Jonathan Nieder <jrnieder@gmail.com> (21/01/2012): > (-cc: Witold) (that's not what I see, fixing) > Ok, sorry about that. My choice was reassign a kernel bug concerning > the same problem to libdrm or the libdrm bug to the kernel, or to > leave them unmerged. I should have said so, I guess. Leaving bugs as-is was very fine, at least with my libdrm hat. > Honestly, a little kindness could have helped a lot. I guess you mean > that you do not suspect this is a kernel bug? You could easily be > right, which is why upstream does not bother to separate DRI bugs from > libdrm ones. Though in that case the error message could be a lot > better. Since you seem to be missing the big picture, several components are at play here: kernel (+ possibly firmware in the radeon case), libdrm, x server, x driver, mesa. And if one of them is sending shitty commands to the GPU, anything can break havoc. Usually, with the GPU hanging. That does *not* mean the kernel is at fault, just because there's some line in the kernel log! Back to the original topic: it is very difficult for Intel techs to support old (and I'm just echo-ing their views on hardware) hardware, even if older versions of the x driver might have worked at some point. That is sad, but that's the state of affairs today. Mraw, KiBi.
Attachment:
signature.asc
Description: Digital signature