Jakub Jelinek <ja...@redhat.com> writes:
> Hi!
>
> The following testcase ICEs, because aarch64_gen_compare_reg_maybe_ze emits
> invalid RTL.
> For y_mode [QH]Imode it expects y to be of that mode (or CONST_INT that fits
> into that mode) and x being SImode; for non-CONST_INT y it zero extends y
> into SImode and compares that against x, for CONST_INT y it zero extends y
> into SImode.  The problem is that when the zero extended constant isn't
> usable directly, it forces it into a REG, but with y_mode mode, and then
> compares against y.  That is wrong, because it should force it into a SImode
> REG and compare that way.
>
> The following patch fixes that, bootstrapped/regtested on aarch64-linux, ok
> for trunk?
>
> 2020-04-01  Jakub Jelinek  <ja...@redhat.com>
>
>       PR target/94435
>       * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): For
>       y_mode E_[QH]Imode and y being a CONST_INT, change y_mode to SImode.
>
>       * gcc.target/aarch64/pr94435.c: New test.

OK, thanks.

Richard

>
> --- gcc/config/aarch64/aarch64.c.jj   2020-03-30 17:02:28.000000000 +0200
> +++ gcc/config/aarch64/aarch64.c      2020-04-01 11:32:48.877900235 +0200
> @@ -2371,7 +2371,10 @@ aarch64_gen_compare_reg_maybe_ze (RTX_CO
>    if (y_mode == E_QImode || y_mode == E_HImode)
>      {
>        if (CONST_INT_P (y))
> -     y = GEN_INT (INTVAL (y) & GET_MODE_MASK (y_mode));
> +     {
> +       y = GEN_INT (INTVAL (y) & GET_MODE_MASK (y_mode));
> +       y_mode = SImode;
> +     }
>        else
>       {
>         rtx t, cc_reg;
> --- gcc/testsuite/gcc.target/aarch64/pr94435.c.jj     2020-04-01 
> 11:36:21.172797217 +0200
> +++ gcc/testsuite/gcc.target/aarch64/pr94435.c        2020-04-01 
> 11:36:36.493573280 +0200
> @@ -0,0 +1,25 @@
> +/* PR target/94435 */
> +/* { dg-do compile } */
> +/* { dg-options "-march=armv8-a+nolse -moutline-atomics" } */
> +
> +int b, c, d, e, f, h;
> +short g;
> +int foo (int) __attribute__ ((__const__));
> +
> +void
> +bar (void)
> +{
> +  while (1)
> +    {
> +      while (1)
> +     {
> +       __atomic_load_n (&e, 0);
> +       if (foo (2))
> +         __sync_val_compare_and_swap (&c, 0, f);
> +       b = 1;
> +       if (h == e)
> +         break;
> +     }
> +      __sync_val_compare_and_swap (&g, -1, f);
> +    }
> +}
>
>       Jakub

Reply via email to