Comment 57 for bug 1415880

Revision history for this message
Rafael David Tinoco (rafaeldtinoco) wrote :

Martin,

I'm not wireless expert but, after reading a bit and analysing your syslog output:

----
Aug 20 12:57:16 martin04 kernel: [ 9464.423445] Call Trace:
Aug 20 12:57:16 martin04 kernel: [ 9464.423451] [<ffffffff817aeed7>] dump_stack+0x45/0x57
Aug 20 12:57:16 martin04 kernel: [ 9464.423453] [<ffffffff81074d8a>] warn_slowpath_common+0x8a/0xc0
Aug 20 12:57:16 martin04 kernel: [ 9464.423455] [<ffffffff81074e7a>] warn_slowpath_null+0x1a/0x20
Aug 20 12:57:16 martin04 kernel: [ 9464.423462] [<ffffffffc04c5e09>] cfg80211_roamed+0x89/0x90 [cfg80211]
Aug 20 12:57:16 martin04 kernel: [ 9464.423489] [<ffffffffc07aba0b>] wl_notify_roaming_status+0xcb/0x150 [wl]
Aug 20 12:57:16 martin04 kernel: [ 9464.423508] [<ffffffffc07a876a>] wl_event_handler+0x6a/0x230 [wl]
Aug 20 12:57:16 martin04 kernel: [ 9464.423526] [<ffffffffc07a8700>] ? wl_free_wdev.isra.23+0x80/0x80 [wl]
Aug 20 12:57:16 martin04 kernel: [ 9464.423529] [<ffffffff81093802>] kthread+0xd2/0xf0
Aug 20 12:57:16 martin04 kernel: [ 9464.423531] [<ffffffff81093730>] ? kthread_create_on_node+0x1c0/0x1c0
Aug 20 12:57:16 martin04 kernel: [ 9464.423533] [<ffffffff817b67d8>] ret_from_fork+0x58/0x90
Aug 20 12:57:16 martin04 kernel: [ 9464.423535] [<ffffffff81093730>] ? kthread_create_on_node+0x1c0/0x1c0
----

This only means that broadcom WL driver tried to warn kernel that it roamed from one AP to another and new BSS could not be found (meaning that the card could not get all its new BSS - peers, APs, etc).

*** Try configuring the wireless adapter to use a SPECIFIC AP *** and check your wifi env for interferences that might be stepping into broadcom's driver decisions on whether to roam or not -> This does not mean a fix for WL is not needed, it only means we would have to wait for Broadcom to come up with a new driver.

Based on good feedback from Igor I'll suggest fixes for all Ubuntu versions. Even if upstream BW driver still faces problems with adapters, we still have to fix this bug's original problem.

I'll attach debdiffs for this bug soon.

Alberto,

I saw you pushed precise into latest version (right now in -proposed), I'll check if it also needs the fix and provide debdiff. Let me know if you have any other comments since you've been handling all bcmwl work and you are more into it than I am.

Thank you

Rafael