Stuart Marshall <[email protected]> writes: > Hello, > > I have been developing an x64 OS with USB support, and noticed that, every > time I introduced a bug into my xHCI driver, QEMU would > often silently continue successfully where real hardware that I tested would > error. > > Would it be helpful if I were to try and add warning messages into QEMU logs > for incorrect but recoverable usage of xHCI? This could > aid other OS & driver developers, without breaking drivers that rely on the > permissiveness of QEMU, but I'm not sure that it is the > mission of QEMU to have conformance warnings.
The biggest problem the USB models have is their MAINTAINERS state: USB S: Orphan F: hw/usb/* F: tests/qtest/usb-*-test.c F: docs/system/devices/usb.rst F: include/hw/usb/usb.h F: include/hw/usb/ so no one is actively reviewing or cleaning up the code. So I'm not surprised the current state of the emulation is lax compared to real hardware. We do use the log message qemu_log_mask(LOG_GUEST_ERROR, ... where we detect the guest doing something weird but its by no means comprehensive and mostly there to aid with debug logs. For actual analytics we have trace points. If your interested in helping as a reviewer or fixing issues in the emulation there is a long backlog of issues to look at. > > From Stuart -- Alex Bennée Virtualisation Tech Lead @ Linaro
