[PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Fancy Fang]

When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core
will assign the corresponding offset to the 'boff' field of the
videobuf_buffer for each requested buffer sequentially. Later, user
may call mmap() to map one or all of the buffers with the 'offset'
parameter which is equal to its 'boff' value. Obviously, the 'offset'
value is only used to find the matched buffer instead of to be the
real offset from the buffer's physical start address as used by
vm_iomap_memory(). So, in some case that if the offset is not zero,
vm_iomap_memory() will fail.

Signed-off-by: Fancy Fang <chen.fang@freescale.com>
---
 drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c b/drivers/media/v4l2-core/videobuf-dma-contig.c
index bf80f0f..8bd9889 100644
--- a/drivers/media/v4l2-core/videobuf-dma-contig.c
+++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
@@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
 	/* Try to remap memory */
 	size = vma->vm_end - vma->vm_start;
 	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
-	retval = vm_iomap_memory(vma, mem->dma_handle, size);
+	retval = remap_pfn_range(vma, vma->vm_start,
+				 mem->dma_handle >> PAGE_SHIFT,
+				 size, vma->vm_page_prot);
 	if (retval) {
 		dev_err(q->dev, "mmap: remap failed with error %d. ",
 			retval);
-- 
1.9.1

Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Hans Verkuil]

On 09/10/14 07:28, Fancy Fang wrote:
> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core
> will assign the corresponding offset to the 'boff' field of the
> videobuf_buffer for each requested buffer sequentially. Later, user
> may call mmap() to map one or all of the buffers with the 'offset'
> parameter which is equal to its 'boff' value. Obviously, the 'offset'
> value is only used to find the matched buffer instead of to be the
> real offset from the buffer's physical start address as used by
> vm_iomap_memory(). So, in some case that if the offset is not zero,
> vm_iomap_memory() will fail.

Is this just a fix for something that can fail theoretically, or do you
actually have a case where this happens? I am very reluctant to make
any changes to videobuf. Drivers should all migrate to vb2.

I have CC-ed Marek as well since he knows a lot more about this stuff
than I do.

Regards,

	Hans

> 
> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
> ---
>  drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c b/drivers/media/v4l2-core/videobuf-dma-contig.c
> index bf80f0f..8bd9889 100644
> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>  	/* Try to remap memory */
>  	size = vma->vm_end - vma->vm_start;
>  	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
> +	retval = remap_pfn_range(vma, vma->vm_start,
> +				 mem->dma_handle >> PAGE_SHIFT,
> +				 size, vma->vm_page_prot);
>  	if (retval) {
>  		dev_err(q->dev, "mmap: remap failed with error %d. ",
>  			retval);
> 

RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[chen.fang]

It is not a theoretically issue, it is a real case that the mapping failed issue happens in 3.14.y kernel but not happens in previous 3.10.y kernel.
So I need your confirmation on it.

Thanks.

Best regards,
Fancy Fang

-----Original Message-----
From: Hans Verkuil [mailto:hverkuil@xs4all.nl] 
Sent: Wednesday, September 10, 2014 3:01 PM
To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

On 09/10/14 07:28, Fancy Fang wrote:
> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core 
> will assign the corresponding offset to the 'boff' field of the 
> videobuf_buffer for each requested buffer sequentially. Later, user 
> may call mmap() to map one or all of the buffers with the 'offset'
> parameter which is equal to its 'boff' value. Obviously, the 'offset'
> value is only used to find the matched buffer instead of to be the 
> real offset from the buffer's physical start address as used by 
> vm_iomap_memory(). So, in some case that if the offset is not zero,
> vm_iomap_memory() will fail.

Is this just a fix for something that can fail theoretically, or do you actually have a case where this happens? I am very reluctant to make any changes to videobuf. Drivers should all migrate to vb2.

I have CC-ed Marek as well since he knows a lot more about this stuff than I do.

Regards,

	Hans

> 
> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
> ---
>  drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c 
> b/drivers/media/v4l2-core/videobuf-dma-contig.c
> index bf80f0f..8bd9889 100644
> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>  	/* Try to remap memory */
>  	size = vma->vm_end - vma->vm_start;
>  	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
> +	retval = remap_pfn_range(vma, vma->vm_start,
> +				 mem->dma_handle >> PAGE_SHIFT,
> +				 size, vma->vm_page_prot);
>  	if (retval) {
>  		dev_err(q->dev, "mmap: remap failed with error %d. ",
>  			retval);
> 

Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Hans Verkuil]

On 09/10/14 09:14, chen.fang@freescale.com wrote:
> It is not a theoretically issue, it is a real case that the mapping failed issue happens in 3.14.y kernel but not happens in previous 3.10.y kernel.
> So I need your confirmation on it.

