On Thu, 2010-10-28 at 14:28 -0500, James Bottomley wrote: > On Thu, 2010-10-28 at 17:05 +0200, Andi Kleen wrote: > > Here's the promised attempt to do a full host_lock pushdown. This > > was done using coccinelle and some manual adjustments. > > > > This is just a dumb pushdown of host lock and serial number into > > the driver, no attempt to remove any locks yet. > > > > Currently lightly tested, but compiles with allmodconfig on 32bit x86. > > > > I made no attempt to automatically add maintainers to Cc. > > > > I'm not posting the patches because they are numerous and boring, > > but it's available in git. > > > > This will likely bitrot quickly so if it's decided to do it this > > way I would prefer if this was merged quickly. > > OK, so can you post an actual diff? It obviously has to be done in one > commit rather than the 87 you have. > So it appears this series will still need the atomic_add_return() usage in scsi_cmd_get_serial() and drop host_lock usage from those LLDs that we know can be converted to 'host_lock-less', yes..? Following Chrisof's and David's comments for zfcp and srp host_lock push down this morning, the latest v9 code has been pushed into: git://git.kernel.org/pub/scm/linux/kernel/git/nab/lio-4.0.git host_lock-less-for-37-v9 +1 for ak for actually coding this up in coccinelle. ;) > > Issues found: > > - I found a clear deadlock in aha1740, not sure if that driver ever > > worked since host_lock was introduced. Anyways I attempted to work > > around it. But if it's really that unused it may be better to > > simply drop it. > > I certainly don't have the hardware ... and I don't think we've had any > bug reports for a while > > > - I'm not sure how to compile cciss_scsi, is it even used? > > It's used for tape support in the cciss driver. It's actually #included > into the cciss.c file (don't ask, I'll only start to cry) and is > conditioned on the CISS_SCSI_TAPE Kconfig variable. > I just did a 'make drivers/block' to compile this code.. Thanks! --nab -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
