Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
From:   David Laight <David.Laight@ACULAB.COM>
To:     'Thomas Gleixner' <tglx@linutronix.de>,
        Peter Zijlstra <peterz@infradead.org>
CC:     "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Andy Lutomirski <luto@kernel.org>,
        "x86@kernel.org" <x86@kernel.org>,
        Andrey Ryabinin <aryabinin@virtuozzo.com>,
        Alexander Potapenko <glider@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>,
        "linux-mm@kvack.org" <linux-mm@kvack.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: RE: [RFCv2 05/10] x86/mm: Provide untagged_addr() helper
Thread-Topic: [RFCv2 05/10] x86/mm: Provide untagged_addr() helper
Thread-Index: AQHYZlYbkN98ikiNG0u1ah6OjFcYAq0ck+jA
Date:   Fri, 13 May 2022 10:14:23 +0000
Message-ID: <6f206d410f5b49789e986166ea473a6a@AcuMS.aculab.com>
References: <20220511022751.65540-1-kirill.shutemov@linux.intel.com>
 <20220511022751.65540-7-kirill.shutemov@linux.intel.com>
 <87a6bmx5lt.ffs@tglx> <Yn0YdPNG/Q3lf+4G@hirez.programming.kicks-ass.net>
 <87sfpevl1g.ffs@tglx> <87wneqtkb8.ffs@tglx>
In-Reply-To: <87wneqtkb8.ffs@tglx>
Accept-Language: en-GB, en-US
MIME-Version: 1.0
Content-Language: en-US
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8BIT
Precedence: bulk

From: Thomas Gleixner
> Sent: 13 May 2022 00:15
...
> But whatever we chose, it's sad, that we need to have support for
> interfaces which swallow any pointer (user or kernel) because otherwise
> this really boils down to a single OR resp. AND operation plus the
> according mov to retrieve the mask.

Are there any of those left?
Most will have gone with setfs(KERNEL_DS) removal.
Almost all code has to know whether an address is user
or kernel - the value can't be used because of architectures
that use the same address values in user and kernel.

How often do addresses actually need de-tagging?
Probably only code that is looking for page table
entries for virtual addresses?
How often does that happen for user addresses?

If the hardware is ignoring the bits then you don't
need to remove them before memory accesses.
That would include all userspace accesses.
Clearly access_ok() has to work with tagged addresses,
but that doesn't require the tag be masked off.
It can just check the transfer doesn't cross 1u<<63.
It (probably) just requires the fault handler to treat
non-canonical address faults as page faults.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)