Your patch resolves the issue of compiling - I have not yet tested the
functionality of the resulting v4l module. I suspect the disadvantage
of this patch is that it will silently fail to work for the types of
devices (DV primarily) that depend on clock support whereas my patch
omits building code/modules for devices that depend on clock support a
kernel config option that prevents building useless modules (although I
realized the DV_SUPPORT should depend on CLK_COMMON since that's the
whole reason for making that a config option) and (and of course
#ifdef's out the code and dependencies on clk support when it's not
present).
Regards,
Daniel
On 2015-08-15 10:34 AM, Hauke Mehrtens wrote:
On 08/02/2015 04:34 PM, Daniel Dickinson wrote:
The v4l kernel modules fail to compile on latest kernels due to a new
dependency on a clock source. Since most routers won't have a suitable
clock source, I have added a kernel config option that omits the functionality
that requires the clock source.
That means v4l modules in OpenWrt can't support DV devices (DV requires
the clock source support) but unless clock source support was addded to
OpenWrt the other alternative is to drop v4l altogether).
Signed-off-by: Daniel Dickinson <[email protected]>
---
package/kernel/linux/modules/video.mk | 3 +-
...avoid-clk-for-devices-that-do-not-need-it.patch | 348 +++++++++++++++++++++
2 files changed, 350 insertions(+), 1 deletion(-)
create mode 100644
target/linux/generic/patches-4.1/273-v4l-avoid-clk-for-devices-that-do-not-need-it.patch
The build problem of ar71xx was fixed here:
https://dev.openwrt.org/changeset/46572
The arch code did not export all the clk functions needed, but claimed
to do so. Is your patch needed with r46572 any more? When an arch really
does not support clocks it a driver would use the static inline
functions doing nothing instead.
Hauke
_______________________________________________
openwrt-devel mailing list
[email protected]
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
