Re: [PATCH] Input: eeti_ts: Mark as CONFIG_BROKEN

From: Russell King - ARM Linux
Date: Tue Apr 10 2012 - 06:11:05 EST


On Sat, Apr 07, 2012 at 01:32:31PM -0700, Olof Johansson wrote:
> Haojian, I think it was probably premature to do the multiplatform
> change like that, since it means that a PXA-only kernel has no mapping
> from irq_to_gpio to pxa_irq_to_gpio. Can you please address this as a
> fix for 3.4?

Please fix ezx-pcap instead - it's broken as it currently stands by using
irq_to_gpio(), and it's one reason why my randconfig builds on OMAP fail.

The big problem is - what does this do:

do {
...
} while (gpio_get_value(irq_to_gpio(pcap->spi->irq)));

if pcap->spi->irq has no GPIO associated with the interrupt? irq_to_gpio()
probably returns some random number which might be some other GPIO in the
system, and gpio_get_value() could oops if irq_to_gpio returns a negative
or large positive number. To put it another way, according to the above
code, irq_to_gpio() must always return a valid gpio for the IRQ even if
the IRQ doesn't have a GPIO associated with it.

This is a fine illustration of why irq_to_gpio() is just plain broken in
its design.
--
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/