On Wed, 2012-07-04 at 19:18 +0100, Matt Wilson wrote:
Currently shared libraries are automatically installed into /usr/lib
or /usr/lib64, depending on the supplied --prefix value and
$(XEN_TARGET_ARCH). Some systems, like recent Debian and Ubuntu releases,
do not use /usr/lib64, but instead /usr/lib/x86_64-linux-gnu.
With this change, packagers can supply the desired location for shared
libraries on the ./configure command line. Packagers need to note that
the default behaviour on 64-bit Linux systems will be to install shared
libraries in /usr/lib, not /usr/lib64, unless a --libdir value is provided
to ./configure.
Additionally, the libfsimage plugins are now loaded explicitly from
$LIBDIR/fs, removing platform-based decision trees in code.
The LIBLEAFDIR{,_x86_32,_x86_64} make variables and the LIB_PATH
autoconf substition are no longer used, so this change removes
m4/default_lib.m4 and regenerates configure. m4/pkg.m4 has been
refreshed from pkg-config 0.26 to resolve an error,
"possibly undefined macro: AS_MESSAGE_LOG_FDdnl" when running
autoconf.
Was this required due to the removal of default_lib.m4?
Signed-off-by: Matt Wilson<msw@xxxxxxxxxx>
This time I get everything installed in /lib instead of /usr/lib. e.g.
dist/install/lib/fs instead of the expected dist/install/usr/lib/fs
dist/install/lib/libxenctrl.so and not
dist/install/usr/lib/libxenctrl.so etc etc.
I suspect this is due to:
$ head -n 4 config/Tools.mk
# Prefix and install folder
PREFIX := /usr
exec_prefix := ${prefix}
LIBDIR := ${exec_prefix}/lib
Note the case of prefix vs PREFIX. AFAICT $PREFIX is completely a
Xen-ism while autoconf always uses $prefix.
On a hunch I reverted the change to remove m4/default_lib.m4 and this
lead to the expected result.
I'm going to revert the m4/default_lib.m4 and by extension the m4/pkg.m4
changes and commit the result (after consulting with Ian J to make sure
the test system can cope with this change).
Please can you resubmit the other bits separately? I suspect the right
answer will be to include prefix:=@prefix@ in config/Tools.mk.in, but
I've not checked that.
For reference my test procedure is to do a full "make dist" before and
after each time doing:
find dist | sort> FILE_LIST.BEFORE
(or ....AFTER as appropriate) and then to do
diff -u<(sed -e 's/lib64/lib/' ../FILE_LIST.BEFORE | sort)
../FILE_LIST.AFTER
It'd be useful if you could do something similar as part of your
testing.
FYI I also tested live migration of PV (booted with pygrub), HVM (w/
qemu in dom0) and HVM (w/ stubdomains) with xl and PV migration and HVM
migration with xend. Since pygrub and the various save herlpes are some
of the things which have moved around. All seemed fine.
Thanks,
Ian.
