[Bug target/30210] New: Altivec builtins return wrong types

[Bug target/30210] New: Altivec builtins return wrong types

[irar at il dot ibm dot com]

The return value of builtin function rs6000_builtin_mul_widen_even
for Altivec is always signed, while it should be signed/unsigned 
as relevant (builtin vmuloub returns vector signed short, instead of vector
unsigned short, as defined by the altivec PIM).

It seems to be a more general problem with altivec builtins declaration.

Ira


-- 
           Summary: Altivec builtins return wrong types
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: irar at il dot ibm dot com
 GCC build triplet: powerpc*-*
  GCC host triplet: powerpc*-*
GCC target triplet: powerpc*-*


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins return wrong types

[bonzini at gnu dot org]

------- Comment #1 from bonzini at gnu dot org  2006-12-14 17:23 -------
Builtin vmuloub has always returned "wrong" types, because it derives its types
simply from the modes of the altivec_vmuloub pattern.  For this reason, the C
front-end uses vec_vmuloub instead: this one is expanded by
resolve_overloaded_builtin to add the appropriate casts on both the arguments
and the results.

The way to solve this bug could be to pass the two operands to
rs6000_builtin_mul_widen_even, and get back the fully-built tree, including
conversions to the appropriate return type.


-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bonzini at gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins return wrong types

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2006-12-14 17:29 -------
(In reply to comment #1)
> Builtin vmuloub has always returned "wrong" types, because it derives its types
> simply from the modes of the altivec_vmuloub pattern. 
Except that is wrong.  Look at the altivec PIM and you will see it returns
vector unsigned short.
Take the following testcase:
#include <altivec.h>
vector unsigned short f(vector unsigned char a, vector unsigned char b)
{
 return vec_vmuloub(a, b);
}

This should work even if we cannot assign from vector signed short to vector
unsigend short.

I think this whole type issue is a mess and needs some improvement.
Maybe next week I can get to that.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2006-12-14 17:29:55
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins return wrong types

[bonzini at gnu dot org]

------- Comment #3 from bonzini at gnu dot org  2006-12-14 17:46 -------
Don't confuse __builtin_altivec_vmuloub (only used by middle-end, has the
"wrong" types) with __builtin_vec_vmuloub (cannot be created by middle-end,
lowered by the C front-end to __builtin_altivec_vmuloub after checking the
parameter types are correct).

The types of __builtin_altivec_vmuloub don't have anything to do with what the
PIM manual says.  And anyway, __builtin_altivec_vmuloub can have the "wrong"
types because it is always decorated with the appropriate NOP_EXPRs: this may
be bad from a memory usage point of view, but it is not incorrect.  Also, note
that it has always worked this way: what changed is *who* does the lowering
(either a preprocessor macro vec_vmuloub, or resolve_overloaded_builtin
starting from builtin __builtin_vec_vmuloub).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins return wrong types

[bonzini at gnu dot org]

------- Comment #4 from bonzini at gnu dot org  2006-12-15 12:05 -------
Seems invalid.  Ira, if you have a specific bug (missed optimization, or
whatever) triggered by this, please reopen and change the subject.


-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins return wrong types

[irar at il dot ibm dot com]

------- Comment #5 from irar at il dot ibm dot com  2006-12-20 12:20 -------
Paolo, thanks for the explanation! The problem originates from PR 22372, so I
will not open another bug report for it.

Thanks,
Ira


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins have inaccurate return types

[bonzini at gnu dot org]

------- Comment #6 from bonzini at gnu dot org  2006-12-20 12:28 -------
Ok, so we have an agreement on why this is bad.

I've reopened the bug myself and made it block 22372.  A fix for 22372 can be
done independent of this PR, but it will be inefficient memory-wise.


-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
OtherBugsDependingO|                            |22372
              nThis|                            |
           Severity|normal                      |enhancement
             Status|RESOLVED                    |REOPENED
         Resolution|INVALID                     |
            Summary|Altivec builtins return     |Altivec builtins have
                   |wrong types                 |inaccurate return types


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins have inaccurate return types

[bonzini at gnu dot org]

------- Comment #7 from bonzini at gnu dot org  2006-12-20 15:25 -------
To make it clearer, there is a fix for this PR other than

>> The way to solve this bug could be to pass the two operands to
>> rs6000_builtin_mul_widen_even, and get back the fully-built tree,
>> including conversions to the appropriate return type.

The solution is exactly to modify __builtin_altivec_vmuloub and friends so that
it has the accurate return types and argument types (unsigned).  This would be,
in practice, a rewrite of the way rs6000.c defines Altivec builtins, because if
you want signed/unsigned information it is not enough to look at the modes in
the RTL patterns.  I don't see any other way to do so, except to use a big
table (possibly a subset of the huge table in rs6000-c.c).

The reason I had closed the bug as invalid, was that pinskia was erroneously
relating it to PR14899, and saying that this would have caused regressions with
Mark Shinwell's patch for stricter conversions between vector types.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins have inaccurate return types

[irar at il dot ibm dot com]

------- Comment #8 from irar at il dot ibm dot com  2007-01-15 10:04 -------
(In reply to comment #2)
> I think this whole type issue is a mess and needs some improvement.
> Maybe next week I can get to that.

