boards: move CPU/CPU_MODEL definition to Makefile.features#12014
Conversation
|
I ran the validation script on a previous version of the branch as I needed to rebase it at the end to |
|
I'm running the script and getting tons of in standard output. Edit: it only seems to happen when evaluating |
This is an issue with the And for comparing the dependency resolution, I call it on all boards (that are not blacklisted).
|
|
I compared again the output with the last versions of this PR and I get the same output as the testing description. |
|
Squash please. |
|
I will remove the test script as it is part of #12004 |
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.
e391308 to
636285e
Compare
|
Thank you for the review. Now I need to tackle the complex ones. |
Contribution description
This moves CPU and CPU_MODEL to Makefile.features for a selection of board that currently cause no difference at all in the dependency parsing.
This was accepted in "Tracking: move CPU/CPU_MODEL to Makefile.features" #11477 and more precisely by this PR "Makefile.features: prerequisites for moving CPU/CPU_MODEL to boards/Makefile.features" #11478
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.
Review procedure
This PR modifications are only
CPU,CPU_MODELto a different fileinclude $(RIOTCPU)/cpu_name/...export CPUandexport CPU_MODELas they are globally exported inRIOT/makefiles/vars.inc.mk
Lines 11 to 12 in 3e75383
The testing procedure dumps all variables used for dependency parsing and compare that the new value is consistent.
The dependency to #12004 is only for testing and could be merged without it.
Testing procedure
Review of the new state
In the PR run
Then run the aggregation script:
generate_aggregated.shNow the value of
CPUandCPU_MODELinfo-boards-supportedand the normal dependency parsing are the same except for boards that are not migrated by this PR:Difference in `CPU/CPU_MODEL/BOARD`
Note that currently there are still differences in both parsing with
Comparison with the previous state
On the commit:
"squash! make: add targets to debug dependencies variables "run the same generation:It did not change the variables used when normal dependency parsing:
And in general nothing changed except
CPUandCPU_MODELvalues being now declared ininfo-boards-supported:I need to exclude the aggregated file as diff reports a few lines with
CPU/CPU_MODELeven if I asked to ignore them.For reference without excluding the aggregated files
$ diff -r -W 80 -y --suppress-common-lines build/deps/ref/ build/deps/cpu_model/ -I 'CPU ' -I 'CPU_MODEL ' diff -r -W 80 -y --suppress-common-lines -I 'CPU ' -I 'CPU_MODEL ' build/deps/ref/deps_info-boards-supported build/deps/cpu_model/deps_info-boards-supported BOARD = opencm904 | BOARD = opencm904 CPU = | CPU = stm32f1 CPU_MODEL = | CPU_MODEL = stm32f103cb BOARD = nz32-sc151 | BOARD = nz32-sc151 CPU = | CPU = stm32l1 CPU_MODEL = | CPU_MODEL = stm32l151rc BOARD = pba-d-01-kw2x | BOARD = pba-d-01-kw2x CPU = | CPU = kinetis CPU_MODEL = | CPU_MODEL = mkw21d256vha5 BOARD = slwstk6220a | BOARD = slwstk6220a CPU = | CPU = ezr32wg CPU_MODEL = | CPU_MODEL = ezr32wg330f256r60 BOARD = stm32f0discovery | BOARD = stm32f0discovery CPU = | CPU = stm32f0 CPU_MODEL = | CPU_MODEL = stm32f051r8 BOARD = saml21-xpro | BOARD = saml21-xpro CPU = | CPU = saml21 CPU_MODEL = | CPU_MODEL = saml21j18a BOARD = remote-reva | BOARD = remote-reva CPU = | CPU = cc2538 CPU_MODEL = | CPU_MODEL = cc2538sf53Issues/PRs references