[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Mylex Extremeraid 3000 + DAC960 woes.



Hi,

> Sorry, I shoulda remembered that... :-\
> 
> 2GB or less should cause no problems.

Well, who could possibly remember all alpha models in the
various configurations ? ( sure not me, although i have quite
a collection ;-)

> I've built it under RH 7.1 and 7.2, with GCC 2.96.

I wonder, if it would be possible to compile a 2.4.19 kernel under
redhat 7.1/7.2 and implant it on my UP2000.....
Even more, i wonder if I could motivate *you* doing it for me,
based in my .config ;-)
If that would work, i even would download some sort of redhat
ISO, install it on a testbed just for compiling kernels....
 
> I realize that Debian is still using 2.95.x, and also that there have
> been numerous problems related to its inability to correctly build
> some parts of the kernel.

Not only the kernel, but some assorted sources, i've done in
customer projects.

> Change compilers... ;-}
> 
> Seriously, perhaps 3.2 is in decent enough shape now?

Yup. That's a high-priority item on my todo list. I just wanted to
do it correctly:
1. get the new FC-AL subsystem running, using the Mylex
controller
2. Install Debian 3.0 on it
3. Move important parts to versions recompiled with 3.2

One has to keep in mind though, that warnings regarding
compiling kernels with 3.x GCC's clutter LKML quite
regularily.
As a side note, it *does* compile with 2.95.x, if you avoid
inlining the functions in the DAC driver. I fail to see, why this
should break the driver itself ( there are some potential
pitfalls there ).
Looking at the DAC-code itself, the function that fails is being 
called quite down the line during driver init(), even after
some functions have been called, that are supposed to be
inlined.

I wonder if anybody *ever* got that mylex extremeraid 3000
beast to work on alphas. As far as i can tell, about all
models of DAC-controllers share the same infrastructure to
some extend. That's not "compatibility" in it's strict sense,
but it should be a qualified guess, that the model 3000, which
is the Fibre-channel version, does run if the model 2000 ( dual
channel SCSI ) runs.

 
> Good luck.

well, uhm, thanks ;-)
( I still keep wondering, why it always appears, as if it's me who 
has these kinds of bizarre problems first ;-)

Regards,
T. Weyergraf

-- 
Thomas Weyergraf                                                kirk@colinet.de
My Favorite IA64 Opcode-guess ( see arch/ia64/lib/memset.S )
"br.ret.spnt.few" - got back from getting beer, did not spend a lot.




Reply to: