Date: Sun, 16 Dec 2012 23:57:29 +0200 From: Alexander Motin <mav@FreeBSD.org> To: Adrian Chadd <adrian@freebsd.org> Cc: svn-src-projects@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r244287 - projects/calloutng/sys/x86/isa Message-ID: <50CE43C9.4030809@FreeBSD.org> In-Reply-To: <CAJ-VmomZC1aatO3zREWc0VGPXhmUcCB9GqVF0hqS=%2BVP2xL3Ag@mail.gmail.com> References: <201212161116.qBGBGEwn063983@svn.freebsd.org> <CAJ-VmonYuh4dTwY9PjBmE4uOq8nNAL_kDKXpi6knwvc99PqJcw@mail.gmail.com> <50CE009B.7010804@FreeBSD.org> <CAJ-Vmok7rdooCifWmNw2LJ95BSuOwwz6wPeq86x8AcAp7FzuJw@mail.gmail.com> <50CE087A.2020200@FreeBSD.org> <CAJ-VmomZC1aatO3zREWc0VGPXhmUcCB9GqVF0hqS=%2BVP2xL3Ag@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 16.12.2012 23:06, Adrian Chadd wrote: > What we need to do is get you and David a bunch of that older embedded > i386 class hardware to test against. > > I can organise some of that to appear in a month or so - soekris > hardware and alix hardware to start with. > > I'd also suggest testing on vmware/virtualbox/qemu/parallels installs > where you explicitly enable the legacy timer hardware and see what the > behaviour is. Testing and new hardware is always great, but one-shot timers are already working in FreeBSD for quite a while since FreeBSD 9.0. The only change to the low-level timer driver in this branch was mentioned here and it is irrelevant for now. Most of work done in this branch is machine-independent, and it is not very important on which hardware to test it. But I am still trying to test on as many kinds as I can just to be safe. Testing under VM is a good idea, as it may have very specific timing characteristics. Still IMO much more interesting would be to try some more different workloads. If you can propose some workloads that actively use callouts, it would be interesting to try them. I've tried mentioned disk and TCP I/O and got no problems. In both cases profiler shows new callouts consuming no more then couple percents of CPU time, that is not bad. Synthetic tests are also doing well. Me and Davide are now quite satisfied with the present state of the code. Unless something significant will be found/reported, I think we could commit it somewhere closer to the end of the week. -- Alexander Motin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?50CE43C9.4030809>