Danek Duvall wrote: [CC:'ing ksh93-integration-discuss at opensolaris.org] > I'm currently running into three bugs building SFW on opensolaris: > > 6817601 builds broken on opensolaris due to empty default $PATH > 6859442 ksh93 dumps core during fftw2 build in SFW > 6859444 ksh93 hang in mysql build in SFW
Can you check whether the binaries from http://www.opensolaris.org/os/project/ksh93-integration/downloads/2009-07-02/ fix the last two bugs (CR #6859442 and CR #6859444, CR #6817601 is a problem caused by the lack of setting PATH (and ksh93 conforms to the POSIX standard and won't invent a value like bash3)), please ? > I have a fix for 6817601, and a workaround for 6859444. I suspect that > adding --disable-fortran to the fftw configure invocation will take care > of 6859442, though I haven't tried it yet. I did some basic testing and CR #6859442 ("ksh93 dumps core during fftw2 build in SFW") only happens for the 32bit version of ksh93 but not for the 64bit one (e.g. the workaround would be to force SHELL=/usr/bin/ksh93 CONFIG_SHELL=/usr/bin/ksh93 for 64bit systems (unfortunately it does not help for 32bit-only platforms/machines)) BTW: Steps to reproduce CR #6859442 ("ksh93 dumps core during fftw2 build in SFW") on SPARC: -- snip -- $ wget 'http://src.opensolaris.org/source/raw/sfw/usr/src/lib/fftw2/fftw-2.1.5.tar.gz' $ ( gzcat <fftw-2.1.5.tar.gz | tar -xf - ; cd fftw-2.1.5 ; cat configure | sed 's/\/bin\/sh/\/usr\/bin\/sparcv7\/ksh93/g' >configure_mod ; (LC_ALL=C SHELL=/usr/bin/sparcv7/ksh93 CONFIG_SHELL=/usr/bin/sparcv7/ksh93 VMDEBUG=a /usr/bin/sparcv7/ksh93 configure_mod --no-reexec --prefix=/usr --enable-threads --enable-shared --disable-static) 2>&1 | tee buildlog.log) -- snip -- Problem is that I do not understand why this crashes only for 32bit (and $ dbx -check access ... # is broken on Solaris 11/B110, leaving no quick way to check for the "usual suspects" of bugs) and only for a fraction of the calls (unless you add VMDEBUG=a (which enables libumem-like allocator checks in libast (which makes it "smell" like a dangling pointer used after deallocation))). I need help with this bug (either via finding someone with a working version of $ dbx -check access ... # (e.g. run $ bcheck -access # against the shell running "configure")) ... ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) roland.mainz at nrubsig.org \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 3992797 (;O/ \/ \O;)
