[PATCH 1/2] drm/shmem: Add madvise state and purge helpers

[PATCH 1/2] drm/shmem: Add madvise state and purge helpers

[Rob Herring]

Add support to the shmem GEM helpers for tracking madvise state and
purging pages. This is based on the msm implementation.

The BO provides a list_head, but the list management is handled outside
of the shmem helpers as there are different locking requirements.

Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <maxime.ripard@bootlin.com>
Cc: Sean Paul <sean@poorly.run>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Eric Anholt <eric@anholt.net>
Signed-off-by: Rob Herring <robh@kernel.org>
---
 drivers/gpu/drm/drm_gem_shmem_helper.c | 57 ++++++++++++++++++++++++++
 include/drm/drm_gem_shmem_helper.h     | 15 +++++++
 2 files changed, 72 insertions(+)

diff --git a/drivers/gpu/drm/drm_gem_shmem_helper.c b/drivers/gpu/drm/drm_gem_shmem_helper.c
index 2f64667ac805..4b442576de1c 100644
--- a/drivers/gpu/drm/drm_gem_shmem_helper.c
+++ b/drivers/gpu/drm/drm_gem_shmem_helper.c
@@ -75,6 +75,7 @@ struct drm_gem_shmem_object *drm_gem_shmem_create(struct drm_device *dev, size_t
 	shmem = to_drm_gem_shmem_obj(obj);
 	mutex_init(&shmem->pages_lock);
 	mutex_init(&shmem->vmap_lock);
+	INIT_LIST_HEAD(&shmem->madv_list);
 
 	/*
 	 * Our buffers are kept pinned, so allocating them
@@ -362,6 +363,62 @@ drm_gem_shmem_create_with_handle(struct drm_file *file_priv,
 }
 EXPORT_SYMBOL(drm_gem_shmem_create_with_handle);
 
+/* Update madvise status, returns true if not purged, else
+ * false or -errno.
+ */
+int drm_gem_shmem_madvise(struct drm_gem_object *obj, int madv)
+{
+	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
+
+	mutex_lock(&shmem->pages_lock);
+
+	if (shmem->madv >= 0)
+		shmem->madv = madv;
+
+	madv = shmem->madv;
+
+	mutex_unlock(&shmem->pages_lock);
+
+	return (madv >= 0);
+}
+EXPORT_SYMBOL(drm_gem_shmem_madvise);
+
+void drm_gem_shmem_purge_locked(struct drm_gem_object *obj)
+{
+	struct drm_device *dev = obj->dev;
+	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
+
+	WARN_ON(!drm_gem_shmem_is_purgeable(shmem));
+
+	drm_gem_shmem_put_pages_locked(shmem);
+
+	shmem->madv = -1;
+
+	drm_vma_node_unmap(&obj->vma_node, dev->anon_inode->i_mapping);
+	drm_gem_free_mmap_offset(obj);
+
+	/* Our goal here is to return as much of the memory as
+	 * is possible back to the system as we are called from OOM.
+	 * To do this we must instruct the shmfs to drop all of its
+	 * backing pages, *now*.
+	 */
+	shmem_truncate_range(file_inode(obj->filp), 0, (loff_t)-1);
+
+	invalidate_mapping_pages(file_inode(obj->filp)->i_mapping,
+			0, (loff_t)-1);
+}
+EXPORT_SYMBOL(drm_gem_shmem_purge_locked);
+
+void drm_gem_shmem_purge(struct drm_gem_object *obj)
+{
+	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
+
+	mutex_lock(&shmem->pages_lock);
+	drm_gem_shmem_purge_locked(obj);
+	mutex_unlock(&shmem->pages_lock);
+}
+EXPORT_SYMBOL(drm_gem_shmem_purge);
+
 /**
  * drm_gem_shmem_dumb_create - Create a dumb shmem buffer object
  * @file: DRM file structure to create the dumb buffer for
diff --git a/include/drm/drm_gem_shmem_helper.h b/include/drm/drm_gem_shmem_helper.h
index 038b6d313447..ce1600fdfc3e 100644
--- a/include/drm/drm_gem_shmem_helper.h
+++ b/include/drm/drm_gem_shmem_helper.h
@@ -44,6 +44,9 @@ struct drm_gem_shmem_object {
 	 */
 	unsigned int pages_use_count;
 
+	int madv;
+	struct list_head madv_list;
+
 	/**
 	 * @pages_mark_dirty_on_put:
 	 *
@@ -121,6 +124,18 @@ void drm_gem_shmem_unpin(struct drm_gem_object *obj);
 void *drm_gem_shmem_vmap(struct drm_gem_object *obj);
 void drm_gem_shmem_vunmap(struct drm_gem_object *obj, void *vaddr);
 
+int drm_gem_shmem_madvise(struct drm_gem_object *obj, int madv);
+
+static inline bool drm_gem_shmem_is_purgeable(struct drm_gem_shmem_object *shmem)
+{
+	return (shmem->madv > 0) &&
+		!shmem->vmap_use_count && shmem->sgt &&
+		!shmem->base.dma_buf && !shmem->base.import_attach;
+}
+
+void drm_gem_shmem_purge_locked(struct drm_gem_object *obj);
+void drm_gem_shmem_purge(struct drm_gem_object *obj);
+
 struct drm_gem_shmem_object *
 drm_gem_shmem_create_with_handle(struct drm_file *file_priv,
 				 struct drm_device *dev, size_t size,
-- 
2.20.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 2/2] drm/panfrost: Add madvise and shrinker support

[Rob Herring]

Add support for madvise and a shrinker similar to other drivers. This
allows userspace to mark BOs which can be freed when there is memory
pressure.

