Multi core guest support #107
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The FS base now holds the address of the start of the per-core section, instead of the core id
This drastically increases the granularity of the locking in most aspects of the hypervisor
The new approach is just to use the VirtualMachineSet functions directly
|
@dlrobertson Have you had a chance to look at this any more? I know it sounded like there were issues with QEMU on machines with fewer than 6 cores. If so, we could probably just switch to 4 (2 guests with 2 cores or something). |
|
It was a little slow to start, but we can bump it down if we need to in another PR |
|
Sorry for taking so long to get back to this! I'm really excited to use this. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #86 #87
This is a more correct version of the work in #106. One of the big issues we need to overcome is that KVM does not permit a nested hypervisor to put a guest into a 'waiting-for-sipi' state. This means we can't just launch the VCPU in that state. Instead, we wait to launch the guest AP's until every core associated with a given VM is ready and they receive a signal from the guest BSP to start (at which point they are launch in a running state at the provided address).
In general, adding multi-core guest support required a more complete implementation of guest local apic emulation, particularly with respect to guest IPI support. To make some of this more clear, this MR makes a more explicit delineation between a GSI and an actual interrupt vector.
This also drastically reduces the coarseness of the locking. There is no longer a per-vm RwLock, instead the various shared components of the VM expose thread-safe interfaces, which allows VMEXITS to no longer be effectively serialized.