[PATCH v3 00/15] Support Asynchronous Page Fault
From: Gavin Shan
Date: Fri May 07 2021 - 02:40:59 EST
There are two stages of page fault. The guest kernel is responsible for
handling stage-1 page fault, while the host kernel is to take care of the
stage-2 page fault. When the guest is trapped to host because of stage-2
page fault, the guest is suspended until the requested memory (page) is
populated. Sometimes, the cost to populate the requested page isn't cheap
and can take hundreds of milliseconds in extreme cases. Similarly, the
guest has to wait until the requested memory is ready in the scenario of
post-copy live migration.
This series introduces the feature (Asynchronous Page Fault) to improve
situation, so that the guest needn't have to wait in the scenarios. With
it, the overall performance is improved on the guest. This series depends
on the feature "SDEI virtualization" and QEMU changes. All code changes
can be found from github:
https://github.com/gwshan/linux ("kvm/arm64_sdei") # SDEI virtualization
https://github.com/gwshan/linux ("kvm/arm64_apf") # This series + "sdei"
https://github.com/gwshan/qemu ("kvm/arm64_apf") # QEMU code changes
About the design, the details can be found from last patch. Generally,
it's driven by two notifications: page-not-present and page-ready. They
are delivered from the host to guest via SDEI event and PPI separately.
In the mean while, each notification is always associated with a token,
used to identify the notification. The token is passed by the shared
memory between host/guest. Besides, the SMCCC and ioctl interface are
mitigated by VMM and guest to configure, enable, disable, even migrate
the functionality.
When the guest is trapped to host because of stage-2 page fault, a
page-not-present notification is raised by the host, and sent to the
guest through dedicated SDEI event (0x40400001) if the requested page
can't be populated immediately. In the mean while, a (background) worker
is also started to populate the requested page. On receiving the SDEI
event, the guest marks the current running process with special flag
(TIF_ASYNC_PF) and associates it with a pre-allocated waitqueue. At
same time, a (reschedule) IPI is sent to current CPU. After the SDEI
event is acknowledged by the guest, the (reschedule) IPI is delivered
and it causes context switch from that process tagged with TIF_ASYNC_PF
to another process.
Later on, a page-ready notification is sent to guest after the requested
page is populated by the (background) worker. On receiving the interrupt,
the guest uses the associated token to locate the process, which was
previously suspended because of page-not-present. The flag (TIF_ASYNC_PF)
is cleared for the suspended process and it's waken up.
The series is organized as below:
PATCH[01-04] makes the GFN hash table management generic so that it
can be shared by x86/arm64.
PATCH[05-06] preparatory work to support asynchronous page fault.
PATCH[07-08] support asynchronous page fault.
PATCH[09-11] support ioctl and SMCCC interfaces for the functionality.
PATHC[12-14] supoort asynchronous page fault for guest
PATCH[15] adds document to explain the design and internals
Testing
=======
The tests are taken using program "testsuite", which is written by myself.
The program basically does two things: (a) Starts a thread to allocate
all the available memory, write random values to them by the specified
times, and then release them; (b) The parallel thread is started to do
calculation if needed.
The VM has 4096MB memory and the QEMU process is put into memory cgroup,
which has 512MB memory limitation. It enforces memory swapping on host
side when the test program "testsuite" allocates all available memory
and writes random values to them on the guest side.
host# echo "never" > /sys/kernel/mm/transparent_hugepage/enabled
host# mkdir /sys/fs/cgroup/test \
host# echo "max" > /sys/fs/cgroup/test/memory.swap.max; \
echo "max" > /sys/fs/cgroup/test/memory.swap.high; \
echo "1024M" > /sys/fs/cgroup/test/memory.max; \
echo "1024M" > /sys/fs/cgroup/test/memory.high
host# echo $$ > /sys/fs/cgroup/test/cgroup.procs
host# /home/gavin/sandbox/qemu.main/build qemu-system-aarch64 \
-accel kvm -machine virt,gic-version=host \
-cpu host -smp 1,sockets=1,cores=1,threads=1 -m 4096M,maxmem=64G \
-object memory-backend-ram,id=mem0,size=4096M \
-numa node,nodeid=0,cpus=0,memdev=mem0 \
:
-kernel /home/gavin/sandbox/linux.guest/arch/arm64/boot/Image \
-initrd /home/gavin/sandbox/images/rootfs.cpio.xz
[1] Access all available memory without calculation thread
The performance is improved slightly because of the background swapin
introduced by asynchronous page fault.
Index -APF +APF Performance
-----------------------------------------------
1 61399ms 59911ms +0.24%
2 61121ms 59941ms +0.19%
3 60812ms 59728ms +0.17%
4 60749ms 59698ms +0.17%
5 60723ms 59678ms +0.17%
[2] Access all available memory with calculation thread
With the asynchronous page fault, the overall time to access all the
available memory is dropped by 40.2% (Index#1) in the cost of calculation
loss. It proves the asynchronous page fault is beneficial to system
responsiveness.
