Regression in 2.6.27-git from commit 43ac2ca3840f64

Regression in 2.6.27-git from commit 43ac2ca3840f64

[Larry Finger]

On my x86_64 system, the mainline 2.6.27-git kernel (v2.6.27-6030-g6da0b38)
shows the following anomaly with the 'iwlist scan' command:

          Cell 03 - Address: 00:14:BF:85:49:FA
                    ESSID:"lwfdjf"
                    Mode:Master
                    Channel:1
                    Frequency:2.412 GHz (Channel 1)
                    Quality=66/100  Signal level:-62 dBm  Noise level=-68 dBm
                    Encryption key:on
                    IE: Unknown: 00066C7766646A66
                    IE: Unknown: 010882848B962430486C
                    IE: Unknown: 030101
                    IE: Unknown: 2A0104
                    IE: Unknown: 2F0104
                    IE: Unknown: 32040C121860
                    IE: Unknown: DD09001018020014000000
                    IE: WPA Version 1
                        Group Cipher : TKIP
                        Pairwise Ciphers (1) : TKIP
                        Authentication Suites (1) : PSK
                    Bit Rates:1 Mb/s; 2 Mb/s; 5.5 Mb/s; 11 Mb/s; 18 Mb/s
                              24 Mb/s; 36 Mb/s; 54 Mb/s; 6 Mb/s; 9 Mb/s
                              12 Mb/s; 48 Mb/s
                    Extra:tsf=0000014488aeb18b
                    Extra: Last beacon: 448ms ago

This problem was bisected to commit 43ac2ca3840f64f699a239535c590fa7ebaaac27,
entitled "mac80211: Handle scan result IEs in one block". The same bug is
present in the wireless-testing tree.

Larry

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1659 bytes --]

On Tue, 2008-10-21 at 09:01 -0700, Larry Finger wrote:
> On my x86_64 system, the mainline 2.6.27-git kernel (v2.6.27-6030-g6da0b38)
> shows the following anomaly with the 'iwlist scan' command:
> 
>           Cell 03 - Address: 00:14:BF:85:49:FA
>                     ESSID:"lwfdjf"
>                     Mode:Master
>                     Channel:1
>                     Frequency:2.412 GHz (Channel 1)
>                     Quality=66/100  Signal level:-62 dBm  Noise level=-68 dBm
>                     Encryption key:on
>                     IE: Unknown: 00066C7766646A66
>                     IE: Unknown: 010882848B962430486C
>                     IE: Unknown: 030101
>                     IE: Unknown: 2A0104
>                     IE: Unknown: 2F0104
>                     IE: Unknown: 32040C121860
>                     IE: Unknown: DD09001018020014000000
>                     IE: WPA Version 1
>                         Group Cipher : TKIP
>                         Pairwise Ciphers (1) : TKIP
>                         Authentication Suites (1) : PSK
>                     Bit Rates:1 Mb/s; 2 Mb/s; 5.5 Mb/s; 11 Mb/s; 18 Mb/s
>                               24 Mb/s; 36 Mb/s; 54 Mb/s; 6 Mb/s; 9 Mb/s
>                               12 Mb/s; 48 Mb/s
>                     Extra:tsf=0000014488aeb18b
>                     Extra: Last beacon: 448ms ago
> 
> This problem was bisected to commit 43ac2ca3840f64f699a239535c590fa7ebaaac27,
> entitled "mac80211: Handle scan result IEs in one block". The same bug is
> present in the wireless-testing tree.

I'm probably too tired, what exactly is the problem?

johannes

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

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Larry Finger]

Johannes Berg wrote:
> On Tue, 2008-10-21 at 09:01 -0700, Larry Finger wrote:
>> On my x86_64 system, the mainline 2.6.27-git kernel (v2.6.27-6030-g6da0b38)
>> shows the following anomaly with the 'iwlist scan' command:
>>
>>           Cell 03 - Address: 00:14:BF:85:49:FA
>>                     ESSID:"lwfdjf"
>>                     Mode:Master
>>                     Channel:1
>>                     Frequency:2.412 GHz (Channel 1)
>>                     Quality=66/100  Signal level:-62 dBm  Noise level=-68 dBm
>>                     Encryption key:on
>>                     IE: Unknown: 00066C7766646A66
>>                     IE: Unknown: 010882848B962430486C
>>                     IE: Unknown: 030101
>>                     IE: Unknown: 2A0104
>>                     IE: Unknown: 2F0104
>>                     IE: Unknown: 32040C121860
>>                     IE: Unknown: DD09001018020014000000
>>                     IE: WPA Version 1
>>                         Group Cipher : TKIP
>>                         Pairwise Ciphers (1) : TKIP
>>                         Authentication Suites (1) : PSK
>>                     Bit Rates:1 Mb/s; 2 Mb/s; 5.5 Mb/s; 11 Mb/s; 18 Mb/s
>>                               24 Mb/s; 36 Mb/s; 54 Mb/s; 6 Mb/s; 9 Mb/s
>>                               12 Mb/s; 48 Mb/s
>>                     Extra:tsf=0000014488aeb18b
>>                     Extra: Last beacon: 448ms ago
>>
>> This problem was bisected to commit 43ac2ca3840f64f699a239535c590fa7ebaaac27,
>> entitled "mac80211: Handle scan result IEs in one block". The same bug is
>> present in the wireless-testing tree.
> 
> I'm probably too tired, what exactly is the problem?

