From mboxrd@z Thu Jan 1 00:00:00 1970 From: linus.walleij@linaro.org (Linus Walleij) Date: Tue, 23 Oct 2012 10:31:45 +0200 Subject: [PATCH] pinctrl/nomadik: use irq_create_mapping() In-Reply-To: <5085A7CA.3070001@wwwdotorg.org> References: <1350659375-7335-1-git-send-email-linus.walleij@stericsson.com> <50817E43.8000804@wwwdotorg.org> <5085A7CA.3070001@wwwdotorg.org> Message-ID: To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Mon, Oct 22, 2012 at 10:08 PM, Stephen Warren wrote: > On 10/22/2012 02:14 AM, Linus Walleij wrote: >> It's an IRQ handler so it should be robust to spurious IRQs due to >> transient hardware states etc I believe. >> >> So if there is a transient IRQ before gpio_to_irq() is called -> boom. > > I wonder though (a) why it would be unmasked in HW, and (b) why the > software would even look at the status bit if no handler were registered? That's true of course ... OK I'll update the patch. Still I'm not feeling good about the irq_create_mapping/irq_find_mapping separation, I think a lot of drivers just get this wrong and it's causing bugs... it'd be way better if there was just one of them and we could count on descriptors being allocated after adding any kind of irqdomain but I have no clue how hard it would be to achieve this. Yours, Linus Walleij