LU-18989 osd: no tx restart in fallocate

LU-18989 osd: no tx restart in fallocate

A transaction restart in osd_fallocate_preallocate()
while holding an object lock may lead to a deadlock
with another thread which takes the lock first and then
starts a transaction.

crash> bt 44982
 #0 __schedule at ffffffffa2331fe8
 #1 schedule at ffffffffa233241a
 #2 wait_transaction_locked at ffffffffc106f08a
 #3  add_transaction_credits at ffffffffc106f67a
 #4  start_this_handle at ffffffffc106fa10
 #5  jbd2__journal_restart at ffffffffc107015e
 #6  osd_fallocate_preallocate.constprop.0 at ffffffffc23e3b1b
 #7  osd_fallocate at ffffffffc23e40cb
 #8  mdt_object_fallocate at ffffffffc21418f9
 #9  mdt_fallocate_hdl at ffffffffc2144125

and

crash> bt
PID: 47838    TASK: ffff9f66f5288000  CPU: 9    COMMAND: "mdt00_020"
 #0  __schedule at ffffffffa2331fe8
 #1  schedule at ffffffffa233241a
 #2  rwsem_down_write_slowpath at ffffffffa2334a9b
 #3  osd_write_lock at ffffffffc23b6b4d
 #4  mdd_xattr_del at ffffffffc208cd88
 #5  mdt_reint_setxattr at ffffffffc21224a9
 #6  mdt_reint_rec at ffffffffc211ec69
 #7  mdt_reint_internal at ffffffffc20f0a64
 #8  mdt_reint at ffffffffc20fc6b9
 #9  tgt_handle_request0 at ffffffffc1dea9e7

HPE-bug-id: LUS-12819
Signed-off-by: Alexander Zarochentsev <alexander.zarochentsev@hpe.com>
Change-Id: I8baa3ba1505c7a55e96841524368a66d447d18c5
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/59158
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andrew Perepechko <andrew.perepechko@hpe.com>
Reviewed-by: Arshad Hussain <arshad.hussain@aeoncomputing.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>