Another useful piece of code lost

Hello!

The kernel that Dan Malek wrote for us in Transcept, Inc. had the
following lines in include/asm-ppc/ide.h :

/*
 * The following are not needed for the non-m68k ports
 */
#if 0
#define ide_ack_intr(base, irq)         (1)
#else
static __inline__ void ide_ack_intr(unsigned int base, unsigned int irq)
{
        ((immap_t *)IMAP_ADDR)->im_pcmcia.pcmc_pscr = ~0;
}
#endif

Strictly speaking, this is wrong because this code assumes that all PPC
machines, even PowerMac's use PCMCIA.

But from the practical point of view, this code makes it possible to
support ATA Flash at least on RPX/Lite, so the comment is wrong - there
are non-m68k ports depending on non-trivial ide_ack_intr().

The MontaVista kernels don't include anything like that. This is one of
the reasons why they don't support ATA Flash on RPX/Lite.

Regards,
Pavel Roskin


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Pavel Roskin wrote:

Strictly speaking, this is wrong because this code assumes that all PPC
machines, even PowerMac's use PCMCIA.

Of course it's wrong.  That was done specifically for your custom
embedded hardware.  There wasn't any mechanism in that code base to
ack the ATA interrupt if it didn't come from known IDE hatdware.

The MontaVista kernels don't include anything like that. This is one of
the reasons why they don't support ATA Flash on RPX/Lite.

That work was done for you under contract on the RPX-Lite.  If you wish
to give this away, go ahead.  My intention is to use newer IDE abstractions
or Magnus' real PCMCIA instead of this unique solution for you.


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/