On architectures where the is no coherent caching such as ARM use the
dma_alloc_noncontiguos API and handle manually the cache flushing using
dma_sync_single().
With this patch on the affected architectures we can measure up to 20x
performance improvement in uvc_video_copy_data_work().
Signed-off-by: Ricardo Ribalda <[email protected]>
---
drivers/media/usb/uvc/uvc_video.c | 74 ++++++++++++++++++++++++++-----
drivers/media/usb/uvc/uvcvideo.h | 1 +
2 files changed, 63 insertions(+), 12 deletions(-)
diff --git a/drivers/media/usb/uvc/uvc_video.c b/drivers/media/usb/uvc/uvc_video.c
index a6a441d92b94..9e90b261428a 100644
--- a/drivers/media/usb/uvc/uvc_video.c
+++ b/drivers/media/usb/uvc/uvc_video.c
@@ -1490,6 +1490,11 @@ static void uvc_video_encode_bulk(struct uvc_urb *uvc_urb,
urb->transfer_buffer_length = stream->urb_size - len;
}
+static inline struct device *stream_to_dmadev(struct uvc_streaming *stream)
+{
+ return stream->dev->udev->bus->controller->parent;
+}
+
static void uvc_video_complete(struct urb *urb)
{
struct uvc_urb *uvc_urb = urb->context;
@@ -1539,6 +1544,11 @@ static void uvc_video_complete(struct urb *urb)
* Process the URB headers, and optionally queue expensive memcpy tasks
* to be deferred to a work queue.
*/
+ if (uvc_urb->pages)
+ dma_sync_single_for_cpu(stream_to_dmadev(stream),
+ urb->transfer_dma,
+ urb->transfer_buffer_length,
+ DMA_FROM_DEVICE);
stream->decode(uvc_urb, buf, buf_meta);
/* If no async work is needed, resubmit the URB immediately. */
@@ -1566,8 +1576,15 @@ static void uvc_free_urb_buffers(struct uvc_streaming *stream)
continue;
#ifndef CONFIG_DMA_NONCOHERENT
- usb_free_coherent(stream->dev->udev, stream->urb_size,
- uvc_urb->buffer, uvc_urb->dma);
+ if (uvc_urb->pages) {
+ vunmap(uvc_urb->buffer);
+ dma_free_noncontiguous(stream_to_dmadev(stream),
+ stream->urb_size,
+ uvc_urb->pages, uvc_urb->dma);
+ } else {
+ usb_free_coherent(stream->dev->udev, stream->urb_size,
+ uvc_urb->buffer, uvc_urb->dma);
+ }
#else
kfree(uvc_urb->buffer);
#endif
@@ -1577,6 +1594,47 @@ static void uvc_free_urb_buffers(struct uvc_streaming *stream)
stream->urb_size = 0;
}
+#ifndef CONFIG_DMA_NONCOHERENT
+static bool uvc_alloc_urb_buffer(struct uvc_streaming *stream,
+ struct uvc_urb *uvc_urb, gfp_t gfp_flags)
+{
+ struct device *dma_dev = dma_dev = stream_to_dmadev(stream);
+
+ if (!dma_can_alloc_noncontiguous(dma_dev)) {
+ uvc_urb->buffer = usb_alloc_coherent(stream->dev->udev,
+ stream->urb_size,
+ gfp_flags | __GFP_NOWARN,
+ &uvc_urb->dma);
+ return uvc_urb->buffer != NULL;
+ }
+
+ uvc_urb->pages = dma_alloc_noncontiguous(dma_dev, stream->urb_size,
+ &uvc_urb->dma,
+ gfp_flags | __GFP_NOWARN, 0);
+ if (!uvc_urb->pages)
+ return false;
+
+ uvc_urb->buffer = vmap(uvc_urb->pages,
+ PAGE_ALIGN(stream->urb_size) >> PAGE_SHIFT,
+ VM_DMA_COHERENT, PAGE_KERNEL);
+ if (!uvc_urb->buffer) {
+ dma_free_noncontiguous(dma_dev, stream->urb_size,
+ uvc_urb->pages, uvc_urb->dma);
+ return false;
+ }
+
+ return true;
+}
+#else
+static bool uvc_alloc_urb_buffer(struct uvc_streaming *stream,
+ struct uvc_urb *uvc_urb, gfp_t gfp_flags)
+{
+ uvc_urb->buffer = kmalloc(stream->urb_size, gfp_flags | __GFP_NOWARN);
+
