[U-Boot] [RFC] more features enabled on sandbox?

[U-Boot] [RFC] more features enabled on sandbox?

[AKASHI Takahiro]

Hi Tom, Simon,

I remember that Tom suggested me, in some UEFI-related discussion,
that we can and should enable as many features/configurations
on Sandbox as possible.
How should we do that?

1. modify a Kconfig;
        config FOO
          bool
          default y if SANDBOX
2. modify a sandbox*_defconfig
        CONFIG_FOO=y
3. take a mixed approach of (2) and (3) depending on a *type* of feature
4. add a customized build/test in ("matrix" of) Travis for a specific feature
5. or else?

For example,
CONFIG_FAT
CONFIG_FAT_WRITE
CONFIG_UEFI_SECURE_BOOT
  (which has bunch of dependencies.)

Those are NOT defined as optional features by UEFI specification,
but currently NOT selected by default, probably due to fears
against increase of code size. But this is not the case for Sandbox.

In addition,
CONFIG_CMD_FAT
CONFIG_CMD_EFIDEBUG
CONFIG_CMD_NVEDIT_EFI

Those commands are mandated in order to run my pytest for UEFI
secure boot. Again, they are NOT selected by default.
So even if sandbox build has no errors, UEFI secure boot test
will never be run due to configuration dependencies.

# This is NOT my patch-specific issue, as you know. There are
# bunch of pytests which are unexpectedly skipped in Travis CI.

What is your expectation here?

Thanks,
-Takahiro Akashi

[U-Boot] [RFC] more features enabled on sandbox?

[Tom Rini]

On Thu, Nov 21, 2019 at 12:06:53PM +0900, AKASHI Takahiro wrote:
> Hi Tom, Simon,
> 
> I remember that Tom suggested me, in some UEFI-related discussion,
> that we can and should enable as many features/configurations
> on Sandbox as possible.
> How should we do that?
> 
> 1. modify a Kconfig;
>         config FOO
>           bool
>           default y if SANDBOX
> 2. modify a sandbox*_defconfig
>         CONFIG_FOO=y
> 3. take a mixed approach of (2) and (3) depending on a *type* of feature
> 4. add a customized build/test in ("matrix" of) Travis for a specific feature
> 5. or else?

I believe that we should be enabling as much as possible in
sandbox_defconfig and sandbox64_defconfig directly.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20191120/b7a4f8ef/attachment.sig>