[yocto] [meta-raspberrypi][PATCH] linux-raspberrypi: fix path to mkknlimg

[Jonathan Liu]

On 28 February 2016 at 00:05, Andrei Gherzan <andrei at gherzan.ro> wrote:
> On Sat, Feb 27, 2016 at 11:25:28PM +1100, Jonathan Liu wrote:
>> On 27 February 2016 at 23:21, Andrei Gherzan <andrei at gherzan.ro> wrote:
>> > On Sat, Feb 27, 2016 at 08:03:10AM +1100, Jonathan Liu wrote:
>> >> On 27 February 2016 at 06:55, Andrei Gherzan <andrei at gherzan.ro> wrote:
>> >> > On Fri, Feb 26, 2016 at 07:04:28PM +0100, Andrei Gherzan wrote:
>> >> >>    --
>> >> >>    Andrei Gherzan
>> >> >>    On Fri, Feb 26, 2016 at 6:48 PM, Petter Mabäcker
>> >> >>    <[1]petter at technux.se> wrote:
>> >> >>
>> >> >>    2016-02-26 18:14 skrev Andrei Gherzan:
>> >> >>
>> >> >> On Fri, Feb 26, 2016 at 08:26:57PM +1100, Jonathan Liu wrote:
>> >> >>
>> >> >>      The path had libexecdir hardcoded so it was no longer correct when
>> >> >>      libexecdir was changed in meta/conf/bitbake.conf from
>> >> >>      ${libdir}/${BPN} to ${prefix_native}/libexec for nativesdk recipes.
>> >> >>      Fixes "No such file or directory" error when executing
>> >> >>      do_rpiboot_mkimage. Signed-off-by: Jonathan Liu
>> >> >>      <[2]net147 at gmail.com> --- recipes-kernel/linux/linux-raspberrypi.inc
>> >> >>      | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git
>> >> >>      a/recipes-kernel/linux/linux-raspberrypi.inc
>> >> >>      b/recipes-kernel/linux/linux-raspberrypi.inc index 70e8bfe..8c7c46c
>> >> >>      100644 --- a/recipes-kernel/linux/linux-raspberrypi.inc +++
>> >> >>      b/recipes-kernel/linux/linux-raspberrypi.inc @@ -60,7 +60,7 @@
>> >> >>      do_rpiboot_mkimage() { if test "x${KERNEL_IMAGETYPE}" != "xuImage" ;
>> >> >>      then if test -n "${KERNEL_DEVICETREE}"; then # Add RPi bootloader
>> >> >>      trailer to kernel image to enable DeviceTree support -
>> >> >>      ${STAGING_DIR_NATIVE}/usr/lib/rpi-mkimage/mkknlimg --dtok
>> >> >>      ${KERNEL_OUTPUT} ${KERNEL_OUTPUT} +
>> >> >>      ${STAGING_LIBEXECDIR_NATIVE}/mkknlimg --dtok ${KERNEL_OUTPUT}
>> >> >>      ${KERNEL_OUTPUT} fi fi } -- 2.7.1
>> >> >>
>> >> >> Back alive. testing this.
>> >> >>
>> >> >> --
>> >> >> Andrei Gherzan
>> >> >>
>> >> >>
>> >> >>    Hi,
>> >> >>    I think we should use the older patchset from Maciej Borzecki (that
>> >> >>    solves this and some other minor stuffs) instead. It also keeps the
>> >> >>    rpi-mkimage structure:
>> >> >>    [3]https://lists.yoctoproject.org/pipermail/yocto/2016-February/028271.
>> >> >>    html
>> >> >>    BR Petter
>> >> >>
>> >> >>    Very good catch Petter. I'll test that PR.
>> >> >
>> >> > Merged the patch set sent by Maciej. Which covers this patch too.
>> >> > Thanks,
>> >> >
>> >> > --
>> >> > Andrei Gherzan
>> >>
>> >> Note that Maciej's patch doesn't really preserve the path on older
>> >> branches like you expect.
>> >>
>> >> On master branch:
>> >> ${libexecdir}/rpi-mkimage expands to /usr/libexec/rpi-mkimage
>> >>
>> >> On daisy branch:
>> >> ${libexecdir}/rpi-mkimage expands to /usr/lib/rpi-mkimage/rpi-mkimage
>> >
>> > This is a very good catch. But for now this patch is not in daisy it is only in
>> > master which is supposed to work with poky master. I do understand your
>> > concern. And I think it is the same thing on jethro too. So you propose relying
>> > on STAGING_LIBEXECDIR_NATIVE and let it append or not BPN?
>> >
>> > --
>> > Andrei Gherzan
>>
>> I propose my patch instead which has the following behavior and goes
>> along with the libexecdir change rather than trying to go against it.
>>
>> My patch has the following paths for mkknlimg.
>>
>> daisy:
>> [...]/sysroots/x86_64-linux/usr/lib/rpi-mkimage/mkknlimg
>> which matches the previous hardcoded path before the patch so the path
>> doesn't change with the patch for older branches
>>
>> master:
>> [...]/sysroots/x86_64-linux/usr/libexec/mkknlimg
>>
>> You can then do something similar for U-Boot.
>
> Indeed I agree and sounds good. One problem though - I merged the Maciej's
> patches.  Would you be able to patch current master and adapt your patch on it?
> If not I'll try to do it this weekend. Thanks a lot.
>
> --
> Andrei Gherzan

I have submitted a patchset against current master.

Regards,
Jonathan