[PATCH] ath9k: remove fragmentation workaround again

[PATCH] ath9k: remove fragmentation workaround again

[Johannes Berg]

Looking at the RC algorithm this is no longer necessary.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
---
Untested, but the error that used to happen will no longer happen since
now the sub-structure is assigned by the driver, in rate_driver_data,
per fragment.

 drivers/net/wireless/ath9k/main.c |    6 ------
 1 file changed, 6 deletions(-)

--- everything.orig/drivers/net/wireless/ath9k/main.c	2008-12-08 21:54:08.000000000 +0100
+++ everything/drivers/net/wireless/ath9k/main.c	2008-12-08 21:54:20.000000000 +0100
@@ -2423,11 +2423,6 @@ static int ath9k_ampdu_action(struct iee
 	return ret;
 }
 
-static int ath9k_no_fragmentation(struct ieee80211_hw *hw, u32 value)
-{
-	return -EOPNOTSUPP;
-}
-
 static struct ieee80211_ops ath9k_ops = {
 	.tx 		    = ath9k_tx,
 	.start 		    = ath9k_start,
@@ -2444,7 +2439,6 @@ static struct ieee80211_ops ath9k_ops = 
 	.get_tsf 	    = ath9k_get_tsf,
 	.reset_tsf 	    = ath9k_reset_tsf,
 	.ampdu_action       = ath9k_ampdu_action,
-	.set_frag_threshold = ath9k_no_fragmentation,
 };
 
 static struct {

Re: [PATCH] ath9k: remove fragmentation workaround again

[Jouni Malinen]

On Tue, Dec 09, 2008 at 12:26:14PM +0100, Johannes Berg wrote:
> Looking at the RC algorithm this is no longer necessary.

> Untested, but the error that used to happen will no longer happen since
> now the sub-structure is assigned by the driver, in rate_driver_data,
> per fragment.

Thanks! Based on a quick test, fragmentation was working with this
patch. I never tested the old issue, so I'm not sure what exactly was
failing, but anyway, I do see fragments in a sniffer log and the device
did not crash ;-).

-- 
Jouni Malinen                                            PGP id EFC895FA

Re: [PATCH] ath9k: remove fragmentation workaround again

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 726 bytes --]

On Tue, 2008-12-09 at 18:39 +0200, Jouni Malinen wrote:
> On Tue, Dec 09, 2008 at 12:26:14PM +0100, Johannes Berg wrote:
> > Looking at the RC algorithm this is no longer necessary.
> 
> > Untested, but the error that used to happen will no longer happen since
> > now the sub-structure is assigned by the driver, in rate_driver_data,
> > per fragment.
> 
> Thanks! Based on a quick test, fragmentation was working with this
> patch. I never tested the old issue, so I'm not sure what exactly was
> failing, but anyway, I do see fragments in a sniffer log and the device
> did not crash ;-).

Heh, well, the broken rc algorithm would end up with a double-free and
slub/slab/slob being very unhappy.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]