Occasional APC Smart-UPS CS 500 USB UPS troubles

Occasional APC Smart-UPS CS 500 USB UPS troubles

[Chris Boot]

Hi all,

I have an APC Smart-UPS CS 500, and most of the time it works really 
nicely. I was stupefied to find it worked out of the box in Ubuntu 
Dapper and pops up a nice little icon when the power goes out, etc... 
The trouble is, that's only most of the time.

I'm guessing the USB controller in the device is buggy or something, 
because occasionally, when I reboot my machine with the UPS plugged in, 
the boot hangs or produces strange errors when detecting USB devices. 
All it takes to get the machine to boot properly is yank the USB plug on 
the UPS, plug it in again, and reboot.

Has anyone seen this before?

Not exactly kernel related I must admit, but this never seems to happen 
on *shudder* Windows, and I wouldn't expect such severe behaviour in 
case of trouble...

I've seen this on all sorts of kernels starting with the early 2.6 
series up to 2.6.16.1 and 2.6.16-ck3. My latest hang was with the latter 
kernel, and modprobe got stuck with the following trace:

modprobe      D ED465280     0   819    815                     (NOTLB)
f7cfade4 f7d0366c f7d03540 ed465280 000f41fd 005b8d80 00000000 f7794d1c
        00000292 f7cfa000 f7d03540 c02d0e0e 00000001 f7d03540 c0113f02 
f7794d24
        f7794d24 f6e31c58 f9734f40 f97e2f44 c0257dc6 c02cf89f f97e2f44 
f7794c58
Call Trace:
  [__down+202/240] __down+0xca/0xf0
  [default_wake_function+0/12] default_wake_function+0x0/0xc
  [__driver_attach+0/89] __driver_attach+0x0/0x59
  [__sched_text_start+7/12] __down_failed+0x7/0xc
  [.text.lock.dd+39/188] .text.lock.dd+0x27/0xbc
  [bus_for_each_dev+55/89] bus_for_each_dev+0x37/0x59
  [driver_attach+17/19] driver_attach+0x11/0x13
  [__driver_attach+0/89] __driver_attach+0x0/0x59
  [bus_add_driver+90/211] bus_add_driver+0x5a/0xd3
  [pg0+959850838/1069790208] usb_register_driver+0x50/0xae [usbcore]
  [pg0+949030918/1069790208] hid_init+0x6/0x3d [usbhid]
  [sys_init_module+4905/5197] sys_init_module+0x1329/0x144d
  [cp_new_stat64+237/255] cp_new_stat64+0xed/0xff
  [vma_prio_tree_insert+23/42] vma_prio_tree_insert+0x17/0x2a
  [vma_link+162/223] vma_link+0xa2/0xdf
  [do_mmap_pgoff+1202/1535] do_mmap_pgoff+0x4b2/0x5ff
  [sys_mmap2+97/144] sys_mmap2+0x61/0x90
  [sysenter_past_esp+84/117] sysenter_past_esp+0x54/0x75

Let me know if you need more info!

Thanks,
Chris

-- 
Chris Boot
bootc@bootc.net
http://www.bootc.net/

Re: Occasional APC Smart-UPS CS 500 USB UPS troubles

[Greg KH]

On Sun, Apr 02, 2006 at 03:39:42PM +0100, Chris Boot wrote:
> Hi all,
> 
> I have an APC Smart-UPS CS 500, and most of the time it works really 
> nicely. I was stupefied to find it worked out of the box in Ubuntu 
> Dapper and pops up a nice little icon when the power goes out, etc... 
> The trouble is, that's only most of the time.
> 
> I'm guessing the USB controller in the device is buggy or something, 
> because occasionally, when I reboot my machine with the UPS plugged in, 
> the boot hangs or produces strange errors when detecting USB devices. 
> All it takes to get the machine to boot properly is yank the USB plug on 
> the UPS, plug it in again, and reboot.

What exactly are these "strange errors"?

> Has anyone seen this before?

It hasn't been reported, no.

