[1874] in linux-scsi channel archive
Re: Latest aicxxx driver (fwd)
daemon@ATHENA.MIT.EDU (Daniel M. Eischen)
Wed May 14 12:48:48 1997
Date: Wed, 14 May 1997 11:43:49 -0500 (CDT)
From: "Daniel M. Eischen" <deischen@iworks.InterWorks.org>
To: yuri@buster.rgti.com
Cc: aic7xxx@FreeBSD.ORG, linux-scsi@vger.rutgers.edu,
owner-aic7xxx@FreeBSD.ORG
> The other drive is a Quantum 540S. Didn't try it without SMP yet.
OK, it might not support too many tagged commands. You might be
able to find out more info at Quantums site.
> It does work with 8 - however I dont think this excercises the paging
> mechanism. (Also I reeaaaly want to use as many as possible ;)))
Yup :-)
> Does the driver downgrade the number of SCBs for a specific device
> if the current number is too high? Some of the other drivers do this
> successfully - it would be a nice idiot-proof feature to implement.
There is code in there to do that, but it's #ifdef'd out. I don't
believe the mid-level SCSI code will support this though (yet) -
which is why it's commented out of the driver. I'll have to take
another look to see if the mid-level SCSI code has changed to
support this. Last I looked it was allocating a Scsi_Cmnd for
the number of tagged commands reported by the driver, and wasn't
using the queue depth count after that point. So it would send
as many commands as there were Scsi_Cmnds for a device regardless
of how queue depth was changed after this point.
Try without SMP. I heard there were fixes for SMP in the pre-7
release ???
Dan Eischen
deischen@iworks.InterWorks.org
