* [PATCH v2 RFC] mm/vmscan: more restrictive condition for retry in do_try_to_free_pages
@ 2017-03-11 13:36 Yisheng Xie
0 siblings, 0 replies; 6+ messages in thread
From: Yisheng Xie @ 2017-03-11 13:36 UTC (permalink / raw)
To: akpm, hannes, mgorman, vbabka, mhocko, riel, shakeelb
Cc: linux-mm, linux-kernel, xieyisheng1, hanjunguo, qiuxishi
From: Yisheng Xie <xieyisheng1@huawei.com>
When we enter do_try_to_free_pages, the may_thrash is always clear, and
it will retry shrink zones to tap cgroup's reserves memory by setting
may_thrash when the former shrink_zones reclaim nothing.
However, when memcg is disabled or on legacy hierarchy, it should not do
this useless retry at all, for we do not have any cgroup's reserves
memory to tap, and we have already done hard work but made no progress.
To avoid this time costly and useless retrying, add a stub function
may_thrash and return true when memcg is disabled or on legacy
hierarchy.
Signed-off-by: Yisheng Xie <xieyisheng1@huawei.com>
Suggested-by: Shakeel Butt <shakeelb@google.com>
---
v2:
- more restrictive condition for retry of shrink_zones (restricting
cgroup_disabled=memory boot option and cgroup legacy hierarchy) - Shakeel
- add a stub function may_thrash() to avoid compile error or warning.
- rename subject from "donot retry shrink zones when memcg is disable"
to "more restrictive condition for retry in do_try_to_free_pages"
Any comment is more than welcome!
Thanks
Yisheng Xie
mm/vmscan.c | 20 +++++++++++++++++++-
1 file changed, 19 insertions(+), 1 deletion(-)
diff --git a/mm/vmscan.c b/mm/vmscan.c
index bc8031e..415f800 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -184,6 +184,19 @@ static bool sane_reclaim(struct scan_control *sc)
#endif
return false;
}
+
+static bool may_thrash(struct scan_control *sc)
+{
+ /*
+ * When memcg is disabled or on legacy hierarchy, there is no cgroup
+ * reserves memory to tap.
+ */
+ if (!cgroup_subsys_enabled(memory_cgrp_subsys) ||
+ !cgroup_subsys_on_dfl(memory_cgrp_subsys))
+ return true;
+
+ return sc->may_thrash;
+}
#else
static bool global_reclaim(struct scan_control *sc)
{
@@ -194,6 +207,11 @@ static bool sane_reclaim(struct scan_control *sc)
{
return true;
}
+
+static bool may_thrash(struct scan_control *sc)
+{
+ return true;
+}
#endif
/*
@@ -2808,7 +2826,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
return 1;
/* Untapped cgroup reserves? Don't OOM, retry. */
- if (!sc->may_thrash) {
+ if (!may_thrash(sc)) {
sc->priority = initial_priority;
sc->may_thrash = 1;
goto retry;
--
1.9.1
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH v2 RFC] mm/vmscan: more restrictive condition for retry in do_try_to_free_pages
@ 2017-03-11 13:51 Yisheng Xie
2017-03-11 17:52 ` Shakeel Butt
0 siblings, 1 reply; 6+ messages in thread
From: Yisheng Xie @ 2017-03-11 13:51 UTC (permalink / raw)
To: akpm, hannes, mgorman, vbabka, mhocko, riel, shakeelb
Cc: linux-mm, linux-kernel, xieyisheng1, guohanjun, qiuxishi
From: Yisheng Xie <xieyisheng1@huawei.com>
When we enter do_try_to_free_pages, the may_thrash is always clear, and
it will retry shrink zones to tap cgroup's reserves memory by setting
may_thrash when the former shrink_zones reclaim nothing.
However, when memcg is disabled or on legacy hierarchy, it should not do
this useless retry at all, for we do not have any cgroup's reserves
memory to tap, and we have already done hard work but made no progress.
To avoid this time costly and useless retrying, add a stub function
may_thrash and return true when memcg is disabled or on legacy
hierarchy.
