Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20190704195555.580363209@infradead.org> <20190704200050.534802824@infradead.org>
 <CAHk-=wiJ4no+TW-8KTfpO-Q5+aaTGVoBJzrnFTvj_zGpVbrGfA@mail.gmail.com> <20190705134916.GU3402@hirez.programming.kicks-ass.net>
In-Reply-To: <20190705134916.GU3402@hirez.programming.kicks-ass.net>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Sat, 6 Jul 2019 14:41:22 -0700
Message-ID: <CAHk-=whsgA+8XtqJY91gqHhh9xLYQLM3kLLFTby=uf2eoZyK7Q@mail.gmail.com>
Subject: Re: [PATCH v2 5/7] x86/mm, tracing: Fix CR2 corruption
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     Thomas Gleixner <tglx@linutronix.de>,
        Borislav Petkov <bp@alien8.de>, Ingo Molnar <mingo@kernel.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        Andrew Lutomirski <luto@kernel.org>,
        Peter Anvin <hpa@zytor.com>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Juergen Gross <jgross@suse.com>,
        Linux List Kernel Mailing <linux-kernel@vger.kernel.org>,
        He Zhe <zhe.he@windriver.com>,
        Joel Fernandes <joel@joelfernandes.org>, devel@etsukata.com
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Jul 5, 2019 at 6:50 AM Peter Zijlstra <peterz@infradead.org> wrote:
>
> Also; all previous attempts at fixing this have been about pushing the
> read_cr2() earlier; notably:
>
>   0ac09f9f8cd1 ("x86, trace: Fix CR2 corruption when tracing page faults")
>   d4078e232267 ("x86, trace: Further robustify CR2 handling vs tracing")

I think both of those are because people - again - felt like tracing
can validly corrupt CPU state, and then they fix up the symptoms
rather than the cause.

Which I disagree with.

> And I'm thinking that with exception of this patch, the rest are
> worthwhile cleanups regardless.

I don't have any issues with the patches themselves, I agree that they
are probably good on their own.

I *do* have issues with the "tracing can change CPU state so we need
to deal with it" model, though. I think that mode of thinking is
wrong. I don't believe tracing should ever change core CPU state and
that be considered ok.

> Also; while looking at this, if we do continue with the C wrappers from
> the very last patch, we can do horrible things like this on top and move
> the read_cr2() back into C code.

Again, I don't think that is the problem. I think it's a much more
fundamental problem in thinking that core code should be changed
because tracing is broken garbage and didn't do things right.

I see Eiichi's patch, and it makes me go "that looks better" - simply
because it fixes the fundamental issue, rather than working around the
symptoms.

               Linus