Re: [PATCH] Correct dir.c to compile on Solaris 9

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



söndag 15 april 2007 23:03 skrev Josef Weidendorfer:
> On Sunday 15 April 2007, Johannes Schindelin wrote:
> > On Sun, 15 Apr 2007, Shawn O. Pearce wrote:
> > >  static int simple_length(const char *match)
> > >  {
> > > -	const char special[256] = {
> > > -		[0] = 1, ['?'] = 1,
> > > -		['\\'] = 1, ['*'] = 1,
> > > -		['['] = 1
> > > -	};
> > >  	int len = -1;
> > >  
> > >  	for (;;) {
> > >  		unsigned char c = *match++;
> > >  		len++;
> > > -		if (special[c])
> > > +		switch (c) {
> > > +		case 0: case '?':
> > > +		case '\\': case '*':
> > > +		case '[':
> > >  			return len;
> > > +		}
> > >  	}
> > >  }
> > 
> > You are replacing a table-based check with a switch based, which might be 
> > substantially slower (depends on how often cmp_name() is called).
> 
> Or faster. When the table gives a cache miss and has to be
> loaded from main memory, I am quite sure that 5 compares in a row are
> faster than the cache miss.
It is five compares time the length of the path being examined. The cache miss
only occurs once per invocation even inthe worst case.  Judging from where
the code is invoked, cache misses should be rare which means the table is much
faster. 

As for the table being a micro-optimization, that still holds true 

> 
> Actually, with the switch, the compiler is free to implement it with a
> table (and gcc usually does this, probably even using a substantially
> smaller table). The table-based check in contrast looks
It usually uses binary search. For gcc to create a lookup table you'll need
a large number of case's.
> like some kind of micro-optimization which makes the code IMHO more
> difficult to read, and which only would be justified with meassured
> improvements.

The table lookup *is* faster (meastured), but that doesn't make a big difference
on the total CPU used. The muliple-case-per line thing (both versions, however makes is hard
to read.  

-- robin

diff --git a/dir.c b/dir.c
index 7426fde..0780f23 100644
--- a/dir.c
+++ b/dir.c
@@ -423,12 +423,22 @@ static int cmp_name(const void *p1, const void *p2)
  */
 static int simple_length(const char *match)
 {
-       const char special[256] = {
-               [0] = 1, ['?'] = 1,
-               ['\\'] = 1, ['*'] = 1,
-               ['['] = 1
-       };
        int len = -1;
+       static const char special[256] = {
+               1,0,0,0,0,0,0,0, /* nul */
+               0,0,0,0,0,0,0,0,
+               0,0,0,0,0,0,0,0,
+               0,0,0,0,0,0,0,0,
+               0,0,0,0,0,0,0,0,
+               0,0,1,0,0,0,0,0, /* * */
+               0,0,0,0,0,0,0,0,
+               0,0,0,0,0,0,0,1, /* ? */
+
+               0,0,0,0,0,0,0,0,
+               0,0,0,0,0,0,0,0,
+               0,0,0,0,0,0,0,0,
+               0,0,0,1,1,0,0,0  /* [ \ */
+       };

        for (;;) {
                unsigned char c = *match++;
-
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]