The unknown IE's don't show up before this commit.

Larry

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 368 bytes --]

On Tue, 2008-10-21 at 19:34 -0700, Larry Finger wrote:

> >>                     IE: Unknown: 00066C7766646A66

> The unknown IE's don't show up before this commit.

Oh, ok, but that's just more information shown to userspace, that was
intentional, just iwlist doesn't understand how to parse them, for
example the one I quoted above is the SSID.

johannes

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

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[John W. Linville]

On Tue, Oct 21, 2008 at 09:01:14AM -0700, Larry Finger wrote:
> On my x86_64 system, the mainline 2.6.27-git kernel (v2.6.27-6030-g6da0b38)
> shows the following anomaly with the 'iwlist scan' command:
> 
>           Cell 03 - Address: 00:14:BF:85:49:FA
>                     ESSID:"lwfdjf"
>                     Mode:Master
>                     Channel:1
>                     Frequency:2.412 GHz (Channel 1)
>                     Quality=66/100  Signal level:-62 dBm  Noise level=-68 dBm
>                     Encryption key:on
>                     IE: Unknown: 00066C7766646A66
>                     IE: Unknown: 010882848B962430486C
>                     IE: Unknown: 030101
>                     IE: Unknown: 2A0104
>                     IE: Unknown: 2F0104
>                     IE: Unknown: 32040C121860
>                     IE: Unknown: DD09001018020014000000
>                     IE: WPA Version 1
>                         Group Cipher : TKIP
>                         Pairwise Ciphers (1) : TKIP
>                         Authentication Suites (1) : PSK
>                     Bit Rates:1 Mb/s; 2 Mb/s; 5.5 Mb/s; 11 Mb/s; 18 Mb/s
>                               24 Mb/s; 36 Mb/s; 54 Mb/s; 6 Mb/s; 9 Mb/s
>                               12 Mb/s; 48 Mb/s
>                     Extra:tsf=0000014488aeb18b
>                     Extra: Last beacon: 448ms ago
> 
> This problem was bisected to commit 43ac2ca3840f64f699a239535c590fa7ebaaac27,
> entitled "mac80211: Handle scan result IEs in one block". The same bug is
> present in the wireless-testing tree.

I'm guessing that iwlist doesn't like the packed IEs?  Does this patch help?

John

P.S.  I'm not sure if this qualifies as actual userland breakage or not...

---

From: John W. Linville <linville@tuxdriver.com>
Subject: [PATCH] mac80211: do not aggregate IEs from scan results

Signed-off-by: John W. Linville <linville@tuxdriver.com>
---
 net/mac80211/scan.c |    2 --
 1 files changed, 0 insertions(+), 2 deletions(-)

diff --git a/net/mac80211/scan.c b/net/mac80211/scan.c
index 416bb41..51ed74d 100644
--- a/net/mac80211/scan.c
+++ b/net/mac80211/scan.c
@@ -728,8 +728,6 @@ static void ieee80211_scan_add_ies(struct iw_request_info *info,
 
 	while (end - pos > IW_GENERIC_IE_MAX) {
 		next = pos + 2 + pos[1];
-		while (next + 2 + next[1] - pos < IW_GENERIC_IE_MAX)
-			next = next + 2 + next[1];
 
 		memset(&iwe, 0, sizeof(iwe));
 		iwe.cmd = IWEVGENIE;
-- 
1.5.4.3

-- 
John W. Linville		Linux should be at the core
linville@tuxdriver.com			of your literate lifestyle.

