[439] in linux-scsi channel archive
RE: Project im working on...
daemon@ATHENA.MIT.EDU (Michael Morrison)
Sat Jul 29 09:47:27 1995
From: mike@ringo.reno.nv.us (Michael Morrison)
To: linux-scsi@vger.rutgers.edu (linux scsi)
Date: Fri, 28 Jul 1995 21:29:08 -0700 (PDT)
[snip]
>> First, has anyone on the scsi list
>> used linux in an application similar to this who would be willing to
>> share their experience, pitfalls, etc?
We are using Linux and an Adaptex 3940W to help develop/debug a
SCSI frame buffer. The frame buffer is used to move video from
a workstation to a digital video disk recorder as quicky as possible.
The frame buffer has 4 fast/wide ports each of which can support multiple
busses. The 3940W has 2 fast/wide scsi busses.
[snip]
>> Also, is it still the case that the
>> documentation for the sg interface is "read the source, luke"?
Get the SCSI-Programing howto from your favorite FTP site. It explains
the interface.
We exclusively use the generic scsi driver to move huge amounts of
data from the linux box to the disk recorder. There are a couple
of bugs in the sg driver, one that probably wont affect you and one
that may affect you. (Linux 1.2.10)
1. You can't send data greater than 63 blocks (32256 bytes)
even though SG_BIG_BUFF is set to 32768 in a single command.
This is due to a round up and subsequent compare error.
2. Error info sent up from the lower level drivers is wiped out
in sg_read. So if you need lots of error information
you have to comment out the line in sg_read() that says:
device->header.result=0;
3. Interface is clumsy
I will be issueing a bug report after I determine if these bugs are
still in the latest driver.
>> Finally, where would I look to try to disable the SCSI bus probe on
>> this controller? I have a feeling that Lynx may not work well with it
>> and I may have to disable it. I suppose I could boot linux while the
>> VME is sitting in it's ROM monitor to avoid any problems here if
>> necessary.
I was not sucessful in putting 2 initiators on the same bus with
the aic7xxxx driver. Even though I was able to set the two initiators
on the 3940w to different IDs, the linux driver forced them to be ID 7.
I haven't investigated this very much so it coult (probalbly?) was my
fault.
>> Thanks in advance for any assistance/advice.
mike
mike@ringo.reno.nv.us
