On Wed, Nov 14, 2018 at 09:23:41PM +0200, Ahmed Soliman wrote: > Hello, > > > > I couldn't reproduce this on a Westmere. Are you sure you're testing > > > a clean compilation? Can you bisect the kernel? > I don't do a make clean normally, but I will do it this time when > bisecting, Sorry, I meant a clean build of the kvm selftests, not the kernel. The kvm selftests makefile is a bit lacking and I've noticed that tests can assert in random places if they aren't built from scratch - ensuring all changed dependencies also get recompiled. Fixing up the makefile is on my todo, but it wasn't very high on it, since 'make clean && make' is fast enough. > also I only use shallow > clones so it will also take some time pulling. Also to note, The arch > I am using is Haswell, I am not > sure if that should make any difference though. I'm not sure either, and unfortunately I don't have the same hardware as you to try and reproduce. > > What would be the difference in CONFIG_* fragments that you both have enabled? > > Here is my config file: > https://pastebin.com/fCNV2z8c I checked the differences but nothing interesting stood out. I'm using a defconfig config with a couple extra file system modules tacked on. drew
