http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59600
Yury Gribov <y.gribov at samsung dot com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |y.gribov at samsung dot com --- Comment #1 from Yury Gribov <y.gribov at samsung dot com> --- Created attachment 31515 --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31515&action=edit Draft patch Fails for me as well. Given that Asan runs long after inliner this behavior is expected. Perhaps it makes sense to prohibit inline for unsanitized functions? We'll loose some performance but no_sanitize_address semantics would be more transparent for users. Here's a crude patch which seems to fix repro and also show no regressions for `make check-c RUNTESTFLAGS=asan.exp'.