On Wed, Sep 8, 2010 at 17:10:36 +0100, Chris Wilson wrote: > On Wed, 8 Sep 2010 17:34:39 +0200, Julien Cristau <jcristau@debian.org> wrote: > > Hi Chris, > > > > does the following look like a known bug with 2.12? > > It might be an issue with 2.6.32 between the memory corruption and > 945GM low-power render failure. > Florian's log says he's running our 2.6.32-20 kernel, which has both i915 self-reclaim patches (included in -17 and -18) and the low-power render fix (included in -19). > As it looks like uninitialised data, it could be the xserver bug where we > sent damage events prior to flushing the 2D batchbuffer, so that there was > an opportunity for compositing WM to grab garbage. > That one's plausible, thanks. Florian, are you running a compositing manager? If yes, does the corruption happen without that? Looks like the fix for that one is 8d7b7a0d71e0b89321b3341b781bc8845386def6 and c65f610e12f9df168d5639534ed3c2bd40afffc8? > If you can verify that it is not either of those bugs, then yeah it's an > unresolved bug. There are a couple of similar bugs still in fd.o that I'm > still waiting on feedback for (but I think are due to the above). > -Chris > Cheers, Julien
Attachment:
signature.asc
Description: Digital signature