Unlike other implementations, we don't depend on struct_mutex. The
driver maintains a list of BOs which can be freed when the shrinker
is called. Access to the list is serialized with the shrinker_lock.

Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Rob Herring <robh@kernel.org>
---
 drivers/gpu/drm/panfrost/Makefile             |   1 +
 drivers/gpu/drm/panfrost/TODO                 |   3 -
 drivers/gpu/drm/panfrost/panfrost_device.h    |   4 +
 drivers/gpu/drm/panfrost/panfrost_drv.c       |  38 +++++++
 drivers/gpu/drm/panfrost/panfrost_gem.c       |   5 +
 drivers/gpu/drm/panfrost/panfrost_gem.h       |   3 +
 .../gpu/drm/panfrost/panfrost_gem_shrinker.c  | 107 ++++++++++++++++++
 include/uapi/drm/panfrost_drm.h               |  22 ++++
 8 files changed, 180 insertions(+), 3 deletions(-)
 create mode 100644 drivers/gpu/drm/panfrost/panfrost_gem_shrinker.c

diff --git a/drivers/gpu/drm/panfrost/Makefile b/drivers/gpu/drm/panfrost/Makefile
index ecf0864cb515..b71935862417 100644
--- a/drivers/gpu/drm/panfrost/Makefile
+++ b/drivers/gpu/drm/panfrost/Makefile
@@ -5,6 +5,7 @@ panfrost-y := \
 	panfrost_device.o \
 	panfrost_devfreq.o \
 	panfrost_gem.o \
+	panfrost_gem_shrinker.o \
 	panfrost_gpu.o \
 	panfrost_job.o \
 	panfrost_mmu.o \
diff --git a/drivers/gpu/drm/panfrost/TODO b/drivers/gpu/drm/panfrost/TODO
index c2e44add37d8..f3beca8d4ce9 100644
--- a/drivers/gpu/drm/panfrost/TODO
+++ b/drivers/gpu/drm/panfrost/TODO
@@ -18,10 +18,7 @@
 
 - Support userspace controlled GPU virtual addresses. Needed for Vulkan. (Tomeu)
 
-- Support for madvise and a shrinker.
-
 - Compute job support. So called 'compute only' jobs need to be plumbed up to
   userspace.
 
 - Performance counter support. (Boris)
-
diff --git a/drivers/gpu/drm/panfrost/panfrost_device.h b/drivers/gpu/drm/panfrost/panfrost_device.h
index 83cc01cafde1..9751e8dc9ec6 100644
--- a/drivers/gpu/drm/panfrost/panfrost_device.h
+++ b/drivers/gpu/drm/panfrost/panfrost_device.h
@@ -84,6 +84,10 @@ struct panfrost_device {
 	struct mutex sched_lock;
 	struct mutex reset_lock;
 
+	struct mutex shrinker_lock;
+	struct list_head shrinker_list;
+	struct shrinker shrinker;
+
 	struct {
 		struct devfreq *devfreq;
 		struct thermal_cooling_device *cooling;
diff --git a/drivers/gpu/drm/panfrost/panfrost_drv.c b/drivers/gpu/drm/panfrost/panfrost_drv.c
index cb43ff4ebf4a..399007d306ea 100644
--- a/drivers/gpu/drm/panfrost/panfrost_drv.c
+++ b/drivers/gpu/drm/panfrost/panfrost_drv.c
@@ -301,6 +301,38 @@ static int panfrost_ioctl_get_bo_offset(struct drm_device *dev, void *data,
 	return 0;
 }
 
+static int panfrost_ioctl_madvise(struct drm_device *dev, void *data,
+				  struct drm_file *file_priv)
+{
+	struct drm_panfrost_madvise *args = data;
+	struct panfrost_device *pfdev = dev->dev_private;
+	struct drm_gem_object *gem_obj;
+
+	gem_obj = drm_gem_object_lookup(file_priv, args->handle);
+	if (!gem_obj) {
+		DRM_DEBUG("Failed to look up GEM BO %d\n", args->handle);
+		return -ENOENT;
+	}
+
+	args->retained = drm_gem_shmem_madvise(gem_obj, args->madv);
+
+	if (args->retained) {
+		struct panfrost_gem_object *bo = to_panfrost_bo(gem_obj);
+
+		mutex_lock(&pfdev->shrinker_lock);
+
+		if (args->madv == PANFROST_MADV_DONTNEED)
+			list_add_tail(&bo->base.madv_list, &pfdev->shrinker_list);
+		else if (args->madv == PANFROST_MADV_WILLNEED)
+			list_del_init(&bo->base.madv_list);
+
+		mutex_unlock(&pfdev->shrinker_lock);
+	}
+
+	drm_gem_object_put_unlocked(gem_obj);
+	return 0;
+}
+
 int panfrost_unstable_ioctl_check(void)
 {
 	if (!unstable_ioctls)
@@ -352,6 +384,7 @@ static const struct drm_ioctl_desc panfrost_drm_driver_ioctls[] = {
 	PANFROST_IOCTL(GET_BO_OFFSET,	get_bo_offset,	DRM_RENDER_ALLOW),
 	PANFROST_IOCTL(PERFCNT_ENABLE,	perfcnt_enable,	DRM_RENDER_ALLOW),
 	PANFROST_IOCTL(PERFCNT_DUMP,	perfcnt_dump,	DRM_RENDER_ALLOW),
+	PANFROST_IOCTL(MADVISE,		madvise,	DRM_RENDER_ALLOW),
 };
 
