Re: dm ioctl: Restore __GFP_HIGH in copy_params()

From: Mike Snitzer
Date: Mon May 22 2017 - 19:36:03 EST


On Mon, May 22 2017 at 4:35pm -0400,
David Rientjes <rientjes@xxxxxxxxxx> wrote:

> On Mon, 22 May 2017, Mike Snitzer wrote:
>
> > > > The lvm2 was designed this way - it is broken, but there is not much that
> > > > can be done about it - fixing this would mean major rewrite. The only
> > > > thing we can do about it is to lower the deadlock probability with
> > > > __GFP_HIGH (or PF_MEMALLOC that was used some times ago).
> >
> > Yes, lvm2 was originally designed to to have access to memory reserves
> > to ensure forward progress. But if the mm subsystem has improved to
> > allow for the required progress without lvm2 trying to stake a claim on
> > those reserves then we'll gladly avoid (ab)using them.
> >
>
> There is no such improvement to the page allocator when allocating at
> runtime. A persistent amount of memory in a mempool could be set aside as
> a preallocation and unavailable from the rest of the system forever as an
> alternative to dynamically allocating with memory reserves, but that has
> obvious downsides. This patch is the exact right thing to do.
>
> > > But let me repeat. GFP_KERNEL allocation for order-0 page will not fail.
> >
> > OK, but will it be serviced immediately? Not failing isn't useful if it
> > never completes.
> >
>
> No, and you can use __GFP_HIGH, which this patch does, to have a
> reasonable expectation of forward progress in the very near term.
>
> > While adding the __GFP_NOFAIL flag would serve to document expectations
> > I'm left unconvinced that the memory allocator will _not fail_ for an
> > order-0 page -- as Mikulas said most ioctls don't need more than 4K.
>
> __GFP_NOFAIL would make no sense in kvmalloc() calls, ever, it would never
> fallback to vmalloc :)
>
> I'm hoping this can get merged during the 4.12 window to fix the broken
> commit d224e9381897.

I've added your Acked-by and staged it for 4.12, please see:
https://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm.git/commit/?h=for-4.12/dm&id=8c1e2162f27b319da913683143c0c6c09b083ebb

Not sure when I'll send it to Linus but certainly no later than for rc4
inclusion.

Thanks,
Mike