public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug middle-end/112087] New: gimple front end: Simple testcase with array fails to round trip
@ 2023-10-25 14:08 acoplan at gcc dot gnu.org
  2023-10-27 12:57 ` [Bug middle-end/112087] " rguenth at gcc dot gnu.org
  0 siblings, 1 reply; 2+ messages in thread
From: acoplan at gcc dot gnu.org @ 2023-10-25 14:08 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112087

            Bug ID: 112087
           Summary: gimple front end: Simple testcase with array fails to
                    round trip
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: acoplan at gcc dot gnu.org
  Target Milestone: ---

Reproducer:

$ cat t.c
long f()
{
  int a[4];
  return (long)a;
}
$ gcc/xgcc -B gcc -c t.c -S -o /dev/null -O2 -fdump-tree-ssa-gimple=gimple.c
$ cat gimple.c
long int __GIMPLE (ssa)
f ()
{
  int a[4];
  long int _1;

  __BB(2):
  _1 = (long int) _Literal (int[4] *)&a;
  a ={v} _Literal (int[4]) {CLOBBER(eol)};
  return _1;

}


$ gcc/xgcc -B gcc -c gimple.c -S -o /dev/null -O2 -fgimple
gimple.c: In function ‘f’:
gimple.c:8:35: error: expected ‘)’ before ‘*’ token
    8 |   _1 = (long int) _Literal (int[4] *)&a;
      |                                   ^~
      |                                   )
gimple.c:9:6: error: expected expression before ‘{’ token
    9 |   a ={v} _Literal (int[4]) {CLOBBER(eol)};
      |      ^
gimple.c:12:1: error: invalid types in nop conversion
   12 | }
      | ^
long int
int[4]
_1 = (long int) &a;

^ permalink raw reply	[flat|nested] 2+ messages in thread

* [Bug middle-end/112087] gimple front end: Simple testcase with array fails to round trip
  2023-10-25 14:08 [Bug middle-end/112087] New: gimple front end: Simple testcase with array fails to round trip acoplan at gcc dot gnu.org
@ 2023-10-27 12:57 ` rguenth at gcc dot gnu.org
  0 siblings, 0 replies; 2+ messages in thread
From: rguenth at gcc dot gnu.org @ 2023-10-27 12:57 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112087

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2023-10-27
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
"Round-tripping" in the sense of

gcc -fdump-tree-X-gimple=- t.c -S | gcc -x c -fgimple -

was never the intention.

Yes, parsing of CLOBBERs isn't implemented, nor is parsing of
"gimple_has_volatile_ops", but since that's redundant info recomputed
at update_stmt time we can just scrap it at dump time (but then it's
a bit odd to not see it with -gimple ..).

It would be nice to have bugreports more specific that "cannot round-trip A or
B".

Two bugs for this.

Parsing of ={v}

Parsing of {CLOBBER} (and variants)

parsing of ={v} is a bit difficult since it would need to be tentative
(there's always a reason why seemingly "easy" things are not yet done).
Finding a better syntax for dumping the flags on a stmt might help.


Workaround for both: use -fstack-reuse=none

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2023-10-27 12:57 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-10-25 14:08 [Bug middle-end/112087] New: gimple front end: Simple testcase with array fails to round trip acoplan at gcc dot gnu.org
2023-10-27 12:57 ` [Bug middle-end/112087] " 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).