On Wed, Feb 25, 2026 at 03:26:11PM +0100, Quentin Schulz wrote: > Hi Heinrich, > > On 2/25/26 7:57 AM, Heinrich Schuchardt wrote: > > Symbol CONFIG_SPL_DM_REGULATOR_FAN53555 is selected by > > SPL_DM_PMIC_FAN53555 and used in a Makefile. But the > > symbol definition is missing. > > > > Well it is "used" in Makefile simply because I wanted to be able to NOT > build it for xPL stages. I'm thinking we should use $(PHASE_) for most > symbols in Makefile, even if we don't have an xPL symbol in any Kconfig file > such that one would explicitly need to add an xPL symbol and enable it for > the feature to be enabled in the xPL stage.
We should, generally, use $(PHASE_) in Makefiles. But it's not a simple conversion because there are cases where SPL depends on CONFIG_DRIVER=y to build a needed driver. This was part of one of the long threads Simon and I had last year (or the year before?) where at one point I lamented that Yamada-san was right and configuring the build the way I was intended would lead to these kind of problems down the line. > Maybe we even need a PROPER_ prefix for PHASE_ when not in xPL for example > for things that make sense to be stage-specific. There are some symbols that > should represent something enabled across stages (e.g. the CPU architecture) > where the PROPER_ prefix wouldn't make sense. I'm often confused as to what > is a symbol that must apply to all stages by design or simply because > someone didn't bother to implement separate xPL switches for that (yet). Yes, this can be confusing. It would likely be less confusing if we at least converted things to be consistent about it in stages. -- Tom
signature.asc
Description: PGP signature