Signed-off-by: Yisheng Xie <xieyisheng1@huawei.com>
Suggested-by: Shakeel Butt <shakeelb@google.com>
---
v2:
- more restrictive condition for retry of shrink_zones (restricting
cgroup_disabled=memory boot option and cgroup legacy hierarchy) - Shakeel
- add a stub function may_thrash() to avoid compile error or warning.
- rename subject from "donot retry shrink zones when memcg is disable"
to "more restrictive condition for retry in do_try_to_free_pages"
Any comment is more than welcome!
Thanks
Yisheng Xie
mm/vmscan.c | 20 +++++++++++++++++++-
1 file changed, 19 insertions(+), 1 deletion(-)
diff --git a/mm/vmscan.c b/mm/vmscan.c
index bc8031e..415f800 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -184,6 +184,19 @@ static bool sane_reclaim(struct scan_control *sc)
#endif
return false;
}
+
+static bool may_thrash(struct scan_control *sc)
+{
+ /*
+ * When memcg is disabled or on legacy hierarchy, there is no cgroup
+ * reserves memory to tap.
+ */
+ if (!cgroup_subsys_enabled(memory_cgrp_subsys) ||
+ !cgroup_subsys_on_dfl(memory_cgrp_subsys))
+ return true;
+
+ return sc->may_thrash;
+}
#else
static bool global_reclaim(struct scan_control *sc)
{
@@ -194,6 +207,11 @@ static bool sane_reclaim(struct scan_control *sc)
{
return true;
}
+
+static bool may_thrash(struct scan_control *sc)
+{
+ return true;
+}
#endif
/*
@@ -2808,7 +2826,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
return 1;
/* Untapped cgroup reserves? Don't OOM, retry. */
- if (!sc->may_thrash) {
+ if (!may_thrash(sc)) {
sc->priority = initial_priority;
sc->may_thrash = 1;
goto retry;
--
1.9.1
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH v2 RFC] mm/vmscan: more restrictive condition for retry in do_try_to_free_pages
2017-03-11 13:51 Yisheng Xie
@ 2017-03-11 17:52 ` Shakeel Butt
2017-03-12 10:15 ` Yisheng Xie
2017-03-16 19:57 ` Johannes Weiner
0 siblings, 2 replies; 6+ messages in thread
From: Shakeel Butt @ 2017-03-11 17:52 UTC (permalink / raw)
To: Yisheng Xie
Cc: Andrew Morton, Johannes Weiner, Mel Gorman, Vlastimil Babka,
Michal Hocko, riel, Linux MM, LKML, xieyisheng1, guohanjun,
Xishi Qiu
On Sat, Mar 11, 2017 at 5:51 AM, Yisheng Xie <ysxie@foxmail.com> wrote:
> From: Yisheng Xie <xieyisheng1@huawei.com>
>
> When we enter do_try_to_free_pages, the may_thrash is always clear, and
> it will retry shrink zones to tap cgroup's reserves memory by setting
> may_thrash when the former shrink_zones reclaim nothing.
>
> However, when memcg is disabled or on legacy hierarchy, it should not do
> this useless retry at all, for we do not have any cgroup's reserves
> memory to tap, and we have already done hard work but made no progress.
>
> To avoid this time costly and useless retrying, add a stub function
> may_thrash and return true when memcg is disabled or on legacy
> hierarchy.
>
> Signed-off-by: Yisheng Xie <xieyisheng1@huawei.com>
> Suggested-by: Shakeel Butt <shakeelb@google.com>
> ---
> v2:
> - more restrictive condition for retry of shrink_zones (restricting
> cgroup_disabled=memory boot option and cgroup legacy hierarchy) - Shakeel
>
> - add a stub function may_thrash() to avoid compile error or warning.
>
> - rename subject from "donot retry shrink zones when memcg is disable"
> to "more restrictive condition for retry in do_try_to_free_pages"
>
> Any comment is more than welcome!
