On Sat, Feb 04, 2012 at 12:54:58PM -0700, Warren Block wrote: > On Sat, 4 Feb 2012, Jason Hellenthal wrote: > > > On Sat, Feb 04, 2012 at 10:48:16AM -0800, Jeremy Chadwick wrote: > >> On Sat, Feb 04, 2012 at 10:57:40AM -0700, Warren Block wrote: > >>> Trying to build 8-stable i386 for the last couple of days gives this > >>> error when building the kernel: > >>> > >>> /usr/local/libexec/ccache/world/cc -c -O -pipe -march=prescott -std=c99 > >>> -g -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes > >>> -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -Wundef > >>> -Wno-pointer-sign -fformat-extensions -nostdinc -I. -I/usr/src/sys > >>> -I/usr/src/sys/contrib/altq -D_KERNEL -DHAVE_KERNEL_OPTION_HEADERS > >>> -include opt_global.h -fno-common -finline-limit=8000 --param > >>> inline-unit-growth=100 --param large-function-growth=1000 > >>> -mno-align-long-strings -mpreferred-stack-boundary=2 -mno-mmx -mno-3dnow > >>> -mno-sse -mno-sse2 -mno-sse3 -ffreestanding -fstack-protector -Werror > >>> vers.c > >>> linking kernel.debug > >>> ld: kernel.debug: Not enough room for program headers (allocated 5, need > >>> 6) > >>> ld: final link failed: Bad value > >>> *** Error code 1 > >>> 1 error > >>> *** Error code 2 > >>> 1 error > >>> *** Error code 2 > >>> 1 error > >> > >> The freebsd-stable mailing list has been filled with tinderbox failures > >> for this exact problem with the LINT kernel, and it appears to be > >> specific to RELENG_8 on i386. > >> > >> http://lists.freebsd.org/pipermail/freebsd-stable/2012-February/thread.html > >> > >> Start from the bottom and search upwards for the string: > >> "[releng_8 tinderbox] failure on i386/i386" > >> > >> Point being: the problem is not you. > >> > > > > This is why I don't like ccache. Technically the problem is there but > > also coexists within the source itself. For whatever reason I would > > believe a cached copy of already built source is having some collisions. > > As already stated "NO CCACHE" the kernel can be built with what I > > posted on three different machines. > > But it still does (did) not build here with NOCCACHE set, so it's not > a ccache problem.
I have seen ccache before run anyway even though .if statements are within make.conf to prevent it. It was something to do with make and friends built from world with ccache and after it would continue to use ccache regardless. This can be verified by watching the cache hits and misses during a compile with NOCCACHE set. Its tough to revert from this but involves removing ccache/distcc from the system. -- ;s =;
pgpIOevQmhCmD.pgp
Description: PGP signature
