From mboxrd@z Thu Jan 1 00:00:00 1970 From: Suravee Suthikulpanit Subject: Re: Xen, oprofile, perf, PEBS, event counters, PVHVM, PV Date: Wed, 16 Jan 2013 11:40:50 -0600 Message-ID: <1358358050.2964.10.camel@Greyghost-ubuntu> References: <20130114204546.GA6706@phenom.dumpdata.com> <50F58DA5.5090809@citrix.com> <20130116044705.GB20668@phenom.dumpdata.com> <50F6D2C7.1020606@amd.com> Reply-To: suravee.suthikulpanit@amd.com Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <50F6D2C7.1020606@amd.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Boris Ostrovsky Cc: Marcus Granado , "xen-devel@lists.xensource.com" , "msw@amazon.com" , "andrew.thomas@oracle.com" , Konrad Rzeszutek Wilk , "jun.nakajima@intel.com" , "kurt.hackel@oracle.com" , "jacob.shin@amd.com" , Marcus Granado , "Marcos.Matsunaga@oracle.com" , "JBeulich@suse.com" , "Sherry.Hurwitz@amd.com" List-Id: xen-devel@lists.xenproject.org On Wed, 2013-01-16 at 11:18 -0500, Boris Ostrovsky wrote: > On 01/15/2013 11:47 PM, Konrad Rzeszutek Wilk wrote: > > On Tue, Jan 15, 2013 at 05:11:01PM +0000, Marcus Granado wrote: > >> On 14/01/13 20:45, Konrad Rzeszutek Wilk wrote: > > >>> And it occurs to me it could be possible be to make some inroads on making > >>> performance monitoring easier: > >>> > >>> 1). fix the glaring omissions in oprofile for the new CPUs > >>> 2). Add a register keyhandle to get some debug info. > >>> 3). piggyback on oprofile hypercalls and insert some bridge in perf (lots > >>> of handwaving here). Or perhaps emulate in the Linux kernel the > >>> wmsrs (so xen_safe_wrmsrs) and have the pvops kernel based on the MSRs > >>> make the hypercalls to setup the buffers, etc. > >>> > >>> 3a). new hypercalls? intercept rdmsr/wrmsrs and stuff the right data > >>> in the initial domain? Other thoughts? > >>> > >>> 4). Extend perf to have '--xen' so it can also look at the xen-hypervisor > >>> ELF file. > >> > >> 5) live event reports from xenoprof/opreport, ala perf top. > >> 6) ports of oprofile kernel modules for other oses (bsd, windows, > >> mirage), so that these oses can be used as active participants. > > (+ Suravee) > > What is the status of oprofile development in general (on Linux)? I > don't see a whole lot of activity there, it appears to be more in > maintenance mode. Linux OProfile driver is in maintenance mode. However, the user-space tools have been converted to use PERF syscalls interface which allows it to support per-process (non-root) profiling (similar to PERF default mode). One advantage of OProfile is the large database of performance counters on various platforms. Suravee > I realize that doing (4) may end up being a lot of work but given much > more active perf development it seems to me that this should be the > preferred route. > -boris