Hi!
This patch tries to handle bitfield accesses as if it were writes/reads
from the corresponding DECL_BIT_FIELD_REPRESENTATIVE.
Richard, does it make sense this way (the code will take ADDR_EXPR
of the COMPONENT_REF with the DECL_BIT_FIELD_REPRESENTATIVE and use
the size of the representative if it decides to instrument it.
Fixes the first 4 asan_test.C failures. Ok for trunk?
2012-12-05 Jakub Jelinek <ja...@redhat.com>
* asan.c (instrument_derefs): Handle bitfield COMPONENT_REFs
accesses as reads/writes to their DECL_BIT_FIELD_REPRESENTATIVE.
--- gcc/asan.c.jj 2012-12-05 15:30:56.000000000 +0100
+++ gcc/asan.c 2012-12-05 17:00:56.957641944 +0100
@@ -792,9 +792,6 @@ instrument_derefs (gimple_stmt_iterator
|| (unsigned HOST_WIDE_INT) size_in_bytes - 1 >= 16)
return;
- /* For now just avoid instrumenting bit field acceses.
- Fixing it is doable, but expected to be messy. */
-
HOST_WIDE_INT bitsize, bitpos;
tree offset;
enum machine_mode mode;
@@ -803,7 +800,17 @@ instrument_derefs (gimple_stmt_iterator
&mode, &unsignedp, &volatilep, false);
if (bitpos % (size_in_bytes * BITS_PER_UNIT)
|| bitsize != size_in_bytes * BITS_PER_UNIT)
- return;
+ {
+ if (TREE_CODE (t) == COMPONENT_REF
+ && DECL_BIT_FIELD_REPRESENTATIVE (TREE_OPERAND (t, 1)) != NULL_TREE)
+ {
+ tree repr = DECL_BIT_FIELD_REPRESENTATIVE (TREE_OPERAND (t, 1));
+ instrument_derefs (iter, build3 (COMPONENT_REF, TREE_TYPE (repr),
+ TREE_OPERAND (t, 0), repr,
+ NULL_TREE), location, is_store);
+ }
+ return;
+ }
base = build_fold_addr_expr (t);
build_check_stmt (location, base, iter, /*before_p=*/true,
Jakub
