https://bugs.freedesktop.org/show_bug.cgi?id=106533
michael.panzlaff@fau.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |michael.panzlaff@fau.de
--- Comment #3 from michael.panzlaff@fau.de --- Created attachment 139995 --> https://bugs.freedesktop.org/attachment.cgi?id=139995&action=edit registers, stack and jit function disassembly
(In reply to Roland Scheidegger from comment #2)
I suppose it crashes in the jit-compiled code (debug symbols should help with identifying that, but not help any further if that's the case). Can you print out the faulting instruction (gdb x/i address or so)? Also, if that would be some SSE instruction, try to see if the memory operand is aligned (or just not addressable).
I've tried to get the game "Thimbleweed Park" to run and it seems like it's crashing for a very similar reason. This is the stack trace:
#0 0x00007ffff7fdc000 in ?? () #1 0x00007ffff27288d8 in llvm_pipeline_generic (middle=middle@entry=0x2bc9110, fetch_info=fetch_info@entry=0x7fffffffdf80, in_prim_info=in_prim_info@entry=0x7fffffffdfa0) at draw/draw_pt_fetch_shade_pipeline_llvm.c:408 #2 0x00007ffff2728f86 in llvm_middle_end_linear_run (middle=0x2bc9110, start=0, count=<optimized out>, prim_flags=0) at draw/draw_pt_fetch_shade_pipeline_llvm.c:588 #3 0x00007ffff2635d56 in vsplit_segment_simple_linear (vsplit=0x2bc6340, vsplit=0x2bc6340, icount=4, istart=0, flags=0) at draw/draw_pt_vsplit_tmp.h:226 #4 vsplit_run_linear (frontend=0x2bc6340, start=0, count=4) at draw/draw_split_tmp.h:70 #5 0x00007ffff262d71a in draw_pt_arrays (draw=draw@entry=0x2ba3b20, prim=6, start=0, count=count@entry=4) at draw/draw_pt.c:175 #6 0x00007ffff262df50 in draw_vbo (draw=0x2ba3b20, info=0x7fffffffe0d0, info@entry=0x7fffffffe1a0) at draw/draw_pt.c:609 #7 0x00007ffff273b319 in r300_swtcl_draw_vbo (pipe=0x2b7ac80, info=0x7fffffffe1a0) at r300_render.c:862 #8 0x00007ffff273d9e6 in r300_stencilref_draw_vbo (pipe=0x2b7ac80, info=0x7fffffffe1a0) at r300_render_stencilref.c:113 #9 0x00007ffff261cce7 in cso_draw_arrays (cso=<optimized out>, mode=mode@entry=6, start=start@entry=0, count=count@entry=4) at cso_cache/cso_context.c:1724 #10 0x00007ffff2413ee4 in st_draw_quad (st=st@entry=0x2cbddb0, x0=x0@entry=-1, y0=y0@entry=-0.899999976, x1=x1@entry=1, y1=y1@entry=0.899999976, z=1, s0=s0@entry=0, t0=t0@entry=0, s1=s1@entry=0, t1=0, color=color@entry=0x2c9a44c, num_instances=num_instances@entry=1) at state_tracker/st_draw.c:435 #11 0x00007ffff23f8df1 in clear_with_quad (clear_buffers=<optimized out>, ctx=0x2c987c0) at state_tracker/st_cb_clear.c:300 #12 st_Clear (ctx=0x2c987c0, mask=2) at state_tracker/st_cb_clear.c:454 #13 0x00007ffff2244dc5 in clear (no_error=false, mask=<optimized out>, ctx=0x2c987c0) at main/clear.c:221 #14 _mesa_Clear (mask=<optimized out>) at main/clear.c:244 #15 0x000000000049e364 in ?? () #16 0x0000000000481fd3 in ?? () #17 0x000000000048359f in ?? () #18 0x00007ffff6c9aa87 in __libc_start_main (main=0x40e130, argc=1, argv=0x7fffffffe5f8, init=<optimized out>, fini=<optimized out>, rtld_fini=<optimized out>, stack_end=0x7fffffffe5e8) at ../csu/libc-start.c:310 #19 0x000000000040f04a in ?? ()
To answer your original question I've provided information (stack, registers, assembly of jit function) in the attached text file.
PS: Hope I did everything correct. First time for me posting something on the bugzilla.