Skip to contentSkip to conversation

xenbus/backend: Add memory pressure handler callback

PATCH Superseded SeongJae Park 2 participants 12 messages Dec 18, 2019

Subsystems (3)

v12 of 14 Superseded Latest is v14 →

lore
Share & tools
lore
#0 lore raw

v12 cover xenbus/backend: Add memory pressure handler callback

Granting pages consumes backend system memory. In systems configured with insufficient spare memory for those pages, it can cause a memory pressure situation. However, finding the optimal amount of the spare memory is challenging for large systems having dynamic resource utilization patterns. Also, such a static configuration might lack flexibility.

To mitigate such problems, this patchset adds a memory reclaim callback to 'xenbus_driver' (patch 1) and then introduce a lock for race condition avoidance (patch 2). After that, patch 3 applies the callback mechanism to mitigate the problem in 'xen-blkback'. The fourth and fifth patches are trivial cleanups; those fix nits we found during the development of this patchset.

Note that patch 1, 4, and 5 are not changed since v9.

Base Version
------------

This patch is based on v5.4. A complete tree is also available at my public git repo:
https://github.com/sjp38/linux/tree/patches/blkback/buffer_squeeze/v12

Patch History
-------------

Changes from v11
(https://lore.kernel.org/xen-devel/[email protected]/)
- Fix wrong trylock use (reported by Juergen)
- Merge patch 3 and 4 (suggested by Juergen)
- Update test result

Changes from v10
(https://lore.kernel.org/xen-devel/[email protected]/)
- Fix race condition (reported by SeongJae, suggested by Juergen)

Changes from v9
(https://lore.kernel.org/xen-devel/[email protected]/)
- Add 'Reviewed-by' and 'Acked-by' from Roger Pau Monné
- Update the commit message for overhead test of the 2nd path

Changes from v8
(https://lore.kernel.org/xen-devel/[email protected]/)
- Drop 'Reviewed-by: Juergen' from the second patch (suggested by Roger Pau Monné)
- Update contact of the new module param to SeongJae Park <[email protected]>
(suggested by Roger Pau Monné)
- Wordsmith the description of the parameter
(suggested by Roger Pau Monné)
- Fix dumb bugs
(suggested by Roger Pau Monné)
- Move module param definition to xenbus.c and reduce the number of lines for this change
(suggested by Roger Pau Monné)
- Add a comment for the new callback, reclaim_memory, as other callbacks also have
- Add another trivial cleanup of xenbus.c file (4th patch)

Changes from v7
(https://lore.kernel.org/xen-devel/[email protected]/)
- Update sysfs-driver-xen-blkback for new parameter (suggested by Roger Pau Monné)
- Use per-xen_blkif buffer_squeeze_end instead of global variable (suggested by Roger Pau Monné)

Changes from v6
(https://lore.kernel.org/linux-block/[email protected]/)
- Remove more unnecessary prefixes (suggested by Roger Pau Monné)
- Constify a variable (suggested by Roger Pau Monné)
- Rename 'reclaim' into 'reclaim_memory' (suggested by Roger Pau Monné)
- More wordsmith of the commit message (suggested by Roger Pau Monné)

Changes from v5
(https://lore.kernel.org/linux-block/[email protected]/)
- Wordsmith the commit messages (suggested by Roger Pau Monné)
- Change the reclaim callback return type (suggested by Roger Pau Monné)
- Change the type of the blkback squeeze duration variable (suggested by Roger Pau Monné)
- Add a patch for removal of unnecessary static variable name prefixes (suggested by Roger Pau Monné)
- Fix checkpatch.pl warnings

Changes from v4
(https://lore.kernel.org/xen-devel/[email protected]/)
- Remove domain id parameter from the callback (suggested by Juergen Gross)
- Rename xen-blkback module parameter (suggested by Stefan Nuernburger)

Changes from v3
(https://lore.kernel.org/xen-devel/[email protected]/)
- Add general callback in xen_driver and use it (suggested by Juergen Gross)

Changes from v2
(https://lore.kernel.org/linux-block/[email protected])
- Rename the module parameter and variables for brevity (aggressive shrinking -> squeezing)

Changes from v1
(https://lore.kernel.org/xen-devel/[email protected]/)
- Adjust the description to not use the term, arbitrarily (suggested by Paul Durrant)
- Specify time unit of the duration in the parameter description, (suggested by Maximilian Heyne)
- Change default aggressive shrinking duration from 1ms to 10ms
- Merge two patches into one single patch

SeongJae Park (5):
xenbus/backend: Add memory pressure handler callback xenbus/backend: Protect xenbus callback with lock xen/blkback: Squeeze page pools if a memory pressure is detected xen/blkback: Remove unnecessary static variable name prefixes xen/blkback: Consistently insert one empty line between functions

 .../ABI/testing/sysfs-driver-xen-blkback      | 10 +++++
 drivers/block/xen-blkback/blkback.c           | 42 +++++++++----------
 drivers/block/xen-blkback/common.h            |  1 +
 drivers/block/xen-blkback/xenbus.c            | 37 +++++++++++++---
 drivers/xen/xenbus/xenbus_probe.c             |  1 +
 drivers/xen/xenbus/xenbus_probe_backend.c     | 39 +++++++++++++++++
 include/xen/xenbus.h                          |  3 ++
 7 files changed, 107 insertions(+), 26 deletions(-)
Signature · 1 line
2.17.1
#1 lore raw

PATCH v12 1/5 xenbus/backend: Add memory pressure handler callback

From: SeongJae Park <[email protected]>

Granting pages consumes backend system memory. In systems configured with insufficient spare memory for those pages, it can cause a memory pressure situation. However, finding the optimal amount of the spare memory is challenging for large systems having dynamic resource utilization patterns. Also, such a static configuration might lack flexibility.

