Messages in this thread | | | Date | Thu, 31 Aug 2023 10:55:29 +1000 | Subject | Re: [PATCH] virtio_balloon: Fix endless deflation and inflation on arm64 | From | Gavin Shan <> |
| |
On 8/31/23 02:30, David Hildenbrand wrote: > On 29.08.23 03:54, Gavin Shan wrote: >> The deflation request to the target, which isn't unaligned to the >> guest page size causes endless deflation and inflation actions. For >> example, we receive the flooding QMP events for the changes on memory >> balloon's size after a deflation request to the unaligned target is >> sent for the ARM64 guest, where we have 64KB base page size. >> >> /home/gavin/sandbox/qemu.main/build/qemu-system-aarch64 \ >> -accel kvm -machine virt,gic-version=host -cpu host \ >> -smp maxcpus=8,cpus=8,sockets=2,clusters=2,cores=2,threads=1 \ >> -m 1024M,slots=16,maxmem=64G \ >> -object memory-backend-ram,id=mem0,size=512M \ >> -object memory-backend-ram,id=mem1,size=512M \ >> -numa node,nodeid=0,memdev=mem0,cpus=0-3 \ >> -numa node,nodeid=1,memdev=mem1,cpus=4-7 \ >> : \ >> -device virtio-balloon-pci,id=balloon0,bus=pcie.10 >> >> { "execute" : "balloon", "arguments": { "value" : 1073672192 } } >> {"return": {}} >> {"timestamp": {"seconds": 1693272173, "microseconds": 88667}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272174, "microseconds": 89704}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272175, "microseconds": 90819}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272176, "microseconds": 91961}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272177, "microseconds": 93040}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073676288}} >> {"timestamp": {"seconds": 1693272178, "microseconds": 94117}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073676288}} >> {"timestamp": {"seconds": 1693272179, "microseconds": 95337}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272180, "microseconds": 96615}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073676288}} >> {"timestamp": {"seconds": 1693272181, "microseconds": 97626}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272182, "microseconds": 98693}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073676288}} >> {"timestamp": {"seconds": 1693272183, "microseconds": 99698}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272184, "microseconds": 100727}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272185, "microseconds": 90430}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> {"timestamp": {"seconds": 1693272186, "microseconds": 102999}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073676288}} >> : >> <The similar QMP events repeat> >> >> Fix it by having the target aligned to the guest page size, 64KB >> in this specific case. With this applied, no flooding QMP event >> is observed and the memory balloon's size can be stablizied to >> 0x3ffe0000 soon after the deflation request is sent. >> >> { "execute" : "balloon", "arguments": { "value" : 1073672192 } } >> {"return": {}} >> {"timestamp": {"seconds": 1693273328, "microseconds": 793075}, \ >> "event": "BALLOON_CHANGE", "data": {"actual": 1073610752}} >> { "execute" : "query-balloon" } >> {"return": {"actual": 1073610752}} >> >> Signed-off-by: Gavin Shan <gshan@redhat.com> >> --- >> drivers/virtio/virtio_balloon.c | 13 ++++++++++++- >> 1 file changed, 12 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_balloon.c >> index 5b15936a5214..625caac35264 100644 >> --- a/drivers/virtio/virtio_balloon.c >> +++ b/drivers/virtio/virtio_balloon.c >> @@ -386,6 +386,17 @@ static void stats_handle_request(struct virtio_balloon *vb) >> virtqueue_kick(vq); >> } >> +static inline s64 align_pages_up(s64 diff) >> +{ >> + if (diff == 0) >> + return diff; >> + >> + if (diff > 0) >> + return ALIGN(diff, VIRTIO_BALLOON_PAGES_PER_PAGE); >> + >> + return -ALIGN(-diff, VIRTIO_BALLOON_PAGES_PER_PAGE); >> +} >> + >> static inline s64 towards_target(struct virtio_balloon *vb) >> { >> s64 target; >> @@ -396,7 +407,7 @@ static inline s64 towards_target(struct virtio_balloon *vb) >> &num_pages); >> target = num_pages; >> - return target - vb->num_pages; > > We know that vb->num_pages is always multiples of VIRTIO_BALLOON_PAGES_PER_PAGE. > > Why not simply align target down? > > target = ALIGN(num_pages, VIRTIO_BALLOON_PAGES_PER_PAGE); > return target - vb->num_pages; >
Good point. Thanks a lot, David. The code will be changed to what's suggested in v2, to be posted soon. I will also add a comment to explain it a bit. Besides, ALIGN() is align-up instead of align-down to give bias to deflation intentionally, to avoid overrunning the machine's memory size if it's not aligned to 64KB. Further more, the align-up causes deflation even user requests a 4KB diff. However, the outcome of ALIGN_DOWN(4KB, 64KB) is zero and no deflation will be triggered.
Thanks, Gavin
| |