Coredump while call perf2bolt in arm 4.14

[dafoerx]

uname -a 4.14.0-115.el7a.0.1.aarch64 #1 SMP Sun Nov 25 20:54:21 UTC 2018 aarch64 aarch64 aarch64 GNU/Linux

I am optimizing bin/postgres in the reference documentation, and core appears in the phase of converting perf.data in the ARM environment. Can anyone help

ref: https://github.com/facebookincubator/BOLT

[root@01 bolt]# perf2bolt -p perf.data -o perf.fdata postgres -nl
PERF2BOLT: Starting data aggregation job for perf.data
PERF2BOLT: spawning perf job to read events without LBR
PERF2BOLT: spawning perf job to read mem events
PERF2BOLT: spawning perf job to read process events
PERF2BOLT: spawning perf job to read task events
BOLT-INFO: Target architecture: aarch64
BOLT-INFO: BOLT version: fe64c4001e17a994c178d1102d7cb12eee196a1c
BOLT-INFO: first alloc address is 0x400000
BOLT-INFO: creating new program header table at address 0x1000000, offset 0xc00000
BOLT-WARNING: non-relocation mode for AArch64 is not fully supported
BOLT-INFO: disabling -align-macro-fusion on non-x86 platform
BOLT-INFO: pre-processing profile using perf data aggregator
BOLT-INFO: binary build-id is:     c07aa22e408b00530c7a0999e4e8f375ed75fd44
PERF2BOLT: spawning perf job to read buildid list
PERF2BOLT: matched build-id and file name
PERF2BOLT: waiting for perf mmap events collection to finish...
PERF2BOLT: parsing perf-script mmap events output
PERF2BOLT: waiting for perf task events collection to finish...
PERF2BOLT: parsing perf-script task events output
PERF2BOLT: input binary is associated with 884 PID(s)
PERF2BOLT: waiting for perf events collection to finish...
PERF2BOLT: parsing basic events (without LBR)...
perf2bolt: /root/test/tools/BOLT/bolt/lib/Target/AArch64/AArch64MCPlusBuilder.cpp:598: bool {anonymous}::AArch64MCPlusBuilder::analyzeIndirectBranchFragment(const llvm::MCInst&, llvm::DenseMap<const llvm::MCInst*, llvm::SmallVector<llvm::MCInst*, 4> >&, const llvm::MCExpr*&, int64_t&, int64_t&, llvm::MCInst*&) const: Assertion `DefJTBaseAdd->getOpcode() == AArch64::ADDXri && "Failed to match jump table base address pattern! (1)"' failed.
 #0 0x0000000000c90eec PrintStackTraceSignalHandler(void*) Signals.cpp:0:0
 #1 0x0000000000c8f0dc SignalHandler(int) Signals.cpp:0:0
 #2 0x0000ffff81c2066c  0x66c raise
 #3 0x0000ffff81c2066c 
 #4 0x0000ffff81c2066c abort (+0x66c)
 #5 0x0000ffff816850e8 __assert_fail_base (/lib64/libc.so.6+0x350e8)
 #6 0x0000ffff81686760 __assert_perror_fail (/lib64/libc.so.6+0x36760)
 #7 0x0000ffff8167e5dc (anonymous namespace)::AArch64MCPlusBuilder::analyzeIndirectBranch(llvm::MCInst&, llvm::bolt::MCPlusBuilder::InstructionIterator, llvm::bolt::MCPlusBuilder::InstructionIterator, unsigned int, llvm::MCInst*&, unsigned int&, unsigned int&, long&, llvm::MCExpr const*&, llvm::MCInst*&) const AArch64MCPlusBuilder.cpp:0:0
 #8 0x0000ffff8167e694 llvm::bolt::BinaryFunction::processIndirectBranch(llvm::MCInst&, unsigned int, unsigned long, unsigned long&) (/lib64/libc.so.6+0x2e694)
 #9 0x0000000000dc6e18 llvm::bolt::BinaryFunction::disassemble() (/root/test/tools/BOLT/build/bin/llvm-bolt+0xdc6e18)
#10 0x00000000019b4364 llvm::bolt::RewriteInstance::disassembleFunctions() (/root/test/tools/BOLT/build/bin/llvm-bolt+0x19b4364)
#11 0x00000000019c8164 llvm::bolt::RewriteInstance::run() (/root/test/tools/BOLT/build/bin/llvm-bolt+0x19c8164)
#12 0x0000000000a920a8 main (/root/test/tools/BOLT/build/bin/llvm-bolt+0xa920a8)
#13 0x0000000000ae0440 __libc_start_main (/root/test/tools/BOLT/build/bin/llvm-bolt+0xae0440)
#14 0x000000000042b620 _start (/root/test/tools/BOLT/build/bin/llvm-bolt+0x42b620)
Stack dump without symbol names (ensure you have llvm-symbolizer in your PATH or set the environment var `LLVM_SYMBOLIZER_PATH` to point to it):
perf2bolt[0xc90eec]
perf2bolt[0xc8f0dc]
[0xffff81c2066c]
/lib64/libc.so.6(gsignal+0x4c)[0xffff816850e8]
/lib64/libc.so.6(abort+0x11c)[0xffff81686760]
/lib64/libc.so.6(+0x2e5dc)[0xffff8167e5dc]
/lib64/libc.so.6(__assert_perror_fail+0x0)[0xffff8167e694]
perf2bolt[0xdc6e18]
perf2bolt[0x19b4364]
perf2bolt[0x19c8164]
perf2bolt[0xa920a8]
perf2bolt[0xae0440]
perf2bolt[0x42b620]
/lib64/libc.so.6(__libc_start_main+0xf0)[0xffff816715d4]
perf2bolt[0x47d4d4]
PLEASE submit a bug report to https://bugs.llvm.org/ and include the crash backtrace.
Stack dump:
0.      Program arguments: perf2bolt -p perf.data -o perf.fdata postgres -nl
Aborted (core dumped)

