Re: [PATCHv13 05/16] x86/uaccess: Provide untagged_addr() and remove tags before address check

From: Linus Torvalds <torvalds@linux-foundation.org>
To: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	Andy Lutomirski <luto@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	x86@kernel.org, Kostya Serebryany <kcc@google.com>,
	Andrey Ryabinin <ryabinin.a.a@gmail.com>,
	Andrey Konovalov <andreyknvl@gmail.com>,
	Alexander Potapenko <glider@google.com>,
	Taras Madan <tarasmadan@google.com>,
	Dmitry Vyukov <dvyukov@google.com>,
	"H . J . Lu" <hjl.tools@gmail.com>,
	Andi Kleen <ak@linux.intel.com>,
	Rick Edgecombe <rick.p.edgecombe@intel.com>,
	Bharata B Rao <bharata@amd.com>,
	Jacob Pan <jacob.jun.pan@linux.intel.com>,
	Ashok Raj <ashok.raj@intel.com>,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCHv13 05/16] x86/uaccess: Provide untagged_addr() and remove tags before address check
Date: Sat, 7 Jan 2023 09:28:10 -0800	[thread overview]
Message-ID: <CAHk-=wjUcjPboro_P5qeFuuwAyqPEDRH8BvX0UFU1hNRfx+6Kw@mail.gmail.com> (raw)
In-Reply-To: <20230107091027.xbikgiizkeegofdd@box.shutemov.name>

On Sat, Jan 7, 2023 at 1:10 AM Kirill A. Shutemov <kirill@shutemov.name> wrote:
>
> On Fri, Dec 30, 2022 at 04:42:05PM -0800, Linus Torvalds wrote:
> > in ex_handler_uaccess() for the GP trap that users can now cause by
> > giving a non-canonical address with the high bit clear. So we'd
> > probably just want a new EX_TYPE_* for these cases, but that still
> > looks fairly straightforward.
>
> Plain _ASM_EXTABLE() seems does the trick.

Ack, for some reason I stupidly thought we'd have to change the
_ASM_EXTABLE_UA logic.

Thanks for setting me straight.

> Here's what I've come up with:

This looks good to me. And I like how you've used assembler macros
instead of the C preprocessor, it makes things more readable.

I'm personally so unused to asm macros that I never use them (and the
same is obviously true of Christoph who did that previous task size
thing), but I can appreciate others doing a better job at it.

So ack on this from me (I assume you tested it - hopefully even with
LAM), but maybe the x86 maintainers disagree violently?

The one possible downside is that *if* somebody passes non-valid user
addresses to get/put_user() intentionally (expecting an EFAULT), we
will now handle that much more slowly with a fault. But it would have
to be some really crazy use-case, and the normal case should be
simpler and faster.

But honestly, to me the upside is mainly "no need to worry about LAM
masking in asm code".

               Linus