remove set_fs for riscv v2

[Palmer Dabbelt]

On Mon, 21 Sep 2020 21:37:52 PDT (-0700), Christoph Hellwig wrote:
> Given tht we've not made much progress with the common branch,
> are you fine just picking this up through the riscv tree for 5.10?
>
> I'll defer other architectures that depend on the common changes to
> 5.11 then.

I'm OK taking it, but there's a few things I'd like to sort out.  IIRC I put it
on a temporary branch over here

    https://git.kernel.org/pub/scm/linux/kernel/git/palmer/linux.git/log/?h=riscv-remove_set_fs

under the assumption it might get lost otherwise, but let me know if that's not
what you were looking for.

Arnd: Are you OK with the asm-generic stuff?  I couldn't find anything in my
mail history, so sorry if I just missed it.

Al: IIRC the plan here was to have me merge in a feature branch with this
stuff, but it'd have to be based on your for-next as there are some
dependencies over there.  I see 5ae4998b5d6f ("powerpc: remove address space
overrides using set_fs()") in vfs/for-next so I think we should be OK, but let
me know if I'm doing something wrong.

> On Wed, Sep 09, 2020 at 08:55:15AM +0200, Christoph Hellwig wrote:
>> now that we've sorted out a remaining issue base.set_fs should not
>> be rebased any more, so you could pull it into the riscv tree or a topic
>> branch.
>>
>> The first four patch should go into base.set_fs, though.  Arnd, can you
>> re-review the updated patches?
> ---end quoted text---