> Not exactly kernel related I must admit, but this never seems to happen 
> on *shudder* Windows, and I wouldn't expect such severe behaviour in 
> case of trouble...
> 
> I've seen this on all sorts of kernels starting with the early 2.6 
> series up to 2.6.16.1 and 2.6.16-ck3. My latest hang was with the latter 
> kernel, and modprobe got stuck with the following trace:
> 
> modprobe      D ED465280     0   819    815                     (NOTLB)
> f7cfade4 f7d0366c f7d03540 ed465280 000f41fd 005b8d80 00000000 f7794d1c
>        00000292 f7cfa000 f7d03540 c02d0e0e 00000001 f7d03540 c0113f02 
> f7794d24
>        f7794d24 f6e31c58 f9734f40 f97e2f44 c0257dc6 c02cf89f f97e2f44 
> f7794c58
> Call Trace:
>  [__down+202/240] __down+0xca/0xf0
>  [default_wake_function+0/12] default_wake_function+0x0/0xc
>  [__driver_attach+0/89] __driver_attach+0x0/0x59
>  [__sched_text_start+7/12] __down_failed+0x7/0xc
>  [.text.lock.dd+39/188] .text.lock.dd+0x27/0xbc
>  [bus_for_each_dev+55/89] bus_for_each_dev+0x37/0x59
>  [driver_attach+17/19] driver_attach+0x11/0x13
>  [__driver_attach+0/89] __driver_attach+0x0/0x59
>  [bus_add_driver+90/211] bus_add_driver+0x5a/0xd3
>  [pg0+959850838/1069790208] usb_register_driver+0x50/0xae [usbcore]
>  [pg0+949030918/1069790208] hid_init+0x6/0x3d [usbhid]
>  [sys_init_module+4905/5197] sys_init_module+0x1329/0x144d
>  [cp_new_stat64+237/255] cp_new_stat64+0xed/0xff
>  [vma_prio_tree_insert+23/42] vma_prio_tree_insert+0x17/0x2a
>  [vma_link+162/223] vma_link+0xa2/0xdf
>  [do_mmap_pgoff+1202/1535] do_mmap_pgoff+0x4b2/0x5ff
>  [sys_mmap2+97/144] sys_mmap2+0x61/0x90
>  [sysenter_past_esp+84/117] sysenter_past_esp+0x54/0x75
> 
> Let me know if you need more info!

That's really odd.  What else is happening in the sysrq-t output at this
moment in time?

Also CCing the linux-usb-devel list, as the people there can help out.

thanks,

greg k-h

Re: Occasional APC Smart-UPS CS 500 USB UPS troubles

[Chris Boot]

On 2 Apr 2006, at 19:52, Greg KH wrote:

> On Sun, Apr 02, 2006 at 03:39:42PM +0100, Chris Boot wrote:
>> Hi all,
>>
>> I have an APC Smart-UPS CS 500, and most of the time it works really
>> nicely. I was stupefied to find it worked out of the box in Ubuntu
>> Dapper and pops up a nice little icon when the power goes out, etc...
>> The trouble is, that's only most of the time.
>>
>> I'm guessing the USB controller in the device is buggy or something,
>> because occasionally, when I reboot my machine with the UPS  
>> plugged in,
>> the boot hangs or produces strange errors when detecting USB devices.
>> All it takes to get the machine to boot properly is yank the USB  
>> plug on
>> the UPS, plug it in again, and reboot.
>
> What exactly are these "strange errors"?

Well I haven't seen them for a long time now, but error messages that  
say very little other than an error code (which, from memory, was -47  
or something).

>> Has anyone seen this before?
>
> It hasn't been reported, no.

Well next time it happens I'll try to gather as much information as  
possible. What sort of stuff would help diagnose the problem?

