Hi all,
First, the good news: bcm43xx-mac80211 in wireless-dev now works about
as well for me as the old driver. Similar connection reliability,
upload speed and download speed. Woo!
But I've got ton's of these in my log:
BUG: at net/mac80211/ieee80211.c:1280 ieee80211_tx()
Call Trace:
[<ffffffff805866a3>] ieee80211_master_start_xmit+0x5f3/0x680
[<ffffffff804d447d>] dev_hard_start_xmit+0x26d/0x2d0
[<ffffffff804e4781>] __qdisc_run+0xd1/0x260
[<ffffffff804d6a87>] dev_queue_xmit+0x1a7/0x370
[<ffffffff8058a45b>] ieee80211_subif_start_xmit+0x34b/0x590
[<ffffffff804ce3a7>] kfree_skb+0x17/0x30
[<ffffffff804d447d>] dev_hard_start_xmit+0x26d/0x2d0
[<ffffffff804e4781>] __qdisc_run+0xd1/0x260
[<ffffffff804d6a87>] dev_queue_xmit+0x1a7/0x370
[<ffffffff804f9bef>] ip_output+0x19f/0x340
[<ffffffff804f9069>] ip_queue_xmit+0x219/0x430
[<ffffffff80295fe0>] __pollwait+0x0/0x110
[<ffffffff8050a82e>] tcp_transmit_skb+0x40e/0x7c0
[<ffffffff8022ce00>] default_wake_function+0x0/0x10
[<ffffffff8050c4bb>] __tcp_push_pending_frames+0x11b/0x940
[<ffffffff8059c6a8>] csum_partial_copy_from_user+0x88/0x120
[<ffffffff80500b66>] tcp_sendmsg+0x876/0xc80
[<ffffffff8022ce0d>] default_wake_function+0xd/0x10
[<ffffffff8022ba37>] __wake_up_common+0x47/0x70
[<ffffffff8051e795>] inet_sendmsg+0x45/0x80
[<ffffffff804c7161>] sock_aio_write+0x101/0x110
[<ffffffff80286071>] do_sync_write+0xf1/0x130
[<ffffffff80246f80>] autoremove_wake_function+0x0/0x40
[<ffffffff8023688e>] current_fs_time+0x3e/0x40
[<ffffffff80286999>] vfs_write+0x159/0x170
[<ffffffff80286fa0>] sys_write+0x50/0x90
[<ffffffff8020a1ae>] system_call+0x7e/0x83
Kernel: wireless-dev latest from git
Arch: x86-64
My boot log is attached for full hardware information.
--
Will Dyson
On Friday 25 May 2007 21:10, Will Dyson wrote:
> On 5/25/07, Michael Wu <[email protected]> wrote:
> > On Friday 25 May 2007 16:37, Will Dyson wrote:
> > > But I've got ton's of these in my log:
> > >
> > > BUG: at net/mac80211/ieee80211.c:1280 ieee80211_tx()
> >
> > Do you have CONFIG_NET_SCHED on?
>
> I do not have it set in this kernel. Should I?
It may help. mac80211 didn't work correctly without it before and I'm not
entirely sure if it works correctly now without that.
-Michael Wu
On Friday 25 May 2007 16:37, Will Dyson wrote:
> But I've got ton's of these in my log:
>
> BUG: at net/mac80211/ieee80211.c:1280 ieee80211_tx()
>
Do you have CONFIG_NET_SCHED on?
If you do, can you check what qdisc is attached to the master device? This
command should do it if you only have one mac80211 device on your system:
tc qdisc show dev wmaster0
I suspect the patch that made mac80211 work without the ieee80211 qdisc might
not be working correctly.
-Michael Wu
On 5/25/07, Michael Wu <[email protected]> wrote:
> On Friday 25 May 2007 16:37, Will Dyson wrote:
> > But I've got ton's of these in my log:
> >
> > BUG: at net/mac80211/ieee80211.c:1280 ieee80211_tx()
> >
> Do you have CONFIG_NET_SCHED on?
I do not have it set in this kernel. Should I?
--
Will Dyson
On Saturday 26 May 2007 01:37, Will Dyson wrote:
> Hi all,
>
> First, the good news: bcm43xx-mac80211 in wireless-dev now works about
> as well for me as the old driver. Similar connection reliability,
> upload speed and download speed. Woo!
Whoo, nice. :)
/me pours a bottle of beer to everybody on the list.
> But I've got ton's of these in my log:
>
> BUG: at net/mac80211/ieee80211.c:1280 ieee80211_tx()
Hm, probably mac80211 bug. Not sure. I didn't look at the code, yet.
