On Sat, Jun 13, 2020 at 2:39 PM Wolfram Sang <wsa+renesas@xxxxxxxxxxxxxxxxxxxx> wrote: > > > > > There is no extra interrupt when registering a shared interrupt handler > > > since 2011. Update the Kconfig text to make it clear and to avoid wrong > > > assumptions when debugging issues found by it. > > > > > > > I'm not sure. > > I have recently fixed a bug in the IIO sensor during ->probe() due to > > an issued test interrupt exactly as soon as the handler is registered. > > $ git grep DEBUG_SHIRQ_FIXME > kernel/irq/manage.c:#ifdef CONFIG_DEBUG_SHIRQ_FIXME > > There is no place to enable this code. > > Maybe your case was like Krzysztof's case where the issue turned out to > be the extra interrupt on deregistering after a deferred probe? He > thought it was the initial interrupt but it wasn't. Commit https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/iio/pressure/bmp280-core.c?id=97b31a6f5fb95b1ec6575b78a7240baddba34384 The relevant IRQ core code https://elixir.bootlin.com/linux/latest/source/kernel/irq/manage.c#L1774 It runs it at deregistering, right. -- With Best Regards, Andy Shevchenko
