linux.git/drivers/block, branch v3.9.8

rbd: use the correct length for format 2 object names

2013-06-27T17:39:17+00:00

commit 3a96d5cd7bdce45d5dded75c3a62d4fb98050280 upstream.

[Backported for 3.9-stable.
'kmalloc(MAX_OBJ_NAME_SIZE + 1, GFP_NOIO)' was changed as
'kmem_cache_alloc(rbd_segment_name_cache, GFP_NOIO)' in 78c2a44
since 3.10-rc1, and 78c2a44 is relied on a big patchset, so restore
it as 3.9 did.]

Format 2 objects use 16 characters for the object name suffix to be
able to express the full 64-bit range of object numbers. Format 1
images only use 12 characters for this. Using 12-character names for
format 2 caused userspace and kernel rbd clients to read differently
named objects, which made an image written by one client look empty to
the other client.

Reported-by: Chris Dunlop 
Signed-off-by: Josh Durgin 
Reviewed-by: Sage Weil 
Reviewed-by: Lingzhu Xiang 
Signed-off-by: Zhouping Liu 
Signed-off-by: Greg Kroah-Hartman

cciss: fix broken mutex usage in ioctl

2013-06-20T19:01:29+00:00

commit 03f47e888daf56c8e9046c674719a0bcc644eed5 upstream.

If a new logical drive is added and the CCISS_REGNEWD ioctl is invoked
(as is normal with the Array Configuration Utility) the process will
hang as below.  It attempts to acquire the same mutex twice, once in
do_ioctl() and once in cciss_unlocked_open().  The BKL was recursive,
the mutex isn't.

  Linux version 3.10.0-rc2 (scameron@localhost.localdomain) (gcc version 4.4.7 20120313 (Red Hat 4.4.7-3) (GCC) ) #1 SMP Fri May 24 14:32:12 CDT 2013
  [...]
  acu             D 0000000000000001     0  3246   3191 0x00000080
  Call Trace:
    schedule+0x29/0x70
    schedule_preempt_disabled+0xe/0x10
    __mutex_lock_slowpath+0x17b/0x220
    mutex_lock+0x2b/0x50
    cciss_unlocked_open+0x2f/0x110 [cciss]
    __blkdev_get+0xd3/0x470
    blkdev_get+0x5c/0x1e0
    register_disk+0x182/0x1a0
    add_disk+0x17c/0x310
    cciss_add_disk+0x13a/0x170 [cciss]
    cciss_update_drive_info+0x39b/0x480 [cciss]
    rebuild_lun_table+0x258/0x370 [cciss]
    cciss_ioctl+0x34f/0x470 [cciss]
    do_ioctl+0x49/0x70 [cciss]
    __blkdev_driver_ioctl+0x28/0x30
    blkdev_ioctl+0x200/0x7b0
    block_ioctl+0x3c/0x40
    do_vfs_ioctl+0x89/0x350
    SyS_ioctl+0xa1/0xb0
    system_call_fastpath+0x16/0x1b

This mutex usage was added into the ioctl path when the big kernel lock
was removed.  As it turns out, these paths are all thread safe anyway
(or can easily be made so) and we don't want ioctl() to be single
threaded in any case.

Signed-off-by: Stephen M. Cameron 
Cc: Jens Axboe 
Cc: Mike Miller 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

rbd: don't destroy ceph_opts in rbd_add()

2013-06-20T19:01:28+00:00

commit 7262cfca430a1a0e0707149af29ae86bc0ded230 upstream.

Whether rbd_client_create() successfully creates a new client or
not, it takes responsibility for getting the ceph_opts structure
it's passed destroyed.  If successful, the structure becomes
associated with the created client; if not, rbd_client_create()
will destroy it.

Previously, rbd_get_client() would call ceph_destroy_options()
if rbd_get_client() failed, and that meant it got called twice.
That led freeing various pointers more than once, which is never a
good idea.

This resolves:
    http://tracker.ceph.com/issues/4559

Reported-by: Dan van der Ster 
Signed-off-by: Alex Elder 
Reviewed-by: Josh Durgin 
Signed-off-by: Greg Kroah-Hartman

drivers/block/brd.c: fix brd_lookup_page() race

2013-06-07T19:52:41+00:00

commit dfd20b2b174d3a9b258ea3b7a35ead33576587b1 upstream.

