On Fri, Apr 25, 2014 at 04:43:01PM +0200, Peter Zijlstra wrote: > On Fri, Apr 25, 2014 at 03:54:13PM +0200, Robert Richter wrote: > > I guess its this patch from Al you referring to: > > > > ea635c6 Fix racy use of anon_inode_getfd() in perf_event.c > > > > I think we do not introduce it back due to: > > > > a6fa941 perf_event: Switch to internal refcount, fix race with close() > > > > which removes the user of event_file. > > linux-2.6# git show a6fa941 > error: short SHA1 a6fa941 is ambiguous. > error: short SHA1 a6fa941 is ambiguous. > fatal: ambiguous argument 'a6fa941': unknown revision or path not in the > working tree. > > Which is I think why Linus makes us use 12 chars instead of the git > default of 8. > > Of course its also entirely useless of git to not list the full IDs for > those it did find to be ambiguous.
ok, so its a6fa941d94b4. But no, I don't think that helps, its still true that the moment you get a fd another thread can immediately close(). That would drop the last ref and free it, meanwhile perf_event_open() is happily poking at it. Now I think you could cure this by adding an extra ref before calling your perf_get_fd() and dropping that extra ref at the end, where we used to have fd_install(). -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
