[PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-12 Thread Nico Pache
gitlab issue: https://gitlab.com/libvirt/libvirt/-/issues/79 The virtio-balloon device now has the ability to report free pages back to the hypervisor for reuse by other programs. This kernel feature allows for more stable and resource friendly systems. This feature is available in QEMU and is e

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread Daniel Henrique Barboza
On 10/12/20 8:35 PM, Nico Pache wrote: gitlab issue: https://gitlab.com/libvirt/libvirt/-/issues/79 The virtio-balloon device now has the ability to report free pages back to the hypervisor for reuse by other programs. This kernel feature allows for more stable and resource friendly systems.

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread Michal Privoznik
On 10/13/20 1:35 AM, Nico Pache wrote: gitlab issue: https://gitlab.com/libvirt/libvirt/-/issues/79 The virtio-balloon device now has the ability to report free pages back to the hypervisor for reuse by other programs. This kernel feature allows for more stable and resource friendly systems. T

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread Michal Privoznik
On 10/13/20 5:10 PM, Michal Privoznik wrote: On 10/13/20 1:35 AM, Nico Pache wrote: gitlab issue: https://gitlab.com/libvirt/libvirt/-/issues/79 The virtio-balloon device now has the ability to report free pages back to the hypervisor for reuse by other programs. Is this something that we mig

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread Nico Pache
I don't believe so, but David might have some more insight into that question. Once the feature is enabled the guest will report pages on its freelist to the hypervisor, and the host/hypervisor will then release them. I'm not exactly sure what information we can or would want to 'stat' out of thi

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread David Hildenbrand
On 13.10.20 19:00, Nico Pache wrote: > I don't believe so, but David might have some more insight into that > question.  > > Once the feature is enabled the guest will report pages on its freelist > to the hypervisor, and the host/hypervisor will then release them. We report (some, not all) free

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread Peter Krempa
On Tue, Oct 13, 2020 at 18:47:39 +0200, Michal Privoznik wrote: > On 10/13/20 5:10 PM, Michal Privoznik wrote: > > On 10/13/20 1:35 AM, Nico Pache wrote: > > > gitlab issue: https://gitlab.com/libvirt/libvirt/-/issues/79 > > > > > > The virtio-balloon device now has the ability to report free page

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread Michal Privoznik
On 10/13/20 7:57 PM, Peter Krempa wrote: On Tue, Oct 13, 2020 at 18:47:39 +0200, Michal Privoznik wrote: On 10/13/20 5:10 PM, Michal Privoznik wrote: On 10/13/20 1:35 AM, Nico Pache wrote: gitlab issue: https://gitlab.com/libvirt/libvirt/-/issues/79 The virtio-balloon device now has the abili

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread Nico Pache
IMO "return-pages" sounds the best out of those and stays relatively consistent with the kernel and qemu terminology for this feature. I personally don't see a huge problem with the current name, but I've also been staring at the words "free page reporting" for too long. Cheers! -- Nico On Wed,

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-13 Thread Michal Privoznik
On 10/14/20 8:07 AM, Nico Pache wrote: IMO "return-pages" sounds the best out of those and stays relatively consistent with the kernel and qemu terminology for this feature. I personally don't see a huge problem with the current name, but I've also been staring at the words "free page reportin

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-14 Thread David Hildenbrand
On 14.10.20 08:30, Michal Privoznik wrote: > On 10/14/20 8:07 AM, Nico Pache wrote: >> IMO "return-pages" sounds the best out of those and stays relatively >> consistent with the kernel and qemu terminology for this feature. >> >> I personally don't see a huge problem with the current name, but I'

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-14 Thread Michal Privoznik
On 10/14/20 10:26 AM, David Hildenbrand wrote: On 14.10.20 08:30, Michal Privoznik wrote: Sorry for hijacking this thread, but I need to report it somewhere (what is the best place?) When I want to start a guest with this feature turned on + memfd + hugepages, the host kernel prints this wa

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-14 Thread David Hildenbrand
On 14.10.20 13:53, Michal Privoznik wrote: > On 10/14/20 10:26 AM, David Hildenbrand wrote: >> On 14.10.20 08:30, Michal Privoznik wrote: > > Sorry for hijacking this thread, but I need to report it somewhere (what > is the best place?) > > When I want to start a guest with this feature turned o

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-14 Thread Michal Privoznik
On 10/14/20 2:06 PM, David Hildenbrand wrote: On 14.10.20 13:53, Michal Privoznik wrote: On 10/14/20 10:26 AM, David Hildenbrand wrote: On 14.10.20 08:30, Michal Privoznik wrote: No, not at all. Thanks for reporting! And the "bad" thing is, that QEMU doesn't do anything too fancy. All it

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-14 Thread David Hildenbrand
On 14.10.20 15:46, Michal Privoznik wrote: > On 10/14/20 2:06 PM, David Hildenbrand wrote: >> On 14.10.20 13:53, Michal Privoznik wrote: >>> On 10/14/20 10:26 AM, David Hildenbrand wrote: On 14.10.20 08:30, Michal Privoznik wrote: >>> >> >> >> No, not at all. Thanks for reporting! >> >> And t

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-14 Thread Michal Privoznik
On 10/14/20 4:07 PM, David Hildenbrand wrote: Note: prealloc=yes is a bad choice in this environment. It contradicts memory overcommit - what we want to optimize with free page reporting. You allocate all VM memory to throw it away once the guest is up. Is there an option to turn this of with

Re: [PATCH v2 0/4] qemu: Add support for free-page-reporting

2020-10-14 Thread David Hildenbrand
On 14.10.20 16:15, Michal Privoznik wrote: > On 10/14/20 4:07 PM, David Hildenbrand wrote: > >> >> >> Note: prealloc=yes is a bad choice in this environment. It >> contradicts memory overcommit - what we want to optimize with >> free page reporting. You allocate all VM memory to throw it away >> o