public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug middle-end/106870] New: ctrl_altering flag is not set correctly
@ 2022-09-07 11:28 rguenth at gcc dot gnu.org
  2022-09-07 13:32 ` [Bug middle-end/106870] " rguenth at gcc dot gnu.org
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: rguenth at gcc dot gnu.org @ 2022-09-07 11:28 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106870

            Bug ID: 106870
           Summary: ctrl_altering flag is not set correctly
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rguenth at gcc dot gnu.org
  Target Milestone: ---

When trying to add the following verification I run into various issues with
the gimple_ctrl_altering_p flag which stmt_ends_bb_p which sets found_ctrl_stmt
ultimatively relies on for the noreturn case.

diff --git a/gcc/tree-cfg.cc b/gcc/tree-cfg.cc
index bbe08357d6e..a939a11d8bb 100644
--- a/gcc/tree-cfg.cc
+++ b/gcc/tree-cfg.cc
@@ -5708,6 +5709,14 @@ gimple_verify_flow_info (void)
            seen_nondebug_stmt = true;
        }

+      if (!found_ctrl_stmt && EDGE_COUNT (bb->succs) != 1)
+       {
+         error ("basic block %d has no control altering stmt but %d "
+                "outgoing edges",
+                bb->index, EDGE_COUNT (bb->succs));
+         err = 1;
+       }
+
       gsi = gsi_last_nondebug_bb (bb);
       if (gsi_end_p (gsi))
        continue;

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug middle-end/106870] ctrl_altering flag is not set correctly
  2022-09-07 11:28 [Bug middle-end/106870] New: ctrl_altering flag is not set correctly rguenth at gcc dot gnu.org
@ 2022-09-07 13:32 ` rguenth at gcc dot gnu.org
  2022-09-08 14:06 ` cvs-commit at gcc dot gnu.org
  2022-09-08 14:08 ` rguenth at gcc dot gnu.org
  2 siblings, 0 replies; 4+ messages in thread
From: rguenth at gcc dot gnu.org @ 2022-09-07 13:32 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106870

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
gcc.dg/torture/pr57036-2.c shows how inlining and non-local gotos come into
play here.

Note the flag was specifically invented to not make CFG verification ICE
when a function in the middle of a block becomes noreturn, for example
via IPA analysis.  It wasn't necessarily required to have the flag set
on stmts ending a BB.

Still not having the flag set prevents nothing to put stmts after such
call and we'd not complain.

gcc.dg/torture/pr83055.c shows how the checking is confused as called from
gimple_flow_call_edges_add after adding loads of fake edges.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug middle-end/106870] ctrl_altering flag is not set correctly
  2022-09-07 11:28 [Bug middle-end/106870] New: ctrl_altering flag is not set correctly rguenth at gcc dot gnu.org
  2022-09-07 13:32 ` [Bug middle-end/106870] " rguenth at gcc dot gnu.org
@ 2022-09-08 14:06 ` cvs-commit at gcc dot gnu.org
  2022-09-08 14:08 ` rguenth at gcc dot gnu.org
  2 siblings, 0 replies; 4+ messages in thread
From: cvs-commit at gcc dot gnu.org @ 2022-09-08 14:06 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106870

--- Comment #2 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:78ef801b7263606d27a6a752a40cbeecc584fb3d

commit r13-2541-g78ef801b7263606d27a6a752a40cbeecc584fb3d
Author: Richard Biener <rguenther@suse.de>
Date:   Wed Sep 7 12:01:34 2022 +0200

    Fix some gimple_ctrl_altering_p mistakes

    CFG cleanup resets the control altering flag for noreturn functions
    when they are ECF_LEAF (like __builtin_unreachable ()).  The
    .ABNORMAL_DISPATCHER call built during CFG construction is not
    marked as control altering.  Several passes inserting traps or
    unreachables fail to set the flag.  And more.

            PR middle-end/106870
            * gimple-harden-conditionals.cc (insert_check_and_trap):
            Set the control-altering flag on the built IFN_TRAP.
            * gimple.cc (gimple_build_builtin_unreachable): Likewise.
            * tree-cfg.cc (handle_abnormal_edges): Set the control-altering
            flag on the .ABNORMAL_DISPATCHER call.
            * tree-cfgcleanup.cc (cleanup_call_ctrl_altering_flag): Avoid
            resetting the control altering flag for ECF_NORETURN calls.
            (cleanup_control_flow_bb): Set the control altering flag on
            discovered noreturn calls.
            * symtab-thunks.cc (expand_thunk): Set the control altering
            flag for the noreturn tailcall case.
            * tree-eh.cc (lower_resx): Likewisw for trap and unwind_resume
            calls.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug middle-end/106870] ctrl_altering flag is not set correctly
  2022-09-07 11:28 [Bug middle-end/106870] New: ctrl_altering flag is not set correctly rguenth at gcc dot gnu.org
  2022-09-07 13:32 ` [Bug middle-end/106870] " rguenth at gcc dot gnu.org
  2022-09-08 14:06 ` cvs-commit at gcc dot gnu.org
@ 2022-09-08 14:08 ` rguenth at gcc dot gnu.org
  2 siblings, 0 replies; 4+ messages in thread
From: rguenth at gcc dot gnu.org @ 2022-09-08 14:08 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106870

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
This fixes some cases but leaves others not fixed.

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2022-09-08 14:08 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-09-07 11:28 [Bug middle-end/106870] New: ctrl_altering flag is not set correctly rguenth at gcc dot gnu.org
2022-09-07 13:32 ` [Bug middle-end/106870] " rguenth at gcc dot gnu.org
2022-09-08 14:06 ` cvs-commit at gcc dot gnu.org
2022-09-08 14:08 ` rguenth at gcc dot gnu.org

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).