On Sun, Feb 16, 2014 at 10:36 PM, Jay Aurabind <jay.aurabind@xxxxxxxxx> wrote: > Dear all, > > I've been going through Robert Love's LKD. Here is an excerpt from it > regarding registration of interrupt handlers: > > "When request_irq() is called with IRQF_SHARED specified, the call succeeds > only if the interrupt line is currently not registered, or if all registered > handlers on the line also specified IRQF_SHARED. Shared handlers, however, > can mix usage of IRQF_DISABLED." > > As far as I understand, the first sentence tells that a line currently > having shared handlers *will only have* handlers registered with IRQF_SHARED > flag in the past. Correct ? Yes. > > If a interrupt line has been registered by a handler specified as non > shared, then whats the point in allowing a new handler with a "shared" flag It doesn't allow as your statement in above indicates. > registering to the same line ? So how does the mixing of shared and unshared > interrupt handlers for the same line go together as mentioned by the 2nd > sentence ? > > Or does it mean that a shared handler which already succeeded the > registration can further register a non shared and shared interrupt > handlers? ( That doesnt make sense, but still... ) ? Simply put, can > someone please elaborate on the second sentence I quoted from the book ? Where is the confusion ? Just to be clear, kernel doesn't allow shared and unshared interrupt handler on the same interrupt line. Read the code in irq core subsystem. kernel/kernel/irq > > > Thanks and Regards, > Aurabindo J > > _______________________________________________ > Kernelnewbies mailing list > Kernelnewbies@xxxxxxxxxxxxxxxxx > http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies > _______________________________________________ Kernelnewbies mailing list Kernelnewbies@xxxxxxxxxxxxxxxxx http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
