public inbox for gcc-cvs@sourceware.org help / color / mirror / Atom feed
From: Richard Biener <rguenth@gcc.gnu.org> To: gcc-cvs@gcc.gnu.org Subject: [gcc r13-4229] tree-optimization/107803 - abnormal cleanup from the SSA propagator Date: Tue, 22 Nov 2022 11:05:33 +0000 (GMT) [thread overview] Message-ID: <20221122110533.4AF783858422@sourceware.org> (raw) https://gcc.gnu.org/g:e4faee8d02ec5d65bf418612f7181823eb08c078 commit r13-4229-ge4faee8d02ec5d65bf418612f7181823eb08c078 Author: Richard Biener <rguenther@suse.de> Date: Tue Nov 22 10:16:14 2022 +0100 tree-optimization/107803 - abnormal cleanup from the SSA propagator The SSA propagator is missing abnormal cleanup which shows in a sanity check in the uninit engine (and missed CFG verification). The following adds that. PR tree-optimization/107803 * tree-ssa-propagate.cc (substitute_and_fold_dom_walker): Add need_ab_cleanup member. (substitute_and_fold_dom_walker::before_dom_children): When a stmt can no longer transfer control flow abnormally set need_ab_cleanup. (substitute_and_fold_engine::substitute_and_fold): Cleanup abnormal control flow. * g++.dg/pr107803.C: New testcase. Diff: --- gcc/testsuite/g++.dg/pr107803.C | 19 +++++++++++++++++++ gcc/tree-ssa-propagate.cc | 20 ++++++++++++++++++-- 2 files changed, 37 insertions(+), 2 deletions(-) diff --git a/gcc/testsuite/g++.dg/pr107803.C b/gcc/testsuite/g++.dg/pr107803.C new file mode 100644 index 00000000000..f814e968b69 --- /dev/null +++ b/gcc/testsuite/g++.dg/pr107803.C @@ -0,0 +1,19 @@ +// { dg-do compile } +// { dg-options "-O -fno-tree-dominator-opts -fno-tree-fre -Wmaybe-uninitialized" } + +void printf(...); +void __sigsetjmp_cancel() __attribute__((__returns_twice__)); +int z, main_ret; +void func(void *) {} + +int +main() +{ + int x; + void (*__cancel_routine)(void *)(func); + __sigsetjmp_cancel(); + __cancel_routine(0); + if (main_ret) + x = z; + printf(x); +} diff --git a/gcc/tree-ssa-propagate.cc b/gcc/tree-ssa-propagate.cc index 9dc4bfd85bf..d8b0aed4564 100644 --- a/gcc/tree-ssa-propagate.cc +++ b/gcc/tree-ssa-propagate.cc @@ -671,12 +671,14 @@ public: stmts_to_remove.create (0); stmts_to_fixup.create (0); need_eh_cleanup = BITMAP_ALLOC (NULL); + need_ab_cleanup = BITMAP_ALLOC (NULL); } ~substitute_and_fold_dom_walker () { stmts_to_remove.release (); stmts_to_fixup.release (); BITMAP_FREE (need_eh_cleanup); + BITMAP_FREE (need_ab_cleanup); } edge before_dom_children (basic_block) final override; @@ -689,6 +691,7 @@ public: vec<gimple *> stmts_to_remove; vec<gimple *> stmts_to_fixup; bitmap need_eh_cleanup; + bitmap need_ab_cleanup; class substitute_and_fold_engine *substitute_and_fold_engine; @@ -838,8 +841,13 @@ substitute_and_fold_dom_walker::before_dom_children (basic_block bb) folded. */ did_replace = false; gimple *old_stmt = stmt; - bool was_noreturn = (is_gimple_call (stmt) - && gimple_call_noreturn_p (stmt)); + bool was_noreturn = false; + bool can_make_abnormal_goto = false; + if (is_gimple_call (stmt)) + { + was_noreturn = gimple_call_noreturn_p (stmt); + can_make_abnormal_goto = stmt_can_make_abnormal_goto (stmt); + } /* Replace real uses in the statement. */ did_replace |= substitute_and_fold_engine->replace_uses_in (stmt); @@ -905,6 +913,12 @@ substitute_and_fold_dom_walker::before_dom_children (basic_block bb) if (maybe_clean_or_replace_eh_stmt (old_stmt, stmt)) bitmap_set_bit (need_eh_cleanup, bb->index); + /* If we turned a call with possible abnormal control transfer + into one that doesn't, remove abnormal edges. */ + if (can_make_abnormal_goto + && !stmt_can_make_abnormal_goto (stmt)) + bitmap_set_bit (need_ab_cleanup, bb->index); + /* If we turned a not noreturn call into a noreturn one schedule it for fixup. */ if (!was_noreturn @@ -1012,6 +1026,8 @@ substitute_and_fold_engine::substitute_and_fold (basic_block block) if (!bitmap_empty_p (walker.need_eh_cleanup)) gimple_purge_all_dead_eh_edges (walker.need_eh_cleanup); + if (!bitmap_empty_p (walker.need_ab_cleanup)) + gimple_purge_all_dead_abnormal_call_edges (walker.need_ab_cleanup); /* Fixup stmts that became noreturn calls. This may require splitting blocks and thus isn't possible during the dominator walk. Do this
reply other threads:[~2022-11-22 11:05 UTC|newest] Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20221122110533.4AF783858422@sourceware.org \ --to=rguenth@gcc.gnu.org \ --cc=gcc-cvs@gcc.gnu.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
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).