mm: bdi: allow setting a minimum for the bdi dirty limit

Under normal circumstances each device is given a part of the total write-back cache that relates to its current avg writeout speed in relation to the other devices. min_ratio - allows one to assign a minimum portion of the write-back cache to a particular device. This is useful in situations where you might want to provide a minimum QoS. (One request for this feature came from flash based storage people who wanted to avoid writing out at all costs - they of course needed some pdflush hacks as well) max_ratio - allows one to assign a maximum portion of the dirty limit to a particular device. This is useful in situations where you want to avoid one device taking all or most of the write-back cache. Eg. an NFS mount that is prone to get stuck, or a FUSE mount which you don't trust to play fair. Add "min_ratio" to /sys/class/bdi. This indicates the minimum percentage of the global dirty threshold allocated to this bdi. [mszeredi@suse.cz] - fix parsing in min_ratio_store() - document new sysfs attribute Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: bdi: allow setting a minimum for the bdi dirty limit
Under normal circumstances each device is given a part of the total write-back cache that relates to its current avg writeout speed in relation to the other devices. min_ratio - allows one to assign a minimum portion of the write-back cache to a particular device. This is useful in situations where you might want to provide a minimum QoS. (One request for this feature came from flash based storage people who wanted to avoid writing out at all costs - they of course needed some pdflush hacks as well) max_ratio - allows one to assign a maximum portion of the dirty limit to a particular device. This is useful in situations where you want to avoid one device taking all or most of the write-back cache. Eg. an NFS mount that is prone to get stuck, or a FUSE mount which you don't trust to play fair. Add "min_ratio" to /sys/class/bdi. This indicates the minimum percentage of the global dirty threshold allocated to this bdi. [mszeredi@suse.cz] - fix parsing in min_ratio_store() - document new sysfs attribute Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Miklos Szeredi <mszeredi@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Peter Zijlstra · Linus Torvalds
1 parent b6f2fcbcfc
Showing 4 changed files with 57 additions and 1 deletions Side-by-side Diff
Documentation/ABI/testing/sysfs-class-bdi
include/linux/backing-dev.h
mm/backing-dev.c
mm/page-writeback.c
@@ -43,4 +43,11 @@
  
 	Current threshold on this BDI for reclaimable + writeback
 	memory
+
+min_ratio (read-write)
+
+	Minimal percentage of global dirty threshold allocated to this
+	bdi.  If the value written to this file would make the the sum
+	of all min_ratio values exceed 100, then EINVAL is returned.
+	The default is zero
@@ -51,6 +51,8 @@
 	struct prop_local_percpu completions;
 	int dirty_exceeded;
  
+	unsigned int min_ratio;
+
 	struct device *dev;
 };
  
@@ -136,6 +138,8 @@
 	return 1;
 #endif
 }
+
+int bdi_set_min_ratio(struct backing_dev_info *bdi, unsigned int min_ratio);
  
 /*
  * Flags in backing_dev_info::capability
@@ -55,6 +55,24 @@
 BDI_SHOW(dirty_kb, K(get_dirty(bdi, 1)))
 BDI_SHOW(bdi_dirty_kb, K(get_dirty(bdi, 2)))
  
+static ssize_t min_ratio_store(struct device *dev,
+		struct device_attribute *attr, const char *buf, size_t count)
+{
+	struct backing_dev_info *bdi = dev_get_drvdata(dev);
+	char *end;
+	unsigned int ratio;
+	ssize_t ret = -EINVAL;
+
+	ratio = simple_strtoul(buf, &end, 10);
+	if (*buf && (end[0] == '\0' || (end[0] == '\n' && end[1] == '\0'))) {
+		ret = bdi_set_min_ratio(bdi, ratio);
+		if (!ret)
+			ret = count;
+	}
+	return ret;
+}
+BDI_SHOW(min_ratio, bdi->min_ratio)
+
 #define __ATTR_RW(attr) __ATTR(attr, 0644, attr##_show, attr##_store)
  
 static struct device_attribute bdi_dev_attrs[] = {
@@ -63,6 +81,7 @@
 	__ATTR_RO(writeback_kb),
 	__ATTR_RO(dirty_kb),
 	__ATTR_RO(bdi_dirty_kb),
+	__ATTR_RW(min_ratio),
 	__ATTR_NULL,
 };
  
@@ -126,6 +145,8 @@
 	int err;
  
 	bdi->dev = NULL;
+
+	bdi->min_ratio = 0;
  
 	for (i = 0; i < NR_BDI_STAT_ITEMS; i++) {
 		err = percpu_counter_init_irq(&bdi->bdi_stat[i], 0);
@@ -243,6 +243,29 @@
 }
  
 /*
+ *
+ */
+static DEFINE_SPINLOCK(bdi_lock);
+static unsigned int bdi_min_ratio;
+
+int bdi_set_min_ratio(struct backing_dev_info *bdi, unsigned int min_ratio)
+{
+	int ret = 0;
+	unsigned long flags;
+
+	spin_lock_irqsave(&bdi_lock, flags);
+	min_ratio -= bdi->min_ratio;
+	if (bdi_min_ratio + min_ratio < 100) {
+		bdi_min_ratio += min_ratio;
+		bdi->min_ratio += min_ratio;
+	} else
+		ret = -EINVAL;
+	spin_unlock_irqrestore(&bdi_lock, flags);
+
+	return ret;
+}
+
+/*
  * Work out the current dirty-memory clamping and background writeout
  * thresholds.
  *
@@ -330,7 +353,7 @@
 	*pdirty = dirty;
  
 	if (bdi) {
-		u64 bdi_dirty = dirty;
+		u64 bdi_dirty;
 		long numerator, denominator;
  
 		/*
  
@@ -338,8 +361,10 @@
 		 */
 		bdi_writeout_fraction(bdi, &numerator, &denominator);
  
+		bdi_dirty = (dirty * (100 - bdi_min_ratio)) / 100;
 		bdi_dirty *= numerator;
 		do_div(bdi_dirty, denominator);
+		bdi_dirty += (dirty * bdi->min_ratio) / 100;
  
 		*pbdi_dirty = bdi_dirty;
 		clip_bdi_dirty_limit(bdi, dirty, pbdi_dirty);
...	...	@@ -43,4 +43,11 @@
43	43
44	44	Current threshold on this BDI for reclaimable + writeback
45	45	memory
	46	+
	47	+min_ratio (read-write)
	48	+
	49	+ Minimal percentage of global dirty threshold allocated to this
	50	+ bdi. If the value written to this file would make the the sum
	51	+ of all min_ratio values exceed 100, then EINVAL is returned.
	52	+ The default is zero
...	...	@@ -51,6 +51,8 @@
51	51	struct prop_local_percpu completions;
52	52	int dirty_exceeded;
53	53
	54	+ unsigned int min_ratio;
	55	+
54	56	struct device *dev;
55	57	};
56	58
...	...	@@ -136,6 +138,8 @@
136	138	return 1;
137	139	#endif
138	140	}
	141	+
	142	+int bdi_set_min_ratio(struct backing_dev_info *bdi, unsigned int min_ratio);
139	143
140	144	/*
141	145	* Flags in backing_dev_info::capability
...	...	@@ -55,6 +55,24 @@
55	55	BDI_SHOW(dirty_kb, K(get_dirty(bdi, 1)))
56	56	BDI_SHOW(bdi_dirty_kb, K(get_dirty(bdi, 2)))
57	57
	58	+static ssize_t min_ratio_store(struct device *dev,
	59	+ struct device_attribute attr, const char buf, size_t count)
	60	+{
	61	+ struct backing_dev_info *bdi = dev_get_drvdata(dev);
	62	+ char *end;
	63	+ unsigned int ratio;
	64	+ ssize_t ret = -EINVAL;
	65	+
	66	+ ratio = simple_strtoul(buf, &end, 10);
	67	+ if (*buf && (end[0] == '\0' \|\| (end[0] == '\n' && end[1] == '\0'))) {
	68	+ ret = bdi_set_min_ratio(bdi, ratio);
	69	+ if (!ret)
	70	+ ret = count;
	71	+ }
	72	+ return ret;
	73	+}
	74	+BDI_SHOW(min_ratio, bdi->min_ratio)
	75	+
58	76	#define __ATTR_RW(attr) __ATTR(attr, 0644, attr##_show, attr##_store)
59	77
60	78	static struct device_attribute bdi_dev_attrs[] = {
...	...	@@ -63,6 +81,7 @@
63	81	__ATTR_RO(writeback_kb),
64	82	__ATTR_RO(dirty_kb),
65	83	__ATTR_RO(bdi_dirty_kb),
	84	+ __ATTR_RW(min_ratio),
66	85	__ATTR_NULL,
67	86	};
68	87
...	...	@@ -126,6 +145,8 @@
126	145	int err;
127	146
128	147	bdi->dev = NULL;
	148	+
	149	+ bdi->min_ratio = 0;
129	150
130	151	for (i = 0; i < NR_BDI_STAT_ITEMS; i++) {
131	152	err = percpu_counter_init_irq(&bdi->bdi_stat[i], 0);
...	...	@@ -243,6 +243,29 @@
243	243	}
244	244
245	245	/*
	246	+ *
	247	+ */
	248	+static DEFINE_SPINLOCK(bdi_lock);
	249	+static unsigned int bdi_min_ratio;
	250	+
	251	+int bdi_set_min_ratio(struct backing_dev_info *bdi, unsigned int min_ratio)
	252	+{
	253	+ int ret = 0;
	254	+ unsigned long flags;
	255	+
	256	+ spin_lock_irqsave(&bdi_lock, flags);
	257	+ min_ratio -= bdi->min_ratio;
	258	+ if (bdi_min_ratio + min_ratio < 100) {
	259	+ bdi_min_ratio += min_ratio;
	260	+ bdi->min_ratio += min_ratio;
	261	+ } else
	262	+ ret = -EINVAL;
	263	+ spin_unlock_irqrestore(&bdi_lock, flags);
	264	+
	265	+ return ret;
	266	+}
	267	+
	268	+/*
246	269	* Work out the current dirty-memory clamping and background writeout
247	270	* thresholds.
248	271	*
...	...	@@ -330,7 +353,7 @@
330	353	*pdirty = dirty;
331	354
332	355	if (bdi) {
333		- u64 bdi_dirty = dirty;
	356	+ u64 bdi_dirty;
334	357	long numerator, denominator;
335	358
336	359	/*
337	360
...	...	@@ -338,8 +361,10 @@
338	361	*/
339	362	bdi_writeout_fraction(bdi, &numerator, &denominator);
340	363
	364	+ bdi_dirty = (dirty * (100 - bdi_min_ratio)) / 100;
341	365	bdi_dirty *= numerator;
342	366	do_div(bdi_dirty, denominator);
	367	+ bdi_dirty += (dirty * bdi->min_ratio) / 100;
343	368
344	369	*pbdi_dirty = bdi_dirty;
345	370	clip_bdi_dirty_limit(bdi, dirty, pbdi_dirty);