To mitigate such problems, this commit adds a memory reclaim callback to 'xenbus_driver'. If a memory pressure is detected, 'xenbus' requests every backend driver to volunarily release its memory.

Note that it would be able to improve the callback facility for more sophisticated handlings of general pressures. For example, it would be possible to monitor the memory consumption of each device and issue the release requests to only devices which causing the pressure. Also, the callback could be extended to handle not only memory, but general resources. Nevertheless, this version of the implementation defers such sophisticated goals as a future work.

---
 drivers/xen/xenbus/xenbus_probe_backend.c | 32 +++++++++++++++++++++++
 include/xen/xenbus.h                      |  1 +
 2 files changed, 33 insertions(+)
drivers/xen/xenbus/xenbus_probe_backend.c +32 -0
index b0bed4faf44c..7e78ebef7c54 100644
@@ -248,6 +248,35 @@ static int backend_probe_and_watch(struct notifier_block *notifier,
 	return NOTIFY_DONE;
 }
 
+static int backend_reclaim_memory(struct device *dev, void *data)
+{
+	const struct xenbus_driver *drv;
+
+	if (!dev->driver)
+		return 0;
+	drv = to_xenbus_driver(dev->driver);
+	if (drv && drv->reclaim_memory)
+		drv->reclaim_memory(to_xenbus_device(dev));
+	return 0;
+}
+
+/*
+ * Returns 0 always because we are using shrinker to only detect memory
+ * pressure.
+ */
+static unsigned long backend_shrink_memory_count(struct shrinker *shrinker,
+				struct shrink_control *sc)
+{
+	bus_for_each_dev(&xenbus_backend.bus, NULL, NULL,
+			backend_reclaim_memory);
+	return 0;
+}
+
+static struct shrinker backend_memory_shrinker = {
+	.count_objects = backend_shrink_memory_count,
+	.seeks = DEFAULT_SEEKS,
+};
+
 static int __init xenbus_probe_backend_init(void)
 {
 	static struct notifier_block xenstore_notifier = {
@@ -264,6 +293,9 @@ static int __init xenbus_probe_backend_init(void)
 
 	register_xenstore_notifier(&xenstore_notifier);
 
+	if (register_shrinker(&backend_memory_shrinker))
+		pr_warn("shrinker registration failed\n");
+
 	return 0;
 }
 subsys_initcall(xenbus_probe_backend_init);
include/xen/xenbus.h +1 -0
index 869c816d5f8c..c861cfb6f720 100644
@@ -104,6 +104,7 @@ struct xenbus_driver {
 	struct device_driver driver;
 	int (*read_otherend_details)(struct xenbus_device *dev);
 	int (*is_ready)(struct xenbus_device *dev);
+	void (*reclaim_memory)(struct xenbus_device *dev);
 };
 
 static inline struct xenbus_driver *to_xenbus_driver(struct device_driver *drv)
Signature · 1 line
2.17.1
#2 lore raw

PATCH v12 2/5 xenbus/backend: Protect xenbus callback with lock

In reply to SeongJae Park

From: SeongJae Park <[email protected]>

'reclaim_memory' callback can race with a driver code as this callback will be called from any memory pressure detected context. To deal with the case, this commit adds a spinlock in the 'xenbus_device'. Whenever 'reclaim_memory' callback is called, the lock of the device which passed to the callback as its argument is locked. Thus, drivers registering their 'reclaim_memory' callback should protect the data that might race with the callback with the lock by themselves.

---
 drivers/xen/xenbus/xenbus_probe.c         |  1 +
 drivers/xen/xenbus/xenbus_probe_backend.c | 11 +++++++++--
 include/xen/xenbus.h                      |  2 ++
 3 files changed, 12 insertions(+), 2 deletions(-)
drivers/xen/xenbus/xenbus_probe.c +1 -0
index 5b471889d723..b86393f172e6 100644
@@ -472,6 +472,7 @@ int xenbus_probe_node(struct xen_bus_type *bus,
 		goto fail;
 
 	dev_set_name(&xendev->dev, "%s", devname);
+	spin_lock_init(&xendev->reclaim_lock);
 
