Bug 112721 - [14 Regression] Wrong code at -O1/s on x86_64-linux-gnu since r14-5831-gaae723d360c
Summary: [14 Regression] Wrong code at -O1/s on x86_64-linux-gnu since r14-5831-gaae72...
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: tree-optimization (show other bugs)
Version: 14.0
: P3 normal
Target Milestone: 14.0
Assignee: Martin Jambor
URL:
Keywords: wrong-code
Depends on:
Blocks:
 
Reported: 2023-11-27 08:04 UTC by Shaohua Li
Modified: 2023-11-29 15:27 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:
Known to work:
Known to fail:
Last reconfirmed: 2023-11-27 00:00:00


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Shaohua Li 2023-11-27 08:04:21 UTC
gcc at -O1/s produces the wrong code.

Bisected to r14-5831-gaae723d360c

Compiler explorer: https://godbolt.org/z/671M9YPPj

$ cat a.c
int printf(const char *, ...);
struct a {
  int b;
};
int c, e;
long d;
unsigned *f(unsigned *g) {
  for (; c;)
    e = d;
  return g;
}
int main() {
  int *h;
  struct a i = {8};
  int *j = &i.b;
  h = f(j);
  *h = 0;
  printf("%d\n", i.b);
}
$
$ gcc -O0 a.c && ./a.out
0
$ gcc -O1 a.c && ./a.out
8
$
Comment 1 Martin Jambor 2023-11-27 08:16:48 UTC
Most likely the same reason as PR 112711.  Mine.
Comment 2 Martin Jambor 2023-11-27 18:17:16 UTC
I have proposed a fix on the mailing list:

https://gcc.gnu.org/pipermail/gcc-patches/2023-November/638318.html
Comment 3 GCC Commits 2023-11-29 15:25:56 UTC
The master branch has been updated by Martin Jambor <jamborm@gcc.gnu.org>:

https://gcc.gnu.org/g:302461ad9a04d82fee904bddac69811d13d5bb6a

commit r14-5971-g302461ad9a04d82fee904bddac69811d13d5bb6a
Author: Martin Jambor <mjambor@suse.cz>
Date:   Wed Nov 29 16:24:33 2023 +0100

    tree-sra: Avoid returns of references to SRA candidates
    
    The enhancement to address PR 109849 contained an importsnt thinko,
    and that any reference that is passed to a function and does not
    escape, must also not happen to be aliased by the return value of the
    function.  This has quickly transpired as bugs PR 112711 and PR
    112721.
    
    Just as IPA-modref does a good enough job to allow us to rely on the
    escaped set of variables, it sems to be doing well also on updating
    EAF_NOT_RETURNED_DIRECTLY call argument flag which happens to address
    exactly the situation we need to avoid.  Of course, if a call
    statement ignores any returned value, we also do not need to check the
    flag.
    
    Hopefully this does not pessimize things too much, I have verified
    that the PR 109849 testcae remains quick and so should also the
    benchmark it is derived from.
    
    gcc/ChangeLog:
    
    2023-11-27  Martin Jambor  <mjambor@suse.cz>
    
            PR tree-optimization/112711
            PR tree-optimization/112721
            * tree-sra.cc (build_access_from_call_arg): New parameter
            CAN_BE_RETURNED, disqualify any candidate passed by reference if it is
            true.  Adjust leading comment.
            (scan_function): Pass appropriate value to CAN_BE_RETURNED of
            build_access_from_call_arg.
    
    gcc/testsuite/ChangeLog:
    
    2023-11-29  Martin Jambor  <mjambor@suse.cz>
    
            PR tree-optimization/112711
            PR tree-optimization/112721
            * g++.dg/tree-ssa/pr112711.C: New test.
            * gcc.dg/tree-ssa/pr112721.c: Likewise.
Comment 4 Martin Jambor 2023-11-29 15:27:01 UTC
Fixed