[llvm-bugs] [Bug 29289] New: 9 piglit builtin-float-* tests crash with UNREACHABLE executed at at AMDGPUAsmBackend.cpp:125

Mon Aug 29 05:51:19 PDT 2016

https://llvm.org/bugs/show_bug.cgi?id=29289

            Bug ID: 29289
           Summary: 9 piglit builtin-float-* tests crash with UNREACHABLE
                    executed at at AMDGPUAsmBackend.cpp:125
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Keywords: regression
          Severity: normal
          Priority: P
         Component: Backend: AMDGPU
          Assignee: unassignedbugs at nondot.org
          Reporter: pavel.ondracka at email.cz
                CC: llvm-bugs at lists.llvm.org, Matthew.Arsenault at amd.com
    Classification: Unclassified

Problematic tests:
builtin-float-acosh-1.0
builtin-float-asinh-1.0
builtin-float-atanh-1.0
builtin-float-cbrt-1.0
builtin-float-cosh-1.0
builtin-float-cospi-1.0
builtin-float-log10-1.0
builtin-float-log1p-1.0
builtin-float-sinpi-1.0
program-scope-arrays

Ir crashes with:
unhandled fixup kind
UNREACHABLE executed at
..../llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUAsmBackend.cpp:125!

print Fixup
$2 = (const llvm::MCFixup &) @0x6caaa0: {Value = 0x874440, Offset = 424, Kind =
18, Loc = {Ptr = 0x0}}