 DEFINE_DRM_GEM_SHMEM_FOPS(panfrost_drm_driver_fops);
@@ -400,6 +433,8 @@ static int panfrost_probe(struct platform_device *pdev)
 	pfdev->ddev = ddev;
 
 	spin_lock_init(&pfdev->mm_lock);
+	mutex_init(&pfdev->shrinker_lock);
+	INIT_LIST_HEAD(&pfdev->shrinker_list);
 
 	/* 4G enough for now. can be 48-bit */
 	drm_mm_init(&pfdev->mm, SZ_32M >> PAGE_SHIFT, (SZ_4G - SZ_32M) >> PAGE_SHIFT);
@@ -430,6 +465,8 @@ static int panfrost_probe(struct platform_device *pdev)
 	if (err < 0)
 		goto err_out1;
 
+	panfrost_gem_shrinker_init(ddev);
+
 	return 0;
 
 err_out1:
@@ -446,6 +483,7 @@ static int panfrost_remove(struct platform_device *pdev)
 	struct drm_device *ddev = pfdev->ddev;
 
 	drm_dev_unregister(ddev);
+	panfrost_gem_shrinker_cleanup(ddev);
 	pm_runtime_get_sync(pfdev->dev);
 	pm_runtime_put_sync_autosuspend(pfdev->dev);
 	pm_runtime_disable(pfdev->dev);
diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.c b/drivers/gpu/drm/panfrost/panfrost_gem.c
index 543ab1b81bd5..67d374184340 100644
--- a/drivers/gpu/drm/panfrost/panfrost_gem.c
+++ b/drivers/gpu/drm/panfrost/panfrost_gem.c
@@ -26,6 +26,11 @@ static void panfrost_gem_free_object(struct drm_gem_object *obj)
 	drm_mm_remove_node(&bo->node);
 	spin_unlock(&pfdev->mm_lock);
 
+	mutex_lock(&pfdev->shrinker_lock);
+	if (!list_empty(&bo->base.madv_list))
+		list_del(&bo->base.madv_list);
+	mutex_unlock(&pfdev->shrinker_lock);
+
 	drm_gem_shmem_free_object(obj);
 }
 
diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.h b/drivers/gpu/drm/panfrost/panfrost_gem.h
index 6dbcaba020fc..5f51f881ea3f 100644
--- a/drivers/gpu/drm/panfrost/panfrost_gem.h
+++ b/drivers/gpu/drm/panfrost/panfrost_gem.h
@@ -27,4 +27,7 @@ panfrost_gem_prime_import_sg_table(struct drm_device *dev,
 				   struct dma_buf_attachment *attach,
 				   struct sg_table *sgt);
 
