[embarrassement] I managed to send this to a bogus address on
Sunday night.  So, second try, and an update.

 First, 64-bit flex failed when my script tried to 'touch
doc/flex.1' - there was a '21933 Killed' message.  Fortunately, that
instruction is no longer in the book and flex built fine without it.

 Then I got to bash in the final system:  That fails as soon as make
begins:

config.status: creating po/Makefile
config.status: executing default commands
bison -y -d ./parse.y
conflicts: 1 shift/reduce
touch parser-built
make: *** [y.tab.c] Killed              # <<<<<<<<<
make: *** Deleting file `y.tab.c'

 I don't see anything odd in my script, but by that point I was
feeling uneasy about the whole build.

 After I'd mis-sent the original mail, I decided to plough on by not
building bash.  The testsuite for module-init-tools failed with a
'Killed', everything else was ok, but since it couldn't build bash,
I trashed it. [ I also tried compiling a static bash on the 64-bit
host, then rebuilding coreutils and bison, but no different. ]

 Now, I'm trying a fresh build with gcc-4.2.0 (and without the
extras for bash's config.cache in the temporary system).  This has
managed to rebuild coreutils for 'df'.  (Yes, I know I should use
what Joe pointed out, but I'd like to get to the bottom of the
problem first), also glibc in the final system, and is currently
bootstrapping gcc.  At the moment, I'm wondering if 4.2.1 miscompiles
glibc on x86_64.  If the base system builds, I want to update my
multilib desktop, so it could all end in tears based on the bug
report Joe pointed to.

ĸen
-- 
das eine Mal als Tragödie, das andere Mal als Farce
_______________________________________________
Clfs-dev mailing list
[email protected]
http://lists.cross-lfs.org/cgi-bin/mailman/listinfo/clfs-dev

Reply via email to