I think I know what the basic issues are here, but I'm trying to get
some bearing on what to do next, and maybe nudge the developers in a
favourable direction.
I have a dual Opteron with the 1210SA controller plus a 36GB WDC SATA
disk. It won't install either of the test or development setups. It
will boot, recognize the controller as an IDE device (I believe that
is the siimage driver, right?), find the disk, but then I get tons of
"hdg: lost interrupt" messages. The system appears to talk to the
disk, but so slowly as to be useless.
Skimming recent discussions on the Linux Kernel list, it transpires
that
a) the controller has a SIL3112 chip
b) the firmware has a nasty habit of turning the interrupts off on
this particular device
c) there's another kernel driver (sata_sil + libata) which is the one
you really ought to use
d) patches exist and have been tested with some success
I saw a note (where?) to the effect that the libata stuff was still in
the works and an errata RPM was being planned. In any case, I'm not
badly scared of patching kernels myself. However, without a kernel
that can handle the disk, I'd still be up the creek without a paddle.
(No other Opterons around and I'm not going into cross-compiling from
386 systems.)
On the other hand if someone would just set up a pxeboot directory
with stuff which could get me through the install stages, I'd be eager
to test it and report back. It might be useful to do this before
release since I suspect I'm not the only one around with such a
setup (nudge, nudge...)
The other obvious workaround is of course just to go out and buy a
disk for the *real* IDE controller and install to that. However, if a
solution is just around the corner, I'd rather save my money for
something else.
--
O__ ---- Peter Dalgaard Blegdamsvej 3
c/ /'_ --- Dept. of Biostatistics 2200 Cph. N
(*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard(a)biostat.ku.dk) FAX: (+45) 35327907
Show replies by date