From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
To: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: "linux-mm@kvack.org" <linux-mm@kvack.org>,
"containers@lists.osdl.org" <containers@lists.osdl.org>,
Andrew Morton <akpm@linux-foundation.org>,
"balbir@linux.vnet.ibm.com" <balbir@linux.vnet.ibm.com>,
"yamamoto@valinux.co.jp" <yamamoto@valinux.co.jp>,
"riel@redhat.com" <riel@redhat.com>,
xemul@openvz.org
Subject: [RFC][for -mm] memory controller enhancements for reclaiming take2 [6/8] high_low watermark for res_counter
Date: Mon, 3 Dec 2007 18:41:41 +0900 [thread overview]
Message-ID: <20071203184141.0c5d22f9.kamezawa.hiroyu@jp.fujitsu.com> (raw)
In-Reply-To: <20071203183355.0061ddeb.kamezawa.hiroyu@jp.fujitsu.com>
This patch adds high/low watermark parameter to res_counter.
and check routine.
splitted out from YAMAMOTO's background page reclaim for memory cgroup set.
TODO?
- if res_counter's user doesn't want high/low watermark, res_counter_write()
should ignore low <= high <= limit limitation ?
Changelog
* added param watermark_state this can ba read without lock lock.
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
From: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
include/linux/res_counter.h | 28 ++++++++++++++++++++++++++++
kernel/res_counter.c | 42 +++++++++++++++++++++++++++++++++++++++++-
2 files changed, 69 insertions(+), 1 deletion(-)
Index: linux-2.6.24-rc3-mm2/include/linux/res_counter.h
===================================================================
--- linux-2.6.24-rc3-mm2.orig/include/linux/res_counter.h
+++ linux-2.6.24-rc3-mm2/include/linux/res_counter.h
@@ -19,6 +19,12 @@
* the helpers described beyond
*/
+enum watermark_state {
+ RES_WMARK_BELOW_LOW,
+ RES_WMARK_ABOVE_LOW,
+ RES_WMARK_ABOVE_HIGH,
+};
+
struct res_counter {
/*
* the current resource consumption level
@@ -33,10 +39,17 @@ struct res_counter {
*/
unsigned long long failcnt;
/*
+ * Watermarks. Must keep low <= high <= limit.
+ */
+ unsigned long long high_watermark;
+ unsigned long long low_watermark;
+ /*
* the lock to protect all of the above.
* the routines below consider this to be IRQ-safe
*/
spinlock_t lock;
+ /* can be read without lock */
+ enum watermark_state watermark_state;
};
/*
@@ -66,6 +79,8 @@ enum {
RES_USAGE,
RES_LIMIT,
RES_FAILCNT,
+ RES_HWMARK,
+ RES_LWMARK,
};
/*
@@ -124,4 +139,17 @@ static inline bool res_counter_check_und
return ret;
}
+/*
+ * Helper function for implementing high/low watermark to resource controller.
+ */
+static inline bool res_counter_below_low_watermark(struct res_counter *cnt)
+{
+ return (cnt->watermark_state == RES_WMARK_BELOW_LOW);
+}
+
+static inline bool res_counter_above_high_watermark(struct res_counter *cnt)
+{
+ return (cnt->watermark_state == RES_WMARK_ABOVE_HIGH);
+}
+
#endif
Index: linux-2.6.24-rc3-mm2/kernel/res_counter.c
===================================================================
--- linux-2.6.24-rc3-mm2.orig/kernel/res_counter.c
+++ linux-2.6.24-rc3-mm2/kernel/res_counter.c
@@ -17,6 +17,9 @@ void res_counter_init(struct res_counter
{
spin_lock_init(&counter->lock);
counter->limit = (unsigned long long)LLONG_MAX;
+ counter->low_watermark = (unsigned long long)LLONG_MAX;
+ counter->high_watermark = (unsigned long long)LLONG_MAX;
+ counter->watermark_state = RES_WMARK_BELOW_LOW;
}
int res_counter_charge_locked(struct res_counter *counter, unsigned long val)
@@ -27,6 +30,12 @@ int res_counter_charge_locked(struct res
}
counter->usage += val;
+
+ if (counter->usage > counter->high_watermark)
+ counter->watermark_state = RES_WMARK_ABOVE_HIGH;
+ else if (counter->usage > counter->low_watermark)
+ counter->watermark_state = RES_WMARK_ABOVE_LOW;
+
return 0;
}
@@ -47,6 +56,11 @@ void res_counter_uncharge_locked(struct
val = counter->usage;
counter->usage -= val;
+
+ if (counter->usage < counter->low_watermark)
+ counter->watermark_state = RES_WMARK_BELOW_LOW;
+ else if (counter->usage < counter->high_watermark)
+ counter->watermark_state = RES_WMARK_ABOVE_LOW;
}
void res_counter_uncharge(struct res_counter *counter, unsigned long val)
@@ -69,6 +83,10 @@ res_counter_member(struct res_counter *c
return &counter->limit;
case RES_FAILCNT:
return &counter->failcnt;
+ case RES_HWMARK:
+ return &counter->high_watermark;
+ case RES_LWMARK:
+ return &counter->low_watermark;
};
BUG();
@@ -123,12 +141,34 @@ ssize_t res_counter_write(struct res_cou
goto out_free;
}
spin_lock_irqsave(&counter->lock, flags);
+ /*
+ * check low <= high <= limit.
