Re: [PATCH] mm: use in_atomic() in print_vma_addr()

[Yang Shi]

On 11/3/17 11:02 AM, Andrew Morton wrote:
> On Fri, 03 Nov 2017 01:44:44 +0800 "Yang Shi" <yang.s@xxxxxxxxxxxxxxx> wrote:
>
> > I may not articulate it in the commit log
>
> You should have done so ;)

Yes, definitely. I could done it much better.

> Here's the changelog I ended up with:
>
> : From: "Yang Shi" <yang.s@xxxxxxxxxxxxxxx>
> : Subject: mm: use in_atomic() in print_vma_addr()
> :
> : 3e51f3c4004c9b ("sched/preempt: Remove PREEMPT_ACTIVE unmasking off
> : in_atomic()") uses in_atomic() just check the preempt count, so it is not
> : necessary to use preempt_count() in print_vma_addr() any more.  Replace
> : preempt_count() to in_atomic() which is a generic API for checking atomic
> : context.
> :
> : in_atomic() is the preferred API for checking atomic context instead of
> : preempt_count() which should be used for retrieving the preemption count
> : value.
> :
> : If we go through the kernel code, almost everywhere "in_atomic" is used
> : for such use case already, except two places:
> :
> : - print_vma_addr()
> : - debug_smp_processor_id()
> :
> : Both came from Ingo long time ago before 3e51f3c4004c9b01 ("sched/preempt:
> : Remove PREEMPT_ACTIVE unmasking off in_atomic()").  But, after this commit
> : was merged, use in_atomic() to follow the convention.
> :
> : Link: http://lkml.kernel.org/r/1509572313-102989-1-git-send-email-yang.s@xxxxxxxxxxxxxxx
> : Signed-off-by: Yang Shi <yang.s@xxxxxxxxxxxxxxx>
> : Acked-by: Michal Hocko <mhocko@xxxxxxxx>
> : Cc: Frederic Weisbecker <fweisbec@xxxxxxxxx>
> : Cc: Ingo Molnar <mingo@xxxxxxx>

Thanks a lot for reworking the commit log.

> Also, checkpatch says
>
> WARNING: use of in_atomic() is incorrect outside core kernel code
> #43: FILE: mm/memory.c:4491:
> +       if (in_atomic())
>
> I don't recall why we did that, but perhaps this should be revisited?

I think the rule for in_atomic is obsolete in checkpatch.pl. A quick 
grep shows in_atomic() is used by arch, drivers, crypto, even though the 
comment in include/linux/preempt.h says in_atomic() should be not used 
by drivers.

However, the message could be ignored with --ignore=IN_ATOMIC. But, it 
sounds better to fix the wrong rule and maybe even the comment in 
include/linux/preempt.h since it sounds confusing.

Thanks,
Yang
>