On 11/14/2017 02:30 PM, Martin Sebor wrote: > On 11/14/2017 02:10 PM, Jeff Law wrote: >> On 11/14/2017 02:04 PM, Jakub Jelinek wrote: >>> Hi! >>> >>> strlen_to_stridx.get (rhs1) returns an address into the hash_map, and >>> strlen_to_stridx.put (lhs, *ps); (in order to be efficient) doesn't >>> make a >>> copy of the argument just in case, first inserts the slot into it which >>> may cause reallocation, and only afterwards runs the copy ctor to assign >>> the value into the new slot. So, passing it a reference to something >>> in the hash_map is wrong. Fixed thusly, bootstrapped/regtested on >>> x86_64-linux and i686-linux, ok for trunk? >>> >>> 2017-11-14 Jakub Jelinek <ja...@redhat.com> >>> >>> PR tree-optimization/82977 >>> * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a >>> copy >>> constructed temporary to strlen_to_stridx.put. >> I've been seeing a couple new tests flip between pass and fail recently. >> I wonder if this is the ultimate cause. > > I've been noticing it for quite a while, even before the commit, > so I suspect something else is going on in addition to this bug. I'm referring specifically to the Wstringop-truncation tests. They're ping-ponging between PASS/FAIL here with alarming regularity and no sense of rhyme or reason.
If I had to guess I'd guess uninit memory, dangling pointer or the like, which is precisely the kind of problem the patch is meant to fix. jeff
