Re: [PATCH v17 07/10] mm: introduce memfd_secret system call to create "secret" memory areas

[Mike Rapoport <rppt@kernel.org>]

On Thu, Feb 11, 2021 at 01:30:42PM +0100, Michal Hocko wrote:
> On Thu 11-02-21 13:20:08, Mike Rapoport wrote:
> [...]
> > Sealing is anyway controlled via fcntl() and I don't think
> > MFD_ALLOW_SEALING makes much sense for the secretmem because it is there to
> > prevent rogue file sealing in tmpfs/hugetlbfs.
> 
> This doesn't really match my understanding. The primary usecase for the
> sealing is to safely and predictably coordinate over shared memory. I
> absolutely do not see why this would be incompatible with an additional
> requirement to unmap the memory from the kernel to prevent additional
> interference from the kernel side. Quite contrary it looks like a very
> nice extension to this model.

I didn't mean that secretmem should not support sealing. I meant that
MFD_ALLOW_SEALING flag does not make sense. Unlike tmpfs, the secretmem fd
does not need protection from somebody unexpectedly sealing it.

> > As for the huge pages, I'm not sure at all that supporting huge pages in
> > secretmem will involve hugetlbfs.
> 
> Have a look how hugetlb proliferates through our MM APIs. I strongly
> suspect this is strong signal that this won't be any different.
> 
> > And even if yes, adding SECRETMEM_HUGE
> > flag seems to me less confusing than saying "from kernel x.y you can use
> > MFD_CREATE | MFD_SECRET | MFD_HUGE" etc for all possible combinations.
> 
> I really fail to see your point. This is a standard model we have. It is
> quite natural that flags are added. Moreover adding a new syscall will
> not make it any less of a problem.

Nowadays adding a new syscall is not as costly as it used to be. And I
think it'll provide better extensibility when new features would be added
to secretmem. 

For instance, for creating a secretmem fd backed with sealing we'd have

	memfd_secretm(SECRETMEM_HUGE);

rather than

	memfd_create(MFD_ALLOW_SEALING | MFD_HUGETLB | MFD_SECRET);


Besides, if we overload memfd_secret we add complexity to flags validation
of allowable flag combinations even with the simplest initial
implementation.
And what it will become when more features are added to secretmem?
 
> > > I by no means do not insist one way or the other but from what I have
> > > seen so far I have a feeling that the interface hasn't been thought
> > > through enough.
> > 
> > It has been, but we have different thoughts about it ;-)
> 
> Then you must be carrying a lot of implicit knowledge which I want you
> to document.

I don't have any implicit knowledge, we just have a different perspective.

-- 
Sincerely yours,
Mike.
_______________________________________________
Linux-nvdimm mailing list -- linux-nvdimm@lists.01.org
To unsubscribe send an email to linux-nvdimm-leave@lists.01.org