[Bug tree-optimization/66419] [6 regression] FAIL: gcc.target/aarch64/aapcs64/va_arg-6.c execution, -O3 -g

2016-12-01 Thread dje at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66419

--- Comment #6 from David Edelsohn  ---
Author: dje
Date: Thu Dec  1 18:58:47 2016
New Revision: 243126

URL: https://gcc.gnu.org/viewcvs?rev=243126=gcc=rev
Log:
PR debug/66419
PR c++/78235
* dbxout.c (dbxout_type_fields): Skip TEMPLATE_DECLs.

Modified:
trunk/gcc/ChangeLog
trunk/gcc/dbxout.c

[Bug tree-optimization/66419] [6 regression] FAIL: gcc.target/aarch64/aapcs64/va_arg-6.c execution, -O3 -g

2015-06-09 Thread rguenth at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66419

Richard Biener rguenth at gcc dot gnu.org changed:

   What|Removed |Added

 Status|ASSIGNED|RESOLVED
 Resolution|--- |FIXED

--- Comment #5 from Richard Biener rguenth at gcc dot gnu.org ---
Fixed.

[Bug tree-optimization/66419] [6 regression] FAIL: gcc.target/aarch64/aapcs64/va_arg-6.c execution, -O3 -g

2015-06-09 Thread rguenth at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66419

--- Comment #4 from Richard Biener rguenth at gcc dot gnu.org ---
Author: rguenth
Date: Tue Jun  9 08:22:20 2015
New Revision: 224271

URL: https://gcc.gnu.org/viewcvs?rev=224271root=gccview=rev
Log:
2015-06-09  Richard Biener  rguent...@suse.de

PR tree-optimization/66419
* tree-vect-slp.c (vect_supported_load_permutation_p): Properly
consider GROUP_GAP when detecting a perfect subchain.

* gcc.dg/vect/bb-slp-37.c: New testcase.

Added:
trunk/gcc/testsuite/gcc.dg/vect/bb-slp-37.c
Modified:
trunk/gcc/ChangeLog
trunk/gcc/testsuite/ChangeLog
trunk/gcc/tree-vect-slp.c

[Bug tree-optimization/66419] [6 regression] FAIL: gcc.target/aarch64/aapcs64/va_arg-6.c execution, -O3 -g

2015-06-08 Thread rguenth at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66419

Richard Biener rguenth at gcc dot gnu.org changed:

   What|Removed |Added

 Status|UNCONFIRMED |ASSIGNED
   Last reconfirmed||2015-06-08
   Assignee|unassigned at gcc dot gnu.org  |rguenth at gcc dot 
gnu.org
 Ever confirmed|0   |1

--- Comment #1 from Richard Biener rguenth at gcc dot gnu.org ---
I suppose we vectorize the init_data function.  I will have a look.

[Bug tree-optimization/66419] [6 regression] FAIL: gcc.target/aarch64/aapcs64/va_arg-6.c execution, -O3 -g

2015-06-08 Thread rguenth at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66419

--- Comment #3 from Richard Biener rguenth at gcc dot gnu.org ---
Ah, because we detect it as subchain which fails to consider gaps.

[Bug tree-optimization/66419] [6 regression] FAIL: gcc.target/aarch64/aapcs64/va_arg-6.c execution, -O3 -g

2015-06-08 Thread rguenth at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66419

--- Comment #2 from Richard Biener rguenth at gcc dot gnu.org ---
The difference is that we vectorize parts of stdarg_func:

L3:
  _57 = ap.__vr_top;
  _58 = (sizetype) _39;
  _59 = _57 + _58;
  _60 = *_59;
  MEM[(float *)ha.20] = _60;
  _64 = _58 + 16;
  _65 = _57 + _64;
  _66 = MEM[(float *)_65];
  MEM[(float *)ha.20 + 4B] = _66;
  _70 = _58 + 32;
  _71 = _57 + _70;
  _72 = MEM[(float *)_71];
  MEM[(float *)ha.20 + 8B] = _72;
  _76 = _58 + 48;
  _77 = _57 + _76;
  _78 = MEM[(float *)_77];
  MEM[(float *)ha.20 + 12B] = _78;

to

L3:
  _57 = ap.__vr_top;
  _58 = (sizetype) _39;
  _59 = _57 + _58;
  _60 = *_59;
  _64 = _58 + 16;
  _65 = _57 + _64;
  _66 = MEM[(float *)_65];
  _70 = _58 + 32;
  _71 = _57 + _70;
  _72 = MEM[(float *)_71];
  _76 = _58 + 48;
  _77 = _57 + _76;
  vectp.40_138 = _59;
  vect__60.41_125 = MEM[(float *)vectp.40_138];
  _78 = MEM[(float *)_77];
  vectp.43_107 = ha.20;
  MEM[(float *)vectp.43_107] = vect__60.41_125;

which looks bogus.  The load permutation

/space/rguenther/src/svn/trunk2/gcc/testsuite/gcc.target/aarch64/aapcs64/abitest.h:98:6:
note: Load permutation 0 4 8 12

due to the gaps isn't reflected in the transform.  And of course cost
considerations should prohibit the transform (but well).

C testcase that fails on x86_64 as well:

extern void abort (void);

int a[16];
int b[4];

void __attribute__((noinline))
foo (void)
{
  b[0] = a[0];
  b[1] = a[4];
  b[2] = a[8];
  b[3] = a[12];
}

int main()
{
  int i;
  for (i = 0; i  16; ++i)
{
  a[i] = i;
  __asm__ volatile ();
}
  foo ();
  if (b[0] != 0 || b[1] != 4 || b[2] != 8 || b[3] != 12)
abort ();
  return 0;
}

[Bug tree-optimization/66419] [6 regression] FAIL: gcc.target/aarch64/aapcs64/va_arg-6.c execution, -O3 -g

2015-06-04 Thread sch...@linux-m68k.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66419

Andreas Schwab sch...@linux-m68k.org changed:

   What|Removed |Added

   Target Milestone|--- |6.0