With which driver does this happen? On which architecture?

Regards,

	Hans

> 
> Thanks.
> 
> Best regards,
> Fancy Fang
> 
> -----Original Message-----
> From: Hans Verkuil [mailto:hverkuil@xs4all.nl] 
> Sent: Wednesday, September 10, 2014 3:01 PM
> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
> 
> On 09/10/14 07:28, Fancy Fang wrote:
>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core 
>> will assign the corresponding offset to the 'boff' field of the 
>> videobuf_buffer for each requested buffer sequentially. Later, user 
>> may call mmap() to map one or all of the buffers with the 'offset'
>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>> value is only used to find the matched buffer instead of to be the 
>> real offset from the buffer's physical start address as used by 
>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>> vm_iomap_memory() will fail.
> 
> Is this just a fix for something that can fail theoretically, or do you actually have a case where this happens? I am very reluctant to make any changes to videobuf. Drivers should all migrate to vb2.
> 
> I have CC-ed Marek as well since he knows a lot more about this stuff than I do.
> 
> Regards,
> 
> 	Hans
> 
>>
>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>> ---
>>  drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c 
>> b/drivers/media/v4l2-core/videobuf-dma-contig.c
>> index bf80f0f..8bd9889 100644
>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>  	/* Try to remap memory */
>>  	size = vma->vm_end - vma->vm_start;
>>  	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
>> +	retval = remap_pfn_range(vma, vma->vm_start,
>> +				 mem->dma_handle >> PAGE_SHIFT,
>> +				 size, vma->vm_page_prot);
>>  	if (retval) {
>>  		dev_err(q->dev, "mmap: remap failed with error %d. ",
>>  			retval);
>>
> 

RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[chen.fang]

On the Freescale imx6 platform which belongs to ARM architecture. The driver is our local v4l2 output driver which is not upstream yet unfortunately.

Best regards,
Fancy Fang

-----Original Message-----
From: Hans Verkuil [mailto:hverkuil@xs4all.nl] 
Sent: Wednesday, September 10, 2014 3:21 PM
To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

On 09/10/14 09:14, chen.fang@freescale.com wrote:
> It is not a theoretically issue, it is a real case that the mapping failed issue happens in 3.14.y kernel but not happens in previous 3.10.y kernel.
> So I need your confirmation on it.

With which driver does this happen? On which architecture?

Regards,

	Hans

> 
> Thanks.
> 
> Best regards,
> Fancy Fang
> 
> -----Original Message-----
> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
> Sent: Wednesday, September 10, 2014 3:01 PM
> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; 
> linux-kernel@vger.kernel.org; Marek Szyprowski
> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
> 
> On 09/10/14 07:28, Fancy Fang wrote:
>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core 
>> will assign the corresponding offset to the 'boff' field of the 
>> videobuf_buffer for each requested buffer sequentially. Later, user 
>> may call mmap() to map one or all of the buffers with the 'offset'
>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>> value is only used to find the matched buffer instead of to be the 
>> real offset from the buffer's physical start address as used by 
>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>> vm_iomap_memory() will fail.
> 
> Is this just a fix for something that can fail theoretically, or do you actually have a case where this happens? I am very reluctant to make any changes to videobuf. Drivers should all migrate to vb2.
> 
> I have CC-ed Marek as well since he knows a lot more about this stuff than I do.
> 
> Regards,
> 
> 	Hans
> 
>>
>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>> ---
>>  drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c
>> b/drivers/media/v4l2-core/videobuf-dma-contig.c
>> index bf80f0f..8bd9889 100644
>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>  	/* Try to remap memory */
>>  	size = vma->vm_end - vma->vm_start;
>>  	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
>> +	retval = remap_pfn_range(vma, vma->vm_start,
>> +				 mem->dma_handle >> PAGE_SHIFT,
>> +				 size, vma->vm_page_prot);
>>  	if (retval) {
>>  		dev_err(q->dev, "mmap: remap failed with error %d. ",
>>  			retval);
>>
> 

RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[chen.fang]

Hans,
Do you have any more comment on this patch?

Best regards,
Fancy Fang

-----Original Message-----
From: Fang Chen-B47543 
Sent: Wednesday, September 10, 2014 3:29 PM
To: 'Hans Verkuil'; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
Subject: RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

On the Freescale imx6 platform which belongs to ARM architecture. The driver is our local v4l2 output driver which is not upstream yet unfortunately.

Best regards,
Fancy Fang

-----Original Message-----
From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
Sent: Wednesday, September 10, 2014 3:21 PM
To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

On 09/10/14 09:14, chen.fang@freescale.com wrote:
> It is not a theoretically issue, it is a real case that the mapping failed issue happens in 3.14.y kernel but not happens in previous 3.10.y kernel.
> So I need your confirmation on it.

