Closed ramosian-glider closed 9 years ago
See r1253 - the test passes natively and with Clang, but crashes on a CHECK under ASan
at ~23500-23700 threads (the value varies a bit).
Reported by timurrrr
on 2012-03-19 13:26:26
Forgot to mention: it's the "real" CreateThread which fails
Reported by timurrrr
on 2012-03-19 13:58:24
Is this 32- or 64-bit?
I would not expect 32-bit to survive so many threads -- just not enough address space.
What's the limit on 32-bit linux?
Reported by konstantin.s.serebryany
on 2012-03-19 16:14:04
It's 32-bit.
The native app works fine and it is extracted from a small Chromium test run with a
hundred of other tests in the same process.
I was able to spawn 100k+ threads in a native execution.
It might be that our shadow allocation point conflicts with where the stack of the
new thread should be...
I've noticed that this particular test is disabled under TSan and Dr.Memory so this
isn't a high priority issue.
Reported by timurrrr
on 2012-03-19 17:29:47
I don't think this is an issue at all, just a feature.
asan takes much more address space than non-asan.
It is quite normal that on 32-bit we can't run as many threads as non-asan.
Unless the limit of the thread number on Windows is much smaller than on linux (with
the same thread stack size), I suggest to close this as "WontFix".
Reported by konstantin.s.serebryany
on 2012-03-19 17:34:01
Ok, sounds reasonable.
I thought of something like this too :)
Will reopen if this affects anything else.
Reported by timurrrr
on 2012-03-19 17:47:19
WontFix
Adding Project:AddressSanitizer as part of GitHub migration.
Reported by ramosian.glider
on 2015-07-30 09:12:58
Originally reported on Google Code with ID 52
Reported by
timurrrr
on 2012-03-19 13:24:52