Re: [PATCH] perf_counter: extensible perf_counter_attr

[Corey Ashford <cjashfor@linux.vnet.ibm.com>]

Ingo Molnar wrote:
> * Corey Ashford <cjashfor@linux.vnet.ibm.com> wrote:
> 
>> Peter Zijlstra wrote:
>>> On Mon, 2009-06-08 at 14:18 -0700, Corey Ashford wrote:
>>>> pca.ext_attrs = &feature.head; /* secretly know that there's data 
>>>> that lies past the attr struct header */
>>> Right, except its not so very secret since we have the type field
>>> telling us.
>> I see.  Thanks for the explanation.  Looks like a good plan to me!
> 
> i think we'd have a much simpler implementation by changing 
> __reserved_1 to attr_size. When the kernel adds new attributes, the 
> size will increase - old user-space will use the old size which the 
> kernel detects and adopts to (by zeroing out that attribute space).
> 
> That way we'll always have a nice flat attributes structure, with no 
> quirky chaining that has field-dependent data types ...
> 
> 	Ingo

If I understand you correctly, you would simply make perf_counter_attr larger 
every time you want to add a new attribute.  Users using the new attributes 
would call sys_perf_counter_open with a larger attr_size value.

What about arch-dependent attributes?  Would you want to place them all in the 
perf_counter_attr struct?  I suppose this could be done by #include'ing an 
arch-specific .h file.

- Corey