[PATCH] cls_u32 u32_classify()

[PATCH] cls_u32 u32_classify()

[Dzianis Kahanovich]

[-- Attachment #1: Type: text/plain, Size: 184 bytes --]

Currently fine u32 "hashkey ... at ..." not work with relative offsets.
There are simpliest fix to use "eat".

-- 
WBR,
Denis Kaganovich,  mahatma@eu.by  http://mahatma.bspu.unibel.by

[-- Attachment #2: cls_u32.diff --]
[-- Type: text/plain, Size: 985 bytes --]

diff -pruN linux-2.6.orig/net/sched/cls_u32.c linux-2.6/net/sched/cls_u32.c
--- linux-2.6.orig/net/sched/cls_u32.c	2008-01-29 23:02:50.000000000 +0200
+++ linux-2.6/net/sched/cls_u32.c	2008-01-30 10:56:11.000000000 +0200
@@ -181,11 +181,13 @@ check_terminal:
 
 		ht = n->ht_down;
 		sel = 0;
-		if (ht->divisor)
-			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
 
-		if (!(n->sel.flags&(TC_U32_VAROFFSET|TC_U32_OFFSET|TC_U32_EAT)))
+		if (!(n->sel.flags&(TC_U32_VAROFFSET|TC_U32_OFFSET|TC_U32_EAT))) {
+			if (!ht->divisor)
+				goto next_ht;
+			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
 			goto next_ht;
+		}
 
 		if (n->sel.flags&(TC_U32_OFFSET|TC_U32_VAROFFSET)) {
 			off2 = n->sel.off + 3;
@@ -198,6 +200,9 @@ check_terminal:
 			off2 = 0;
 		}
 
+		if (ht->divisor)
+			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
+
 		if (ptr < skb_tail_pointer(skb))
 			goto next_ht;
 	}

[PATCH] cls_u32 u32_classify() +

[Dzianis Kahanovich]

[-- Attachment #1: Type: text/plain, Size: 198 bytes --]

Currently fine u32 "hashkey ... at ..." not work with relative offsets.
There are simpliest fix to use "eat".
(sorry, v2)

-- 
WBR,
Denis Kaganovich,  mahatma@eu.by  http://mahatma.bspu.unibel.by



[-- Attachment #2: cls_u32.diff --]
[-- Type: text/plain, Size: 1029 bytes --]

diff -pruN linux-2.6.orig/net/sched/cls_u32.c linux-2.6/net/sched/cls_u32.c
--- linux-2.6.orig/net/sched/cls_u32.c	2008-01-29 23:02:50.000000000 +0200
+++ linux-2.6/net/sched/cls_u32.c	2008-01-30 11:28:00.000000000 +0200
@@ -181,11 +181,13 @@ check_terminal:
 
 		ht = n->ht_down;
 		sel = 0;
-		if (ht->divisor)
-			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
 
-		if (!(n->sel.flags&(TC_U32_VAROFFSET|TC_U32_OFFSET|TC_U32_EAT)))
+		if (!(n->sel.flags&(TC_U32_VAROFFSET|TC_U32_OFFSET|TC_U32_EAT))) {
+			if (!ht->divisor)
+				goto next_ht;
+			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
 			goto next_ht;
+		}
 
 		if (n->sel.flags&(TC_U32_OFFSET|TC_U32_VAROFFSET)) {
 			off2 = n->sel.off + 3;
@@ -198,6 +200,9 @@ check_terminal:
 			off2 = 0;
 		}
 
+		if (ht->divisor && ptr+n->sel.hoff < skb_tail_pointer(skb))
+			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
+
 		if (ptr < skb_tail_pointer(skb))
 			goto next_ht;
 	}

[PATCH] cls_u32 u32_classify() ++

[Dzianis Kahanovich]

[-- Attachment #1: Type: text/plain, Size: 207 bytes --]

Currently fine u32 "hashkey ... at ..." not work with relative offsets.
There are simpliest fix to use "eat".
(sorry, i'm idiot)

-- 
WBR,
Denis Kaganovich,  mahatma@eu.by  http://mahatma.bspu.unibel.by





[-- Attachment #2: cls_u32.diff --]
[-- Type: text/plain, Size: 1031 bytes --]

diff -pruN linux-2.6.orig/net/sched/cls_u32.c linux-2.6/net/sched/cls_u32.c
--- linux-2.6.orig/net/sched/cls_u32.c	2008-01-29 23:02:50.000000000 +0200
+++ linux-2.6/net/sched/cls_u32.c	2008-01-30 11:37:31.000000000 +0200
@@ -181,11 +181,13 @@ check_terminal:
 
 		ht = n->ht_down;
 		sel = 0;
-		if (ht->divisor)
-			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
 
-		if (!(n->sel.flags&(TC_U32_VAROFFSET|TC_U32_OFFSET|TC_U32_EAT)))
+		if (!(n->sel.flags&(TC_U32_VAROFFSET|TC_U32_OFFSET|TC_U32_EAT))) {
+			if (!ht->divisor)
+				goto next_ht;
+			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
 			goto next_ht;
+		}
 
 		if (n->sel.flags&(TC_U32_OFFSET|TC_U32_VAROFFSET)) {
 			off2 = n->sel.off + 3;
@@ -198,6 +200,9 @@ check_terminal:
 			off2 = 0;
 		}
 
+		if (ht->divisor && ptr+n->sel.hoff+4 < skb_tail_pointer(skb))
+			sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel,n->fshift);
+
 		if (ptr < skb_tail_pointer(skb))
 			goto next_ht;
 	}

Re: [PATCH] cls_u32 u32_classify() +

[jamal]

On Wed, 2008-30-01 at 11:31 -0200, Dzianis Kahanovich wrote:
> Currently fine u32 "hashkey ... at ..." not work with relative offsets.
> There are simpliest fix to use "eat".
> (sorry, v2)
> 

Hi, 
Please send me the commands you are trying to run that motivated this
patch.

cheers,
jamal

Re: [PATCH] cls_u32 u32_classify()

[David Miller]

From: Dzianis Kahanovich <mahatma@bspu.unibel.by>
Date: Wed, 30 Jan 2008 11:16:30 -0200

> Currently fine u32 "hashkey ... at ..." not work with relative offsets.
> There are simpliest fix to use "eat".

So the question is whether 'sel' is defined to be calculated
before all offsets and EAT operations are processed or before.

I do not understand the U32 classifier enough to know what
this kind of change might or might not break.

Can some u32 expert review this?

Thanks.

Re: [PATCH] cls_u32 u32_classify()

[jamal]

On Mon, 2008-18-02 at 21:46 -0800, David Miller wrote:


> Can some u32 expert review this?

http://marc.info/?l=linux-netdev&m=120178638323045&w=2

cheers,
jamal

Re: [PATCH] cls_u32 u32_classify()

[David Miller]

From: jamal <hadi@cyberus.ca>
Date: Tue, 19 Feb 2008 06:54:48 -0500

> On Mon, 2008-18-02 at 21:46 -0800, David Miller wrote:
> 
> 
> > Can some u32 expert review this?
> 
> http://marc.info/?l=linux-netdev&m=120178638323045&w=2

Fair enough.

Dzianis, please provide the feedback Jamal is asking for.