With which driver does this happen? On which architecture?

Regards,

	Hans

> 
> Thanks.
> 
> Best regards,
> Fancy Fang
> 
> -----Original Message-----
> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
> Sent: Wednesday, September 10, 2014 3:01 PM
> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; 
> linux-kernel@vger.kernel.org; Marek Szyprowski
> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
> 
> On 09/10/14 07:28, Fancy Fang wrote:
>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core 
>> will assign the corresponding offset to the 'boff' field of the 
>> videobuf_buffer for each requested buffer sequentially. Later, user 
>> may call mmap() to map one or all of the buffers with the 'offset'
>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>> value is only used to find the matched buffer instead of to be the 
>> real offset from the buffer's physical start address as used by 
>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>> vm_iomap_memory() will fail.
> 
> Is this just a fix for something that can fail theoretically, or do you actually have a case where this happens? I am very reluctant to make any changes to videobuf. Drivers should all migrate to vb2.
> 
> I have CC-ed Marek as well since he knows a lot more about this stuff than I do.
> 
> Regards,
> 
> 	Hans
> 
>>
>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>> ---
>>  drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c
>> b/drivers/media/v4l2-core/videobuf-dma-contig.c
>> index bf80f0f..8bd9889 100644
>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>  	/* Try to remap memory */
>>  	size = vma->vm_end - vma->vm_start;
>>  	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
>> +	retval = remap_pfn_range(vma, vma->vm_start,
>> +				 mem->dma_handle >> PAGE_SHIFT,
>> +				 size, vma->vm_page_prot);
>>  	if (retval) {
>>  		dev_err(q->dev, "mmap: remap failed with error %d. ",
>>  			retval);
>>
> 

Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Hans Verkuil]

Hi Fancy Fang,

I do have a few comments:

1) One reason why the switch to vm_iomap_memory() was made originally was that
that function did a bunch of sanity checks. Since this patch moves back to
remap_pfn_range() those sanity checks are lost and should be reinstated.

2) You need Marek's Ack as well since he understands the memory code much
better than I do.

3) There are efforts underway to make a modern i.mx6 video driver based on
vb2 and all the other modern V4L2 APIs, wouldn't it be much better if freescale
jumps on board and starts working on that code as well? See e.g. this mail
thread: http://www.spinics.net/lists/linux-media/msg79078.html

I assume this refers to the same hardware. The official freescale driver
for that hardware is horrendous.

If you are writing code for unrelated hardware, then you should move over to
vb2 yourself. videobuf should not be used anymore for new drivers.

4) I still do not entirely understand the control flow and I will have to take
another look. I'll see if I can do that tomorrow.

Regards,

	Hans

On 09/23/2014 05:11 AM, chen.fang@freescale.com wrote:
> Hans,
> Do you have any more comment on this patch?
> 
> Best regards,
> Fancy Fang
> 
> -----Original Message-----
> From: Fang Chen-B47543 
> Sent: Wednesday, September 10, 2014 3:29 PM
> To: 'Hans Verkuil'; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
> Subject: RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
> 
> On the Freescale imx6 platform which belongs to ARM architecture. The driver is our local v4l2 output driver which is not upstream yet unfortunately.
> 
> Best regards,
> Fancy Fang
> 
> -----Original Message-----
> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
> Sent: Wednesday, September 10, 2014 3:21 PM
> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
> 
> On 09/10/14 09:14, chen.fang@freescale.com wrote:
>> It is not a theoretically issue, it is a real case that the mapping failed issue happens in 3.14.y kernel but not happens in previous 3.10.y kernel.
>> So I need your confirmation on it.
> 
> With which driver does this happen? On which architecture?
> 
> Regards,
> 
> 	Hans
> 
>>
>> Thanks.
>>
>> Best regards,
>> Fancy Fang
>>
>> -----Original Message-----
>> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
>> Sent: Wednesday, September 10, 2014 3:01 PM
>> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
>> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; 
>> linux-kernel@vger.kernel.org; Marek Szyprowski
>> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
>>
>> On 09/10/14 07:28, Fancy Fang wrote:
>>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core 
>>> will assign the corresponding offset to the 'boff' field of the 
>>> videobuf_buffer for each requested buffer sequentially. Later, user 
>>> may call mmap() to map one or all of the buffers with the 'offset'
>>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>>> value is only used to find the matched buffer instead of to be the 
>>> real offset from the buffer's physical start address as used by 
>>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>>> vm_iomap_memory() will fail.
>>
>> Is this just a fix for something that can fail theoretically, or do you actually have a case where this happens? I am very reluctant to make any changes to videobuf. Drivers should all migrate to vb2.
>>
>> I have CC-ed Marek as well since he knows a lot more about this stuff than I do.
>>
>> Regards,
>>
>> 	Hans
>>
>>>
>>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>>> ---
>>>  drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> index bf80f0f..8bd9889 100644
>>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>>  	/* Try to remap memory */
>>>  	size = vma->vm_end - vma->vm_start;
>>>  	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>>> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
>>> +	retval = remap_pfn_range(vma, vma->vm_start,
>>> +				 mem->dma_handle >> PAGE_SHIFT,
>>> +				 size, vma->vm_page_prot);
>>>  	if (retval) {
>>>  		dev_err(q->dev, "mmap: remap failed with error %d. ",
>>>  			retval);
>>>
>>
> 

RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[chen.fang]

Thanks a lot, Hans.
 
I know the reason that you use vm_iomap_memory() instead of remap_pfn_range(). But
according to my analysis, the sanity check is not suitable for v4l2 buffers mapping. You can check the
code logic to see this. 

Migration to vb2 is not a quick job and it is not on our schedule yet.
I just need you to check the code logic and make a decision for this.

Best regards,
Fancy Fang


-----Original Message-----
From: Hans Verkuil [mailto:hverkuil@xs4all.nl] 
Sent: Tuesday, September 23, 2014 2:10 PM
To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

Hi Fancy Fang,

I do have a few comments:

1) One reason why the switch to vm_iomap_memory() was made originally was that that function did a bunch of sanity checks. Since this patch moves back to
remap_pfn_range() those sanity checks are lost and should be reinstated.

2) You need Marek's Ack as well since he understands the memory code much better than I do.

3) There are efforts underway to make a modern i.mx6 video driver based on
vb2 and all the other modern V4L2 APIs, wouldn't it be much better if freescale jumps on board and starts working on that code as well? See e.g. this mail
thread: http://www.spinics.net/lists/linux-media/msg79078.html

I assume this refers to the same hardware. The official freescale driver for that hardware is horrendous.

If you are writing code for unrelated hardware, then you should move over to
vb2 yourself. videobuf should not be used anymore for new drivers.

4) I still do not entirely understand the control flow and I will have to take another look. I'll see if I can do that tomorrow.

Regards,

	Hans

On 09/23/2014 05:11 AM, chen.fang@freescale.com wrote:
> Hans,
> Do you have any more comment on this patch?
> 
> Best regards,
> Fancy Fang
> 
> -----Original Message-----
> From: Fang Chen-B47543
> Sent: Wednesday, September 10, 2014 3:29 PM
> To: 'Hans Verkuil'; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; 
> linux-kernel@vger.kernel.org; Marek Szyprowski
> Subject: RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
> 
> On the Freescale imx6 platform which belongs to ARM architecture. The driver is our local v4l2 output driver which is not upstream yet unfortunately.
> 
> Best regards,
> Fancy Fang
> 
> -----Original Message-----
> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
> Sent: Wednesday, September 10, 2014 3:21 PM
> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; 
> linux-kernel@vger.kernel.org; Marek Szyprowski
> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
> 
> On 09/10/14 09:14, chen.fang@freescale.com wrote:
>> It is not a theoretically issue, it is a real case that the mapping failed issue happens in 3.14.y kernel but not happens in previous 3.10.y kernel.
>> So I need your confirmation on it.
> 
> With which driver does this happen? On which architecture?
> 
> Regards,
> 
> 	Hans
> 
>>
>> Thanks.
>>
>> Best regards,
>> Fancy Fang
>>
>> -----Original Message-----
>> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
>> Sent: Wednesday, September 10, 2014 3:01 PM
>> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
>> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; 
>> linux-kernel@vger.kernel.org; Marek Szyprowski
>> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
>>
>> On 09/10/14 07:28, Fancy Fang wrote:
>>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core 
>>> will assign the corresponding offset to the 'boff' field of the 
>>> videobuf_buffer for each requested buffer sequentially. Later, user 
>>> may call mmap() to map one or all of the buffers with the 'offset'
>>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>>> value is only used to find the matched buffer instead of to be the 
>>> real offset from the buffer's physical start address as used by 
>>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>>> vm_iomap_memory() will fail.
>>
>> Is this just a fix for something that can fail theoretically, or do you actually have a case where this happens? I am very reluctant to make any changes to videobuf. Drivers should all migrate to vb2.
>>
>> I have CC-ed Marek as well since he knows a lot more about this stuff than I do.
>>
>> Regards,
>>
>> 	Hans
>>
>>>
>>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>>> ---
>>>  drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>>  1 file changed, 3 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> index bf80f0f..8bd9889 100644
>>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>>  	/* Try to remap memory */
>>>  	size = vma->vm_end - vma->vm_start;
>>>  	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>>> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
>>> +	retval = remap_pfn_range(vma, vma->vm_start,
>>> +				 mem->dma_handle >> PAGE_SHIFT,
>>> +				 size, vma->vm_page_prot);
>>>  	if (retval) {
>>>  		dev_err(q->dev, "mmap: remap failed with error %d. ",
>>>  			retval);
>>>
>>
> 

Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Hans Verkuil]

On 09/24/2014 05:29 AM, chen.fang@freescale.com wrote:
> Thanks a lot, Hans.
>
> I know the reason that you use vm_iomap_memory() instead of remap_pfn_range(). But
> according to my analysis, the sanity check is not suitable for v4l2 buffers mapping. You can check the
> code logic to see this.

The logic relating to vm_pgoff might not be suitable, but other checks are.

OK, I've looked at this a bit more and what I don't understand is why this
has never caused problems before. Would this perhaps just work on intel
architecture? Most platform drivers that would typically use the arm
architecture use vb2 by now, so that might explain it.

Anyway, put back the relevant checks and I'll Ack it.

>
> Migration to vb2 is not a quick job and it is not on our schedule yet.

You really should do this. At some point the old videobuf will go away.
More important, vb2 has support for the dmabuf API allowing you to share
buffers between v4l and drm/kms.

There are many other advantages of moving to vb2 including proper instrumentation
and a control flow that doesn't make you go insane like vb1 does.

And if you are still using the old v4l2-int-device API, then that's already
no longer available as an API and is only used by a staging driver that will
be removed soon. This is going to cause you major headaches if you want to
support recent and upcoming kernels.

Regards,

	Hans

> I just need you to check the code logic and make a decision for this.
>
> Best regards,
> Fancy Fang
>
>
> -----Original Message-----
> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
> Sent: Tuesday, September 23, 2014 2:10 PM
> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org; Marek Szyprowski
> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
>
> Hi Fancy Fang,
>
> I do have a few comments:
>
> 1) One reason why the switch to vm_iomap_memory() was made originally was that that function did a bunch of sanity checks. Since this patch moves back to
> remap_pfn_range() those sanity checks are lost and should be reinstated.
>
> 2) You need Marek's Ack as well since he understands the memory code much better than I do.
>
> 3) There are efforts underway to make a modern i.mx6 video driver based on
> vb2 and all the other modern V4L2 APIs, wouldn't it be much better if freescale jumps on board and starts working on that code as well? See e.g. this mail
> thread: http://www.spinics.net/lists/linux-media/msg79078.html
>
> I assume this refers to the same hardware. The official freescale driver for that hardware is horrendous.
>
> If you are writing code for unrelated hardware, then you should move over to
> vb2 yourself. videobuf should not be used anymore for new drivers.
>
> 4) I still do not entirely understand the control flow and I will have to take another look. I'll see if I can do that tomorrow.
>
> Regards,
>
> 	Hans
>
> On 09/23/2014 05:11 AM, chen.fang@freescale.com wrote:
>> Hans,
>> Do you have any more comment on this patch?
>>
>> Best regards,
>> Fancy Fang
>>
>> -----Original Message-----
>> From: Fang Chen-B47543
>> Sent: Wednesday, September 10, 2014 3:29 PM
>> To: 'Hans Verkuil'; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
>> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org;
>> linux-kernel@vger.kernel.org; Marek Szyprowski
>> Subject: RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
>>
>> On the Freescale imx6 platform which belongs to ARM architecture. The driver is our local v4l2 output driver which is not upstream yet unfortunately.
>>
>> Best regards,
>> Fancy Fang
>>
>> -----Original Message-----
>> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
>> Sent: Wednesday, September 10, 2014 3:21 PM
>> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
>> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org;
>> linux-kernel@vger.kernel.org; Marek Szyprowski
>> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
>>
>> On 09/10/14 09:14, chen.fang@freescale.com wrote:
>>> It is not a theoretically issue, it is a real case that the mapping failed issue happens in 3.14.y kernel but not happens in previous 3.10.y kernel.
>>> So I need your confirmation on it.
>>
>> With which driver does this happen? On which architecture?
>>
>> Regards,
>>
>> 	Hans
>>
>>>
>>> Thanks.
>>>
>>> Best regards,
>>> Fancy Fang
>>>
>>> -----Original Message-----
>>> From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
>>> Sent: Wednesday, September 10, 2014 3:01 PM
>>> To: Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
>>> Cc: Guo Shawn-R65073; linux-media@vger.kernel.org;
>>> linux-kernel@vger.kernel.org; Marek Szyprowski
>>> Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().
>>>
>>> On 09/10/14 07:28, Fancy Fang wrote:
>>>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core
>>>> will assign the corresponding offset to the 'boff' field of the
>>>> videobuf_buffer for each requested buffer sequentially. Later, user
>>>> may call mmap() to map one or all of the buffers with the 'offset'
>>>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>>>> value is only used to find the matched buffer instead of to be the
>>>> real offset from the buffer's physical start address as used by
>>>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>>>> vm_iomap_memory() will fail.
>>>
>>> Is this just a fix for something that can fail theoretically, or do you actually have a case where this happens? I am very reluctant to make any changes to videobuf. Drivers should all migrate to vb2.
>>>
>>> I have CC-ed Marek as well since he knows a lot more about this stuff than I do.
>>>
>>> Regards,
>>>
>>> 	Hans
>>>
>>>>
>>>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>>>> ---
>>>>   drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>>> b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>>> index bf80f0f..8bd9889 100644
>>>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>>>   	/* Try to remap memory */
>>>>   	size = vma->vm_end - vma->vm_start;
>>>>   	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>>>> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
>>>> +	retval = remap_pfn_range(vma, vma->vm_start,
>>>> +				 mem->dma_handle >> PAGE_SHIFT,
>>>> +				 size, vma->vm_page_prot);
>>>>   	if (retval) {
>>>>   		dev_err(q->dev, "mmap: remap failed with error %d. ",
>>>>   			retval);
>>>>
>>>
>>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Marek Szyprowski]

