public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Richard Sandiford <richard.sandiford@arm.com>
To: Richard Biener <rguenther@suse.de>
Cc: Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>,
	gcc Patches <gcc-patches@gcc.gnu.org>
Subject: Re: [2/2] PR96463 -- changes to type checking vec_perm_expr in middle end
Date: Tue, 04 Jan 2022 11:50:56 +0000	[thread overview]
Message-ID: <mptlezv3ejj.fsf@arm.com> (raw)
In-Reply-To: <72p4o95q-697o-1o21-po9r-58r6rq3nq9n@fhfr.qr> (Richard Biener's message of "Mon, 3 Jan 2022 11:40:23 +0100 (CET)")

Richard Biener <rguenther@suse.de> writes:
> On Fri, 17 Dec 2021, Richard Sandiford wrote:
>
>> Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org> writes:
>> > Hi,
>> > The attached patch rearranges order of type-check for vec_perm_expr
>> > and relaxes type checking for
>> > lhs = vec_perm_expr<rhs1, rhs2, mask>
>> >
>> > when:
>> > rhs1 == rhs2,
>> > lhs is variable length vector,
>> > rhs1 is fixed length vector,
>> > TREE_TYPE (lhs) == TREE_TYPE (rhs1)
>> >
>> > I am not sure tho if this check is correct ? My intent was to capture
>> > case when vec_perm_expr is used to "extend" fixed length vector to
>> > it's VLA equivalent.
>> 
>> VLAness isn't really the issue.  We want the same thing to work for
>> -msve-vector-bits=256, -msve-vector-bits=512, etc., even though the
>> vectors are fixed-length in that case.
>> 
>> The principle is that for:
>> 
>>   A = VEC_PERM_EXPR <B, C, D>;
>> 
>> the requirements are:
>> 
>> - A, B, C and D must be vectors
>> - A, B and C must have the same element type
>> - D must have an integer element type
>> - A and D must have the same number of elements (NA)
>> - B and C must have the same number of elements (NB)
>> 
>> The semantics are that we create a joined vector BC (all elements of B
>> followed by all element of C) and that:
>> 
>>   A[i] = BC[D[i] % (NB+NB)]
>> 
>> for 0 ≤ i < NA.
>> 
>> This operation makes sense even if NA != NB.
>
> But note that we don't currently expect NA != NB and the optab just
> has a single mode.

True, but we only need this for constant permutes.  They are already
special in that they allow the index elements to be wider than the data
elements.

