Re: Re: [PATCH] make c/s 21089 work again with c/s 21092

Re: Re: [PATCH] make c/s 21089 work again with c/s 21092

[Konrad Rzeszutek Wilk]

On Tue, Apr 13, 2010 at 04:53:35PM +0100, Jan Beulich wrote:
> >>> Bastian Blank <waldi@debian.org> 13.04.10 17:10 >>>
> >On Tue, Apr 13, 2010 at 02:36:55PM +0100, Jan Beulich wrote:
> >> Unfortunately the latter c/s' change to mpparse.c yielded the former
> >> patch non-functional - Xen's serial port IRQ is not in IQR_DISABLED
> >> state, yet must be allowed to get its trigger mode and polarity set
> >> up in order for it to be usable.
> >
> >This looks not right. The original patch was designed to keep this
> >function away from the irqs of the serial port. This patch adds another
> >check for disallowing access but does not touch the IQR_DISABLED check.
> >
> >Please explain which machines have non-legacy configured interrupts for
> >the first two serial ports.
> 
> I'm talking about PCI serial cards - in the one case I'm working with,

Can't you use polling for it? For example this is what I do for my PCI
serial card:

com1=115200,8n1,0xd800,0

Re: Re: [PATCH] make c/s 21089 work again with c/s 21092

[Jan Beulich]

>>> Konrad Rzeszutek Wilk  04/13/10 6:12 PM >>>
>On Tue, Apr 13, 2010 at 04:53:35PM +0100, Jan Beulich wrote:
>> I'm talking about PCI serial cards - in the one case I'm working with,
>
>Can't you use polling for it? For example this is what I do for my PCI
>serial card:
>
>com1=115200,8n1,0xd800,0

It works with low rates of output, but in order to not use sync_console and still be able to reliably get through high rate bursts of output, interrupt driven communication seems to be the only reliable way (other than using unreasonably high values with "serial_tx_buffer=").

And also from a more general perspective - why should legacy IRQs work, but not PCI ones?

Jan

Re: Re: [PATCH] make c/s 21089 work again with c/s 21092

[Konrad Rzeszutek Wilk]

On Wed, Apr 14, 2010 at 07:41:30AM +0100, Jan Beulich wrote:
> >>> Konrad Rzeszutek Wilk  04/13/10 6:12 PM >>>
> >On Tue, Apr 13, 2010 at 04:53:35PM +0100, Jan Beulich wrote:
> >> I'm talking about PCI serial cards - in the one case I'm working with,
> >
> >Can't you use polling for it? For example this is what I do for my PCI
> >serial card:
> >
> >com1=115200,8n1,0xd800,0
> 
> It works with low rates of output, but in order to not use sync_console and still be able to reliably get through high rate bursts of output, interrupt driven communication seems to be the only reliable way (other than using unreasonably high values with "serial_tx_buffer=").
> 
> And also from a more general perspective - why should legacy IRQs work, but not PCI ones?

You have a good point. I actually had the darnest time getting these
cards to work with Linux so was quite happy to see Xen working so well
with them.

I will take a look at your patch.
> 
> Jan
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel