Re: [PATCH v8 14/21] mm/mmap: Avoid zeroing vma tree in mmap_region()

[Lorenzo Stoakes <lorenzo.stoakes@oracle.com>]

On Tue, Oct 01, 2024 at 12:42:33PM GMT, Bert Karwatzki wrote:
> Am Dienstag, dem 01.10.2024 um 11:33 +0100 schrieb Lorenzo Stoakes:
> >
> >
> > OK so what is less perfect is this looks very genuine and present in rc1 so
> > yeah.
> >
> > Thanks so much for helping out here - I may ask you to try some patches if
> > that's ok at some point?
> >
> > Cheers, Lorenzo
> >
> Yes, that would be ok. As a sidenote, when I first got this bug I conducted an
> internet search and found this vaguely similar report:
> https://lkml.org/lkml/2024/8/28/1558
> The report says its from 6.10.0, that would be not related, but at least from
> the date it's possible that it could have tested the v7 patchset.
>
> Bert Karwatzki
> >

Thanks it looks similar but as I noted there, I couldn't repro it [0], the sender
sent a bunch of seemingly bogus reports in a flurry and it seemed to be on exit
whwere the issue occurred rather than anything the change did.

Of course that could be due to maple tree corruption that only got flagged on
exit, similar to your situation.

Unfortunately we got no further information or details and the report was sparse
so I don't think it hugely helps us.

Thankfully here we have a repro-able situation (albeit, painful, sorry about
that!) so we should be able to figure this out.

[0]:https://lore.kernel.org/linux-mm/e50f9da1-521d-46ae-9e12-d82c4c093223@lucifer.local/