Re: Patch "coredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping" has been added to the 4.4-stable tree

Re: Patch "coredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping" has been added to the 4.4-stable tree

[Greg KH]

On Tue, Jun 04, 2019 at 04:52:31PM +0200, Michal Hocko wrote:
> Please note that I have posted my backport today
> http://lkml.kernel.org/r/20190604094953.26688-1-mhocko@kernel.org and it
> differs from this one. Please have a look!

Ah, good point, I just noticed that.  Ben, any thoughts as to the
difference?

thanks,

greg k-h

Re: Patch "coredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping" has been added to the 4.4-stable tree

[Ben Hutchings]

On Tue, 2019-06-04 at 17:07 +0200, Greg KH wrote:
> On Tue, Jun 04, 2019 at 04:52:31PM +0200, Michal Hocko wrote:
> > Please note that I have posted my backport today
> > http://lkml.kernel.org/r/20190604094953.26688-1-mhocko@kernel.org and it
> > differs from this one. Please have a look!
> 
> Ah, good point, I just noticed that.  Ben, any thoughts as to the
> difference?

- The additional fix in binder makes sense to me, and is also needed in
4.9.  I'm not sure why I missed it.

- I don't understand why collapse_huge_range() needs to be fixed, but
then I really don't understand the khugepaged code at all!  So I would
trust Michal on this.

- The userfaultfd fixes look different because I picked "userfaultfd:
don't pin the user memory in userfaultfd_file_create()" first and
Michal did not.  I don't think it makes sense to be calling
mmget_still_valid() in these functions if they don't use
mmget_not_zero() or similar.  But again, Michal is the expert here.

Ben.

-- 
Ben Hutchings, Software Developer                         Codethink Ltd
https://www.codethink.co.uk/                 Dale House, 35 Dale Street
                                     Manchester, M1 2HF, United Kingdom

Re: Patch "coredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping" has been added to the 4.4-stable tree

[Michal Hocko]

On Tue 04-06-19 16:51:38, Ben Hutchings wrote:
[...]
> - I don't understand why collapse_huge_range() needs to be fixed, but
> then I really don't understand the khugepaged code at all!  So I would
> trust Michal on this.

To be honest, I am not really sure myself here. But we are using a
remote mm there and I do not see anything that would prevent from racing
with exit/coredump. Maybe I am wrong. Let's wait for Andrea for his
review feedback. This patch is quite tricky for the stable backport.

Thanks!

-- 
Michal Hocko
SUSE Labs

Re: Patch "coredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping" has been added to the 4.4-stable tree

[Ben Hutchings]

On Wed, 2019-06-05 at 08:50 +0200, Michal Hocko wrote:
> On Tue 04-06-19 16:51:38, Ben Hutchings wrote:
> [...]
> > - I don't understand why collapse_huge_range() needs to be fixed, but
> > then I really don't understand the khugepaged code at all!  So I would
> > trust Michal on this.
> 
> To be honest, I am not really sure myself here. But we are using a
> remote mm there and I do not see anything that would prevent from racing
> with exit/coredump. Maybe I am wrong. Let's wait for Andrea for his
> review feedback. This patch is quite tricky for the stable backport.

So, Greg, it seems like you should drop this from the current stable
round.

Ben.

-- 
Ben Hutchings, Software Developer                         Codethink Ltd
https://www.codethink.co.uk/                 Dale House, 35 Dale Street
                                     Manchester, M1 2HF, United Kingdom

Re: Patch "coredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping" has been added to the 4.4-stable tree

[Greg KH]

On Wed, Jun 05, 2019 at 01:08:48PM +0100, Ben Hutchings wrote:
> On Wed, 2019-06-05 at 08:50 +0200, Michal Hocko wrote:
> > On Tue 04-06-19 16:51:38, Ben Hutchings wrote:
> > [...]
> > > - I don't understand why collapse_huge_range() needs to be fixed, but
> > > then I really don't understand the khugepaged code at all!  So I would
> > > trust Michal on this.
> > 
> > To be honest, I am not really sure myself here. But we are using a
> > remote mm there and I do not see anything that would prevent from racing
> > with exit/coredump. Maybe I am wrong. Let's wait for Andrea for his
> > review feedback. This patch is quite tricky for the stable backport.
> 
> So, Greg, it seems like you should drop this from the current stable
> round.

Ok, will do, thanks.

greg k-h