[Bug ada/24992] New: gcc aborts compiling record whose component definition refers to 'Size of current type

[stephaniechc-gccbug at yahoo dot com]

procedure Main is
   type Bar(E : Integer) is null record;
   type CurrentType(D : Integer) is
  record
 Field : Bar(CurrentType'Size);  
 -- not sure if this is valid but the compiler shouldn't crash
  end record;
begin
   null;
end Main;

command line is

gcc -c main.adb

gcc 4.0.0 says

+===GNAT BUG DETECTED==+
| 4.0.0 (i686-pc-linux-gnu) in Attribute_to_gnu, at ada/trans.c:741|
| Error detected at main.adb:5:22  |
| Please submit a bug report; see http://gcc.gnu.org/bugs.html.|
| Include the entire contents of this bug box in the report.   |
| Include the exact gcc or gnatmake command that you entered.  |
| Also include sources listed below in gnatchop format |
| (concatenated together with no headers between files).   |
+==+

Both gcc 2.8.1 and gcc 4.0.0 aborts. The second line of the 2.8.1 box says

| 3.15p  (20020523) (i686-pc-linux-gnu) Gigi abort, Code=401   |

gcc 4.0.0 was built from 4.0.0 source (last change log entry 2005-04-20) with
default settings and c,c++,objc,ada
enabled.


-- 
   Summary: gcc aborts compiling record whose component definition
refers to 'Size of current type
   Product: gcc
   Version: 4.0.0
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: ada
AssignedTo: unassigned at gcc dot gnu dot org
    ReportedBy: stephaniechc-gccbug at yahoo dot com
  GCC host triplet: Linux 2.4.21-4-686-smp #1 SMP  i686 GNU/Linux
GCC target triplet: Linux 2.4.21-4-686-smp #1 SMP  i686 GNU/Linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24992

[Bug c/31691] New: optimized code taking the wrong branch

[stephaniechc-gccbug at yahoo dot com]

The following example takes the "wrong_branch" when compiled with 
gcc -O3 -fno-inline

gcc --version
gcc (GCC) 4.1.2
Copyright (C) 2006 Free Software Foundation, Inc.

= example =
#include 

static int get_kind(int v) {
volatile int k = v;
return k;
}

static int some_call(void)
{
return 0;
}

static void wrong_branch(void) {
printf("wrong_branch\n");
}

static int example(int arg)
{
int  kind;
kind = get_kind(arg);
if (kind == 9 || kind == 10 || kind == 5) {
if (some_call() == 0) {
int tmp;
//kind = get_kind(arg); // ok if this call is inserted
if (kind == 9 || kind == 10) {
tmp = arg;
} else {
wrong_branch();
} /* if */
return 0;
}
}
return 0;
} 

int main(void) {
int v = 10;
example(v);
return 0;
}

===

The example prints "wrong_branch" when executed. If the second call to get_kind
is uncommented, nothing is printed (which is correct).

The generated assembly code is

= example.s ==
example:
save%sp, -112, %sp
callget_kind, 0
 mov%i0, %o0
add %o0, -9, %g1
subcc   %g0, %g1, %g0
subx%g0, -1, %i0
cmp %g1, 1
bleu.LL11
 cmp%o0, 5
be  .LL11
 nop
.LL9:
jmp %i7+8
 restore %g0, 0, %o0
.LL11:
callsome_call, 0
 nop
cmp %o0, 0
bne .LL9
 cmp%i0, 0
bne .LL9
 nop
callwrong_branch, 0
 mov0, %i0
jmp %i7+8
 restore
===


-- 
   Summary: optimized code taking the wrong branch
   Product: gcc
   Version: 4.1.2
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: c
AssignedTo: unassigned at gcc dot gnu dot org
    ReportedBy: stephaniechc-gccbug at yahoo dot com
 GCC build triplet: sparc-sun-solaris2.8
  GCC host triplet: sparc-sun-solaris2.8+
GCC target triplet: sparc-sun-solaris2.8+


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31691

[Bug middle-end/31691] [4.1 regression] optimized code taking the wrong branch

[stephaniechc-gccbug at yahoo dot com]

--- Comment #4 from stephaniechc-gccbug at yahoo dot com  2007-04-25 14:26 
---
Created an attachment (id=13438)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=13438&action=view)
output of -fdump-tree-optimized


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31691