Thanks,
Richard

>
> I'd rather go with the simpler patch I posted as reply to the earlier
> mail rather such large refactoring at this point.
>
> Richard.
>
>> Thanks,
>> Richard
>> 
>> >
>> > Thanks,
>> > Prathamesh
>> >
>> > diff --git a/gcc/tree-cfg.c b/gcc/tree-cfg.c
>> > index 672e384ef09..9f91878c468 100644
>> > --- a/gcc/tree-cfg.c
>> > +++ b/gcc/tree-cfg.c
>> > @@ -4325,10 +4325,11 @@ verify_gimple_assign_ternary (gassign *stmt)
>> >        break;
>> >  
>> >      case VEC_PERM_EXPR:
>> > -      if (!useless_type_conversion_p (lhs_type, rhs1_type)
>> > -	  || !useless_type_conversion_p (lhs_type, rhs2_type))
>> > +      if (TREE_CODE (rhs1_type) != VECTOR_TYPE
>> > +	  || TREE_CODE (rhs2_type) != VECTOR_TYPE
>> > +	  || TREE_CODE (rhs3_type) != VECTOR_TYPE)
>> >  	{
>> > -	  error ("type mismatch in %qs", code_name);
>> > +	  error ("vector types expected in %qs", code_name);
>> >  	  debug_generic_expr (lhs_type);
>> >  	  debug_generic_expr (rhs1_type);
>> >  	  debug_generic_expr (rhs2_type);
>> > @@ -4336,11 +4337,14 @@ verify_gimple_assign_ternary (gassign *stmt)
>> >  	  return true;
>> >  	}
>> >  
>> > -      if (TREE_CODE (rhs1_type) != VECTOR_TYPE
>> > -	  || TREE_CODE (rhs2_type) != VECTOR_TYPE
>> > -	  || TREE_CODE (rhs3_type) != VECTOR_TYPE)
>> > +      if (TREE_CODE (TREE_TYPE (rhs3_type)) != INTEGER_TYPE
>> > +	  || (TREE_CODE (rhs3) != VECTOR_CST
>> > +	      && (GET_MODE_BITSIZE (SCALAR_INT_TYPE_MODE
>> > +				    (TREE_TYPE (rhs3_type)))
>> > +		  != GET_MODE_BITSIZE (SCALAR_TYPE_MODE
>> > +				       (TREE_TYPE (rhs1_type))))))
>> >  	{
>> > -	  error ("vector types expected in %qs", code_name);
>> > +	  error ("invalid mask type in %qs", code_name);
>> >  	  debug_generic_expr (lhs_type);
>> >  	  debug_generic_expr (rhs1_type);
>> >  	  debug_generic_expr (rhs2_type);
>> > @@ -4348,15 +4352,18 @@ verify_gimple_assign_ternary (gassign *stmt)
>> >  	  return true;
>> >  	}
>> >  
>> > -      if (maybe_ne (TYPE_VECTOR_SUBPARTS (rhs1_type),
>> > -		    TYPE_VECTOR_SUBPARTS (rhs2_type))
>> > -	  || maybe_ne (TYPE_VECTOR_SUBPARTS (rhs2_type),
>> > -		       TYPE_VECTOR_SUBPARTS (rhs3_type))
>> > -	  || maybe_ne (TYPE_VECTOR_SUBPARTS (rhs3_type),
>> > -		       TYPE_VECTOR_SUBPARTS (lhs_type)))
>> > +      /* Accept lhs = vec_perm_expr<v, v, mask> if lhs is vector length agnostic,
>> > +	 and has same element type as v.  */
>> > +      if (!TYPE_VECTOR_SUBPARTS (lhs_type).is_constant ()
>> > +	  && operand_equal_p (rhs1, rhs2, 0)
>> > +	  && TYPE_VECTOR_SUBPARTS (rhs1_type).is_constant ()
>> > +	  && TREE_TYPE (lhs_type) == TREE_TYPE (rhs1_type)) 
>> > +	return false;
>> > +
>> > +      if (!useless_type_conversion_p (lhs_type, rhs1_type)
>> > +	  || !useless_type_conversion_p (lhs_type, rhs2_type))
>> >  	{
>> > -	  error ("vectors with different element number found in %qs",
>> > -		 code_name);
>> > +	  error ("type mismatch in %qs", code_name);
>> >  	  debug_generic_expr (lhs_type);
>> >  	  debug_generic_expr (rhs1_type);
>> >  	  debug_generic_expr (rhs2_type);
>> > @@ -4364,21 +4371,21 @@ verify_gimple_assign_ternary (gassign *stmt)
>> >  	  return true;
>> >  	}
>> >  
>> > -      if (TREE_CODE (TREE_TYPE (rhs3_type)) != INTEGER_TYPE
>> > -	  || (TREE_CODE (rhs3) != VECTOR_CST
>> > -	      && (GET_MODE_BITSIZE (SCALAR_INT_TYPE_MODE
>> > -				    (TREE_TYPE (rhs3_type)))
>> > -		  != GET_MODE_BITSIZE (SCALAR_TYPE_MODE
>> > -				       (TREE_TYPE (rhs1_type))))))
>> > +      if (maybe_ne (TYPE_VECTOR_SUBPARTS (rhs1_type),
>> > +		    TYPE_VECTOR_SUBPARTS (rhs2_type))
>> > +	  || maybe_ne (TYPE_VECTOR_SUBPARTS (rhs2_type),
>> > +		       TYPE_VECTOR_SUBPARTS (rhs3_type))
>> > +	  || maybe_ne (TYPE_VECTOR_SUBPARTS (rhs3_type),
>> > +		       TYPE_VECTOR_SUBPARTS (lhs_type)))
>> >  	{
>> > -	  error ("invalid mask type in %qs", code_name);
>> > +	  error ("vectors with different element number found in %qs",
>> > +		 code_name);
>> >  	  debug_generic_expr (lhs_type);
>> >  	  debug_generic_expr (rhs1_type);
>> >  	  debug_generic_expr (rhs2_type);
>> >  	  debug_generic_expr (rhs3_type);
>> >  	  return true;
>> >  	}
>> > -
>> >        return false;
>> >  
>> >      case SAD_EXPR:
>> 

  reply	other threads:[~2022-01-04 11:50 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-17 10:05 Prathamesh Kulkarni
2021-12-17 11:07 ` Richard Sandiford
2021-12-27 10:25   ` Prathamesh Kulkarni
2022-01-03 10:40   ` Richard Biener
2022-01-04 11:50     ` Richard Sandiford [this message]
2022-01-04 12:40       ` Richard Biener
2022-01-04 13:42         ` Richard Sandiford
2022-05-03 10:41           ` Prathamesh Kulkarni
2022-05-03 12:55             ` Richard Sandiford
2022-05-09 11:21               ` Prathamesh Kulkarni
2022-05-09 13:52                 ` Richard Sandiford
2022-05-09 15:51                   ` Prathamesh Kulkarni
2022-05-23 17:27                     ` Prathamesh Kulkarni
2022-05-31 11:35                       ` Prathamesh Kulkarni
2022-06-01  7:58                         ` Richard Biener

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=mptlezv3ejj.fsf@arm.com \
    --to=richard.sandiford@arm.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=prathamesh.kulkarni@linaro.org \
    --cc=rguenther@suse.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).