[Bug tree-optimization/34534] New: Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[Bug tree-optimization/34534] New: Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[hjl at lucon dot org]

On Linux/ia64, I saw

FAIL: gcc.dg/struct/wo_prof_global_var.c execution test
FAIL: gcc.dg/struct/wo_prof_local_var.c execution test
FAIL: gcc.dg/struct/wo_prof_mult_field_peeling.c execution test
FAIL: gcc.dg/struct/wo_prof_two_strs.c execution test

They are run-time failures.


-- 
           Summary: Multiple gcc.dg/struct/wo_prof_xxxx execution failures
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: hjl at lucon dot org
GCC target triplet: ia64-unknown-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[olga at gcc dot gnu dot org]

------- Comment #1 from olga at gcc dot gnu dot org  2007-12-20 09:32 -------
Would you please provide me with the following:

-dump files for all failures (as you did for PR34472)
-configuration options you use
-at least initial debugging (where it fails)

Thank you,
Olga


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[hjl at lucon dot org]

------- Comment #2 from hjl at lucon dot org  2007-12-20 17:45 -------
Created an attachment (id=14802)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14802&action=view)
Dump files

Starting program:
/export/build/gnu/gcc/build-ia64-linux/gcc/testsuite/gcc/a.out 

Program received signal SIGSEGV, Segmentation fault.
0x4000000000000751 in main ()
    at
/net/gnu-13/export/gnu/src/gcc/gcc/gcc/testsuite/gcc.dg/struct/wo_prof_global_var.c:22
22          p[i].a = p[i].b + 1;
(gdb) p/x $pc
$4 = 0x4000000000000751
(gdb) disass  0x4000000000000740 0x4000000000000752
Dump of assembler code from 0x4000000000000740 to 0x4000000000000752:
0x4000000000000740 <main+192>:  [MII]       shladd r15=r16,2,r0
0x4000000000000741 <main+193>:              adds r16=1,r16;;
0x4000000000000742 <main+194>:              add r14=r33,r15
0x4000000000000750 <main+208>:  [MMI]       add r15=r8,r15;;
0x4000000000000751 <main+209>:              ldfs f6=[r14]
End of assembler dump.
(gdb) p/x $r14
$5 = 0x5f60
(gdb) 

"ldfs f6=[r14]" loads memory at r14 into FP register f6. It looks that
-fipa-struct-reorg miscompiles the code.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[hjl at lucon dot org]

------- Comment #3 from hjl at lucon dot org  2007-12-21 00:25 -------
I configured gcc with

                --enable-clocale=gnu --with-system-zlib --with-demangler-in-ld
\
                --enable-shared \
                --enable-threads=posix \
                --enable-haifa \
                --enable-checking=assert \
                --prefix=/usr/gcc-4.3 \
                --with-local-prefix=/usr/local


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[olga at gcc dot gnu dot org]

------- Comment #4 from olga at gcc dot gnu dot org  2007-12-30 11:52 -------
(In reply to comment #2)
> Created an attachment (id=14802)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14802&action=view) [edit]
> Dump files

I looked a bit at the dump files you generated for wo_prof_global_var.c test.
It seems that in the first loop we convert (field 'b' + 1), that is float, to
int and store this value into field 'a', i.e.

 D.1845_14 = D.1844_13->b;
 D.1846_15 = D.1845_14 + 1.0e+0;
 D.1847_16 = (int) D.1846_15;
 D.1844_13->a ={v} D.1847_16;

while in the second we cast field 'a' to the float and compare it
to the value of (field 'b' + 1), i.e.

  D.1849_24 = (float) D.1848_23;
  ...
  D.1845_29 = D.1844_28->b;
  D.1846_30 = D.1845_29 + 1.0e+0;
  if (D.1849_24 != D.1846_30)

Is the last 'if' always should be true? I mean on any system...

So I wonder whether this test works correctly even without -fipa-struct-reorg.
Would you please try to compile it with all other options (-O3 -fdump-ipa-all
-fwhole-program -combine -fipa-type-escape) and run it on your machine?

It would be also interesting to check this test when the type of field 'b' is
int instead of float.

Thank you,
Olga


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[olga at gcc dot gnu dot org]

------- Comment #5 from olga at gcc dot gnu dot org  2007-12-30 12:07 -------
(In reply to comment #4)
> Is the last 'if' always should be true? I mean on any system...

true->false, sorry.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[dominiq at lps dot ens dot fr]

------- Comment #6 from dominiq at lps dot ens dot fr  2007-12-30 12:34 -------
I also see the problem on i686-apple-darwin9, gcc version 4.3.0 20071230 with
-m64 

FAIL: gcc.dg/struct/wo_prof_global_var.c execution test
FAIL: gcc.dg/struct/wo_prof_local_var.c execution test
FAIL: gcc.dg/struct/wo_prof_mult_field_peeling.c execution test
FAIL: gcc.dg/struct/wo_prof_two_strs.c execution test
FAIL: gcc.dg/struct/w_prof_global_var.c execution,    -O3 -fwhole-program
-combine -fipa-type-escape -fprofile-use -fipa-struct-reorg -fdump-ipa-all
FAIL: gcc.dg/struct/w_prof_local_var.c execution,    -O3 -fwhole-program
-combine -fipa-type-escape -fprofile-use -fipa-struct-reorg -fdump-ipa-all
FAIL: gcc.dg/struct/w_prof_two_strs.c execution,    -O3 -fwhole-program
-combine -fipa-type-escape -fprofile-use -fipa-struct-reorg -fdump-ipa-all


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[hjl at lucon dot org]

------- Comment #7 from hjl at lucon dot org  2007-12-30 15:45 -------
(In reply to comment #4)
> 
> So I wonder whether this test works correctly even without -fipa-struct-reorg.
> Would you please try to compile it with all other options (-O3 -fdump-ipa-all
> -fwhole-program -combine -fipa-type-escape) and run it on your machine?
> 

It works fine:

[hjl@gnu-14 gcc]$  /export/build/gnu/gcc/build-ia64-linux/gcc/xgcc
-B/export/build/gnu/gcc/build-ia64-linux/gcc/
/net/gnu-13/export/gnu/src/gcc/gcc/gcc/testsuite/gcc.dg/struct/wo_prof_global_var.c
  -O3  -fdump-ipa-all -fwhole-program -combine -fipa-type-escape
-fno-show-column  -lm   -o ./wo_prof_global_var.exe
[hjl@gnu-14 gcc]$ ./wo_prof_global_var.exe
[hjl@gnu-14 gcc]$


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[hjl at lucon dot org]

------- Comment #8 from hjl at lucon dot org  2007-12-30 15:49 -------
Created an attachment (id=14844)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14844&action=view)
Dump files without -fipa-struct-reorg


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[olga at gcc dot gnu dot org]

------- Comment #9 from olga at gcc dot gnu dot org  2008-01-07 13:48 -------
(In reply to comment #8)

Would you please try the Alon's patch from PR 34701?
I am not sure but may be it's related.

Thank you,
Olga


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534

[Bug tree-optimization/34534] Multiple gcc.dg/struct/wo_prof_xxxx execution failures

[hjl dot tools at gmail dot com]

------- Comment #10 from hjl dot tools at gmail dot com  2008-01-24 22:37 -------
Revision 131801 is OK.


-- 

hjl dot tools at gmail dot com changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34534