Date: Tue, 7 Oct 2008 09:32:27 +0530 From: "Joseph Koshy" <joseph.koshy@gmail.com> To: "Ryan Stone" <rysto32@gmail.com> Cc: freebsd-current@freebsd.org Subject: Re: pmcstat: should it fail when a cpu is disabled? Message-ID: <84dead720810062102r2baf0e32y10594d3f1d07120d@mail.gmail.com> In-Reply-To: <bc2d970810061328p794782aaxed78dc1a32077ec6@mail.gmail.com> References: <bc2d970810061328p794782aaxed78dc1a32077ec6@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> Recently I've run into a small issue where pmcstat will fail if a cpu > is disabled unless you specifically tell pmcstat to not attach to that > processor. This is annoying for me because we have some machines with > hyper-threaded cpus but we disable hyperthreading on them. When we > try to run pmcstat on those machines, pmcstat will try to attach to > the disabled hyperthreaded cpu and fail. I can work around it by > figuring out what the active cpus are and specifying only those cpus > to the -c options, but that's annoying. I was thinking that if the -c > '*' option is specified(either explicitly or implicitly by not > specifying any -c option) that pmcstat should only attach to active > cpus and ignore disabled cpus. I have a patch that will do this but I > wanted to gauge opinions on whether this behaviour is wanted or not > before submitting it. Any strong opinions one way or the other? In -current and 7.X, pmcstat(8) checks sysctl "machdep.hlt_cpus" and should not by default attempt to allocate PMCs on halted CPUs. Which version of FreeBSD are you running? Koshy
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?84dead720810062102r2baf0e32y10594d3f1d07120d>