>> Not exactly kernel related I must admit, but this never seems to  
>> happen
>> on *shudder* Windows, and I wouldn't expect such severe behaviour in
>> case of trouble...
>>
>> I've seen this on all sorts of kernels starting with the early 2.6
>> series up to 2.6.16.1 and 2.6.16-ck3. My latest hang was with the  
>> latter
>> kernel, and modprobe got stuck with the following trace:
>>
>> modprobe      D ED465280     0   819    815                      
>> (NOTLB)
>> f7cfade4 f7d0366c f7d03540 ed465280 000f41fd 005b8d80 00000000  
>> f7794d1c
>>        00000292 f7cfa000 f7d03540 c02d0e0e 00000001 f7d03540 c0113f02
>> f7794d24
>>        f7794d24 f6e31c58 f9734f40 f97e2f44 c0257dc6 c02cf89f f97e2f44
>> f7794c58
>> Call Trace:
>>  [__down+202/240] __down+0xca/0xf0
>>  [default_wake_function+0/12] default_wake_function+0x0/0xc
>>  [__driver_attach+0/89] __driver_attach+0x0/0x59
>>  [__sched_text_start+7/12] __down_failed+0x7/0xc
>>  [.text.lock.dd+39/188] .text.lock.dd+0x27/0xbc
>>  [bus_for_each_dev+55/89] bus_for_each_dev+0x37/0x59
>>  [driver_attach+17/19] driver_attach+0x11/0x13
>>  [__driver_attach+0/89] __driver_attach+0x0/0x59
>>  [bus_add_driver+90/211] bus_add_driver+0x5a/0xd3
>>  [pg0+959850838/1069790208] usb_register_driver+0x50/0xae [usbcore]
>>  [pg0+949030918/1069790208] hid_init+0x6/0x3d [usbhid]
>>  [sys_init_module+4905/5197] sys_init_module+0x1329/0x144d
>>  [cp_new_stat64+237/255] cp_new_stat64+0xed/0xff
>>  [vma_prio_tree_insert+23/42] vma_prio_tree_insert+0x17/0x2a
>>  [vma_link+162/223] vma_link+0xa2/0xdf
>>  [do_mmap_pgoff+1202/1535] do_mmap_pgoff+0x4b2/0x5ff
>>  [sys_mmap2+97/144] sys_mmap2+0x61/0x90
>>  [sysenter_past_esp+84/117] sysenter_past_esp+0x54/0x75
>>
>> Let me know if you need more info!
>
> That's really odd.  What else is happening in the sysrq-t output at  
> this
> moment in time?

Well, everything else is waiting for the modprobe to finish. There's  
rather a lot of processes and fortunately by then klogd is running so  
I can get the logs. Shall I send over the full bootlog and sysrq-t  
trace for that boot? It's long and there doesn't appear to be  
anything particularly juicy other than the above trace.

Actually with the above trace modprobe eventually returned, but USB  
was unusable. I tried to rmmod it and it hung again, hard this time.  
Again no interesting kernel messages, but the following sysrq-t trace:

rmmod         D 0F203D40     0  4902   3752                     (NOTLB)
f1068e9c f17001bc f1700090 0f203d40 000f4239 15382100 00000000 f7794d1c
        00000282 f1068000 f1700090 c02d0e0e 00000001 f1700090  
c0113f02 f7794d24
        f7cfadf8 f7794c00 f972d359 f8cfad2c f779be28 c02cf89f  
f7e3a000 f1068000
Call Trace:
[__down+202/240] __down+0xca/0xf0
[default_wake_function+0/12] default_wake_function+0x0/0xc
[__sched_text_start+7/12] __down_failed+0x7/0xc
[pg0+959837300/1069790208] .text.lock.hub+0xc7/0xef [usbcore]
[pg0+959838978/1069790208] usb_remove_hcd+0x7c/0xcd [usbcore]
[pg0+959869461/1069790208] usb_hcd_pci_remove+0x15/0x68 [usbcore]
[pci_device_remove+22/40] pci_device_remove+0x16/0x28
[__device_release_driver+83/107] __device_release_driver+0x53/0x6b
[driver_detach+117/165] driver_detach+0x75/0xa5
[bus_remove_driver+87/117] bus_remove_driver+0x57/0x75
[driver_unregister+11/19] driver_unregister+0xb/0x13
[pci_unregister_driver+12/103] pci_unregister_driver+0xc/0x67
[pg0+949181306/1069790208] uhci_hcd_cleanup+0xa/0x33 [uhci_hcd]
[sys_delete_module+353/393] sys_delete_module+0x161/0x189
[.text.lock.mlock+14/126] .text.lock.mlock+0xe/0x7e
[do_munmap+392/418] do_munmap+0x188/0x1a2
[sysenter_past_esp+84/117] sysenter_past_esp+0x54/0x75

It appears none of my other mishaps were actually logged... :-(

> Also CCing the linux-usb-devel list, as the people there can help out.
>
> thanks,
>
> greg k-h

-- 
Chris Boot
bootc@bootc.net
http://www.bootc.net/