Re: [gcv v3 27/35] arm: Replace __get_cpu_var uses

[Christoph Lameter]

On Fri, 30 Aug 2013, Will Deacon wrote:

> This is the flavour we have for ARM's hw_breakpoint code, where we have an
> array of perf_event * instead of int...
>
> > Index: linux/arch/arm/kernel/hw_breakpoint.c
> > ===================================================================
> > --- linux.orig/arch/arm/kernel/hw_breakpoint.c  2013-08-26 13:48:40.956794980 -0500
> > +++ linux/arch/arm/kernel/hw_breakpoint.c       2013-08-26 13:48:40.952795024 -0500
> > @@ -344,13 +344,13 @@ int arch_install_hw_breakpoint(struct pe
> >                 /* Breakpoint */
> >                 ctrl_base = ARM_BASE_BCR;
> >                 val_base = ARM_BASE_BVR;
> > -               slots = (struct perf_event **)__get_cpu_var(bp_on_reg);
> > +               slots = (struct perf_event **)__this_cpu_read(bp_on_reg);
>
> ...so I don't think this is quite right, and indeed, we get a bunch of errors
> from GCC:
>
> arch/arm/kernel/hw_breakpoint.c: In function âarch_install_hw_breakpointâ:
> arch/arm/kernel/hw_breakpoint.c:347:33: error: incompatible types when assigning to type âstruct perf_event *[16]â from type âstruct perf_event **â
> arch/arm/kernel/hw_breakpoint.c:347:1: error: incompatible types when assigning to type âstruct perf_event *[16]â from type âstruct perf_event **â
> arch/arm/kernel/hw_breakpoint.c:347:1: error: incompatible types when assigning to type âstruct perf_event *[16]â from type âstruct perf_event **â
> arch/arm/kernel/hw_breakpoint.c:347:1: error: incompatible types when assigning to type âstruct perf_event *[16]â from type âstruct perf_event **â

Did you apply the first patch of this series which is a bug fix?

> changing to match your recipe still doesn't work, however:
>
> arch/arm/kernel/hw_breakpoint.c: In function âarch_install_hw_breakpointâ:
> arch/arm/kernel/hw_breakpoint.c:347:33: error: cast specifies array type

Yep that is the macro bug that was fixed in the first patch.

> >
> >         WARN_ON(preemptible());
> >
> > -       if (local_inc_return(&__get_cpu_var(mde_ref_count)) == 1)
> > +       if (this_cpu_inc_return(mde_ref_count) == 1)
> >                 enable = DBG_MDSCR_MDE;
>
> I'm not sure that this is safe. We rely on local_inc_return to be atomic
> with respect to the current CPU, which will end up being a wrapper around
> atomic64_inc_return. However, this_cpu_inc_return simply uses a lock, so
> other people accessing the count in a different manner (local_dec_and_test
> below) may break local atomicity unless we start disabling interrupts or
> something horrible like that.

I do not see any special code for ARM for this_cpu_inc_return. The
fallback solution in the core code is to disable interrupts for the
inc_return and arch/arm/include/asm/percpu.h includes
asm-generic/percpu.h.

Where did you see it using a lock?