While re-working the USB session API and adapting the different layers and components to it, the qemu-usb library's handling was changed in a way that is too limited. During the tests, I only had USB headsets at hand, which supported a limited audio frequency. Thereby, the wrong assumption came up that in isochronous audio streams the packets always use the USB EP's maximum packet size. The isochronous packet cache necessary to meet the Qemu's host controller model implementation was therefore constructed with a granularity of maximum packet size. This limitation has to be removed.
nfeske
commented
1 week ago
While re-working the USB session API and adapting the different layers and components to it, the qemu-usb library's handling was changed in a way that is too limited. During the tests, I only had USB headsets at hand, which supported a limited audio frequency. Thereby, the wrong assumption came up that in isochronous audio streams the packets always use the USB EP's maximum packet size. The isochronous packet cache necessary to meet the Qemu's host controller model implementation was therefore constructed with a granularity of maximum packet size. This limitation has to be removed.