2.6.33-rc1 went out yesterday so today we get the respective backport
code now with 802.11 and bluetooth combined on the compat-wireless
package [1]. At this point only bug fixes are welcomed for that
package. At the last minute I decided to add Atheros Ethernet drivers
atlx, atl1e and atl1c. For wireless specific updates refer to wireless
2.6.33-rc1 ChangeLog [2]. If there are any bugs please report them
through linux-wireless. I've tested ath5k on 2.6.27.41 on x86_64 and
send this e-mail out through it. I should note
./scripts/drivers-select is busted as I never got a chance to update
it based on the latest changes.
While at it 2.6.32.2 got released today so I just updated a release
for that so users can benefit from the few fixes [3], you can check
the ChangeLog for that on the wireless for details on the changes [4].
For more details and sha1 checksums you can refer to the wireless wiki
compat-wireless stable page [5].
[1] http://www.orbit-lab.org/kernel/compat-wireless-2.6-stable/v2.6.33/compat-wireless-2.6.33-rc1.tar.bz2
[2] http://www.orbit-lab.org/kernel/compat-wireless-2.6-stable/v2.6.33/ChangeLog-2.6.33-rc1-wireless
[3] http://www.orbit-lab.org/kernel/compat-wireless-2.6-stable/v2.6.32/compat-wireless-2.6.32.2.tar.bz2
[4] http://www.orbit-lab.org/kernel/compat-wireless-2.6-stable/v2.6.32/ChangeLog-2.6.32.2-wireless
[5] http://wireless.kernel.org/en/users/Download/stable
Luis
On Mon, Dec 21, 2009 at 9:36 AM, Corey Wright <[email protected]> wrote:
> (my apologies to those with thread-aware mail user agents as this is not a
> true reply as i'm not subscribed to linux-wireless, but copied from
> marc.info's raw message display.)
>
>> If there are any bugs please report them
>> through linux-wireless.
>
>> I've tested ath5k on 2.6.27.41 on x86_64 and
>> send this e-mail out through it.
>
>> While at it 2.6.32.2 got released today so I just updated a release
>> for that so users can benefit from the few fixes [3], you can check
>> the ChangeLog for that on the wireless for details on the changes [4].
>
> i tested ath5k from compat-wireless-2.6.32 on 2.6.27.41 and from
> compat-wireless-2.6.32.2 on 2.6.27.42 and i needed this patch [1]
> to avoid duplicate ieee80211_rx symbol in ieee80211 (from kernel) &
> mac80211 (from compat-wireless).
>
> [1] http://patchwork.kernel.org/patch/54569/
>
> i see that the patch was applied to compat-wireless-2.6.33-rc1, but i
> wonder if it should be backported to the 2.6.32.2 release.
I'll suck it in, thanks for the report.
> actually, i have no personal stake in the fix as after i got it to work i
> learned that the new release is not compatible with my older distro-provided
> user-land (doesn't recognize it as wireless),
Old userspace should not bust. If you find what the issue was let us
know. AFAICT old userspace tend to look for 3 possible things to see
if a device is a wireless device.
1) Wireless-tools requires SIOCGIWRANGE
2) Network manager requires SIOCGIWNAME
3) Linux hald < 0.5.12 used to use check the existance of
/sys/class/net/<interface>/wireless/
The only one I can think of your userspace might have an issue with is
the last one but even then you should get this if you have
CONFIG_WIRELESS_EXT_SYSFS=y on 2.6.27.41
> which i don't feel like
> upgrading/self-maintaining and possibly descending into dependency-hell, so
> i'm sticking with compat-wireless-2.6.30 (compat-wireless-2.6.31-rc7 release
> didn't seem to work as well as 2.6.30), but thought the finding might be of
> interest to others (possibly stumbling upon this searching for
> "compat-wireless-2.6.32 ieee80211_rx ieee80211 mac80211 duplicate symbol"
> or something similar).
>
> thanks for compat-wireless!
Thanks for reporting your issue.
Luis
On Mon, 21 Dec 2009 10:46:18 -0800
"Luis R. Rodriguez" <[email protected]> wrote:
> > actually, i have no personal stake in the fix as after i got it to work
> > i learned that the new release is not compatible with my older
> > distro-provided user-land (doesn't recognize it as wireless),
>
> Old userspace should not bust. If you find what the issue was let us
> know. AFAICT old userspace tend to look for 3 possible things to see
> if a device is a wireless device.
>
> 1) Wireless-tools requires SIOCGIWRANGE
> 2) Network manager requires SIOCGIWNAME
> 3) Linux hald < 0.5.12 used to use check the existance of
> /sys/class/net/<interface>/wireless/
>
> The only one I can think of your userspace might have an issue with is
> the last one but even then you should get this if you have
> CONFIG_WIRELESS_EXT_SYSFS=y on 2.6.27.41
the primary differences between compat-wireless 2.6.30 & 2.6.32.2 ath5k on
2.6.27.42 (by observation of the resulting environment) is the older
version creates a corresponding wmaster0 interface and the wlan0 sysfs
entry has a wireless directory (ie /sys/class/net/wlan0/wireless).
hal is version 0.5.11rc2 (from "hald --version") and indeed does not
recognize wlan0 as 802.11, but instead 802.03 (from "hal-device").
without digging into the hal source (though i guess i could simply strace
it and see if it tests for the wireless subdirectory), it appears your idea
was correct, but i have CONFIG_WIRELESS_EXT_SYSFS=y.
from my 2.6.27.42 .config (which i compiled & ran both compat-wireless
versions against):
#
# Wireless
#
CONFIG_CFG80211=m
CONFIG_NL80211=y
CONFIG_WIRELESS_EXT=y
CONFIG_WIRELESS_EXT_SYSFS=y
CONFIG_MAC80211=m
if you have any other ideas, patches, etc, then i'll try them. or tell me
that hal < 0.5.12 is unsupported and i'll stick with compat-wireless-2.6.30.
for the record: wireless-tools (iwconfig) recognizes wlan0 as "IEEE
802.11bg" with both compat-wireless versions and NetworkManager only used
SIOC[SG]IW* ioctl requests (eg ESSID, MODE) with the older version (as seen
by strace), so i figure hal reported wlan0 as 802.03 and NetworkManager
believed it (never trying wireless ioctl requests).
thanks for your work (going all the way back to prism54)!
corey
--
[email protected]
On Wed, Dec 30, 2009 at 7:17 AM, Corey Wright <[email protected]> wrote:
> On Mon, 21 Dec 2009 10:46:18 -0800
> "Luis R. Rodriguez" <[email protected]> wrote:
>
>> > actually, i have no personal stake in the fix as after i got it to work
>> > i learned that the new release is not compatible with my older
>> > distro-provided user-land (doesn't recognize it as wireless),
>>
>> Old userspace should not bust. If you find what the issue was let us
>> know. AFAICT old userspace tend to look for 3 possible things to see
>> if a device is a wireless device.
>>
>> ? 1) Wireless-tools requires SIOCGIWRANGE
>> ? 2) Network manager requires SIOCGIWNAME
>> ? 3) Linux hald < 0.5.12 used to use check the existance of
>> /sys/class/net/<interface>/wireless/
>>
>> The only one I can think of your userspace might have an issue with is
>> the last one but even then you should get this if you have
>> CONFIG_WIRELESS_EXT_SYSFS=y on 2.6.27.41
>
> the primary differences between compat-wireless 2.6.30 & 2.6.32.2 ath5k on
> 2.6.27.42 (by observation of the resulting environment) is the older
> version creates a corresponding wmaster0 interface and the wlan0 sysfs
> entry has a wireless directory (ie /sys/class/net/wlan0/wireless).
>
> hal is version 0.5.11rc2 (from "hald --version") and indeed does not
> recognize wlan0 as 802.11, but instead 802.03 (from "hal-device").
>
> without digging into the hal source (though i guess i could simply strace
> it and see if it tests for the wireless subdirectory), it appears your idea
> was correct, but i have CONFIG_WIRELESS_EXT_SYSFS=y.
>
> from my 2.6.27.42 .config (which i compiled & ran both compat-wireless
> versions against):
> #
> # Wireless
> #
> CONFIG_CFG80211=m
> CONFIG_NL80211=y
> CONFIG_WIRELESS_EXT=y
> CONFIG_WIRELESS_EXT_SYSFS=y
> CONFIG_MAC80211=m
>
> if you have any other ideas, patches, etc, then i'll try them. ?or tell me
> that hal < 0.5.12 is unsupported and i'll stick with compat-wireless-2.6.30.
>
> for the record: wireless-tools (iwconfig) recognizes wlan0 as "IEEE
> 802.11bg" with both compat-wireless versions and NetworkManager only used
> SIOC[SG]IW* ioctl requests (eg ESSID, MODE) with the older version (as seen
> by strace), so i figure hal reported wlan0 as 802.03 and NetworkManager
> believed it (never trying wireless ioctl requests).
>
> thanks for your work (going all the way back to prism54)!
>
> corey
> --
> [email protected]
> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to [email protected]
> More majordomo info at ?http://vger.kernel.org/majordomo-info.html
>
Having looked at hal source a bit, I think it doesn't do much more
than monitoring things going on in /sys -
most of the device probing code is in hal/hald/linux/device.c, if that
helps. hal is not big and not too difficult to find your way around.
wmaster is on its way out, I think.
(my apologies to those with thread-aware mail user agents as this is not a
true reply as i'm not subscribed to linux-wireless, but copied from
marc.info's raw message display.)
> If there are any bugs please report them
> through linux-wireless.
> I've tested ath5k on 2.6.27.41 on x86_64 and
> send this e-mail out through it.
> While at it 2.6.32.2 got released today so I just updated a release
> for that so users can benefit from the few fixes [3], you can check
> the ChangeLog for that on the wireless for details on the changes [4].
i tested ath5k from compat-wireless-2.6.32 on 2.6.27.41 and from
compat-wireless-2.6.32.2 on 2.6.27.42 and i needed this patch [1]
to avoid duplicate ieee80211_rx symbol in ieee80211 (from kernel) &
mac80211 (from compat-wireless).
[1] http://patchwork.kernel.org/patch/54569/
i see that the patch was applied to compat-wireless-2.6.33-rc1, but i
wonder if it should be backported to the 2.6.32.2 release.
actually, i have no personal stake in the fix as after i got it to work i
learned that the new release is not compatible with my older distro-provided
user-land (doesn't recognize it as wireless), which i don't feel like
upgrading/self-maintaining and possibly descending into dependency-hell, so
i'm sticking with compat-wireless-2.6.30 (compat-wireless-2.6.31-rc7 release
didn't seem to work as well as 2.6.30), but thought the finding might be of
interest to others (possibly stumbling upon this searching for
"compat-wireless-2.6.32 ieee80211_rx ieee80211 mac80211 duplicate symbol"
or something similar).
thanks for compat-wireless!
corey
--
[email protected]