From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: kvmmmu tracing Date: Tue, 24 Nov 2009 11:50:46 +0200 Message-ID: <4B0BAC76.3080202@redhat.com> References: <1258974375.7094.158.camel@johannes.local> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: kvm , linux-kernel , Steven Rostedt To: Johannes Berg Return-path: Received: from mx1.redhat.com ([209.132.183.28]:58615 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932491AbZKXJvR (ORCPT ); Tue, 24 Nov 2009 04:51:17 -0500 In-Reply-To: <1258974375.7094.158.camel@johannes.local> Sender: kvm-owner@vger.kernel.org List-ID: On 11/23/2009 01:06 PM, Johannes Berg wrote: > Commit f691fe1da7e2715137d21ae5a80bec64db4625db is really broken wrt. > the userspace interface for tracing because of the weird > KVM_MMU_PAGE_PRINTK macro. > > Can you explain what is wrong with it? > Maybe we should have a "unsafe for export" flag for events, if they do > strange things like that? > > As it stands, I can't use trace-cmd on an x86 machine that has kvm > enabled because it will try to read all the kvm stuff. Arguably, it > should only try to parse it when it needs it (i.e. not for me) but still > it's very inconvenient to export something to userspace that it cannot > possibly understand. > Is userspace reading mmutrace.h? When the structure attributes can be exported via /sys/kernel/debug/tracing? -- error compiling committee.c: too many arguments to function