Re: 2.6.24-rc4-mm1

[Reuben Farrelly <reuben-linuxkernel@reub.net>]

On 11/12/2007 8:11 AM, Andrew Morton wrote:
> On Tue, 11 Dec 2007 01:48:39 +1100
> Reuben Farrelly <reuben-linuxkernel@reub.net> wrote:
> 
>>
>> On 5/12/2007 4:17 PM, Andrew Morton wrote:
>>> Temporarily at
>>>
>>>   http://userweb.kernel.org/~akpm/2.6.24-rc4-mm1/
>>>
>>> Will appear later at
>>>
>>>   ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.24-rc4/2.6.24-rc4-mm1/
>>>
>>>
>>> - Lots of device IDs have been removed from the e1000 driver and moved over
>>>   to e1000e.  So if your e1000 stops working, you forgot to set CONFIG_E1000E.
>>>
>>> - The s390 build is still broken.
>> I'm seeing this most incredibly unhelpful (to debug) but fortunately
>> reproduceable problem (so far 4/4 times) on this -mm kernel.  I thought this 
>> problem may have been related to another bug which I have reported (A TCP oops) 
>> but even after applying a likely fix for that I am still seeing this problem.
>>
>> The machine boots up perfectly fine and runs good until I load it up.
>> In this case I can reliably cause this to occur by pulling a 3G ISO across the
>> GigE network from my Linux box to my PC.  After maybe 50M or so, the console 
>> just displays this (ignore initial boot banner):
>>
>> ----------
>>
>>   * Starting local ...                                                     [ ok ]
>>
>>
>> This is tornado.reub.net (Linux x86_64 2.6.24-rc4-mm1) 00:24:01
>>
>> tornado login: *** buffer overf
>>
>> -------
>>
>> Yes - after displaying the 'f' in what I can only guess is the word 'overflow',
>> the box spontaneously reboots.  There is no further console output until it 
>> starts to come back up again.
>>
>> The problem does not exist in 2.6.23-gentoo kernels nor in a vanilla 
>> 2.6.24-rc4-git6 (phew!), so this looks to be an -mm only problem at this stage.
>>
>> I enabled a number of kernel debugging options but then I got no output at all 
>> when the machine crashed.
>>
>> I'm at a bit of a loss as to which subsystem this might be coming from, so I'm 
>> not sure who to CC.
>>
>> Box information is (still) up at http://www.reub.net/files/kernel/2.6.24-rc4-mm1/
>>
> 
> hm.  grepping around for "buffer overflow" doesn't turn up anything except in
> drivers which you won't be using on that machine.
> 
> I'd be suspecting networking, obviously.  If you're feeling keen could you please
> grep a 2.6.24-rc4 tree and apply 2.6.24-rc4-mm1's origin.patch and git-net.patch
> and see if the bug is still present?

No - seems to be fine with just origin.patch and git-net.patch.

Just for good measure I then reverted git-net.patch and applied 
git-netdev-all.patch instead, and still wasn't able to trigger the reboot or 
console message, no matter how hard I tried.

I guess for now I'll sit on it, and if it appears in the next -mm it'll probably 
annoy me enough and inspire me to dig deeper (or, "guess" deeper, given the lack 
of direction as to where to even begin).

Reuben