compilation will stop complaining about uname.o being unrecognized: .... snippet:..... i586-pc-linux-gnu-gcc -std=gnu99 -march=i586 -m3dnow -mmmx -Os -pipe -lrt -Wl,--as-needed -o who who.o ../lib/libcoreutils.a ../lib/libcoreutils.a uname.o: file not recognized: File format not recognized collect2: ld returned 1 exit status make[2]: *** [uname] Error 1 make[2]: *** Waiting for unfinished jobs.... who.o: file not recognized: File format not recognized collect2: ld returned 1 exit status make[2]: *** [who] Error 1 mv -f .deps/setuidgid.Tpo .deps/setuidgid.Po mv -f .deps/su.Tpo .deps/su.Po ../../lib/libcoreutils.a(closeout.o): In function `close_stdout': closeout.c:(.text+0x34): undefined reference to `quotearg_colon' ../
Created attachment 155283 [details] Portage Log for coreutils
Created attachment 155289 [details] ebuild environment
Created attachment 155291 [details] Uname.o file which seems to be of an unknown format MC view (F3) will pretend: MPEG sequence, v1, system multiplex
Created attachment 155293 [details] emerge --info
does it fail at the same place every time ? what if you try MAKEOPTS=-j1 ?
Thanks for the suggestion. I tried export MAKEOPTS=" -jl " && emerge coreutils, the result is the same. And yes, it always fails at the same place.
go into the directory, remove uname.o, and run the compile line manually. see if you get the same result. also see if you get the same result if you set the output to like /tmp/foo.o. we've seen disk corruption cause weird things like this (so forcing a fsck run on your disks may be a good idea)
removing uname.o didn't help, neither did fsck, but wiping /tmp completely did! Might have been a ccache issue. Thanks for your help, and sorry for bothering you.