On Wed, Oct 26, 2022 at 04:28:59PM +0300, Andy Shevchenko wrote: > On Wed, Oct 26, 2022 at 03:39:08PM +0300, Jarkko Nikula wrote: > > Do not return with interrupt handled if host controller is off and thus > > interrupt is originating from other device or is spurious. > > > > Add a check to detect when controller is runtime suspended or > > transitioning/reset. In latter case all raw interrupt status register > > bits may read one. In both cases return IRQ_NONE to indicate interrupt > > was not from this device. ... > > dev_dbg(dev->dev, "enabled=%#x stat=%#x\n", enabled, stat); > > if (!enabled || !(stat & ~DW_IC_INTR_ACTIVITY)) > > return IRQ_NONE; > > + if (pm_runtime_suspended(dev->dev) || stat == ~0) > > + return IRQ_NONE; > > I haven't checked the type of 'stat', but usually be careful with ~0. > Due to integer promotion it may give the unexpected results. That said, GENMASK(hi, 0) in this case is better. -- With Best Regards, Andy Shevchenko
