linux-sparse.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Nicolai Stange <nicstange@gmail.com>
To: linux-sparse@vger.kernel.org
Cc: Christopher Li <sparse@chrisli.org>,
	Josh Triplett <josh@joshtriplett.org>,
	Luc Van Oostenryck <luc.vanoostenryck@gmail.com>,
	Nicolai Stange <nicstange@gmail.com>
Subject: [PATCH v3 18/21] evaluate: relax some constant expression rules for pointer expressions
Date: Mon, 01 Feb 2016 03:44:38 +0100	[thread overview]
Message-ID: <87fuxdgne1.fsf@gmail.com> (raw)
In-Reply-To: <87lh75jh9l.fsf@gmail.com> (Nicolai Stange's message of "Mon, 01 Feb 2016 03:28:38 +0100")

The official constraints on constant expressions [6.6] are insanely
strict in that they do not allow some constructs commonly used in the
wild.

Relax them by treating
- address constants cast to different pointer type as address constants
  again,
- address constants cast to arithmetic type as arithmetic constant
  expressions
- conditional expressions whose true and false branches both yield
  address constants as address constants,
- and conditional expressions whose condition is an address constant
  as an constant expression to the extent their true and false branches
  allow.

Signed-off-by: Nicolai Stange <nicstange@gmail.com>
---
 evaluate.c | 38 ++++++++++++++++++++++++++++++++------
 1 file changed, 32 insertions(+), 6 deletions(-)

diff --git a/evaluate.c b/evaluate.c
index 0101e61..ff51d84 100644
--- a/evaluate.c
+++ b/evaluate.c
@@ -1129,14 +1129,23 @@ static struct symbol *evaluate_conditional_expression(struct expression *expr)
 	}
 
 	/*
+	 * A conditional operator yields a particular constant
+	 * expression type only if all of its three subexpressions are
+	 * of that type [6.6(6), 6.6(8)].
+	 * As an extension, relax this restriction by allowing any
+	 * constant expression type for the condition expression.
+	 *
 	 * A conditional operator never yields an address constant
 	 * [6.6(9)].
+	 * However, as an extension, if the condition is any constant
+	 * expression, and the true and false expressions are both
+	 * address constants, mark the result as an address constant.
 	 */
-	expr->constexpr_flags = (expr->conditional->constexpr_flags &
-				(*true)->constexpr_flags &
-				expr->cond_false->constexpr_flags &
-				~CONSTEXPR_FLAG_DECAY_CONSTS_MASK &
-				~CONSTEXPR_FLAG_ADDR_CONST);
+	if (expr->conditional->constexpr_flags &
+		(CONSTEXPR_FLAG_ARITH_CONST_EXPR | CONSTEXPR_FLAG_ADDR_CONST))
+		expr->constexpr_flags = (*true)->constexpr_flags &
+			expr->cond_false->constexpr_flags &
+			~CONSTEXPR_FLAG_DECAY_CONSTS_MASK;
 
 	lclass = classify_type(ltype, &ltype);
 	rclass = classify_type(rtype, &rtype);
@@ -2786,9 +2795,18 @@ static struct symbol *evaluate_cast(struct expression *expr)
 		/*
 		 * Casts of integer literals to pointer type yield
 		 * address constants [6.6(9)].
+		 *
+		 * As an extension, treat address constants cast to a
+		 * different pointer type as address constants again.
+		 *
+		 * As another extension, treat integer constant
+		 * expressions (in contrast to literals) cast to
+		 * pointer type as address constants.
 		 */
 		if (class1 & TYPE_PTR &&
-			(target->constexpr_flags & CONSTEXPR_FLAG_INT_CONST))
+			(target->constexpr_flags &
+				(CONSTEXPR_FLAG_INT_CONST_EXPR |
+					CONSTEXPR_FLAG_ADDR_CONST)))
 			expr->constexpr_flags = CONSTEXPR_FLAG_ADDR_CONST;
 	} else {
 		/*
@@ -2798,6 +2816,14 @@ static struct symbol *evaluate_cast(struct expression *expr)
 		expr->constexpr_flags = target->constexpr_flags &
 			~CONSTEXPR_FLAG_DECAY_CONSTS_MASK &
 			~CONSTEXPR_FLAG_ADDR_CONST;
+
+		/*
+		 * As an extension, treat address constants cast to
+		 * integer type as an arithmetic constant.
+		 */
+		if (target->constexpr_flags & CONSTEXPR_FLAG_ADDR_CONST)
+			expr->constexpr_flags = CONSTEXPR_FLAG_ARITH_CONST_EXPR;
+
 		/*
 		 * Cast to float type -> not an integer constant
 		 * expression [6.6(6)].
-- 
2.7.0


  parent reply	other threads:[~2016-02-01  2:44 UTC|newest]

