Re: crash by cdc_acm driver in kernels 4.8-rc1/5

[Wim Osterholt]

On Thu, Nov 17, 2016 at 10:14:34AM +0100, Wim Osterholt wrote:
> > For completeness I should also try with SMP unset. That is for tomorrow
> > then.
> 
> With CONFIG_SMP unset nothing goes wrong here either.
> It looks like it has been fixed in 4.9-rc5, but I should also double check
> several conbinations on my (slower) laptop.

Nov 17 15:07:49 localhost kernel: usb 4-1: new full-speed USB device number 2 using uhci_hcd
Nov 17 15:07:49 localhost kernel: usb 4-1: New USB device found, idVendor=0572, idProduct=1340
Nov 17 15:07:49 localhost kernel: usb 4-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3
Nov 17 15:07:49 localhost kernel: usb 4-1: Product: USB Modem
Nov 17 15:07:49 localhost kernel: usb 4-1: Manufacturer: Conexant
Nov 17 15:07:49 localhost kernel: usb 4-1: SerialNumber: 12345678
Nov 17 15:07:50 localhost mtp-probe[12956]: checking bus 4, device 2: "/sys/devices/pci0000:00/0000:00:1d.2/usb4/4-1"
Nov 17 15:07:50 localhost mtp-probe[12956]: bus: 4, device: 2 was not an MTP device
Nov 17 15:07:51 localhost kernel: Check point  1
Nov 17 15:07:51 localhost kernel: Check point  2
Nov 17 15:07:51 localhost kernel: Check point  3
Nov 17 15:07:51 localhost kernel: Check point  4
Nov 17 15:07:51 localhost kernel: Check point  5
Nov 17 15:07:51 localhost kernel: Check point  6
Nov 17 15:07:51 localhost kernel: Check point  7
Nov 17 15:07:51 localhost kernel: Check point  8
Nov 17 15:07:51 localhost kernel: Check point  9
Nov 17 15:07:51 localhost kernel: Check point  10
Nov 17 15:07:51 localhost kernel: BUG: unable to handle kernel NULL pointer dereference at 00000249
Nov 17 15:07:51 localhost kernel: IP: [<e186ece2>] acm_probe+0x559/0xe53 [cdc_acm]
Nov 17 15:07:51 localhost kernel: *pde = 00000000 
Nov 17 15:07:51 localhost kernel: Oops: 0000 [#1] SMP
...


You can get a few logs concatenated at
http://webserver.djo.tudelft.nl/insane488.logs

Usually the oops follows immediately after inserting the modem, but in one case
it took a few seconds extra before the oops. I think that is the one that
printed the full range of checkpoints.

On another 3 machines running 4.9-rc5 here  everything went fine on insertion.
Switching back to 4.8.8 produced the bug with outputs similar to the above.

Now back to the laptops to see if I can get confirmation of having seen the
oops at 4.9-rc5 in some way. I'm not sure now. 


Regards, Wim.