On Thu, Jan 24, 2002 at 08:22:51PM -0500, Mike A. Harris wrote: > On Mon, 21 Jan 2002, Tom Diehl wrote: > > >Both Alan Cox and Andrea Arcangeli among others have stated on lkmk that this > >is not a problem for linux. According to the lkml traffic gentoo did something > >weird to tweak this bug. > > It is not fully clear how much Linux is affected at this point, > other than that some people experiencing lockups who disable > usage of 4M pages, no longer experience lockups. I thik that if you will look at later discussions on linux-kernel list you will find out that this is a problem and it is even understood why and why 'mem=nopentium' helps. Initial explanations from AMD were not really correct thus creating an additional confusion. Actually lwn.net has a pretty concise and accurate summary of what is known at this moment. What is still not entirely clear is if using 4k pages really solves the issue or only makes it less likely and how to fix that properly - in particular taking into account all architectures which can be affected by that. Michal
