cb0 94c59000 943356f0 00000000 943d0000 94c59000 943356f0
        94c59030 943d0000 943c27c0 94fade10 00000000 94fade20 94c59000 9428e5a4
        00000000 94c59000 00000041 94289768 94c59000 00000041 00001002 00001043
        00000000 9428d810 00000000 94fade10 7f8da4e8 9428e6b8 00000000 7f8da4a8
        7f8da4e8 00008914 00000000 942f7f2c 00000000 00000008 00408000 00008913
        ...
Call Trace:
[<94272518>] cpmac_open+0x208/0x3f8
[<9428e5a4>] dev_open+0x164/0x264
[<9428d810>] dev_change_flags+0xd0/0x1bc
[<942f7f2c>] devinet_ioctl+0x2d8/0x908
[<942771f8>] sock_ioctl+0x29c/0x2fc
[<941a0fb4>] vfs_ioctl+0x2c/0x7c
[<941a16ec>] do_vfs_ioctl+0x5dc/0x630
[<941a1790>] sys_ioctl+0x50/0x88
[<94101e10>] stack_done+0x20/0x3c

Signed-off-by: peter fuerst <post@pfrst.de>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1050/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: AR7: Fix phat finger of cpmac fixed_phy_add
Alexander Clouter [Sat, 13 Mar 2010 00:09:15 +0000 (00:09 +0000)]
MIPS: AR7: Fix phat finger of cpmac fixed_phy_add

Seems I trimmed one too many lines in
29ca2d81bd2a62fa86bc9a72ddadcf03d7daf795 (lmo) rsp
7084338eb8eb0cc021ba86c340157bad397f3f0b (kernel.org) which led to no
functioning Ethernet on my WAG54Gv2.  This patch restores the AWOL line.

Signed-off-by: Alexander Clouter <alex@digriz.org.uk>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1065/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: AR7: Fix phat finger of reset bit in vlynq_high_data
Alexander Clouter [Fri, 12 Mar 2010 19:39:48 +0000 (19:39 +0000)]
MIPS: AR7: Fix phat finger of reset bit in vlynq_high_data

Seems in my whitespace cleanup 0f2536082d01448daeced8d9e82c3ba1751fefa3
(lmo) rsp.  8c2961da46abd85a71d20f2b169bf80618e (kernel.org) caused AR7
to no longer get as far as init.  Fixed my phat fingering.

Signed-off-by: Alexander Clouter <alex@digriz.org.uk>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1064/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Loongson: Add module info to the loongson2_clock driver
Wu Zhangjin [Tue, 9 Mar 2010 16:53:21 +0000 (00:53 +0800)]
MIPS: Loongson: Add module info to the loongson2_clock driver

This patch fixes a kernel warning when loading the the loongson2_clock
driver:

"Feb 25 23:42:27 localhost kernel: [    4.965000] loongson2_clock: module
license 'unspecified' taints kernel."

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Reported-by: Liu Shiwei <liushiwei@gmail.com>
Cc: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1045/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Lemote 2F: Ensure atomic execution of _rdmsr and _wrmsr
Wu Zhangjin [Thu, 11 Mar 2010 03:30:50 +0000 (11:30 +0800)]
MIPS: Lemote 2F: Ensure atomic execution of _rdmsr and _wrmsr

On Lemote 2F CS5536 MSRs are accessed through a index / data register pair.
The access sequence must be protected by a spinlock to be atomic.

Without this rebooting in fs2f_reboot() may fail.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips@linux-mips.org
Cc: David Daney <ddaney@caviumnetworks.com>
Patchwork: http://patchwork.linux-mips.org/patch/1058/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Initialize an atomic_t properly with ATOMIC_INIT(0).
Robert P. J. Day [Sat, 27 Feb 2010 17:02:51 +0000 (12:02 -0500)]
MIPS: Initialize an atomic_t properly with ATOMIC_INIT(0).

