Re: Questions about backports of fixes for "CoW after fork() issue"

Re: Questions about backports of fixes for "CoW after fork() issue"

[Greg Kroah-Hartman]

On Wed, Jun 16, 2021 at 02:47:15PM +0800, Liu Shixin wrote:
> Hi, Suren,
> 
> I read the previous discussion about fixing CVE-2020-29374 in stable 4.14 and 4.19 in
> <https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/>
> 
> https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/
> 
> And the results of the discussion is that you backports of 17839856fd58 for 4.14 and
> 
> 4.19 kernels.
> 
> But the bug about dax and strace in the discussion has not been solved, right? I don't
> 
> find a conclusion on this issue, am I missing something? Does this problem still exist in
> 
> the stable 4.14 and 4.19 kernel?

As the code is all there for you, can you just test them and see for
yourself?

thanks,

greg k-h

Re: Questions about backports of fixes for "CoW after fork() issue"

[Suren Baghdasaryan]

On Tue, Jun 15, 2021 at 11:52 PM Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
>
> On Wed, Jun 16, 2021 at 02:47:15PM +0800, Liu Shixin wrote:
> > Hi, Suren,
> >
> > I read the previous discussion about fixing CVE-2020-29374 in stable 4.14 and 4.19 in
> > <https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/>
> >
> > https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/
> >
> > And the results of the discussion is that you backports of 17839856fd58 for 4.14 and
> >
> > 4.19 kernels.
> >
> > But the bug about dax and strace in the discussion has not been solved, right? I don't
> >
> > find a conclusion on this issue, am I missing something? Does this problem still exist in
> >
> > the stable 4.14 and 4.19 kernel?

That is my understanding after discussions with Andrea but I did not
verify that myself. As Greg pointed out, the best way would be to try
it out.
Thanks,
Suren.

>
> As the code is all there for you, can you just test them and see for
> yourself?
>
> thanks,
>
> greg k-h

Re: Questions about backports of fixes for "CoW after fork() issue"

[Liu Shixin]

On 2021/6/16 15:11, Suren Baghdasaryan wrote:
> On Tue, Jun 15, 2021 at 11:52 PM Greg Kroah-Hartman
> <gregkh@linuxfoundation.org> wrote:
>> On Wed, Jun 16, 2021 at 02:47:15PM +0800, Liu Shixin wrote:
>>> Hi, Suren,
>>>
>>> I read the previous discussion about fixing CVE-2020-29374 in stable 4.14 and 4.19 in
>>> <https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/>
>>>
>>> https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/
>>>
>>> And the results of the discussion is that you backports of 17839856fd58 for 4.14 and
>>>
>>> 4.19 kernels.
>>>
>>> But the bug about dax and strace in the discussion has not been solved, right? I don't
>>>
>>> find a conclusion on this issue, am I missing something? Does this problem still exist in
>>>
>>> the stable 4.14 and 4.19 kernel?
> That is my understanding after discussions with Andrea but I did not
> verify that myself. As Greg pointed out, the best way would be to try
> it out.
> Thanks,
> Suren.
>
>> As the code is all there for you, can you just test them and see for
>> yourself?
>>
>> thanks,
>>
>> greg k-h
> .
>
Thank you both for replies. I have tested it in stable 4.19 kernel and the bug is existed as expected.

Thanks,
Liu Shixin

Re: Questions about backports of fixes for "CoW after fork() issue"

[Greg Kroah-Hartman]

On Wed, Jun 16, 2021 at 05:28:54PM +0800, Liu Shixin wrote:
> On 2021/6/16 15:11, Suren Baghdasaryan wrote:
> > On Tue, Jun 15, 2021 at 11:52 PM Greg Kroah-Hartman
> > <gregkh@linuxfoundation.org> wrote:
> >> On Wed, Jun 16, 2021 at 02:47:15PM +0800, Liu Shixin wrote:
> >>> Hi, Suren,
> >>>
> >>> I read the previous discussion about fixing CVE-2020-29374 in stable 4.14 and 4.19 in
> >>> <https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/>
> >>>
> >>> https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/
> >>>
> >>> And the results of the discussion is that you backports of 17839856fd58 for 4.14 and
> >>>
> >>> 4.19 kernels.
> >>>
> >>> But the bug about dax and strace in the discussion has not been solved, right? I don't
> >>>
> >>> find a conclusion on this issue, am I missing something? Does this problem still exist in
> >>>
> >>> the stable 4.14 and 4.19 kernel?
> > That is my understanding after discussions with Andrea but I did not
> > verify that myself. As Greg pointed out, the best way would be to try
> > it out.
> > Thanks,
> > Suren.
> >
> >> As the code is all there for you, can you just test them and see for
> >> yourself?
> >>
> >> thanks,
> >>
> >> greg k-h
> > .
> >
> Thank you both for replies. I have tested it in stable 4.19 kernel and the bug is existed as expected.

Great, can you provide a working backport of the patches needed to solve
this for 4.19 so that we can apply them?

thanks,

greg k-h

Re: Questions about backports of fixes for "CoW after fork() issue"

[Vlastimil Babka]

On 6/16/21 11:36 AM, Greg Kroah-Hartman wrote:
> On Wed, Jun 16, 2021 at 05:28:54PM +0800, Liu Shixin wrote:
>> On 2021/6/16 15:11, Suren Baghdasaryan wrote:
>> > On Tue, Jun 15, 2021 at 11:52 PM Greg Kroah-Hartman
>> > <gregkh@linuxfoundation.org> wrote:
>> >> On Wed, Jun 16, 2021 at 02:47:15PM +0800, Liu Shixin wrote:
>> >>> Hi, Suren,
>> >>>
>> >>> I read the previous discussion about fixing CVE-2020-29374 in stable 4.14 and 4.19 in
>> >>> <https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/>
>> >>>
>> >>> https://lore.kernel.org/linux-mm/20210401181741.168763-1-surenb@google.com/
>> >>>
>> >>> And the results of the discussion is that you backports of 17839856fd58 for 4.14 and
>> >>>
>> >>> 4.19 kernels.
>> >>>
>> >>> But the bug about dax and strace in the discussion has not been solved, right? I don't
>> >>>
>> >>> find a conclusion on this issue, am I missing something? Does this problem still exist in
>> >>>
>> >>> the stable 4.14 and 4.19 kernel?
>> > That is my understanding after discussions with Andrea but I did not
>> > verify that myself. As Greg pointed out, the best way would be to try
>> > it out.
>> > Thanks,
>> > Suren.
>> >
>> >> As the code is all there for you, can you just test them and see for
>> >> yourself?
>> >>
>> >> thanks,
>> >>
>> >> greg k-h
>> > .
>> >
>> Thank you both for replies. I have tested it in stable 4.19 kernel and the bug is existed as expected.

If you can reproduce it, great. That means a root cause can be found and fixed,
hopefully in a minimal way.

> Great, can you provide a working backport of the patches needed to solve
> this for 4.19 so that we can apply them?

We probably don't want to blindly backport the upstream patches (that also fixed
dax+ptrace as a side-effect) because they changed the semantics a lot and led to
further fixes, which is IMHO too risky to do now in stable. Linus also thought so:

https://lore.kernel.org/linux-mm/CAHk-=whUKYdWbKfFzXXnK8n04oCMwEgSnG8Y3tgE=YZUjiDvbA@mail.gmail.com/#t

> thanks,
> 
> greg k-h
>