Hi, I'm trying to build gcc 5.4.0 together with glibc 2.26. It is a cross-build to both AMD64 or i686 that shows problems, while ARM works. (For those who want to suggest a newer gcc first: no, doesn't work: https://gcc.gnu.org/bugzilla/ show_bug.cgi?id=70936). The problem arises from this output of configure of libgomp: checking for ISO C99 support in <math.h>... no This has always been "yes" before. This leads to I-don't-know-which-changes, which leads to std::to_string not being available (and probably some related problems), which afterwards will lead to gdb 8.0 not being able to be (cross-)compiled. My initial idea was this: echo 'glibcxx_cv_c99_math=yes' >> ${CONFIG_SITE} export CONFIG_SITE Which would be ok, if not most sub-configures are called with CONFIG_SITE=no-such-file Why is this? I have just removed all of them an the only difference I was able to observe is that now the CONFIG_SITE workaround work, while all other results are the same. Looking in the git history this seems to be the case forever, but I would propose for this to be reconsidered. Greetings, Eike -- Rolf Eike Beer, emlix GmbH, http://www.emlix.com Fon +49 551 30664-0, Fax +49 551 30664-11 Bertha-von-Suttner-Str. 9, 37085 Göttingen, Germany Sitz der Gesellschaft: Göttingen, Amtsgericht Göttingen HR B 3160 Geschäftsführung: Heike Jordan, Dr. Uwe Kracke – Ust-IdNr.: DE 205 198 055 emlix – smart embedded open source
Attachment:
signature.asc
Description: This is a digitally signed message part.
