On Thu, 2011-12-22 at 18:20 +0000, David Miller wrote:

From: Ian Campbell <redacted>
Date: Thu, 22 Dec 2011 10:33:36 +0000

quoted

On Wed, 2011-12-21 at 19:28 +0000, David Miller wrote:

quoted

From: Eric Dumazet <redacted>
Date: Wed, 21 Dec 2011 15:02:18 +0100

quoted

No idea on this +2 point.

I think I know, and I believe I instructed Alexey Kuznetsov to do
this.

When sendfile() is performed, we might start the SKB with the last few
bytes of one page, and end the SKB with the first few bytes of another
page.

In order to fit a full 64K frame into an SKB in this situation we have
to accomodate this case.

Thanks David, that makes sense.

However I think you only actually need 1 extra page for that. If the
data in frag[0] starts at $offset then frag[16] will need to have
$offset bytes in it. e.g.
	4096-$offset + 4096*15 + $offset = 65536
which == 17 pages rather than 18.

The following documents the status quo but I could update to switch to +
1 instead if there are no flaws in the above logic...

Indeed, you're right.  Please change this to 1 and document it, and we
can put that change into net-next, thanks a lot!

Please see below.

This change actually makes the size increase due to adding the
destructor small enough that a maximum MTU frame + shared_info still
fits in 2048 bytes (just!). I think I can see some scope for shaving off
a few more bytes as well but that'll have to wait for next year.

8<-----------------------------------------

From 75d3ee297af89b7f801bb0a13517f51181ef4d3c Mon Sep 17 00:00:00 2001

From: Ian Campbell <redacted>
Date: Thu, 22 Dec 2011 10:07:19 +0000
Subject: [PATCH] net: document reason for 2 additional pages in MAX_SKB_FRAGS

Signed-off-by: Ian Campbell <redacted>
---
 include/linux/skbuff.h |   12 ++++++++----
 1 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index fe86488..0592b3d 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h

@@ -128,13 +128,17 @@ struct sk_buff_head {
 
 struct sk_buff;
 
-/* To allow 64K frame to be packed as single skb without frag_list. Since
- * GRO uses frags we allocate at least 16 regardless of page size.
+/* To allow 64K frame to be packed as single skb without frag_list we
+ * require 64K/PAGE_SIZE pages plus 1 additional page to allow for
+ * buffers which do not start on a page boundary.
+ *
+ * Since GRO uses frags we allocate at least 16 regardless of page
+ * size.
  */
-#if (65536/PAGE_SIZE + 2) < 16
+#if (65536/PAGE_SIZE + 1) < 16
 #define MAX_SKB_FRAGS 16UL
 #else
-#define MAX_SKB_FRAGS (65536/PAGE_SIZE + 2)
+#define MAX_SKB_FRAGS (65536/PAGE_SIZE + 1)
 #endif
 
 typedef struct skb_frag_struct skb_frag_t;

-- 
1.7.2.5