Re: perf: use after free in perf_remove_from_context

[Dave Jones]

On Wed, May 14, 2014 at 01:09:31PM -0400, Sasha Levin wrote:
 
 > > *** Error in `./trinity': double free or corruption (top): 0x000000000135af60 ***
 > > [main] Random reseed: 3671679404
 > > [main] Random reseed: 67838733
 > > *** Error in `./trinity': double free or corruption (top): 0x000000000135ef60 ***
 > > [main] Random reseed: 943044433
 > > *** Error in `./trinity': double free or corruption (top): 0x00000000012bb210 ***
 > > [main] Random reseed: 3627649145
 > > *** Error in `./trinity': double free or corruption (top): 0x000000000139dbe0 ***
 > > [main] Random reseed: 935036390
 > > [main] Random reseed: 3922349807
 > > [main] Random reseed: 835563887
 > > *** Error in `./trinity': double free or corruption (top): 0x0000000001323890 ***
 > > [main] Random reseed: 974945617
 > > *** Error in `./trinity': double free or corruption (top): 0x0000000001308190 ***
 > > [main] Random reseed: 2912296953
 > > [main] Random reseed: 681900861
 > > 
 > > Is this 'expected' behaviour?
 > 
 > I've talked about it with Dave a while ago, he said it happens because the threads that
 > do all the testing sometimes corrupt themselves which result in that error. It shouldn't
 > be harmful as the thread will just get killed and a fresh one will be started in it's
 > place.
 > 
 > Since then I set MALLOC_CHECK_=0 to avoid seeing that in my logs.
 
Pretty strong odds it's mremap causing those double frees.  You can
-xmremap that to see if it goes away.

	Dave

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/