[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#769466: /boot/vmlinuz-3.16.0-4-amd64: Kernel Oops when specifying console=ttyS0




On November 13, 2014 5:03:56 PM PST, Ben Hutchings <ben@decadent.org.uk> wrote:
>> On 11/13/2014 01:45 PM, Ben Hutchings wrote:
>> > On Thu, 2014-11-13 at 12:53 -0800, Nikolaus Rath wrote:
>> >> Package: src:linux
>> >> Version: 3.16.7-2
>> >> Severity: normal
>> >> File: /boot/vmlinuz-3.16.0-4-amd64
>> >>
>> >> Hello,
>> >>
>> >> When adding 
>> >>
>> >>   console=ttyS0,115200n8 console=tty0 no_console_suspend
>> >>
>> >> to the kernel command line, I'm getting the following kernel Oops
>during
>> >> boot (full quote in kernel.log below):
>> >>
>> >> [...]
>> >> [   13.687971] 0000:07:01.0: ttyS0 at I/O 0xc050 (irq = 19,
>base_baud = 115200) is a 16550A
>> >> [   13.687980] kernel tried to execute NX-protected page - exploit
>attempt? (uid: 0)
>> >> [   13.688130] BUG: unable to handle kernel paging request at
>ffffffff819419bf
>> >> [   13.688299] IP: [<ffffffff819419bf>]
>serial8250_console_setup+0x0/0xaa
>> >> [   13.688604] PGD 1816067 PUD 1817063 PMD 215f93063 PTE
>8000000001941163
>> >> [   13.688911] Oops: 0011 [#1] SMP 
>> >> [...]
>> >>
>> >> Please let me know if I can do something to debug this further.
>> > [...]
>> > 
>> > The log didn't include the full oops message; please can you send
>that
>> > as I don't understand why anything would call this function at this
>> > point.  (It is only meant to be called during kernel
>initialisation,
>> > before starting the init program.  At this point the code has been
>> > freed, hence the page fault.)
>> 
>> Full log is attached. Sorry, I did not realize that the automatically
>> included log was truncated.
>
>This bug seems to be specific to parport_serial.  Only built-in drivers
>can be console drivers, but parport_serial is built as a module.
>However, because it works on top of with 8250_pci, which *is* built-in,
>the kernel tries to use it as a console driver anyway, and this leads
>to
>the crash.  We should fix the crash, but you still won't be able to use
>ports on the combined parallel/serial card for a console.

Duh. Not at all, or will it work if I build my own kernel with parport_serial built-in? 

Do I understand correctly that a pci card with only serial ports would work without changes? 

Best, 
Nikolaus 

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.


Reply to: