From: Patrick McHardy <kaber@trash.net>
To: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Kernel Netdev Mailing List <netdev@vger.kernel.org>
Subject: Re: [XFRM]: Improve MTU estimation
Date: Fri, 04 Aug 2006 13:51:24 +0200 [thread overview]
Message-ID: <44D334BC.8040003@trash.net> (raw)
In-Reply-To: <44D33477.2060803@trash.net>
[-- Attachment #1: Type: text/plain, Size: 233 bytes --]
Patrick McHardy wrote:
> Now I get it, thanks :) I missed that the IP header isn't part of the
> length when it is aligned. So the worst-case increases by block-size
> - 4 (- 8 for IPv6). How does this look?
Forgot the patch ..
[-- Attachment #2: x --]
[-- Type: text/plain, Size: 4965 bytes --]
diff --git a/include/net/xfrm.h b/include/net/xfrm.h
index 9c5ee9f..ea1b028 100644
--- a/include/net/xfrm.h
+++ b/include/net/xfrm.h
@@ -262,8 +262,7 @@ struct xfrm_type
void (*destructor)(struct xfrm_state *);
int (*input)(struct xfrm_state *, struct sk_buff *skb);
int (*output)(struct xfrm_state *, struct sk_buff *pskb);
- /* Estimate maximal size of result of transformation of a dgram */
- u32 (*get_max_size)(struct xfrm_state *, int size);
+ u32 (*get_mtu)(struct xfrm_state *, int size);
};
extern int xfrm_register_type(struct xfrm_type *type, unsigned short family);
diff --git a/net/ipv4/esp4.c b/net/ipv4/esp4.c
index fc2f8ce..1ce06cc 100644
--- a/net/ipv4/esp4.c
+++ b/net/ipv4/esp4.c
@@ -251,21 +251,23 @@ out:
return -EINVAL;
}
-static u32 esp4_get_max_size(struct xfrm_state *x, int mtu)
+static u32 esp4_get_mtu(struct xfrm_state *x, int mtu)
{
struct esp_data *esp = x->data;
u32 blksize = ALIGN(crypto_tfm_alg_blocksize(esp->conf.tfm), 4);
+ u32 align;
- if (x->props.mode) {
- mtu = ALIGN(mtu + 2, blksize);
- } else {
- /* The worst case. */
- mtu = ALIGN(mtu + 2, 4) + blksize - 4;
- }
- if (esp->conf.padlen)
- mtu = ALIGN(mtu, esp->conf.padlen);
+ align = blksize;
+ if (esp->conf.padlen > align)
+ align = esp->conf.padlen;
- return mtu + x->props.header_len + esp->auth.icv_trunc_len;
+ mtu -= x->props.header_len + esp->auth.icv_trunc_len;
+ mtu &= ~(align - 1);
+ mtu -= 2;
+ if (!x->props.mode)
+ mtu -= blksize - 4;
+
+ return mtu;
}
static void esp4_err(struct sk_buff *skb, u32 info)
@@ -307,6 +309,7 @@ static void esp_destroy(struct xfrm_stat
static int esp_init_state(struct xfrm_state *x)
{
struct esp_data *esp = NULL;
+ u32 align;
/* null auth and encryption can have zero length keys */
if (x->aalg) {
@@ -385,7 +388,10 @@ static int esp_init_state(struct xfrm_st
}
}
x->data = esp;
- x->props.trailer_len = esp4_get_max_size(x, 0) - x->props.header_len;
+ align = ALIGN(crypto_tfm_alg_blocksize(esp->conf.tfm), 4);
+ if (esp->conf.padlen)
+ align = ALIGN(align, esp->conf.padlen);
+ x->props.trailer_len = align - 1 + esp->auth.icv_trunc_len;
return 0;
error:
@@ -402,7 +408,7 @@ static struct xfrm_type esp_type =
.proto = IPPROTO_ESP,
.init_state = esp_init_state,
.destructor = esp_destroy,
- .get_max_size = esp4_get_max_size,
+ .get_mtu = esp4_get_mtu,
.input = esp_input,
.output = esp_output
};
diff --git a/net/ipv6/esp6.c b/net/ipv6/esp6.c
index a278d5e..0c9c159 100644
--- a/net/ipv6/esp6.c