+ return uvc_urb->buffer != NULL;
+}
+#endif
+
/*
* Allocate transfer buffers. This function can be called with buffers
* already allocated when resuming from suspend, in which case it will
@@ -1607,19 +1665,11 @@ static int uvc_alloc_urb_buffers(struct uvc_streaming *stream,
/* Retry allocations until one succeed. */
for (; npackets > 1; npackets /= 2) {
+ stream->urb_size = psize * npackets;
for (i = 0; i < UVC_URBS; ++i) {
struct uvc_urb *uvc_urb = &stream->uvc_urb[i];
- stream->urb_size = psize * npackets;
-#ifndef CONFIG_DMA_NONCOHERENT
- uvc_urb->buffer = usb_alloc_coherent(
- stream->dev->udev, stream->urb_size,
- gfp_flags | __GFP_NOWARN, &uvc_urb->dma);
-#else
- uvc_urb->buffer =
- kmalloc(stream->urb_size, gfp_flags | __GFP_NOWARN);
-#endif
- if (!uvc_urb->buffer) {
+ if (!uvc_alloc_urb_buffer(stream, uvc_urb, gfp_flags)) {
uvc_free_urb_buffers(stream);
break;
}
diff --git a/drivers/media/usb/uvc/uvcvideo.h b/drivers/media/usb/uvc/uvcvideo.h
index a3dfacf069c4..3e3ef1f1daa5 100644
--- a/drivers/media/usb/uvc/uvcvideo.h
+++ b/drivers/media/usb/uvc/uvcvideo.h
@@ -532,6 +532,7 @@ struct uvc_urb {
char *buffer;
dma_addr_t dma;
+ struct page **pages;
unsigned int async_operations;
struct uvc_copy_op copy_operations[UVC_MAX_PACKETS];
--
2.29.2.454.gaff20da3a2-goog
On 2020-11-24 15:38, Ricardo Ribalda wrote:
> On architectures where the is no coherent caching such as ARM use the
> dma_alloc_noncontiguos API and handle manually the cache flushing using
> dma_sync_single().
>
> With this patch on the affected architectures we can measure up to 20x
> performance improvement in uvc_video_copy_data_work().
>
> Signed-off-by: Ricardo Ribalda <[email protected]>
> ---
> drivers/media/usb/uvc/uvc_video.c | 74 ++++++++++++++++++++++++++-----
> drivers/media/usb/uvc/uvcvideo.h | 1 +
> 2 files changed, 63 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/media/usb/uvc/uvc_video.c b/drivers/media/usb/uvc/uvc_video.c
> index a6a441d92b94..9e90b261428a 100644
> --- a/drivers/media/usb/uvc/uvc_video.c
> +++ b/drivers/media/usb/uvc/uvc_video.c
> @@ -1490,6 +1490,11 @@ static void uvc_video_encode_bulk(struct uvc_urb *uvc_urb,
> urb->transfer_buffer_length = stream->urb_size - len;
> }
>
> +static inline struct device *stream_to_dmadev(struct uvc_streaming *stream)
> +{
> + return stream->dev->udev->bus->controller->parent;
> +}
> +
> static void uvc_video_complete(struct urb *urb)
> {
> struct uvc_urb *uvc_urb = urb->context;
> @@ -1539,6 +1544,11 @@ static void uvc_video_complete(struct urb *urb)
> * Process the URB headers, and optionally queue expensive memcpy tasks
> * to be deferred to a work queue.
> */
> + if (uvc_urb->pages)
> + dma_sync_single_for_cpu(stream_to_dmadev(stream),
> + urb->transfer_dma,
> + urb->transfer_buffer_length,
> + DMA_FROM_DEVICE);
This doesn't work. Even in iommu-dma, the streaming API still expects to
work on physically-contiguous memory that could have been passed to
dma_map_single() in the first place. As-is, this will invalidate
transfer_buffer_length bytes from the start of the *first* physical
page, and thus destroy random other data if lines from subsequent
unrelated pages are dirty in caches.
