On Fri, Jun 21, 2013 at 09:24:38AM +0900, Hyunhee Kim wrote:
> In the original vmpressure, events are triggered whenever there is a reclaim
> activity. This becomes overheads to user space module and also increases
Not true.
We have lots of filter to not trigger event even if reclaim is going on.
Your statement would make confuse.
> power consumption if there is somebody to listen to it. This patch provides
> options to trigger events only when the pressure level changes.
> This trigger option can be set when registering each event by writing
> a trigger option, "edge" or "always", next to the string of levels.
> "edge" means that the event is triggered only when the pressure level is changed.
> "always" means that events are triggered whenever there is a reclaim process.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Not true, either.
> To keep backward compatibility, "always" is set by default if nothing is input
> as an option. Each event can have different option. For example,
> "low" level uses "always" trigger option to see reclaim activity at user space
> while "medium"/"critical" uses "edge" to do an important job
> like killing tasks only once.
Question.
1. user: set critical edge
2. kernel: memory is tight and trigger event with critical
3. user: kill a program when he receives a event
4. kernel: memory is very tight again and want to trigger a event
with critical but fail because last_level was critical and it was edge.
Right?
>
> Signed-off-by: Hyunhee Kim <hyunhee.kim@xxxxxxxxxxx>
> Signed-off-by: Kyungmin Park <kyungmin.park@xxxxxxxxxxx>
> ---
> Documentation/cgroups/memory.txt | 12 ++++++++++--
> mm/vmpressure.c | 36 ++++++++++++++++++++++++++++++++----
> 2 files changed, 42 insertions(+), 6 deletions(-)
>
> diff --git a/Documentation/cgroups/memory.txt b/Documentation/cgroups/memory.txt
> index ddf4f93..185870f 100644
> --- a/Documentation/cgroups/memory.txt
> +++ b/Documentation/cgroups/memory.txt
> @@ -807,13 +807,21 @@ register a notification, an application must:
>
> - create an eventfd using eventfd(2);
> - open memory.pressure_level;
> -- write string like "<event_fd> <fd of memory.pressure_level> <level>"
> +- write string like
> + "<event_fd> <fd of memory.pressure_level> <level> <trigger_option>"
> to cgroup.event_control.
>
> Application will be notified through eventfd when memory pressure is at
> the specific level (or higher). Read/write operations to
> memory.pressure_level are no implemented.
>
> +Events can be triggered whenever there is a reclaim activity or
Not true.
> +only when the pressure level changes. Trigger option is decided
> +by writing it next to the level. The event whose trigger option is
> +"always" is triggered whenever there is a reclaim process.
Not true.
> +If "edge" is set, the event is triggered only when the level is changed.
> +If the trigger option is not set, "always" is set by default.
> +
> Test:
>
> Here is a small script example that makes a new cgroup, sets up a
> @@ -823,7 +831,7 @@ Test:
> # cd /sys/fs/cgroup/memory/
> # mkdir foo
> # cd foo
> - # cgroup_event_listener memory.pressure_level low &
> + # cgroup_event_listener memory.pressure_level low edge &
> # echo 8000000 > memory.limit_in_bytes
> # echo 8000000 > memory.memsw.limit_in_bytes
> # echo $$ > tasks
> diff --git a/mm/vmpressure.c b/mm/vmpressure.c
> index 736a601..a08252e 100644
> --- a/mm/vmpressure.c
> +++ b/mm/vmpressure.c
> @@ -137,6 +137,8 @@ static enum vmpressure_levels vmpressure_calc_level(unsigned long scanned,
> struct vmpressure_event {
> struct eventfd_ctx *efd;
> enum vmpressure_levels level;
> + int last_level;
int? but level is enum vmpressure_levels?
> + bool edge_trigger;
> struct list_head node;
> };
>
> @@ -153,11 +155,14 @@ static bool vmpressure_event(struct vmpressure *vmpr,
>
> list_for_each_entry(ev, &vmpr->events, node) {
> if (level >= ev->level) {
> + if (ev->edge_trigger && level == ev->last_level)
> + continue;
> +
> eventfd_signal(ev->efd, 1);
> signalled = true;
> }
> + ev->last_level = level;
> }
> -
Unnecessary change.
> mutex_unlock(&vmpr->events_lock);
>
> return signalled;
> @@ -290,9 +295,11 @@ void vmpressure_prio(gfp_t gfp, struct mem_cgroup *memcg, int prio)
> *
> * This function associates eventfd context with the vmpressure
> * infrastructure, so that the notifications will be delivered to the
> - * @eventfd. The @args parameter is a string that denotes pressure level
> + * @eventfd. The @args parameters are a string that denotes pressure level
> * threshold (one of vmpressure_str_levels, i.e. "low", "medium", or
> - * "critical").
> + * "critical") and a trigger option that decides whether events are triggered
> + * continuously or only on edge ("always" or "edge" if "edge", events
> + * are triggered when the pressure level changes.
> *
> * This function should not be used directly, just pass it to (struct
> * cftype).register_event, and then cgroup core will handle everything by
> @@ -303,22 +310,43 @@ int vmpressure_register_event(struct cgroup *cg, struct cftype *cft,
> {
> struct vmpressure *vmpr = cg_to_vmpressure(cg);
> struct vmpressure_event *ev;
> + char *strlevel, *strtrigger;
> int level;
> + bool trigger;
What trigger?
Would be better to use "bool egde" instead?
> +
> + strlevel = args;
> + strtrigger = strchr(args, ' ');
> +
> + if (strtrigger) {
> + *strtrigger = '\0';
> + strtrigger++;
> + }
>
> for (level = 0; level < VMPRESSURE_NUM_LEVELS; level++) {
> - if (!strcmp(vmpressure_str_levels[level], args))
> + if (!strcmp(vmpressure_str_levels[level], strlevel))
> break;
> }
>
> if (level >= VMPRESSURE_NUM_LEVELS)
> return -EINVAL;
>
> + if (strtrigger == NULL)
> + trigger = false;
> + else if (!strcmp(strtrigger, "always"))
> + trigger = false;
> + else if (!strcmp(strtrigger, "edge"))
> + trigger = true;
> + else
> + return -EINVAL;
> +
> ev = kzalloc(sizeof(*ev), GFP_KERNEL);
> if (!ev)
> return -ENOMEM;
>
> ev->efd = eventfd;
> ev->level = level;
> + ev->last_level = -1;
VMPRESSURE_NONE is better?
> + ev->edge_trigger = trigger;
>
> mutex_lock(&vmpr->events_lock);
> list_add(&ev->node, &vmpr->events);
> --
> 1.7.9.5
>
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@xxxxxxxxx. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@xxxxxxxxx. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
--
Kind regards,
Minchan Kim
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>