Re: [PATCH v4 2/2] arm64/bpf: don't allocate BPF JIT programs in module memory

[Will Deacon <will.deacon@arm.com>]

On Wed, Dec 05, 2018 at 01:24:17PM +0100, Daniel Borkmann wrote:
> On 12/04/2018 04:45 PM, Ard Biesheuvel wrote:
> > On Mon, 3 Dec 2018 at 13:49, Will Deacon <will.deacon@arm.com> wrote:
> >> On Fri, Nov 30, 2018 at 08:20:06PM +0100, Ard Biesheuvel wrote:
> >>> On Fri, 30 Nov 2018 at 19:26, Will Deacon <will.deacon@arm.com> wrote:
> >>>> On Fri, Nov 23, 2018 at 11:18:04PM +0100, Ard Biesheuvel wrote:
> >>>>> diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c
> >>>>> index a6fdaea07c63..76c2ab40c02d 100644
> >>>>> --- a/arch/arm64/net/bpf_jit_comp.c
> >>>>> +++ b/arch/arm64/net/bpf_jit_comp.c
> >>>>> @@ -940,3 +940,16 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog)
> >>>>>                                          tmp : orig_prog);
> >>>>>       return prog;
> >>>>>  }
> >>>>> +
> >>>>> +void *bpf_jit_alloc_exec(unsigned long size)
> >>>>> +{
> >>>>> +     return __vmalloc_node_range(size, PAGE_SIZE, BPF_JIT_REGION_START,
> >>>>> +                                 BPF_JIT_REGION_END, GFP_KERNEL,
> >>>>> +                                 PAGE_KERNEL_EXEC, 0, NUMA_NO_NODE,
> >>>>> +                                 __builtin_return_address(0));
> >>>>
> >>>> I guess we'll want VM_IMMEDIATE_UNMAP here if Rich gets that merged.
> >>>
> >>> I think akpm already queued up that patch.
> >>>
> >>>> In the
> >>>> meantime, I wonder if it's worth zeroing the region in bpf_jit_free_exec()?
> >>>> (although we'd need the size information...).
> >>>>
> >>>
> >>> Not sure. What exactly would that achieve?
> >>
> >> I think the zero encoding is guaranteed to be undefined, so it would limit
> >> the usefulness of any stale, executable TLB entries. However, we'd also need
> >> cache maintenance to make that stuff visible to the I side, so it's probably
> >> not worth it, especially if akpm has queued the stuff from Rich.
> >>
> >> Maybe just add an:
> >>
> >> /* FIXME: Remove this when VM_IMMEDIATE_UNMAP is supported */
> >> #ifndef VM_IMMEDIATE_UNMAP
> >> #define VM_IMMEDIATE_UNMAP      0
> >> #endif
> >>
> >> so we remember to come back and sort this out? Up to you.
> > 
> > I'll just make a note to send out that patch once the definition lands via -akpm
> 
> Could I get an ACK from you for this patch, then I'd take the series into bpf-next.

Gah, thanks for the ping: I thought I acked this initially, but turns out I
didn't.

Acked-by: Will Deacon <will.deacon@arm.com>

Will