Kernel/HID: Add support for virtio input devices (+ use them for RISC-V in run.py) #24254
Conversation
github-actions
bot
added
the
👀 pr-needs-review
spholz
force-pushed
the
virtio-input
branch
4 times, most recently
from
16ba956
to
94a852f
Compare
If the VirtIO driver is stable enough, I'd suggest that we should move to that and ditch PS/2 for bare metal usage. |
spholz
force-pushed
the
virtio-input
branch
2 times, most recently
from
d09375f
to
1301675
Compare
Kernel/Devices/HID/VirtIO/Input.cpp
Outdated
| HIDManagement::the().attach_standalone_hid_device(*m_mouse_device); | ||
| } else { | ||
| dbgln("Virtio::Input: Unknown product ID"); | ||
| VERIFY_NOT_REACHED(); |
There was a problem hiding this comment.
There sadly is no nice way to ignore devices with unknown devids, since the caller of initialize_virtio_resources uses MUST. We also would have to leave the Input device in a half-initialized state, which doesn't seem ideal.
So I would rather wait for #24210 to be merged, since I don't think we should be using VERIFY_NOT_REACHED here.
There was a problem hiding this comment.
I removed the devid matching code now.
Only the mouse will reliably work until #24210 is merged.
The keyboard actually somewhat already works with the way it is added in run.py because it is listed first, and QEMU apparently assigns PCI addresses based on the order of the command line arguments.
As a result, the keyboard is located at "/dev/input/keyboard/0", which is where WindowServer currently expects it to be.
spholz
force-pushed
the
virtio-input
branch
4 times, most recently
from
125fc1d
to
f3e4e7f
Compare
|
conflicts |
ADKaster
added
the
⚠️ pr-has-conflicts
|
rebased |
ADKaster
removed
the
⚠️ pr-has-conflicts
github-actions
bot
removed
the
👀 pr-needs-review
This PR makes keyboard and (absolute) mouse input work on RISC-V!
The QEMU virtio keyboard, mouse and tablet (absolute mouse) work on both x86-64 and riscv64.
I only tested the keyboard driver with my german keyboard (+german layout), but that should hopefully not be an issue. I would still appreciate it if someone could test this with a different layout.
We could also consider changing run.py to use the virtio input devices on x86-64 in the future, since those might be a better option than using PS/2 when running in QEMU.
The new virtio driver is not in the Bus/VirtIO directory, as #24152 will move all other virtio drivers out of there.