public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r12-9415] tree-optimization/109434 - bogus DSE of throwing call LHS
@ 2023-04-17 10:45 Richard Biener
0 siblings, 0 replies; only message in thread
From: Richard Biener @ 2023-04-17 10:45 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:3d2210dafd872e8470e2a6ae5eea74d2669bc055
commit r12-9415-g3d2210dafd872e8470e2a6ae5eea74d2669bc055
Author: Richard Biener <rguenther@suse.de>
Date: Tue Apr 11 15:06:59 2023 +0200
tree-optimization/109434 - bogus DSE of throwing call LHS
The byte tracking of call LHS didn't properly handle possibly
throwing calls correctly which cases bogus DSE and in turn, for the
testcase a bogus uninit diagnostic and (unreliable) wrong-code.
PR tree-optimization/109434
* tree-ssa-dse.cc (initialize_ao_ref_for_dse): Properly
handle possibly throwing calls when processing the LHS
and may-defs are not OK. Add mode to initialize a may-def.
(dse_optimize_stmt): Query may-defs.
* g++.dg/opt/pr109434.C: New testcase.
Diff:
---
gcc/testsuite/g++.dg/opt/pr109434.C | 28 ++++++++++++++++++++++++++++
gcc/tree-ssa-dse.cc | 15 ++++++++++-----
2 files changed, 38 insertions(+), 5 deletions(-)
diff --git a/gcc/testsuite/g++.dg/opt/pr109434.C b/gcc/testsuite/g++.dg/opt/pr109434.C
new file mode 100644
index 00000000000..cffa327fd9b
--- /dev/null
+++ b/gcc/testsuite/g++.dg/opt/pr109434.C
@@ -0,0 +1,28 @@
+// { dg-do compile }
+// { dg-require-effective-target c++17 }
+// { dg-options "-O2 -Wall" }
+
+#include <optional>
+#include <stdexcept>
+
+std::optional<int> foo()
+{
+ volatile int x = 1;
+ if (x)
+ throw std::runtime_error("haha");
+ return 42;
+}
+
+int main()
+{
+ std::optional<int> optInt;
+ try {
+ // We falsely DSEd the LHS of the call even though foo throws
+ // which results in an uninitialized diagnostic
+ optInt = foo();
+ } catch (...) {
+ return optInt.has_value();
+ }
+ std::optional<double> optDbl{optInt};
+ return optDbl ? optDbl.value () : 2.0;
+}
diff --git a/gcc/tree-ssa-dse.cc b/gcc/tree-ssa-dse.cc
index e647f0e7368..ba448c999dd 100644
--- a/gcc/tree-ssa-dse.cc
+++ b/gcc/tree-ssa-dse.cc
@@ -93,7 +93,9 @@ static bitmap need_eh_cleanup;
static bitmap need_ab_cleanup;
/* STMT is a statement that may write into memory. Analyze it and
- initialize WRITE to describe how STMT affects memory.
+ initialize WRITE to describe how STMT affects memory. When
+ MAY_DEF_OK is true then the function initializes WRITE to what
+ the stmt may define.
Return TRUE if the statement was analyzed, FALSE otherwise.
@@ -101,7 +103,7 @@ static bitmap need_ab_cleanup;
can be achieved by analyzing more statements. */
static bool
-initialize_ao_ref_for_dse (gimple *stmt, ao_ref *write)
+initialize_ao_ref_for_dse (gimple *stmt, ao_ref *write, bool may_def_ok = false)
{
/* It's advantageous to handle certain mem* functions. */
if (gimple_call_builtin_p (stmt, BUILT_IN_NORMAL))
@@ -148,7 +150,8 @@ initialize_ao_ref_for_dse (gimple *stmt, ao_ref *write)
}
else if (tree lhs = gimple_get_lhs (stmt))
{
- if (TREE_CODE (lhs) != SSA_NAME)
+ if (TREE_CODE (lhs) != SSA_NAME
+ && (may_def_ok || !stmt_could_throw_p (cfun, stmt)))
{
ao_ref_init (write, lhs);
return true;
@@ -1293,8 +1296,10 @@ dse_optimize_stmt (function *fun, gimple_stmt_iterator *gsi, sbitmap live_bytes)
ao_ref ref;
/* If this is not a store we can still remove dead call using
- modref summary. */
- if (!initialize_ao_ref_for_dse (stmt, &ref))
+ modref summary. Note we specifically allow ref to be initialized
+ to a conservative may-def since we are looking for followup stores
+ to kill all of it. */
+ if (!initialize_ao_ref_for_dse (stmt, &ref, true))
{
dse_optimize_call (gsi, live_bytes);
return;
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2023-04-17 10:45 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-04-17 10:45 [gcc r12-9415] tree-optimization/109434 - bogus DSE of throwing call LHS Richard Biener
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).