Re: [PATCH] hvc_console: returning 0 from put_chars is not an error

[Christian Borntraeger]

Am Mittwoch 14 Oktober 2009 23:53:46 schrieben Sie:
> hvc_console_print() calls the HVC client driver's put_chars() callback
> to write some characters to the console.  If the callback returns 0, that
> indicates that no characters were written (perhaps the output buffer is
> full), but hvc_console_print() treats that as an error and discards the
> rest of the buffer.
> 
> So change hvc_console_print() to just loop and call put_chars() again if it
> returns a 0 return code.
> 
> This change makes hvc_console_print() behave more like hvc_push(), which
>  does check for a 0 return code and re-schedules itself.

There is a difference between console and tty, if the console call does not 
return, it might bring the full system to a stop. (if its the preferred console, 
init will stop)

> This patch might fix drivers that return 0 to indicate that they're busy, such 
> as arch/powerpc/platforms/pseries/hvconsole.c. It will break drivers that
> return 0 if their output buffer is full, but where those buffers cannot be
> emptied while the kernel is in a loop.

Yep. I think it really depends on the backend if looping will result in any 
progress or not. My experience wth hvc_console is, that its quite hard to get 
changes tested on all backends. (e.g. XEN, pseries, iseries, virtio_console, 
s390_iucv...), so even if this change turns out to be correct, it should 
probably sit in linux-next for a while. In additon I really dont oversee, what 
backends wil break due to this patch.

The fact that struct console->write returns void indicates that the console 
layer is not interested in errors. We have two policies we can implement:

1. drop console messages if case of congestion but keep the system going
2. dont drop messages and wait, even if the system might come to a complete stop 

Looking at drivers/char/vt.c 
        /* console busy or not yet initialized */
        if (!printable)
                return;
        if (!spin_trylock(&printing_lock))
                return;
could mean that  Linux consoles should not block.

Maybe its time to ask some of the elder magicians (CCing Alan Cox and linux-
kernel) about blocking and error handling in console code.


> --- a/drivers/char/hvc_console.c
> +++ b/drivers/char/hvc_console.c
> @@ -161,7 +161,7 @@ static void hvc_console_print(struct console *co, const
>  char *b, }
>  		} else {
>  			r = cons_ops[index]->put_chars(vtermnos[index], c, i);
> -			if (r <= 0) {
> +			if (r < 0) {
>  				/* throw away chars on error */
>  				i = 0;
>  			} else if (r > 0) {
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/