On Friday 18 September 2015 11:49:50 Hans Verkuil wrote:
> *If* the conversion takes place only in v4l2-ioctl.c, then it makes sense
> have these structs + ioctls moved to v4l2-ioctl.h.
Ok.
> I noticed that v4l2-compat-ioctl32.c wasn't changed. Is that right? I have
> unfortunately no time to do
On 09/18/15 11:26, Arnd Bergmann wrote:
> On Friday 18 September 2015 09:18:45 Hans Verkuil wrote:
>> Hi Arnd,
>>
>> Thanks once again for working on this! Unfortunately, this approach won't
>> work, see my comments below.
>>
>> BTW, I would expect to see compile errors when compiling for 32 bit. D
On Friday 18 September 2015 09:18:45 Hans Verkuil wrote:
> Hi Arnd,
>
> Thanks once again for working on this! Unfortunately, this approach won't
> work, see my comments below.
>
> BTW, I would expect to see compile errors when compiling for 32 bit. Did
> you try that?
I only tested on 32-bit, b
Hi Arnd,
Thanks once again for working on this! Unfortunately, this approach won't
work, see my comments below.
BTW, I would expect to see compile errors when compiling for 32 bit. Did
you try that?
On 09/17/2015 11:19 PM, Arnd Bergmann wrote:
> This is the final change to enable user space with
This is the final change to enable user space with 64-bit time_t
in the core v4l2 code.
Four ioctls are affected here: VIDIOC_QUERYBUF, VIDIOC_QBUF,
VIDIOC_DQBUF, and VIDIOC_PREPARE_BUF. All of them pass a
struct v4l2_buffer, which can either contain a 32-bit time_t
or a 64-bit time on 32-bit arch
