Re: hd64465_ss & hd64461_ss

[Paul M. <le...@li...>]

On Tue, Nov 13, 2007 at 11:52:38PM +0100, Kristoffer Ericson wrote:
> Just starting looking at pcmcia code again. When compairing to hd64465
> code I found this segment:
> 
Both hd64461 and hd64465 do totally bogus IRQ handling, most of which
should be moved over to arch/sh/cchips and corrected to work with the new
API.

> static void hs_mask_and_ack_irq(unsigned int irq)
> {
> 	hs_socket_disable_ireq(hs_mapped_irq[irq].sock);
>         if (hs_mapped_irq[irq].old_handler != &no_irq_type)
> 	    hs_mapped_irq[irq].chip->ack(irq);
> }
> 
Again, this is simply crap. You are better off deleting everything that
mentions 'irq' in the hd6446x pcmcia drivers and rewriting it completely.

We have proper APIs in place for chaining the handlers, demux, etc. all
of which hd6446x tries to side-step. Surprisingly that doesn't work very
well. Look at the other PCMCIA drivers (ie, not hd6446x) for examples on
how to implement a less fucked up driver.

Also, your keyboard was blessed with an enter key, please consider using
it when writing email.