Hello,
On 2016-04-27 14:58, Hans Verkuil wrote:
On 04/27/2016 02:23 PM, Marek Szyprowski wrote:
Hello,
On 2016-04-27 14:10, Hans Verkuil wrote:
On 04/27/2016 02:00 PM, Marek Szyprowski wrote:
Add a helper function for device drivers to set DMA's max_seg_size.
Setting it to largest possible value lets DMA-mapping API always create
contiguous mappings in DMA address space. This is essential for all
devices, which use dma-contig videobuf2 memory allocator and shared
buffers.
Signed-off-by: Marek Szyprowski <m.szyprowski@xxxxxxxxxxx>
---
This patch was posted earlier as a part of
http://thread.gmane.org/gmane.linux.drivers.video-input-infrastructure/97316
thread, but applying it is really needed to get all Exynos multimedia
drivers working with IOMMU enabled.
Best regards,
Marek Szyprowski
---
drivers/media/v4l2-core/videobuf2-dma-contig.c | 15 +++++++++++++++
include/media/videobuf2-dma-contig.h | 1 +
2 files changed, 16 insertions(+)
diff --git a/drivers/media/v4l2-core/videobuf2-dma-contig.c b/drivers/media/v4l2-core/videobuf2-dma-contig.c
index 5361197..f611456 100644
--- a/drivers/media/v4l2-core/videobuf2-dma-contig.c
+++ b/drivers/media/v4l2-core/videobuf2-dma-contig.c
@@ -753,6 +753,21 @@ void vb2_dma_contig_cleanup_ctx(void *alloc_ctx)
}
EXPORT_SYMBOL_GPL(vb2_dma_contig_cleanup_ctx);
+int vb2_dma_contig_set_max_seg_size(struct device *dev, unsigned int size)
+{
+ if (!dev->dma_parms) {
+ dev->dma_parms = devm_kzalloc(dev, sizeof(dev->dma_parms),
+ GFP_KERNEL);
The v3 patch from December uses kzalloc here. Is this perhaps on old version?
Right, my fault. I will do another resend (and fix the typo in the second patch).
+ if (!dev->dma_parms)
+ return -ENOMEM;
+ }
+ if (dma_get_max_seg_size(dev) < size)
+ return dma_set_max_seg_size(dev, size);
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(vb2_dma_contig_set_max_seg_size);
Admittedly I haven't looked closely at this, but is this something that you
want for all dma-contig devices? Or to rephrase this question: what type of
devices will need this?
This is needed for all devices using vb2-dc, iommu and user-ptr mode, however
in the previous discussions (see https://patchwork.linuxtv.org/patch/30870/
) it has been suggested to make it via common helper instead of forcing it
in vb2-dc.
This certainly will need to be carefully documented in videobuf2-dma-contig.h.
What happens if it is called when you don't have an iommu? Does something break?
Nope, nothing breaks in such case. When no iommu is available this
parameter is
ignored by dma-mapping layer. Due to some other issues, it cannot be set by
generic platform init code:
http://lists.infradead.org/pipermail/linux-arm-kernel/2014-November/305913.html
Best regards
--
Marek Szyprowski, PhD
Samsung R&D Institute Poland
--
To unsubscribe from this list: send the line "unsubscribe linux-samsung-soc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
