[dm-devel] [PATCH] 0/9: Multipath patches

[Kevin Corry]

Here are some quick-fixes for the new dm-multipath module. Based on 2.6.0-udm3 
+ Christophe's bio_list patches from Jan 10.

Revision 1:
  When a bio is submitted to the block-layer, several of the fields in that
  bio will have changed by the time it completes. In particular, the bi_sector
  field will be different, the bi_size field will be zero, and the bi_bdev and
  bi_idx fields may have changed as well. This creates a problem for some
  of the Device-Mapper modules (like dm-multipath) that may want to redrive a
  bio.
  
  Save the necessary bio fields in the target_io structure just before calling
  generic_make_request(), and restore those bio fields just before calling the
  target's end_io() routine. This ensures that the target module sees the bio
  exactly as it was following the map() routine.

Revision 2:
  path->fail_count needs to be initialized to non-zero, or a failed path will
  never actually be failed.
  
  Set to 1 for now, but this number can be increased if we don't want to fail
  the path on the very first I/O error.

Revision 3:
  Make a #define for the minimum-io value.

Revision 4:
  We can't depend on the test_bio being the same every time it is submitted.
  Need to reset some of the fields in the bio each time.

Revision 5:
  Add a pointer from the priority-group back to the multipath context.

Revision 6:
  Change the failed_ios list in the multipath structure from a stack to a
  queue using the new bio_list services.

Revision 7:
  In test_path, instead of driving the test I/O, queue it on a list. When all
  test I/Os have been queued, they can then be driven down without having to
  hold the path_lock.

Revision 8:
  m->path_lock needs to be locked with spin_lock_irqsave() and unlocked with
  spin_lock_irqrestore().

Revision 9:
  If a test bio completes successfully and the path is currently invalid,
  reactivate the path.

-- 
Kevin Corry
kevcorry at us.ibm.com
http://evms.sourceforge.net/