[Qemu-devel] [PATCH v2 00/11] blockjobs: Fix transactional race condition

[John Snow]

There are a few problems with transactional job completion right now.

First, if jobs complete so quickly they complete before remaining jobs
get a chance to join the transaction, the completion mode can leave well
known state and the QLIST can get corrupted and the transactional jobs
can complete in batches or phases instead of all together.

Second, if two or more jobs defer to the main loop at roughly the same
time, it's possible for one job's cleanup to directly invoke the other
job's cleanup from within the same thread, leading to a situation that
will deadlock the entire transaction.

Thanks to Vladimir for pointing out these modes of failure.

This series also does a little digging into refactoring Jobs into public
and private interfaces. It's somewhat unrelated, but it was easier to
include this with this series than separate it out and send it later.
This comprises patches 2-6. The actual fixes here are in patches 1 and
7-10. A new test to catch Vladimir's failure scenario is in patch 11.

v2:
 - Lots of differences in patches 2-9.
 - Cancel should now work on an "unstarted" blockjob.
 - New refactoring patches.
 - Added "start" property for BlockJob Drivers.

________________________________________________________________________________

For convenience, this branch is available at:
https://github.com/jnsnow/qemu.git branch job-manual-start
https://github.com/jnsnow/qemu/tree/job-manual-start

This version is tagged job-manual-start-v2:
https://github.com/jnsnow/qemu/releases/tag/job-manual-start-v2

John Snow (10):
  blockjob: centralize QMP event emissions
  Blockjobs: Internalize user_pause logic
  blockjobs: Always use block_job_get_aio_context
  blockjobs: split interface into public/private
  blockjobs: fix documentation
  blockjob: add .clean property
  blockjob: add .start field
  blockjob: add block_job_start
  blockjob: refactor backup_start as backup_job_create
  iotests: add transactional failure race test

Vladimir Sementsov-Ogievskiy (1):
  blockjob: fix dead pointer in txn list

 block/backup.c               |  59 ++++---
 block/commit.c               |   6 +-
 block/io.c                   |   6 +-
 block/mirror.c               |   7 +-
 block/replication.c          |  13 +-
 block/stream.c               |   6 +-
 blockdev.c                   | 128 +++++++--------
 blockjob.c                   |  72 +++++++--
 include/block/block.h        |   3 +-
 include/block/block_int.h    |  29 ++--
 include/block/blockjob.h     | 345 +++-------------------------------------
 include/block/blockjob_int.h | 366 +++++++++++++++++++++++++++++++++++++++++++
 qemu-img.c                   |   4 +-
 tests/qemu-iotests/124       |  91 +++++++++++
 tests/qemu-iotests/124.out   |   4 +-
 tests/test-blockjob-txn.c    |  14 +-
 tests/test-blockjob.c        |   2 +-
 17 files changed, 688 insertions(+), 467 deletions(-)
 create mode 100644 include/block/blockjob_int.h

-- 
2.7.4