All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] drm/virtio: use kvmalloc for large allocations
@ 2020-11-05  1:47 ` Sergey Senozhatsky
  0 siblings, 0 replies; 12+ messages in thread
From: Sergey Senozhatsky @ 2020-11-05  1:47 UTC (permalink / raw)
  To: Gerd Hoffmann, David Airlie, Daniel Vetter
  Cc: Suleiman Souhlal, Sergey Senozhatsky, linux-kernel, dri-devel,
	virtualization

We observed that some of virtio_gpu_object_shmem_init() allocations
can be rather costly - order 6 - which can be difficult to fulfill
under memory pressure conditions. Switch to kvmalloc_array() in
virtio_gpu_object_shmem_init() and let the kernel vmalloc the entries
array.

Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
---
 drivers/gpu/drm/virtio/virtgpu_object.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/virtio/virtgpu_object.c b/drivers/gpu/drm/virtio/virtgpu_object.c
index 2d3aa7baffe4..d9ad27e00905 100644
--- a/drivers/gpu/drm/virtio/virtgpu_object.c
+++ b/drivers/gpu/drm/virtio/virtgpu_object.c
@@ -184,8 +184,9 @@ static int virtio_gpu_object_shmem_init(struct virtio_gpu_device *vgdev,
 		*nents = shmem->pages->orig_nents;
 	}
 
-	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
-			      GFP_KERNEL);
+	*ents = kvmalloc_array(*nents,
+			       sizeof(struct virtio_gpu_mem_entry),
+			       GFP_KERNEL);
 	if (!(*ents)) {
 		DRM_ERROR("failed to allocate ent list\n");
 		return -ENOMEM;
-- 
2.29.1.341.ge80a0c044ae-goog

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

^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH] drm/virtio: use kvmalloc for large allocations
@ 2020-11-05  1:47 ` Sergey Senozhatsky
  0 siblings, 0 replies; 12+ messages in thread
From: Sergey Senozhatsky @ 2020-11-05  1:47 UTC (permalink / raw)
  To: Gerd Hoffmann, David Airlie, Daniel Vetter
  Cc: Suleiman Souhlal, dri-devel, virtualization, linux-kernel,
	Sergey Senozhatsky

We observed that some of virtio_gpu_object_shmem_init() allocations
can be rather costly - order 6 - which can be difficult to fulfill
under memory pressure conditions. Switch to kvmalloc_array() in
virtio_gpu_object_shmem_init() and let the kernel vmalloc the entries
array.

Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
---
 drivers/gpu/drm/virtio/virtgpu_object.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/virtio/virtgpu_object.c b/drivers/gpu/drm/virtio/virtgpu_object.c
index 2d3aa7baffe4..d9ad27e00905 100644
--- a/drivers/gpu/drm/virtio/virtgpu_object.c
+++ b/drivers/gpu/drm/virtio/virtgpu_object.c
@@ -184,8 +184,9 @@ static int virtio_gpu_object_shmem_init(struct virtio_gpu_device *vgdev,
 		*nents = shmem->pages->orig_nents;
 	}
 
-	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
-			      GFP_KERNEL);
+	*ents = kvmalloc_array(*nents,
+			       sizeof(struct virtio_gpu_mem_entry),
+			       GFP_KERNEL);
 	if (!(*ents)) {
 		DRM_ERROR("failed to allocate ent list\n");
 		return -ENOMEM;
-- 
2.29.1.341.ge80a0c044ae-goog


^ permalink raw reply related	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
  2020-11-05  1:47 ` Sergey Senozhatsky
  (?)
@ 2020-11-05  6:52   ` Gerd Hoffmann
  -1 siblings, 0 replies; 12+ messages in thread
From: Gerd Hoffmann @ 2020-11-05  6:52 UTC (permalink / raw)
  To: Sergey Senozhatsky
  Cc: David Airlie, linux-kernel, dri-devel, virtualization,
	Daniel Vetter, Suleiman Souhlal

