Re: [PATCH 3/3] kvm tools, ui: Add support for SDL framebuffer output target

[Ingo Molnar <mingo@elte.hu>]

so, i still don't have a GUI. What happens is that once the 
framebuffer initializes in the guest the bootup gets *real* slow, 
like only printk-ing two lines per second.

Then it gets even slower and stops around here:

[   20.040000] No RTC device found, ALARM timers will not wake from suspend
[   20.040000] registered taskstats version 1
[   20.040000]   Magic number: 15:707:842
[   41.601000] md: Waiting for all devices to be available before autodetect
[   41.601000] md: If you don't use raid, use raid=noautodetect

not much CPU used after this point - just the SIGALRM hitting the kvm 
thread every now and then.

here is how the slowdown looks like with printk timestamps:

[    0.677000] uvesafb: failed to execute /sbin/v86d
[    0.678000] uvesafb: make sure that the v86d helper is installed and executable
[    0.679000] uvesafb: Getting VBE info block failed (eax=0x4f00, err=-2)
[    0.680000] uvesafb: vbe_init() failed with -22
[    0.680000] uvesafb: probe of uvesafb.0 failed with error -22
[    0.681000] vesafb: mode is 640x480x32, linelength=2560, pages=0
[    0.682000] vesafb: scrolling: redraw
[    0.682000] vesafb: Truecolor: size=8:8:8:8, shift=24:0:8:16
[    0.683000] vesafb: framebuffer at 0xd0000000, mapped to 0xffffc90000900000, using 1200k, total 1200k
[    1.855000] Console: switching to colour frame buffer device 80x30
[    3.635000] fb0: VESA VGA frame buffer device
[    4.268000] virtio-pci 0000:00:01.0: enabling device (0000 -> 0001)
�[    4.269000] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[    4.535000] serial8250: ttyS0 at I/O 0x3f8 (irq = 0) is a 16550A
�[    7.270000] serial8250: ttyS1 at I/O 0x2f8 (irq = 0) is a 16550A
�[    8.190000] serial8250: ttyS2 at I/O 0x3e8 (irq = 0) is a 16550A
[    8.818000] Non-volatile memory driver v1.3
[    9.419000] Linux agpgart interface v0.103
[    9.420000] [drm] Initialized drm 1.1.0 20060810
[    9.420000] [drm:i915_init] *ERROR* drm/i915 can't work without intel_agp module!

so the badness starts after the 'framebuffer mapped' message.

Thanks,

	Ingo