Hi again,

sorry for the noise but did some investigations and was able to 
reproduce....

On 2021-07-06 15:42, Peter Bergin wrote:

Hi,

On 2021-07-06 04:44, Anuj Mittal wrote:

quoted

Hi Peter,

On Sun, 2021-07-04 at 10:21 +0100, Richard Purdie wrote:

quoted

On Sun, 2021-07-04 at 10:59 +0200, Peter Bergin wrote:

quoted

On 2021-07-03 16:55, Mittal, Anuj wrote:

quoted

On Fri, 2021-07-02 at 13:59 +0200, Peter Bergin wrote:

quoted

This reverts commit 10aa1291979fb90bed1beb49be4d406ed0e1e4d5.

As there is no build dependency between libubootenv and the
configuration
of u-boot there is no reason to check for UBOOT_CONFIG or
UBOOT_MACHINE
by adding the class uboot-config. Revert this in order to
remove
useless
workaround in bsp layer (meta-freescale).

This was added to address failures in world builds:

https://lists.openembedded.org/g/openembedded-core/message/141757

If this is not required, then I think we'd need to address those
failures.

Thanks for the pointer about the reason for this patch. I searched
a bit
but could not find it.

Can you please help me define 'world builds'?

"bitbake world"

i.e. build everything

quoted

By reading the mail thread my thoughts are that the obvious thing
is
that the build has a bad configuration as u-boot is skipped. This
should
be the thing to solve.

Not every MACHINE will support u-boot or have a configuration for it.

Do you have a fix for this? This patch was applied and now the world
builds for MACHINEs not supporting u-boot are failing. Do you mind if I
send a revert of your patch?

I don't mind but I think it is the wrong thing to do, at least without 
analyzing the failing builds and get to the root of the problem.

My view on this, correct me if I'm wrong here. If you just want to 
build libubootenv you should be able to do it for any machine, with or 
without u-boot support as there is no build dependency between 
libubootenv and u-boot. If you have libubootenv in your image there 
will be a RRECOMMENDS to u-boot-env and in that case you must have 
NO_RECOMMENDATION=0 or a valid u-boot configuration. If the world 
build is failing I guess there is a configuration for a target without 
u-boot that includes libubootenv in an image. Or what is the exact 
failure? Can you send a reference with a log to the failing build?

/Peter

     $ bitbake libubootenv
     Loading cache: 100% 
|######################################################################################################| 
     Time: 0:00:00
     Loaded 3462 entries from dependency cache.
     NOTE: Resolving any missing task queue dependencies
     ERROR: Nothing RPROVIDES 'u-boot-default-env' (but 
meta/recipes-bsp/u-boot/libubootenv_0.3.1.bb RDEPENDS on or otherwise 
requires it)
     NOTE: Runtime target 'u-boot-default-env' is unbuildable, removing...
     Missing or unbuildable dependency chain was: ['u-boot-default-env']
     ERROR: Required build target 'libubootenv' has no buildable providers.
     Missing or unbuildable dependency chain was: ['libubootenv', 
'u-boot-default-env']

     Summary: There was 1 WARNING message shown.
     Summary: There were 2 ERROR messages shown, returning a non-zero 
exit code.

The setup above was to use a machine from (meta-intel) with grub as 
bootloader.

Then I can correct my earlier statement and a new wisdom that run-time 
dependecies are scanned already in the early phases. Apparently it is 
not possible to build libubootenv due to RRECOMMENDS to 
u-boot-default-env. The build for a 'non-u-boot config' will fail if we 
add 'inherit uboot-config' and it will fail without. What is the 
difference for a 'world' build?

/Peter