From: Julian Seward <jseward@acm.org>
Date: Fri, 3 Mar 2006 21:02:18 +0000 (+0000)
Subject: get_abit_and_vbyte/get_abit can be a significant bottleneck when
X-Git-Tag: svn/VALGRIND_3_2_0~237
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=eedadc5fb340eb61692615f41a06e9fa9a7d68dd;p=thirdparty%2Fvalgrind.git

get_abit_and_vbyte/get_abit can be a significant bottleneck when
painting/checking huge numbers of bytes resulting from client check
requests.  Inlining them helps significantly.



git-svn-id: svn://svn.valgrind.org/valgrind/trunk@5710
---

diff --git a/memcheck/mc_main.c b/memcheck/mc_main.c
index ec51703d31..39f3c01ba3 100644
--- a/memcheck/mc_main.c
+++ b/memcheck/mc_main.c
@@ -352,7 +352,7 @@ static inline UWord byte_offset_w ( UWord wordszB, Bool bigendian,
 
 /* --------------- Fundamental functions --------------- */
 
-static 
+static inline
 void get_abit_and_vbyte ( /*OUT*/UWord* abit, 
                           /*OUT*/UWord* vbyte,
                           Addr a )
@@ -362,7 +362,7 @@ void get_abit_and_vbyte ( /*OUT*/UWord* abit,
    *abit  = read_bit_array(sm->abits, a & 0xFFFF);
 } 
 
-static 
+static inline
 UWord get_abit ( Addr a )
 {
    SecMap* sm = get_secmap_readable(a);