On Sun, 2007-12-23 at 20:11 -0500, Rik van Riel wrote: > On Mon, 24 Dec 2007 04:29:36 +0530 > Balbir Singh <[EMAIL PROTECTED]> wrote: > > Rik van Riel wrote: > > > > In the real world, users with large JVMs on their servers, which > > > sometimes go a little into swap, can trigger this system. All of > > > the CPUs end up scanning the active list, and all pages have the > > > referenced bit set. Even if the system eventually recovers, it > > > might as well have been dead. > > > > > > Going into swap a little should only take a little bit of time. > > > > Very fascinating, so we need to scale better with larger memory. > > I suspect part of the answer will lie with using large/huge pages. > > Linus vetoed going to a larger soft page size, with good reason. > > Just look at how much the 64kB page size on PPC64 sucks for most > workloads - it works for PPC64 because people buy PPC64 monster > systems for the kinds of monster workloads that work well with a > large page size, but it definately isn't general purpose.
Indeed, machines already exist with >> 1TB of RAM, so even going to 1MB pages leaves these machines in trouble. Going to big pages a few years ago would have pushed the problem back a few years, but now we need real fixes. -- Mathematics is the supreme nostalgia of our time. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