 	/* Register with generic device framework. */
 	err = device_register(&xendev->dev);
drivers/xen/xenbus/xenbus_probe_backend.c +9 -2
index 7e78ebef7c54..e862cb932cc4 100644
@@ -251,12 +251,19 @@ static int backend_probe_and_watch(struct notifier_block *notifier,
 static int backend_reclaim_memory(struct device *dev, void *data)
 {
 	const struct xenbus_driver *drv;
+	struct xenbus_device *xdev;
+	unsigned long flags;
 
 	if (!dev->driver)
 		return 0;
 	drv = to_xenbus_driver(dev->driver);
-	if (drv && drv->reclaim_memory)
-		drv->reclaim_memory(to_xenbus_device(dev));
+	if (drv && drv->reclaim_memory) {
+		xdev = to_xenbus_device(dev);
+		if (!spin_trylock_irqsave(&xdev->reclaim_lock, flags))
+			return 0;
+		drv->reclaim_memory(xdev);
+		spin_unlock_irqrestore(&xdev->reclaim_lock, flags);
+	}
 	return 0;
 }
 
include/xen/xenbus.h +2 -0
index c861cfb6f720..d9468313061d 100644
@@ -76,6 +76,8 @@ struct xenbus_device {
 	enum xenbus_state state;
 	struct completion down;
 	struct work_struct work;
+	/* 'reclaim_memory' callback is called while this lock is acquired */
+	spinlock_t reclaim_lock;
 };
 
 static inline struct xenbus_device *to_xenbus_device(struct device *dev)
Signature · 1 line
2.17.1
maintainer
#3 lore raw

Re: [PATCH v12 2/5] xenbus/backend: Protect xenbus callback with lock

Quoted 9 lines · SeongJae Park

From: SeongJae Park <[email protected]>

'reclaim_memory' callback can race with a driver code as this callback will be called from any memory pressure detected context. To deal with the case, this commit adds a spinlock in the 'xenbus_device'. Whenever 'reclaim_memory' callback is called, the lock of the device which passed to the callback as its argument is locked. Thus, drivers registering their 'reclaim_memory' callback should protect the data that might race with the callback with the lock by themselves.

Any reason you don't take the lock around the .probe() and .remove() calls of the backend (xenbus_dev_probe() and xenbus_dev_remove())? This would eliminate the need to do that in each backend instead.

Juergen

#4 lore raw

Re: Re: [Xen-devel] [PATCH v12 2/5] xenbus/backend: Protect xenbus callback with lock

Quoted 11 lines · On Wed, 18 Dec 2019 13:27:37 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 9 lines · SeongJae Park

From: SeongJae Park <[email protected]>

'reclaim_memory' callback can race with a driver code as this callback will be called from any memory pressure detected context. To deal with the case, this commit adds a spinlock in the 'xenbus_device'. Whenever 'reclaim_memory' callback is called, the lock of the device which passed to the callback as its argument is locked. Thus, drivers registering their 'reclaim_memory' callback should protect the data that might race with the callback with the lock by themselves.

Quoted 3 lines

Any reason you don't take the lock around the .probe() and .remove() calls of the backend (xenbus_dev_probe() and xenbus_dev_remove())? This would eliminate the need to do that in each backend instead.

First of all, I would like to keep the critical section as small as possible. With my small test, I could see slightly increasing memory pressure as the critical section becomes wider. Also, some drivers might share the data their 'reclaim_memory' callback touches with other functions. I think only the driver owners can know what data is shared and what is the minimum critical section to protect it.

If you think differently or I am missing something, please let me know.

Thanks,
SeongJae Park

Quoted 3 lines

Juergen

maintainer
#5 lore raw

Re: [Xen-devel] [PATCH v12 2/5] xenbus/backend: Protect xenbus callback with lock

Quoted 17 lines · SeongJae Park
Quoted 11 lines · On Wed, 18 Dec 2019 13:27:37 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 9 lines · SeongJae Park

From: SeongJae Park <[email protected]>

'reclaim_memory' callback can race with a driver code as this callback will be called from any memory pressure detected context. To deal with the case, this commit adds a spinlock in the 'xenbus_device'. Whenever 'reclaim_memory' callback is called, the lock of the device which passed to the callback as its argument is locked. Thus, drivers registering their 'reclaim_memory' callback should protect the data that might race with the callback with the lock by themselves.

Quoted 3 lines

Any reason you don't take the lock around the .probe() and .remove() calls of the backend (xenbus_dev_probe() and xenbus_dev_remove())? This would eliminate the need to do that in each backend instead.

Quoted 6 lines

First of all, I would like to keep the critical section as small as possible. With my small test, I could see slightly increasing memory pressure as the critical section becomes wider. Also, some drivers might share the data their 'reclaim_memory' callback touches with other functions. I think only the driver owners can know what data is shared and what is the minimum critical section to protect it.

But this kind of serialization can still be added on top.

And with the trylock in the reclaim path I believe you can even avoid the irq variants of the spinlock. But I might be wrong, so you should try that with lockdep enabled. If it is working there is no harm done when making the critical section larger, as memory allocations will work as before.

Juergen

#6 lore raw

Re: Re: [Xen-devel] [PATCH v12 2/5] xenbus/backend: Protect xenbus callback with lock

Quoted 18 lines · On Wed, 18 Dec 2019 14:30:44 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 17 lines · SeongJae Park
Quoted 14 lines · On Wed, 18 Dec 2019 13:27:37 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 9 lines · SeongJae Park

From: SeongJae Park <[email protected]>

'reclaim_memory' callback can race with a driver code as this callback will be called from any memory pressure detected context. To deal with the case, this commit adds a spinlock in the 'xenbus_device'. Whenever 'reclaim_memory' callback is called, the lock of the device which passed to the callback as its argument is locked. Thus, drivers registering their 'reclaim_memory' callback should protect the data that might race with the callback with the lock by themselves.

Any reason you don't take the lock around the .probe() and .remove() calls of the backend (xenbus_dev_probe() and xenbus_dev_remove())? This would eliminate the need to do that in each backend instead.

Quoted 8 lines
Quoted 6 lines

First of all, I would like to keep the critical section as small as possible. With my small test, I could see slightly increasing memory pressure as the critical section becomes wider. Also, some drivers might share the data their 'reclaim_memory' callback touches with other functions. I think only the driver owners can know what data is shared and what is the minimum critical section to protect it.

But this kind of serialization can still be added on top.

I'm still worrying about the unnecessarily large critical section, but it might be small enough to be ignored. If no others have strong objection, I will take the lock around the '->probe()' and '->remove()'.

Quoted 6 lines

And with the trylock in the reclaim path I believe you can even avoid the irq variants of the spinlock. But I might be wrong, so you should try that with lockdep enabled. If it is working there is no harm done when making the critical section larger, as memory allocations will work as before.

Yes, you're right. I will try test with lockdep.

Thanks,
SeongJae Park

Quoted 3 lines

Juergen

maintainer
#7 lore raw

Re: [Xen-devel] [PATCH v12 2/5] xenbus/backend: Protect xenbus callback with lock

Quoted 27 lines · SeongJae Park
Quoted 25 lines · On Wed, 18 Dec 2019 14:30:44 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 23 lines · SeongJae Park
Quoted 11 lines · On Wed, 18 Dec 2019 13:27:37 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 9 lines · SeongJae Park

From: SeongJae Park <[email protected]>

'reclaim_memory' callback can race with a driver code as this callback will be called from any memory pressure detected context. To deal with the case, this commit adds a spinlock in the 'xenbus_device'. Whenever 'reclaim_memory' callback is called, the lock of the device which passed to the callback as its argument is locked. Thus, drivers registering their 'reclaim_memory' callback should protect the data that might race with the callback with the lock by themselves.

Quoted 3 lines

Any reason you don't take the lock around the .probe() and .remove() calls of the backend (xenbus_dev_probe() and xenbus_dev_remove())? This would eliminate the need to do that in each backend instead.

First of all, I would like to keep the critical section as small as possible. With my small test, I could see slightly increasing memory pressure as the critical section becomes wider. Also, some drivers might share the data their 'reclaim_memory' callback touches with other functions. I think only the driver owners can know what data is shared and what is the minimum critical section to protect it.

Quoted 5 lines
Quoted 1 line

But this kind of serialization can still be added on top.

I'm still worrying about the unnecessarily large critical section, but it might be small enough to be ignored. If no others have strong objection, I will take the lock around the '->probe()' and '->remove()'.

The lock is per device, so contention is possible only for the reclaim case. In case probe or remove are running reclaim will have nothing to free (in probe case nothing is allocated yet, in remove case everything should be freed anyway). So the larger critical section is no problem at all IMO.

Quoted 7 lines
Quoted 5 lines

And with the trylock in the reclaim path I believe you can even avoid the irq variants of the spinlock. But I might be wrong, so you should try that with lockdep enabled. If it is working there is no harm done when making the critical section larger, as memory allocations will work as before.

Yes, you're right. I will try test with lockdep.

Thanks,

Juergen

#8 lore raw

Re: Re: [Xen-devel] [PATCH v12 2/5] xenbus/backend: Protect xenbus callback with lock

Quoted 34 lines · On Wed, 18 Dec 2019 16:11:51 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 27 lines · SeongJae Park
Quoted 25 lines · On Wed, 18 Dec 2019 14:30:44 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 23 lines · SeongJae Park
Quoted 11 lines · On Wed, 18 Dec 2019 13:27:37 +0100 "Jürgen Groß" <[email protected]> wrote:
Quoted 9 lines · SeongJae Park

From: SeongJae Park <[email protected]>

'reclaim_memory' callback can race with a driver code as this callback will be called from any memory pressure detected context. To deal with the case, this commit adds a spinlock in the 'xenbus_device'. Whenever 'reclaim_memory' callback is called, the lock of the device which passed to the callback as its argument is locked. Thus, drivers registering their 'reclaim_memory' callback should protect the data that might race with the callback with the lock by themselves.

Quoted 3 lines

Any reason you don't take the lock around the .probe() and .remove() calls of the backend (xenbus_dev_probe() and xenbus_dev_remove())? This would eliminate the need to do that in each backend instead.

First of all, I would like to keep the critical section as small as possible. With my small test, I could see slightly increasing memory pressure as the critical section becomes wider. Also, some drivers might share the data their 'reclaim_memory' callback touches with other functions. I think only the driver owners can know what data is shared and what is the minimum critical section to protect it.

Quoted 5 lines
Quoted 1 line

But this kind of serialization can still be added on top.

