Hi,

> > >
> > > Why we can't break keyboard by flooding input after boot up?
> > >
> > Actually, I have pointed the root reason about this issue in other email.
> >
> > When the linux kernel booting, will init the i8042 controller
> (drivers/input/serio/i8042.c), and
> > check the i8042 controller:
> > i8042_init()
> >     |-> i8042_controller_check()
> >
> > If the kernel has booted up, the check will not happen, so the Linux kernel 
> > will
> not get confused.
> 
> I didn't find that the checking is necessary in protocol docs.
> any hint?
> 
The check is belong to Linux kernel, I don't think that will be written to 
protocol docs.
Maybe we can find some hint from the log about i8042_controller_check() at 
Linux git tree.

> If we still use 256 size buffer in qemu, guest ps2 driver will drop
> some input keys? or just delay them?
> 
IMHO, just delay them from my testing result. But it will violate the real 
hardware protocol.


Best regards,
-Gonglei



Reply via email to