The index on the page must be set before it is inserted in the radix
tree.  Otherwise there is a small race which can occur during lookup
where the page can be found with the incorrect index.  This will trigger
the BUG_ON() in brd_lookup_page().

Signed-off-by: Brian Behlendorf 
Reported-by: Chris Wedgwood 
Cc: Jens Axboe 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

drbd: fix for deadlock when using automatic split-brain-recovery

2013-05-19T18:38:41+00:00

commit 7c689e63a847316c1b2500f86891b0a574ce7e69 upstream.

With an automatic after split-brain recovery policy of
"after-sb-1pri call-pri-lost-after-sb",
when trying to drbd_set_role() to R_SECONDARY,
we run into a deadlock.

This was first recognized and supposedly fixed by
2009-06-10 "Fixed a deadlock when using automatic split brain recovery when both nodes are"
replacing drbd_set_role() with drbd_change_state() in that code-path,
but the first hunk of that patch forgets to remove the drbd_set_role().

We apparently only ever tested the "two primaries" case.

Signed-off-by: Philipp Reisner 
Signed-off-by: Lars Ellenberg 
Signed-off-by: Jens Axboe 
Signed-off-by: Jonghwan Choi 
Signed-off-by: Greg Kroah-Hartman

drbd: fix memory leak

2013-05-19T18:38:41+00:00

commit 94ad0a101415978be04945b2787be1e8e8a874db upstream.

We forgot to free the disk_conf,
so for each attach/detach cycle we leaked 336 bytes.

Signed-off-by: Philipp Reisner 
Signed-off-by: Lars Ellenberg 
Signed-off-by: Jens Axboe 
Signed-off-by: Jonghwan Choi 
Signed-off-by: Greg Kroah-Hartman

drbd: Fix build error when CONFIG_CRYPTO_HMAC is not set

2013-05-19T18:38:41+00:00

commit ef57f9e6bb9278720c8a5278728f252ab85d7ac6 upstream.

Signed-off-by: Philipp Reisner 
Signed-off-by: Lars Ellenberg 
Signed-off-by: Jens Axboe 
Signed-off-by: Jonghwan Choi 
Signed-off-by: Greg Kroah-Hartman

Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client

2013-04-17T19:52:02+00:00

Pull Ceph fix from Sage Weil:
 "It's a simple fix for a hard to hit race, but low-risk and clearly
  correct"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  rbd: do a safe list traversal in rbd_img_request_submit()

rbd: do a safe list traversal in rbd_img_request_submit()

2013-04-17T18:39:09+00:00

It's possible that the reference to the object request dropped
inside the loop in rbd_img_request_submit() will be the last
one, in which case the content of the object pointer can't be
trusted.

Use a safe form of the object request list traversal to avoid
problems.

This resolves:
    http://tracker.ceph.com/issues/4705

Signed-off-by: Alex Elder 
Reviewed-by: Josh Durgin

Merge tag 'for-linus-20130409' of git://git.kernel.dk/linux-block

2013-04-09T19:05:41+00:00

Pull block fixes from Jens Axboe:
 "I've got a few smaller fixes queued up for 3.9 that should go in.  The
  major one is the loop regression, the others are nice fixes on their
  own though.  It contains:

   - Fix for unitialized var in the block sysfs code, courtesy of Arnd
     and gcc-4.8.

   - Two fixes for mtip32xx, fixing probe and command timeout.  Also a
     debug measure that could have waited for 3.10, but it's driver
     only, so I let it slip in.

   - Revert the loop partition cleanup fix, it could cause a deadlock on
     auto-teardown as part of umount.  The fix is clear, but at this
     point we just want to revert it and get a real fix in for 3.10."

* tag 'for-linus-20130409' of git://git.kernel.dk/linux-block:
  Revert "loop: cleanup partitions when detaching loop device"
  mtip32xx: fix two smatch warnings
  mtip32xx: Add debugfs entry device_status
  mtip32xx: return 0 from pci probe in case of rebuild
  mtip32xx: recovery from command timeout
  block: avoid using uninitialized value in from queue_var_store