https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87048
--- Comment #4 from Christophe Lyon <clyon at gcc dot gnu.org> --- Looks like there was a misunderstanding, I was probably not clear. r263082 actually removed the regression I reported, because that commit reverts the offending one. So current trunk is OK. I'm looking again at the traces, and the different function names noticed by Thomas seem harmless (it seems to me that the different symbols refer to the same functions, eg _uname is weak alias for uname, etc...)