[maksfb]

@yota9, any idea what is happening there?

[yota9]

Hello @dafoerx. Could you please attach the binary and perf file so I will be able to check the binary? Thank you.

[dafoerx]

Sorry, the core file is too big to upload temporarily, I can provide some more detailed location information for the time being.

1>os version uname -r 4.14.0-115.el7a.0.1.aarch64 cat /etc/redhat-release CentOS Linux release 7.6.1810 (AltArch) 2>bolt code version git clone https://github.com/facebookincubator/BOLT.git git log fe64c4001e17a994c178d1102d7cb12eee196a1c 3>Repetition steps perf record -C 0-47 -e cycles:u -a -o perf.data -- sleep 180 perf2bolt -p perf.data -o perf.fdata postgres -nl

The core file is generated in aARCH64mcplusBuilder.cpp:598. I wonder what scenarios assert comes in. analyzeIndirectBranchFragment assert(DefJTBaseAdd->getOpcode() == AArch64::ADDXri && "Failed to match jump table base address pattern! (1)");

there is a position in the process of use is not strictly according to the operation, I do not know whether it affects, I modify command A(perf record -C 0-47 -e cycles:_u -j any,_u -a -o perf.data -- sleep 180) to command B(perf record -C 0-47 -e cycles:u -a -o perf.data -- sleep 180) Because my environment doesn't seem to support the -j operation.like this "Error: cycles:u: PMU Hardware doesn't support sampling/overflow-interrupts. Try 'perf stat'"

[dafoerx]

Maybe I'm similar to this issue Failed to match indirect branch when BOLT AArch64 binary #111

Is there a way around it so my experiment can go on? perf2bolt -p perf.data -o perf.fdata postgres -nl .....

[yota9]

Sorry, I don't need the core file, I need the executable at least (better with the perf file) to reproduce the problem and check its asm and to find the failure reason