From mboxrd@z Thu Jan 1 00:00:00 1970 From: Steven Rostedt Subject: Re: [PATCH] KVM guest: prevent tracing recursion with kvmclock Date: Tue, 15 Nov 2011 09:01:05 -0500 Message-ID: <1321365665.4181.11.camel@frodo> References: <1321362075-27115-1-git-send-email-avi@redhat.com> <1321364308.4181.7.camel@frodo> <4EC26CFE.8010406@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: kvm@vger.kernel.org, Gleb Natapov To: Avi Kivity Return-path: Received: from hrndva-omtalb.mail.rr.com ([71.74.56.124]:33972 "EHLO hrndva-omtalb.mail.rr.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755532Ab1KOOBI (ORCPT ); Tue, 15 Nov 2011 09:01:08 -0500 In-Reply-To: <4EC26CFE.8010406@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: On Tue, 2011-11-15 at 15:45 +0200, Avi Kivity wrote: > What about function traces? Will any noninlined calls cause the same > problem? > Both function_graph and function tracing have recursion protection around the users of the call. function tracing is much lighter weight than function_graph tracing. The problem here was that function_graph tracing does some accounting, and calls the clock handler outside the recursion protection. I need to add recursion protection around the accounting too. But I still want to warn when it happens as recursion protection protects from crashing, but recursion is still bad because it still goes through a bit of arch specific setup before the recursion is detected, slowing function_graph tracing down much more than function tracing when it recurses. -- Steve