cfb_* / fbcon vs. 64 bits

cfb_* / fbcon vs. 64 bits

[Benjamin Herrenschmidt]

Hi !

Did anybody tested a non-accelerated framebuffer device (that is one
using the cfb_* functions) on a 64 bits arch ?

Here, with radeonfb, the mode gets properly sync'ed, but I get garbage
on the first lines where the logo should be, and then other junk that
looks like some address calculations aren't done properly in the
blitting code.

Same machine, same driver, with a 32 bits kernel it works just fine.

And yes, the driver itself seem to be passing properly the 64 bits
result of ioremap, but taht could be the problem. On ppc64, ioremap
returns "weird" addresses which are token that only readx/writex
can actually understand. Maybe I should try to "munge" this a bit
first...

Ben.





-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?   SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

Re: cfb_* / fbcon vs. 64 bits

[Benjamin Herrenschmidt]

> And yes, the driver itself seem to be passing properly the 64 bits
> result of ioremap, but taht could be the problem. On ppc64, ioremap
> returns "weird" addresses which are token that only readx/writex
> can actually understand. Maybe I should try to "munge" this a bit
> first...

No, in this case, the "eeh" stuff isn't activated, so the address
should be correct. I haven't yet figured out what's up. offb seem
to not work neither.

We tried on an IBM ppc64 box as well, same results. Matroxfb seem
to work though....

Ben.




-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?   SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/