On Wed, Jun 10, 2009 at 10:00:29AM -0400, Mike Snitzer wrote: > My v2 of the DM topology support patch does away with those and just > uses blk_stack_limits(). In the process we go back to _not_ using the > blk_queue_*() setters (note the additional checks that those setters > have). That was a secondary point, that could be cleared up by reviewing the checks to ensure they all automatically hold so never need checking at that point. The main problem was that applying the combined limits for the table previously - and correctly - paid no regard to the existing queue limits on the dm device. The new code seems to combine the limits between the live and inactive tables, which is nonsense as in dm terms they are totally independent. Alasdair -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
