From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Wilk Subject: xc_tbuf_set_cpu_mask & xentrace & cpu_mask Date: Tue, 11 Mar 2014 10:10:53 -0400 Message-ID: <20140311141053.GA13345@phenom.dumpdata.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: george.dunlap@eu.citrix.com, xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org Hey George, I recently got my hands on a machine with 240CPUS (ugh!) and wanted to play a bit on figuring out what is happening with certain guests if I partition it the box in nice per-socket pools. While I was setting this up I started looking at xentrace and figuring out what to set the cpu_mask and then I realized it is limited to 64 cpus. Or at least that is my understanding (with not enough sleep - maybe I am missing something?). I see this xc_tbuf_set_cpu_mask which accepts a 32-bit mask and in the code I see: sysctl.u.tbuf_op.cpu_mask.nr_bits = sizeof(bytemap) * 8; where bytemap is a uint8. That all reads to me as a 64-bit mask (thought we only pass in a 32-bit mask?). Seems like two bugs there. The hypervisor part is OK - it does the right thing. Am I reading the code right? And if so, do you have some patches in your backlog that had addressed this by any chance? Thanks.