The method to acquire the input interrupt was working in a platform data based board, but was failing in a device-tree one when the gpio controller was probed after pxa_cplds.
Use platform_get_irq() to benefit from the probe deferral mechanism. Moreover, as seen in dm9000.c development, platform_get_irq() doesn't honor the irq type IO resource (ie. edge rising for example), and it must be passed again at irq request in a not device-tree build, hence the irq_get_trigger_type() call. Signed-off-by: Robert Jarzmik <[email protected]> --- arch/arm/mach-pxa/pxa_cplds_irqs.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) diff --git a/arch/arm/mach-pxa/pxa_cplds_irqs.c b/arch/arm/mach-pxa/pxa_cplds_irqs.c index e362f865fcd2..941508585e34 100644 --- a/arch/arm/mach-pxa/pxa_cplds_irqs.c +++ b/arch/arm/mach-pxa/pxa_cplds_irqs.c @@ -120,13 +120,9 @@ static int cplds_probe(struct platform_device *pdev) if (!fpga) return -ENOMEM; - res = platform_get_resource(pdev, IORESOURCE_IRQ, 0); - if (res) { - fpga->irq = (unsigned int)res->start; - irqflags = res->flags; - } - if (!fpga->irq) - return -ENODEV; + fpga->irq = platform_get_irq(pdev, 0); + if (fpga->irq <= 0) + return fpga->irq; base_irq = platform_get_irq(pdev, 1); if (base_irq < 0) @@ -142,6 +138,7 @@ static int cplds_probe(struct platform_device *pdev) writel(fpga->irq_mask, fpga->base + FPGA_IRQ_MASK_EN); writel(0, fpga->base + FPGA_IRQ_SET_CLR); + irqflags = irq_get_trigger_type(fpga->irq); ret = devm_request_irq(&pdev->dev, fpga->irq, cplds_irq_handler, irqflags, dev_name(&pdev->dev), fpga); if (ret == -ENOSYS) -- 2.1.4
