boards with side-effect: move CPU/CPU_MODEL definition to Makefile.features #12092
Conversation
jcarrano
added
Area: boards
Your computer seems to be much better than mine, running for 1 hour now.. hahaha |
|
@fjmolinas :D did you set the |
|
I get the expected no diff when running on the top commit: I can also see most differences where fixed when doing I also get the same results when normal parsing. |
fjmolinas
added
the
Reviewed: 3-testing
* CPU files should already have 'CPU' defined by the board. * Do not conditionally define CPU as it is not needed. This is part of cleanup prior to moving the CPU/CPU_MODEL to Makefile.features.
cpu/$(CPU)/Makefile.features and cpu/$(CPU)/Makefile.dep are automatically included Part of moving CPU/CPU_MODEL definition to Makefile.features to have it available before Makefile.include.
cpu/$(CPU)/Makefile.features and cpu/$(CPU)/Makefile.dep are automatically included Part of moving CPU/CPU_MODEL definition to Makefile.features to have it available before Makefile.include.
cpu/$(CPU)/Makefile.features and cpu/$(CPU)/Makefile.dep are automatically included Part of moving CPU/CPU_MODEL definition to Makefile.features to have it available before Makefile.include.
cpu/$(CPU)/Makefile.features and cpu/$(CPU)/Makefile.dep are automatically included Part of moving CPU/CPU_MODEL definition to Makefile.features to have it available before Makefile.include.
cpu/$(CPU)/Makefile.features and cpu/$(CPU)/Makefile.dep are automatically included Add the missing space when defining 'CPU_MODEL'. Part of moving CPU/CPU_MODEL definition to Makefile.features to have it available before Makefile.include.
cpu/$(CPU)/Makefile.features and cpu/$(CPU)/Makefile.dep are automatically included CPU_MODEL is currently kept uppercase. Making it lowercase is a future task. Part of moving CPU/CPU_MODEL definition to Makefile.features to have it available before Makefile.include.
cpu/$(CPU)/Makefile.features and cpu/$(CPU)/Makefile.dep are automatically included Part of moving CPU/CPU_MODEL definition to Makefile.features to have it available before Makefile.include.
cpu/$(CPU)/Makefile.features and cpu/$(CPU)/Makefile.dep are automatically included The multiple boards handling has been changed to use 'CPU_MODEL ?='. Part of moving CPU/CPU_MODEL definition to Makefile.features to have it available before Makefile.include.
cladmi
force-pushed
the
pr/cpu_model/minor_changes_included
branch
from
f3675b6 to
cdca4d8
Compare
|
Thank you for testing. I removed the commits from #12004 |
It took me much longer than I want to admit to get the testing procedure right hahahaha |
fjmolinas
added
Reviewed: 1-fundamentals
I can imagine. Running tests with 2 separate versions is also always hard to explain concisely :/ Any feedback on what I should put more clearly? |
|
Thank you for the review. |
Contribution description
This moves CPU and CPU_MODEL to Makefile.features for a selection of board that only cause harmless side-effects in the dependency parsing like
?=replaced by=and spacingReview procedure
I would advise to look at each commit individually and the global output comparison.
This PR modifications are, like the first PR
CPU,CPU_MODELto a different fileinclude $(RIOTCPU)/cpu_nameexport CPUandexport CPU_MODELas they are globally exported inRIOT/makefiles/vars.inc.mk
Lines 11 to 12 in 3e75383
And also the "side-effects":
includeorder when moved to the global include location which change order in the variables?=in espI can split in several PRs if wanted, it was just easier for me to run the validation once.
The dependency to #12004 is only for testing and could be merged without it.
Testing procedure
All the modified boards for this change are the following
So to run the comparison, it is easier to first limit the testing to these boards.
In the PR run
On the second commit: "squash! make: add targets to debug dependencies variables " run the same generation: (it is shown at last on github, but is not locally)
And generate the aggregated result
generate_aggregated.shDifference between normal and "global" dependency parsing
Both dependency parsing now have the same values for `FEATURES_PROVIDED`, `CPU`, `CPU_MODEL` for these boards
You can see that most differences have been fixed with this PR by checking the previous state with all
CPUandCPU_MODELthat were not defined ininfo-boards-supported.There are still currently differences in some variables, so they are excluded, but this is out of this PR scope and was the case before too.
The difference between before and now for normal parsing
As the normal parsing already had
CPUandCPU_MODEL, there are only difference inFEATURES_REQUIREDandFEATURES_PROVIDEDbut the actual value is unchanged as the value of the sorted ones is the same.Looking in the difference in `examples/default`
And when excluding
^FEATURES_REQUIREDand^FEATURES_PROVIDED, so keeping thesortedones we have the same values after migration.Issues/PRs references