backtrace:
#0  0x00007ffff524a6f5 in raise () from /lib64/libc.so.6
#1  0x00007ffff524c2fa in abort () from /lib64/libc.so.6
#2  0x00007ffff1fffd60 in llvm::llvm_unreachable_internal
(msg=msg at entry=0x7ffff3e61b84 "unhandled fixup kind", 
    file=file at entry=0x7ffff3e61b20
"/home/ondracka/graphics/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUAsmBackend.cpp",
line=line at entry=125)
    at /home/ondracka/graphics/llvm/lib/Support/ErrorHandling.cpp:118
#3  0x00007ffff31f3fdd in adjustFixupValue (Fixup=..., Value=<optimized out>,
Ctx=<optimized out>)
    at
/home/ondracka/graphics/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUAsmBackend.cpp:125
#4  0x00007ffff31f4258 in (anonymous namespace)::AMDGPUAsmBackend::applyFixup
(this=0xf1ee80, Fixup=..., Data=0x8a7960 "\022", DataSize=1104, 
    Value=<optimized out>, IsPCRel=<optimized out>) at
/home/ondracka/graphics/llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUAsmBackend.cpp:146
#5  0x00007ffff2db4461 in llvm::MCAssembler::layout (this=this at entry=0xf1f2d0,
Layout=...) at /home/ondracka/graphics/llvm/lib/MC/MCAssembler.cpp:710
#6  0x00007ffff2db4799 in llvm::MCAssembler::Finish (this=0xf1f2d0) at
/home/ondracka/graphics/llvm/lib/MC/MCAssembler.cpp:719
#7  0x00007ffff267826d in llvm::AsmPrinter::doFinalization (this=0xf1fbf0,
M=...) at
/home/ondracka/graphics/llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp:1252
#8  0x00007ffff215371c in llvm::FPPassManager::doFinalization (this=0xeea360,
M=...) at /home/ondracka/graphics/llvm/lib/IR/LegacyPassManager.cpp:1561
#9  0x00007ffff215d915 in (anonymous namespace)::MPPassManager::runOnModule
(M=..., this=<optimized out>)
    at /home/ondracka/graphics/llvm/lib/IR/LegacyPassManager.cpp:1617
#10 llvm::legacy::PassManagerImpl::run (this=0x6b4ef0, M=...) at
/home/ondracka/graphics/llvm/lib/IR/LegacyPassManager.cpp:1702
#11 0x00007ffff215da89 in llvm::legacy::PassManager::run
(this=this at entry=0x7fffffffbfd0, M=...)
    at /home/ondracka/graphics/llvm/lib/IR/LegacyPassManager.cpp:1733
#12 0x00007ffff656eede in (anonymous namespace)::emit_code (mod=...,
target=..., ft=ft at entry=llvm::TargetMachine::CGFT_ObjectFile, r_log="")
    at
../../../../../src/gallium/state_trackers/clover/llvm/codegen/native.cpp:138
#13 0x00007ffff656f4dd in clover::llvm::build_module_native (mod=...,
target=..., c=..., r_log="")
    at
../../../../../src/gallium/state_trackers/clover/llvm/codegen/native.cpp:150
#14 0x00007ffff656b7e2 in clover::llvm::link_program (modules=...,
ir=PIPE_SHADER_IR_NATIVE, target="caicos-r600--", opts=..., r_log="")
    at ../../../../../src/gallium/state_trackers/clover/llvm/invocation.cpp:289
#15 0x00007ffff655f151 in clover::program::link (this=this at entry=0x6800b0,
devs=..., opts=" -cl-std=CL1.1", progs=...)
    at ../../../../../src/gallium/state_trackers/clover/core/program.cpp:82
#16 0x00007ffff653d8f9 in clBuildProgram (d_prog=0x6800b8, num_devs=1,
d_devs=0x666a10, p_opts=<optimized out>, pfn_notify=0x0, user_data=0x0)
    at ../../../../../src/gallium/state_trackers/clover/api/program.cpp:186
#17 0x00007ffff7bce540 in piglit_cl_build_program_with_source_extended
(context=0x6677c0, count=1, strings=0x7fffffffd608, 
    options=0x680560 " -cl-std=CL1.1", fail=false) at
/home/ondracka/graphics/piglit/tests/util/piglit-util-cl.c:805
#18 0x00007ffff7bce6ab in piglit_cl_build_program_with_source
(context=0x6677c0, count=1, strings=0x7fffffffd608, options=0x680560 "
-cl-std=CL1.1")
    at /home/ondracka/graphics/piglit/tests/util/piglit-util-cl.c:851
#19 0x00007ffff7bd313d in piglit_cl_program_test_run (argc=2,
argv=0x7fffffffd8d8, void_config=0x6102c0 <config>, version=11, 
    platform_id=0x7ffff7bc2b60 <(anonymous namespace)::_clover_platform>,
device_id=0x646478)
    at
/home/ondracka/graphics/piglit/tests/util/piglit-framework-cl-program.c:257
#20 0x00007ffff7bd1ad4 in piglit_cl_framework_run (argc=2, argv=0x7fffffffd8d8)
at /home/ondracka/graphics/piglit/tests/util/piglit-framework-cl.c:344
#21 0x0000000000402785 in main (argc=2, argv=0x7fffffffd8d8) at
/home/ondracka/graphics/piglit/tests/cl/program/program-tester.c:277

git blame showed that the UNREACHABLE was added in this commit and indeed a
revert is enough to make the tests pass again.

commit d97fe48fe0577cb0b66145490e2331bbd1de0b21
Author: Matt Arsenault <Matthew.Arsenault at amd.com>
Date:   Sat Aug 27 00:21:22 2016 +0000

    AMDGPU: Improve error reporting for maximum branch distance

    Unfortunately this seems to only help the assembler diagnostic.

My system:
GPU: Advanced Micro Devices, Inc. [AMD/ATI] Caicos XTX [Radeon HD 8490 / R5
235X OEM]
kernel:4.6.4-301.fc24.x86_64
llvm:922af1cb46bb89a7bdbf68dfe77b15d1347441d7
clang:47e160c4ac62f7354755fe6063896fba58e9a71d
libclc:6c3f6de9017feea72e22be98d02460239fa10717
mesa:5f0ea3db16706968278da99cccc511e5b53e190d

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20160829/f17e5796/attachment.html>