  Hi,

> -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> -			      GFP_KERNEL);
> +	*ents = kvmalloc_array(*nents,
> +			       sizeof(struct virtio_gpu_mem_entry),
> +			       GFP_KERNEL);

Shouldn't that be balanced with a kvfree() elsewhere?

take care,
  Gerd

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
@ 2020-11-05  6:52   ` Gerd Hoffmann
  0 siblings, 0 replies; 12+ messages in thread
From: Gerd Hoffmann @ 2020-11-05  6:52 UTC (permalink / raw)
  To: Sergey Senozhatsky
  Cc: David Airlie, linux-kernel, dri-devel, virtualization,
	Suleiman Souhlal

  Hi,

> -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> -			      GFP_KERNEL);
> +	*ents = kvmalloc_array(*nents,
> +			       sizeof(struct virtio_gpu_mem_entry),
> +			       GFP_KERNEL);

Shouldn't that be balanced with a kvfree() elsewhere?

take care,
  Gerd

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

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
@ 2020-11-05  6:52   ` Gerd Hoffmann
  0 siblings, 0 replies; 12+ messages in thread
From: Gerd Hoffmann @ 2020-11-05  6:52 UTC (permalink / raw)
  To: Sergey Senozhatsky
  Cc: David Airlie, Daniel Vetter, Suleiman Souhlal, dri-devel,
	virtualization, linux-kernel

  Hi,

> -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> -			      GFP_KERNEL);
> +	*ents = kvmalloc_array(*nents,
> +			       sizeof(struct virtio_gpu_mem_entry),
> +			       GFP_KERNEL);

Shouldn't that be balanced with a kvfree() elsewhere?

take care,
  Gerd


^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
  2020-11-05  6:52   ` Gerd Hoffmann
  (?)
@ 2020-11-05  7:00     ` Sergey Senozhatsky
  -1 siblings, 0 replies; 12+ messages in thread
From: Sergey Senozhatsky @ 2020-11-05  7:00 UTC (permalink / raw)
  To: Gerd Hoffmann
  Cc: David Airlie, linux-kernel, dri-devel, virtualization,
	Sergey Senozhatsky, Daniel Vetter, Suleiman Souhlal

Hi,

On (20/11/05 07:52), Gerd Hoffmann wrote:
> > -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> > -			      GFP_KERNEL);
> > +	*ents = kvmalloc_array(*nents,
> > +			       sizeof(struct virtio_gpu_mem_entry),
> > +			       GFP_KERNEL);
> 
> Shouldn't that be balanced with a kvfree() elsewhere?

I think it already is. ents pointer is assigned to vbuf->data_buf,
and free_vbuf() already uses kvfree(vbuf->data_buf) to free it.

	-ss
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
@ 2020-11-05  7:00     ` Sergey Senozhatsky
  0 siblings, 0 replies; 12+ messages in thread
From: Sergey Senozhatsky @ 2020-11-05  7:00 UTC (permalink / raw)
  To: Gerd Hoffmann
  Cc: David Airlie, linux-kernel, dri-devel, virtualization,
	Sergey Senozhatsky, Suleiman Souhlal

Hi,

On (20/11/05 07:52), Gerd Hoffmann wrote:
> > -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> > -			      GFP_KERNEL);
> > +	*ents = kvmalloc_array(*nents,
> > +			       sizeof(struct virtio_gpu_mem_entry),
> > +			       GFP_KERNEL);
> 
> Shouldn't that be balanced with a kvfree() elsewhere?

I think it already is. ents pointer is assigned to vbuf->data_buf,
and free_vbuf() already uses kvfree(vbuf->data_buf) to free it.

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

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
@ 2020-11-05  7:00     ` Sergey Senozhatsky
  0 siblings, 0 replies; 12+ messages in thread
From: Sergey Senozhatsky @ 2020-11-05  7:00 UTC (permalink / raw)
  To: Gerd Hoffmann
  Cc: Sergey Senozhatsky, David Airlie, Daniel Vetter, Suleiman Souhlal,
	dri-devel, virtualization, linux-kernel

Hi,

On (20/11/05 07:52), Gerd Hoffmann wrote:
> > -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> > -			      GFP_KERNEL);
> > +	*ents = kvmalloc_array(*nents,
> > +			       sizeof(struct virtio_gpu_mem_entry),
> > +			       GFP_KERNEL);
> 
> Shouldn't that be balanced with a kvfree() elsewhere?

I think it already is. ents pointer is assigned to vbuf->data_buf,
and free_vbuf() already uses kvfree(vbuf->data_buf) to free it.

	-ss

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
  2020-11-05  7:00     ` Sergey Senozhatsky
  (?)
@ 2020-11-05 11:34       ` Gerd Hoffmann
  -1 siblings, 0 replies; 12+ messages in thread
From: Gerd Hoffmann @ 2020-11-05 11:34 UTC (permalink / raw)
  To: Sergey Senozhatsky
  Cc: David Airlie, linux-kernel, dri-devel, virtualization,
	Sergey Senozhatsky, Daniel Vetter, Suleiman Souhlal

On Thu, Nov 05, 2020 at 04:00:54PM +0900, Sergey Senozhatsky wrote:
> Hi,
> 
> On (20/11/05 07:52), Gerd Hoffmann wrote:
> > > -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> > > -			      GFP_KERNEL);
> > > +	*ents = kvmalloc_array(*nents,
> > > +			       sizeof(struct virtio_gpu_mem_entry),
> > > +			       GFP_KERNEL);
> > 
> > Shouldn't that be balanced with a kvfree() elsewhere?
> 
> I think it already is. ents pointer is assigned to vbuf->data_buf,
> and free_vbuf() already uses kvfree(vbuf->data_buf) to free it.

Ah, right, we needed that before elsewhere.
Ok then, pushed to drm-misc-next.

thanks,
  Gerd

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
@ 2020-11-05 11:34       ` Gerd Hoffmann
  0 siblings, 0 replies; 12+ messages in thread
From: Gerd Hoffmann @ 2020-11-05 11:34 UTC (permalink / raw)
  To: Sergey Senozhatsky
  Cc: David Airlie, linux-kernel, dri-devel, virtualization,
	Sergey Senozhatsky, Suleiman Souhlal

On Thu, Nov 05, 2020 at 04:00:54PM +0900, Sergey Senozhatsky wrote:
> Hi,
> 
> On (20/11/05 07:52), Gerd Hoffmann wrote:
> > > -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> > > -			      GFP_KERNEL);
> > > +	*ents = kvmalloc_array(*nents,
> > > +			       sizeof(struct virtio_gpu_mem_entry),
> > > +			       GFP_KERNEL);
> > 
> > Shouldn't that be balanced with a kvfree() elsewhere?
> 
> I think it already is. ents pointer is assigned to vbuf->data_buf,
> and free_vbuf() already uses kvfree(vbuf->data_buf) to free it.

Ah, right, we needed that before elsewhere.
Ok then, pushed to drm-misc-next.

thanks,
  Gerd

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

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH] drm/virtio: use kvmalloc for large allocations
@ 2020-11-05 11:34       ` Gerd Hoffmann
  0 siblings, 0 replies; 12+ messages in thread
From: Gerd Hoffmann @ 2020-11-05 11:34 UTC (permalink / raw)
  To: Sergey Senozhatsky
  Cc: Sergey Senozhatsky, David Airlie, Daniel Vetter, Suleiman Souhlal,
	dri-devel, virtualization, linux-kernel

On Thu, Nov 05, 2020 at 04:00:54PM +0900, Sergey Senozhatsky wrote:
> Hi,
> 
> On (20/11/05 07:52), Gerd Hoffmann wrote:
> > > -	*ents = kmalloc_array(*nents, sizeof(struct virtio_gpu_mem_entry),
> > > -			      GFP_KERNEL);
> > > +	*ents = kvmalloc_array(*nents,
> > > +			       sizeof(struct virtio_gpu_mem_entry),
> > > +			       GFP_KERNEL);
> > 
> > Shouldn't that be balanced with a kvfree() elsewhere?
> 
> I think it already is. ents pointer is assigned to vbuf->data_buf,
> and free_vbuf() already uses kvfree(vbuf->data_buf) to free it.

