015-mac80211-tx-info-skb-cb.patch

015-mac80211-tx-info-skb-cb.patch

[Ivo van Doorn]

Hi Johannes,

The following patch is intended to be ontop or integrated into your patch:
http://johannes.sipsolutions.net/patches/kernel/all/LATEST/015-mac80211-tx-info-skb-cb.patch
This fixes a compilation error when mac80211 debug output is disabled.

Ivo
---
diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
index 3b3592f..50edc9d 100644
--- a/net/mac80211/tx.c
+++ b/net/mac80211/tx.c
@@ -223,10 +223,9 @@ static ieee80211_tx_result
 ieee80211_tx_h_check_assoc(struct ieee80211_tx_data *tx)
 {
 #ifdef CONFIG_MAC80211_VERBOSE_DEBUG
-	struct sk_buff *skb = tx->skb;
-	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *) skb->data;
+	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *) tx->skb->data;
 #endif /* CONFIG_MAC80211_VERBOSE_DEBUG */
-	struct ieee80211_tx_info *info = (void *)skb->cb;
+	struct ieee80211_tx_info *info = (void *)tx->skb->cb;
 	u32 sta_flags;
 
 	if (unlikely(info->flags & IEEE80211_TX_CTL_INJECTED))

Re: 015-mac80211-tx-info-skb-cb.patch

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1064 bytes --]

On Thu, 2008-05-08 at 15:03 +0200, Ivo van Doorn wrote:
> Hi Johannes,
> 
> The following patch is intended to be ontop or integrated into your patch:
> http://johannes.sipsolutions.net/patches/kernel/all/LATEST/015-mac80211-tx-info-skb-cb.patch
> This fixes a compilation error when mac80211 debug output is disabled.
> 
> Ivo
> ---
> diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
> index 3b3592f..50edc9d 100644
> --- a/net/mac80211/tx.c
> +++ b/net/mac80211/tx.c
> @@ -223,10 +223,9 @@ static ieee80211_tx_result
>  ieee80211_tx_h_check_assoc(struct ieee80211_tx_data *tx)
>  {
>  #ifdef CONFIG_MAC80211_VERBOSE_DEBUG
> -	struct sk_buff *skb = tx->skb;
> -	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *) skb->data;
> +	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *) tx->skb->data;
>  #endif /* CONFIG_MAC80211_VERBOSE_DEBUG */
> -	struct ieee80211_tx_info *info = (void *)skb->cb;
> +	struct ieee80211_tx_info *info = (void *)tx->skb->cb;

Did you send me that before? I seem to have a similar fix already.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: 015-mac80211-tx-info-skb-cb.patch

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 602 bytes --]


> No I haven't, I had been compiling mac80211 with VERBOSE_DEBUG on ever
> since I added the patches from your tree to rt2x00.git so I never noticed this
> bug. I just received the bugreport from somebody at the rt2x00 forum. :)

Then somebody else told me. Or I found it in your tree :)

> Since rt2x00.git contains your patches from a week ago, I did check if the
> bug was still present in your latest version.
> Anyway the important thing is that you have the fix. ;)

Right :) I'm pretty much done with the skb->cb conversion too except for
ath5k (nick) and rt2x00 (you)

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: 015-mac80211-tx-info-skb-cb.patch

[Ivo van Doorn]

On Thursday 08 May 2008, Johannes Berg wrote:
> On Thu, 2008-05-08 at 15:03 +0200, Ivo van Doorn wrote:
> > Hi Johannes,
> > 
> > The following patch is intended to be ontop or integrated into your patch:
> > http://johannes.sipsolutions.net/patches/kernel/all/LATEST/015-mac80211-tx-info-skb-cb.patch
> > This fixes a compilation error when mac80211 debug output is disabled.
> > 
> > Ivo
> > ---
> > diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
> > index 3b3592f..50edc9d 100644
> > --- a/net/mac80211/tx.c
> > +++ b/net/mac80211/tx.c
> > @@ -223,10 +223,9 @@ static ieee80211_tx_result
> >  ieee80211_tx_h_check_assoc(struct ieee80211_tx_data *tx)
> >  {
> >  #ifdef CONFIG_MAC80211_VERBOSE_DEBUG
> > -	struct sk_buff *skb = tx->skb;
> > -	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *) skb->data;
> > +	struct ieee80211_hdr *hdr = (struct ieee80211_hdr *) tx->skb->data;
> >  #endif /* CONFIG_MAC80211_VERBOSE_DEBUG */
> > -	struct ieee80211_tx_info *info = (void *)skb->cb;
> > +	struct ieee80211_tx_info *info = (void *)tx->skb->cb;
> 
> Did you send me that before? I seem to have a similar fix already.