Hello,

On 2014-09-10 09:00, Hans Verkuil wrote:
> On 09/10/14 07:28, Fancy Fang wrote:
>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core
>> will assign the corresponding offset to the 'boff' field of the
>> videobuf_buffer for each requested buffer sequentially. Later, user
>> may call mmap() to map one or all of the buffers with the 'offset'
>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>> value is only used to find the matched buffer instead of to be the
>> real offset from the buffer's physical start address as used by
>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>> vm_iomap_memory() will fail.
> Is this just a fix for something that can fail theoretically, or do you
> actually have a case where this happens? I am very reluctant to make
> any changes to videobuf. Drivers should all migrate to vb2.
>
> I have CC-ed Marek as well since he knows a lot more about this stuff
> than I do.

I'm sorry for a delay, I was really busy with other things.

>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>> ---
>>   drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c b/drivers/media/v4l2-core/videobuf-dma-contig.c
>> index bf80f0f..8bd9889 100644
>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>   	/* Try to remap memory */
>>   	size = vma->vm_end - vma->vm_start;
>>   	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>> -	retval = vm_iomap_memory(vma, mem->dma_handle, size);
>> +	retval = remap_pfn_range(vma, vma->vm_start,
>> +				 mem->dma_handle >> PAGE_SHIFT,
>> +				 size, vma->vm_page_prot);
>>   	if (retval) {
>>   		dev_err(q->dev, "mmap: remap failed with error %d. ",
>>   			retval);

I think we don't need to revert the code to use remap_pfn_range() again 
(like
it was in pre v3.10 times). The simplest way will be to correctly fix
vma->vm_pgoff and set it to zero before calling vm_iomap_memory(). It is
done the same way in vb2_dma_contig.c:vb2_dc_mmap().

To sum up - please change your patch: keep vm_iomap_memory() call and add
"vma->vm_pgoff = 0;" line before it with suitable comment.

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland

Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Hans Verkuil]

On 09/24/2014 02:40 PM, Marek Szyprowski wrote:
> Hello,
>
> On 2014-09-10 09:00, Hans Verkuil wrote:
>> On 09/10/14 07:28, Fancy Fang wrote:
>>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core
>>> will assign the corresponding offset to the 'boff' field of the
>>> videobuf_buffer for each requested buffer sequentially. Later, user
>>> may call mmap() to map one or all of the buffers with the 'offset'
>>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>>> value is only used to find the matched buffer instead of to be the
>>> real offset from the buffer's physical start address as used by
>>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>>> vm_iomap_memory() will fail.
>> Is this just a fix for something that can fail theoretically, or do you
>> actually have a case where this happens? I am very reluctant to make
>> any changes to videobuf. Drivers should all migrate to vb2.
>>
>> I have CC-ed Marek as well since he knows a lot more about this stuff
>> than I do.
>
> I'm sorry for a delay, I was really busy with other things.
>
>>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>>> ---
>>>   drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> index bf80f0f..8bd9889 100644
>>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>>       /* Try to remap memory */
>>>       size = vma->vm_end - vma->vm_start;
>>>       vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>>> -    retval = vm_iomap_memory(vma, mem->dma_handle, size);
>>> +    retval = remap_pfn_range(vma, vma->vm_start,
>>> +                 mem->dma_handle >> PAGE_SHIFT,
>>> +                 size, vma->vm_page_prot);
>>>       if (retval) {
>>>           dev_err(q->dev, "mmap: remap failed with error %d. ",
>>>               retval);
>
> I think we don't need to revert the code to use remap_pfn_range() again (like
> it was in pre v3.10 times). The simplest way will be to correctly fix
> vma->vm_pgoff and set it to zero before calling vm_iomap_memory(). It is
> done the same way in vb2_dma_contig.c:vb2_dc_mmap().
>
> To sum up - please change your patch: keep vm_iomap_memory() call and add
> "vma->vm_pgoff = 0;" line before it with suitable comment.

