Add support for KVM_SIGNAL_MSI #42
Conversation
|
@andreeaflorescu @sameo @rbradford I don't see how I can properly test this new ioctl... |
rbradford
left a comment
rbradford
left a comment
There was a problem hiding this comment.
I think testing such ioctls will often present challenges and we need to be tolerant of lowering the coverage where necessary. With such little logic in the handling of this ioctl I do not feel overly concerned that anything might break it.
|
@andreeaflorescu @rbradford PR updated, let me know what you think? |
|
@sboeuf need to update your coverage amount. |
|
Yeah so I lowered the coverage from 91.3 to 91.2 and now all green! |
src/ioctls/vm.rs
Outdated
| // exactly the size of the struct | ||
| let ret = unsafe { ioctl_with_ref(self, KVM_SIGNAL_MSI(), &msi) }; | ||
| if ret != -1 { | ||
| Ok(()) |
There was a problem hiding this comment.
As per KVM API documentation this ioctl:
Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
We should keep the same return values in this wrapper as well.
There was a problem hiding this comment.
should I create a new error for 0?
There was a problem hiding this comment.
I think you should return 0 and let the caller handle that exit code.
There was a problem hiding this comment.
P.S. It's worth specifying this in the function documentation as well.
There was a problem hiding this comment.
Ok it's updated now, let me know what you think.
sboeuf
force-pushed
the
kvm_signal_msi
branch
3 times, most recently
from
8f48374 to
f883bcc
Compare
sboeuf
force-pushed
the
kvm_signal_msi
branch
3 times, most recently
from
ff3b30a to
2feeda7
Compare
Any consumer of this crate who would like to implement MSI/MSI-X support will need KVM_SIGNAL_MSI to notify directly the local APIC emulated by KVM, using MSI/MSI-X vectors. Signed-off-by: Sebastien Boeuf <[email protected]>
4 participants
This pull request primary goal is to add the support for the KVM ioctl
KVM_SIGNAL_MSI.The reason being to allow consumers of this crate to directly inject MSI messages instead of using legacy pin based interrupts.
The second part of this pull request implements the
Clonetrait forVmFdso that anyone can safely cloneVmFd. The need for cloning is to be able to move theVmFdinto a closure or a thread.