can't buildworld - boot2

[brian at aljex.com]

Quoting brian at aljex.com:

> Quoting Ryan Sommers <ryans at gamersimpact.com>:
>
> > brian at aljex.com said:
> > > kernel: ver=1.01 size=780 load=9000 entry=9010 map=16M pgctl=1:1
> > > client: fmt=bin size=1779 text=0 data=0 bss=0 entry=0
> > > output: fmt=bin size=200d text=114 data=1ef9 org=0 entry=0
> > > -525 bytes available
> > > *** Error code 1
> > >
> > > Stop in /usr/src/sys/boot/i386/boot2.
> > > *** Error code 1
> >
> > Are you building with a -j option? There are no errors that I can see in
> > the output you provided. Using a -j option can usually be a symptom of
> > this kind of failure where the real errors end up being buried in the
> > output further up.
> >
> > Try building without -j and then see where the errors are.
> >
> > --
> > Ryan Sommers
> > ryans at gamersimpact.com
> >
> >
>
> no -j
>
> just "make buildworld"
>
> -bash-2.05b# uname -a
> FreeBSD venti 6.0-CURRENT FreeBSD 6.0-CURRENT #13: Tue Oct 19 06:56:35 EDT
> 2004
>    root at venti:/usr/obj/usr/src/sys/VENTI  i386
> -bash-2.05b# whereis cc
> cc: /usr/bin/cc /usr/share/man/man1/cc.1.gz /usr/src/gnu/usr.bin/cc
> -bash-2.05b# ls -l /usr/bin/cc
> -r-xr-xr-x  2 root  wheel  171304 Oct 19 04:37 /usr/bin/cc
> -bash-2.05b# cc --version
> cc (GCC) 3.4.2 [FreeBSD] 20040728
> ...
>
> (ie: the kernel and gcc are in fact the 10-19 build)
>
> -bash-2.05b# rm -rf /usr/src /usr/obj
> -bash-2.05b# cvsup all.sup
> ...
> -bash-2.05b# cd /usr/src
> -bash-2.05b# make buildworld 2&1 >/buildworld.txt
> ...
> (error shown previously)
>
> where all.sup has:
> -------
> *default host=ftp4.us.FreeBSD.org
> *default base=/usr
> *default prefix=/usr
> *default release=cvs tag=.
> *default delete use-rel-suffix
> *default compress
> src-all
> ports-all
> -------
>
> make.conf has:
> -------
> #COPTS = -O -march=pentium4
> CPUTYPE = pentium4
> COPTFLAGS = -O
> #CFLAGS = -O -msse2 -mfpmath-sse
> CFLAGS = -O
> NOPROFILE = true
> -------
>
>
>
> I'm used to using ksh on sco and as a result the command above didn't capture
> stderr so I'll do it again and repost. Just for, no particular reason I
> guess,
> I'll try from the default csh instead of bash this time too.
>
> thanks

deleted /usr/src and /usr/obj again
used vipw to change roots shell back to /bin/csh
logged out & back in
cvsupped and make buildworld worked fine (with the -O2 -fno-strict-aliasing btw,
I don't know if that's been implimented yet, I put it in make.conf)

possible clue: you mentioned -j
I had tried a -j4 , which was the first time I ever tried -j anything btw,
previously and it failed, but it didn't fail anywhere near the spot I showed.
After it failed I shrugged and thought "not surprising" and did a make clean
and make buildworld and got the failure in boot2 I originally posted about, and
continued to get the same failure no matter what I did over several complete
delete/checkout/build attempts even though I never used -j again.

Silly as it seems, the only common factor I see was that all those attempts,
spread over 3 or so days, took place in the same login session, the same
instance of bash, as the original make -j4 failure.

Either the initial failure screwed up the instance of bash and all I needed to
do was log out & in, or the use of bash vs csh is no good, or some change was
committed in the last few hours that fixed it, or -O2 -fno-strict-aliasing
works where -O doesn't.

The kernel just successfully built too while I was writing this.

I'll try to duplicate it just so it becomes a known repeatable thing instead of
a passing fluke that no one knows why it happened or if there is anything to be
aware of.

Thanks.


----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.