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Larry Finger]

John W. Linville wrote:
> From: John W. Linville <linville@tuxdriver.com>
> Subject: [PATCH] mac80211: do not aggregate IEs from scan results
> 
> Signed-off-by: John W. Linville <linville@tuxdriver.com>
> ---
>  net/mac80211/scan.c |    2 --
>  1 files changed, 0 insertions(+), 2 deletions(-)
> 
> diff --git a/net/mac80211/scan.c b/net/mac80211/scan.c
> index 416bb41..51ed74d 100644
> --- a/net/mac80211/scan.c
> +++ b/net/mac80211/scan.c
> @@ -728,8 +728,6 @@ static void ieee80211_scan_add_ies(struct iw_request_info *info,
>  
>  	while (end - pos > IW_GENERIC_IE_MAX) {
>  		next = pos + 2 + pos[1];
> -		while (next + 2 + next[1] - pos < IW_GENERIC_IE_MAX)
> -			next = next + 2 + next[1];
>  
>  		memset(&iwe, 0, sizeof(iwe));
>  		iwe.cmd = IWEVGENIE;

No, this patch does not fix the problem.

Larry

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Jouni Malinen]

On Tue, Oct 21, 2008 at 09:01:14AM -0700, Larry Finger wrote:

> On my x86_64 system, the mainline 2.6.27-git kernel (v2.6.27-6030-g6da0b38)
> shows the following anomaly with the 'iwlist scan' command:
> 
>           Cell 03 - Address: 00:14:BF:85:49:FA
>                     ESSID:"lwfdjf"
>                     Mode:Master
>                     Channel:1
>                     Frequency:2.412 GHz (Channel 1)
>                     Quality=66/100  Signal level:-62 dBm  Noise level=-68 dBm
>                     Encryption key:on
>                     IE: Unknown: 00066C7766646A66

...

> This problem was bisected to commit 43ac2ca3840f64f699a239535c590fa7ebaaac27,
> entitled "mac80211: Handle scan result IEs in one block". The same bug is
> present in the wireless-testing tree.

Problem? Bug? This is by design. What does not work for you?

-- 
Jouni Malinen                                            PGP id EFC895FA

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Larry Finger]

Jouni Malinen wrote:
> On Tue, Oct 21, 2008 at 09:01:14AM -0700, Larry Finger wrote:
> 
>> On my x86_64 system, the mainline 2.6.27-git kernel (v2.6.27-6030-g6da0b38)
>> shows the following anomaly with the 'iwlist scan' command:
>>
>>           Cell 03 - Address: 00:14:BF:85:49:FA
>>                     ESSID:"lwfdjf"
>>                     Mode:Master
>>                     Channel:1
>>                     Frequency:2.412 GHz (Channel 1)
>>                     Quality=66/100  Signal level:-62 dBm  Noise level=-68 dBm
>>                     Encryption key:on
>>                     IE: Unknown: 00066C7766646A66
> 
> ...
> 
>> This problem was bisected to commit 43ac2ca3840f64f699a239535c590fa7ebaaac27,
>> entitled "mac80211: Handle scan result IEs in one block". The same bug is
>> present in the wireless-testing tree.
> 
> Problem? Bug? This is by design. What does not work for you?

When a utility starts spitting messages that it doesn't understand what the
kernel is providing _AND_ the commit message does not mention that this will
happen, can someone be blamed for thinking it is a bug?

Clearly this is not a regression. Are there plans to update the wireless utilities?

Larry

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Jouni Malinen]

On Wed, Oct 22, 2008 at 09:47:47AM -0700, Larry Finger wrote:

> When a utility starts spitting messages that it doesn't understand what the
> kernel is providing _AND_ the commit message does not mention that this will
> happen, can someone be blamed for thinking it is a bug?

But the utility here understands that it receives additional IEs; it
just does not know the name or format for them. I do not understand why
the commit message should have commented on any particular user space
tool and their behavior as long as nothing is broken (and as far as I'm
aware, nothing was broken by that change).

> Clearly this is not a regression. Are there plans to update the wireless utilities?

Update them (or well, iwlist) to do what? Sure, it could have a list of
some additional known SSIDs, but that does not sound very useful. If the
"IE: Unknown" part is really making this somehow horrible for users, I
would just hide those lines by default (but there should be an option to
enable them). Anyway, I do not have any plans on providing a patch to
change iwlist behavior because I think the current behavior is
desirable.

-- 
Jouni Malinen                                            PGP id EFC895FA

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 604 bytes --]

On Wed, 2008-10-22 at 09:47 -0700, Larry Finger wrote:

> When a utility starts spitting messages that it doesn't understand what the
> kernel is providing _AND_ the commit message does not mention that this will
> happen, can someone be blamed for thinking it is a bug?

