| Summary: | C compiler cannot create executables | ||
|---|---|---|---|
| Product: | buildroot | Reporter: | outtierbert |
| Component: | Other | Assignee: | unassigned |
| Status: | RESOLVED INVALID | ||
| Severity: | normal | CC: | buildroot |
| Priority: | P5 | ||
| Version: | 2018.02.1 | ||
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Host: | Target: | ||
| Build: | |||
| Attachments: | file located in output/build/host-gcc-initial-6.4.0/build/arm-buildroot-linux-uclibcgnueabihf/libgcc/config.log | ||
|
Description
outtierbert
2018-05-23 14:44:42 UTC
First question: does this issue appear when doing a full rebuild from scratch, i.e "make clean; make" ? Yes, it does. I even completely removed the output folder once, just to be sure. OK. Could you try to provide a defconfig that exhibits the problem but does not use your own private BR2_EXTERNAL tree, so that we could try to reproduce the issue ? Since it's the toolchain that is failing to build, normally only those options would be needed to reproduce the problem: BR2_arm=y BR2_cortex_a8=y BR2_PACKAGE_HOST_LINUX_HEADERS_CUSTOM_4_9=y the other options should not affect the toolchain build, except if your patches on gcc/binutils in $(BR2_EXTERNAL_BERNARDO_PATH)/board/bernardo/patches. I suspect the problem must be in the external. The key is that config.log contains: CFLAGS='-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -c -Wall -O2 -c -Wall -O2' The -Os is from TARGET_OPTIMIZATION, then there is an empty TARGET_DEBUGGING, but the "-c -Wall -O2 -c -Wall -O2" is what causes the problem. Indeed, because of the -c, a conftest.o is generated while configure checks for a.out and conftest.exe etc. So Bert, please check why TARGET_CFLAGS is modified (and if this is indeed the issue). You were totally right Arnout. Thanks for the quick support guys! |