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1008/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Fix elfcore.c build warning
Ralf Baechle [Thu, 11 Mar 2010 07:48:14 +0000 (08:48 +0100)]
MIPS: Fix elfcore.c build warning

kernel/elfcore.c includes <linux/elf.h> which includes the <asm/elf.h>.  In
<asm/elf.h>, struct pt_regs is declared inside the parameter list of the
elf_dump_regs function which causes a kernel build warning.

Fixed by adding a forward declaration of struct pt_regs.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Octeon: Remove redundant declaration of octeon_reserve32_memory
Yang Shi [Wed, 3 Mar 2010 08:43:20 +0000 (16:43 +0800)]
MIPS: Octeon: Remove redundant declaration of octeon_reserve32_memory

octeon_reserve32_memory is defined In Octeon's setup.c, so remove the
redundant extern declaration of this variable.

Signed-off-by: Yang Shi <yang.shi@windriver.com>
To: f.fainelli@gmail.com
Cc: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1022/
Acked-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Trace: Don't trace irqsoff for the idle process
Wu Zhangjin [Tue, 9 Mar 2010 16:27:28 +0000 (00:27 +0800)]
MIPS: Trace: Don't trace irqsoff for the idle process

Like x86 did in arch/x86/kernel/{process_32.c,process_64.c}, also don't
trace irqsoff for idle.

If there's no useful work to be done, we don't care about the irqsoff
duration. If we trace the idle process, the max duration of irqsoff will
be the idle time and make the irqsoff tracer useless.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Cc: linux-mips@linux-mips.org
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Patchwork: http://patchwork.linux-mips.org/patch/1044/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: delay: Fix use of current_cpu_data in preemptable code.
Ralf Baechle [Wed, 10 Mar 2010 15:16:04 +0000 (16:16 +0100)]
MIPS: delay: Fix use of current_cpu_data in preemptable code.

This may lead to warnings like:

BUG: using smp_processor_id() in preemptible [00000000] code: reboot/1989
caller is __udelay+0x14/0x70
Call Trace:
[<ffffffff8110ad28>] dump_stack+0x8/0x34
[<ffffffff812dde04>] debug_smp_processor_id+0xf4/0x110
[<ffffffff812d90bc>] __udelay+0x14/0x70
[<ffffffff81378274>] md_notify_reboot+0x12c/0x148
[<ffffffff81161054>] notifier_call_chain+0x64/0xc8
[<ffffffff811614dc>] __blocking_notifier_call_chain+0x64/0xc0
[<ffffffff8115566c>] kernel_restart_prepare+0x1c/0x38
[<ffffffff811556cc>] kernel_restart+0x14/0x50
[<ffffffff8115581c>] SyS_reboot+0x10c/0x1f0
[<ffffffff81103684>] handle_sysn32+0x44/0x84

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Octeon: Remove #if 0 code.
David Daney [Wed, 3 Mar 2010 19:07:43 +0000 (11:07 -0800)]
MIPS: Octeon: Remove #if 0 code.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1029/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Octeon: Remove vestiges of CONFIG_CAVIUM_RESERVE32_USE_WIRED_TLB
David Daney [Wed, 3 Mar 2010 19:07:07 +0000 (11:07 -0800)]
MIPS: Octeon: Remove vestiges of CONFIG_CAVIUM_RESERVE32_USE_WIRED_TLB

The config option CAVIUM_RESERVE32_USE_WIRED_TLB is not supported.
Remove the dead code controlled by it.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/1028/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Cavium: Remove unused watchdog code.
Ralf Baechle [Tue, 9 Mar 2010 19:16:15 +0000 (20:16 +0100)]
MIPS: Cavium: Remove unused watchdog code.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
14 years agoMIPS: Fix build breakage if CONFIG_DEBUG_FS is enabled.
Ralf Baechle [Mon, 8 Mar 2010 18:38:13 +0000 (19:38 +0100)]
MIPS: Fix build breakage if CONFIG_DEBUG_FS is enabled.

