Re: linux-next: percpu/tip tree build failure

[Ingo Molnar <mingo@elte.hu>]

* Frederic Weisbecker <fweisbec@gmail.com> wrote:

> On Tue, Dec 08, 2009 at 06:25:15PM +1100, Stephen Rothwell wrote:
> > Hi all,
> > 
> > Today's linux-next build (x86_64 allmodconfig) failed like this:
> > 
> > kernel/hw_breakpoint.c:86: error: 'task_bp_pinned' redeclared as different kind of symbol
> > kernel/hw_breakpoint.c:55: note: previous declaration of 'task_bp_pinned' was here
> > kernel/hw_breakpoint.c: In function 'toggle_bp_task_slot':
> > kernel/hw_breakpoint.c:165: warning: assignment from incompatible pointer type
> > 
> > Caused by commit dd17c8f72993f9461e9c19250e3f155d6d99df22 ("percpu:
> > remove per_cpu__ prefix") from the percpu tree interacting with commit
> > 56053170ea2a2c0dc17420e9b94aa3ca51d80408 ("hw-breakpoints: Fix task-bound
> > breakpoint slot allocation") from the tip tree.
> > 
> > I have applied the following patch for today (and can carry it as
> > necessary:
> > 
> > From: Stephen Rothwell <sfr@canb.auug.org.au>
> > Date: Tue, 8 Dec 2009 18:19:48 +1100
> > Subject: [PATCH] percpu: make per_cpu variable name unique in hw_breakpoint.c
> > 
> > Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
> 
> 
> Doh!
> 
> Thanks Stephen.
> 
> Ingo, can you queue up this patch?
> 
> Thanks.
> 
> Acked-by: Frederic Weisbecker <fweisbec@gmail.com>

I have applied it - but really, the new percpu namespace changes headed 
towards upstream are quite a nuisance IMO. The 3-4 (trivial to solve) 
breakages i've seen so far affecting code i maintain give us an 
estimation about the ongoing maintainence cost - which wont be high but 
not zero either.

The change that was forced here:

 -static DEFINE_PER_CPU(unsigned int, task_bp_pinned[HBP_NUM]);
 +static DEFINE_PER_CPU(unsigned int, nr_task_bp_pinned[HBP_NUM]);

Is it really an improvement to the old code?

Dunno.

	Ingo