[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <yq1popgrsw0.fsf@sermon.lab.mkp.net>
Date: Wed, 10 Aug 2016 21:33:51 -0400
From: "Martin K. Petersen" <martin.petersen@...cle.com>
To: Shaun Tancheff <shaun.tancheff@...gate.com>
Cc: Tom Yan <tom.ty89@...il.com>, Shaun Tancheff <shaun@...cheff.com>,
linux-ide@...r.kernel.org, LKML <linux-kernel@...r.kernel.org>,
Christoph Hellwig <hch@....de>, Tejun Heo <tj@...nel.org>,
Josh Bingaman <josh.bingaman@...gate.com>
Subject: Re: [PATCH v5 2/2] Add support for SCT Write Same
>>>>> "Shaun" == Shaun Tancheff <shaun.tancheff@...gate.com> writes:
Shaun,
Shaun> You are correct in that we can advertise the larger limit in
Shaun> ata_scsi_dev_config() when only SCT write same is supported
Shaun> rather than fall back to WS10.
I deliberately capped WRITE SAME to 64K blocks unless otherwise reported
by the device because:
a) Several older drives supported the WRITE SAME(16) command but
ignored the upper bytes of the transfer length effectively
turning it into a WRITE SAME(10).
b) 64K blocks was the sweet spot for older drives as well, a
size commonly used by RAID array firmwares that were the only
commonplace users of the WRITE SAME family.
Shaun> I really am not sure what would be considered the correct
Shaun> solution though. I believe that the WRITE SAME defaults are
Shaun> currently being chosen around physical limits.
Lacking a solid reporting facility in the spec, I suggest you just leave
it unset and let the SCSI defaults apply.
Shaun> I think we can also bump the command timeout for WRITE SAME?
The default WRITE SAME timeout is 120s.
--
Martin K. Petersen Oracle Linux Engineering
Powered by blists - more mailing lists