topic Debug version of newlib in MCUXpresso General

Debug version of newlib

davenadler — Tue, 20 Oct 2020 15:31:59 GMT

Using latest MCUXpresso for an LPC55S69, need to debug a possible newlib problem.
For now we'll concentrate on variant full newlib (not nano), none.
Thus a few questions for NXP:

have NXP published the debug version of newlib (if so, where, if not, why not)?
where is the build script NXP have used for newlib?
have NXP made any modifications to newlib, or used unchanged version 3.1.0?

Thanks!
Best Regards, Dave
PS: @v_canoz - This is regarding your concurrency issue...

Re: Debug version of newlib

converse — Tue, 20 Oct 2020 16:08:21 GMT

NXP take their GNU Tools releases (unchanged) directly from the ARM GNU Tools project. Everything you need can be found there:

https://developer.arm.com/tools-and-software/open-source-software/developer-tools/gnu-toolchain/gnu-rm

If you think there is a bug, search the GNU newlib forum

Re: Debug version of newlib

davenadler — Wed, 21 Oct 2020 00:49:59 GMT

Thanks @converse, good to know NXP did not make mods.
But I did not find any newlib debug builds on the ARM site.
Do you know where I can find debug version(s) of newlib, which is my main question?
Thanks,
Best Regards, Dave

Re: Debug version of newlib

ErichStyger — Wed, 21 Oct 2020 06:13:59 GMT

Hi Dave,

to my knowledge there is no pre-built debug version of the library. What I did in the past (but not recently) was building it myself:

https://mcuoneclipse.com/2014/08/23/gnu-libs-with-debug-information-rebuilding-the-gnu-arm-libraries/

you will be better better off building it on Linux.

Erich

Re: Debug version of newlib

davenadler — Wed, 21 Oct 2020 14:22:02 GMT

Hi @ErichStyger - I saw your excellent (as always) page, where I could see building newlib is a PITA. Perhaps I'll spin up a Linux VM to build it, but that always wastes a few hours too unfortunately. @v_canoz has a concurrency problem, which looks like a real platform issue (and not related to free storage). Seems really unlikely but might be GCC runtime as built for ARM or FreeRTOS. He's come up with a trivial test case that crashes despite plenty of stack space and no obvious configuration settings error. I checked all the usual suspects. More likely MCUXpresso build setting wrong somehow, but I don't see it yet.