Thread overview: 71+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-02-01  2:28 [PATCH v3 00/21] improve constexpr handling Nicolai Stange
2016-02-01  2:29 ` [PATCH v3 01/21] expression: introduce additional expression constness tracking flags Nicolai Stange
2016-03-15 21:23   ` Luc Van Oostenryck
2016-02-01  2:30 ` [PATCH v3 02/21] expression: init constexpr_flags at expression allocation Nicolai Stange
2016-03-15 16:59   ` Luc Van Oostenryck
2016-02-01  2:31 ` [PATCH v3 03/21] expression: examine constness of casts at evaluation only Nicolai Stange
2016-03-15 20:43   ` Luc Van Oostenryck
2016-02-01  2:32 ` [PATCH v3 04/21] expression: examine constness of binops and alike " Nicolai Stange
2016-03-15 17:06   ` Luc Van Oostenryck
2016-02-01  2:33 ` [PATCH v3 05/21] expression: examine constness of preops " Nicolai Stange
2016-03-15 17:09   ` Luc Van Oostenryck
2016-02-01  2:34 ` [PATCH v3 06/21] expression: examine constness of conditionals " Nicolai Stange
2016-03-15 17:11   ` Luc Van Oostenryck
2016-02-01  2:35 ` [PATCH v3 07/21] expression: add support for tagging arithmetic constant expressions Nicolai Stange
2016-03-15 17:13   ` Luc Van Oostenryck
2016-02-01  2:36 ` [PATCH v3 08/21] expression, evaluate: add support for tagging address constants Nicolai Stange
2016-03-15 17:15   ` Luc Van Oostenryck
2016-02-01  2:37 ` [PATCH v3 09/21] evaluate: check static storage duration objects' intializers' constness Nicolai Stange
2016-03-15 17:28   ` Luc Van Oostenryck
2016-02-01  2:38 ` [PATCH v3 10/21] expression, evaluate: recognize static objects as address constants Nicolai Stange
2016-03-15 17:38   ` Luc Van Oostenryck
2016-02-01  2:39 ` [PATCH v3 11/21] evaluate: recognize address constants created through casts Nicolai Stange
2016-03-15 17:44   ` Luc Van Oostenryck
2016-02-01  2:39 ` [PATCH v3 12/21] evaluate: recognize address constants created through pointer arithmetic Nicolai Stange
2016-03-15 17:46   ` Luc Van Oostenryck
2016-02-01  2:40 ` [PATCH v3 13/21] evaluate: recognize members of static compound objects as address constants Nicolai Stange
2016-03-15 17:46   ` Luc Van Oostenryck
2016-02-01  2:41 ` [PATCH v3 14/21] evaluate: recognize string literals " Nicolai Stange
2016-03-15 17:46   ` Luc Van Oostenryck
2016-02-01  2:42 ` [PATCH v3 15/21] expression: recognize references to labels " Nicolai Stange
2016-03-15 17:47   ` Luc Van Oostenryck
2016-02-01  2:42 ` [PATCH v3 16/21] expression: examine constness of __builtin_offsetof at evaluation only Nicolai Stange
2016-03-15 19:52   ` Luc Van Oostenryck
2016-02-01  2:43 ` [PATCH v3 17/21] symbol: flag builtins constant_p, safe_p and warning as constexprs Nicolai Stange
2016-03-15 19:45   ` Luc Van Oostenryck
2016-02-01  2:44 ` Nicolai Stange [this message]
2016-03-15 17:47   ` [PATCH v3 18/21] evaluate: relax some constant expression rules for pointer expressions Luc Van Oostenryck
2016-03-15 19:44     ` Luc Van Oostenryck
2016-03-15 18:10   ` Luc Van Oostenryck
2016-02-01  2:45 ` [PATCH v3 19/21] expression, evaluate: support compound literals as address constants Nicolai Stange
2016-03-15 20:02   ` Luc Van Oostenryck
2016-02-01  2:46 ` [PATCH v3 20/21] symbol: do not inherit storage modifiers from base types at examination Nicolai Stange
2016-03-15 20:31   ` Luc Van Oostenryck
2016-02-01  2:47 ` [PATCH v3 21/21] evaluation: treat comparsions between types as integer constexpr Nicolai Stange
2016-03-15 20:34   ` Luc Van Oostenryck
2016-02-19  8:22 ` [PATCH v3 00/21] improve constexpr handling Nicolai Stange
2016-02-24  9:45   ` Christopher Li
2016-02-24 12:13     ` Nicolai Stange
2016-03-15 16:54   ` Luc Van Oostenryck
2016-03-15 22:36 ` Luc Van Oostenryck
2016-10-28 20:28   ` Luc Van Oostenryck
2016-11-23  3:12 ` Christopher Li
2016-11-23  4:05   ` Luc Van Oostenryck
2016-11-23  6:49     ` Christopher Li
2016-11-23  8:39       ` Nicolai Stange
2016-11-23 15:36         ` Christopher Li
2016-11-23 16:43           ` Nicolai Stange
2016-11-23 17:38             ` Christopher Li
2016-11-23 18:23               ` Christopher Li
2016-11-23 18:33               ` Nicolai Stange
2016-11-24  1:18                 ` Christopher Li
2016-11-24  9:45                   ` Nicolai Stange
2016-11-24 11:24                     ` Christopher Li
2016-11-24 17:22                       ` Luc Van Oostenryck
2016-12-06  6:00                     ` Christopher Li
2016-12-06 16:54                       ` Luc Van Oostenryck
2017-03-29 14:42                       ` Luc Van Oostenryck
2017-03-31  5:06                         ` Christopher Li
2017-03-31  8:55                           ` Luc Van Oostenryck
2017-03-31 10:40                             ` Christopher Li
2017-03-31 19:47                               ` 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=87fuxdgne1.fsf@gmail.com \
    --to=nicstange@gmail.com \
    --cc=josh@joshtriplett.org \
    --cc=linux-sparse@vger.kernel.org \
    --cc=luc.vanoostenryck@gmail.com \
    --cc=sparse@chrisli.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).