Much better, I agree completely.

Regards,

	Hans

RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[chen.fang]

Thanks for your idea, Marek and Hans. This change makes sense. And I've already
verified it. I'll resubmit my change soon.

Best regards,
Fancy Fang

-----Original Message-----
From: Hans Verkuil [mailto:hverkuil@xs4all.nl] 
Sent: Wednesday, September 24, 2014 8:43 PM
To: Marek Szyprowski; Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org
Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().



On 09/24/2014 02:40 PM, Marek Szyprowski wrote:
> Hello,
>
> On 2014-09-10 09:00, Hans Verkuil wrote:
>> On 09/10/14 07:28, Fancy Fang wrote:
>>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core 
>>> will assign the corresponding offset to the 'boff' field of the 
>>> videobuf_buffer for each requested buffer sequentially. Later, user 
>>> may call mmap() to map one or all of the buffers with the 'offset'
>>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>>> value is only used to find the matched buffer instead of to be the 
>>> real offset from the buffer's physical start address as used by 
>>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>>> vm_iomap_memory() will fail.
>> Is this just a fix for something that can fail theoretically, or do 
>> you actually have a case where this happens? I am very reluctant to 
>> make any changes to videobuf. Drivers should all migrate to vb2.
>>
>> I have CC-ed Marek as well since he knows a lot more about this stuff 
>> than I do.
>
> I'm sorry for a delay, I was really busy with other things.
>
>>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>>> ---
>>>   drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c 
>>> b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> index bf80f0f..8bd9889 100644
>>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>>       /* Try to remap memory */
>>>       size = vma->vm_end - vma->vm_start;
>>>       vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>>> -    retval = vm_iomap_memory(vma, mem->dma_handle, size);
>>> +    retval = remap_pfn_range(vma, vma->vm_start,
>>> +                 mem->dma_handle >> PAGE_SHIFT,
>>> +                 size, vma->vm_page_prot);
>>>       if (retval) {
>>>           dev_err(q->dev, "mmap: remap failed with error %d. ",
>>>               retval);
>
> I think we don't need to revert the code to use remap_pfn_range() 
> again (like it was in pre v3.10 times). The simplest way will be to 
> correctly fix
> vma->vm_pgoff and set it to zero before calling vm_iomap_memory(). It 
> vma->is
> done the same way in vb2_dma_contig.c:vb2_dc_mmap().
>
> To sum up - please change your patch: keep vm_iomap_memory() call and 
> add "vma->vm_pgoff = 0;" line before it with suitable comment.

Much better, I agree completely.

Regards,

	Hans

RE: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[chen.fang]

Thanks for your idea, Marek and Hans. This change makes sense. And I've already verified it. I'll resubmit my change soon.

Best regards,
Fancy Fang

-----Original Message-----
From: Hans Verkuil [mailto:hverkuil@xs4all.nl]
Sent: Wednesday, September 24, 2014 8:43 PM
To: Marek Szyprowski; Fang Chen-B47543; m.chehab@samsung.com; viro@ZenIV.linux.org.uk
Cc: Guo Shawn-R65073; linux-media@vger.kernel.org; linux-kernel@vger.kernel.org
Subject: Re: [PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().



