On Tue, Sep 15, 2009 at 23:49, Barry Song wrote: > On Wed, Sep 16, 2009 at 12:26 AM, Dmitry Torokhov wrote: >> On Tue, Sep 15, 2009 at 07:19:14AM -0400, Mike Frysinger wrote: >>> On Tue, Sep 15, 2009 at 02:20, Dmitry Torokhov wrote: >>> > BTW, maybe you shoudl convert to threaded IRQs here? >>> >>> yes, after your suggestion for the previous driver, we've been looking >>> at all our input drivers to convert to threaded IRQs. do we need to >>> convert all of them before acceptance, or can we merge now and post an >>> updated patch after ? >> >> It really depends on the driver. If there is a race between IRQ and the >> WQ in the driver I will request you to fix it one way or another before >> accepting the driver (and quite often using threaded IRQ gets rid of the >> race). In the cases like this particular driver though I am not even >> convinced that we need threaded IRQ. The driver is not expected to >> generate lots of events rapidly so using keventd as it does now is >> probably the best solution. > > Even though CONFIG_GENERIC_HARDIRQS is almost always "y", in case > CONFIG_GENERIC_HARDIRQS is "n" not "y", threaded_irq will become > original irq, drivers will not work. So is it necessary to check the > dependence of GENERIC_HARDIRQS while using threaded_irq? when is GENERIC_HARDIRQS not "y" ? a quick check shows only s390/m68k dont set it to "y" in their arch Kconfig ? s390 isnt going to use any of these drivers, and m68k will eventually update (m68knommu already has !?). -mike -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
