>> This series strips down my Xtensa QOM'ification patch from the >> qom-cpu-others.v1 >> RFC series, avoiding the XtensaConfig vs. subclass problem and taking care >> not to introduce "klass", including in the header file. >> >> I've tested this with the dc232b semihosting test image. > > BTW, is it expected that there's a large multi-second lag between > keyboard input and response?
Actually yes: sim machine in the mainline linux kernel has 20 second polling loop in its keyboard driver. This issue didn't show up in early target-xtensa versions because of bug in clock counter implementation and I must have forgotten to send an updated kernel after I've fixed it. > Also, typing "root" at the login prompt resulted in "oot", and > not having the "x" file in the current directory leads to a hang > (Ctrl+A, x does not work, needed to kill the process). Non-semihosting kernels behave more politely. Looks like it's time for me to update testing kernels/rootfs. -- Thanks. -- Max