Re: [RFC PATCH 29/35] Add the Xen virtual console driver.

[Andi Kleen <ak@suse.de>]

On Wednesday 22 March 2006 07:31, Chris Wright wrote:

Basically it's the same thing as the buffer in printk.c except
that it does input too.

You could safe a lot of code by just teaching the Hypervisor
about that one. Not sure it is fully practical though,


> --- xen-subarch-2.6.orig/drivers/char/tty_io.c
> +++ xen-subarch-2.6/drivers/char/tty_io.c
> @@ -132,6 +132,8 @@ LIST_HEAD(tty_drivers);			/* linked list
>     vt.c for deeply disgusting hack reasons */
>  DECLARE_MUTEX(tty_sem);
>  
> +int console_use_vt = 1;

This doesn't have anything to do with early console.

Why is it needed? Looks like a ugly hack.



> +#define __RETCODE 0

What is that? 

> +/*** Useful function for console debugging -- goes straight to Xen. ***/
> +asmlinkage int xprintk(const char *fmt, ...)

In normal Linux this is called early_printk()

> +{
> +	va_list args;
> +	int printk_len;
> +	static char printk_buf[1024];

Who does the locking here?

-Andi