Caused by 38b7827fcdd660f591d645bd3ae6644456a4773c - no, cpu_local_* was
not unused.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Cc: Christoph Lameter <cl@linux-foundation.org>
Acked-by: David Daney <ddaney@caviumnetworks.com>
14 years agonilfs2: fix typo "numer" -> "number" in alloc.c
Ryusuke Konishi [Sun, 11 Apr 2010 16:51:03 +0000 (01:51 +0900)]
nilfs2: fix typo "numer" -> "number" in alloc.c

Fixes the typo found in a warning message of a persistent object
allocator function.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Fri, 9 Apr 2010 18:53:06 +0000 (11:53 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/mlx4: Check correct variable for allocation failure
  RDMA/nes: Correct cap.max_inline_data assignment in nes_query_qp()
  RDMA/cm: Set num_paths when manually assigning path records
  IB/cm: Fix device_create() return value check

14 years agoMerge branch 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Fri, 9 Apr 2010 18:52:48 +0000 (11:52 -0700)]
Merge branch 'for-linus' of git://git390.marist.edu/linux-2.6

* 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6:
  [S390] Update default configuration.
  [S390] nss: add missing .previous statement to asm function
  [S390] increase default size of vmalloc area
  [S390] s390: disable change bit override
  [S390] fix io_return critical section cleanup
  [S390] sclp_async: potential buffer overflow
  [S390] arch/s390/kernel: Add missing unlock

14 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Fri, 9 Apr 2010 18:50:29 +0000 (11:50 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block: (34 commits)
  cfq-iosched: Fix the incorrect timeslice accounting with forced_dispatch
  loop: Update mtime when writing using aops
  block: expose the statistics in blkio.time and blkio.sectors for the root cgroup
  backing-dev: Handle class_create() failure
  Block: Fix block/elevator.c elevator_get() off-by-one error
  drbd: lc_element_by_index() never returns NULL
  cciss: unlock on error path
  cfq-iosched: Do not merge queues of BE and IDLE classes
  cfq-iosched: Add additional blktrace log messages in CFQ for easier debugging
  i2o: Remove the dangerous kobj_to_i2o_device macro
  block: remove 16 bytes of padding from struct request on 64bits
  cfq-iosched: fix a kbuild regression
  block: make CONFIG_BLK_CGROUP visible
  Remove GENHD_FL_DRIVERFS
  block: Export max number of segments and max segment size in sysfs
  block: Finalize conversion of block limits functions
  block: Fix overrun in lcm() and move it to lib
  vfs: improve writeback_inodes_wb()
  paride: fix off-by-one test
  drbd: fix al-to-on-disk-bitmap for 4k logical_block_size
  ...

14 years agoMerge branch 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Fri, 9 Apr 2010 18:50:01 +0000 (11:50 -0700)]
Merge branch 'drm-linus' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (29 commits)
  drm/nouveau: bail out of auxch transaction if we repeatedly recieve defers
  drm/nv50: implement gpio set/get routines
  drm/nv50: parse/use some more de-magiced parts of gpio table entries
  drm/nouveau: store raw gpio table entry in bios gpio structs
  drm/nv40: Init some tiling-related PGRAPH state.
  drm/nv50: Add NVA3 support in ctxprog/ctxvals generator.
  drm/nv50: another dodgy DP hack
  drm/nv50: punt hotplug irq handling out to workqueue
  drm/nv50: preserve an unknown SOR_MODECTRL value for DP encoders
  drm/nv50: Allow using the NVA3 new compute class.
  drm/nv50: cleanup properly if PDISPLAY init fails
  drm/nouveau: fixup the init failure paths some more
  drm/nv50: fix instmem init on IGPs if stolen mem crosses 4GiB mark
  drm/nv40: add LVDS table quirk for Dell Latitude D620
  drm/nv40: rework lvds table parsing
  drm/nouveau: detect vram amount once, and save the value
  drm/nouveau: remove some unused members from drm_nouveau_private
  drm/nouveau: Make use of TTM busy_placements.
  drm/nv50: add more 0x100c80 flushy magic
  drm/nv50: fix fbcon when framebuffer above 4GiB mark
  ...

14 years agoradix_tree_tag_get() is not as safe as the docs make out [ver #2]
David Howells [Tue, 6 Apr 2010 21:36:20 +0000 (22:36 +0100)]
radix_tree_tag_get() is not as safe as the docs make out [ver #2]

radix_tree_tag_get() is not safe to use concurrently with radix_tree_tag_set()
or radix_tree_tag_clear().  The problem is that the double tag_get() in
radix_tree_tag_get():

if (!tag_get(node, tag, offset))
saw_unset_tag = 1;
if (height == 1) {
int ret = tag_get(node, tag, offset);

may see the value change due to the action of set/clear.  RCU is no protection
against this as no pointers are being changed, no nodes are being replaced
according to a COW protocol - set/clear alter the node directly.

The documentation in linux/radix-tree.h, however, says that
radix_tree_tag_get() is an exception to the rule that "any function modifying
the tree or tags (...) must exclude other modifications, and exclude any
functions reading the tree".

The problem is that the next statement in radix_tree_tag_get() checks that the
tag doesn't vary over time:

BUG_ON(ret && saw_unset_tag);

This has been seen happening in FS-Cache:

https://www.redhat.com/archives/linux-cachefs/2010-April/msg00013.html

To this end, remove the BUG_ON() from radix_tree_tag_get() and note in various
comments that the value of the tag may change whilst the RCU read lock is held,
and thus that the return value of radix_tree_tag_get() may not be relied upon
unless radix_tree_tag_set/clear() and radix_tree_delete() are excluded from
running concurrently with it.

Reported-by: Romain DEGEZ <romain.degez@smartjog.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoslub: Fix kmem_ptr_validate() for non-kernel pointers
Pekka Enberg [Wed, 7 Apr 2010 16:23:41 +0000 (19:23 +0300)]
slub: Fix kmem_ptr_validate() for non-kernel pointers

As suggested by Linus, fix up kmem_ptr_validate() to handle non-kernel pointers
more graciously. The patch changes kmem_ptr_validate() to use the newly
introduced kern_ptr_validate() helper to check that a pointer is a valid kernel
pointer before we attempt to convert it into a 'struct page'.

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Acked-by: Christoph Lameter <cl@linux-foundation.org>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoslab: Generify kernel pointer validation
Pekka Enberg [Wed, 7 Apr 2010 16:23:40 +0000 (19:23 +0300)]
slab: Generify kernel pointer validation

As suggested by Linus, introduce a kern_ptr_validate() helper that does some
sanity checks to make sure a pointer is a valid kernel pointer.  This is a
preparational step for fixing SLUB kmem_ptr_validate().

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoRevert "memory-hotplug: add 0x prefix to HEX block_size_bytes"
Linus Torvalds [Fri, 9 Apr 2010 17:05:33 +0000 (10:05 -0700)]
Revert "memory-hotplug: add 0x prefix to HEX block_size_bytes"

This reverts commit ba168fc37dea145deeb8fa9e7e71c748d2e00d74.

It changes user-visible sysfs interfaces, and breaks some existing user
space applications which apparently rely on the fact that the output
does not contain the "0x" prefix.

Requested-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branches 'cma', 'misc', 'mlx4' and 'nes' into for-linus
Roland Dreier [Fri, 9 Apr 2010 16:14:21 +0000 (09:14 -0700)]
Merge branches 'cma', 'misc', 'mlx4' and 'nes' into for-linus

14 years agoARM: Fix ioremap_cached()/ioremap_wc() for SMP platforms
Russell King [Fri, 9 Apr 2010 14:00:11 +0000 (15:00 +0100)]
ARM: Fix ioremap_cached()/ioremap_wc() for SMP platforms

