Re: [PATCH v2] patches: fix v4l2 patch

[Florian Grandel <fgrandel@gmail.com>]

Hi Hauke,

I seem to have no luck with this patch. :-(

On 07/01/2015 11:31 PM, Hauke Mehrtens wrote:
> On 07/01/2015 04:06 PM, Florian Grandel wrote:
>> Fix to support next-20160626.
>>
>> Signed-off-by: Florian Grandel <fgrandel@gmail.com>
>> ---
>>   .../media/0005-dma-no-sync/v4l2.patch.orig         | 67 ++++++++++++++++++++++
>>   1 file changed, 67 insertions(+)
>>   create mode 100644 patches/collateral-evolutions/media/0005-dma-no-sync/v4l2.patch.orig
>
> Why do you want to add a .orig file?

Of course the orig file slipped in inadvertently.

> A similar fix is already in backports. It went into in this commit:
> https://git.kernel.org/cgit/linux/kernel/git/backports/backports.git/commit/?id=c1a416852f24d166c51d8203703a1bddff1ad6f8
>
> The current release based on kernel next-20160626 was compile tested
> against all the last minor versions of kernel 3.0 to 4.1 on x86. There
> could still be some problems with different kernel configurations or
> with different architectures.
>
>>
>> diff --git a/patches/collateral-evolutions/media/0005-dma-no-sync/v4l2.patch.orig b/patches/collateral-evolutions/media/0005-dma-no-sync/v4l2.patch.orig
>> new file mode 100644
>> index 0000000..7c99937
>> --- /dev/null
>> +++ b/patches/collateral-evolutions/media/0005-dma-no-sync/v4l2.patch.orig
>> @@ -0,0 +1,67 @@
>> +<<<<<<< HEAD
>> +--- a/drivers/media/v4l2-core/videobuf2-dma-contig.c
>> ++++ b/drivers/media/v4l2-core/videobuf2-dma-contig.c
>> +@@ -470,7 +470,9 @@ static void vb2_dc_put_userptr(void *buf
>> + 	if (sgt) {
>> + 		DEFINE_DMA_ATTRS(attrs);
>> +
>> ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,6,0)
>> + 		dma_set_attr(DMA_ATTR_SKIP_CPU_SYNC, &attrs);
>> ++#endif
>> + 		/*
>> + 		 * No need to sync to CPU, it's already synced to the CPU
>> + 		 * since the finish() memop will have been called before this.
>> +@@ -534,7 +536,9 @@ static void *vb2_dc_get_userptr(void *al
>> + 	unsigned long dma_align = dma_get_cache_alignment();
>> + 	DEFINE_DMA_ATTRS(attrs);
>> +
>> ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,6,0)
>> + 	dma_set_attr(DMA_ATTR_SKIP_CPU_SYNC, &attrs);
>> ++#endif
>> +
>> + 	/* Only cache aligned DMA transfers are reliable */
>> + 	if (!IS_ALIGNED(vaddr | size, dma_align)) {
>> +=======
>> +>>>>>>> patches: fix v4l2 patch
>
> It looks like you want to add a merge conflict here.

Yep. I was probably asleep already when I prepared that commit.

>> +--- a/drivers/media/v4l2-core/videobuf2-dma-sg.c
>> ++++ b/drivers/media/v4l2-core/videobuf2-dma-sg.c
>> +@@ -109,7 +109,9 @@ static void *vb2_dma_sg_alloc(void *allo
>> + 	int num_pages;
>> + 	DEFINE_DMA_ATTRS(attrs);
>> +
>> ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,6,0)
>> + 	dma_set_attr(DMA_ATTR_SKIP_CPU_SYNC, &attrs);
>> ++#endif
>> +
>> + 	if (WARN_ON(alloc_ctx == NULL))
>> + 		return NULL;
>> +@@ -185,7 +187,9 @@ static void vb2_dma_sg_put(void *buf_pri
>> + 	if (atomic_dec_and_test(&buf->refcount)) {
>> + 		DEFINE_DMA_ATTRS(attrs);
>> +
>> ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,6,0)
>> + 		dma_set_attr(DMA_ATTR_SKIP_CPU_SYNC, &attrs);
>> ++#endif
>> + 		dprintk(1, "%s: Freeing buffer of %d pages\n", __func__,
>> + 			buf->num_pages);
>> + 		dma_unmap_sg_attrs(buf->dev, sgt->sgl, sgt->orig_nents,
>> +@@ -235,7 +239,9 @@ static void *vb2_dma_sg_get_userptr(void
>> + 	DEFINE_DMA_ATTRS(attrs);
>> + 	struct frame_vector *vec;
>> +
>> ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,6,0)
>> + 	dma_set_attr(DMA_ATTR_SKIP_CPU_SYNC, &attrs);
>> ++#endif
>> + 	buf = kzalloc(sizeof *buf, GFP_KERNEL);
>> + 	if (!buf)
>> + 		return NULL;
>> +@@ -292,7 +298,9 @@ static void vb2_dma_sg_put_userptr(void
>> + 	int i = buf->num_pages;
>> + 	DEFINE_DMA_ATTRS(attrs);
>> +
>> ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,6,0)
>> + 	dma_set_attr(DMA_ATTR_SKIP_CPU_SYNC, &attrs);
>> ++#endif
>> +
>> + 	dprintk(1, "%s: Releasing userspace buffer of %d pages\n",
>> + 	       __func__, buf->num_pages);
>>

As you say: this has already been fixed elsewhere and the artifact you 
got here is nonsense. :-D

Please forget about the patch. I'm really sorry for the noise.

Florian
--
To unsubscribe from this list: send the line "unsubscribe backports" in