public inbox for linux-scsi@vger.kernel.org
 help / color / mirror / Atom feed
* queue depth help!!
@ 2003-12-27 21:38 jshankar
  2003-12-27 21:46 ` James Bottomley
  0 siblings, 1 reply; 4+ messages in thread
From: jshankar @ 2003-12-27 21:38 UTC (permalink / raw)
  To: linux-scsi

Hello,

if the scsi target queue depth is 1, does it mean that i can handle only one
outstanding command  at a time.

Does it mean if another command is issued( queue depth=1), it has to wait till
the previous command is done.


Thanks
Jay


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: queue depth help!!
  2003-12-27 21:38 jshankar
@ 2003-12-27 21:46 ` James Bottomley
  0 siblings, 0 replies; 4+ messages in thread
From: James Bottomley @ 2003-12-27 21:46 UTC (permalink / raw)
  To: jshankar; +Cc: linux-scsi

On Sat, 2003-12-27 at 15:38, jshankar wrote:
> if the scsi target queue depth is 1, does it mean that i can handle only one
> outstanding command  at a time.

Yes

> Does it mean if another command is issued( queue depth=1), it has to wait till
> the previous command is done.

Yes.

The mid-layer will start the queued command as soon as the one occupying
the device returns...this is often why untagged devices actually set
their depth to 2 so they can have another command all prepared and ready
to roll as soon as the in-progress one returns.

James



^ permalink raw reply	[flat|nested] 4+ messages in thread

* RE: queue depth help!!
@ 2003-12-31 16:23 jshankar
  2003-12-31 16:26 ` James Bottomley
  0 siblings, 1 reply; 4+ messages in thread
From: jshankar @ 2003-12-31 16:23 UTC (permalink / raw)
  To: James Bottomley; +Cc: linux-scsi

Hello James,

Case :queue depth=1

On issuing a scsi command write, does the  data needs to be written 
succesfully 
to disk before issuing a scsi response. Is there a chance that the previous 
command's data is still cached on scsi controller and the next command issued 
is executed.

Please let me know if there are any  good tools for scsi analyzer.

Thanks
Jay



>===== Original Message From James Bottomley <James.Bottomley@SteelEye.com> 
=====
>On Sat, 2003-12-27 at 15:38, jshankar wrote:
>> if the scsi target queue depth is 1, does it mean that i can handle only 
one
>> outstanding command  at a time.
>
>Yes
>
>> Does it mean if another command is issued( queue depth=1), it has to wait 
till
>> the previous command is done.
>
>Yes.
>
>The mid-layer will start the queued command as soon as the one occupying
>the device returns...this is often why untagged devices actually set
>their depth to 2 so they can have another command all prepared and ready
>to roll as soon as the in-progress one returns.
>
>James


^ permalink raw reply	[flat|nested] 4+ messages in thread

* RE: queue depth help!!
  2003-12-31 16:23 queue depth help!! jshankar
@ 2003-12-31 16:26 ` James Bottomley
  0 siblings, 0 replies; 4+ messages in thread
From: James Bottomley @ 2003-12-31 16:26 UTC (permalink / raw)
  To: jshankar; +Cc: linux-scsi

On Wed, 2003-12-31 at 10:23, jshankar wrote:
> On issuing a scsi command write, does the  data needs to be written 
> succesfully 
> to disk before issuing a scsi response. Is there a chance that the previous 
> command's data is still cached on scsi controller and the next command issued 
> is executed.

That depends on the drive cache setting.  For a write back cache,
no--status will be sent before the command reaches the medium (as long
as the FUA bit isn't set); for all other cache settings, yes.

> Please let me know if there are any  good tools for scsi analyzer.

I don't actually have one of these, so I'll let others answer...

James



^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2003-12-31 16:27 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-12-31 16:23 queue depth help!! jshankar
2003-12-31 16:26 ` James Bottomley
  -- strict thread matches above, loose matches on Subject: below --
2003-12-27 21:38 jshankar
2003-12-27 21:46 ` James Bottomley

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox