Ronald G Minnich <[EMAIL PROTECTED]> writes:

> On Sat, 15 Dec 2001, Dave Jones wrote:
> 
> > Can you repeat this with .17rc1 ?
> 
> Dave, first, I'm sorry for the random complaint. It was uncalled for.
> Obviously we wouldn't even be here if not for folks like you. Thank you.
> 
> Yes, we'll try .17rc1 as we get time. The big effort this week is to try
> to get 2.4.13 to really work on the l440gx, we're having eepro problems
> again, the machine comes up and can send but not recieve RARP packets :-(
> 
> FYI, another big problem we've had as 2.4 progresses is the gradual "bit
> rot" of the framebuffer stuff. The Cirrus Logic devfb stuff just sort of
> stopped working with 2.4.7 or so and has not come back. Same with the
> matrox stuff -- I am not sure what's happening here. I don't know if the
> other devfb is the same, but I'm wondering who I should relay this kind of
> experience to. We've missed having working devfb. It was wonderful while
> it lasted!

Hmm.  The only framebuffer stuff I have had working was the permedia
chip.  I have a problem with blanks not being written to the screen
(and so have left over text mode garabage) but otherwise it all works.

> The other thing we've seen here is that 2.4.16 won't work with Two Kernel
> Monte any more, we're not sure why. 

Ron.  As of 2.4.14 the kexec stuff was working fine.  I'll check back in
a few days and see if it works on 2.4.16.  I haven't published because
I haven't yet verified that the alpha works...

> Everything up to 2.4.13 is fine.
> 2.4.16 just hangs, and that's when we found the vmware problem too.
> Something changed in 2.4.16 startup, and that's the main reason I don't
> recommend it for linuxbios-based systems. Maybe linuxbios should start
> contributing to regression testing with new kernel versions? Can we help
> somehow?
> 
> Finally, I don't know if anyone knows this, but big chunks of MTD won't
> even compile from 2.4.13 on if you turn them on. If memory serves it's the
> flash translations layer stuff but I can get you more specifics. If you
> want to see yourself just turn ALL the mtd stuff on, about 25% of it won't
> compile.

Ron big chunks of the MTD code was initially poorly factored.  And that stuff
has broken.  Things link amd_flash.c.  Last I looked this wasn't a
real problem.

Eric

Reply via email to