>
> Thanks
> Yisheng Xie
>
> mm/vmscan.c | 20 +++++++++++++++++++-
> 1 file changed, 19 insertions(+), 1 deletion(-)
>
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index bc8031e..415f800 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -184,6 +184,19 @@ static bool sane_reclaim(struct scan_control *sc)
> #endif
> return false;
> }
> +
> +static bool may_thrash(struct scan_control *sc)
> +{
> + /*
> + * When memcg is disabled or on legacy hierarchy, there is no cgroup
> + * reserves memory to tap.
> + */
> + if (!cgroup_subsys_enabled(memory_cgrp_subsys) ||
> + !cgroup_subsys_on_dfl(memory_cgrp_subsys))
> + return true;
> +
> + return sc->may_thrash;
> +}
> #else
> static bool global_reclaim(struct scan_control *sc)
> {
> @@ -194,6 +207,11 @@ static bool sane_reclaim(struct scan_control *sc)
> {
> return true;
> }
> +
> +static bool may_thrash(struct scan_control *sc)
> +{
> + return true;
> +}
> #endif
>
> /*
> @@ -2808,7 +2826,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
> return 1;
>
> /* Untapped cgroup reserves? Don't OOM, retry. */
> - if (!sc->may_thrash) {
> + if (!may_thrash(sc)) {
Thanks Yisheng. The name of the function may_thrash() is confusing in
the sense that it is returning exactly the opposite of what its name
implies. How about reversing the condition of may_thrash() function
and change the scan_control's field may_thrash to thrashed?
> sc->priority = initial_priority;
> sc->may_thrash = 1;
> goto retry;
> --
> 1.9.1
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH v2 RFC] mm/vmscan: more restrictive condition for retry in do_try_to_free_pages
2017-03-11 17:52 ` Shakeel Butt
@ 2017-03-12 10:15 ` Yisheng Xie
2017-03-16 19:57 ` Johannes Weiner
1 sibling, 0 replies; 6+ messages in thread
From: Yisheng Xie @ 2017-03-12 10:15 UTC (permalink / raw)
To: Shakeel Butt
Cc: Andrew Morton, Johannes Weiner, Mel Gorman, Vlastimil Babka,
Michal Hocko, riel, Linux MM, LKML, xieyisheng1, guohanjun,
Xishi Qiu
hi, Shakeel,
On 03/12/2017 01:52 AM, Shakeel Butt wrote:
> On Sat, Mar 11, 2017 at 5:51 AM, Yisheng Xie <ysxie@foxmail.com> wrote:
>> From: Yisheng Xie <xieyisheng1@huawei.com>
>>
>> When we enter do_try_to_free_pages, the may_thrash is always clear, and
>> it will retry shrink zones to tap cgroup's reserves memory by setting
>> may_thrash when the former shrink_zones reclaim nothing.
>>
>> However, when memcg is disabled or on legacy hierarchy, it should not do
>> this useless retry at all, for we do not have any cgroup's reserves
>> memory to tap, and we have already done hard work but made no progress.
>>
>> To avoid this time costly and useless retrying, add a stub function
>> may_thrash and return true when memcg is disabled or on legacy
>> hierarchy.
>>
>> Signed-off-by: Yisheng Xie <xieyisheng1@huawei.com>
>> Suggested-by: Shakeel Butt <shakeelb@google.com>
>> ---
>>
>> return 1;
>>
>> /* Untapped cgroup reserves? Don't OOM, retry. */
>> - if (!sc->may_thrash) {
>> + if (!may_thrash(sc)) {
> Thanks Yisheng. The name of the function may_thrash() is confusing in
> the sense that it is returning exactly the opposite of what its name
> implies.
Right.
> How about reversing the condition of may_thrash() function
> and change the scan_control's field may_thrash to thrashed?
hmm, maybe I can change the may_thrash() function to mem_cgroup_thrashed().