+void panfrost_gem_shrinker_init(struct drm_device *dev);
+void panfrost_gem_shrinker_cleanup(struct drm_device *dev);
+
 #endif /* __PANFROST_GEM_H__ */
diff --git a/drivers/gpu/drm/panfrost/panfrost_gem_shrinker.c b/drivers/gpu/drm/panfrost/panfrost_gem_shrinker.c
new file mode 100644
index 000000000000..d191632b6197
--- /dev/null
+++ b/drivers/gpu/drm/panfrost/panfrost_gem_shrinker.c
@@ -0,0 +1,107 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (C) 2019 Arm Ltd.
+ *
+ * Based on msm_gem_freedreno.c:
+ * Copyright (C) 2016 Red Hat
+ * Author: Rob Clark <robdclark@gmail.com>
+ */
+
+#include <linux/list.h>
+
+#include <drm/drm_device.h>
+#include <drm/drm_gem_shmem_helper.h>
+
+#include "panfrost_device.h"
+#include "panfrost_gem.h"
+#include "panfrost_mmu.h"
+
+static unsigned long
+panfrost_gem_shrinker_count(struct shrinker *shrinker, struct shrink_control *sc)
+{
+	struct panfrost_device *pfdev =
+		container_of(shrinker, struct panfrost_device, shrinker);
+	struct drm_gem_shmem_object *shmem;
+	unsigned long count = 0;
+
+	if (!mutex_trylock(&pfdev->shrinker_lock))
+		return 0;
+
+	list_for_each_entry(shmem, &pfdev->shrinker_list, madv_list) {
+		if (drm_gem_shmem_is_purgeable(shmem))
+			count += shmem->base.size >> PAGE_SHIFT;
+	}
+
+	mutex_unlock(&pfdev->shrinker_lock);
+
+	return count;
+}
+
+static void panfrost_gem_purge(struct drm_gem_object *obj)
+{
+	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
+	mutex_lock(&shmem->pages_lock);
+
+	panfrost_mmu_unmap(to_panfrost_bo(obj));
+	drm_gem_shmem_purge_locked(obj);
+
+	mutex_unlock(&shmem->pages_lock);
+}
+
+static unsigned long
+panfrost_gem_shrinker_scan(struct shrinker *shrinker, struct shrink_control *sc)
+{
+	struct panfrost_device *pfdev =
+		container_of(shrinker, struct panfrost_device, shrinker);
+	struct drm_gem_shmem_object *shmem, *tmp;
+	unsigned long freed = 0;
+
+	if (!mutex_trylock(&pfdev->shrinker_lock))
+		return SHRINK_STOP;
+
+	list_for_each_entry_safe(shmem, tmp, &pfdev->shrinker_list, madv_list) {
+		if (freed >= sc->nr_to_scan)
+			break;
+		if (drm_gem_shmem_is_purgeable(shmem)) {
+			panfrost_gem_purge(&shmem->base);
+			freed += shmem->base.size >> PAGE_SHIFT;
+			list_del_init(&shmem->madv_list);
+		}
+	}
+
+	mutex_unlock(&pfdev->shrinker_lock);
+
+	if (freed > 0)
+		pr_info_ratelimited("Purging %lu bytes\n", freed << PAGE_SHIFT);
+
+	return freed;
+}
+
+/**
+ * panfrost_gem_shrinker_init - Initialize panfrost shrinker
+ * @dev: DRM device
+ *
+ * This function registers and sets up the panfrost shrinker.
+ */
+void panfrost_gem_shrinker_init(struct drm_device *dev)
+{
+	struct panfrost_device *pfdev = dev->dev_private;
+	pfdev->shrinker.count_objects = panfrost_gem_shrinker_count;
+	pfdev->shrinker.scan_objects = panfrost_gem_shrinker_scan;
+	pfdev->shrinker.seeks = DEFAULT_SEEKS;
+	WARN_ON(register_shrinker(&pfdev->shrinker));
+}
+
+/**
+ * panfrost_gem_shrinker_cleanup - Clean up panfrost shrinker
+ * @dev: DRM device
+ *
+ * This function unregisters the panfrost shrinker.
+ */
+void panfrost_gem_shrinker_cleanup(struct drm_device *dev)
+{
+	struct panfrost_device *pfdev = dev->dev_private;
+
+	if (pfdev->shrinker.nr_deferred) {
+		unregister_shrinker(&pfdev->shrinker);
+	}
+}
diff --git a/include/uapi/drm/panfrost_drm.h b/include/uapi/drm/panfrost_drm.h
index b5d370638846..1b3ff910ebe7 100644
--- a/include/uapi/drm/panfrost_drm.h
+++ b/include/uapi/drm/panfrost_drm.h
@@ -20,6 +20,7 @@ extern "C" {
 #define DRM_PANFROST_GET_BO_OFFSET		0x05
 #define DRM_PANFROST_PERFCNT_ENABLE		0x06
 #define DRM_PANFROST_PERFCNT_DUMP		0x07
+#define DRM_PANFROST_MADVISE			0x08
 
 #define DRM_IOCTL_PANFROST_SUBMIT		DRM_IOW(DRM_COMMAND_BASE + DRM_PANFROST_SUBMIT, struct drm_panfrost_submit)
 #define DRM_IOCTL_PANFROST_WAIT_BO		DRM_IOW(DRM_COMMAND_BASE + DRM_PANFROST_WAIT_BO, struct drm_panfrost_wait_bo)
@@ -27,6 +28,7 @@ extern "C" {
 #define DRM_IOCTL_PANFROST_MMAP_BO		DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_MMAP_BO, struct drm_panfrost_mmap_bo)
 #define DRM_IOCTL_PANFROST_GET_PARAM		DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_GET_PARAM, struct drm_panfrost_get_param)
 #define DRM_IOCTL_PANFROST_GET_BO_OFFSET	DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_GET_BO_OFFSET, struct drm_panfrost_get_bo_offset)
+#define DRM_IOCTL_PANFROST_MADVISE		DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_MADVISE, struct drm_panfrost_madvise)
 
 /*
  * Unstable ioctl(s): only exposed when the unsafe unstable_ioctls module
@@ -159,6 +161,26 @@ struct drm_panfrost_perfcnt_dump {
 	__u64 buf_ptr;
 };
 
+/* madvise provides a way to tell the kernel in case a buffers contents
+ * can be discarded under memory pressure, which is useful for userspace
+ * bo cache where we want to optimistically hold on to buffer allocate
+ * and potential mmap, but allow the pages to be discarded under memory
+ * pressure.
+ *
+ * Typical usage would involve madvise(DONTNEED) when buffer enters BO
+ * cache, and madvise(WILLNEED) if trying to recycle buffer from BO cache.
+ * In the WILLNEED case, 'retained' indicates to userspace whether the
+ * backing pages still exist.
+ */
+#define PANFROST_MADV_WILLNEED 0	/* backing pages are needed, status returned in 'retained' */
+#define PANFROST_MADV_DONTNEED 1	/* backing pages not needed */
+
+struct drm_panfrost_madvise {
+	__u32 handle;         /* in, GEM handle */
+	__u32 madv;           /* in, PANFROST_MADV_x */
+	__u32 retained;       /* out, whether backing store still exists */
+};
+
 #if defined(__cplusplus)
 }
 #endif
-- 
2.20.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 2/2] drm/panfrost: Add madvise and shrinker support

[Alyssa Rosenzweig]

[-- Attachment #1.1: Type: text/plain, Size: 1220 bytes --]

