On 2024-08-26, Derek Barbosa <debarbos@xxxxxxxxxx> wrote: > This is not running in QEMU. I installed & booted this kernel directly. > Considering that this produced a vmcore, I could try my hand at doing a crash > analysis (with some pointers). That would be nice. It would be interesting to see if the trace is in the ringbuffer. If not, it probably means the system is hanging in the console driver while trying to print out the messages synchronously. > Do you think running this kernel in QEMU would behave any differently > than it is now? Until now, I have never needed bare metal to debug/fix printk issues for serial. This is because the 8250 UART is so simple and QEMU emulation of it is good enough. Of course, this is one (of many) reasons why I am very grateful that you are running all these tests on bare metal! Thanks for this! _If_ the problems are reproducible using QEMU/KVM, it would be really easy to dig deeper. Even if it is just to confirm, "Oh yeah, we know about that problem... not fixable with legacy printing." But sometimes we see that it is an easily fixable problem to at least help improve the reliability of legacy printing. John