No I haven't, I had been compiling mac80211 with VERBOSE_DEBUG on ever
since I added the patches from your tree to rt2x00.git so I never noticed this
bug. I just received the bugreport from somebody at the rt2x00 forum. :)

Since rt2x00.git contains your patches from a week ago, I did check if the
bug was still present in your latest version.
Anyway the important thing is that you have the fix. ;)

Ivo

Re: 015-mac80211-tx-info-skb-cb.patch

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 871 bytes --]


> > Right :) I'm pretty much done with the skb->cb conversion too except for
> > ath5k (nick) and rt2x00 (you)
> 
> I'm going to rebase rt2x00.git this weekend to update from wireless-testing,
> since the skb->cb patches now depends on HW crypto which is not ready
> for upstream yet, I am going to reverse the patch order so the skb->cb patches
> and some other updates can go upstream more easily.

Cool, thanks. If you have a single patch that can pretty much go into
wireless-testing you can send it to me and I'll roll it into my patch.
If it depends on another rt2x00 merge then just let me know, I can wait
with this. I'd like to do it soonish though, it removes ~200 lines of
code (touching roughly 1000) and a lot of kmemdup/memcpy/kmalloc calls.

I'm just rebasing things a bit myself as well so I can give people a
patch to test.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: 015-mac80211-tx-info-skb-cb.patch

[Ivo van Doorn]

On Thursday 08 May 2008, Johannes Berg wrote:
> 
> > No I haven't, I had been compiling mac80211 with VERBOSE_DEBUG on ever
> > since I added the patches from your tree to rt2x00.git so I never noticed this
> > bug. I just received the bugreport from somebody at the rt2x00 forum. :)
> 
> Then somebody else told me. Or I found it in your tree :)

ok. :)

> > Since rt2x00.git contains your patches from a week ago, I did check if the
> > bug was still present in your latest version.
> > Anyway the important thing is that you have the fix. ;)
> 
> Right :) I'm pretty much done with the skb->cb conversion too except for
> ath5k (nick) and rt2x00 (you)

I'm going to rebase rt2x00.git this weekend to update from wireless-testing,
since the skb->cb patches now depends on HW crypto which is not ready
for upstream yet, I am going to reverse the patch order so the skb->cb patches
and some other updates can go upstream more easily.

Ivo

Re: 015-mac80211-tx-info-skb-cb.patch

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1168 bytes --]

On Thu, 2008-05-08 at 17:34 +0200, Ivo van Doorn wrote:
> On Thursday 08 May 2008, Johannes Berg wrote:
> > 
> > > > Right :) I'm pretty much done with the skb->cb conversion too except for
> > > > ath5k (nick) and rt2x00 (you)
> > > 
> > > I'm going to rebase rt2x00.git this weekend to update from wireless-testing,
> > > since the skb->cb patches now depends on HW crypto which is not ready
> > > for upstream yet, I am going to reverse the patch order so the skb->cb patches
> > > and some other updates can go upstream more easily.
> > 
> > Cool, thanks. If you have a single patch that can pretty much go into
> > wireless-testing you can send it to me and I'll roll it into my patch.
> 
> Well at the moment I have it split into 2 patches, one that is only
> preperation work by splitting a particular function to guarentee the skb->cb
> rules are being followed. And the second patch is the actual conversion.
> Those 2 can easily be merged into a single patch which you can roll into your
> patch.

Whatever you like better. Maybe it's easier to follow the code/patch
flow if your preparation patch goes in first on its own?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: 015-mac80211-tx-info-skb-cb.patch

[Ivo van Doorn]

On Thursday 08 May 2008, Johannes Berg wrote:
> 
> > > Right :) I'm pretty much done with the skb->cb conversion too except for
> > > ath5k (nick) and rt2x00 (you)
> > 
> > I'm going to rebase rt2x00.git this weekend to update from wireless-testing,
> > since the skb->cb patches now depends on HW crypto which is not ready
> > for upstream yet, I am going to reverse the patch order so the skb->cb patches
> > and some other updates can go upstream more easily.
> 
> Cool, thanks. If you have a single patch that can pretty much go into
> wireless-testing you can send it to me and I'll roll it into my patch.

Well at the moment I have it split into 2 patches, one that is only
preperation work by splitting a particular function to guarentee the skb->cb
rules are being followed. And the second patch is the actual conversion.
Those 2 can easily be merged into a single patch which you can roll into your
patch.

> If it depends on another rt2x00 merge then just let me know, I can wait
> with this. I'd like to do it soonish though, it removes ~200 lines of
> code (touching roughly 1000) and a lot of kmemdup/memcpy/kmalloc calls.

Hehe, well I am kind of hoping the patch can be in included soon as well,
(as well as your QOS patches). Unwinding the HW crypto dependency shouldn't
be that hard, after which the patch won't have any dependencies anymore.

> I'm just rebasing things a bit myself as well so I can give people a
> patch to test.

Ivo