segfault at GetMaxWatchpointCount

[laolei]

Hi,
When I try to debug the matrix sample, segment fault happened within the close source .so module. Could you give a clue to solve/workaround?

(ROCm-gdb) run
Starting program: /home/gedu/work/rocm/ROCm-GPUDebugSDK/samples/MatrixMultiplication/MatrixMul
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Initializing HSA runtime...
[New Thread 0x7ffff470d700 (LWP 19982)]
[ROCm-gdb: <Handling a SIGALRM - Used for GPU Debugging>]
[ROCm-gdb: <Handling a SIGALRM - Used for GPU Debugging>]
[ROCm-gdb: GPU Debugging has been successfully initialized]
[New Thread 0x7ffff34b1700 (LWP 19983)]
HSA device attributes:
name: AMD A10-8700P Radeon R6, 10 Compute Cores 4C+6G
type: CPU
chip ID: 0x9874
HSA profile: Full
HSA device attributes:
name: gfx801
type: GPU
chip ID: 0x9874
HSA profile: Full

Thread 1 "MatrixMul" hit Breakpoint 2, 0x00007ffff3845f40 in HwDbgDataBreakpointManager::GetMaxWatchpointCount() const@plt () from ./../../lib/x86_64/libAMDGPUDebugHSA-x64.so
1: x/3i $pc
=> 0x7ffff3845f40 _ZNK26HwDbgDataBreakpointManager21GetMaxWatchpointCountEv@plt: jmp QWORD PTR [rip+0x42e3ca] # 0x7ffff3c74310 _ZNK26HwDbgDataBreakpointManager21GetMaxWatchpointCountEv@got.plt
0x7ffff3845f46 <_ZNK26HwDbgDataBreakpointManager21GetMaxWatchpointCountEv@plt+6>: push 0x25f
0x7ffff3845f4b <_ZNK26HwDbgDataBreakpointManager21GetMaxWatchpointCountEv@plt+11>: jmp 0x7ffff3843940
(ROCm-gdb)

[j-stephan]

Did you manage to fix your problem? I'm currently experiencing a similar issue with ROCm 1.8 and a custom kernel. I'm unsure if this is my fault or due to the fact that rocm-gdb hasn't been updated in a while.