Re: [PATCH, rs6000] fix failure test cases caused by disabling mode promotion for pseudos [PR100952]

[Segher Boessenkool <segher@kernel.crashing.org>]

Hi!

On Tue, Jul 06, 2021 at 11:11:05AM +0800, HAO CHEN GUI wrote:
>    The patch changed matching conditions in pr81384.c and pr56605.c. 
> The original conditions failed to match due to mode promotion disabled.

> 	PR target/100952
> 	* gcc/testsuite/gcc.target/powerpc/pr56605.c: Change matching
> 	conditions.
> 	* gcc/testsuite/gcc.target/powerpc/pr81348.c: Likewise.
> 

> diff --git a/gcc/testsuite/gcc.target/powerpc/pr56605.c b/gcc/testsuite/gcc.target/powerpc/pr56605.c
> index 29efd815adc..2b7ddbd7410 100644
> --- a/gcc/testsuite/gcc.target/powerpc/pr56605.c
> +++ b/gcc/testsuite/gcc.target/powerpc/pr56605.c
> @@ -11,5 +11,5 @@ void foo (short* __restrict sb, int* __restrict ia)
>      ia[i] = (int) sb[i];
>  }
>  
> -/* { dg-final { scan-rtl-dump-times "\\\(compare:CC \\\((?:and|zero_extend):DI \\\(reg:\[SD\]I" 1 "combine" } } */
> +/* { dg-final { scan-rtl-dump-times "\\\(compare:CC \\\((?:and|zero_extend):SI \\\(subreg:SI \\\(reg:\[SD\]I" 1 "combine" } } */

So, this testcase only runs on 64-bit machines (even only on lp64
configurations).  But do we now always get a subreg?  And, can that
change again some time in the future?

Writing it as
/* { dg-final { scan-rtl-dump-times {\(compare:CC \((?:and|zero_extend):SI \(subreg:SI \(reg:[SD]I} 1 "combine" } } */
is easier to read btw.

If you get a subreg:SI of a reg:SI here, something is wrong.  And you
cannot have a zero_extend:SI of anything :SI either.

So what the original matched were
  (compare:CC (and:DI (reg:DI
and
  (compare:CC (zero_extend:DI (reg:SI
and now you want to allow a subreg:SI in that last one as well (and you
do not really care what it is a subreg of, you don't check what offset
anyway), so maybe just
/* { dg-final { scan-rtl-dump-times {\(compare:CC \((?:and|zero_extend):(?:DI \((?:sub)?reg:[SD]I} 1 "combine" } } */
will do what you want?

> --- a/gcc/testsuite/gcc.target/powerpc/pr81348.c
> +++ b/gcc/testsuite/gcc.target/powerpc/pr81348.c
> @@ -19,5 +19,5 @@ void d(void)
>          ***c = e;
>  }
>  
> -/* { dg-final { scan-assembler {\mlxsihzx\M}  } } */
> -/* { dg-final { scan-assembler {\mvextsh2d\M} } } */
> +/* { dg-final { scan-assembler {\mlha\M}  } } */
> +/* { dg-final { scan-assembler {\mmtvsrwa\M} } } */

(This test should not test for powerpc64*-*-* but powerpc*-*-* btw,
and that means it can just be left out, so just
/* { dg-do compile { target lp64 } } */
and nothing more).

Okay for trunk with those changes (the RE and lp64).  Thanks!
(Test if it works of course; I did not :-) )


Segher