I'm still worrying about the unnecessarily large critical section, but it might be small enough to be ignored. If no others have strong objection, I will take the lock around the '->probe()' and '->remove()'.

Quoted 5 lines

The lock is per device, so contention is possible only for the reclaim case. In case probe or remove are running reclaim will have nothing to free (in probe case nothing is allocated yet, in remove case everything should be freed anyway). So the larger critical section is no problem at all IMO.

Agreed. I think I was worried about nothing really existing now.

Quoted 10 lines
Quoted 8 lines
Quoted 6 lines

And with the trylock in the reclaim path I believe you can even avoid the irq variants of the spinlock. But I might be wrong, so you should try that with lockdep enabled. If it is working there is no harm done when making the critical section larger, as memory allocations will work as before.

Yes, you're right. I will try test with lockdep.

Thanks,

Good news, lockdep says it's okay :)

Will post next version soon.

Thanks,
SeongJae Park

Quoted 3 lines

Juergen

#9 lore raw

PATCH v12 3/5 xen/blkback: Squeeze page pools if a memory pressure is detected

In reply to SeongJae Park

From: SeongJae Park <[email protected]>

Each blkif has a free pages pool for the grant mapping. The size of the pool starts from zero and is increased on demand while processing the I/O requests. If current I/O requests handling is finished or 100 milliseconds has passed since last I/O requests handling, it checks and shrinks the pool to not exceed the size limit, max_buffer_pages.

Therefore, host administrators can cause memory pressure in blkback by attaching a large number of block devices and inducing I/O. Such problematic situations can be avoided by limiting the maximum number of devices that can be attached, but finding the optimal limit is not so easy. Improper set of the limit can results in memory pressure or a resource underutilization. This commit avoids such problematic situations by squeezing the pools (returns every free page in the pool to the system) for a while (users can set this duration via a module parameter) if memory pressure is detected.

Discussions
===========

The blkback's original shrinking mechanism returns only pages in the pool which are not currently be used by blkback to the system. In other words, the pages that are not mapped with granted pages. Because this commit is changing only the shrink limit but still uses the same freeing mechanism it does not touch pages which are currently mapping grants.

Once memory pressure is detected, this commit keeps the squeezing limit for a user-specified time duration. The duration should be neither too long nor too short. If it is too long, the squeezing incurring overhead can reduce the I/O performance. If it is too short, blkback will not free enough pages to reduce the memory pressure. This commit sets the value as 10 milliseconds by default because it is a short time in terms of I/O while it is a long time in terms of memory operations. Also, as the original shrinking mechanism works for at least every 100 milliseconds, this could be a somewhat reasonable choice. I also tested other durations (refer to the below section for more details) and confirmed that 10 milliseconds is the one that works best with the test. That said, the proper duration depends on actual configurations and workloads. That's why this commit allows users to set the duration as a module parameter.

Memory Pressure Test
====================

To show how this commit fixes the memory pressure situation well, I configured a test environment on a xen-running virtualization system. On the blkfront running guest instances, I attach a large number of network-backed volume devices and induce I/O to those. Meanwhile, I measure the number of pages that swapped in (pswpin) and out (pswpout) on the blkback running guest. The test ran twice, once for the blkback before this commit and once for that after this commit. As shown below, this commit has dramatically reduced the memory pressure:

                pswpin  pswpout
    before      76,672  185,799
    after          867    3,967

Optimal Aggressive Shrinking Duration
-------------------------------------

To find a best squeezing duration, I repeated the test with three different durations (1ms, 10ms, and 100ms). The results are as below:

    duration    pswpin  pswpout
    1           707     5,095
    10          867     3,967
    100         362     3,348

As expected, the memory pressure decreases as the duration increases, but the reduction become slow from the 10ms. Based on this results, I chose the default duration as 10ms.

Performance Overhead Test
=========================

This commit could incur I/O performance degradation under severe memory pressure because the squeezing will require more page allocations per I/O. To show the overhead, I artificially made a worst-case squeezing situation and measured the I/O performance of a blkfront running guest.

For the artificial squeezing, I set the blkback.max_buffer_pages using the /sys/module/xen_blkback/parameters/max_buffer_pages file. In this test, I set the value to 1024 and 0. The 1024 is the default value. Setting the value as 0 is same to a situation doing the squeezing always (worst-case).

If the underlying block device is slow enough, the squeezing overhead could be hidden. For the reason, I use a fast block device, namely the rbd[1]:

    # xl block-attach guest phy:/dev/ram0 xvdb w

For the I/O performance measurement, I run a simple dd command 5 times directly to the device as below and collect the 'MB/s' results.

    $ for i in {1..5}; do dd if=/dev/zero of=/dev/xvdb \
                             bs=4k count=$((256*512)); sync; done

The results are as below. 'max_pgs' represents the value of the blkback.max_buffer_pages parameter.

    max_pgs   Min       Max       Median     Avg    Stddev
    0         417       423       420        419.4  2.5099801
    1024      414       425       416        417.8  4.4384682

No difference proven at 95.0% confidence

In short, even worst case squeezing on ramdisk based fast block device makes no visible performance degradation. Please note that this is just a very simple and minimal test. On systems using super-fast block devices and a special I/O workload, the results might be different. If you have any doubt, test on your machine with your workload to find the optimal squeezing duration for you.