For, if change the scan_control's may_thrash to thrashed, it may also looks
confusing in shrink_node, and it will be like:
if (mem_cgroup_low(root, memcg)) {
if (!sc->thrashed) -----> looks confuse here?
continue;
mem_cgroup_events(memcg, MEMCG_LOW, 1);
}
Thanks
Yisheng Xie
@
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH v2 RFC] mm/vmscan: more restrictive condition for retry in do_try_to_free_pages
2017-03-11 17:52 ` Shakeel Butt
2017-03-12 10:15 ` Yisheng Xie
@ 2017-03-16 19:57 ` Johannes Weiner
2017-03-16 20:26 ` Shakeel Butt
1 sibling, 1 reply; 6+ messages in thread
From: Johannes Weiner @ 2017-03-16 19:57 UTC (permalink / raw)
To: Shakeel Butt
Cc: Yisheng Xie, Andrew Morton, Mel Gorman, Vlastimil Babka,
Michal Hocko, riel, Linux MM, LKML, xieyisheng1, guohanjun,
Xishi Qiu
On Sat, Mar 11, 2017 at 09:52:15AM -0800, Shakeel Butt wrote:
> On Sat, Mar 11, 2017 at 5:51 AM, Yisheng Xie <ysxie@foxmail.com> wrote:
> > @@ -2808,7 +2826,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
> > return 1;
> >
> > /* Untapped cgroup reserves? Don't OOM, retry. */
> > - if (!sc->may_thrash) {
> > + if (!may_thrash(sc)) {
>
> Thanks Yisheng. The name of the function may_thrash() is confusing in
> the sense that it is returning exactly the opposite of what its name
> implies. How about reversing the condition of may_thrash() function
> and change the scan_control's field may_thrash to thrashed?
How so?
The user sets memory.low to a minimum below which the application will
thrash. Hence, being allowed to break that minimum and causing the app
to thrash, means you "may thrash".
OTOH, I'm not sure what "thrashed" would mean.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH v2 RFC] mm/vmscan: more restrictive condition for retry in do_try_to_free_pages
2017-03-16 19:57 ` Johannes Weiner
@ 2017-03-16 20:26 ` Shakeel Butt
0 siblings, 0 replies; 6+ messages in thread
From: Shakeel Butt @ 2017-03-16 20:26 UTC (permalink / raw)
To: Johannes Weiner
Cc: Yisheng Xie, Andrew Morton, Mel Gorman, Vlastimil Babka,
Michal Hocko, riel, Linux MM, LKML, xieyisheng1, guohanjun,
Xishi Qiu
On Thu, Mar 16, 2017 at 12:57 PM, Johannes Weiner <hannes@cmpxchg.org> wrote:
> On Sat, Mar 11, 2017 at 09:52:15AM -0800, Shakeel Butt wrote:
>> On Sat, Mar 11, 2017 at 5:51 AM, Yisheng Xie <ysxie@foxmail.com> wrote:
>> > @@ -2808,7 +2826,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
>> > return 1;
>> >
>> > /* Untapped cgroup reserves? Don't OOM, retry. */
>> > - if (!sc->may_thrash) {
>> > + if (!may_thrash(sc)) {
>>
>> Thanks Yisheng. The name of the function may_thrash() is confusing in
>> the sense that it is returning exactly the opposite of what its name
>> implies. How about reversing the condition of may_thrash() function
>> and change the scan_control's field may_thrash to thrashed?
>
> How so?
>
> The user sets memory.low to a minimum below which the application will
> thrash. Hence, being allowed to break that minimum and causing the app
> to thrash, means you "may thrash".
>
Basically how I interpreted may_thrash() is "may I thrash" or may I
reclaim memory from memcgs which were already below memory.low. So, if
it returns true, we go for second pass with the authorization to
reclaim memory even from memcgs with usage below memory.low.
> OTOH, I'm not sure what "thrashed" would mean.
By 'thrashed', I wanted to say, hey I have already tried to reclaim
memory from memcgs below their memory.low, so no need to try again but
Yisheng correctly pointed out that it will cause confusion in
shrink_node().
Sorry for confusion.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2017-03-16 20:26 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2017-03-11 13:36 [PATCH v2 RFC] mm/vmscan: more restrictive condition for retry in do_try_to_free_pages Yisheng Xie
-- strict thread matches above, loose matches on Subject: below --
2017-03-11 13:51 Yisheng Xie
2017-03-11 17:52 ` Shakeel Butt
2017-03-12 10:15 ` Yisheng Xie
2017-03-16 19:57 ` Johannes Weiner
2017-03-16 20:26 ` Shakeel Butt
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).