> Call Trace:
> [<ffffffff805866a3>] ieee80211_master_start_xmit+0x5f3/0x680
> [<ffffffff804d447d>] dev_hard_start_xmit+0x26d/0x2d0
> [<ffffffff804e4781>] __qdisc_run+0xd1/0x260
> [<ffffffff804d6a87>] dev_queue_xmit+0x1a7/0x370
> [<ffffffff8058a45b>] ieee80211_subif_start_xmit+0x34b/0x590
> [<ffffffff804ce3a7>] kfree_skb+0x17/0x30
> [<ffffffff804d447d>] dev_hard_start_xmit+0x26d/0x2d0
> [<ffffffff804e4781>] __qdisc_run+0xd1/0x260
> [<ffffffff804d6a87>] dev_queue_xmit+0x1a7/0x370
> [<ffffffff804f9bef>] ip_output+0x19f/0x340
> [<ffffffff804f9069>] ip_queue_xmit+0x219/0x430
> [<ffffffff80295fe0>] __pollwait+0x0/0x110
> [<ffffffff8050a82e>] tcp_transmit_skb+0x40e/0x7c0
> [<ffffffff8022ce00>] default_wake_function+0x0/0x10
> [<ffffffff8050c4bb>] __tcp_push_pending_frames+0x11b/0x940
> [<ffffffff8059c6a8>] csum_partial_copy_from_user+0x88/0x120
> [<ffffffff80500b66>] tcp_sendmsg+0x876/0xc80
> [<ffffffff8022ce0d>] default_wake_function+0xd/0x10
> [<ffffffff8022ba37>] __wake_up_common+0x47/0x70
> [<ffffffff8051e795>] inet_sendmsg+0x45/0x80
> [<ffffffff804c7161>] sock_aio_write+0x101/0x110
> [<ffffffff80286071>] do_sync_write+0xf1/0x130
> [<ffffffff80246f80>] autoremove_wake_function+0x0/0x40
> [<ffffffff8023688e>] current_fs_time+0x3e/0x40
> [<ffffffff80286999>] vfs_write+0x159/0x170
> [<ffffffff80286fa0>] sys_write+0x50/0x90
> [<ffffffff8020a1ae>] system_call+0x7e/0x83
>
> Kernel: wireless-dev latest from git
> Arch: x86-64
>
> My boot log is attached for full hardware information.
>
On 5/26/07, Michael Wu <[email protected]> wrote:
> On Friday 25 May 2007 21:10, Will Dyson wrote:
> > On 5/25/07, Michael Wu <[email protected]> wrote:
> > > On Friday 25 May 2007 16:37, Will Dyson wrote:
> > > > But I've got ton's of these in my log:
> > > >
> > > > BUG: at net/mac80211/ieee80211.c:1280 ieee80211_tx()
> > >
> > > Do you have CONFIG_NET_SCHED on?
> >
> > I do not have it set in this kernel. Should I?
> It may help. mac80211 didn't work correctly without it before and I'm not
> entirely sure if it works correctly now without that.
Ok. I'll try it.
For what it's worth, I can no-longer reproduce this error (although it
was reproducible across several reboots yesterday).
--
Will Dyson
On 5/26/07, Will Dyson <[email protected]> wrote:
> On 5/26/07, Michael Wu <[email protected]> wrote:
> > On Friday 25 May 2007 21:10, Will Dyson wrote:
> > > On 5/25/07, Michael Wu <[email protected]> wrote:
> > > > On Friday 25 May 2007 16:37, Will Dyson wrote:
> > > > > But I've got ton's of these in my log:
> > > > >
> > > > > BUG: at net/mac80211/ieee80211.c:1280 ieee80211_tx()
> > > >
> > > > Do you have CONFIG_NET_SCHED on?
> > >
> > > I do not have it set in this kernel. Should I?
> > It may help. mac80211 didn't work correctly without it before and I'm not
> > entirely sure if it works correctly now without that.
>
> Ok. I'll try it.
>
> For what it's worth, I can no-longer reproduce this error (although it
> was reproducible across several reboots yesterday).
Followup:
The warning did eventually show up again in the kernel without
CONFIG_NET_SCHED. Runing a week with CONFIG_NET_SCHED without a
recurrence. So that seems to have solved it.
If mac80211 needs CONFIG_NET_SCHED, shouldn't mac80211 select or depend on it?
--
Will Dyson
On Sun, 2007-06-03 at 15:34 -0400, Will Dyson wrote:
> The warning did eventually show up again in the kernel without
> CONFIG_NET_SCHED. Runing a week with CONFIG_NET_SCHED without a
> recurrence. So that seems to have solved it.
>
> If mac80211 needs CONFIG_NET_SCHED, shouldn't mac80211 select or depend on it?
It's supposed to not need it ;)
johannes