[PATCH] rs6000: Move V2DI vec_neg under power8-vector [PR105271]

[PATCH] rs6000: Move V2DI vec_neg under power8-vector [PR105271]

[Kewen.Lin]

Hi,

As PR105271 shows, __builtin_altivec_neg_v2di requires option
-mpower8-vector as its pattern expansion relies on subv2di which
has guard VECTOR_UNIT_P8_VECTOR_P (V2DImode).  This fix is to move
the related lines for __builtin_altivec_neg_v2di to the section
of stanza power8-vector.

Bootstrapped and regtested on powerpc64-linux-gnu P8 and
powerpc64le-linux-gnu P9 and P10.

Is it ok for trunk?

BR,
Kewen
-----
	PR target/105271

gcc/ChangeLog:

	* config/rs6000/rs6000-builtins.def (NEG_V2DI): Move to [power8-vector]
	stanza.

gcc/testsuite/ChangeLog:

	* gcc.target/powerpc/pr105271.c: New test.
---
 gcc/config/rs6000/rs6000-builtins.def       |  6 +++---
 gcc/testsuite/gcc.target/powerpc/pr105271.c | 14 ++++++++++++++
 2 files changed, 17 insertions(+), 3 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/powerpc/pr105271.c

diff --git a/gcc/config/rs6000/rs6000-builtins.def b/gcc/config/rs6000/rs6000-builtins.def
index 0f527c5d78f..f4a9f24bcc5 100644
--- a/gcc/config/rs6000/rs6000-builtins.def
+++ b/gcc/config/rs6000/rs6000-builtins.def
@@ -1190,9 +1190,6 @@
   const vd __builtin_altivec_neg_v2df (vd);
     NEG_V2DF negv2df2 {}

-  const vsll __builtin_altivec_neg_v2di (vsll);
-    NEG_V2DI negv2di2 {}
-
   void __builtin_altivec_stvx_v2df (vd, signed long, void *);
     STVX_V2DF altivec_stvx_v2df {stvec}

@@ -2136,6 +2133,9 @@
   const vus __builtin_altivec_nand_v8hi_uns (vus, vus);
     NAND_V8HI_UNS nandv8hi3 {}

+  const vsll __builtin_altivec_neg_v2di (vsll);
+    NEG_V2DI negv2di2 {}
+
   const vsc __builtin_altivec_orc_v16qi (vsc, vsc);
     ORC_V16QI orcv16qi3 {}

diff --git a/gcc/testsuite/gcc.target/powerpc/pr105271.c b/gcc/testsuite/gcc.target/powerpc/pr105271.c
new file mode 100644
index 00000000000..1c5f88cadcf
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/pr105271.c
@@ -0,0 +1,14 @@
+/* { dg-require-effective-target powerpc_vsx_ok } */
+/* { dg-options "-mdejagnu-cpu=power7" } */
+
+/* It's to verify no ICE here, ignore error messages about
+   the required options for vec_neg here.  */
+/* { dg-excess-errors "pr105271" } */
+
+#include <altivec.h>
+
+vector signed long long
+test (vector signed long long x)
+{
+  return vec_neg (x);
+}
--
2.27.0

Re: [PATCH] rs6000: Move V2DI vec_neg under power8-vector [PR105271]

[Jakub Jelinek]

Hi!

On Fri, Apr 15, 2022 at 04:08:15PM +0800, Kewen.Lin via Gcc-patches wrote:
> As PR105271 shows, __builtin_altivec_neg_v2di requires option
> -mpower8-vector as its pattern expansion relies on subv2di which
> has guard VECTOR_UNIT_P8_VECTOR_P (V2DImode).  This fix is to move
> the related lines for __builtin_altivec_neg_v2di to the section
> of stanza power8-vector.
> 
> Bootstrapped and regtested on powerpc64-linux-gnu P8 and
> powerpc64le-linux-gnu P9 and P10.
> 
> Is it ok for trunk?
> 
> BR,
> Kewen
> -----
> 	PR target/105271
> 
> gcc/ChangeLog:
> 
> 	* config/rs6000/rs6000-builtins.def (NEG_V2DI): Move to [power8-vector]
> 	stanza.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* gcc.target/powerpc/pr105271.c: New test.

I'd like to ping this patch, one of the last few remaining P1s we have for
GCC 12.

Thanks.

	Jakub

Re: [PATCH] rs6000: Move V2DI vec_neg under power8-vector [PR105271]

[Segher Boessenkool]

On Tue, Apr 26, 2022 at 05:16:18PM +0200, Jakub Jelinek wrote:
> Hi!
> 
> On Fri, Apr 15, 2022 at 04:08:15PM +0800, Kewen.Lin via Gcc-patches wrote:
> > As PR105271 shows, __builtin_altivec_neg_v2di requires option
> > -mpower8-vector as its pattern expansion relies on subv2di which
> > has guard VECTOR_UNIT_P8_VECTOR_P (V2DImode).  This fix is to move
> > the related lines for __builtin_altivec_neg_v2di to the section
> > of stanza power8-vector.
> > 
> > Bootstrapped and regtested on powerpc64-linux-gnu P8 and
> > powerpc64le-linux-gnu P9 and P10.
> > 
> > Is it ok for trunk?
> > 
> > BR,
> > Kewen
> > -----
> > 	PR target/105271
> > 
> > gcc/ChangeLog:
> > 
> > 	* config/rs6000/rs6000-builtins.def (NEG_V2DI): Move to [power8-vector]
> > 	stanza.
> > 
> > gcc/testsuite/ChangeLog:
> > 
> > 	* gcc.target/powerpc/pr105271.c: New test.
> 
> I'd like to ping this patch, one of the last few remaining P1s we have for
> GCC 12.

Heh.  I approved it this morning (off-list).  Kewen will commit it
soonish (somewhere during your night probably) :-)


Segher

Re: [PATCH] rs6000: Move V2DI vec_neg under power8-vector [PR105271]

[Kewen.Lin]

on 2022/4/26 11:25 PM, Segher Boessenkool wrote:
> On Tue, Apr 26, 2022 at 05:16:18PM +0200, Jakub Jelinek wrote:
>> Hi!
>>
>> On Fri, Apr 15, 2022 at 04:08:15PM +0800, Kewen.Lin via Gcc-patches wrote:
>>> As PR105271 shows, __builtin_altivec_neg_v2di requires option
>>> -mpower8-vector as its pattern expansion relies on subv2di which
>>> has guard VECTOR_UNIT_P8_VECTOR_P (V2DImode).  This fix is to move
>>> the related lines for __builtin_altivec_neg_v2di to the section
>>> of stanza power8-vector.
>>>
>>> Bootstrapped and regtested on powerpc64-linux-gnu P8 and
>>> powerpc64le-linux-gnu P9 and P10.
>>>
>>> Is it ok for trunk?
>>>
>>> BR,
>>> Kewen
>>> -----
>>> 	PR target/105271
>>>
>>> gcc/ChangeLog:
>>>
>>> 	* config/rs6000/rs6000-builtins.def (NEG_V2DI): Move to [power8-vector]
>>> 	stanza.
>>>
>>> gcc/testsuite/ChangeLog:
>>>
>>> 	* gcc.target/powerpc/pr105271.c: New test.
>>
>> I'd like to ping this patch, one of the last few remaining P1s we have for
>> GCC 12.
> 
> Heh.  I approved it this morning (off-list).  Kewen will commit it
> soonish (somewhere during your night probably) :-)
> 

Thank you both!  I just committed it as r12-8275.

BR,
Kewen