Implement SIMPLE_NETWORK_PROTOCOL #606
Conversation
| if simple_network.receive(&mut buffer, None, None, None, None) | ||
| == Err(Status::NOT_READY.into()) | ||
| { | ||
| bt.stall(1_000_000); |
There was a problem hiding this comment.
Ideally we could use the wait_for_packet here. I recall from the last PR attempting this that we couldn't get that event to fire... any idea what the problem could be there?
There was a problem hiding this comment.
This event didn't seem to fire on QEMU at all, despite the packet being received. I thought I was doing something wrong, but if with the same code I try to wait for a wait_for_key event, that event resolves fine...
At this point, I think either the QEMU/OVMF implementation and the EFI spec disagree, QEMU doesn't implement this event for whatever reason (I found this thread from 10 years ago that seems to imply that the event not being implemented is fairly common, but I am not sure how I would verify), or there was something else wrong with my code that I don't undersand. ~30m of digging got me nowhere last time I tried.
This is also why the event is not exposed, since I haven't managed to get it to work :)
There was a problem hiding this comment.
I dug around in the EDK2 and OVMF source a bit but don't have much to add. It does look like the WaitForPacket Event and all related machinery have been implemented for at least the virtio-net-pci driver in QEMU. However, even while using this driver, the event never seems to be signaled.
I'll try to plug gdb into qemu tomorrow and see what might be going on.
There was a problem hiding this comment.
This is also why the event is not exposed, since I haven't managed to get it to work :)
I would recommend exposing it anyway; it may well work on some non-QEMU/EDK2 environments.
There was a problem hiding this comment.
I exposed it with a comment that cautions the user to check before using it :)
veluca93
force-pushed
the
snp
branch
2 times, most recently
from
0221361 to
4a92872
Compare
|
(Sorry for my slow speed at clicking the "allow checks to run" button -- ideally I could just tell github to trust you or this PR, but it doesn't seem to be an option. Once a PR has been accepted then future PRs from you will be "trusted" for CI runs I think.) |
No worries, I've been on the other side of this and I understand very well :) Hopefully this last version passes all the checks... |
Fixes #431.
The code is written on top of @d-sonuga's PR, with the main change being in the tests, where an UDP packet is sent to the UDP echo server and the answer is read back.