msp430: Add newlib support

[jnohlgard]

MSPGCC (the old compiler) is usually used with msp-libc, while the new upstream GCC msp430 port (sometimes referred to as the Red Hat MSP430 compiler) is usually accompanied by newlib.
The new msp430 port started with version 4.9.x, the latest release of the old mspgcc is 4.6.3 in the beginning of 2012 and doesn't look like it will receive any updates.

Untested on actual hardware since I don't have access to any msp430 boards.

[OlegHahm]

At least in terms of memory footprint it seems to be okay. (msp430-size is showing similar results when building default and gnrc_minimal with and without this PR.)

[jnohlgard]

@OlegHahm the changes with the old mspgcc 4.6 toolchain should be minimal, the big change is that it builds fine on msp430-elf-gcc-5.3.0 with newlib (though as I wrote in the top post I have not tested the resulting binary on real hardware)

[haukepetersen]

side question: when using the gcc msp430 compiler, would it be possible/make sense to use our own linkerscript for msp430 boards? This way we could get closer to the arm cortex boards and it would allow us to add our own sections (as @jfischer-phytec-iot proposed for auto-init).

[jnohlgard]

I guess so. There's like a bazillion ldscripts in the toolchain, but we
only support a small number of them so I guess it would be doable and even
beneficial in the long run to align them with the cortex platforms.

side question: when using the gcc msp430 compiler, would it be
possible/make sense to use our own linkerscript for msp430 boards? This way
we could get closer to the arm cortex boards and it would allow us to add
our own sections (as @jfischer-phytec-iot
https://github.com/jfischer-phytec-iot proposed for auto-init).

—
Reply to this email directly or view it on GitHub
#4766 (comment).

[haukepetersen]

nice

[jnohlgard]

@haukepetersen do you want to refactor the ldscripts?

[haukepetersen]

nope, not planning on doing so. It was more a general question, and I have the feeling that @kaspar030 and @jfischer-phytec-iot will need to do so to follow on their init/static array ideas...

[jnohlgard]

Something broke after rebasing on latest master, it won't compile right now because of missing symbols but I did not debug further

[Kijewski]

I love this hack!

[OlegHahm]

Any time estimate for this one?

[jnohlgard]

I won't have time to work on it, and I don't have any msp430 hardware to test on anyway so I would like to see someone adopt this.

[jnohlgard]

I have done some updates to rebase on latest master. Everything builds now on both the new and the old toolchain, except for the chronos board with the newlib toolchain, something about missing USART_1. I don't have the board and no desire to dig any further in this topic, so do we have any msp430 developer who would like to take over this PR?

[PeterKietzmann]

Probably @malosek :-) ?

[jnohlgard]

won't make it for release

[miri64]

Postponed due to feature freeze

[OlegHahm]

I will try to take the PR over if you agree, @gebart.

[jnohlgard]

Please, feel free to take over, I don't have any time or energy to work on this particular issue.

…

On Dec 22, 2016 23:02, "Oleg Hahm" ***@***.***> wrote: I will try to take the PR over if you agree, @gebart <https://github.com/gebart>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#4766 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AATYQmj2LffbXIII2fMtmbMv63Btcrvlks5rKvNygaJpZM4HVpOL> .

[jnohlgard]

postponed

[OlegHahm]

Sorry that there was no progress from my side until now. But it's still on my list.

[jnohlgard]

closed in favour of #6445