Il 18/10/2012 17:17, Laszlo Ersek ha scritto: > On 10/18/12 10:12, Gleb Natapov wrote: >> On Thu, Oct 18, 2012 at 03:38:37AM +0200, Laszlo Ersek wrote: >>> I have no idea why the UEFI-booted kernel tries to initialize the >>> display with Int10h, since that is a (VGA) BIOS interrupt. >> Looks like google knows something about it. Search for "INT 10" here: >> http://www.uefi.org/events/UEFI-Plugfest-WindowsBootEnvironment.pdf >> >> * VGA support still requires INT 10h support >> * Windows Server 2008 supports headless systems with no VGA > > (The hang reproduces with "-vga none" as well.) > > Two screenshots linked (sorry, found no easy way to extract text from > the call stack window): > - call stack from the (futile) display initialization loop [1], with > "-vga cirrus", > - call stack during the "hang" [2]. The call stack looks the same with > "-vga cirrus" and "-vga none". Appears to be consistent with the kvm ftrace.
Is it really a hang? I haven't looked at the kvm ftrace, but this looks more like Windows is running headless. Perhaps you can install Windows under DUET (with DUET in a USB flash drive), configure remote desktop, then boot it again under OVMF, and try to connect to the Windows remote desktop. More info on DUET: http://www.rodsbooks.com/bios2uefi/ Paolo > I'll try to follow the guest with finer granularity before the idle > loop. (Single stepping seems a bit too ambitious.) > > Thanks! > Laszlo > > [1] > http://people.redhat.com/~lersek/ovmf_win2k8r2sp1_debug/during-display-init.png > [2] http://people.redhat.com/~lersek/ovmf_win2k8r2sp1_debug/hang.png > ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_sfd2d_oct _______________________________________________ edk2-devel mailing list edk2-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/edk2-devel
