Received-SPF: pass (google.com: domain of linux-kernel+bounces-96366-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Precedence: bulk
MIME-Version: 1.0
References: <20240307133916.3782068-1-yosryahmed@google.com>
 <20240307133916.3782068-2-yosryahmed@google.com> <ba8a51fe-7b22-46b1-be5f-1e4c837d085c@intel.com>
 <ZeomquHvZs9-BKKK@google.com> <CAJD7tkZMiRDF4jXK6yQh+cZdC7uJsTMfAXrPaC0TkTFQCoe7Tg@mail.gmail.com>
 <56ea29c2-8545-4689-a418-eb7784613650@intel.com>
In-Reply-To: <56ea29c2-8545-4689-a418-eb7784613650@intel.com>
From: Yosry Ahmed <yosryahmed@google.com>
Date: Thu, 7 Mar 2024 15:37:04 -0800
Message-ID: <CAJD7tkaRV4APEYECL_rThr5ArWaLpUG+eBiwgxqe49N6sMRBbg@mail.gmail.com>
Subject: Re: [RFC PATCH 1/3] x86/mm: fix LAM cr3 mask inconsistency during
 context switch
To: Dave Hansen <dave.hansen@intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>, Thomas Gleixner <tglx@linutronix.de>, 
	Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>, Peter Zijlstra <peterz@infradead.org>, 
	Andy Lutomirski <luto@kernel.org>, "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>, x86@kernel.org, 
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"

> > Or we can avoid returning a value from the helper and avoid passing an
> > mm. The callers would be more verbose, they'll have to call
> > mm_lam_cr3_mask() and mm_untag_mask() and pass the results into the
> > helper (set_tlbstate_lam_mode() or update_cpu_tlbstate_lam()). Another
> > advantage of this is that we can move the READ_ONCE to
> > switch_mm_irqs_off() and keep the comment here.
>
>
> One thing I don't like about set_tlbstate_lam_mode() is that it's not
> obvious that it's writing to "cpu_tlbstate" and its right smack in the
> middle of a bunch of other writers to the same thing.
>
> But I'm also not sure that open-coding it at its three call sites makes
> things better overall.
>
> I honestly don't have any brilliant suggestions.

Let me ponder this a little bit and try to come up with something, I
think a max of renaming and open-coding could make an improvement.

Thanks!