Write combining/cached device mappings are not setting the shared bit,
which could potentially cause problems on SMP systems since the cache
lines won't participate in the cache coherency protocol.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
14 years ago[S390] Update default configuration.
Martin Schwidefsky [Fri, 9 Apr 2010 11:43:04 +0000 (13:43 +0200)]
[S390] Update default configuration.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
14 years ago[S390] nss: add missing .previous statement to asm function
Heiko Carstens [Fri, 9 Apr 2010 11:43:03 +0000 (13:43 +0200)]
[S390] nss: add missing .previous statement to asm function

The savesys_ipl_nss asm function is put into the .init.text section
however it is missing a ".previous" section which would restore the
previous section.
Luckily all functions in early.c are init functions so it doesn't
matter currently.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
14 years ago[S390] increase default size of vmalloc area
Martin Schwidefsky [Fri, 9 Apr 2010 11:43:02 +0000 (13:43 +0200)]
[S390] increase default size of vmalloc area

The default size of the vmalloc area is currently 1 GB. The memory resource
controller uses about 10 MB of vmalloc space per gigabyte of memory. That
turns a system with more than ~100 GB memory unbootable with the default
vmalloc size. It costs us nothing to increase the default size to some
more adequate value, e.g. 128 GB.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
14 years ago[S390] s390: disable change bit override
Christian Borntraeger [Fri, 9 Apr 2010 11:43:01 +0000 (13:43 +0200)]
[S390] s390: disable change bit override

commit 6a985c6194017de2c062916ad1cd00dee0302c40
([S390] s390: use change recording override for kernel mapping)
deactivated the change bit recording for the kernel mapping to
improve the performance. This works most of the time, but there
are cases (e.g. kernel runs in home space, futex atomic compare xcmg)
where we modify user memory with the kernel mapping instead of the
user mapping.
Instead of fixing these cases, this patch just deactivates change bit
override to avoid future problems with other kernel code that might
use the kernel mapping for user memory.

CC: stable@kernel.org
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
14 years ago[S390] fix io_return critical section cleanup
Martin Schwidefsky [Fri, 9 Apr 2010 11:43:00 +0000 (13:43 +0200)]
[S390] fix io_return critical section cleanup

If a machine check interrupts the io interrupt handler on one of the
instructions between io_return and io_leave the critical section
cleanup code will move the return psw to io_work_loop. By doing that
the switch from the asynchronous interrupt stack to the process stack
is skipped. If e.g. TIF_NEED_RESCHED is set things break because
the scheduler is called with the asynchronous interrupts stack.
Moving the psw back to io_return instead fixes the problem.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
14 years ago[S390] sclp_async: potential buffer overflow
Dan Carpenter [Fri, 9 Apr 2010 11:42:59 +0000 (13:42 +0200)]
[S390] sclp_async: potential buffer overflow

"len" hasn't been properly range checked so we shouldn't use it as an
array offset.  This can only be written to by root but it would still be
annoying to accidentally write more than 3 characters and corrupt your
memory.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
14 years ago[S390] arch/s390/kernel: Add missing unlock
Julia Lawall [Fri, 9 Apr 2010 11:42:58 +0000 (13:42 +0200)]
[S390] arch/s390/kernel: Add missing unlock

In the default case the lock is not unlocked.  The return is
converted to a goto, to share the unlock at the end of the function.

A simplified version of the semantic patch that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r exists@
expression E1;
identifier f;
@@

f (...) { <+...
* spin_lock_irq (E1,...);
... when != E1
* return ...;
...+> }
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
14 years agoARM: 6043/1: AT91 slow-clock resume: Don't wait for a disabled PLL to lock
Anders Larsen [Thu, 8 Apr 2010 10:48:16 +0000 (11:48 +0100)]
ARM: 6043/1: AT91 slow-clock resume: Don't wait for a disabled PLL to lock

at91 slow-clock resume: Don't wait for a disabled PLL to lock.

We run into this problem with the PLLB on the at91: ohci-at91 disables
the PLLB when going to suspend. The slowclock code however tries to do
the same: It saves the PLLB register value and when restoring the value
during resume, it waits for the PLLB to lock again. However the PLL will
never lock and the loop would run into its timeout because the slowclock
code just stored and restored an empty register.
This fixes the problem by only restoring PLLA/PLLB when they were enabled
at suspend time.

