From: Thomas Gleixner Subject: Re: [PATCH v3 06/16] irqchip: add sl28cpld interrupt controller support Date: Mon, 27 Apr 2020 21:00:05 +0200 Message-ID: <87sggopxe2.fsf@nanos.tec.linutronix.de> References: <20200423174543.17161-1-michael@walle.cc> <20200423174543.17161-7-michael@walle.cc> <87pnbtqhr1.fsf@nanos.tec.linutronix.de> <87f141bce0a4fda04b550647306be296@walle.cc> Mime-Version: 1.0 Content-Type: text/plain Return-path: Received: from lindbergh.monkeyblade.net ([23.128.96.19]:48684 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726688AbgD0TAh (ORCPT ); Mon, 27 Apr 2020 15:00:37 -0400 In-Reply-To: <87f141bce0a4fda04b550647306be296@walle.cc> Sender: linux-pwm-owner@vger.kernel.org List-Id: linux-pwm@vger.kernel.org To: Michael Walle Cc: Andy Shevchenko , linux-gpio@vger.kernel.org, devicetree@vger.kernel.org, linux-kernel@vger.kernel.org, linux-hwmon@vger.kernel.org, linux-pwm@vger.kernel.org, linux-watchdog@vger.kernel.org, linux-arm-kernel@lists.infradead.org, Linus Walleij , Bartosz Golaszewski , Rob Herring , Jean Delvare , Guenter Roeck , Lee Jones , Thierry Reding , Uwe =?utf-8?Q?Kleine-K=C3=B6nig?= , Wim Van Sebroeck , Shawn Guo , Li Yang , Jason Cooper , Marc Michael, Michael Walle writes: > Am 2020-04-27 13:40, schrieb Thomas Gleixner: >>> + >>> + ret = devm_regmap_add_irq_chip(&pdev->dev, irqchip->regmap, irq, >>> + IRQF_SHARED | IRQF_ONESHOT, 0, >> >> What's the point of IRQF_SHARED | IRQF_ONESHOT here? > > IRQF_SHARED because this interrupt is shared with all the blocks > which can generate interrupts, i.e. the GPIO contollers. Why are people still designing hardware with shared interrupts? Shared interrupts are broken by design and that's well known for decades. > IRQF_ONESHOT, because its is a threaded interrupt with no primary > handler. But I just noticed, that regmap-irq will also set the > IRQF_ONESHOT. But that the commit 09cadf6e088b ("regmap-irq: > set IRQF_ONESHOT flag to ensure IRQ request") reads like it is > just there to be sure. So I don't know if it should also be set > here. Ok. Wasn't aware of that magic threaded interrupt connection. Thanks, tglx