On Wed, 2012-01-04 at 18:17 -0800, Roland Dreier wrote: > On Wed, Jan 4, 2012 at 4:54 PM, Roland Dreier <roland@xxxxxxxxxx> wrote: > > After discussing this offline with a few people, I now realize we can't > > just silently succeed, at least the way my previous patch for data > > commands did. (ie that patch is wrong, although better than crashing) > > > > The problem is that returning immediately like that bypasses the > > sequencer, which means we incorrectly skip reservation checking, > > returning unit attention, etc. > > Never mind... I see that generic_new_cmd comes after the sequencer > so this doesn't quite apply. > > However I'm still a bit uncomfortable with adding special handling for > REQUEST SENSE, scattering length checks all over the inquiry handling, > having a separate buffer and copying in MODE SENSE and all the other > ad hoc stuff there is... it would be much better if we could just handle the > 0-length stuff uniformly for all the control CDBs. > So while I don't have any objections with the points above, re-factoring out the data phase from existing control cdb emulation code is going to be easier said than done. ;) So that said, I'm still leaning toward first fixing the immediate bug for 0-length handling with a special case for REQUEST_SENSE in transport_generic_cmd_sequencer() for v3.3, and look at more heavy lifting to unify 0-length handling in per CDB functions for the next round.. I'm still catching up on items from the last weeks, but I'll have something to address the immediate issue for tomorrow. --nab -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