[1] https://www.kernel.org/doc/html/latest/admin-guide/blockdev/ramdisk.html

---
 .../ABI/testing/sysfs-driver-xen-blkback      | 10 ++++++
 drivers/block/xen-blkback/blkback.c           |  7 ++--
 drivers/block/xen-blkback/common.h            |  1 +
 drivers/block/xen-blkback/xenbus.c            | 32 ++++++++++++++++++-
 4 files changed, 47 insertions(+), 3 deletions(-)
Documentation/ABI/testing/sysfs-driver-xen-blkback +10 -0
index 4e7babb3ba1f..f01224231f3f 100644
@@ -25,3 +25,13 @@ Description:
                 allocated without being in use. The time is in
                 seconds, 0 means indefinitely long.
                 The default is 60 seconds.
+
+What:           /sys/module/xen_blkback/parameters/buffer_squeeze_duration_ms
+Date:           December 2019
+KernelVersion:  5.5
+Contact:        SeongJae Park <[email protected]>
+Description:
+                When memory pressure is reported to blkback this option
+                controls the duration in milliseconds that blkback will not
+                cache any page not backed by a grant mapping.
+                The default is 10ms.
drivers/block/xen-blkback/blkback.c +5 -2
index fd1e19f1a49f..79f677aeb5cc 100644
@@ -656,8 +656,11 @@ int xen_blkif_schedule(void *arg)
 			ring->next_lru = jiffies + msecs_to_jiffies(LRU_INTERVAL);
 		}
 
-		/* Shrink if we have more than xen_blkif_max_buffer_pages */
-		shrink_free_pagepool(ring, xen_blkif_max_buffer_pages);
+		/* Shrink the free pages pool if it is too large. */
+		if (time_before(jiffies, blkif->buffer_squeeze_end))
+			shrink_free_pagepool(ring, 0);
+		else
+			shrink_free_pagepool(ring, xen_blkif_max_buffer_pages);
 
 		if (log_stats && time_after(jiffies, ring->st_print))
 			print_stats(ring);
drivers/block/xen-blkback/common.h +1 -0
index 1d3002d773f7..536c84f61fed 100644
@@ -319,6 +319,7 @@ struct xen_blkif {
 	/* All rings for this device. */
 	struct xen_blkif_ring	*rings;
 	unsigned int		nr_rings;
+	unsigned long		buffer_squeeze_end;
 };
 
 struct seg_buf {
drivers/block/xen-blkback/xenbus.c +31 -1
index b90dbcd99c03..20045827a391 100644
@@ -492,6 +492,7 @@ static int xen_vbd_create(struct xen_blkif *blkif, blkif_vdev_t handle,
 static int xen_blkbk_remove(struct xenbus_device *dev)
 {
 	struct backend_info *be = dev_get_drvdata(&dev->dev);
+	unsigned long flags;
 
 	pr_debug("%s %p %d\n", __func__, dev, dev->otherend_id);
 
@@ -504,6 +505,7 @@ static int xen_blkbk_remove(struct xenbus_device *dev)
 		be->backend_watch.node = NULL;
 	}
 
+	spin_lock_irqsave(&dev->reclaim_lock, flags);
 	dev_set_drvdata(&dev->dev, NULL);
 
 	if (be->blkif) {
@@ -512,6 +514,7 @@ static int xen_blkbk_remove(struct xenbus_device *dev)
 		/* Put the reference we set in xen_blkif_alloc(). */
 		xen_blkif_put(be->blkif);
 	}
+	spin_unlock_irqrestore(&dev->reclaim_lock, flags);
 
 	return 0;
 }
@@ -597,6 +600,7 @@ static int xen_blkbk_probe(struct xenbus_device *dev,
 	int err;
 	struct backend_info *be = kzalloc(sizeof(struct backend_info),
 					  GFP_KERNEL);
+	unsigned long flags;
 
 	/* match the pr_debug in xen_blkbk_remove */
 	pr_debug("%s %p %d\n", __func__, dev, dev->otherend_id);
@@ -607,6 +611,7 @@ static int xen_blkbk_probe(struct xenbus_device *dev,
 		return -ENOMEM;
 	}
 	be->dev = dev;
+	spin_lock_irqsave(&dev->reclaim_lock, flags);
 	dev_set_drvdata(&dev->dev, be);
 
 	be->blkif = xen_blkif_alloc(dev->otherend_id);
@@ -614,8 +619,10 @@ static int xen_blkbk_probe(struct xenbus_device *dev,
 		err = PTR_ERR(be->blkif);
 		be->blkif = NULL;
 		xenbus_dev_fatal(dev, err, "creating block interface");
+		spin_unlock_irqrestore(&dev->reclaim_lock, flags);
 		goto fail;
 	}
+	spin_unlock_irqrestore(&dev->reclaim_lock, flags);
 
 	err = xenbus_printf(XBT_NIL, dev->nodename,
 			    "feature-max-indirect-segments", "%u",
@@ -824,6 +831,28 @@ static void frontend_changed(struct xenbus_device *dev,
 }
 
