sata_sil on amd64

sata_sil on amd64

[Tyler Montbriand]

I have been stuck using kernel 2.6.4 for the last year or so for various 
reasons, and now I'm in a vicious catch-22.  If I don't upgrade, I can't use 
NPTL, therefore my system will continue to be susceptible to the 
sigqueue-overflow flaw in linuxthreads.  But if I do upgrade, problems in 
>0.53 sata_sil implementations halt my system in a matter of minutes with 
dmesg entries about imaginary bad blocks.

0.53 probably has it's own flaws, but it's been rock-solid running a single 
drive on my computer for a year.  How feasible would it be to forward-port 
this version to a newer kernel, say 2.6.10?  If I fix the header problems 
could it work, or is .10 too structurally different from .4?

Re: sata_sil on amd64

[Bartlomiej Zolnierkiewicz]

On Sat, 22 Jan 2005 19:19:35 -0600, Tyler Montbriand <tsm@accesscomm.ca> wrote:
> I have been stuck using kernel 2.6.4 for the last year or so for various
> reasons, and now I'm in a vicious catch-22.  If I don't upgrade, I can't use
> NPTL, therefore my system will continue to be susceptible to the
> sigqueue-overflow flaw in linuxthreads.  But if I do upgrade, problems in
> >0.53 sata_sil implementations halt my system in a matter of minutes with
> dmesg entries about imaginary bad blocks.
> 
> 0.53 probably has it's own flaws, but it's been rock-solid running a single
> drive on my computer for a year.  How feasible would it be to forward-port
> this version to a newer kernel, say 2.6.10?  If I fix the header problems
> could it work, or is .10 too structurally different from .4?

Alternatively if you pinpoint the kernel snapshot which introduces this behavior
(-bk patches are available at kernel.org) we may be able to fix sata_sil.

Bartlomiej