search

Traumflug / Teacup_Firmware

Firmware for RepRap and other 3D printers
http://forums.reprap.org/read.php?147
GNU General Public License v2.0
312 stars 199 forks source link

Compile error on Mint with Teacup Master Branch for ARM LCP1114 #320

Closed Willshell closed 5 years ago

Willshell commented 5 years ago

Good evening!

I'm trying to compile the firmware with the master branch for the ARM LCP1114. I have followed the wiki about the ARM toolchain setup, the config.h is:

include "config/board.gen7-arm.h"

include "config/printer.corexy.h" ( a mendel base tuned to COREXY)

Since the configtool is broken, the parameters were set by hands in the printer.corexy.h, the only thing that i have disabled is the watchdog (not implemented)

I receive this error:

~/Teacup_Firmware$ sudo make LINK build/teacup.elf lto1: fatal error: bytecode stream in file 'build/analog-stm32.o' generated with LTO version 5.2 instead of the expected 6.0 compilation terminated. lto-wrapper: fatal error: arm-none-eabi-gcc returned 1 exit status compilation terminated. /usr/bin/../lib/gcc/arm-none-eabi/7.3.1/../../../../arm-none-eabi/bin/ld: error: lto-wrapper failed collect2: error: ld returned 1 exit status Makefile-common:93: recipe for target 'build/teacup.elf' failed make: *** [build/teacup.elf] Error 1

Just to be sure, i have prepared a VM and a fresh install of MINT Linux. There i can compile and upload without issue, but the serial port is messed up. I can establish a connection fiddling around with the reset PIN, but on the serial port i receive START and nothing else.

I'm guessing about something broken with the compiler, right? If so, which compiler version i need to use? Maybe in the meantime i try on the windows machine.

Thanks for any help

Willshell commented 5 years ago

Finally, after many tries i did it, was the damn compiler.

I have dowloaded the arm-none-eabi version in the makefile (5_4-2016q3).

Now everything is working as it should.

Wurstnase commented 5 years ago

Looks like in your build directory were some files already. Make will check and rebuild only files which need to build again. When you build before some files already and change then the compiler you simply need to make clean before a make.

Willshell commented 5 years ago

Nope, was a compiler issue. The master branch can be compiled without any warning, but the code will not run on the CPU, no matter of what. Rolling back to an older version of the compiler and targeting the right directory on the makefile did the trick. I was able to compile teacup and make it run on an LPC1114FBD48/323 with extended RAM and PIO'S. Now I will try to figure out how to make the SPI work on this ARM CPU.

Willshell commented 5 years ago

Forgot to mention, that the broken binary was compiled and uploaded without issue. The serial port was responding for a short while after a cold restart, I have no idea what the CPU was doing. My guess is that something has changed in GCC o mbed and created a conflict somewhere, but due to the lack of my own experience on programming with microcontroller, I wasn't able to create a solution and update the code.

Wurstnase commented 5 years ago

I can read now 2 issues. First is two different compiler versions mixed.

The second should be fixed in the experimental. This is indeed a compiler bug with GCC 7.2.1 as far as I remember. With this bug, the weak symbols are not handled correctly in assembly code and can overwrite the non weak symbols. In the experimental branch I fixed this by linking the startup code before the rest. https://github.com/Traumflug/Teacup_Firmware/commit/0d308c8cfa90c7d1da8781fa25fc5e6ee69aed84