Re: [PATCH v4 05/15] mm: introduce execmem_alloc() and execmem_free()

From: Song Liu
Date: Thu Apr 18 2024 - 12:13:55 EST


On Thu, Apr 18, 2024 at 8:37 AM Mike Rapoport <rppt@xxxxxxxxxx> wrote:
>
[...]
> >
> > Is +/- 2G enough for all realistic use cases? If so, I guess we don't
> > really need
> > EXECMEM_ANYWHERE below?
> >
> > > >
> > > > * I'm not sure about BPF's requirements; it seems happy doing the same as
> > > > modules.
> > >
> > > BPF are happy with vmalloc().
> > >
> > > > So if we *must* use a common execmem allocator, what we'd reall want is our own
> > > > types, e.g.
> > > >
> > > > EXECMEM_ANYWHERE
> > > > EXECMEM_NOPLT
> > > > EXECMEM_PREL32
> > > >
> > > > ... and then we use those in arch code to implement module_alloc() and friends.
> > >
> > > I'm looking at execmem_types more as definition of the consumers, maybe I
> > > should have named the enum execmem_consumer at the first place.
> >
> > I think looking at execmem_type from consumers' point of view adds
> > unnecessary complexity. IIUC, for most (if not all) archs, ftrace, kprobe,
> > and bpf (and maybe also module text) all have the same requirements.
> > Did I miss something?
>
> It's enough to have one architecture with different constrains for kprobes
> and bpf to warrant a type for each.
>

AFAICT, some of these constraints can be changed without too much work.

> Where do you see unnecessary complexity?
>
> > IOW, we have
> >
> > enum execmem_type {
> > EXECMEM_DEFAULT,
> > EXECMEM_TEXT,
> > EXECMEM_KPROBES = EXECMEM_TEXT,
> > EXECMEM_FTRACE = EXECMEM_TEXT,
> > EXECMEM_BPF = EXECMEM_TEXT, /* we may end up without
> > _KPROBE, _FTRACE, _BPF */
> > EXECMEM_DATA, /* rw */
> > EXECMEM_RO_DATA,
> > EXECMEM_RO_AFTER_INIT,
> > EXECMEM_TYPE_MAX,
> > };
> >
> > Does this make sense?
>
> How do you suggest to deal with e.g. riscv that has separate address spaces
> for modules, kprobes and bpf?

IIUC, modules and bpf use the same address space on riscv, while kprobes use
vmalloc address. I haven't tried this yet, but I think we can let
kprobes use the
same space as modules and bpf, which is:

ffffffff00000000 | -4 GB | ffffffff7fffffff | 2 GB | modules, BPF

Did I get this right?

Thanks,
Song