[PATCH] middle-end/110541 - VEC_PERM_EXPR documentation is off

[PATCH] middle-end/110541 - VEC_PERM_EXPR documentation is off

[Richard Biener]

The following adjusts the tree.def documentation about VEC_PERM_EXPR
which wasn't adjusted when the restrictions of permutes with constant
mask were relaxed.

OK?

Thanks,
Richard.

	PR middle-end/110541
	* tree.def (VEC_PERM_EXPR): Adjust documentation to reflect
	reality.
---
 gcc/tree.def | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

diff --git a/gcc/tree.def b/gcc/tree.def
index 1fc2ca7a724..9e1a54ac2f9 100644
--- a/gcc/tree.def
+++ b/gcc/tree.def
@@ -565,13 +565,20 @@ DEFTREECODE (VEC_COND_EXPR, "vec_cond_expr", tcc_expression, 3)
 
    N = length(mask)
    foreach i in N:
-     M = mask[i] % (2*N)
-     A = M < N ? v0[M] : v1[M-N]
+     M = mask[i] % (length(v0) + length(v1))
+     A[i] = M < length(v0) ? v0[M] : v1[M - length(v0)]
 
-   V0 and V1 are vectors of the same type.  MASK is an integer-typed
-   vector.  The number of MASK elements must be the same with the
-   number of elements in V0 and V1.  The size of the inner type
-   of the MASK and of the V0 and V1 must be the same.
+   V0 and V1 are vectors of the same type.
+
+   When MASK is not constant:
+     MASK is an integer-typed vector.  The number of MASK elements must
+     be the same with the number of elements in V0 and V1.  The size of
+     the inner type of the MASK and of the V0 and V1 must be the same.
+
+   When MASK is constant:
+     MASK is an integer-typed vector.   MASK elements outside of
+     [0, length(V0) + length(V1) - 1] invoke undefined behavior (the
+     modulo operation above doesn't apply).
 */
 DEFTREECODE (VEC_PERM_EXPR, "vec_perm_expr", tcc_expression, 3)
 
-- 
2.35.3

Re: [PATCH] middle-end/110541 - VEC_PERM_EXPR documentation is off

[Richard Biener]

On Wed, 5 Jul 2023, Richard Sandiford wrote:

> Richard Biener via Gcc-patches <gcc-patches@gcc.gnu.org> writes:
> > The following adjusts the tree.def documentation about VEC_PERM_EXPR
> > which wasn't adjusted when the restrictions of permutes with constant
> > mask were relaxed.
> 
> I was going to complain about having two copies of the documentation,
> but then I realised that generic.texi doesn't document VEC_PERM_EXPR.
> So... oops.

Yeah, also noticed that ...

> >
> > OK?
> >
> > Thanks,
> > Richard.
> >
> > 	PR middle-end/110541
> > 	* tree.def (VEC_PERM_EXPR): Adjust documentation to reflect
> > 	reality.
> > ---
> >  gcc/tree.def | 19 +++++++++++++------
> >  1 file changed, 13 insertions(+), 6 deletions(-)
> >
> > diff --git a/gcc/tree.def b/gcc/tree.def
> > index 1fc2ca7a724..9e1a54ac2f9 100644
> > --- a/gcc/tree.def
> > +++ b/gcc/tree.def
> > @@ -565,13 +565,20 @@ DEFTREECODE (VEC_COND_EXPR, "vec_cond_expr", tcc_expression, 3)
> >  
> >     N = length(mask)
> >     foreach i in N:
> > -     M = mask[i] % (2*N)
> > -     A = M < N ? v0[M] : v1[M-N]
> > +     M = mask[i] % (length(v0) + length(v1))
> > +     A[i] = M < length(v0) ? v0[M] : v1[M - length(v0)]
> >  
> > -   V0 and V1 are vectors of the same type.  MASK is an integer-typed
> > -   vector.  The number of MASK elements must be the same with the
> > -   number of elements in V0 and V1.  The size of the inner type
> > -   of the MASK and of the V0 and V1 must be the same.
> > +   V0 and V1 are vectors of the same type.
> > +
> > +   When MASK is not constant:
> > +     MASK is an integer-typed vector.  The number of MASK elements must
> > +     be the same with the number of elements in V0 and V1.  The size of
> 
> Preexisting, but s/same with/same as/

Fixed.

> > +     the inner type of the MASK and of the V0 and V1 must be the same.
> > +
> > +   When MASK is constant:
> > +     MASK is an integer-typed vector.   MASK elements outside of
> > +     [0, length(V0) + length(V1) - 1] invoke undefined behavior (the
> > +     modulo operation above doesn't apply).
> 
> I don't remember the last rule.  I thought the modulo did still apply.
> (But the canonical form is to remove obvious modulo opportunities.)
> 
> E.g. a VLA reverse-and-rotate pattern might have { N-2, N-3, N-4, ... }.
> That will wrap at the final position to 2N-1, but that seems OK.

