Re: [syzbot] [mm?] possible deadlock in upgrade_mmap_lock_carefully

From: Paul Moore
Date: Wed Oct 02 2024 - 22:52:56 EST


On Wed, Oct 2, 2024 at 9:32 PM syzbot
<syzbot+a6456f6334aa19425886@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> > On Wed, Oct 2, 2024 at 8:14 AM Shu Han <ebpqwerty472123@xxxxxxxxx> wrote:
> >>
> >> It seems to be the same as [1].
> >> New LSM hook position for remap_file_pages + IMA = deadlock.
> >> The new LSM hook position is added for a bypass caused by
> >> no check in remap_file_pages + READ_IMPLIES_EXEC in do_mmap.
> >
> > Thanks Shu Han, let's mark this as a dup and sort out a fix in the first report.
> >
> > #syz dup: [syzbot] [integrity?] [lsm?] possible deadlock in
>
> can't find the dup bug
>
> > process_measurement (4)

Stupid line wrapping, let's try again ...

#syz dup: possible deadlock in process_measurement (4)

> >> I suggest fix it by removing the check and moving READ_IMPLIES_EXEC
> >> out of do_mmap[2].
> >>
> >> Link: https://lore.kernel.org/lkml/20240928065620.7abadb2d8552f03d785c77c9@xxxxxxxxxxxxxxxxxxxx/
> >> [1]
> >> Link: https://lore.kernel.org/all/20240928180044.50-1-ebpqwerty472123@xxxxxxxxx/
> >> [2]

--
paul-moore.com