From: Eric Dumazet <dada1@cosmosbay.com>
To: David Miller <davem@davemloft.net>
Cc: Michael Chan <mchan@broadcom.com>, netdev@vger.kernel.org
Subject: [PATCH] tg3 : avoid an expensive divide
Date: Tue, 06 Feb 2007 21:53:24 +0100 [thread overview]
Message-ID: <45C8EAC4.9020803@cosmosbay.com> (raw)
In-Reply-To: <20070206.114659.107250775.davem@davemloft.net>
[-- Attachment #1: Type: text/plain, Size: 1873 bytes --]
Hi David
During an oprofile session of linux-2.6.20 on a dual opteron system, I noticed
an expensive divide was done in tg3_poll().
I am using gcc-4.1.1, so the following comment from drivers/net/tg3.c seems
over-optimistic :
/* Do not place this n-ring entries value into the tp struct itself,
* we really want to expose these constants to GCC so that modulo et
* al. operations are done with shifts and masks instead of with
* hw multiply/modulo instructions. Another solution would be to
* replace things like '% foo' with '& (foo - 1)'.
*/
#define TG3_RX_RCB_RING_SIZE(tp) \
((tp->tg3_flags2 & TG3_FLG2_5705_PLUS) ? 512 : 1024)
Assembly code before patch :
(oprofile results included)
6434 0.0088 :ffffffff803684b9: mov 0x6f0(%r15),%eax
587 8.0e-04 :ffffffff803684c0: and $0x40000,%eax
2170 0.0030 :ffffffff803684c5: cmp $0x1,%eax
:ffffffff803684c8: lea 0x1(%r13),%eax
:ffffffff803684cc: sbb %ecx,%ecx
2051 0.0028 :ffffffff803684ce: xor %edx,%edx
:ffffffff803684d0: and $0x200,%ecx
20 2.7e-05 :ffffffff803684d6: add $0x200,%ecx
1986 0.0027 :ffffffff803684dc: div %ecx
103427 0.1410 :ffffffff803684de: cmp %edx,0xffffffffffffff7c(%rbp)
Assembly code after the suggested patch :
ffffffff803684b9: mov 0x6f0(%r15),%eax
ffffffff803684c0: and $0x40000,%eax
ffffffff803684c5: cmp $0x1,%eax
ffffffff803684c8: sbb %eax,%eax
ffffffff803684ca: inc %r13d
ffffffff803684cd: and $0x200,%eax
ffffffff803684d2: add $0x1ff,%eax
ffffffff803684d7: and %eax,%r13d
ffffffff803684da: cmp %r13d,0xffffffffffffff7c(%rbp)
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
[-- Attachment #2: tg3_avoid_divide.patch --]
[-- Type: text/plain, Size: 373 bytes --]
--- linux-2.6.20/drivers/net/tg3.c.orig 2007-02-06 22:30:39.000000000 +0100
+++ linux-2.6.20-ed/drivers/net/tg3.c 2007-02-06 22:32:42.000000000 +0100
@@ -3384,7 +3384,7 @@
}
next_pkt_nopost:
sw_idx++;
- sw_idx %= TG3_RX_RCB_RING_SIZE(tp);
+ sw_idx &= (TG3_RX_RCB_RING_SIZE(tp) - 1);
/* Refresh hw_idx to see if there is new work */
if (sw_idx == hw_idx) {
next prev parent reply other threads:[~2007-02-06 21:00 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-01-03 16:33 [PATCH/RFC 00/10] Transparent proxying patches version 4 KOVACS Krisztian
2007-01-03 16:34 ` [PATCH/RFC 01/10] Implement local diversion of IPv4 skbs KOVACS Krisztian
2007-01-10 6:46 ` Patrick McHardy
2007-01-10 9:31 ` Balazs Scheidler
2007-01-10 12:32 ` Patrick McHardy
2007-01-10 13:27 ` Ingo Oeser
2007-01-10 13:42 ` Patrick McHardy
2007-01-11 14:05 ` KOVACS Krisztian
2007-01-10 10:17 ` KOVACS Krisztian
2007-01-10 12:19 ` Patrick McHardy
2007-01-16 12:49 ` KOVACS Krisztian
2007-01-16 13:19 ` Patrick McHardy
2007-01-03 16:34 ` [PATCH/RFC 02/10] Port redirection support for TCP KOVACS Krisztian
2007-01-03 16:35 ` [PATCH/RFC 03/10] Don't do the TCP socket lookup if we already have one attached KOVACS Krisztian
2007-01-03 16:35 ` [PATCH/RFC 04/10] Don't do the UDP " KOVACS Krisztian
2007-01-03 16:36 ` [PATCH/RFC 05/10] Remove local address check on IP output KOVACS Krisztian
2007-01-10 6:47 ` Patrick McHardy
2007-01-10 10:01 ` KOVACS Krisztian
2007-02-06 14:36 ` IP_FREEBIND and CAP_NET_ADMIN (was: Re: [PATCH/RFC 05/10] Remove local address check on IP output) KOVACS Krisztian
2007-02-06 19:46 ` IP_FREEBIND and CAP_NET_ADMIN David Miller
2007-02-06 20:53 ` Eric Dumazet [this message]
2007-02-06 21:19 ` [PATCH] tg3 : avoid an expensive divide David Miller
2007-02-06 22:09 ` Michael Chan
2007-02-06 21:27 ` David Miller
2007-02-07 9:54 ` Andi Kleen
2007-02-07 9:45 ` David Miller
2007-02-07 9:56 ` Eric Dumazet
2007-02-07 10:27 ` Andi Kleen
2007-02-06 22:05 ` Michael Chan
2007-02-06 21:25 ` David Miller
2007-02-06 21:35 ` Eric Dumazet
2007-02-06 22:17 ` David Miller
2007-01-03 16:36 ` [PATCH/RFC 06/10] Create a tproxy flag in struct sk_buff KOVACS Krisztian
2007-01-03 16:37 ` [PATCH/RFC 07/10] Export UDP socket lookup function KOVACS Krisztian
2007-01-03 16:37 ` [PATCH/RFC 08/10] iptables tproxy table KOVACS Krisztian
2007-01-10 12:40 ` Patrick McHardy
2007-01-03 16:38 ` [PATCH/RFC 09/10] iptables TPROXY target KOVACS Krisztian
2007-01-10 12:45 ` Patrick McHardy
2007-01-03 16:38 ` [PATCH/RFC 10/10] iptables tproxy match KOVACS Krisztian
2007-01-03 17:23 ` [PATCH/RFC 00/10] Transparent proxying patches version 4 Evgeniy Polyakov
2007-01-08 20:30 ` KOVACS Krisztian
2007-01-03 19:33 ` Lennert Buytenhek
2007-01-04 12:13 ` KOVACS Krisztian
2007-01-04 12:16 ` Lennert Buytenhek
2007-01-07 14:11 ` Harald Welte
2007-01-07 16:11 ` Lennert Buytenhek
2007-01-07 23:58 ` Harald Welte
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=45C8EAC4.9020803@cosmosbay.com \
--to=dada1@cosmosbay.com \
--cc=davem@davemloft.net \
--cc=mchan@broadcom.com \
--cc=netdev@vger.kernel.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).