The only feasible way to do a DMA sync on disjoint pages in a single
call is with a scatterlist.
Robin.
> stream->decode(uvc_urb, buf, buf_meta);
>
> /* If no async work is needed, resubmit the URB immediately. */
> @@ -1566,8 +1576,15 @@ static void uvc_free_urb_buffers(struct uvc_streaming *stream)
> continue;
>
> #ifndef CONFIG_DMA_NONCOHERENT
> - usb_free_coherent(stream->dev->udev, stream->urb_size,
> - uvc_urb->buffer, uvc_urb->dma);
> + if (uvc_urb->pages) {
> + vunmap(uvc_urb->buffer);
> + dma_free_noncontiguous(stream_to_dmadev(stream),
> + stream->urb_size,
> + uvc_urb->pages, uvc_urb->dma);
> + } else {
> + usb_free_coherent(stream->dev->udev, stream->urb_size,
> + uvc_urb->buffer, uvc_urb->dma);
> + }
> #else
> kfree(uvc_urb->buffer);
> #endif
> @@ -1577,6 +1594,47 @@ static void uvc_free_urb_buffers(struct uvc_streaming *stream)
> stream->urb_size = 0;
> }
>
> +#ifndef CONFIG_DMA_NONCOHERENT
> +static bool uvc_alloc_urb_buffer(struct uvc_streaming *stream,
> + struct uvc_urb *uvc_urb, gfp_t gfp_flags)
> +{
> + struct device *dma_dev = dma_dev = stream_to_dmadev(stream);
> +
> + if (!dma_can_alloc_noncontiguous(dma_dev)) {
> + uvc_urb->buffer = usb_alloc_coherent(stream->dev->udev,
> + stream->urb_size,
> + gfp_flags | __GFP_NOWARN,
> + &uvc_urb->dma);
> + return uvc_urb->buffer != NULL;
> + }
> +
> + uvc_urb->pages = dma_alloc_noncontiguous(dma_dev, stream->urb_size,
> + &uvc_urb->dma,
> + gfp_flags | __GFP_NOWARN, 0);
> + if (!uvc_urb->pages)
> + return false;
> +
> + uvc_urb->buffer = vmap(uvc_urb->pages,
> + PAGE_ALIGN(stream->urb_size) >> PAGE_SHIFT,
> + VM_DMA_COHERENT, PAGE_KERNEL);
> + if (!uvc_urb->buffer) {
> + dma_free_noncontiguous(dma_dev, stream->urb_size,
> + uvc_urb->pages, uvc_urb->dma);
> + return false;
> + }
> +
> + return true;
> +}
> +#else
> +static bool uvc_alloc_urb_buffer(struct uvc_streaming *stream,
> + struct uvc_urb *uvc_urb, gfp_t gfp_flags)
> +{
> + uvc_urb->buffer = kmalloc(stream->urb_size, gfp_flags | __GFP_NOWARN);
> +
> + return uvc_urb->buffer != NULL;
> +}
> +#endif
> +
> /*
> * Allocate transfer buffers. This function can be called with buffers
> * already allocated when resuming from suspend, in which case it will
> @@ -1607,19 +1665,11 @@ static int uvc_alloc_urb_buffers(struct uvc_streaming *stream,
>
> /* Retry allocations until one succeed. */
> for (; npackets > 1; npackets /= 2) {
> + stream->urb_size = psize * npackets;
> for (i = 0; i < UVC_URBS; ++i) {
> struct uvc_urb *uvc_urb = &stream->uvc_urb[i];
>
> - stream->urb_size = psize * npackets;
> -#ifndef CONFIG_DMA_NONCOHERENT
> - uvc_urb->buffer = usb_alloc_coherent(
> - stream->dev->udev, stream->urb_size,
> - gfp_flags | __GFP_NOWARN, &uvc_urb->dma);
> -#else
> - uvc_urb->buffer =
> - kmalloc(stream->urb_size, gfp_flags | __GFP_NOWARN);
> -#endif
> - if (!uvc_urb->buffer) {
> + if (!uvc_alloc_urb_buffer(stream, uvc_urb, gfp_flags)) {
> uvc_free_urb_buffers(stream);
> break;
> }
> diff --git a/drivers/media/usb/uvc/uvcvideo.h b/drivers/media/usb/uvc/uvcvideo.h
> index a3dfacf069c4..3e3ef1f1daa5 100644
> --- a/drivers/media/usb/uvc/uvcvideo.h
> +++ b/drivers/media/usb/uvc/uvcvideo.h
> @@ -532,6 +532,7 @@ struct uvc_urb {
>
> char *buffer;
> dma_addr_t dma;
> + struct page **pages;
>
> unsigned int async_operations;
> struct uvc_copy_op copy_operations[UVC_MAX_PACKETS];
>
Hi Robin
On Tue, Nov 24, 2020 at 5:29 PM Robin Murphy <[email protected]> wrote:
>
> On 2020-11-24 15:38, Ricardo Ribalda wrote:
> > On architectures where the is no coherent caching such as ARM use the
> > dma_alloc_noncontiguos API and handle manually the cache flushing using
> > dma_sync_single().
