Re: 2.6.18-mm2 -- EIP: [<c11a962e>] klist_node_init+0x2b/0x3a SS:ESP 0068:f63a5f80

[Andrew Morton <akpm@osdl.org>]

On Thu, 28 Sep 2006 21:31:19 -0700
"Miles Lane" <miles.lane@gmail.com> wrote:

> On 9/28/06, Andrew Morton <akpm@osdl.org> wrote:
> > On Thu, 28 Sep 2006 19:13:23 -0700
> > "Miles Lane" <miles.lane@gmail.com> wrote:
> >
> > > eth1: RealTek RTL8139 at 0xf9076800, 00:c0:9f:95:18:1b, IRQ 19
> > > eth1:  Identified 8139 chip type 'RTL-8100B/8139D'
> > > BUG: unable to handle kernel NULL pointer dereference at virtual
> > > address 000000d0
> > >  printing eip:
> > > c11a962e
> > > *pde = 00000000
> > > Oops: 0000 [#1]
> > > PREEMPT
> > > last sysfs file: /class/firmware/0000:01:06.0/loading
> >
> > So you got to the stage of loading firmware.
> >
> > > Modules linked in: shpchp pci_hotplug intel_agp i2c_i801 agpgart
> > > snd_intel8x0 i2c_core snd_intel8x0m snd_ac97_codec snd_ac97_bus
> > > snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device
> > > snd_pcm_oss snd_mixer_oss ata_generic ata_piix libata 8139too sdhci
> > > scsi_mod snd_pcm snd_timer psmouse snd soundcore snd_page_alloc 8139cp
> > > mii yenta_socket rsrc_nonstatic pcmcia_core ohci1394 serio_raw ipw2200
> > > ieee1394 ide_cd cdrom rtc unix ehci_hcd ohci_hcd uhci_hcd usbcore ext3
> > > jbd mbcache
> > > CPU:    0
> > > EIP:    0060:[<c11a962e>]    Not tainted VLI
> > > EFLAGS: 00010296   (2.6.18-mm2 #8)
> > > EIP is at klist_node_init+0x2b/0x3a
> > > eax: dff16b08   ebx: 000000a0   ecx: c102ebd7   edx: f63a5f44
> > > esi: dff16afc   edi: f910e214   ebp: f63a5f88   esp: f63a5f80
> > > ds: 007b   es: 007b   ss: 0068
> > > Process probe-0000:01:0 (pid: 1697, ti=f63a4000 task=f639c030 task.ti=f63a4000)
> > > Stack: 000000a0 dff16afc f63a5f98 c11a964f dff16a80 dff16afc f63a5fac c1124295
> > >        00000000 dff16a80 f910e214 f63a5fc4 c1124338 f5c17e80 f5c17e80 f633bd90
> > >        c11242f0 f63a5fe0 c102a834 ffffffff ffffffff c102a784 00000000 00000000
> > > Call Trace:
> > >  [<c11a964f>] klist_add_tail+0x12/0x38
> > >  [<c1124295>] device_bind_driver+0x45/0xa0
> > >  [<c1124338>] really_probe+0x48/0xb3
> > >  [<c102a834>] kthread+0xb0/0xdc
> > >  [<c1003abb>] kernel_thread_helper+0x7/0x10
> > > DWARF2 unwinder stuck at kernel_thread_helper+0x7/0x10
> > >
> > > Leftover inexact backtrace:
> > >
> > >  [<c1003f02>] show_trace_log_lvl+0x12/0x25
> > >  [<c1003fa1>] show_stack_log_lvl+0x8c/0x97
> > >  [<c100412c>] show_registers+0x180/0x214
> > >  [<c1004355>] die+0x195/0x2b0
> > >  [<c10148a9>] do_page_fault+0x419/0x4e4
> > >  [<c11ac329>] error_code+0x39/0x40
> > >  [<c11a964f>] klist_add_tail+0x12/0x38
> > >  [<c1124295>] device_bind_driver+0x45/0xa0
> > >  [<c1124338>] really_probe+0x48/0xb3
> > >  [<c102a834>] kthread+0xb0/0xdc
> > >  [<c1003abb>] kernel_thread_helper+0x7/0x10
> > >  =======================
> > > Code: 55 89 e5 56 53 89 c3 89 d6 8d 42 04 89 42 04 89 40 04 c7 42 10
> > > 00 00 00 00 8d 42 14 e8 f2 14 e8 ff 8d 46 0c e8 80 3d f1 ff 89 1e <8b>
> > > 53 30 85 d2 74 04 89 f0 ff d2 5b 5e 5d c3 55 89 e5 56 53 89
> > > EIP: [<c11a962e>] klist_node_init+0x2b/0x3a SS:ESP 0068:f63a5f80
> > >  <6>ipw2200: Detected geography ZZM (11 802.11bg channels, 0 802.11a channels)
> >
> > Does setting CONFIG_PCI_MULTITHREAD_PROBE=n fix it?
> 
> Yes and no.  The BUG no longer occurs,

OK, thanks.  Note to Greg: CONFIG_PCI_MULTITHREAD_PROBE+ipw2200 = oops.

> but either ipw2200 or
> NetworkManager is still not working.  I am running Fedora development
> (rawhide).  For some reason, NM still cannot establish a connection
> with this kernel build (this failed with 2.6.18-mm2 previously, but I
> thought it was related to the BUG).  "iwlist scan" can corrently
> detect my access point.  Also, I can use iwconfig and dhclient to
> establish a working connection.  NM fails with this output in the
> message log (I'll try getting a more detailed debug log out of NM):

Another reporter mentioned that he could only get his wireless to associate
with his AP by using essid "any".  But that was a different type of card. 

ipw2200 works OK for me, fwiw.

There's a largeish ipw2200 update in mainline now.  That'd be worth testing.

It looks like that ipw2200 update was in 2.6.18-mm1 though.  Did that work?