On Tue, 20 Aug 2019 at 16:02, Mark Brown <broonie@xxxxxxxxxx> wrote: > > On Sun, Aug 18, 2019 at 09:01:14PM +0300, Vladimir Oltean wrote: > > There is no point in checking which interrupt source was triggered in > > SPI_SR, since only EOQ and TCFQ modes trigger interrupts anyway (see the > > writes to SPI_RSER). In DMA mode, the RSER is configured for RFDF_DIRS=1 > > and TFFF_DIRS=1, aka FIFO events generate eDMA requests and not CPU > > interrupts. > > It's also good to check interrupt sources in case the interrupt line > might be shared, that means that it's possible that the interrupt > handler will be called when there's no interrupt at all from the IP. It > also helps with robustness in case there's some system error though > that's (hopefully!) a lot less common. This driver does set IRQF_SHARED > so it does support that, I don't know how many systems could do it but > it seems a shame to remove the support from the driver. Ok, I hadn't thought of that, I'll add the check back.