[PATCH] small speedup in 'nla_strlcpy' in file '\lib\nlattr.c'

[PATCH] small speedup in 'nla_strlcpy' in file '\lib\nlattr.c'

[Christophe JAILLET]

Hi,

I don't know if it worses a patch, but function 'nla_strlcpy' in file 
'\lib\nlattr.c' could be improved a bit.
There is no need to memset(0) the whole buffer. Only the end of it has 
to be cleared.

Usually, 'dstsize' is small (most of the time IFNAMSIZ = 16 bytes), so 
the gain should be real small, if noticeable.

I'm not sure of who to cc: the mail. I've just added Alexey Kuznetsov 
which is one of the author of the modified file and still in the 
MAINTAINERS list.


Hope this helps,
Best regards.

Signed-off-by: Christophe Jaillet<christophe.jaillet@wanadoo.fr>

--- a/lib/nlattr.c    2012-02-06 21:40:04.421875000 +0100
+++ b/lib/nlattr.c    2012-02-06 21:50:43.093750000 +0100
@@ -248,8 +248,8 @@
      if (dstsize > 0) {
          size_t len = (srclen >= dstsize) ? dstsize - 1 : srclen;

-        memset(dst, 0, dstsize);
          memcpy(dst, src, len);
+        memset(dst + len, 0, dstsize - len);
      }

      return srclen;

Re: [PATCH] small speedup in 'nla_strlcpy' in file '\lib\nlattr.c'

[Dan Carpenter]

[-- Attachment #1: Type: text/plain, Size: 1339 bytes --]

On Mon, Feb 06, 2012 at 10:10:27PM +0100, Christophe JAILLET wrote:
> Hi,
> 
> I don't know if it worses a patch, but function 'nla_strlcpy' in
> file '\lib\nlattr.c' could be improved a bit.
> There is no need to memset(0) the whole buffer. Only the end of it
> has to be cleared.
> 
> Usually, 'dstsize' is small (most of the time IFNAMSIZ = 16 bytes),
> so the gain should be real small, if noticeable.
> 
> I'm not sure of who to cc: the mail. I've just added Alexey
> Kuznetsov which is one of the author of the modified file and still
> in the MAINTAINERS list.
> 

So if we're not going to notice the speedup, then it's not worth it.
Mostly this is only used when people type "ifup eth0" and so a
speedup wouldn't even matter.  It is used in taskstats so maybe if
you did a benchmark and showed that it was faster then that would
be a compelling argument.  But every patch needs to have a reason to
apply it, and "maybe this helps but maybe not" is not good enough
salesmanship, you need actual data.

Use the ./scripts/get_maintainer.pl to find which places to send
this.

Also your email client mangled the patch so it doesn't apply.  Read
Documentation/email-clients.txt.  Send the patch to yourself first
and verify that you can do a "cat raw_email.txt | patch -p1".

regards,
dan carpenter



[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]