Re: [PATCH 1/2] memcg: move mem_cgroup_swappiness to memcontrol.h

Ridong Chen <[email protected]>
Newsgroups org.kernel.vger.cgroups,org.kernel.vger.linux-kernel,org.kvack.linux-mm
Message-ID <[email protected]>

On 7/14/2026 6:21 PM, Barry Song wrote:
> On Tue, Jul 14, 2026 at 3:43 PM Ridong Chen <[email protected]> wrote:
>>
>>
>>
>> On 7/14/2026 9:48 AM, Barry Song wrote:
>>> On Tue, Jul 14, 2026 at 9:43 AM Barry Song <[email protected]> wrote:
>>>>
>>>> On Tue, Jul 14, 2026 at 9:20 AM Ridong Chen <[email protected]> wrote:
>>>>>
>>>>>
>>>>>
>>>>> On 7/13/2026 11:08 PM, Barry Song wrote:
>>>>>> On Sat, Jul 11, 2026 at 5:12 PM Ridong Chen <[email protected]> wrote:
>>>>>>>
>>>>>>> From: Ridong Chen <[email protected]>
>>>>>>>
>>>>>>> The per-memcg swappiness knob is v1-only; v2 always uses global
>>>>>>> vm_swappiness and ignores the per-cgroup field.
>>>>>>>
>>>>>>> Guard memcg->swappiness with CONFIG_MEMCG_V1, and move the helper
>>>>>>> to memcontrol.h where it belongs.
>>>>>>>
>>>>>>> No functional change for v1; v2-only kernels drop the unused field.
>>>>>>>
>>>>>>> Signed-off-by: Ridong Chen <[email protected]>
>>>>>>> Acked-by: Johannes Weiner <[email protected]>
>>>>>>
>>>>>> Reviewed-by: Barry Song <[email protected]>
>>>>>>
>>>>>> With some nits.
>>>>>>
>>>>>>> ---
>>>>>> [...]
>>>>>>>            struct mem_cgroup_per_node *nodeinfo[];
>>>>>>> @@ -365,6 +366,9 @@ enum objext_flags {
>>>>>>>
>>>>>>>     #define OBJEXTS_FLAGS_MASK (__NR_OBJEXTS_FLAGS - 1)
>>>>>>>
>>>>>>> +/* Defined in mm/vmscan.c; used by mem_cgroup_swappiness(). */
>>>>>>> +extern int vm_swappiness;
>>>>>>
>>>>>> This is a bit unusual. I'm not sure whether mm/swap.h would be
>>>>>> a more appropriate place for this.
>>>>>>
>>>>> Thank you for your reply.
>>>>>
>>>>> The vm_swappiness variable is not utilized within mm/swap.c.
>>>>> Furthermore, since memcontrol.h does not include swap.h, retaining the
>>>>> extern int vm_swappiness declaration in mm/swap.h will result in a
>>>>> compilation failure.
>>>>
>>>> If this is the case, it still seems better to keep
>>>> extern int vm_swappiness in include/linux/swap.h.
>>>>
>>>> Then we don't need the comment:
>>>> /* Defined in mm/vmscan.c; used by mem_cgroup_swappiness(). */
>>>>
>>>> It also makes it clearer that vm_swappiness is an extern variable
>>>> belonging to the swap module, rather than the memcontrol module.
>>>
>>> BTW, if mem_c_group_swappiness() and vm_swappiness are only used
>>> within mm/, could all of these be moved to mm/swap.h and
>>> mm/internal.h instead?
>>>
>>> We are making a big effort to move many unrelated things out of
>>> include/linux/swap.h recently. Could you check?
>>>
>>> https://lore.kernel.org/linux-mm/20260708-ch-swap-series-plus-folio-lru-cleanup-v9-0-2bc72b4f8730@gmail.com/
>>
>> Good suggestion. Moving them to mm/internal.h makes sense. Will update
>> in the next version.
> 
> Either mm/swap.h or mm/internal.h.
> vm_swappiness probably belongs in mm/swap.h rather than
> mm/internal.h, right?
> 
> BTW, I am not particularly eager about this cleanup;
> it could be done later as a separate patch.
> 
> If you decide not to do the cleanup, I think it would be better to
> leave "extern int vm_swappiness" in include/linux/swap.h rather than
> declaring it in include/linux/memcontrol.h?
I noticed that some swap-related macros are already defined in 
mm/internal.h, so I'm planning to place the code right after them for 
better grouping. Does that work for you?

```
...
#define MEMCG_RECLAIM_MAY_SWAP (1 << 1)
#define MEMCG_RECLAIM_PROACTIVE (1 << 2)
#define MIN_SWAPPINESS 0
#define MAX_SWAPPINESS 200

/* Just reclaim from anon folios in proactive memory reclaim */
#define SWAPPINESS_ANON_ONLY (MAX_SWAPPINESS + 1)

extern int vm_swappiness;

static inline int mem_cgroup_swappiness(struct mem_cgroup *memcg)
{
#ifdef CONFIG_MEMCG_V1
	/* Cgroup2 doesn't have per-cgroup swappiness */
	if (cgroup_subsys_on_dfl(memory_cgrp_subsys))
		return READ_ONCE(vm_swappiness);

	/* root ? */
	if (mem_cgroup_disabled() || mem_cgroup_is_root(memcg))
		return READ_ONCE(vm_swappiness);

	return READ_ONCE(memcg->swappiness);
#else
	return READ_ONCE(vm_swappiness);
#endif
}
...
```


-- 
Best regards
Ridong
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.