On Mon, Aug 06, 2018 at 12:17:39PM +0200, David Hildenbrand wrote: > On 31.07.2018 10:43, Christian Borntraeger wrote: > > Paolo, Radim, > > > > here are the s390 updates for KVM for 4.19. Most important change is > > the initial support for host large pages. As this touches KVM and > > s390/mm in an intermingled way we have created a topic branch that > > is merged here and in Martins tree. In this way all merge conflicts > > are nicely handled and avoided. > > > > > > The following changes since commit 6f0d349d922ba44e4348a17a78ea51b7135965b1: > > > > Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net (2018-06-25 15:58:17 +0800) > > > > are available in the Git repository at: > > > > git://git.kernel.org/pub/scm/linux/kernel/git/kvms390/linux.git tags/kvm-s390-next-4.19-1 > > > > for you to fetch changes up to 2375846193663a1282c0ef7093640ed3210dc09f: > > > > Merge tag 'hlp_stage1' of git://git.kernel.org/pub/scm/linux/kernel/git/kvms390/linux into kvms390/next (2018-07-30 23:20:48 +0200) > > > > Hi, > > We had an internal discussion and some Daniel (cc) wondered if we should > drop the hpage module parameter and instead glue this to the nested > parameter. > > E.g. nested=1 -> hpage cannot be enabled for a VM > nested=0 -> hpage can be enabled for a VM > > Are we ready to expose this feature as default to all VMs? Opinions? > > This means that nested=0 (default) environments will get hpage support > and hpage support cannot be disabled by an admin. NB an admin would still have to reserve huge pages for use at boot time, and explicitly configure QEMU guest to use them in libvirt XML, which is common with other architectures. > Benefit is that necessary setup to use huge pages is limited. My chief concern was to avoid extra s390-specific options as a prerequisite for using the feature. Most mgmt app developers focus on x86, so any case where other architecture diverge in behaviour from x86 makes it more likely that the feature will not be usable under s390. There's already 2 explicit steps required to enable use of huge pages with a KVM guest, so it wasn't clear to me what extra protection adding a 3rd hurdle brings. > Downside is, that this is somewhat hidden behind another parameter and > cannot be disabled. IIUC, eventually nested VMs would support huge pages too, so its only a short term limitation of nested=1 that this doesn't work with HP. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
