[RFC,PATCH 0/7 v2] Introduce firmware features

[Colin Ian King]

On 28/04/14 02:35, Jeremy Kerr wrote:
> I've been playing with FWTS on a non-ACPI machine, and got a lot of
> aborts due to there not being various ACPI-specific bits of
> functionality present.
> 
> It seems appropriate to define a new type for this firmware, but I don't
> want to add type checks all over the existing fwts tests.
> 
> Rather than checking the type of firmware (and only running ACPI tests
> when we know that the firmware is of type UEFI or BIOS), I'd like to use
> more-flexible features instead; this way, each test can check if
> required features are available, rather than assuming that $type
> firmware has $feature_set.
> 
> Currently, we just key the feature set off the firmware type, but we
> could do something more flexible in future. I've only defined an ACPI
> feature for now.
> 
> Patches 5 and onwards are a basic usage of these features; we skip the
> batch tests that require ACPI if the firmware doesn't have the
> FWTS_FW_FEATURE_ACPI feature. Then, we introduce a firmware type with a
> different feature-set.
> 
> Cheers,
> 
> 
> Jeremy
> 
> v2: 
>  - Put a feature bitmask in the test defintion, rather than relying on
>    ops.init callbacks to do feature detection.
> 
>  - Add example usage with new firmware & feature types
> 
> ---
> Jeremy Kerr (7):
>       fwts: Add fwts_firmware_has_features
>       fwts: Only run firmware detection once
>       fwts: Allow tests to be conditional on available features
>       fwts: Print names of missing features, rather than a cryptic bitmask
>       acpi: only run ACPI tests if we have ACPI
>       fwts: Add FWTS_FW_FEATURE_DEVICETREE
>       fwts: Add OPAL firmware type
> 
> 
I'm happy with this concept. My comments on the patches follow..