OK, I'll remove that sentence.

Pushed as follows.

Richard.

From 38f2d33e5119e6ae39f2702caced7e9b224cbc4f Mon Sep 17 00:00:00 2001
From: Richard Biener <rguenther@suse.de>
Date: Wed, 5 Jul 2023 08:53:01 +0200
Subject: [PATCH] middle-end/110541 - VEC_PERM_EXPR documentation is off
To: gcc-patches@gcc.gnu.org

The following adjusts the tree.def documentation about VEC_PERM_EXPR
which wasn't adjusted when the restrictions of permutes with constant
mask were relaxed.

	PR middle-end/110541
	* tree.def (VEC_PERM_EXPR): Adjust documentation to reflect
	reality.
---
 gcc/tree.def | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/gcc/tree.def b/gcc/tree.def
index 1fc2ca7a724..be94b7ece0a 100644
--- a/gcc/tree.def
+++ b/gcc/tree.def
@@ -565,13 +565,18 @@ DEFTREECODE (VEC_COND_EXPR, "vec_cond_expr", tcc_expression, 3)
 
    N = length(mask)
    foreach i in N:
-     M = mask[i] % (2*N)
-     A = M < N ? v0[M] : v1[M-N]
+     M = mask[i] % (length(v0) + length(v1))
+     A[i] = M < length(v0) ? v0[M] : v1[M - length(v0)]
 
-   V0 and V1 are vectors of the same type.  MASK is an integer-typed
-   vector.  The number of MASK elements must be the same with the
-   number of elements in V0 and V1.  The size of the inner type
-   of the MASK and of the V0 and V1 must be the same.
+   V0 and V1 are vectors of the same type.
+
+   When MASK is not constant:
+     MASK is an integer-typed vector.  The number of MASK elements must
+     be the same as the number of elements in V0 and V1.  The size of
+     the inner type of the MASK and of the V0 and V1 must be the same.
+
+   When MASK is constant:
+     MASK is an integer-typed vector.
 */
 DEFTREECODE (VEC_PERM_EXPR, "vec_perm_expr", tcc_expression, 3)
 
-- 
2.35.3

Re: [PATCH] middle-end/110541 - VEC_PERM_EXPR documentation is off

[Richard Sandiford]

Richard Biener via Gcc-patches <gcc-patches@gcc.gnu.org> writes:
> The following adjusts the tree.def documentation about VEC_PERM_EXPR
> which wasn't adjusted when the restrictions of permutes with constant
> mask were relaxed.

I was going to complain about having two copies of the documentation,
but then I realised that generic.texi doesn't document VEC_PERM_EXPR.
So... oops.

>
> OK?
>
> Thanks,
> Richard.
>
> 	PR middle-end/110541
> 	* tree.def (VEC_PERM_EXPR): Adjust documentation to reflect
> 	reality.
> ---
>  gcc/tree.def | 19 +++++++++++++------
>  1 file changed, 13 insertions(+), 6 deletions(-)
>
> diff --git a/gcc/tree.def b/gcc/tree.def
> index 1fc2ca7a724..9e1a54ac2f9 100644
> --- a/gcc/tree.def
> +++ b/gcc/tree.def
> @@ -565,13 +565,20 @@ DEFTREECODE (VEC_COND_EXPR, "vec_cond_expr", tcc_expression, 3)
>  
>     N = length(mask)
>     foreach i in N:
> -     M = mask[i] % (2*N)
> -     A = M < N ? v0[M] : v1[M-N]
> +     M = mask[i] % (length(v0) + length(v1))
> +     A[i] = M < length(v0) ? v0[M] : v1[M - length(v0)]
>  
> -   V0 and V1 are vectors of the same type.  MASK is an integer-typed
> -   vector.  The number of MASK elements must be the same with the
> -   number of elements in V0 and V1.  The size of the inner type
> -   of the MASK and of the V0 and V1 must be the same.
> +   V0 and V1 are vectors of the same type.
> +
> +   When MASK is not constant:
> +     MASK is an integer-typed vector.  The number of MASK elements must
> +     be the same with the number of elements in V0 and V1.  The size of

Preexisting, but s/same with/same as/

> +     the inner type of the MASK and of the V0 and V1 must be the same.
> +
> +   When MASK is constant:
> +     MASK is an integer-typed vector.   MASK elements outside of
> +     [0, length(V0) + length(V1) - 1] invoke undefined behavior (the
> +     modulo operation above doesn't apply).

I don't remember the last rule.  I thought the modulo did still apply.
(But the canonical form is to remove obvious modulo opportunities.)

E.g. a VLA reverse-and-rotate pattern might have { N-2, N-3, N-4, ... }.
That will wrap at the final position to 2N-1, but that seems OK.

LGTM otherwise FWIW.

Thanks,
Richard