Add pthread mutex deadlock detection for debug builds #24607
Conversation
sbc100
force-pushed
the
deadlock_detection
branch
from
a57b588 to
0e10a5e
Compare
sbc100
force-pushed
the
deadlock_detection
branch
from
0e10a5e to
efb88c5
Compare
| if ((m->_m_type&15) == PTHREAD_MUTEX_NORMAL | ||
| && !a_cas(&m->_m_lock, 0, EBUSY)) | ||
| return 0; | ||
| #endif |
There was a problem hiding this comment.
Should upstream do the same for all platforms? I'm just trying to understand if this is a general improvement or if it works around a specific problem in wasm.
There was a problem hiding this comment.
I think its a general improvement for debug builds yes. I guess most libc's don't ship debug builds like we do (especially automatically selecting debug libs based on command line flags, which emcc does is quite unusual).
There was a problem hiding this comment.
Upstream musl doesn't really have any/many assertions AFAICT, so its not something they tend to consider I suppose?
There was a problem hiding this comment.
Indeed, it looks like upstream musl only has assertion in 2 very specific places:
src/regex/regcomp.c
src/malloc/mallocng/..
No other assertions in whole of libc.
This would really have helped in at least detecting the deadlock that was found in emscripten-core#24570 / emscripten-core#24565
sbc100
force-pushed
the
deadlock_detection
branch
from
efb88c5 to
a8d3d3f
Compare
This would really have helped in at least detecting the deadlock that was found in #24570 / #24565