Heh, no, not really. It was fairly obvious us, but I guess we were just
routine-blinded.

> Clearly this is not a regression. Are there plans to update the wireless utilities?

I suspect it could just be made to ignore them. Or skip a select few,
but obviously it will never be able to parse all of them...

johannes

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

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Jean Tourrilhes]

On Wed, Oct 22, 2008 at 07:06:29PM +0200, Johannes Berg wrote:
> On Wed, 2008-10-22 at 09:47 -0700, Larry Finger wrote:
> 
> > When a utility starts spitting messages that it doesn't understand what the
> > kernel is providing _AND_ the commit message does not mention that this will
> > happen, can someone be blamed for thinking it is a bug?
> 
> Heh, no, not really. It was fairly obvious us, but I guess we were just
> routine-blinded.
> 
> > Clearly this is not a regression. Are there plans to update the wireless utilities?
> 
> I suspect it could just be made to ignore them. Or skip a select few,
> but obviously it will never be able to parse all of them...
> 
> johannes

	Hi,

	If you want me to understand what it's about, you may want to
give me a bit more context.
	Alternatively, just send me a patch for Wireless Tools, and
I'll put it in the next release. Wireless Tools is still maintained
and I still do releases, even if I don't put much cycles into it.
	Have a great day...

	Jean

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1415 bytes --]

On Wed, 2008-10-22 at 10:37 -0700, Jean Tourrilhes wrote:

> 	If you want me to understand what it's about, you may want to
> give me a bit more context.

Sorry. Basically, Larry is saying that due to a kernel change, iwlist
will print out things like
                    IE: Unknown: 0017465249545A21426F7820466F6E20574C414E2037313133
                    IE: Unknown: 010482848B96
                    IE: Unknown: 030106
                    IE: Unknown: 2A0106
                    IE: IEEE 802.11i/WPA2 Version 1
                        Group Cipher : TKIP
                        Pairwise Ciphers (1) : CCMP
                        Authentication Suites (1) : PSK
                    IE: Unknown: 32080C1218243048606C
                    IE: WPA Version 1
                        Group Cipher : TKIP
                        Pairwise Ciphers (1) : TKIP
                        Authentication Suites (1) : PSK
                    IE: Unknown: DD0A0800280101000200FF0F
                    IE: Unknown: DD180050F2020101800003A4000027A4000042435E0062322F00


where all the IE: Unknown lines are unexpected.

> 	Alternatively, just send me a patch for Wireless Tools, and
> I'll put it in the next release. Wireless Tools is still maintained
> and I still do releases, even if I don't put much cycles into it.

The problem is that we don't quite know what to do about them :)

johannes

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

Re: Regression in 2.6.27-git from commit 43ac2ca3840f64

[Jean Tourrilhes]

On Wed, Oct 22, 2008 at 07:46:27PM +0200, Johannes Berg wrote:
> On Wed, 2008-10-22 at 10:37 -0700, Jean Tourrilhes wrote:
> 
> > 	If you want me to understand what it's about, you may want to
> > give me a bit more context.
> 
> Sorry. Basically, Larry is saying that due to a kernel change, iwlist
> will print out things like
>                     IE: Unknown: 0017465249545A21426F7820466F6E20574C414E2037313133
>                     IE: Unknown: 010482848B96
>                     IE: Unknown: 030106
>                     IE: Unknown: 2A0106
>                     IE: IEEE 802.11i/WPA2 Version 1
>                         Group Cipher : TKIP
>                         Pairwise Ciphers (1) : CCMP
>                         Authentication Suites (1) : PSK
>                     IE: Unknown: 32080C1218243048606C
>                     IE: WPA Version 1
>                         Group Cipher : TKIP
>                         Pairwise Ciphers (1) : TKIP
>                         Authentication Suites (1) : PSK
>                     IE: Unknown: DD0A0800280101000200FF0F
>                     IE: Unknown: DD180050F2020101800003A4000027A4000042435E0062322F00
> 
> 
> where all the IE: Unknown lines are unexpected.

	Unexpected, but not broken. That's good.

> > 	Alternatively, just send me a patch for Wireless Tools, and
> > I'll put it in the next release. Wireless Tools is still maintained
> > and I still do releases, even if I don't put much cycles into it.
> 
> The problem is that we don't quite know what to do about them :)

	We are not supposed to "do" anything, but to show the
underlying data in a user friendly way. I can't believe that we could
not parse this data properly, we manage to parse other IEs
correctly. If you are debugging a problem with wpa_supplicant, or a
Access Point issue, what would *you* want to see there.

> johannes

	Have fun...

	Jean