Cc: Andrew Victor <avictor.za@gmail.com>
Signed-off-by: Anders Larsen <al@alarsen.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
14 years agocfq-iosched: Fix the incorrect timeslice accounting with forced_dispatch
Divyesh Shah [Fri, 9 Apr 2010 07:29:57 +0000 (09:29 +0200)]
cfq-iosched: Fix the incorrect timeslice accounting with forced_dispatch

When CFQ dispatches requests forcefully due to a barrier or changing iosched,
it runs through all cfqq's dispatching requests and then expires each queue.
However, it does not activate a cfqq before flushing its IOs resulting in
using stale values for computing slice_used.
This patch fixes it by calling activate queue before flushing reuqests from
each queue.

This is useful mostly for barrier requests because when the iosched is changing
it really doesnt matter if we have incorrect accounting since we're going to
break down all structures anyway.

We also now expire the current timeslice before moving on with the dispatch
to accurately account slice used for that cfqq.

Signed-off-by: Divyesh Shah<dpshah@google.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: Add more debug-only per-cgroup stats
Divyesh Shah [Fri, 9 Apr 2010 04:15:35 +0000 (21:15 -0700)]
blkio: Add more debug-only per-cgroup stats

1) group_wait_time - This is the amount of time the cgroup had to wait to get a
  timeslice for one of its queues from when it became busy, i.e., went from 0
  to 1 request queued. This is different from the io_wait_time which is the
  cumulative total of the amount of time spent by each IO in that cgroup waiting
  in the scheduler queue. This stat is a great way to find out any jobs in the
  fleet that are being starved or waiting for longer than what is expected (due
  to an IO controller bug or any other issue).
2) empty_time - This is the amount of time a cgroup spends w/o any pending
   requests. This stat is useful when a job does not seem to be able to use its
   assigned disk share by helping check if that is happening due to an IO
   controller bug or because the job is not submitting enough IOs.
3) idle_time - This is the amount of time spent by the IO scheduler idling
   for a given cgroup in anticipation of a better request than the exising ones
   from other queues/cgroups.

All these stats are recorded using start and stop events. When reading these
stats, we do not add the delta between the current time and the last start time
if we're between the start and stop events. We avoid doing this to make sure
that these numbers are always monotonically increasing when read. Since we're
using sched_clock() which may use the tsc as its source, it may induce some
inconsistency (due to tsc resync across cpus) if we included the current delta.

Signed-off-by: Divyesh Shah<dpshah@google.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: Add io_queued and avg_queue_size stats
Divyesh Shah [Fri, 9 Apr 2010 04:15:10 +0000 (21:15 -0700)]
blkio: Add io_queued and avg_queue_size stats

These stats are useful for getting a feel for the queue depth of the cgroup,
i.e., how filled up its queues are at a given instant and over the existence of
the cgroup. This ability is useful when debugging problems in the wild as it
helps understand the application's IO pattern w/o having to read through the
userspace code (coz its tedious or just not available) or w/o the ability
to run blktrace (since you may not have root access and/or not want to disturb
performance).

Signed-off-by: Divyesh Shah<dpshah@google.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: Add io_merged stat
Divyesh Shah [Fri, 9 Apr 2010 04:14:23 +0000 (21:14 -0700)]
blkio: Add io_merged stat

This includes both the number of bios merged into requests belonging to this
cgroup as well as the number of requests merged together.
In the past, we've observed different merging behavior across upstream kernels,
some by design some actual bugs. This stat helps a lot in debugging such
problems when applications report decreased throughput with a new kernel
version.

This needed adding an extra elevator function to capture bios being merged as I
did not want to pollute elevator code with blkiocg knowledge and hence needed
the accounting invocation to come from CFQ.

Signed-off-by: Divyesh Shah<dpshah@google.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoblkio: Changes to IO controller additional stats patches
Divyesh Shah [Fri, 9 Apr 2010 06:31:19 +0000 (08:31 +0200)]
blkio: Changes to IO controller additional stats patches

that include some minor fixes and addresses all comments.

