On Mon, Oct 17, 2022 at 8:31 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
On Mon, 17 Oct 2022 16:17:31 +0800, Jason Wang <jasowang@redhat.com> wrote:
Adding Stefan.
On Mon, Oct 17, 2022 at 3:47 PM Xuan Zhuo <xuanzhuo@linux.alibaba.com> wrote:
Hello everyone,
# Background
Nowadays, there is a common scenario to accelerate communication between
different VMs and containers, including light weight virtual machine based
containers. One way to achieve this is to colocate them on the same host.
However, the performance of inter-VM communication through network stack is not
optimal and may also waste extra CPU cycles. This scenario has been discussed
many times, but still no generic solution available [1] [2] [3].
With pci-ivshmem + SMC(Shared Memory Communications: [4]) based PoC[5],
We found that by changing the communication channel between VMs from TCP to SMC
with shared memory, we can achieve superior performance for a common
socket-based application[5]:
- latency reduced by about 50%
- throughput increased by about 300%
- CPU consumption reduced by about 50%
Since there is no particularly suitable shared memory management solution
matches the need for SMC(See ## Comparison with existing technology), and virtio
is the standard for communication in the virtualization world, we want to
implement a virtio-ism device based on virtio, which can support on-demand
memory sharing across VMs, containers or VM-container. To match the needs of SMC,
the virtio-ism device need to support:
1. Dynamic provision: shared memory regions are dynamically allocated and
provisioned.
2. Multi-region management: the shared memory is divided into regions,
and a peer may allocate one or more regions from the same shared memory
device.
3. Permission control: The permission of each region can be set seperately.
Looks like virtio-ROCE
https://lore.kernel.org/all/20220511095900.343-1-xieyongji@bytedance.com/T/
and virtio-vhost-user can satisfy the requirement?
# Virtio ism device
ISM devices provide the ability to share memory between different guests on a
host. A guest's memory got from ism device can be shared with multiple peers at
the same time. This shared relationship can be dynamically created and released.
The shared memory obtained from the device is divided into multiple ism regions
for share. ISM device provides a mechanism to notify other ism region referrers
of content update events.
# Usage (SMC as example)
Maybe there is one of possible use cases:
1. SMC calls the interface ism_alloc_region() of the ism driver to return the
location of a memory region in the PCI space and a token.
2. The ism driver mmap the memory region and return to SMC with the token
3. SMC passes the token to the connected peer
3. the peer calls the ism driver interface ism_attach_region(token) to
get the location of the PCI space of the shared memory
# About hot plugging of the ism device
Hot plugging of devices is a heavier, possibly failed, time-consuming, and
less scalable operation. So, we don't plan to support it for now.
# Comparison with existing technology
## ivshmem or ivshmem 2.0 of Qemu
1. ivshmem 1.0 is a large piece of memory that can be seen by all devices that
use this VM, so the security is not enough.
2. ivshmem 2.0 is a shared memory belonging to a VM that can be read-only by all
other VMs that use the ivshmem 2.0 shared memory device, which also does not
meet our needs in terms of security.
## vhost-pci and virtiovhostuser
Does not support dynamic allocation and therefore not suitable for SMC.
I think this is an implementation issue, we can support VHOST IOTLB
message then the regions could be added/removed on demand.
1. After the attacker connects with the victim, if the attacker does not
dereference memory, the memory will be occupied under virtiovhostuser. In the
case of ism devices, the victim can directly release the reference, and the
maliciously referenced region only occupies the attacker's resources
Let's define the security boundary here. E.g do we trust the device ornot? If yes, in the case of virtiovhostuser, can we simple doVHOST_IOTLB_UNMAP then we can safely release the memory from theattacker.
Thanks, Jason:)
In our the design, there are several roles involved:
1) a virtio-ism-smc front-end driver
2) a Virtio-ism backend device driver and its associated vmm
3) a global device manager
4) a group of remote/peer virtio-ism backend devices/vmms
5) a group of remote/peer virtio-ism-smc front-end drivers
Among which , we treat 1, 2 and 3 as trusted, 4 and 5 as untrusted.
Because 4 and 5 are trusted, we can’t guarantee that IOTLB Invalidate requests have been executed as expected.
Say when disconnecting an SMC connection, a malicious peer may ignore the IOTLB invalidation request and keep access the shared memory region.
