Re: [PATCH] kbuild: modules-cpio-pkg: Respect INSTALL_MOD_PATH

[Nicolas Schier <nsc@kernel.org>]

On Wed, Mar 25, 2026 at 03:48:26PM +0100, Janne Grunau wrote:
> On Sat, Mar 21, 2026 at 04:18:46PM +0100, Nicolas Schier wrote:
> > On Fri, Mar 20, 2026 at 03:30:32PM +0100, Janne Grunau wrote:
> > > The modules-cpio-pkg target added in commit 2a9c8c0b59d3 ("kbuild: add
> > > target to build a cpio containing modules") is incompatible with
> > > initramfs with merged /lib and /usr/lib directories [1]. "/lib" cannot
> > > be a link and directory at the same time.
> > > Respect a non-empty INSTALL_MOD_PATH in the modules-cpio-pkg target so
> > > that `make INSTALL_MOD_PATH=/usr modules-cpio-pkg` results in the same
> > > module install location as `make INSTALL_MOD_PATH=/usr modules_install`.
> > > 
> > > Tested with Fedora distribution initramfs produced by dracut.
> > > 
> > > Link: https://systemd.io/THE_CASE_FOR_THE_USR_MERGE/ [1]
> > > Signed-off-by: Janne Grunau <j@jannau.net>
> > > ---
> > > Hej,
> > > 
> > > this patch allows to produce modules-cpio initramfs which are compatible
> > > with initramfs with merged /lib and /usr/lib (/lib as symlink to
> > > /usr/lib). I expect initramfs of distributions with merged /usr to have
> > > a merged /usr as well. This is at least true for Fedora initramfs built
> > > with dracut.
> > > 
> > > I'm not sure whether the trickery to avoid repeated '/' is justified. It
> > > is necessary to add a slash between "$@" and a non empty
> > > $(INSTALL_MOD_PATH) to avoid make failures due to non existing
> > > .tmp_modules_cpio when INSTALL_MOD_PATH without leading slash is used.
> > > modules-cpio-pkg`.
> > > 
> > > Better or shorter ways to document this not completely obvious behavior
> > > would be appreciated.
> > > 
> > > Janne
> > > ---
> > >  scripts/Makefile.package | 6 ++++--
> > >  1 file changed, 4 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/scripts/Makefile.package b/scripts/Makefile.package
> > > index 0ec946f9b905f74f8698d8d6967d22f5b76f64e0..ab18cf81622ae319380528c401f9aeb6d32070c6 100644
> > > --- a/scripts/Makefile.package
> > > +++ b/scripts/Makefile.package
> > > @@ -195,7 +195,9 @@ tar%-pkg: linux-$(KERNELRELEASE)-$(ARCH).tar.% FORCE
> > >  .tmp_modules_cpio: FORCE
> > >  	$(Q)$(MAKE) -f $(srctree)/Makefile
> > >  	$(Q)rm -rf $@
> > > -	$(Q)$(MAKE) -f $(srctree)/Makefile INSTALL_MOD_PATH=$@ modules_install
> > > +	$(Q)$(MAKE) -f $(srctree)/Makefile \
> > > +		INSTALL_MOD_PATH=$@$(if $(INSTALL_MOD_PATH),/$(INSTALL_MOD_PATH:/%=%)) \
> > > +		modules_install
> > 
> > Thanks for the patch along with its detailed description!
> > 
> > For completeness: I'd rather use $(addprefix):
> > 
> >     INSTALL_MOD_PATH=$@$(addprefix /,$(INSTALL_MOD_PATH:/%=%))
> > 
> > but as POSIX states:
> > 
> > | Multiple successive <slash> characters are considered to be the same
> > | as one <slash>, except it is implementation-defined whether the case
> > | of exactly two leading <slash> characters is treated specially.
> > https://pubs.opengroup.org/onlinepubs/9799919799.2024edition/
> 
> argh, I did read this but confused myself by the second part to think it
> was only for leading slashes.

:)  you are by far not the only one who gets confused by that.


-- 
Nicolas