https://bugs.freedesktop.org/show_bug.cgi?id=66806
Priority: medium
Bug ID: 66806
Keywords: regression
CC: za...@vmware.com
Assignee: mesa-dev@lists.freedesktop.org
Summary: [softpipe] glxgears floating point exception
Severity: critical
Classification: Unclassified
OS: Linux (All)
Reporter: v...@freedesktop.org
Hardware: x86-64 (AMD64)
Status: NEW
Version: git
Component: Mesa core
Product: Mesa
mesa: a06ee5a09e7072aab29abac7a3afe7b04b6d6ee6 (master)
Run glxgears on softpipe.
$ glxgears
Floating point exception (core dumped)
Core was generated by `glxgears'.
Program terminated with signal 8, Arithmetic exception.
#0 0x00007fe644cece5f in micro_mul (dst=0x7fff545deb50, src0=0x7fff545deb30,
src1=0x7fff545deb40)
at tgsi/tgsi_exec.c:983
983 dst->f[0] = src0->f[0] * src1->f[0];
(gdb) bt
#0 0x00007fe644cece5f in micro_mul (dst=0x7fff545deb50, src0=0x7fff545deb30,
src1=0x7fff545deb40)
at tgsi/tgsi_exec.c:983
#1 0x00007fe644cf151c in exec_vector_binary (mach=0x7fe642910010,
inst=0x19a30c0,
op=0x7fe644cece3f <micro_mul>, dst_datatype=TGSI_EXEC_DATA_FLOAT,
src_datatype=TGSI_EXEC_DATA_FLOAT)
at tgsi/tgsi_exec.c:2549
#2 0x00007fe644cf5048 in exec_instruction (mach=0x7fe642910010,
inst=0x19a30c0, pc=0x7fff545deca4)
at tgsi/tgsi_exec.c:3513
#3 0x00007fe644cf7ebc in tgsi_exec_machine_run (mach=0x7fe642910010) at
tgsi/tgsi_exec.c:4366
#4 0x00007fe644cc69fa in vs_exec_run_linear (shader=0x19a1dd0,
input=0x19a5e64, output=0x19a7764,
constants=0x18cbe88, const_size=0x18cbf88, count=80, input_stride=84,
output_stride=84)
at draw/draw_vs_exec.c:155
#5 0x00007fe644cba048 in draw_vertex_shader_run (vshader=0x19a1dd0,
constants=0x18cbe88, const_size=0x18cbf88,
input_verts=0x7fff545dee00, output_verts=0x7fff545dee20) at
draw/draw_pt_fetch_shade_pipeline.c:207
#6 0x00007fe644cba2b0 in fetch_pipeline_generic (middle=0x18d9300,
fetch_info=0x0, in_prim_info=0x7fff545def50)
at draw/draw_pt_fetch_shade_pipeline.c:266
#7 0x00007fe644cba68f in fetch_pipeline_linear_run (middle=0x18d9300,
start=82, count=80, prim_flags=0)
at draw/draw_pt_fetch_shade_pipeline.c:400
#8 0x00007fe644cc312c in vsplit_segment_simple_linear (vsplit=0x18d6710,
flags=0, istart=82, icount=80)
at draw/draw_pt_vsplit_tmp.h:240
#9 0x00007fe644cc3414 in vsplit_run_linear (frontend=0x18d6710, start=82,
count=80) at draw/draw_split_tmp.h:60
#10 0x00007fe644cb5b29 in draw_pt_arrays (draw=0x18cb670, prim=7, start=82,
count=80) at draw/draw_pt.c:149
#11 0x00007fe644cb68fb in draw_vbo (draw=0x18cb670, info=0x7fff545df0b0) at
draw/draw_pt.c:560
#12 0x00007fe644dca6e0 in softpipe_draw_vbo (pipe=0x18324a0,
info=0x7fff545df200) at sp_draw_arrays.c:128
#13 0x00007fe644c9c7da in cso_draw_vbo (cso=0x1966930, info=0x7fff545df200) at
cso_cache/cso_context.c:1413
#14 0x00007fe644babaa2 in st_draw_vbo (ctx=0x18facc0, prims=0x1959580,
nr_prims=2, ib=0x0,
index_bounds_valid=1 '\001', min_index=0, max_index=161, tfb_vertcount=0x0)
at ../../src/mesa/state_tracker/st_draw.c:286
#15 0x00007fe644b8b95c in vbo_save_playback_vertex_list (ctx=0x18facc0,
data=0x1958de8)
at ../../src/mesa/vbo/vbo_save_draw.c:309
#16 0x00007fe644a29eb3 in ext_opcode_execute (ctx=0x18facc0, node=0x1958de0) at
../../src/mesa/main/dlist.c:598
#17 0x00007fe644a3f3f7 in execute_list (ctx=0x18facc0, list=1) at
../../src/mesa/main/dlist.c:7333
#18 0x00007fe644a45880 in _mesa_CallList (list=1) at
../../src/mesa/main/dlist.c:8733
#19 0x0000000000401a72 in ?? ()
#20 0x0000000000403176 in ?? ()
#21 0x00007fe647141ea5 in __libc_start_main (main=0x402aa0, argc=1,
ubp_av=0x7fff545dfbf8,
init=<optimized out>, fini=<optimized out>, rtld_fini=<optimized out>,
stack_end=0x7fff545dfbe8)
at libc-start.c:260
#22 0x00000000004017b9 in ?? ()
#23 0x00007fff545dfbe8 in ?? ()
#24 0x000000000000001c in ?? ()
#25 0x0000000000000001 in ?? ()
#26 0x00007fff545e0440 in ?? ()
#27 0x0000000000000000 in ?? ()
63386b2f66a6d450889cd5368bc599beb7f1efbf is the first bad commit
commit 63386b2f66a6d450889cd5368bc599beb7f1efbf
Author: Zack Rusin <za...@vmware.com>
Date: Mon Jul 8 23:45:55 2013 -0400
util: treat denorm'ed floats like zero
The D3D10 spec is very explicit about treatment of denorm floats and
the behavior is exactly the same for them as it would be for -0 or
+0. This makes our shading code match that behavior, since OpenGL
doesn't care and on a few cpu's it's faster (worst case the same).
Float16 conversions will likely break but we'll fix them in a follow
up commit.
Signed-off-by: Zack Rusin <za...@vmware.com>
Reviewed-by: Jose Fonseca <jfons...@vmware.com>
Reviewed-by: Roland Scheidegger <srol...@vmware.com>
:040000 040000 a92f3a68cc7b1ea78d521df7ea88d347359d0e7f
6ef306940ae30adc64843a7943458d2e681891cc M src
bisect run success
--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
