| To: | Anton Blanchard <anton@xxxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH] TSO Reloaded |
| From: | "David S. Miller" <davem@xxxxxxxxxxxxx> |
| Date: | Thu, 5 May 2005 23:13:45 -0700 |
| Cc: | netdev@xxxxxxxxxxx |
| In-reply-to: | <20050506054516.GE8255@krispykreme> |
| References: | <20050504230731.12be1bc3.davem@xxxxxxxxxxxxx> <20050506043653.GD8255@krispykreme> <20050506054516.GE8255@krispykreme> |
| Sender: | netdev-bounce@xxxxxxxxxxx |
On Fri, 6 May 2005 15:45:16 +1000
Anton Blanchard <anton@xxxxxxxxx> wrote:
> With a 9000 byte MTU I see page sized frags as expected. There are still
> some pages broken in two which is interesting.
I know what is causing the breaks, it's the "cache line align" crap
in tcp_sendmsg().
Comment out this alignment code in net/ipv4/tcp.c and see how it goes:
/* If page is cached, align
* offset to L1 cache boundary
*/
off = (off + L1_CACHE_BYTES - 1) &
~(L1_CACHE_BYTES - 1);
This alignment causes TSO coalescing to fail at these spots.
You should not see this effect when sendfile() is used for
the transfer.
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [PATCH] TSO Reloaded, Anton Blanchard |
|---|---|
| Next by Date: | Re: e1000 (?) jumbo frames performance issue, Michael Iatrou |
| Previous by Thread: | Re: [PATCH] TSO Reloaded, Anton Blanchard |
| Next by Thread: | Re: [PATCH] TSO Reloaded, Anton Blanchard |
| Indexes: | [Date] [Thread] [Top] [All Lists] |