Messages in this thread | | | Date | Wed, 11 Sep 2013 07:36:01 +0200 | From | Ingo Molnar <> | Subject | Re: [PATCH 3/6] tree-sweep: Include linux/sched.h for might_sleep users |
| |
* Andi Kleen <andi@firstfloor.org> wrote:
> On Tue, Sep 10, 2013 at 04:52:12PM -0700, Andrew Morton wrote: > > On Sat, 31 Aug 2013 20:22:23 +0200 Geert Uytterhoeven <geert@linux-m68k.org> wrote: > > > > > On Fri, Aug 16, 2013 at 11:17 PM, Andi Kleen <andi@firstfloor.org> wrote: > > > > might_sleep is moving from linux/kernel.h to linux/sched.h, so any users > > > > need to include linux/sched.h > > > > > > Really? <linux/sched.h> is the worst choice w.r.t. include hell. > > > > Yes. Mechanically adding sched.h inclusions into .h files will cause > > problems. > > > > I've had decent success with this sort of thing by adding new, > > finer-grained headers. Maybe include/linux/might_sleep.h would be a > > good starting point to fix whatever it was that this fixed. > > Ok. I'll look into that and send new patches. > > Some of it may be obsolete with Peter's recent preempt work.
The best order from a maintenance and merge POV would be to rebase your bits on top of Peter's preempt_count optimization work, which is the more comprehensive approach. That allows you to drop the obsolete bits, and wait for Peter to pick up your bits.
The two interact so we want to handle them together in any case, in the scheduler tree.
Thanks,
Ingo
| |