Re: [PATCH rebased] kbuild: rpm-pkg: Fix build with non-default MODLIB

From: Masahiro Yamada <masahiroy@kernel.org>
To: "Michal Suchánek" <msuchanek@suse.de>
Cc: linux-kbuild@vger.kernel.org, linux-kernel@vger.kernel.org,
	"Nicolas Schier" <nicolas@fjasle.eu>,
	linux-modules@vger.kernel.org, "Takashi Iwai" <tiwai@suse.com>,
	"Lucas De Marchi" <lucas.de.marchi@gmail.com>,
	"Michal Koutný" <mkoutny@suse.com>,
	"Jiri Slaby" <jslaby@suse.com>,
	"Jan Engelhardt" <jengelh@inai.de>,
	"Nathan Chancellor" <nathan@kernel.org>,
	"Nick Desaulniers" <ndesaulniers@google.com>
Subject: Re: [PATCH rebased] kbuild: rpm-pkg: Fix build with non-default MODLIB
Date: Tue, 10 Oct 2023 00:14:01 +0900	[thread overview]
Message-ID: <CAK7LNAQQMFUt4R1m_U8kBY5=BvxD_dMuE4MD4kpd48WK1E+AGA@mail.gmail.com> (raw)
In-Reply-To: <20231009140733.GV6241@kitsune.suse.cz>

On Mon, Oct 9, 2023 at 11:07 PM Michal Suchánek <msuchanek@suse.de> wrote:
>
> On Mon, Oct 09, 2023 at 09:34:10PM +0900, Masahiro Yamada wrote:
> > On Mon, Oct 9, 2023 at 5:52 PM Michal Suchánek <msuchanek@suse.de> wrote:
> > >
> > > Hello,
> > >
> > > On Mon, Oct 09, 2023 at 05:31:02PM +0900, Masahiro Yamada wrote:
> > > > On Fri, Oct 6, 2023 at 12:49 AM Michal Suchanek <msuchanek@suse.de> wrote:
> > > > >
> > > > > The default MODLIB value is composed of two variables and the hardcoded
> > > > > string '/lib/modules/'.
> > > > >
> > > > > MODLIB = $(INSTALL_MOD_PATH)/lib/modules/$(KERNELRELEASE)
> > > > >
> > > > > Defining this middle part as a variable was rejected on the basis that
> > > > > users can pass the whole MODLIB to make, such as
> > > >
> > > >
> > > > In other words, do you want to say
> > > >
> > > > "If defining this middle part as a variable had been accepted,
> > > > this patch would have been unneeded." ?
> > >
> > > If it were accepted I would not have to guess what the middle part is,
> > > and could use the variable that unambiguosly defines it instead.
> >
> >
> > How?
> >
> > scripts/package/kernel.spec hardcodes 'lib/modules'
> > in a couple of places.
> >
> > I am asking how to derive the module path.
>
> Not sure what you are asking here. The path is hardcoded, everywhere.
>
> The current Makefile has
>
> MODLIB  = $(INSTALL_MOD_PATH)/lib/modules/$(KERNELRELEASE)
>
> and there is no reliable way to learn what the middle part was after the
> fact - $(INSTALL_MOD_PATH) can be non-empty.
>
> The rejected patch was changing this to a variable, and also default to
> adjusting the content to what kmod exports in pkgconfig after applying a
> proposed patch to make this hardcoded part configurable:
>
> export KERNEL_MODULE_DIRECTORY := $(shell pkg-config --print-variables kmod 2>/dev/null | grep '^module_directory$$' >/dev/null && pkg-config --variable=module_directory kmod || echo /lib/modules)
>
> MODLIB  = $(INSTALL_MOD_PATH)$(KERNEL_MODULE_DIRECTORY)/$(KERNELRELEASE)
>
> It would be completely posible to only define the middle part as a
> variable that could then be used in rpm-pkg:
>
> export KERNEL_MODULE_DIRECTORY := /lib/modules
>
> MODLIB  = $(INSTALL_MOD_PATH)$(KERNEL_MODULE_DIRECTORY)/$(KERNELRELEASE)
>
> Thanks
>
> Michal
>
>

Let me add more context to my question.

I am interested in the timing when
'pkg-config --print-variables kmod | grep module_directory'
is executed.

1.  Build a SRPM on machine A

2.  Copy the SRPM from machine A to machine B

3.  Run rpmbuild on machine B to build the SRPM into a RPM

4.  Copy the RPM from machine B to machine C

5.  Install the RPM to machine C

Of course, we are most interested in the module path
of machine C, but it is difficult/impossible to
guess it at the time of building.

We can assume machine B == machine C.

We are the second most interested in the module
path on machine B.

The module path of machine A is not important.

So, I am asking where you would inject
'pkg-config --print-variables kmod | grep module_directory'.

-- 
Best Regards
Masahiro Yamada