An r300g user seems to hit this issue when st/mesa uses Draw for feedback mode, in Compiz. There is the backtrace attached in the bug report below. The reporter doesn't use LLVM too and he bisected the same bad commit as you (gallium: implement bounds checking for constant buffers).<br> <br><a href="https://bugs.freedesktop.org/show_bug.cgi?id=29311">https://bugs.freedesktop.org/show_bug.cgi?id=29311</a><br><br>Best regards.<br><br>-Marek<br><br><div class="gmail_quote">On Fri, Jul 30, 2010 at 11:52 PM, Jerome Glisse <span dir="ltr"><<a href="mailto:glisse@freedesktop.org">glisse@freedesktop.org</a>></span> wrote:<br> <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"><div class="im">On 07/30/2010 05:02 PM, Jakob Bornecrantz wrote:<br> <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> On 30 jul 2010, at 13.32, Brian Paul wrote:<br> <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> On 07/30/2010 12:38 PM, Jerome Glisse wrote:<br> <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> Hi Brian,<br> <br> I am facing a strange segfault with r600g on top of lastest git,<br> git bisect pointed to gallium: implement bounds checking for constant buffers<br> My feeling is that it should only affect software pipeline but<br> somehow r600g seem to take different path now, attached if full<br> but i can't make much sense out of it, do you have a clue on what<br> might went wrong ?<br> </blockquote> <br> I took a quick look but didn't find anything.<br> <br> Maybe try a make clean and rebuild just in case?<br> </blockquote> <br> I'm getting the same with swrastg on in 32bit VM, "git clean -fdx":ed even.<br> <br> Cheers Jakob.<br> </blockquote> <br></div> Brian, it only affects SSE path, i thought change to struct tgsi_exec_machine<br> was causing some of the generated code to access wrong data but it doesn't<br> seems to be that (moved ConstsSize to bottom of struct def without effect).<br><font color="#888888"> <br> Jerome</font><div><div></div><div class="h5"><br> _______________________________________________<br> mesa-dev mailing list<br> <a href="mailto:mesa-dev@lists.freedesktop.org" target="_blank">mesa-dev@lists.freedesktop.org</a><br> <a href="http://lists.freedesktop.org/mailman/listinfo/mesa-dev" target="_blank">http://lists.freedesktop.org/mailman/listinfo/mesa-dev</a><br> </div></div></blockquote></div><br>