> +/* madvise provides a way to tell the kernel in case a buffers contents
> + * can be discarded under memory pressure, which is useful for userspace
> + * bo cache where we want to optimistically hold on to buffer allocate
> + * and potential mmap, but allow the pages to be discarded under memory
> + * pressure.
> + *
> + * Typical usage would involve madvise(DONTNEED) when buffer enters BO
> + * cache, and madvise(WILLNEED) if trying to recycle buffer from BO cache.
> + * In the WILLNEED case, 'retained' indicates to userspace whether the
> + * backing pages still exist.
> + */
> +#define PANFROST_MADV_WILLNEED 0	/* backing pages are needed, status returned in 'retained' */
> +#define PANFROST_MADV_DONTNEED 1	/* backing pages not needed */
> +
> +struct drm_panfrost_madvise {
> +	__u32 handle;         /* in, GEM handle */
> +	__u32 madv;           /* in, PANFROST_MADV_x */
> +	__u32 retained;       /* out, whether backing store still exists */
> +};

Just to clarify about the `retained` flag: if userspace does a
madvise(WILLNEED) and we find out that retained=0, what's supposed to
happen?

Should userspace evict the BO from its local cache and allocate one
fresh? Or just remmap? Or something else?

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 2/2] drm/panfrost: Add madvise and shrinker support

[Rob Herring]

On Mon, Aug 5, 2019 at 9:52 AM Alyssa Rosenzweig
<alyssa.rosenzweig@collabora.com> wrote:
>
> > +/* madvise provides a way to tell the kernel in case a buffers contents
> > + * can be discarded under memory pressure, which is useful for userspace
> > + * bo cache where we want to optimistically hold on to buffer allocate
> > + * and potential mmap, but allow the pages to be discarded under memory
> > + * pressure.
> > + *
> > + * Typical usage would involve madvise(DONTNEED) when buffer enters BO
> > + * cache, and madvise(WILLNEED) if trying to recycle buffer from BO cache.
> > + * In the WILLNEED case, 'retained' indicates to userspace whether the
> > + * backing pages still exist.
> > + */
> > +#define PANFROST_MADV_WILLNEED 0     /* backing pages are needed, status returned in 'retained' */
> > +#define PANFROST_MADV_DONTNEED 1     /* backing pages not needed */
> > +
> > +struct drm_panfrost_madvise {
> > +     __u32 handle;         /* in, GEM handle */
> > +     __u32 madv;           /* in, PANFROST_MADV_x */
> > +     __u32 retained;       /* out, whether backing store still exists */
> > +};
>
> Just to clarify about the `retained` flag: if userspace does a
> madvise(WILLNEED) and we find out that retained=0, what's supposed to
> happen?
>
> Should userspace evict the BO from its local cache and allocate one
> fresh?

Yes. Just like msm/freedreno.

> Or just remmap? Or something else?
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 2/2] drm/panfrost: Add madvise and shrinker support

[Alyssa Rosenzweig]

[-- Attachment #1.1: Type: text/plain, Size: 333 bytes --]

> > Just to clarify about the `retained` flag: if userspace does a
> > madvise(WILLNEED) and we find out that retained=0, what's supposed to
> > happen?
> >
> > Should userspace evict the BO from its local cache and allocate one
> > fresh?
> 
> Yes. Just like msm/freedreno.

Got it. In that case, the series has my A-b :)

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 1/2] drm/shmem: Add madvise state and purge helpers

[Daniel Vetter]

On Mon, Aug 05, 2019 at 08:33:57AM -0600, Rob Herring wrote:
> Add support to the shmem GEM helpers for tracking madvise state and
> purging pages. This is based on the msm implementation.
> 
> The BO provides a list_head, but the list management is handled outside
> of the shmem helpers as there are different locking requirements.
> 
> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
> Cc: Maxime Ripard <maxime.ripard@bootlin.com>
> Cc: Sean Paul <sean@poorly.run>
> Cc: David Airlie <airlied@linux.ie>
> Cc: Daniel Vetter <daniel@ffwll.ch>
> Cc: Eric Anholt <eric@anholt.net>
> Signed-off-by: Rob Herring <robh@kernel.org>
> ---
>  drivers/gpu/drm/drm_gem_shmem_helper.c | 57 ++++++++++++++++++++++++++
>  include/drm/drm_gem_shmem_helper.h     | 15 +++++++
>  2 files changed, 72 insertions(+)
> 
> diff --git a/drivers/gpu/drm/drm_gem_shmem_helper.c b/drivers/gpu/drm/drm_gem_shmem_helper.c
> index 2f64667ac805..4b442576de1c 100644
> --- a/drivers/gpu/drm/drm_gem_shmem_helper.c
> +++ b/drivers/gpu/drm/drm_gem_shmem_helper.c
> @@ -75,6 +75,7 @@ struct drm_gem_shmem_object *drm_gem_shmem_create(struct drm_device *dev, size_t
>  	shmem = to_drm_gem_shmem_obj(obj);
>  	mutex_init(&shmem->pages_lock);
>  	mutex_init(&shmem->vmap_lock);

Maybe a bit late, but for reasons (interop with ttm, which will be more
important once we have dynamic dma-buf) it would be real nice to use the
reservation_obj lock for all this stuff. msm, being struct_mutex based,
isn't a great example here. The downside is that it will be a lot harder
to get  msm to use these then, but much better to not spread struct_mutex
inspired locking too far.

