On Fri, 2012-01-20 at 09:44 -0800, Andy Grover wrote: > On 01/20/2012 08:12 AM, Christoph Hellwig wrote: > > On Thu, Jan 19, 2012 at 01:39:17PM -0800, Andy Grover wrote: > >> This saves all fabrics from calling core_tmr_alloc_req() and > >> having to check the result. The downside is se_cmd gets bigger for all > >> requests, but hopefully later patches will reduce it. > > > > Without patches to void the overhead it's not acceptable. Fortunately > > it should be doable fairly simply by using an union for command vs > > TMR fields. > > This was my thought too. We should be able to move cmd variables into a > union w/tmr struct very soon, with a low risk of introducing bugs. > Hi Andy, Ping on this..? I've been merging patches from lio-core into target-pending/for-next the past week, and i've stopped ahead of this patch to inline se_tmr_req into se_cmd due of the extra overhead in question here.. I'd really like to get this resolved in v3.4 for-next, so would you mind taking care of the lio-core conversion to union for command vs TMR fields so this can be squashed into for-next..? 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
