On Sun, 13 Mar 2016, Jianyu Zhan wrote: > On Sun, Mar 13, 2016 at 5:33 PM, Thomas Gleixner <[email protected]> wrote: > > IRQ_WORK can work w/o APIC > > > > Emphasis on CAN. If the APIC is available it's used, if not then there is no > > point in setting up the gate for nothing. > > > > So why would your patch do any good? > > I understood it is no point setting up if APIC is not available, but > just got confused by > your wording 'can', now all clear. > > As for the patch set. My initial purpose is just wanting to make the > layout clear and > clean up stale comments and dead code: > > #ifndef CONFIG_X86_LOCAL_APIC > #define first_system_vector NR_VECTORS > #endif > > as we've talked about this before, it won't ever be change on > !CONFIG_X86_LOCAL_APIC, > so no point define it here(it is initialized to NR_VECTORS).
Hell no. If CONFIG_X86_LOCAL_APIC=n then first_system_vector is not a variable and not available and not initialized at all. > Since all points are clear now, Obviously not.