> >
> > With this patch on the affected architectures we can measure up to 20x
> > performance improvement in uvc_video_copy_data_work().
> >
> > Signed-off-by: Ricardo Ribalda <[email protected]>
> > ---
> > drivers/media/usb/uvc/uvc_video.c | 74 ++++++++++++++++++++++++++-----
> > drivers/media/usb/uvc/uvcvideo.h | 1 +
> > 2 files changed, 63 insertions(+), 12 deletions(-)
> >
> > diff --git a/drivers/media/usb/uvc/uvc_video.c b/drivers/media/usb/uvc/uvc_video.c
> > index a6a441d92b94..9e90b261428a 100644
> > --- a/drivers/media/usb/uvc/uvc_video.c
> > +++ b/drivers/media/usb/uvc/uvc_video.c
> > @@ -1490,6 +1490,11 @@ static void uvc_video_encode_bulk(struct uvc_urb *uvc_urb,
> > urb->transfer_buffer_length = stream->urb_size - len;
> > }
> >
> > +static inline struct device *stream_to_dmadev(struct uvc_streaming *stream)
> > +{
> > + return stream->dev->udev->bus->controller->parent;
> > +}
> > +
> > static void uvc_video_complete(struct urb *urb)
> > {
> > struct uvc_urb *uvc_urb = urb->context;
> > @@ -1539,6 +1544,11 @@ static void uvc_video_complete(struct urb *urb)
> > * Process the URB headers, and optionally queue expensive memcpy tasks
> > * to be deferred to a work queue.
> > */
> > + if (uvc_urb->pages)
> > + dma_sync_single_for_cpu(stream_to_dmadev(stream),
> > + urb->transfer_dma,
> > + urb->transfer_buffer_length,
> > + DMA_FROM_DEVICE);
>
> This doesn't work. Even in iommu-dma, the streaming API still expects to
> work on physically-contiguous memory that could have been passed to
> dma_map_single() in the first place. As-is, this will invalidate
> transfer_buffer_length bytes from the start of the *first* physical
> page, and thus destroy random other data if lines from subsequent
> unrelated pages are dirty in caches.
>
> The only feasible way to do a DMA sync on disjoint pages in a single
> call is with a scatterlist.
Thanks for pointing this out. I guess I was lucky on my hardware and
the areas were always contiguous.
Will rework and send back to the list.
Thanks again.
>
> Robin.
