linux-kernel - [PATCH] coccinelle: deref

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <1526885939-1741-1-git-send-email-Julia.Lawall@lip6.fr>
Date:   Mon, 21 May 2018 08:58:59 +0200
From:   Julia Lawall <Julia.Lawall@...6.fr>
To:     Masahiro Yamada <yamada.masahiro@...ionext.com>
Cc:     kernel-janitors@...r.kernel.org,
        Gilles Muller <Gilles.Muller@...6.fr>,
        Nicolas Palix <nicolas.palix@...g.fr>,
        Michal Marek <michal.lkml@...kovi.net>, cocci@...teme.lip6.fr,
        linux-kernel@...r.kernel.org
Subject: [PATCH] coccinelle: deref_null: improve performance

Move rules looking for some special cases of safe dereferences before
the collection of NULL-tested values.  The special cases are fairly
rare, but somewhat costly to find, because isomorphisms create many
variants of the rules.  There is thus no need to search for them over
and over for each NULL tested expression.  Collecting them just once
is sufficient and more efficient.

Signed-off-by: Julia Lawall <Julia.Lawall@...6.fr>

---
 scripts/coccinelle/null/deref_null.cocci |   40 +++++++++++++++----------------
 1 file changed, 20 insertions(+), 20 deletions(-)

diff --git a/scripts/coccinelle/null/deref_null.cocci b/scripts/coccinelle/null/deref_null.cocci
index b16ccb7..cbc6184 100644
--- a/scripts/coccinelle/null/deref_null.cocci
+++ b/scripts/coccinelle/null/deref_null.cocci
@@ -14,18 +14,10 @@ virtual context
 virtual org
 virtual report
 
-@ifm@
-expression *E;
-statement S1,S2;
-position p1;
-@@
-
-if@p1 ((E == NULL && ...) || ...) S1 else S2
-
 // The following two rules are separate, because both can match a single
 // expression in different ways
 @pr1 expression@
-expression *ifm.E;
+expression E;
 identifier f;
 position p1;
 @@
@@ -33,7 +25,7 @@ position p1;
  (E != NULL && ...) ? <+...E->f@.....+> : ...
 
 @pr2 expression@
-expression *ifm.E;
+expression E;
 identifier f;
 position p2;
 @@
@@ -46,6 +38,14 @@ position p2;
  sizeof(<+...E->f@.....+>)
 )
 
+@ifm@
+expression *E;
+statement S1,S2;
+position p1;
+@@
+
+if@p1 ((E == NULL && ...) || ...) S1 else S2
+
 // For org and report modes
 
 @r depends on !context && (org || report) exists@
@@ -212,16 +212,8 @@ else S3
 // The following three rules are duplicates of ifm, pr1 and pr2 respectively.
 // It is need because the previous rule as already made a "change".
 
-@...1 depends on context && !org && !report@
-expression *E;
-statement S1,S2;
-position p1;
-@@
-
-if@p1 ((E == NULL && ...) || ...) S1 else S2
-
 @pr11 depends on context && !org && !report expression@
-expression *ifm1.E;
+expression E;
 identifier f;
 position p1;
 @@
@@ -229,7 +221,7 @@ position p1;
  (E != NULL && ...) ? <+...E->f@.....+> : ...
 
 @pr12 depends on context && !org && !report expression@
-expression *ifm1.E;
+expression E;
 identifier f;
 position p2;
 @@
@@ -242,6 +234,14 @@ position p2;
  sizeof(<+...E->f@.....+>)
 )
 
+@...1 depends on context && !org && !report@
+expression *E;
+statement S1,S2;
+position p1;
+@@
+
+if@p1 ((E == NULL && ...) || ...) S1 else S2
+
 @depends on context && !org && !report exists@
 expression subE <= ifm1.E;
 expression *ifm1.E;