On Wed, Feb 22, 2012 at 11:36:48PM -0000, Igor Galić wrote:
Hello happy people,
I'm back from a two day goose-chase where I was wondering very much why my VMs don't boot.
https://www.redhat.com/archives/libvirt-users/2012-February/msg00106.html
In that process I upgraded my entire stack (libvirt, qemu, seabios) but found that the boot hung at SeaBIOS trying to boot from ROM. After stepping through all the arguments libvirt supplied to I found that my <memory>2048</memory> the qemu argument -m 2 (megs) -- which wasn't quite enough. (Similarly, my previously supplied <memory>20480</memory> weren't too much they too were too little to boot a Linux kernel).
I've already talked to the libvirt folks, and they'll try to get in a feature to supply a <memory unit="M"> or similar.
Now I'd like to ask you guys if you could print the amount of RAM available -- like most other BIOSes out there already do - it would be extremely helpful to aid troubleshooting.
SeaBIOS boots too fast for anything useful to be seen on the screen. That said, one can add "-chardev stdio,id=seabios -device isa-debugcon,iobase=0x402,chardev=seabios" to the qemu command line to see the SeaBIOS debugging output. That output has the memory info and much more.
-Kevin