From mboxrd@z Thu Jan  1 00:00:00 1970
From: Joerg Roedel <jroedel@suse.de>
Subject: Re: [PATCH v4 63/75] x86/sev-es: Handle #DB Events
Date: Wed, 15 Jul 2020 12:08:08 +0200
Message-ID: <20200715100808.GL16200@suse.de>
References: <20200714120917.11253-1-joro@8bytes.org>
 <20200714120917.11253-64-joro@8bytes.org>
 <20200715084752.GD10769@hirez.programming.kicks-ass.net>
 <20200715091337.GI16200@suse.de>
 <20200715095136.GG10769@hirez.programming.kicks-ass.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <20200715095136.GG10769@hirez.programming.kicks-ass.net>
Sender: linux-kernel-owner@vger.kernel.org
To: Peter Zijlstra <peterz@infradead.org>
Cc: Joerg Roedel <joro@8bytes.org>, x86@kernel.org, hpa@zytor.com, Andy Lutomirski <luto@kernel.org>, Dave Hansen <dave.hansen@linux.intel.com>, Jiri Slaby <jslaby@suse.cz>, Dan Williams <dan.j.williams@intel.com>, Tom Lendacky <thomas.lendacky@amd.com>, Juergen Gross <jgross@suse.com>, Kees Cook <keescook@chromium.org>, David Rientjes <rientjes@google.com>, Cfir Cohen <cfir@google.com>, Erdem Aktas <erdemaktas@google.com>, Masami Hiramatsu <mhiramat@kernel.org>, Mike Stunes <mstunes@vmware.com>, Sean Christopherson <sean.j.christopherson@intel.com>, Martin Radev <martin.b.radev@gmail.com>, linux-kernel@vger.kernel.org, kvm@vger.kernel.org, virtualization@lists.linux-foundation.org
List-Id: virtualization@lists.linuxfoundation.org

On Wed, Jul 15, 2020 at 11:51:36AM +0200, Peter Zijlstra wrote:
> On Wed, Jul 15, 2020 at 11:13:37AM +0200, Joerg Roedel wrote:
> > Then my understanding of intrumentation_begin/end() is wrong, I thought
> > that the kernel will forbid setting breakpoints before
> > instrumentation_begin(), which is necessary here because a break-point
> > in the #VC handler might cause recursive #VC-exceptions when #DB is
> > intercepted.
> > Maybe you can elaborate on why this makes no sense?
> 
> Kernel avoids breakpoints in any noinstr text, irrespective of
> instrumentation_begin().
> 
> instrumentation_begin() merely allows one to call !noinstr functions.

Right, but the handler calls into various other functions. I actually
started to annotate them all with noinstr, but that was a can of worms
when calling into generic kernel functions. And the only problem with
intrumentation in the #VC handler is the #VC-for-#DB exit-code, so I
decided to only handle this one with instrumentation forbidden and allow
it for the rest of the handler.

Regards,

	Joerg