+ */
+ switch (member) {
+ case RES_LIMIT:
+ if (counter->high_watermark > tmp)
+ goto unlock_free;
+ break;
+ case RES_HWMARK:
+ if (tmp < counter->low_watermark ||
+ tmp > counter->limit)
+ goto unlock_free;
+ break;
+ case RES_LWMARK:
+ if (tmp > counter->high_watermark)
+ goto unlock_free;
+ break;
+ default:
+ break;
+ }
val = res_counter_member(counter, member);
*val = tmp;
- spin_unlock_irqrestore(&counter->lock, flags);
ret = nbytes;
+unlock_free:
+ spin_unlock_irqrestore(&counter->lock, flags);
out_free:
kfree(buf);
out:
return ret;
}
+
--
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>
next prev parent reply other threads:[~2007-12-03 9:41 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-12-03 9:33 [RFC][for -mm] memory controller enhancements for reclaiming take2 [0/8] introduction KAMEZAWA Hiroyuki
2007-12-03 9:35 ` [RFC][for -mm] memory controller enhancements for reclaiming take2 [1/8] clean up : remove unused variable KAMEZAWA Hiroyuki
2007-12-04 15:55 ` Balbir Singh
2007-12-03 9:36 ` [RFC][for -mm] memory controller enhancements for reclaiming take2 [2/8] add BUG_ON() in mem_cgroup_zoneinfo KAMEZAWA Hiroyuki
2007-12-04 16:01 ` Balbir Singh
2007-12-03 9:37 ` [RFC][for -mm] memory controller enhancements for reclaiming take2 [3/8] define free_mem_cgroup_per_zone_info KAMEZAWA Hiroyuki
2007-12-04 16:32 ` Balbir Singh
2007-12-03 9:38 ` [RFC][for -mm] memory controller enhancements for reclaiming take2 [4/8] possible race fix in res_counter KAMEZAWA Hiroyuki
2007-12-04 19:02 ` Balbir Singh
2007-12-03 9:39 ` [RFC][for -mm] memory controller enhancements for reclaiming take2 [5/8] throttling simultaneous callers of try_to_free_mem_cgroup_pages KAMEZAWA Hiroyuki
2007-12-03 14:24 ` Rik van Riel
2007-12-04 1:33 ` KAMEZAWA Hiroyuki
2007-12-04 13:27 ` Balbir Singh
2007-12-05 0:26 ` KAMEZAWA Hiroyuki
2007-12-03 9:41 ` KAMEZAWA Hiroyuki [this message]
2007-12-03 9:42 ` [RFC][for -mm] memory controller enhancements for reclaiming take2 [7/8] bacground reclaim for memory controller KAMEZAWA Hiroyuki
2007-12-04 3:07 ` YAMAMOTO Takashi
2007-12-04 3:18 ` KAMEZAWA Hiroyuki
2007-12-04 3:31 ` YAMAMOTO Takashi
2007-12-03 9:45 ` [RFC][for -mm] memory controller enhancements for reclaiming take2 [8/8] wake up waiters at unchage KAMEZAWA Hiroyuki
2007-12-04 6:46 ` [RFC][for -mm] memory controller enhancements for reclaiming take2 [0/8] introduction KAMEZAWA Hiroyuki
2007-12-04 14:25 ` Balbir Singh
2007-12-05 0:44 ` KAMEZAWA Hiroyuki
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20071203184141.0c5d22f9.kamezawa.hiroyu@jp.fujitsu.com \
--to=kamezawa.hiroyu@jp.fujitsu.com \
--cc=akpm@linux-foundation.org \
--cc=balbir@linux.vnet.ibm.com \
--cc=containers@lists.osdl.org \
--cc=linux-mm@kvack.org \
--cc=riel@redhat.com \
--cc=xemul@openvz.org \
--cc=yamamoto@valinux.co.jp \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox