duktape: Crash near DUK_UNLIKELY?
I use duktape on android devices, the crash reports reflects a frequently crash at duk_api_stack.c:264:
file: duktape.c
17449
17450 DUK_ASSERT_CTX_VALID(ctx);
17451 DUK_ASSERT(DUK_INVALID_INDEX < 0);
17452
17453 vs_size = (duk_idx_t) (thr->valstack_top - thr->valstack_bottom);
17454 DUK_ASSERT(vs_size >= 0);
17455
17456 if (index < 0) {
17457 index = vs_size + index;
17458 if (DUK_UNLIKELY(index < 0)) { // crash at this line
17459 return NULL;
17460 }
17461 } else {
17462 DUK_ASSERT(index != DUK_INVALID_INDEX);
17463 if (DUK_UNLIKELY(index >= vs_size)) {
17464 return NULL;
17465 }
17466 }
17467
17468 DUK_ASSERT(index >= 0);
the crash stack:
signal 11 (SIGSEGV) code 1 (SEGV_MAPERR) fault addr 0000004c
r0 6f6cf1b8 r1 00000000 r2 400bdbd4 r3 80596fc0
r4 00000000 r5 6ddb5df0 r6 400bdbd4 r7 6f6cf1b8
r8 400bdbd4 r9 00000000 10 00000388 fp 00000067
ip befcc730 sp befcbe00 lr 6dd965b5 pc 6dd961e2 cpsr 400f0030
00 pc 000241e2 my.so
00 pc 000241e2 my.so
01 lr 000245b5 my.so
Can anybody give me some clue on this crash?
About this issue
- Original URL
- State: closed
- Created 8 years ago
- Comments: 19 (11 by maintainers)
Hm, 48MB seems a little tight. If you end up doing a lot of allocations and you’re not using a pooled allocator, I could see running out of memory, due to fragmentation if nothing else. Not saying that’s the issue here, but it’s a possibility.
The crash in
duk_push_heapptr()looks like you’re mismanaging a heap pointer somewhere and the object it points to has been GC’d. Duktape has no sanity checks for that so if you push a stale heap pointer you will segfault. I would go through the code and double check that all heap pointers are being stashed to keep them reachable.If so, it should be preceded by an alloc/realloc call returning NULL.