linux-sparse.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
To: linux-sparse@vger.kernel.org
Cc: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
Subject: [PATCH 13/16] cmp: canonicalize sext(x) cmpu C (with C >= SMAX)
Date: Sun,  8 Nov 2020 02:21:23 +0100	[thread overview]
Message-ID: <20201108012126.94339-4-luc.vanoostenryck@gmail.com> (raw)
In-Reply-To: <20201108012126.94339-1-luc.vanoostenryck@gmail.com>

A unsigned compare of a sign-extended value against a value
bigger than the original SMAX is equivalent to a signed
compare against 0.

Canonicalize to the signed compare against 0.

Note: ultimately both forms are just a test of the sign bit.

Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
---
 simplify.c                       | 12 ++++++++++++
 validation/optim/cmp-sext-uimm.c |  1 -
 2 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/simplify.c b/simplify.c
index a59918697762..8a8b39b81e20 100644
--- a/simplify.c
+++ b/simplify.c
@@ -1149,6 +1149,18 @@ static int simplify_compare_constant(struct instruction *insn, long long value)
 			return replace_pseudo(insn, &insn->src1, def->src);
 		}
 		switch (insn->opcode) {
+		case OP_SET_BE:
+			if (value >= sign_bit(osize)) {
+				replace_binop_value(insn, OP_SET_GE, 0);
+				return replace_pseudo(insn, &insn->src1, def->src);
+			}
+			break;
+		case OP_SET_A:
+			if (value >= sign_bit(osize)) {
+				replace_binop_value(insn, OP_SET_LT, 0);
+				return replace_pseudo(insn, &insn->src1, def->src);
+			}
+			break;
 		case OP_SET_LT: case OP_SET_LE:
 			if (value >= sign_bit(osize))
 				return replace_with_value(insn, 1);
diff --git a/validation/optim/cmp-sext-uimm.c b/validation/optim/cmp-sext-uimm.c
index cc89a80694a1..05da042f5030 100644
--- a/validation/optim/cmp-sext-uimm.c
+++ b/validation/optim/cmp-sext-uimm.c
@@ -19,7 +19,6 @@ int sext_gtu_m3(int x) { return (sext(x) > (NEG - 3)) == (x < 0); }
 /*
  * check-name: cmp-sext-uimm
  * check-command: test-linearize -Wno-decl $file
- * check-known-to-fail
  *
  * check-output-ignore
  * check-output-returns: 1
-- 
2.29.2


  parent reply	other threads:[~2020-11-08  1:21 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-11-08  1:19 [PATCH 00/16] simplify & canonicalize compares Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 01/16] cmp: adapt testcase for compares' canonicalization Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 02/16] cmp: add testcases for the simplification of compares Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 03/16] cmp: add signed/unsigned to opcode table Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 04/16] cmp: move some code in a separate function: simplify_compare_constant() Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 05/16] cmp: use a few helpers for the simplification of compares Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 06/16] cmp: canonicalize unsigned (x {<,>=} C) --> (x {<=,>} C-1) Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 07/16] cmp: simplify unsigned (x {<=,>} UMAX) into {1,0} Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 08/16] cmp: canonicalize unsigned compare with UMAX or UMAX-1 Luc Van Oostenryck
2020-11-08  1:19 ` [PATCH 09/16] cmp: canonicalize unsigned (x {<=,>} SMAX) Luc Van Oostenryck
2020-11-08  1:21 ` [PATCH 10/16] cmp: simplify sext(x) cmp C --> x cmp C Luc Van Oostenryck
2020-11-08  1:21   ` [PATCH 11/16] cmp: simplify zext(x) " Luc Van Oostenryck
2020-11-08  1:21   ` [PATCH 12/16] cmp: simplify sext(x) cmps {SMAX,SMIN} Luc Van Oostenryck
2020-11-08  1:21   ` Luc Van Oostenryck [this message]
2020-11-08  1:21   ` [PATCH 14/16] cmp: simplify zext(x) cmps C Luc Van Oostenryck
2020-11-08  1:21   ` [PATCH 15/16] cmp: simplify zext(x) cmpu C Luc Van Oostenryck
2020-11-08  1:21   ` [PATCH 16/16] cmp: simplify compares and sign/zero extend Luc Van Oostenryck

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=20201108012126.94339-4-luc.vanoostenryck@gmail.com \
    --to=luc.vanoostenryck@gmail.com \
    --cc=linux-sparse@vger.kernel.org \
    /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 NNTP newsgroup(s).