[2/2] package/uboot-tools: env/script generation need BINARIES_DIR

From: Kalpesh Panchal <kalpesh.panchal@rockwellcollins.com>

The host build of uboot-tools can occur early in the build process and may
require the creation of BINARIES_DIR before generation of an enabled envimage
and/or boot script binary.

Signed-off-by: Kalpesh Panchal <kalpesh.panchal@rockwellcollins.com>
Signed-off-by: Matt Weber <matthew.weber@rockwellcollins.com>
---
 package/uboot-tools/uboot-tools.mk | 2 ++
 1 file changed, 2 insertions(+)

On Mon, 25 Jan 2021 08:57:42 -0600
Matt Weber <matthew.weber@rockwellcollins.com> wrote:

> From: Kalpesh Panchal <kalpesh.panchal@rockwellcollins.com>
> 
> The host build of uboot-tools can occur early in the build process and may
> require the creation of BINARIES_DIR before generation of an enabled envimage
> and/or boot script binary.
> 
> Signed-off-by: Kalpesh Panchal <kalpesh.panchal@rockwellcollins.com>
> Signed-off-by: Matt Weber <matthew.weber@rockwellcollins.com>

Hum, while this works, I'm in fact not too happy with the proposed
solution. I would prefer that we move the mkenvimage/$(MKIMAGE)
invocations to a BUILD_CMDS step, that produces its results in $(@D),
and then INSTALL_CMDS does a $(INSTALL) ... of that file, which with
the -D option would create $(BINARIES_DIR).

I know it's a bit bike-shedding, but it feels a little bit better. What
do you think ?

Thomas

Thomas/Yann,


On Mon, Jan 25, 2021 at 3:18 PM Thomas Petazzoni
<thomas.petazzoni@bootlin.com> wrote:
>
> On Mon, 25 Jan 2021 08:57:42 -0600
> Matt Weber <matthew.weber@rockwellcollins.com> wrote:
>
> > From: Kalpesh Panchal <kalpesh.panchal@rockwellcollins.com>
> >
> > The host build of uboot-tools can occur early in the build process and may
> > require the creation of BINARIES_DIR before generation of an enabled envimage
> > and/or boot script binary.
> >
> > Signed-off-by: Kalpesh Panchal <kalpesh.panchal@rockwellcollins.com>
> > Signed-off-by: Matt Weber <matthew.weber@rockwellcollins.com>
>
> Hum, while this works, I'm in fact not too happy with the proposed
> solution. I would prefer that we move the mkenvimage/$(MKIMAGE)
> invocations to a BUILD_CMDS step, that produces its results in $(@D),
> and then INSTALL_CMDS does a $(INSTALL) ... of that file, which with
> the -D option would create $(BINARIES_DIR).
>
> I know it's a bit bike-shedding, but it feels a little bit better. What
> do you think ?

I've added Yann to the discussion as he suggested the mkdir :-)

Whatever is better to maintain.

-Matt

Matthew, Thomas, All,

On 2021-01-25 15:22 -0600, Matthew Weber via buildroot spake thusly:
> On Mon, Jan 25, 2021 at 3:18 PM Thomas Petazzoni
> <thomas.petazzoni@bootlin.com> wrote:
> > On Mon, 25 Jan 2021 08:57:42 -0600
> > Matt Weber <matthew.weber@rockwellcollins.com> wrote:
> > > From: Kalpesh Panchal <kalpesh.panchal@rockwellcollins.com>
> > > The host build of uboot-tools can occur early in the build process and may
> > > require the creation of BINARIES_DIR before generation of an enabled envimage
> > > and/or boot script binary.
> > >
> > > Signed-off-by: Kalpesh Panchal <kalpesh.panchal@rockwellcollins.com>
> > > Signed-off-by: Matt Weber <matthew.weber@rockwellcollins.com>
> >
> > Hum, while this works, I'm in fact not too happy with the proposed
> > solution. I would prefer that we move the mkenvimage/$(MKIMAGE)
> > invocations to a BUILD_CMDS step, that produces its results in $(@D),
> > and then INSTALL_CMDS does a $(INSTALL) ... of that file, which with
> > the -D option would create $(BINARIES_DIR).
> >
> > I know it's a bit bike-shedding, but it feels a little bit better. What
> > do you think ?
> 
> I've added Yann to the discussion as he suggested the mkdir :-)

On principle, I do agree with Thomas.

However, I would point out that this is way wider than just the script
image. Indeed, we also have HOST_UBOOT_TOOLS_GENERATE_ENV_IMAGE which
is only called at install time.

If we wanted to mandate clean, separate build and install steps for the
script image generation, so would we want for the environment image too.

Which is a wider endeavour than just catering for the script image.

Of course, if one would want to actually tackle the issue and actually
separate the build and install steps, that would be awesome.

Otherwise, I am fine with just the mkdir.

Regards,
Yann E. MORIN.

On Mon, 25 Jan 2021 22:42:55 +0100
"Yann E. MORIN" <yann.morin.1998@free.fr> wrote:

> On principle, I do agree with Thomas.
> 
> However, I would point out that this is way wider than just the script
> image. Indeed, we also have HOST_UBOOT_TOOLS_GENERATE_ENV_IMAGE which
> is only called at install time.

Well, the patch is also adding a mkdir inside the code generating the
environment image.

Thomas

All,

On Mon, Jan 25, 2021 at 4:23 PM Thomas Petazzoni
<thomas.petazzoni@bootlin.com> wrote:
>
> On Mon, 25 Jan 2021 22:42:55 +0100
> "Yann E. MORIN" <yann.morin.1998@free.fr> wrote:
>
> > On principle, I do agree with Thomas.
> >
> > However, I would point out that this is way wider than just the script
> > image. Indeed, we also have HOST_UBOOT_TOOLS_GENERATE_ENV_IMAGE which
> > is only called at install time.
>
> Well, the patch is also adding a mkdir inside the code generating the
> environment image.

v2 will be out in a bit, i'll mark this one as superseded.

Regards,
Matt