To reproduce:
$ cat test.cpp
int main() { return 0; }
$ g++ -g -O0 -l GLX -fsanitize=address,undefined test.cpp
ldd a.out
linux-vdso.so.1 (0x00007ffc9d5d7000)
libasan.so.6 => /usr/lib/libasan.so.6 (0x00007f8da32df000)
libGLX.so.0 => /usr/lib/libGLX.so.0 (0x00007f8da32ac000)
libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x00007f8da30cf000)
libm.so.6 => /usr/lib/libm.so.6 (0x00007f8da2f89000)
libubsan.so.1 => /usr/lib/libubsan.so.1 (0x00007f8da261b000)
libgcc_s.so.1 => /usr/lib/libgcc_s.so.1 (0x00007f8da2601000)
libc.so.6 => /usr/lib/libc.so.6 (0x00007f8da2436000)
libdl.so.2 => /usr/lib/../lib/libdl.so.2 (0x00007f8da2430000)
librt.so.1 => /usr/lib/../lib/librt.so.1 (0x00007f8da2425000)
libpthread.so.0 => /usr/lib/../lib/libpthread.so.0 (0x00007f8da2403000)
libGLdispatch.so.0 => /usr/lib/libGLdispatch.so.0 (0x00007f8da234c000)
libX11.so.6 => /usr/lib/libX11.so.6 (0x00007f8da220b000)
/lib64/ld-linux-x86-64.so.2 => /usr/lib64/ld-linux-x86-64.so.2 (0x00007f8da3cfa000)
libxcb.so.1 => /usr/lib/libxcb.so.1 (0x00007f8da21df000)
libXau.so.6 => /usr/lib/libXau.so.6 (0x00007f8da21da000)
libXdmcp.so.6 => /usr/lib/libXdmcp.so.6 (0x00007f8da21d2000)
$ ./a.out
# works fine
$ rr record ./a.out
# never finishes
according to perf, this is what happens:
I guess there's something fishy going on with asan and rr both trying to intercept pthread_mutex_lock here. Oddly enough, a simpler application with a normal mutex seems to work. So I guess it's also important that the mutex is locked from a global static that's called from dl_init in a library?
To reproduce:
according to perf, this is what happens:
I guess there's something fishy going on with asan and rr both trying to intercept pthread_mutex_lock here. Oddly enough, a simpler application with a normal mutex seems to work. So I guess it's also important that the mutex is locked from a global static that's called from dl_init in a library?