Andrew, are you still planning to solve this, or should I prepare a fix for 
rs6000_builtin_mul_widen_even as suggested by Paolo in comment #1 to close PR
22372?

Ira


-- 

irar at il dot ibm dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins have inaccurate return types

[bonzini at gnu dot org]

------- Comment #9 from bonzini at gnu dot org  2009-05-03 14:25 -------
*** Bug 40009 has been marked as a duplicate of this bug. ***


-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rguenth at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins have inaccurate return types

[bonzini at gnu dot org]

------- Comment #10 from bonzini at gnu dot org  2009-05-03 14:26 -------
The wrong types in __builtin_altivec_ctu cause this:

FAIL: gcc.c-torture/compile/pr39943.c  -O3 -fomit-frame-pointer  (internal
compiler error)
FAIL: gcc.c-torture/compile/pr39943.c  -O3 -fomit-frame-pointer  (test for
excess errors)
FAIL: gcc.c-torture/compile/pr39943.c  -O3 -fomit-frame-pointer -funroll-loops 
(internal compiler error)
FAIL: gcc.c-torture/compile/pr39943.c  -O3 -fomit-frame-pointer -funroll-loops 
(test for excess errors)
FAIL: gcc.c-torture/compile/pr39943.c  -O3 -fomit-frame-pointer
-funroll-all-loops -finline-functions  (internal compiler error)
FAIL: gcc.c-torture/compile/pr39943.c  -O3 -fomit-frame-pointer
-funroll-all-loops -finline-functions  (test for excess errors)
FAIL: gcc.c-torture/compile/pr39943.c  -O3 -g  (internal compiler error)
FAIL: gcc.c-torture/compile/pr39943.c  -O3 -g  (test for excess errors)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins have inaccurate return types

[rguenth at gcc dot gnu dot org]

------- Comment #11 from rguenth at gcc dot gnu dot org  2009-05-03 15:54 -------
And as there is _builtin_altivec_cts in addition to _builtin_altivec_ctu these
builtins should better have correct types from the start.  In fact they
look more constrained than "opaque" anyway - so why are they using opaque
arguments / return types at all?

Not an enhancement, wrong-code in the middle-end sense.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|enhancement                 |normal
           Keywords|                            |wrong-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] Altivec builtins have inaccurate return types

[rguenth at gcc dot gnu dot org]

------- Comment #12 from rguenth at gcc dot gnu dot org  2009-05-03 15:56 -------
*** Bug 22372 has been marked as a duplicate of this bug. ***


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] [4.5 Regression] Altivec builtins have inaccurate return types

[rguenth at gcc dot gnu dot org]

------- Comment #13 from rguenth at gcc dot gnu dot org  2009-05-03 15:57 -------
And now "causing" regressions with type checking enabled.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Altivec builtins have       |[4.5 Regression] Altivec
                   |inaccurate return types     |builtins have inaccurate
                   |                            |return types
   Target Milestone|---                         |4.5.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] [4.5 Regression] Altivec builtins have inaccurate return types

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] [4.5 Regression] Altivec builtins have inaccurate return types

[pinskia at gcc dot gnu dot org]

------- Comment #14 from pinskia at gcc dot gnu dot org  2009-06-12 12:20 -------
*** Bug 40425 has been marked as a duplicate of this bug. ***


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dominiq at lps dot ens dot
                   |                            |fr


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] [4.5 Regression] Altivec builtins have inaccurate return types

[rguenth at gcc dot gnu dot org]

------- Comment #15 from rguenth at gcc dot gnu dot org  2009-07-08 21:03 -------
*** Bug 40690 has been marked as a duplicate of this bug. ***


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |janis at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] [4.5 Regression] Altivec builtins have inaccurate return types

[rguenth at gcc dot gnu dot org]

------- Comment #16 from rguenth at gcc dot gnu dot org  2009-07-08 21:04 -------
Mike - you said you have patches for this?


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |meissner at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] [4.5 Regression] Altivec builtins have inaccurate return types

[meissner at linux dot vnet dot ibm dot com]

------- Comment #17 from meissner at linux dot vnet dot ibm dot com  2009-07-08 23:42 -------
Subject: Re:  [4.5 Regression] Altivec builtins have inaccurate return types

On Wed, Jul 08, 2009 at 09:04:03PM -0000, rguenth at gcc dot gnu dot org wrote:
> 
> 
> ------- Comment #16 from rguenth at gcc dot gnu dot org  2009-07-08 21:04 -------
> Mike - you said you have patches for this?
> 
> 
> -- 
> 
> rguenth at gcc dot gnu dot org changed:
> 
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                  CC|                            |meissner at gcc dot gnu dot
>                    |                            |org
> 
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210
> 
> ------- You are receiving this mail because: -------
> You are on the CC list for the bug, or are watching someone who is.

Yes, it should be in the next power7 patch.  I hope to send out the patch for
review shortly, but I have a few regressions in my code.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210

[Bug target/30210] [4.5 Regression] Altivec builtins have inaccurate return types

[pinskia at gcc dot gnu dot org]

------- Comment #18 from pinskia at gcc dot gnu dot org  2009-10-22 21:34 -------
THis has been fixed for awhile now.


-- 

pinskia at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30210