Re: vvar, gup && coredump

[Andy Lutomirski <luto@amacapital.net>]

On Thu, Mar 12, 2015 at 11:19 AM, Pedro Alves <palves@redhat.com> wrote:
> On 03/12/2015 05:46 PM, Oleg Nesterov wrote:
>> On 03/12, Oleg Nesterov wrote:
>>>
>>> Yes, this is true. OK, lets not dump it.
>>
>> OTOH. We can probably add ->access() into special_mapping_vmops, this
>> way __access_remote_vm() could work even if gup() fails ?
>>
>> Jan, Sergio. How much do we want do dump this area ? The change above
>> should be justified.
>
> Memory mappings that weren't touched since they were initially mapped can
> be retrieved from the program binary and the shared libraries, even if
> the core dump is moved to another machine.  However, in vvar case,
> sounds like  there's nowhere to read it from offline?  In that case,
> it could be justified to dump it.

This is why we currently dump the vdso text.  On arm64 (the only other
architecture that uses a real vma for vvar data IIRC), we use a more
normal vma and we dump it.  x86 is the odd one out here.

We could just leave the kernel alone.  The data that gets dumped is of
dubious value, but it could be slightly helpful when debugging vdso
crashes*, but, of course, dumping it is inherently racy.


* The vdso never crashes :)

--Andy

>
> Thanks,
> Pedro Alves
>



-- 
Andy Lutomirski
AMA Capital Management, LLC