https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104103
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target Milestone|--- |12.0
Assignee|unassigned at gcc dot gnu.org |jakub at gcc dot gnu.org
Component|testsuite |middle-end
Ever confirmed|0 |1
CC| |jakub at gcc dot gnu.org
Status|UNCONFIRMED |ASSIGNED
Last reconfirmed| |2022-01-18
--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I've just debugged this, the following patch fixes that.
.ASAN_MARK, while it takes address of variables, often after they go out of
scope, doesn't access those variables in any way, it poisons or unpoisons the
corresponding shadow memory.
--- gcc/gimple-ssa-warn-access.cc.jj 2022-01-16 20:55:46.783932110 +0100
+++ gcc/gimple-ssa-warn-access.cc 2022-01-18 20:56:13.697780325 +0100
@@ -4232,6 +4232,11 @@ pass_waccess::check_call (gcall *stmt)
if (gimple_call_builtin_p (stmt, BUILT_IN_NORMAL))
check_builtin (stmt);
+ /* .ASAN_MARK doesn't access any vars, only modifies shadow memory. */
+ if (gimple_call_internal_p (stmt)
+ && gimple_call_internal_fn (stmt) == IFN_ASAN_MARK)
+ return;
+
if (!m_early_checks_p)
if (tree callee = gimple_call_fndecl (stmt))
{