 
+/* Once a memory pressure is detected, squeeze free page pools for a while. */
+static unsigned int buffer_squeeze_duration_ms = 10;
+module_param_named(buffer_squeeze_duration_ms,
+		buffer_squeeze_duration_ms, int, 0644);
+MODULE_PARM_DESC(buffer_squeeze_duration_ms,
+"Duration in ms to squeeze pages buffer when a memory pressure is detected");
+
+/*
+ * Callback received when the memory pressure is detected.
+ */
+static void reclaim_memory(struct xenbus_device *dev)
+{
+	struct backend_info *be = dev_get_drvdata(&dev->dev);
+
+	/* Device is registered but not probed yet */
+	if (!be)
+		return;
+
+	be->blkif->buffer_squeeze_end = jiffies +
+		msecs_to_jiffies(buffer_squeeze_duration_ms);
+}
+
 /* ** Connection ** */
 
 
@@ -1115,7 +1144,8 @@ static struct xenbus_driver xen_blkbk_driver = {
 	.ids  = xen_blkbk_ids,
 	.probe = xen_blkbk_probe,
 	.remove = xen_blkbk_remove,
-	.otherend_changed = frontend_changed
+	.otherend_changed = frontend_changed,
+	.reclaim_memory = reclaim_memory,
 };
 
 int xen_blkif_xenbus_init(void)
Signature · 1 line
2.17.1
#10 lore raw

PATCH v12 4/5 xen/blkback: Remove unnecessary static variable name prefixes

In reply to SeongJae Park

From: SeongJae Park <[email protected]>

A few of static variables in blkback have 'xen_blkif_' prefix, though it is unnecessary for static variables. This commit removes such prefixes.

---
 drivers/block/xen-blkback/blkback.c | 37 +++++++++++++----------------
 1 file changed, 17 insertions(+), 20 deletions(-)
drivers/block/xen-blkback/blkback.c +17 -20
index 79f677aeb5cc..fbd67f8e4e4e 100644
@@ -62,8 +62,8 @@
  * IO workloads.
  */
 
-static int xen_blkif_max_buffer_pages = 1024;
-module_param_named(max_buffer_pages, xen_blkif_max_buffer_pages, int, 0644);
+static int max_buffer_pages = 1024;
+module_param_named(max_buffer_pages, max_buffer_pages, int, 0644);
 MODULE_PARM_DESC(max_buffer_pages,
 "Maximum number of free pages to keep in each block backend buffer");
 
@@ -78,8 +78,8 @@ MODULE_PARM_DESC(max_buffer_pages,
  * algorithm.
  */
 
-static int xen_blkif_max_pgrants = 1056;
-module_param_named(max_persistent_grants, xen_blkif_max_pgrants, int, 0644);
+static int max_pgrants = 1056;
+module_param_named(max_persistent_grants, max_pgrants, int, 0644);
 MODULE_PARM_DESC(max_persistent_grants,
                  "Maximum number of grants to map persistently");
 
@@ -88,8 +88,8 @@ MODULE_PARM_DESC(max_persistent_grants,
  * use. The time is in seconds, 0 means indefinitely long.
  */
 
-static unsigned int xen_blkif_pgrant_timeout = 60;
-module_param_named(persistent_grant_unused_seconds, xen_blkif_pgrant_timeout,
+static unsigned int pgrant_timeout = 60;
+module_param_named(persistent_grant_unused_seconds, pgrant_timeout,
 		   uint, 0644);
 MODULE_PARM_DESC(persistent_grant_unused_seconds,
 		 "Time in seconds an unused persistent grant is allowed to "
@@ -137,9 +137,8 @@ module_param(log_stats, int, 0644);
 
 static inline bool persistent_gnt_timeout(struct persistent_gnt *persistent_gnt)
 {
-	return xen_blkif_pgrant_timeout &&
-	       (jiffies - persistent_gnt->last_used >=
-		HZ * xen_blkif_pgrant_timeout);
+	return pgrant_timeout && (jiffies - persistent_gnt->last_used >=
+			HZ * pgrant_timeout);
 }
 
 static inline int get_free_page(struct xen_blkif_ring *ring, struct page **page)
@@ -234,7 +233,7 @@ static int add_persistent_gnt(struct xen_blkif_ring *ring,
 	struct persistent_gnt *this;
 	struct xen_blkif *blkif = ring->blkif;
 
-	if (ring->persistent_gnt_c >= xen_blkif_max_pgrants) {
+	if (ring->persistent_gnt_c >= max_pgrants) {
 		if (!blkif->vbd.overflow_max_grants)
 			blkif->vbd.overflow_max_grants = 1;
 		return -EBUSY;
@@ -397,14 +396,13 @@ static void purge_persistent_gnt(struct xen_blkif_ring *ring)
 		goto out;
 	}
 
