>>>>> "Tzafrir" == Tzafrir Cohen <tzafrir@cohens.org.il> writes: > On Sun, Aug 29, 2010 at 01:00:13AM +0800, liushiwei wrote: >> 2010/8/29 Tzafrir Cohen <tzafrir@cohens.org.il>: > Hi, >> > >> > Thanks for the quick reply, >> > >> > On Sat, Aug 28, 2010 at 11:06:58PM +0800, liushiwei wrote: >> >> yeeloong run debia6, just above this 3 patches. >> >> [..] >> > >> > I get the following error, either when I try to build your package >> or > when I try to apply your second patch: >> > [..] >> /tmp/ccBqSRfw.s: Assembler messages: > /tmp/ccBqSRfw.s:4469: Error: >> illegal operands `xor $f0,$f0,$f0' > /tmp/ccBqSRfw.s:4470: Error: >> opcode not supported on this processor: > mips1 (mips1) `ldc1 >> $f4,($6)' > /tmp/ccBqSRfw.s:4471: >> >> this patch use loongson mmi , so gcc4.4 and -march=loongson2f. [..] > Is this patch Loongson-specific? Looks like the 3rd hunk of the 99_xf86-video-siliconmotion-1.7.3-fix-loongson.diff is the culprit here. It adds Loongsoon-specific SIMD optimizations to X-video. I can't find any CPU detection, so the patch makes the siliconmotion video driver loongson specific. Whoever did this did not spend any thoughts about upstream, just tried to somehow make it work for himself. Looks like it could be safe to remove the simd stuff by omitting hunks 3 and 4 of the patch. Other parts of the patch change the behaviour of many parts without any options to select the old behaviour, so I guess this stuff also might make upstream unhappy. The comment at the top of the patch says: A patch for siliconmotion 1.7.3 is available in the bug trackers at freedesktop.org, but a better one will be available monday. Still, it lacks the most interresting part : the MMX pack function, which should not be implemented there but probably either directly in Xv or in pixman (and make Xv use pixman). However, searching the bugtracker did not turn up anything. https://bugs.freedesktop.org/buglist.cgi?quicksearch=siliconmotion Note that I don't own any Longsoon hardware yet. But I'm considering it. Seeing these kind of hacky patches makes me feel a little uncomfortable currently. Hopefully things will stabilize in the future. cheers, David -- GnuPG public key: http://user.cs.tu-berlin.de/~dvdkhlng/dk.gpg Fingerprint: B17A DC95 D293 657B 4205 D016 7DEF 5323 C174 7D40
Attachment:
pgp3cQNkfiuUu.pgp
Description: PGP signature