Ah, right, we needed that before elsewhere.
Ok then, pushed to drm-misc-next.

thanks,
  Gerd


^ permalink raw reply	[flat|nested] 12+ messages in thread

* [PATCH] drm/virtio: use kvmalloc for large allocations
  2021-03-02 22:38 [PATCH 1/1] " Doug Horn
@ 2021-03-04 15:05 ` Sergey Senozhatsky
  0 siblings, 0 replies; 12+ messages in thread
From: Sergey Senozhatsky @ 2021-03-04 15:05 UTC (permalink / raw)
  Cc: adelva, kaiyili, Sergey Senozhatsky, stable, Gerd Hoffmann,
	Doug Horn

From: Sergey Senozhatsky <senozhatsky@chromium.org>

commit ea86f3defd55f141a44146e66cbf8ffb683d60da upstream.

We observed that some of virtio_gpu_object_shmem_init() allocations
can be rather costly - order 6 - which can be difficult to fulfill
under memory pressure conditions. Switch to kvmalloc_array() in
virtio_gpu_object_shmem_init() and let the kernel vmalloc the entries
array.

Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20201105014744.1662226-1-senozhatsky@chromium.org
Cc: stable@vger.kernel.org [5.4]
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Doug Horn <doughorn@google.com>
---
 drivers/gpu/drm/virtio/virtgpu_vq.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c
index 92022a83bbd5..bb46e7a0f1b5 100644
--- a/drivers/gpu/drm/virtio/virtgpu_vq.c
+++ b/drivers/gpu/drm/virtio/virtgpu_vq.c
@@ -992,8 +992,9 @@ int virtio_gpu_object_attach(struct virtio_gpu_device *vgdev,
 	}
 
 	/* gets freed when the ring has consumed it */
-	ents = kmalloc_array(nents, sizeof(struct virtio_gpu_mem_entry),
-			     GFP_KERNEL);
+	ents = kvmalloc_array(nents,
+			      sizeof(struct virtio_gpu_mem_entry),
+			      GFP_KERNEL);
 	if (!ents) {
 		DRM_ERROR("failed to allocate ent list\n");
 		return -ENOMEM;
-- 
2.30.1.766.gb4fecdf3b7-goog


^ permalink raw reply related	[flat|nested] 12+ messages in thread

end of thread, other threads:[~2021-03-04 15:08 UTC | newest]

Thread overview: 12+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2020-11-05  1:47 [PATCH] drm/virtio: use kvmalloc for large allocations Sergey Senozhatsky
2020-11-05  1:47 ` Sergey Senozhatsky
2020-11-05  6:52 ` Gerd Hoffmann
2020-11-05  6:52   ` Gerd Hoffmann
2020-11-05  6:52   ` Gerd Hoffmann
2020-11-05  7:00   ` Sergey Senozhatsky
2020-11-05  7:00     ` Sergey Senozhatsky
2020-11-05  7:00     ` Sergey Senozhatsky
2020-11-05 11:34     ` Gerd Hoffmann
2020-11-05 11:34       ` Gerd Hoffmann
2020-11-05 11:34       ` Gerd Hoffmann
  -- strict thread matches above, loose matches on Subject: below --
2021-03-02 22:38 [PATCH 1/1] " Doug Horn
2021-03-04 15:05 ` [PATCH] " Sergey Senozhatsky

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.