Messages in this thread |  | | Date | Tue, 17 Feb 2015 16:29:08 -0500 (EST) | From | Nicolas Pitre <> | Subject | Re: [PATCH 34/35] tick: Provide tick_suspend_local() |
| |
On Tue, 17 Feb 2015, Nicolas Pitre wrote:
> On Tue, 17 Feb 2015, Lorenzo Pieralisi wrote: > > > On Mon, Feb 16, 2015 at 12:15:09PM +0000, Peter Zijlstra wrote: > > > From: Thomas Gleixner <tglx@linutronix.de> > > > > > > This function is intended to use by the freezer once the freezer folks > > > solved their race issues. Also required to get rid of the ARM BL > > > switcher tick hackery. > > > > Totally agree with the patch(es), but I noticed that the ARM bL switcher > > does not depend on PM_SLEEP, so I do not think you can compile > > tick_{suspend/resume}_local() out if !PM_SLEEP, unless dependency > > is enforced by the ARM bL switcher config but I do not think that > > the config dependency really exists, Nico please correct me if I am > > wrong. > > The ARM bL switcher does not depend on PM_SUSPEND nor does it enforce it > because it currently doesn't need it. So yeah, either PM_SUSPEND is > selected bringing quite a lot of code with it, or another symbol is used > for those functions (CONFIG_HOTPLUG_CPU maybe?).
Note the same problem exists with arch/x86/xen/suspend.c that calls tick_resume_local() yet this file is not subject to a dependency on CONFIG_PM_SUSPEND.
Nicolas
|  |