From: Richard Guenther <richard.guenther@gmail.com>
To: Jason Merrill <jason@redhat.com>
Cc: gcc-patches List <gcc-patches@gcc.gnu.org>
Subject: Re: RFA: PATCH to make fold_indirect_ref_1 fold more things
Date: Mon, 01 Nov 2010 22:17:00 -0000 [thread overview]
Message-ID: <AANLkTikOE7cqFuMLuEgFH=o81F7OiNWHn610mt+S9AjU@mail.gmail.com> (raw)
In-Reply-To: <4CCDD13E.5030403@redhat.com>
On Sun, Oct 31, 2010 at 9:27 PM, Jason Merrill <jason@redhat.com> wrote:
> On 10/29/2010 07:32 PM, Richard Guenther wrote:
>>>
>>> I guess I can just do this folding in the constexpr expander...
>>
>> Yeah, I think that should be safer.
>
> OK, how about this patch that only adds folding INDIRECT_REF of
> POINTER_PLUS_EXPR to ARRAY_REF (and combines the two POINTER_PLUS_EXPR
> hunks)?
That works for me.
Thanks,
Richard.
> commit 49fd7e2c159bc5de19055c64c19e26778167cc28
> Author: Jason Merrill <jason@redhat.com>
> Date: Tue Oct 26 14:49:37 2010 -0400
>
> * fold-const.c (fold_indirect_ref_1): Handle folding
> POINTER_PLUS_EXPR to ARRAY_REF.
>
> diff --git a/gcc/fold-const.c b/gcc/fold-const.c
> index decb0fb..dd69a20 100644
> --- a/gcc/fold-const.c
> +++ b/gcc/fold-const.c
> @@ -15649,53 +15649,59 @@ fold_indirect_ref_1 (location_t loc, tree type,
> tree op0)
> }
> }
>
> - /* ((foo*)&vectorfoo)[1] => BIT_FIELD_REF<vectorfoo,...> */
> if (TREE_CODE (sub) == POINTER_PLUS_EXPR
> && TREE_CODE (TREE_OPERAND (sub, 1)) == INTEGER_CST)
> {
> tree op00 = TREE_OPERAND (sub, 0);
> tree op01 = TREE_OPERAND (sub, 1);
> - tree op00type;
>
> STRIP_NOPS (op00);
> - op00type = TREE_TYPE (op00);
> - if (TREE_CODE (op00) == ADDR_EXPR
> - && TREE_CODE (TREE_TYPE (op00type)) == VECTOR_TYPE
> - && type == TREE_TYPE (TREE_TYPE (op00type)))
> + if (TREE_CODE (op00) == ADDR_EXPR)
> {
> - HOST_WIDE_INT offset = tree_low_cst (op01, 0);
> - tree part_width = TYPE_SIZE (type);
> - unsigned HOST_WIDE_INT part_widthi = tree_low_cst (part_width,
> 0)/BITS_PER_UNIT;
> - unsigned HOST_WIDE_INT indexi = offset * BITS_PER_UNIT;
> - tree index = bitsize_int (indexi);
> + tree op00type;
> + op00 = TREE_OPERAND (op00, 0);
> + op00type = TREE_TYPE (op00);
>
> - if (offset/part_widthi <= TYPE_VECTOR_SUBPARTS (TREE_TYPE
> (op00type)))
> - return fold_build3_loc (loc,
> - BIT_FIELD_REF, type, TREE_OPERAND (op00, 0),
> - part_width, index);
> + /* ((foo*)&vectorfoo)[1] => BIT_FIELD_REF<vectorfoo,...> */
> + if (TREE_CODE (op00type) == VECTOR_TYPE
> + && type == TREE_TYPE (op00type))
> + {
> + HOST_WIDE_INT offset = tree_low_cst (op01, 0);
> + tree part_width = TYPE_SIZE (type);
> + unsigned HOST_WIDE_INT part_widthi = tree_low_cst (part_width,
> 0)/BITS_PER_UNIT;
> + unsigned HOST_WIDE_INT indexi = offset * BITS_PER_UNIT;
> + tree index = bitsize_int (indexi);
>
> - }
> - }
> + if (offset/part_widthi <= TYPE_VECTOR_SUBPARTS (op00type))
> + return fold_build3_loc (loc,
> + BIT_FIELD_REF, type, op00,
> + part_width, index);
>
> -
> - /* ((foo*)&complexfoo)[1] => __imag__ complexfoo */
> - if (TREE_CODE (sub) == POINTER_PLUS_EXPR
> - && TREE_CODE (TREE_OPERAND (sub, 1)) == INTEGER_CST)
> - {
> - tree op00 = TREE_OPERAND (sub, 0);
> - tree op01 = TREE_OPERAND (sub, 1);
> - tree op00type;
> -
> - STRIP_NOPS (op00);
> - op00type = TREE_TYPE (op00);
> - if (TREE_CODE (op00) == ADDR_EXPR
> - && TREE_CODE (TREE_TYPE (op00type)) == COMPLEX_TYPE
> - && type == TREE_TYPE (TREE_TYPE (op00type)))
> - {
> - tree size = TYPE_SIZE_UNIT (type);
> - if (tree_int_cst_equal (size, op01))
> - return fold_build1_loc (loc, IMAGPART_EXPR, type,
> - TREE_OPERAND (op00, 0));
> + }
> + /* ((foo*)&complexfoo)[1] => __imag__ complexfoo */
> + else if (TREE_CODE (op00type) == COMPLEX_TYPE
> + && type == TREE_TYPE (op00type))
> + {
> + tree size = TYPE_SIZE_UNIT (type);
> + if (tree_int_cst_equal (size, op01))
> + return fold_build1_loc (loc, IMAGPART_EXPR, type, op00);
> + }
> + /* ((foo *)&fooarray)[1] => fooarray[1] */
> + else if (TREE_CODE (op00type) == ARRAY_TYPE
> + && type == TREE_TYPE (op00type))
> + {
> + tree type_domain = TYPE_DOMAIN (op00type);
> + tree min_val = size_zero_node;
> + if (type_domain && TYPE_MIN_VALUE (type_domain))
> + min_val = TYPE_MIN_VALUE (type_domain);
> + op01 = size_binop_loc (loc, EXACT_DIV_EXPR, op01,
> + TYPE_SIZE_UNIT (type));
> + op01 = size_binop_loc (loc, PLUS_EXPR, op01, min_val);
> + op0 = build4 (ARRAY_REF, type, op00, op01,
> + NULL_TREE, NULL_TREE);
> + SET_EXPR_LOCATION (op0, loc);
> + return op0;
> + }
> }
> }
>
>
>
prev parent reply other threads:[~2010-11-01 22:12 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-10-27 2:16 Jason Merrill
2010-10-27 20:10 ` Richard Guenther
2010-10-28 16:18 ` Jason Merrill
2010-10-29 16:06 ` Richard Guenther
2010-10-29 21:51 ` Jason Merrill
2010-10-30 7:53 ` Richard Guenther
2010-10-31 23:44 ` Jason Merrill
2010-11-01 22:17 ` Richard Guenther [this message]
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='AANLkTikOE7cqFuMLuEgFH=o81F7OiNWHn610mt+S9AjU@mail.gmail.com' \
--to=richard.guenther@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jason@redhat.com \
/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).