Re: [PATCH 3/4] module: use a structure to encapsulate layout.

[Rusty Russell <rusty@rustcorp.com.au>]

Peter Zijlstra <peterz@infradead.org> writes:
> On Mon, Nov 09, 2015 at 02:53:56PM +1030, Rusty Russell wrote:
>> diff --git a/kernel/module.c b/kernel/module.c
>> index 14b224967e7b..a0a3d6d9d5e8 100644
>> --- a/kernel/module.c
>> +++ b/kernel/module.c
>> @@ -108,13 +108,6 @@ static LIST_HEAD(modules);
>>   * Use a latched RB-tree for __module_address(); this allows us to use
>>   * RCU-sched lookups of the address from any context.
>>   *
>> - * Because modules have two address ranges: init and core, we need two
>> - * latch_tree_nodes entries. Therefore we need the back-pointer from
>> - * mod_tree_node.
>
> We still have the back-pointers, so removing all of that seems a little
> excessive.

Well, I thought about filling the hole with a "am_init" flag, and
putting the layouts in a [2] array, but seemed too cutesy.

>> - *
>> - * Because init ranges are short lived we mark them unlikely and have placed
>> - * them outside the critical cacheline in struct module.
>
> This information also isn't preserved.

Ah yeah, Intel still use 64-byte cachelines.  Still, this comment covers
what we actually care about:

 +#ifdef CONFIG_MODULES_TREE_LOOKUP
 +/* Only touch one cacheline for common rbtree-for-core-layout case. */
 +#define __module_layout_align ____cacheline_aligned
 +#else
 +#define __module_layout_align
 +#endif

> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>

Thanks!
Rusty.