[Bug rtl-optimization/40987] Wrong optimization with if-conversion

[Bug rtl-optimization/40987] Wrong optimization with if-conversion

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40987

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|i686                        |
               Host|i686                        |HWI == 32
              Build|i686                        |

--- Comment #11 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-01-21 21:27:18 UTC ---
This bug is now hard to hit on the trunk as i?86 requires host wide word as
64bit.

[Bug rtl-optimization/40987] ifcvt sometimes goes wrong for integer modes > 64bit

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=40987

--- Comment #12 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #11)
> This bug is now hard to hit on the trunk as i?86 requires host wide word as
> 64bit.

HWINT is now required to be 64bit for all targets even.
So this might happen if there is some TImode but then the patch might be best
thing really.

[Bug rtl-optimization/40987] ifcvt sometimes goes wrong for integer modes > 64bit

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=40987

--- Comment #13 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Here is a x86_64-linux-gnu testcase:
```
#include <stdio.h>

typedef __int128_t mytype;
#define value (((__int128_t)(((unsigned long long)__LONG_LONG_MAX__)+1)) |
(((__int128_t)0xffffffffffffffffull)<<64))

mytype func(long arg) __attribute__((noinline,noclone));
mytype func(long arg) {
   mytype val = 0;
   if (arg < 0) {
    // __int128_t t = 0x8000000000000000ull |
(((__int128_t)0xffffffffffffffff)<<64);
      val = value;
   }
   return val;
}

int main() {
   mytype result = func(-1);
   if (result != value) {
      printf("failed.\n");
      return 1;
   } 
   printf("passed.\n");
   return 0;
}
```

It failed until GCC 4.9.0

[Bug rtl-optimization/40987] ifcvt sometimes goes wrong for integer modes > 64bit

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=40987

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED
   Target Milestone|---                         |4.9.0

--- Comment #14 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Looks like this was fixed with r0-127324-ge15eb172b0dd (aka PR 59545).
Basically a signed integer overflow inside the compiler was causing the wrong
code happening.

[Bug rtl-optimization/40987] Wrong optimization with if-conversion

[steven at gcc dot gnu dot org]

-- 

steven at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2009-
                   |                            |09/msg01655.html
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
           Keywords|                            |patch
   Last reconfirmed|0000-00-00 00:00:00         |2009-09-26 10:18:29
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40987

[Bug rtl-optimization/40987] Wrong optimization with if-conversion

[mikpe at it dot uu dot se]

------- Comment #10 from mikpe at it dot uu dot se  2009-09-24 14:38 -------
Patch posted for review:
http://gcc.gnu.org/ml/gcc-patches/2009-09/msg01655.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40987

[Bug rtl-optimization/40987] Wrong optimization with if-conversion

[mikpe at it dot uu dot se]

------- Comment #9 from mikpe at it dot uu dot se  2009-09-19 16:57 -------
Seems like an if-conversion bug, in particular noce_try_store_flag_constants()
appears to break on HWI32 platforms when long long literals are involved.

In this test case, noce_t_s_f_c() is invoked with an IF where a (the false
value) and b (the true value) are both DImode CONST_INTs. a is zero. b is
stored as 0x80000000, which is truncated but sign-extends to the correct value.

noce_t_s_f_c() then computes a subtraction and some log2() on these truncated
values, and selects the second code generation template "x = (test != 0) <<
3;".
The code becomes (from the ce1 dump file):

(insn 27 8 28 2 pr40987.c:4 (parallel [
            (set (reg:DI 62)
                (sign_extend:DI (reg/v:SI 60 [ arg ])))
            (clobber (reg:CC 17 flags))
            (clobber (scratch:SI))
        ]) 90 {*extendsidi2_1} (nil))

(insn 28 27 29 2 pr40987.c:4 (parallel [
            (set (reg/v:DI 58 [ val ])
                (lshiftrt:DI (reg:DI 62)
                    (const_int 63 [0x3f])))
            (clobber (reg:CC 17 flags))
        ]) 394 {*lshrdi3_1} (nil))

# since this is a logical downshift, it produces 0 or 1 from the sign bit

(insn 29 28 17 2 pr40987.c:4 (parallel [
            (set (reg/v:DI 58 [ val ])
                (ashift:DI (reg/v:DI 58 [ val ])
                    (const_int 31 [0x1f])))
            (clobber (reg:CC 17 flags))
        ]) 356 {*ashldi3_1} (nil))

# which is shifted up to produce 0 or 0x80000000, the high 32 bits are lost

All gcc-4.x releases have this bug. It does not trigger in gcc-3.4.6. There it
seems that the true/false values are swapped compared to 4.x, and a test near
the top of noce_t_s_f_c() causes it to bail out and not perform the conversion.
(The test looks the same in 4.x, but the IF-operands are swapped so it does not
bail.)

I don't know if this code can be fixed to handle longer-than-HWI literals. The
simplest solution might be to just detect the situation and bail out.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40987

[Bug rtl-optimization/40987] Wrong optimization with if-conversion

[rguenth at gcc dot gnu dot org]

------- Comment #8 from rguenth at gcc dot gnu dot org  2009-08-07 12:10 -------
Seems to be a HWI32 issue as I cannot reproduce the fail on x86_64 with -m32.
Which would indeed hint at a RTL if-conversion problem (I guess it uses
scc in a wrong way).


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rguenth at gcc dot gnu dot
                   |                            |org
          Component|c                           |rtl-optimization
           Keywords|                            |wrong-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40987