Here's what Douglas Gilbert had to say about the SCSI generic driver patch:
----- Forwarded message from Douglas Gilbert <dgilbert@xxxxxxxxxxxx> -----
Alan Eldridge wrote:
>
> I ran into the process hang on allocation in the sg driver with 2.4.5. As
> I'm using the SGI XFS-port tree, I have just patched in those two patches
> (one just comments in the .h) in order to stabilize sg.
>
> There have been several persons asking about the applicability of that patch
> backwards from 2.4.5, and its status re integration into Linus's tree.
>
> Could you comment on what versions of the kernel need this patch, and the
> patch's integration status, so that I might post it back to the list? Or, if
> you wish, post directly to linux-xfs@xxxxxxxxxxxx
Alan,
The patch is in lk 2.4.5-ac18 and onwards. It is sg version
3.1.19 and is available as a source tarball at
http://www.torque.net/sg
Since Linus doesn't track the SCSI subsystem (and Alan
Cox does) then I don't bother submitting bug fixes to Linus.
Eric Youngdale and I still have bug fixes pending from
January stranded in the "ac" tree. So I don't know
whether it will make it into lk 2.4.6 Hopefully Linus
will start the lk 2.5 tree and leave "ac" to do
lk 2.4.7 .
Any sg version 3.1.* can go in the lk 2.4 series. There
has been _no_ introduced incompatibility that I am aware
of that effects the sg driver across the lk 2.4 series.
Therefore sg version 3.1.19 can be backported to lk 2.4.5
Feel free to pass this information back to linux-xfs@xxxxxxxxxxx
[I have enough lists to monitor already.]
> Thanks much for your assistance.
>
> P.S. The way the hang happened to me was in cdda2wav, where it was trying to
> increase a data buffer size. It only locked up the process under X; early on
[...my own text elided...]
Yep, that sounds like the one:-)
Doug Gilbert
----- End forwarded message -----
--
Alan Eldridge
"Gee, Bill, what do you want to do tonight?"
"The same thing we do every night, Steve. Try to take over the world!"
|