Bug#594684: xserver-xorg-video-siliconmotion: siliconmotion driver segfaults on a Lemote YeeLoong
On 20/09/12 16:51, Roman Mamedov wrote:
> I do not have a good solution to resolving this bug, I just think no one
> would've been harmed much, and people would be very thankful, if this would
>From what I remember, the problem is MIPS-specific (based on C
preprocessor defines), and makes it impossible to use this variety of
siliconmotion hardware on any MIPS system. (No code exists to
initialise the IOPortBase, which is later needed for doing memory-mapped
I/O, and causes the crash).
One of the patches would fix it by hard-coding a loongson-specific base
address. That doesn't seem ideal; but if there is no way this hardware
could be working on any other MIPS system, maybe that is okay for now?
The patch probably needs cleaning up to apply against the latest Xorg in
unstable. And it should be annotated to credit the original author (who
I think we were unsure of?).
IIRC the other, more objectionable patch (which would have affected
other platforms) doesn't *seem* to be needed any more. At least, I have
a working Wheezy system on which I don't remember having to apply the
sarea patch. (I must look into this...).
I was hoping (last year) that all remaining issues affecting Lemote
loongson systems might have been fixed for Wheezy, so that Debian
GNU/Linux installs & runs out-of-the-box, but unfortunately not :(
I'm also surprised how few mipsel systems show in popcon (46). Of
these, there appear to be at least 10x running the new loongson-2f
kernel flavour, 10x r5k-cobalt, 5x 4kc-malta, 2x 5kc-malta, and 1x
sb1-bcm91250a.
http://qa.debian.org/popcon-graph.php?packages=linux-image-3.2.0-4-loongson-2f+linux-image-3.2.0-3-loongson-2f+linux-image-3.2.0-2-loongson-2f+linux-image-3.2.0-1-loongson-2f
Regards,
--
Steven Chamberlain
steven@pyro.eu.org
Reply to: