On Thu, Dec 29, 2011 at 04:43:30PM -0600, Bruce Dubbs wrote: > Ken Moffat wrote: > > On Thu, Dec 29, 2011 at 03:56:39PM -0600, Bruce Dubbs wrote: > >> I think that idea is fine, but wonder about the use of awk. It may not > >> be available. Perhaps: > >> > >> ldd /bin/ls | grep '/libc.*so' > >> > >> I don't think stripping the cruft is really necessary. > >> > >> LFS 32-bit: libc.so.6 => /lib/libc.so.6 (0xb7e46000) > >> Ubuntu : libc.so.6 => /lib/tls/i686/cmov/libc.so.6 (0xb7e24000) > >> LFS 64-bit: libc.so.6 => /lib/libc.so.6 (0x00007fa7f6920000) > >> RHEL : libc.so.6 => /lib/tls/libc.so.6 (0x0025e000) > >> > >> -- Bruce > > > > In LFS we already require that /usr/bin/awk should exist. In > > version-check.sh it's 2 lines before we test the version of libc. > > Dunno if jhalfs differs, but it seems a pretty essential > > requirement. > > Sure, it's a requirement, but if it isn't there, then the check for libc > fails. Right now, the resources needed are head, cut, echo, readlink, > and cat; other than the resources being directly tested. > > I just don't want to see a cascading failure.
Yeah, that's a good point. Perhaps using cut, as was posted elsewhere in the thread, would work. The output of ldd is prefixed with a tab, which I thought was spaces, which is why I was using awk -- but it's a tab, so cut should work too; it just requires an extra grep: ldd /bin/ls | grep libc\\.so | cut -d' ' -f3 (The awk address-range operator might be good for education, but I wouldn't be surprised to find that elsewhere in the book already, so either way.)
pgptoDQVofzxj.pgp
Description: PGP signature
-- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page