donnie@supernova ~ $ splat nautilus
Emerged at: Sun Sep 19 21:03:46 2004
Build time: 2 hours, 9 minutes, and 43 seconds
This is a dual Athlon 2000 system. A build of the same on an Athlon 1800 (single) took less than 30 minutes.
I'm not sure whether this is a toolchain bug, gnome bug or neither.
Created attachment 39991 [details]
Unfortunately the log was 13MB uncompressed, so I figured bzipping it down to
50KB would work out better.
Created attachment 39992 [details]
Created attachment 39993 [details]
well i guess this in the log explains the extra time it took :
make: Warning: File `/usr/bin/orbit-idl-2' has modification time 6.8e+03 s in the future
i've never seen it before. I'd say your machine, guess something todo with SMP. That might be caused by toolchain bugs, they're doing crazy stuff there afaic.
I'm not sure what to do with this bug, i'd guess you are better of compiling -j1.
I was also wondering whether the MCE notices in dmesg were related, but I don't know enough about Machine Check Exception to tell.
I'll give it another try with -j1 tonight, try to reproduce/not reproduce, etc.
I ran into the same thing on mips building 2.6.3-r1 (o32 userland). Switching MAKEOPTS to -j1 fixed this. It seems like it has to be a toolchain bug, as I encountered something *very* similar with cc1plus running the cpu up to 100% indefinitely while building gcc 3.4.2-r2 with fortran support. On that machine, the system compiler was/is 3.4.1-r2/3.4.2-r2. Perhaps the toolchain herd should be added to this bug if it can be confirmed. Donnie, could you try to build gcc-3.4.2-r2 with USE="f77" and MAKEOPTS set to -j2 and see if you can reproduce my problem there?
I forgot to mention, the build machine is uniprocessor.
fun fun happy happy toolchain oddities!
Anyone on toolchain got an idea w/ this? I never hit geoman's reported fortran oddity on my O2, but then again, I used -j1 on that system.
That modification time is 10 minutes in the future... I really doubt this is
toolchain related, and I can't trigger this bug... =/
I'm gonna close it cantfix unless someone really cares to delve into it...