Changelog: (most based on Vivek Goyal's comments)
o renamed blkiocg_reset_write to blkiocg_reset_stats
o more clarification in the documentation on io_service_time and io_wait_time
o Initialize blkg->stats_lock
o rename io_add_stat to blkio_add_stat and declare it static
o use bool for direction and sync
o derive direction and sync info from existing rq methods
o use 12 for major:minor string length
o define io_service_time better to cover the NCQ case
o add a separate reset_stats interface
o make the indexed stats a 2d array to simplify macro and function pointer code
o blkio.time now exports in jiffies as before
o Added stats description in patch description and
  Documentation/cgroup/blkio-controller.txt
o Prefix all stats functions with blkio and make them static as applicable
o replace IO_TYPE_MAX with IO_TYPE_TOTAL
o Moved #define constant to top of blk-cgroup.c
o Pass dev_t around instead of char *
o Add note to documentation file about resetting stats
o use BLK_CGROUP_MODULE in addition to BLK_CGROUP config option in #ifdef
  statements
o Avoid struct request specific knowledge in blk-cgroup. blk-cgroup.h now has
  rq_direction() and rq_sync() functions which are used by CFQ and when using
  io-controller at a higher level, bio_* functions can be added.

Signed-off-by: Divyesh Shah<dpshah@google.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agoMerge remote branch 'nouveau/for-airlied' of ../drm-nouveau-next into drm-linus
Dave Airlie [Fri, 9 Apr 2010 04:27:51 +0000 (14:27 +1000)]
Merge remote branch 'nouveau/for-airlied' of ../drm-nouveau-next into drm-linus

* 'nouveau/for-airlied' of ../drm-nouveau-next: (21 commits)
  drm/nouveau: bail out of auxch transaction if we repeatedly recieve defers
  drm/nv50: implement gpio set/get routines
  drm/nv50: parse/use some more de-magiced parts of gpio table entries
  drm/nouveau: store raw gpio table entry in bios gpio structs
  drm/nv40: Init some tiling-related PGRAPH state.
  drm/nv50: Add NVA3 support in ctxprog/ctxvals generator.
  drm/nv50: another dodgy DP hack
  drm/nv50: punt hotplug irq handling out to workqueue
  drm/nv50: preserve an unknown SOR_MODECTRL value for DP encoders
  drm/nv50: Allow using the NVA3 new compute class.
  drm/nv50: cleanup properly if PDISPLAY init fails
  drm/nouveau: fixup the init failure paths some more
  drm/nv50: fix instmem init on IGPs if stolen mem crosses 4GiB mark
  drm/nv40: add LVDS table quirk for Dell Latitude D620
  drm/nv40: rework lvds table parsing
  drm/nouveau: detect vram amount once, and save the value
  drm/nouveau: remove some unused members from drm_nouveau_private
  drm/nouveau: Make use of TTM busy_placements.
  drm/nv50: add more 0x100c80 flushy magic
  drm/nv50: fix fbcon when framebuffer above 4GiB mark
  ...

14 years agodrm/nouveau: bail out of auxch transaction if we repeatedly recieve defers
Ben Skeggs [Mon, 15 Mar 2010 22:45:07 +0000 (08:45 +1000)]
drm/nouveau: bail out of auxch transaction if we repeatedly recieve defers

There's one known case where we never stop recieving DEFER, and loop here
forever.  Lets not do that..

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
14 years agodrm/nv50: implement gpio set/get routines
Ben Skeggs [Wed, 7 Apr 2010 02:57:35 +0000 (12:57 +1000)]
drm/nv50: implement gpio set/get routines

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
14 years agodrm/nv50: parse/use some more de-magiced parts of gpio table entries
Ben Skeggs [Wed, 7 Apr 2010 02:05:32 +0000 (12:05 +1000)]
drm/nv50: parse/use some more de-magiced parts of gpio table entries

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>