20210725.1 release missing usb-host device

[molind]

In previous release I had no issues with bridging my printer to qemu. Now it returns an error. Is it issue with

qemu-system-aarch64: -device usb-host,vendorid=0x04e8,productid=0x3301: 'usb-host' is not a valid device model name

[knazarov]

Are you running qemu as root? As far as I remember, it requires root to passthrough the usb. Otherwise it’s strange, because the implementation didn’t really change all that much. We’ll need to probably do some more digging if that’s the case.

[molind]

No I running it from normal user. And when i build formula from 917e219ca0caa41a3dba28192465416a4c449101 it works fine. When I build it from head commit 0bcbb8f590f27738adb757b355c81d30a0e469f5 it returns an error: "'usb-host' is not a valid device model name".

Seems something inside qemu build scripts changed between those two commits.

In your patchV4

diff --git a/configure b/configure
index 85db248ac1..f711778086 100755
--- a/configure
+++ b/configure
@@ -377,6 +377,7 @@ u2f="auto"
 libusb="auto"
 usb_redir="auto"
 opengl="$default_feature"
+egl="no"
 cpuid_h="no"
 avx2_opt="$default_feature"
 capstone="auto"

but in patchV3

diff --git a/configure b/configure
index 4f374b4889..49fb478ac9 100755
--- a/configure
+++ b/configure
@@ -394,6 +394,7 @@ u2f="auto"
 libusb="$default_feature"
 usb_redir="$default_feature"
 opengl="$default_feature"

May be this difference somehow changes number of built qemu modules.

My full command to run it:

qemu-system-aarch64 \
         -machine virt,accel=hvf,highmem=off \
         -cpu cortex-a72 -smp 2 -m 4G \
         -device intel-hda -device hda-output \
         -device virtio-gpu-pci \
         -device virtio-keyboard-pci \
         -device virtio-net-pci,netdev=net \
         -device virtio-mouse-pci \
         -device qemu-xhci \
         -device usb-host,vendorid=0x04e8,productid=0x3301 \
         -display cocoa,gl=es \
         -netdev user,id=net,ipv6=off \
         -drive "if=pflash,format=raw,file=./edk2-aarch64-code.fd,readonly=on" \
         -drive "if=pflash,format=raw,file=./edk2-arm-vars.fd,discard=on" \
         -drive "if=virtio,format=raw,file=./hdd.raw,discard=on"

[knazarov]

This is an interesting finding. Could you please check if usb-host works on unpatched qemu on mac? If it does, we’ll be sure that the virgl patchset breaks something.

You can install it with brew install qemu. But please make sure you uninstall qemu-virgl before trying.

[molind]

It prints lots of qemu-system-aarch64: qemu_mprotect__osdep: mprotect failed: Permission denied and enters into qemu console. But no errors about usb devices.

[knazarov]

Try running it with sudo. Permission denied may be due to the real lack of permissions.

[molind]

No change. It returns the same error.

[knazarov]

OK, I was planning to get myself an arm-based mac today. I’ll try to reproduce the problem and let you know.

[knazarov]

@molind I was able to get my hands on a M1 Mac today, and verified that it really is an upstream problem in qemu that was fixed a few days ago. I'm cherry-picking the fix here: #41

[lazevedo]

Could we get a release with this? Would be much appreciated

[knazarov]

Could we get a release with this? Would be much appreciated

Sure. There’s an updated patchset from the author, so I’ll be working on it around this weekend.