+++ b/net/ipv6/esp6.c
@@ -222,22 +222,23 @@ out:
return ret;
}
-static u32 esp6_get_max_size(struct xfrm_state *x, int mtu)
+static u32 esp6_get_mtu(struct xfrm_state *x, int mtu)
{
struct esp_data *esp = x->data;
u32 blksize = ALIGN(crypto_tfm_alg_blocksize(esp->conf.tfm), 4);
-
- if (x->props.mode) {
- mtu = ALIGN(mtu + 2, blksize);
- } else {
- /* The worst case. */
- u32 padsize = ((blksize - 1) & 7) + 1;
- mtu = ALIGN(mtu + 2, padsize) + blksize - padsize;
- }
- if (esp->conf.padlen)
- mtu = ALIGN(mtu, esp->conf.padlen);
-
- return mtu + x->props.header_len + esp->auth.icv_trunc_len;
+ u32 align;
+
+ align = blksize;
+ if (esp->conf.padlen > align)
+ align = esp->conf.padlen;
+
+ mtu -= x->props.header_len + esp->auth.icv_trunc_len;
+ mtu &= ~(align - 1);
+ mtu -= 2;
+ if (!x->props.mode && blksize >= 8)
+ mtu -= blksize - 8;
+
+ return mtu;
}
static void esp6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
@@ -363,7 +364,7 @@ static struct xfrm_type esp6_type =
.proto = IPPROTO_ESP,
.init_state = esp6_init_state,
.destructor = esp6_destroy,
- .get_max_size = esp6_get_max_size,
+ .get_mtu = esp6_get_mtu,
.input = esp6_input,
.output = esp6_output
};
diff --git a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c
index 0021aad..7b0b100 100644
--- a/net/xfrm/xfrm_state.c
+++ b/net/xfrm/xfrm_state.c
@@ -1129,37 +1129,20 @@ void xfrm_state_delete_tunnel(struct xfr
}
EXPORT_SYMBOL(xfrm_state_delete_tunnel);
-/*
- * This function is NOT optimal. For example, with ESP it will give an
- * MTU that's usually two bytes short of being optimal. However, it will
- * usually give an answer that's a multiple of 4 provided the input is
- * also a multiple of 4.
- */
int xfrm_state_mtu(struct xfrm_state *x, int mtu)
{
- int res = mtu;
-
- res -= x->props.header_len;
-
- for (;;) {
- int m = res;
-
- if (m < 68)
- return 68;
-
- spin_lock_bh(&x->lock);
- if (x->km.state == XFRM_STATE_VALID &&
- x->type && x->type->get_max_size)
- m = x->type->get_max_size(x, m);
- else
- m += x->props.header_len;
- spin_unlock_bh(&x->lock);
+ int res;
- if (m <= mtu)
- break;
- res -= (m - mtu);
- }
+ spin_lock_bh(&x->lock);
+ if (x->km.state == XFRM_STATE_VALID &&
+ x->type && x->type->get_mtu)
+ res = x->type->get_mtu(x, mtu);
+ else
+ res = mtu - x->props.header_len;
+ spin_unlock_bh(&x->lock);
+ if (res < 68)
+ res = 68;
return res;
}
next prev parent reply other threads:[~2006-08-04 11:53 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-08-04 8:50 [XFRM]: Improve MTU estimation Patrick McHardy
2006-08-04 9:27 ` Patrick McHardy
2006-08-04 10:01 ` Herbert Xu
2006-08-04 10:09 ` Patrick McHardy
2006-08-04 10:13 ` Herbert Xu
2006-08-04 11:11 ` Patrick McHardy
2006-08-04 11:16 ` Herbert Xu
2006-08-04 11:21 ` Patrick McHardy
2006-08-04 11:25 ` Herbert Xu
2006-08-04 11:50 ` Patrick McHardy
2006-08-04 11:51 ` Patrick McHardy [this message]
2006-08-04 11:55 ` Herbert Xu
2006-08-04 11:54 ` Herbert Xu
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=44D334BC.8040003@trash.net \
--to=kaber@trash.net \
--cc=herbert@gondor.apana.org.au \
--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).