Other bit: Wire this all up in a shrinker while at it?
-Daniel

> +	INIT_LIST_HEAD(&shmem->madv_list);
>  
>  	/*
>  	 * Our buffers are kept pinned, so allocating them
> @@ -362,6 +363,62 @@ drm_gem_shmem_create_with_handle(struct drm_file *file_priv,
>  }
>  EXPORT_SYMBOL(drm_gem_shmem_create_with_handle);
>  
> +/* Update madvise status, returns true if not purged, else
> + * false or -errno.
> + */
> +int drm_gem_shmem_madvise(struct drm_gem_object *obj, int madv)
> +{
> +	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
> +
> +	mutex_lock(&shmem->pages_lock);
> +
> +	if (shmem->madv >= 0)
> +		shmem->madv = madv;
> +
> +	madv = shmem->madv;
> +
> +	mutex_unlock(&shmem->pages_lock);
> +
> +	return (madv >= 0);
> +}
> +EXPORT_SYMBOL(drm_gem_shmem_madvise);
> +
> +void drm_gem_shmem_purge_locked(struct drm_gem_object *obj)
> +{
> +	struct drm_device *dev = obj->dev;
> +	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
> +
> +	WARN_ON(!drm_gem_shmem_is_purgeable(shmem));
> +
> +	drm_gem_shmem_put_pages_locked(shmem);
> +
> +	shmem->madv = -1;
> +
> +	drm_vma_node_unmap(&obj->vma_node, dev->anon_inode->i_mapping);
> +	drm_gem_free_mmap_offset(obj);
> +
> +	/* Our goal here is to return as much of the memory as
> +	 * is possible back to the system as we are called from OOM.
> +	 * To do this we must instruct the shmfs to drop all of its
> +	 * backing pages, *now*.
> +	 */
> +	shmem_truncate_range(file_inode(obj->filp), 0, (loff_t)-1);
> +
> +	invalidate_mapping_pages(file_inode(obj->filp)->i_mapping,
> +			0, (loff_t)-1);
> +}
> +EXPORT_SYMBOL(drm_gem_shmem_purge_locked);
> +
> +void drm_gem_shmem_purge(struct drm_gem_object *obj)
> +{
> +	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
> +
> +	mutex_lock(&shmem->pages_lock);
> +	drm_gem_shmem_purge_locked(obj);
> +	mutex_unlock(&shmem->pages_lock);
> +}
> +EXPORT_SYMBOL(drm_gem_shmem_purge);
> +
>  /**
>   * drm_gem_shmem_dumb_create - Create a dumb shmem buffer object
>   * @file: DRM file structure to create the dumb buffer for
> diff --git a/include/drm/drm_gem_shmem_helper.h b/include/drm/drm_gem_shmem_helper.h
> index 038b6d313447..ce1600fdfc3e 100644
> --- a/include/drm/drm_gem_shmem_helper.h
> +++ b/include/drm/drm_gem_shmem_helper.h
> @@ -44,6 +44,9 @@ struct drm_gem_shmem_object {
>  	 */
>  	unsigned int pages_use_count;
>  
> +	int madv;
> +	struct list_head madv_list;
> +
>  	/**
>  	 * @pages_mark_dirty_on_put:
>  	 *
> @@ -121,6 +124,18 @@ void drm_gem_shmem_unpin(struct drm_gem_object *obj);
>  void *drm_gem_shmem_vmap(struct drm_gem_object *obj);
>  void drm_gem_shmem_vunmap(struct drm_gem_object *obj, void *vaddr);
>  
> +int drm_gem_shmem_madvise(struct drm_gem_object *obj, int madv);
> +
> +static inline bool drm_gem_shmem_is_purgeable(struct drm_gem_shmem_object *shmem)
> +{
> +	return (shmem->madv > 0) &&
> +		!shmem->vmap_use_count && shmem->sgt &&
> +		!shmem->base.dma_buf && !shmem->base.import_attach;
> +}
> +
> +void drm_gem_shmem_purge_locked(struct drm_gem_object *obj);
> +void drm_gem_shmem_purge(struct drm_gem_object *obj);
> +
>  struct drm_gem_shmem_object *
>  drm_gem_shmem_create_with_handle(struct drm_file *file_priv,
>  				 struct drm_device *dev, size_t size,
> -- 
> 2.20.1
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 1/2] drm/shmem: Add madvise state and purge helpers

[Rob Clark]

On Mon, Aug 5, 2019 at 9:35 AM Daniel Vetter <daniel@ffwll.ch> wrote:
>
> On Mon, Aug 05, 2019 at 08:33:57AM -0600, Rob Herring wrote:
> > Add support to the shmem GEM helpers for tracking madvise state and
> > purging pages. This is based on the msm implementation.
> >
> > The BO provides a list_head, but the list management is handled outside
> > of the shmem helpers as there are different locking requirements.
> >
> > Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
> > Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
> > Cc: Maxime Ripard <maxime.ripard@bootlin.com>
> > Cc: Sean Paul <sean@poorly.run>
> > Cc: David Airlie <airlied@linux.ie>
> > Cc: Daniel Vetter <daniel@ffwll.ch>
> > Cc: Eric Anholt <eric@anholt.net>
> > Signed-off-by: Rob Herring <robh@kernel.org>
> > ---
> >  drivers/gpu/drm/drm_gem_shmem_helper.c | 57 ++++++++++++++++++++++++++
> >  include/drm/drm_gem_shmem_helper.h     | 15 +++++++
> >  2 files changed, 72 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/drm_gem_shmem_helper.c b/drivers/gpu/drm/drm_gem_shmem_helper.c
> > index 2f64667ac805..4b442576de1c 100644
> > --- a/drivers/gpu/drm/drm_gem_shmem_helper.c
> > +++ b/drivers/gpu/drm/drm_gem_shmem_helper.c
> > @@ -75,6 +75,7 @@ struct drm_gem_shmem_object *drm_gem_shmem_create(struct drm_device *dev, size_t
> >       shmem = to_drm_gem_shmem_obj(obj);
> >       mutex_init(&shmem->pages_lock);
> >       mutex_init(&shmem->vmap_lock);
>
> Maybe a bit late, but for reasons (interop with ttm, which will be more
> important once we have dynamic dma-buf) it would be real nice to use the
> reservation_obj lock for all this stuff. msm, being struct_mutex based,
> isn't a great example here. The downside is that it will be a lot harder
> to get  msm to use these then, but much better to not spread struct_mutex
> inspired locking too far.
>

but somewhere you need to protect access to list of bo's that are
available to shrink.. iirc that is the only thing we should need
struct_mutex for now in shrinker path in drm/msm.  The per-bo state is
protected by a per-bo lock.

BR,
-R
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 1/2] drm/shmem: Add madvise state and purge helpers

[Rob Herring]

On Mon, Aug 5, 2019 at 11:47 AM Rob Clark <robdclark@gmail.com> wrote:
>
> On Mon, Aug 5, 2019 at 9:35 AM Daniel Vetter <daniel@ffwll.ch> wrote:
> >
> > On Mon, Aug 05, 2019 at 08:33:57AM -0600, Rob Herring wrote:
> > > Add support to the shmem GEM helpers for tracking madvise state and
> > > purging pages. This is based on the msm implementation.
> > >
> > > The BO provides a list_head, but the list management is handled outside
> > > of the shmem helpers as there are different locking requirements.
> > >
> > > Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
> > > Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
> > > Cc: Maxime Ripard <maxime.ripard@bootlin.com>
> > > Cc: Sean Paul <sean@poorly.run>
> > > Cc: David Airlie <airlied@linux.ie>
> > > Cc: Daniel Vetter <daniel@ffwll.ch>
> > > Cc: Eric Anholt <eric@anholt.net>
> > > Signed-off-by: Rob Herring <robh@kernel.org>
> > > ---
> > >  drivers/gpu/drm/drm_gem_shmem_helper.c | 57 ++++++++++++++++++++++++++
> > >  include/drm/drm_gem_shmem_helper.h     | 15 +++++++
> > >  2 files changed, 72 insertions(+)
> > >
> > > diff --git a/drivers/gpu/drm/drm_gem_shmem_helper.c b/drivers/gpu/drm/drm_gem_shmem_helper.c
> > > index 2f64667ac805..4b442576de1c 100644
> > > --- a/drivers/gpu/drm/drm_gem_shmem_helper.c
> > > +++ b/drivers/gpu/drm/drm_gem_shmem_helper.c
> > > @@ -75,6 +75,7 @@ struct drm_gem_shmem_object *drm_gem_shmem_create(struct drm_device *dev, size_t
> > >       shmem = to_drm_gem_shmem_obj(obj);
> > >       mutex_init(&shmem->pages_lock);
> > >       mutex_init(&shmem->vmap_lock);
> >
> > Maybe a bit late, but for reasons (interop with ttm, which will be more
> > important once we have dynamic dma-buf) it would be real nice to use the
> > reservation_obj lock for all this stuff. msm, being struct_mutex based,
> > isn't a great example here. The downside is that it will be a lot harder
> > to get  msm to use these then, but much better to not spread struct_mutex
> > inspired locking too far.
> >
>
> but somewhere you need to protect access to list of bo's that are
> available to shrink.. iirc that is the only thing we should need
> struct_mutex for now in shrinker path in drm/msm.  The per-bo state is
> protected by a per-bo lock.

Right. And for panfrost, I have a lock just for the shrinker list and
don't use struct_mutex. Hopefully I've understood the locking
requirements sufficiently.

Rob
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 1/2] drm/shmem: Add madvise state and purge helpers

[Sean Paul]

On Mon, Aug 05, 2019 at 08:33:57AM -0600, Rob Herring wrote:
> Add support to the shmem GEM helpers for tracking madvise state and
> purging pages. This is based on the msm implementation.
> 
> The BO provides a list_head, but the list management is handled outside
> of the shmem helpers as there are different locking requirements.
> 
> Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
> Cc: Maxime Ripard <maxime.ripard@bootlin.com>
> Cc: Sean Paul <sean@poorly.run>
> Cc: David Airlie <airlied@linux.ie>
> Cc: Daniel Vetter <daniel@ffwll.ch>
> Cc: Eric Anholt <eric@anholt.net>
> Signed-off-by: Rob Herring <robh@kernel.org>
> ---
>  drivers/gpu/drm/drm_gem_shmem_helper.c | 57 ++++++++++++++++++++++++++
>  include/drm/drm_gem_shmem_helper.h     | 15 +++++++
>  2 files changed, 72 insertions(+)
> 
> diff --git a/drivers/gpu/drm/drm_gem_shmem_helper.c b/drivers/gpu/drm/drm_gem_shmem_helper.c
> index 2f64667ac805..4b442576de1c 100644
> --- a/drivers/gpu/drm/drm_gem_shmem_helper.c
> +++ b/drivers/gpu/drm/drm_gem_shmem_helper.c
> @@ -75,6 +75,7 @@ struct drm_gem_shmem_object *drm_gem_shmem_create(struct drm_device *dev, size_t
>  	shmem = to_drm_gem_shmem_obj(obj);
>  	mutex_init(&shmem->pages_lock);
>  	mutex_init(&shmem->vmap_lock);
> +	INIT_LIST_HEAD(&shmem->madv_list);
>  
>  	/*
>  	 * Our buffers are kept pinned, so allocating them
> @@ -362,6 +363,62 @@ drm_gem_shmem_create_with_handle(struct drm_file *file_priv,
>  }
>  EXPORT_SYMBOL(drm_gem_shmem_create_with_handle);
>  
> +/* Update madvise status, returns true if not purged, else
> + * false or -errno.
> + */
> +int drm_gem_shmem_madvise(struct drm_gem_object *obj, int madv)
> +{
> +	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
> +
> +	mutex_lock(&shmem->pages_lock);
> +
> +	if (shmem->madv >= 0)
> +		shmem->madv = madv;
> +
> +	madv = shmem->madv;
> +
> +	mutex_unlock(&shmem->pages_lock);
> +
> +	return (madv >= 0);
> +}
> +EXPORT_SYMBOL(drm_gem_shmem_madvise);
> +
> +void drm_gem_shmem_purge_locked(struct drm_gem_object *obj)
> +{
> +	struct drm_device *dev = obj->dev;
> +	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
> +
> +	WARN_ON(!drm_gem_shmem_is_purgeable(shmem));
> +
> +	drm_gem_shmem_put_pages_locked(shmem);
> +
> +	shmem->madv = -1;
> +
> +	drm_vma_node_unmap(&obj->vma_node, dev->anon_inode->i_mapping);
> +	drm_gem_free_mmap_offset(obj);
> +
> +	/* Our goal here is to return as much of the memory as
> +	 * is possible back to the system as we are called from OOM.
> +	 * To do this we must instruct the shmfs to drop all of its
> +	 * backing pages, *now*.
> +	 */
> +	shmem_truncate_range(file_inode(obj->filp), 0, (loff_t)-1);
> +
> +	invalidate_mapping_pages(file_inode(obj->filp)->i_mapping,
> +			0, (loff_t)-1);
> +}
> +EXPORT_SYMBOL(drm_gem_shmem_purge_locked);
> +
> +void drm_gem_shmem_purge(struct drm_gem_object *obj)
> +{
> +	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
> +
> +	mutex_lock(&shmem->pages_lock);
> +	drm_gem_shmem_purge_locked(obj);
> +	mutex_unlock(&shmem->pages_lock);
> +}
> +EXPORT_SYMBOL(drm_gem_shmem_purge);
> +
>  /**
>   * drm_gem_shmem_dumb_create - Create a dumb shmem buffer object
>   * @file: DRM file structure to create the dumb buffer for
> diff --git a/include/drm/drm_gem_shmem_helper.h b/include/drm/drm_gem_shmem_helper.h
> index 038b6d313447..ce1600fdfc3e 100644
> --- a/include/drm/drm_gem_shmem_helper.h
> +++ b/include/drm/drm_gem_shmem_helper.h
> @@ -44,6 +44,9 @@ struct drm_gem_shmem_object {
>  	 */
>  	unsigned int pages_use_count;
>  
> +	int madv;
> +	struct list_head madv_list;

Hey Rob,
Not sure why I didn't notice this earlier, but:

../include/drm/drm_gem_shmem_helper.h:87: warning: Function parameter or member 'madv' not described in 'drm_gem_shmem_object'
../include/drm/drm_gem_shmem_helper.h:87: warning: Function parameter or member 'madv_list' not described in 'drm_gem_shmem_object'

Could you please add some kerneldoc for these?

Sean

> +
>  	/**
>  	 * @pages_mark_dirty_on_put:
>  	 *
> @@ -121,6 +124,18 @@ void drm_gem_shmem_unpin(struct drm_gem_object *obj);
>  void *drm_gem_shmem_vmap(struct drm_gem_object *obj);
>  void drm_gem_shmem_vunmap(struct drm_gem_object *obj, void *vaddr);
>  
> +int drm_gem_shmem_madvise(struct drm_gem_object *obj, int madv);
> +
> +static inline bool drm_gem_shmem_is_purgeable(struct drm_gem_shmem_object *shmem)
> +{
> +	return (shmem->madv > 0) &&
> +		!shmem->vmap_use_count && shmem->sgt &&
> +		!shmem->base.dma_buf && !shmem->base.import_attach;
> +}
> +
> +void drm_gem_shmem_purge_locked(struct drm_gem_object *obj);
> +void drm_gem_shmem_purge(struct drm_gem_object *obj);
> +
>  struct drm_gem_shmem_object *
>  drm_gem_shmem_create_with_handle(struct drm_file *file_priv,
>  				 struct drm_device *dev, size_t size,
> -- 
> 2.20.1
> 

-- 
Sean Paul, Software Engineer, Google / Chromium OS
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel