cpu: refactor cortex cpu makefiles#2850
Conversation
kaspar030
added
the
Type: cleanup
|
(did not touch the kinetis MCUs as they have a different |
There was a problem hiding this comment.
why do you declare newlib as pseudomodule?
There was a problem hiding this comment.
Because syscalls.c has to be in front of the linker command line, and the newlib module's Makefile.include adds it there (through UNDEF). So in order not to add newlib.a again, I've added it to PSEUDOMODULES.
There was a problem hiding this comment.
Does it break anything if the .a file appears twice in the command line?
There was a problem hiding this comment.
Don't think so.
I did this while debugging why this commit enlarges the binary (turned out to be the RIOT_VERSION string), but then I thought, it is not needed there, so why add it there.
You guys opt for not making this a pseudomodule?
There was a problem hiding this comment.
I actually don't care.
There was a problem hiding this comment.
I just asked out of interest. :) Make it a pseudomodule.
|
nice one! Will test some of the boards this afternoon |
|
Maybe testing once for #2852 is enough, if you guys ACK the principle? |
|
All testing is done on #2852, so I close this. |
4 participants
Makefile, Makefile.include and syscalls.c were nearly identical for all cortex CPUs.
This PR unifies the Makefiles and factors newlib support into a module.
This affects the following boards, could you please test and then mark here?