On 09/24/2014 02:40 PM, Marek Szyprowski wrote:
> Hello,
>
> On 2014-09-10 09:00, Hans Verkuil wrote:
>> On 09/10/14 07:28, Fancy Fang wrote:
>>> When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core 
>>> will assign the corresponding offset to the 'boff' field of the 
>>> videobuf_buffer for each requested buffer sequentially. Later, user 
>>> may call mmap() to map one or all of the buffers with the 'offset'
>>> parameter which is equal to its 'boff' value. Obviously, the 'offset'
>>> value is only used to find the matched buffer instead of to be the 
>>> real offset from the buffer's physical start address as used by 
>>> vm_iomap_memory(). So, in some case that if the offset is not zero,
>>> vm_iomap_memory() will fail.
>> Is this just a fix for something that can fail theoretically, or do 
>> you actually have a case where this happens? I am very reluctant to 
>> make any changes to videobuf. Drivers should all migrate to vb2.
>>
>> I have CC-ed Marek as well since he knows a lot more about this stuff 
>> than I do.
>
> I'm sorry for a delay, I was really busy with other things.
>
>>> Signed-off-by: Fancy Fang <chen.fang@freescale.com>
>>> ---
>>>   drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
>>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c 
>>> b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> index bf80f0f..8bd9889 100644
>>> --- a/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> +++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
>>> @@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
>>>       /* Try to remap memory */
>>>       size = vma->vm_end - vma->vm_start;
>>>       vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
>>> -    retval = vm_iomap_memory(vma, mem->dma_handle, size);
>>> +    retval = remap_pfn_range(vma, vma->vm_start,
>>> +                 mem->dma_handle >> PAGE_SHIFT,
>>> +                 size, vma->vm_page_prot);
>>>       if (retval) {
>>>           dev_err(q->dev, "mmap: remap failed with error %d. ",
>>>               retval);
>
> I think we don't need to revert the code to use remap_pfn_range() 
> again (like it was in pre v3.10 times). The simplest way will be to 
> correctly fix
> vma->vm_pgoff and set it to zero before calling vm_iomap_memory(). It 
> vma->is
> done the same way in vb2_dma_contig.c:vb2_dc_mmap().
>
> To sum up - please change your patch: keep vm_iomap_memory() call and 
> add "vma->vm_pgoff = 0;" line before it with suitable comment.

Much better, I agree completely.

Regards,

	Hans

[PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Fancy Fang]

When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core
will assign the corresponding offset to the 'boff' field of the
videobuf_buffer for each requested buffer sequentially. Later, user
may call mmap() to map one or all of the buffers with the 'offset'
parameter which is equal to its 'boff' value. Obviously, the 'offset'
value is only used to find the matched buffer instead of to be the
real offset from the buffer's physical start address as used by
vm_iomap_memory(). So, in some case that if the offset is not zero,
vm_iomap_memory() will fail.

Signed-off-by: Fancy Fang <chen.fang@freescale.com>
---
 drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c b/drivers/media/v4l2-core/videobuf-dma-contig.c
index bf80f0f..8bd9889 100644
--- a/drivers/media/v4l2-core/videobuf-dma-contig.c
+++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
@@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
 	/* Try to remap memory */
 	size = vma->vm_end - vma->vm_start;
 	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
-	retval = vm_iomap_memory(vma, mem->dma_handle, size);
+	retval = remap_pfn_range(vma, vma->vm_start,
+				 mem->dma_handle >> PAGE_SHIFT,
+				 size, vma->vm_page_prot);
 	if (retval) {
 		dev_err(q->dev, "mmap: remap failed with error %d. ",
 			retval);
-- 
1.9.1

[PATCH] [media] videobuf-dma-contig: replace vm_iomap_memory() with remap_pfn_range().

[Fancy Fang]

When user requests V4L2_MEMORY_MMAP type buffers, the videobuf-core
will assign the corresponding offset to the 'boff' field of the
videobuf_buffer for each requested buffer sequentially. Later, user
may call mmap() to map one or all of the buffers with the 'offset'
parameter which is equal to its 'boff' value. Obviously, the 'offset'
value is only used to find the matched buffer instead of to be the
real offset from the buffer's physical start address as used by
vm_iomap_memory(). So, in some case that if the offset is not zero,
vm_iomap_memory() will fail.

Signed-off-by: Fancy Fang <chen.fang@freescale.com>
---
 drivers/media/v4l2-core/videobuf-dma-contig.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/media/v4l2-core/videobuf-dma-contig.c b/drivers/media/v4l2-core/videobuf-dma-contig.c
index bf80f0f..8bd9889 100644
--- a/drivers/media/v4l2-core/videobuf-dma-contig.c
+++ b/drivers/media/v4l2-core/videobuf-dma-contig.c
@@ -305,7 +305,9 @@ static int __videobuf_mmap_mapper(struct videobuf_queue *q,
 	/* Try to remap memory */
 	size = vma->vm_end - vma->vm_start;
 	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
-	retval = vm_iomap_memory(vma, mem->dma_handle, size);
+	retval = remap_pfn_range(vma, vma->vm_start,
+				 mem->dma_handle >> PAGE_SHIFT,
+				 size, vma->vm_page_prot);
 	if (retval) {
 		dev_err(q->dev, "mmap: remap failed with error %d. ",
 			retval);
-- 
1.9.1