-	if (ring->persistent_gnt_c < xen_blkif_max_pgrants ||
-	    (ring->persistent_gnt_c == xen_blkif_max_pgrants &&
+	if (ring->persistent_gnt_c < max_pgrants ||
+	    (ring->persistent_gnt_c == max_pgrants &&
 	    !ring->blkif->vbd.overflow_max_grants)) {
 		num_clean = 0;
 	} else {
-		num_clean = (xen_blkif_max_pgrants / 100) * LRU_PERCENT_CLEAN;
-		num_clean = ring->persistent_gnt_c - xen_blkif_max_pgrants +
-			    num_clean;
+		num_clean = (max_pgrants / 100) * LRU_PERCENT_CLEAN;
+		num_clean = ring->persistent_gnt_c - max_pgrants + num_clean;
 		num_clean = min(ring->persistent_gnt_c, num_clean);
 		pr_debug("Going to purge at least %u persistent grants\n",
 			 num_clean);
@@ -599,8 +597,7 @@ static void print_stats(struct xen_blkif_ring *ring)
 		 current->comm, ring->st_oo_req,
 		 ring->st_rd_req, ring->st_wr_req,
 		 ring->st_f_req, ring->st_ds_req,
-		 ring->persistent_gnt_c,
-		 xen_blkif_max_pgrants);
+		 ring->persistent_gnt_c, max_pgrants);
 	ring->st_print = jiffies + msecs_to_jiffies(10 * 1000);
 	ring->st_rd_req = 0;
 	ring->st_wr_req = 0;
@@ -660,7 +657,7 @@ int xen_blkif_schedule(void *arg)
 		if (time_before(jiffies, blkif->buffer_squeeze_end))
 			shrink_free_pagepool(ring, 0);
 		else
-			shrink_free_pagepool(ring, xen_blkif_max_buffer_pages);
+			shrink_free_pagepool(ring, max_buffer_pages);
 
 		if (log_stats && time_after(jiffies, ring->st_print))
 			print_stats(ring);
@@ -887,7 +884,7 @@ static int xen_blkbk_map(struct xen_blkif_ring *ring,
 			continue;
 		}
 		if (use_persistent_gnts &&
-		    ring->persistent_gnt_c < xen_blkif_max_pgrants) {
+		    ring->persistent_gnt_c < max_pgrants) {
 			/*
 			 * We are using persistent grants, the grant is
 			 * not mapped but we might have room for it.
@@ -914,7 +911,7 @@ static int xen_blkbk_map(struct xen_blkif_ring *ring,
 			pages[seg_idx]->persistent_gnt = persistent_gnt;
 			pr_debug("grant %u added to the tree of persistent grants, using %u/%u\n",
 				 persistent_gnt->gnt, ring->persistent_gnt_c,
-				 xen_blkif_max_pgrants);
+				 max_pgrants);
 			goto next;
 		}
 		if (use_persistent_gnts && !blkif->vbd.overflow_max_grants) {
Signature · 1 line
2.17.1
#11 lore raw

PATCH v12 5/5 xen/blkback: Consistently insert one empty line between functions

In reply to SeongJae Park

From: SeongJae Park <[email protected]>

The number of empty lines between functions in the xenbus.c is inconsistent. This trivial style cleanup commit fixes the file to consistently place only one empty line.

---
 drivers/block/xen-blkback/xenbus.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
drivers/block/xen-blkback/xenbus.c +2 -5
index 20045827a391..453f97dd533d 100644
@@ -432,7 +432,6 @@ static void xenvbd_sysfs_delif(struct xenbus_device *dev)
 	device_remove_file(&dev->dev, &dev_attr_physical_device);
 }
 
-
 static void xen_vbd_free(struct xen_vbd *vbd)
 {
 	if (vbd->bdev)
@@ -489,6 +488,7 @@ static int xen_vbd_create(struct xen_blkif *blkif, blkif_vdev_t handle,
 		handle, blkif->domid);
 	return 0;
 }
+
 static int xen_blkbk_remove(struct xenbus_device *dev)
 {
 	struct backend_info *be = dev_get_drvdata(&dev->dev);
@@ -575,6 +575,7 @@ static void xen_blkbk_discard(struct xenbus_transaction xbt, struct backend_info
 	if (err)
 		dev_warn(&dev->dev, "writing feature-discard (%d)", err);
 }
+
 int xen_blkbk_barrier(struct xenbus_transaction xbt,
 		      struct backend_info *be, int state)
 {
@@ -663,7 +664,6 @@ static int xen_blkbk_probe(struct xenbus_device *dev,
 	return err;
 }
 
-
 /*
  * Callback received when the hotplug scripts have placed the physical-device
  * node.  Read it and the mode node, and create a vbd.  If the frontend is
@@ -755,7 +755,6 @@ static void backend_changed(struct xenbus_watch *watch,
 	}
 }
 
-
 /*
  * Callback received when the frontend's state changes.
  */
@@ -830,7 +829,6 @@ static void frontend_changed(struct xenbus_device *dev,
 	}
 }
 
-
 /* Once a memory pressure is detected, squeeze free page pools for a while. */
 static unsigned int buffer_squeeze_duration_ms = 10;
 module_param_named(buffer_squeeze_duration_ms,
@@ -855,7 +853,6 @@ static void reclaim_memory(struct xenbus_device *dev)
 
 /* ** Connection ** */
 
-
 /*
  * Write the physical details regarding the block device to the store, and
  * switch to Connected state.
Signature · 1 line
2.17.1