Re: serial problems cont..

[Avi Kivity <avi@qumranet.com>]

Michael Malone wrote:
> Hi everyone,
>
> I've written a couple of questions regarding the serial device in 
> KVM.  After slightly more investigation I think I have found what's 
> going awry.  Correct me if I'm wrong, but I believe that KVM generates 
> an interrupt for every single character it sends through the serial 
> port.  This throws CPU usage through the roof and I suspect this means 
> that the timers aren't being handled correctly and it failed on a 
> string of 0's for me due to the timing slips.  GNU/Linux and Windows 
> don't have anywhere near the processor usage for their serial ports.  
> Now, I know nothing of serial programming and don't have any time to 
> investigate it too heavily just now, but I have pulled down the source 
> and had a look through that, but it looks to be doing the right thing 
> (I suppose?).   I was mainly wondering how GNU/Linux and windows 
> handle serial interrupts or if some of the serial character events 
> could be buffered, rather than overload the processor?  I guess this 
> is a low priority for you, but any help would be greatly appreciated 
> (And when I have some more time, I will spend some of it helping to 
> develop KVM! Quid pro quo, Clarice...)
>

What version of kvm are you testing?  There were some changes to the 
serial emulation recently.  See for example 
02f0b4c0cc26f3a2578d515d96781f5a6258888d in kvm-73.

-- 
error compiling committee.c: too many arguments to function