>
> > stream->decode(uvc_urb, buf, buf_meta);
> >
> > /* If no async work is needed, resubmit the URB immediately. */
> > @@ -1566,8 +1576,15 @@ static void uvc_free_urb_buffers(struct uvc_streaming *stream)
> > continue;
> >
> > #ifndef CONFIG_DMA_NONCOHERENT
> > - usb_free_coherent(stream->dev->udev, stream->urb_size,
> > - uvc_urb->buffer, uvc_urb->dma);
> > + if (uvc_urb->pages) {
> > + vunmap(uvc_urb->buffer);
> > + dma_free_noncontiguous(stream_to_dmadev(stream),
> > + stream->urb_size,
> > + uvc_urb->pages, uvc_urb->dma);
> > + } else {
> > + usb_free_coherent(stream->dev->udev, stream->urb_size,
> > + uvc_urb->buffer, uvc_urb->dma);
> > + }
> > #else
> > kfree(uvc_urb->buffer);
> > #endif
> > @@ -1577,6 +1594,47 @@ static void uvc_free_urb_buffers(struct uvc_streaming *stream)
> > stream->urb_size = 0;
> > }
> >
> > +#ifndef CONFIG_DMA_NONCOHERENT
> > +static bool uvc_alloc_urb_buffer(struct uvc_streaming *stream,
> > + struct uvc_urb *uvc_urb, gfp_t gfp_flags)
> > +{
> > + struct device *dma_dev = dma_dev = stream_to_dmadev(stream);
> > +
> > + if (!dma_can_alloc_noncontiguous(dma_dev)) {
> > + uvc_urb->buffer = usb_alloc_coherent(stream->dev->udev,
> > + stream->urb_size,
> > + gfp_flags | __GFP_NOWARN,
> > + &uvc_urb->dma);
> > + return uvc_urb->buffer != NULL;
> > + }
> > +
> > + uvc_urb->pages = dma_alloc_noncontiguous(dma_dev, stream->urb_size,
> > + &uvc_urb->dma,
> > + gfp_flags | __GFP_NOWARN, 0);
> > + if (!uvc_urb->pages)
> > + return false;
> > +
> > + uvc_urb->buffer = vmap(uvc_urb->pages,
> > + PAGE_ALIGN(stream->urb_size) >> PAGE_SHIFT,
> > + VM_DMA_COHERENT, PAGE_KERNEL);
> > + if (!uvc_urb->buffer) {
> > + dma_free_noncontiguous(dma_dev, stream->urb_size,
> > + uvc_urb->pages, uvc_urb->dma);
> > + return false;
> > + }
> > +
> > + return true;
> > +}
> > +#else
> > +static bool uvc_alloc_urb_buffer(struct uvc_streaming *stream,
> > + struct uvc_urb *uvc_urb, gfp_t gfp_flags)
> > +{
> > + uvc_urb->buffer = kmalloc(stream->urb_size, gfp_flags | __GFP_NOWARN);
> > +
> > + return uvc_urb->buffer != NULL;
> > +}
> > +#endif
> > +
> > /*
> > * Allocate transfer buffers. This function can be called with buffers
> > * already allocated when resuming from suspend, in which case it will
> > @@ -1607,19 +1665,11 @@ static int uvc_alloc_urb_buffers(struct uvc_streaming *stream,
> >
> > /* Retry allocations until one succeed. */
> > for (; npackets > 1; npackets /= 2) {
> > + stream->urb_size = psize * npackets;
> > for (i = 0; i < UVC_URBS; ++i) {
> > struct uvc_urb *uvc_urb = &stream->uvc_urb[i];
> >
> > - stream->urb_size = psize * npackets;
> > -#ifndef CONFIG_DMA_NONCOHERENT
> > - uvc_urb->buffer = usb_alloc_coherent(
> > - stream->dev->udev, stream->urb_size,
> > - gfp_flags | __GFP_NOWARN, &uvc_urb->dma);
> > -#else
> > - uvc_urb->buffer =
> > - kmalloc(stream->urb_size, gfp_flags | __GFP_NOWARN);
> > -#endif
> > - if (!uvc_urb->buffer) {
> > + if (!uvc_alloc_urb_buffer(stream, uvc_urb, gfp_flags)) {
> > uvc_free_urb_buffers(stream);
> > break;
> > }
> > diff --git a/drivers/media/usb/uvc/uvcvideo.h b/drivers/media/usb/uvc/uvcvideo.h
> > index a3dfacf069c4..3e3ef1f1daa5 100644
> > --- a/drivers/media/usb/uvc/uvcvideo.h
> > +++ b/drivers/media/usb/uvc/uvcvideo.h
> > @@ -532,6 +532,7 @@ struct uvc_urb {
> >
> > char *buffer;
> > dma_addr_t dma;
> > + struct page **pages;
> >
> > unsigned int async_operations;
> > struct uvc_copy_op copy_operations[UVC_MAX_PACKETS];
> >
--
Ricardo Ribalda