[Bug lto/64684] New: wrong code by LTO on x86_64-linux-gnu

[Bug lto/64684] New: wrong code by LTO on x86_64-linux-gnu

[su at cs dot ucdavis.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64684

            Bug ID: 64684
           Summary: wrong code by LTO on x86_64-linux-gnu
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: lto
          Assignee: unassigned at gcc dot gnu.org
          Reporter: su at cs dot ucdavis.edu

The current gcc trunk miscompiles the following code when using LTO on
x86_64-linux-gnu in both 32-bit and 64-bit modes. 

This is a regression from 4.9.x. 

$ gcc-trunk -v
Using built-in specs.
COLLECT_GCC=gcc-trunk
COLLECT_LTO_WRAPPER=/usr/local/gcc-trunk/libexec/gcc/x86_64-unknown-linux-gnu/5.0.0/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: ../gcc-trunk/configure --prefix=/usr/local/gcc-trunk
--enable-languages=c,c++ --disable-werror --enable-multilib
Thread model: posix
gcc version 5.0.0 20150119 (experimental) [trunk revision 219832] (GCC) 

$ 
$ gcc-trunk -O1 -c fn1.c
$ gcc-trunk -Os -c fn2.c
$ gcc-trunk -O0 -c main.c
$ gcc-trunk -O0 fn1.o fn2.o main.o
$ ./a.out
$ 
$ 
$ gcc-4.9 -flto -O1 -c fn1.c
$ gcc-4.9 -flto -Os -c fn2.c
$ gcc-4.9 -flto -O0 -c main.c
$ gcc-4.9 -flto fn1.o fn2.o main.o
$ ./a.out
$ 
$ gcc-trunk -flto -O1 -c fn1.c
$ gcc-trunk -flto -Os -c fn2.c
$ gcc-trunk -flto -O0 -c main.c
$ gcc-trunk -flto fn1.o fn2.o main.o
$ ./a.out
Aborted (core dumped)
$ 
$ cat fn1.c
extern void fn2 (void);
extern int a;

void
fn1 ()
{
  a = -1;
  fn2 ();
  a &= 1;
}
$ cat fn2.c
extern int a;

void
fn2 (void)
{
  a = 0;
}
$ cat main.c
extern void fn1 (void);

int a;

int
main ()
{
  fn1 ();

  if (a != 0) 
    __builtin_abort (); 

  return 0;
}
$

[Bug lto/64684] [5 Regression] wrong code by LTO on x86_64-linux-gnu

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64684

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |lto, wrong-code
                 CC|                            |hubicka at gcc dot gnu.org,
                   |                            |rguenth at gcc dot gnu.org
   Target Milestone|---                         |5.0
            Summary|wrong code by LTO on        |[5 Regression] wrong code
                   |x86_64-linux-gnu            |by LTO on x86_64-linux-gnu

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
I bet this is just the start of interesting bugs with optimization attributes
on all functions...

[Bug lto/64684] [5 Regression] wrong code by LTO on x86_64-linux-gnu

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64684

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-01-20
     Ever confirmed|0                           |1

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Setting value number of .MEM_7 to .MEM_7 (changed)
Value numbering .MEM_8 stmt = fn2 ();
Setting value number of .MEM_8 to .MEM_8 (changed)
Value numbering _4 stmt = _4 = a;
Setting value number of _4 to -1 (changed)
Value numbering _5 stmt = _5 = _4 & 1;
Match-and-simplified _4 & 1 to 1
RHS _4 & 1 simplified to 1 has constants 1
Setting value number of _5 to 1 (changed)

so fn2 () is said to not clobber 'a'.  Must be IPA references fault:

all module statics: a
Function name:fn2/3:
  static not read: all module statics
  static not written: all module statics

Function name:fn1/0:
  static not read: all module statics
  static not written: all module statics

that's wrong for both fn1 and fn2.  WPA does

Promotable global:a (uid=3959)

Function name:fn2/3:
  locals read:
  locals written:
Function name:fn1/0:
  locals read:
  locals written:

And compile-time doesn't have .static-var dumps at all?!  Ah, WPA stage
is supposed to generate summaries?  But doesn't input nodes?

[Bug lto/64684] [5 Regression] wrong code by LTO on x86_64-linux-gnu

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64684

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ah - you do

           if (!is_proper_for_analysis (var))
              continue;
            bitmap_set_bit (ignore_module_statics, DECL_UID (var));

but is_proper_for_analysis adds to all_module_statics (that's quite a broken
thing for this kind of function...).

I have a patch.

[Bug lto/64684] [5 Regression] wrong code by LTO on x86_64-linux-gnu

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64684

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
*** Bug 64685 has been marked as a duplicate of this bug. ***

[Bug lto/64684] [5 Regression] wrong code by LTO on x86_64-linux-gnu

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64684

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Tue Jan 20 13:26:19 2015
New Revision: 219892

URL: https://gcc.gnu.org/viewcvs?rev=219892&root=gcc&view=rev
Log:
2015-01-20  Richard Biener  <rguenther@suse.de>

    PR ipa/64684
    * ipa-reference.c (add_static_var): Inline ...
    (analyze_function): ... here after splitting out from ...
    (is_proper_for_analysis): ... this.

    * gcc.dg/lto/pr64684_0.c: New testcase.
    * gcc.dg/lto/pr64684_1.c: Likewise.
    * gcc.dg/lto/pr64684_2.c: Likewise.
    * gcc.dg/lto/pr64685_0.c: Likewise.
    * gcc.dg/lto/pr64685_1.c: Likewise.

Added:
    trunk/gcc/testsuite/gcc.dg/lto/pr64684_0.c
    trunk/gcc/testsuite/gcc.dg/lto/pr64684_1.c
    trunk/gcc/testsuite/gcc.dg/lto/pr64684_2.c
    trunk/gcc/testsuite/gcc.dg/lto/pr64685_0.c
    trunk/gcc/testsuite/gcc.dg/lto/pr64685_1.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/ipa-reference.c
    trunk/gcc/testsuite/ChangeLog

[Bug lto/64684] [5 Regression] wrong code by LTO on x86_64-linux-gnu

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64684

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|---                         |FIXED

--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.