Re: [patch 2/3] RCU move trace defines to rcupdate_types.h

[Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>]

* Jeremy Fitzhardinge (jeremy@goop.org) wrote:
> Mathieu Desnoyers wrote:
>> * Steven Rostedt (rostedt@goodmis.org) wrote:
>>   
>>> On Thu, 16 Apr 2009, Jeremy Fitzhardinge wrote:
>>>
>>>     
>>>> I don't think this helps.  rcupdate_types.h uses preempt_disable/enable, but
>>>> doesn't include linux/preempt.h for them - but someone's going to have to, so
>>>> you've got an implicit dependency on the user to #include the right headers in
>>>> advance.
>>>>       
>>> Would including linux/preempt.h in rcupdate_types.h be a problem?
>>>
>>> -- Steve
>>>
>>>     
>>
>> I did not include preempt.h in rcupdate_types.h because rcupdate.h did
>> not include it, so I thought it had a special status such a kernel.h.
>> However, I notice the rcupdate.h includes spinlock.h, which in turn
>> includes preempt.h, so we would need to include preempt.h in
>> rcupdate_types.h.
>>
>> But I think preempt.h is pretty much only type definitions. I don't
>> think that would be a problem, but maybe Jeremy knows better.
>
> No, preempt.h has fairly complex #include requirements:
> #include <linux/thread_info.h>
> #include <linux/linkage.h>
> #include <linux/list.h>
>
> which in turn include:
> /* linux/thread_info */
> #include <linux/bitops.h>
> #include <asm/thread_info.h>
>
> /* list.h */
> #include <linux/stddef.h>
> #include <linux/poison.h>
> #include <linux/prefetch.h>
> #include <asm/system.h>
>
> And from there more complex still:
> #include <asm/asm.h>
> #include <asm/segment.h>
> #include <asm/cpufeature.h>
> #include <asm/cmpxchg.h>
> #include <asm/nops.h>
> #include <linux/kernel.h>
> #include <linux/irqflags.h>
> #include <asm/page.h>
> #include <asm/processor.h>
> #include <asm/ftrace.h>
> #include <asm/atomic.h>
> [...and much more...]
>
> Given that paravirt.h is included in some of those headers, it  
> eventually gets cyclic.
>

OK.

Given the simplicity of the preempt_disable/enable_notrace found in
preempt.h, we could move them to 

include/preempt_types.h too, and that would solve all problems, wouldn't
it ?

Mathieu

>    J

-- 
Mathieu Desnoyers
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68