We have considered the IOTLB based design but encountered several issues:
1) It depends on the way to provision guest vm memory. We need a memory resource descriptor to support vhost-user IOTLB messages, thus can’t support anonymous memory based vm.
2) Lack of fine-grain access control of memory resource descriptor. When send a memory resource descriptor to an untrusted peer, we can’t enforce region based access control. Memfd supports file level seal operations, but still lack of region based permission control. Hugetlbfs based fd doesn’t support seal at all.
3) Lack of reliable way to reclaim granted access permissions from untrusted peers, as stated above.
4) How implement resource accounting. Say a vm has shared some memory regions from peers, and those peers exited unexpectedly, then those shared memory will be accounted to the victim vm, and may cause unexpected OOM.
Based on the above consideration, we adopted another design and introduced the device manager to solve above issues:
1) the device manager is the owner of memory buffers.
2) the device manager creates a memfd for each memory buffer/region, and configure SEALs according to requested access permissions.
3) When a guest vm reclaims a shared memory buffer, the device manager will provision a new memfd to the guest vm. And it will take the responsibility to reclaim the old buffer from peer and eventually release the old buffer.
4) Simplify the control communication channel. Every guest vm only needs to talk with the device manager and no need to discover and communicate with other peers.
Thanks,
Gerry
2. The ism device of a VM can be shared with multiple (1000+) VMs at the same
time, which is a challenge for virtiovhostuser
Please elaborate more the the challenges, anything makevirtiovhostuser different?
3. The sharing relationship of ism is dynamically increased, and virtiovhostuser
determines the sharing relationship at startup.
Not necessarily with IOTLB API?
4. For security issues, the device under virtiovhostuser may mmap more memory,
while ism only maps one region to other devices
With VHOST_IOTLB_MAP, the map could be done per region.Thanks
Thanks.
Thanks
# Design
This is a structure diagram based on ism sharing between two vms.
|-------------------------------------------------------------------------------------------------------------|
| |------------------------------------------------| |------------------------------------------------| |
| | Guest | | Guest | |
| | | | | |
| | ---------------- | | ---------------- | |
| | | driver | [M1] [M2] [M3] | | | driver | [M2] [M3] | |
| | ---------------- | | | | | ---------------- | | | |
| | |cq| |map |map |map | | |cq| |map |map | |
| | | | | | | | | | | | | | |
| | | | ------------------- | | | | -------------------- | |
| |----|--|----------------| device memory |-----| |----|--|----------------| device memory |----| |
| | | | ------------------- | | | | -------------------- | |
| | | | | | | |
| | | | | | | |
| | Qemu | | | Qemu | | |
| |--------------------------------+---------------| |-------------------------------+----------------| |
| | | |
| | | |
| |------------------------------+------------------------| |
| | |
| | |
| -------------------------- |
| | M1 | | M2 | | M3 | |
| -------------------------- |
| |
| HOST |
---------------------------------------------------------------------------------------------------------------
# POC code
Kernel: https://github.com/fengidri/linux-kernel-virtio-ism/commits/ism
Qemu: https://github.com/fengidri/qemu/commits/ism
If there are any problems, please point them out.
Hope to hear from you, thank you.
[1] https://projectacrn.github.io/latest/tutorials/enable_ivshmem.html
[2] https://dl.acm.org/doi/10.1145/2847562
[3] https://hal.archives-ouvertes.fr/hal-00368622/document
[4] https://lwn.net/Articles/711071/
[5] https://lore.kernel.org/netdev/20220720170048.20806-1-tonylu@linux.alibaba.com/T/
Xuan Zhuo (2):
Reserve device id for ISM device
virtio-ism: introduce new device virtio-ism
content.tex | 3 +
virtio-ism.tex | 340 +++++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 343 insertions(+)
create mode 100644 virtio-ism.tex
--
2.32.0.3.g01195cf9f
---------------------------------------------------------------------
To unsubscribe, e-mail: virtio-dev-unsubscribe@lists.oasis-open.org
For additional commands, e-mail: virtio-dev-help@lists.oasis-open.org
---------------------------------------------------------------------
To unsubscribe, e-mail: virtio-dev-unsubscribe@lists.oasis-open.org
For additional commands, e-mail: virtio-dev-help@lists.oasis-open.org