Open GoogleCodeExporter opened 9 years ago
One thing that limits us is the number of bits we can spend for thread id.
Currently it is 24:
struct ChunkHeader {
// 1-st 8 bytes.
u32 chunk_state : 8; // Must be first.
u32 alloc_tid : 24;
u32 free_tid : 24;
...
Original comment by konstant...@gmail.com
on 12 Mar 2014 at 4:32
I keep seeing such errors:
=================================================================
==26629==ERROR: AddressSanitizer: attempting free on address which was not
malloc()-ed: 0x000001e28790 in thread T16777215
==26629==AddressSanitizer CHECK failed:
/local/mar_/llvm-3.4/projects/compiler-rt/lib/asan/asan_report.cc:595
"((free_stack->size)) > ((0))" (0x0, 0x0)
Can you confirm this is the same issue?
Is there some workaround possible?
Thanks,
Martin
Original comment by mric...@googlemail.com
on 8 May 2014 at 3:52
P.S. What pointed me to this bug was the thread id, which is always: T16777215
(0xffffff).
Also, it looks like these crashes happen mostly on process shutdown.
Best regards,
Martin
Original comment by mric...@googlemail.com
on 8 May 2014 at 3:55
I don't think your crashes are related to this bug. It may be a real bug in the
code.
How many threads do you have? (run with ASAN_OPTIONS=verbosity=1 if you don't
know).
Do you have a reproducer?
Original comment by konstant...@gmail.com
on 8 May 2014 at 4:35
Will try to reproduce it.
Original comment by mric...@googlemail.com
on 9 May 2014 at 11:57
I reproduced it with verbosity=1, attaching the log.
==61871==ERROR: AddressSanitizer: attempting free on address which was not
malloc()-ed: 0x7f2192ae7010 in thread T16777215
==61871==AddressSanitizer CHECK failed:
/local/mar_/llvm-3.4/projects/compiler-rt/lib/asan/asan_report.cc:595
"((free_stack->size)) > ((0))" (0x0, 0x0)
Unfortunately this is a complex test, so no reproducer yet. Let me know if I
should enable some other trace. The error happens when a server is shutting
down, and the process crashes with signal 11 (after the ASan report).
Original comment by mric...@googlemail.com
on 1 Nov 2014 at 11:52
Attachments:
Please create a separate bug for this issue, this one is about something
completely different.
Original comment by konstant...@gmail.com
on 5 Nov 2014 at 1:18
Original issue reported on code.google.com by
konstant...@gmail.com
on 11 Mar 2014 at 8:51