Index -APF Calculation +APF Calculation Performance
----------------------------------------------------------------------------
1 108392ms 8279539k 76k/ms 64751ms 4658761k 71k/ms +40.2% -43.7% -6.5%
2 107402ms 8160632k 75k/ms 64020ms 4520263k 70k/ms +40.3% -44.6% -6.6%
3 106854ms 8163024k 76k/ms 63654ms 4478215k 70k/ms +40.4% -45.1% -7.6%
4 106844ms 8156974k 76k/ms 63563ms 4461158k 70k/ms +40.5% -45.3% -7.6%
5 106769ms 8146449k 76k/ms 62885ms 4434800k 70k/ms +41.1% -45.5% -7.6%
[3] Live migration while accessing all available memory with calculation
thread
This is also beneficial to live migration. With the asynchronous page
fault is enabled, less time are needed to completee the live migration
and much more calculation capability is provided.
Index Item -APF +APF Performance
-----------------------------------------------------------------------------
1 test.used_time 59755ms 55708ms +6.7%
test.calculation/speed 4515595k 75k/ms 7049430k 126k/ms +68%
migrate.total_time 62275ms 63012ms -1.1%
migrate.setup_time 148ms 145ms +2.0%
migrate.down_time 38ms 83ms -118%
migrate.throughput 487mbps 589mbps +20.9%
2 test.used_time 59063ms 53123ms +10.0%
test.calculation/speed 4456194k 75k/ms 6755980k 127k/ms +69.3%
migrate.total_time 61555ms 57373ms +6.7%
migrate.setup_time 120ms 93ms +22.5%
migrate.down_time 47ms 30ms +36.1%
migrate.throughput 492mbps 608mbps +23.5%
Changelog
=========
v3:
* Rebase to v5.13.rc1 (Gavin)
* Drop patches from Will to detected SMCCC KVM service (Gavin)
* Retest and recapture the benchmarks (Gavin)
v2:
* Rebase to v5.11.rc6 (Gavin)
* Split the patches (James)
* Allocate "struct kvm_arch_async_control" dymaicall and use
it to check if the feature has been enabled. The kernel
option (CONFIG_KVM_ASYNC_PF) isn't used. (James)
* Add document to explain the design (James)
* Make GFN hash table management generic (James)
* Add ioctl commands to support migration (Gavin)
Gavin Shan (15):
KVM: async_pf: Move struct kvm_async_pf around
KVM: async_pf: Add helper function to check completion queue
KVM: async_pf: Make GFN slot management generic
KVM: x86: Use generic async PF slot management
KVM: arm64: Export kvm_handle_user_mem_abort()
KVM: arm64: Add paravirtualization header files
KVM: arm64: Support page-not-present notification
KVM: arm64: Support page-ready notification
KVM: arm64: Support async PF hypercalls
KVM: arm64: Support async PF ioctl commands
KVM: arm64: Export async PF capability
arm64: Detect async PF para-virtualization feature
arm64: Reschedule process on aync PF
arm64: Enable async PF
KVM: arm64: Add async PF document
Documentation/virt/kvm/arm/apf.rst | 143 +++++++
Documentation/virt/kvm/arm/index.rst | 1 +
arch/arm64/Kconfig | 11 +
arch/arm64/include/asm/esr.h | 6 +
arch/arm64/include/asm/kvm_emulate.h | 27 +-
arch/arm64/include/asm/kvm_host.h | 83 ++++
arch/arm64/include/asm/kvm_para.h | 37 ++
arch/arm64/include/asm/processor.h | 1 +
arch/arm64/include/asm/thread_info.h | 4 +-
arch/arm64/include/uapi/asm/Kbuild | 2 -
arch/arm64/include/uapi/asm/kvm.h | 19 +
arch/arm64/include/uapi/asm/kvm_para.h | 23 ++
arch/arm64/include/uapi/asm/kvm_sdei.h | 1 +
arch/arm64/kernel/Makefile | 1 +
arch/arm64/kernel/kvm.c | 452 +++++++++++++++++++++
arch/arm64/kernel/signal.c | 17 +
arch/arm64/kvm/Kconfig | 2 +
arch/arm64/kvm/Makefile | 1 +
arch/arm64/kvm/arm.c | 37 +-
arch/arm64/kvm/async_pf.c | 533 +++++++++++++++++++++++++
arch/arm64/kvm/hypercalls.c | 5 +
arch/arm64/kvm/mmu.c | 76 +++-
arch/arm64/kvm/sdei.c | 5 +
arch/x86/include/asm/kvm_host.h | 2 -
arch/x86/kvm/Kconfig | 1 +
arch/x86/kvm/mmu/mmu.c | 2 +-
arch/x86/kvm/x86.c | 88 +---
include/linux/arm-smccc.h | 15 +
include/linux/kvm_host.h | 72 +++-
include/uapi/linux/kvm.h | 3 +
virt/kvm/Kconfig | 3 +
virt/kvm/async_pf.c | 95 ++++-
virt/kvm/kvm_main.c | 4 +-
33 files changed, 1619 insertions(+), 153 deletions(-)
create mode 100644 Documentation/virt/kvm/arm/apf.rst
create mode 100644 arch/arm64/include/asm/kvm_para.h
create mode 100644 arch/arm64/include/uapi/asm/kvm_para.h
create mode 100644 arch/arm64/kernel/kvm.c
create mode 100644 arch/arm64/kvm/async_pf.c
--
2.23.0