Re: NSLU2 running 5.0_RC not stable

[Donald T Hayford <don%donhayford.com@localhost>]

Donald T Hayford wrote:
> Luazi wrote:
> > On Sat, Mar 7, 2009 at 7:16 PM, Donald T Hayford <don%donhayford.com@localhost> 
> > wrote:
> >  
> > > When I boot a version I built in August 2008, but using the same world
> > > (i.e., boot up a different kernel but use the same hard drive as 
> > > root, thus
> > > the same md5 program), I get the right answer.  Or at least the same 
> > > answer
> > > consistently.
> > >     
> >
> > Interesting. I have just finished rebuilding my 4.99.70 tree and it 
> > is broken.
> >
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = bf4a4904e1b1a03e1bb21cb4be542274
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = a1f8849ef4548c2982b15935fbdd3f37
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = bf4a4904e1b1a03e1bb21cb4be542274
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = bf4a4904e1b1a03e1bb21cb4be542274
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = bf4a4904e1b1a03e1bb21cb4be542274
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = bf4a4904e1b1a03e1bb21cb4be542274
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = 72e0f6214aa74040d3e53f2b51b9e08f
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = bf4a4904e1b1a03e1bb21cb4be542274
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = bf4a4904e1b1a03e1bb21cb4be542274
> > slug# md5 pkgsrc.tar.bz2
> > MD5 (pkgsrc.tar.bz2) = f503e04aa9b1ded9d602f319da688e5c
> >
> >   
> I don't particularly like this test, since it isn't very definitive, 
> but unfortunately I don't have a better one.  Does anybody know if 
> there is something in the new ATF that might help here?
>
> I dug back through some old kernels.  My build of 4.99.72 (about 
> 8/20/2008) failed a test similar to that above twice in 27 attempts.
> My build of 4.99.55 (about 5/5/2008) failed zero times out of 50.  
> But, again, this isn't a very good test.
I built a kernel/world from -current for the NSLU2 on March 14, 2009, 
which went fine except for a small problem with flist (the file 
src/distrib/sets/list/base/ad.armeb is out-of-date):

NetBSD netbsd. 5.99.8 NetBSD 5.99.8 (NSLU2_ALL) #0: Sat Mar 14 13:14:23 
EDT 2009  
hayford@fed10:/home/hayford/netbsd-20090314/obj-armeb/sys/arch/evbarm/compile/NSLU2_ALL 
evbarm

The kernel fails the md5 test virtually 100% of the time.  I ran the atf 
test suite with the following failed results:

bash-3.2# cat testsuite.csv|grep fail
tc, atf/atf-c/t_io, write_fmt_fail, passed
tc, atf/atf-c++/t_macros, fail, passed
tc, atf/tools/t_atf_config, arch, failed, evbarm != armeb (evbarm != armeb)
tp, atf/tools/t_atf_config, failed
tc, atf/tools/t_atf_cleanup, mount, failed, atf-check failed; see the 
output of the test for details
tc, atf/tools/t_atf_cleanup, symlink, failed, atf-check failed; see the 
output of the test for details
tp, atf/tools/t_atf_cleanup, failed
tc, atf/data/t_pkg_config, c_version, failed, pkg-config could not 
locate atf-c.
pc; maybe need to set PKG_CONFIG_PATH?
tc, atf/data/t_pkg_config, c_build, failed, pkg-config could not locate 
atf-c.pc; maybe need to set PKG_CONFIG_PATH?
tc, atf/data/t_pkg_config, cxx_version, failed, pkg-config could not 
locate atf-c++.pc; maybe need to set PKG_CONFIG_PATH?
tc, atf/data/t_pkg_config, cxx_build, failed, pkg-config could not 
locate atf-c++.pc; maybe need to set PKG_CONFIG_PATH?
tp, atf/data/t_pkg_config, failed
tc, ipf/t_ipf, bpf1, failed, atf-check failed; see the output of the 
test for details
tc, ipf/t_ipf, bpf_f1, failed, Test case did not report any status; 
bogus test
tp, ipf/t_ipf, failed
tc, kernel/t_umount, umount, failed, atf-check failed; see the output of 
the test for details
tp, kernel/t_umount, failed
tc, modules/t_modctl, cmd_load, failed, Module load failed
tc, modules/t_modctl, cmd_load_props, failed, Module load failed
tc, modules/t_modctl, cmd_stat, failed, Module load failed
tc, modules/t_modctl, cmd_unload, failed, Module load failed
tp, modules/t_modctl, failed
tc, modules/t_modload, plain, failed, atf-check failed; see the output 
of the test for details
tc, modules/t_modload, iflag, failed, atf-check failed; see the output 
of the test for details
tc, modules/t_modload, sflag, failed, atf-check failed; see the output 
of the test for details
tp, modules/t_modload, failed
tc, util/df/t_df, normal, failed, atf-check failed; see the output of 
the test for details
tp, util/df/t_df, failed
tc, util/sh/t_expand, strip, failed, $stripped != ${line%%/\**} (#define 
bindir"/usr/bin"  != #define bindir ")
tp, util/sh/t_expand, failed
tc, util/sh/t_set_e, all, failed, 30 of 77 failed cases
tp, util/sh/t_set_e, failed

Nothing jumped out at me, bu I don't know enough about the tests to 
decide if any of these are important, so I'll just pass them on for now.

Regards, Don