linux-kernel - Re: [PATCH v1 2/3] selftests/nolibc: fix up O= option support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <058a264d-45bd-4f1f-8af3-56ed337b3251@t-8ch.de>
Date:   Fri, 4 Aug 2023 17:05:06 +0200
From:   Thomas Weißschuh <thomas@...ch.de>
To:     Zhangjin Wu <falcon@...ylab.org>
Cc:     arnd@...db.de, linux-kernel@...r.kernel.org,
        linux-kselftest@...r.kernel.org, tanyuan@...ylab.org, w@....eu
Subject: Re: [PATCH v1 2/3] selftests/nolibc: fix up O= option support

On 2023-08-04 15:43:42+0800, Zhangjin Wu wrote:
> Hi, Thomas
> 
> > On 2023-08-03 22:45:52+0800, Zhangjin Wu wrote:
> > > To avoid pollute the source code tree and avoid mrproper for every
> > > architecture switch, the O= argument must be supported.
> > > 
> > > Both IMAGE and .config are from the building directory, let's use
> > > objtree instead of srctree for them.
> > > 
> > > If no O= option specified, means building kernel in source code tree,
> > > objtree should be srctree in such case.
> > > 
> > > Suggested-by: Willy Tarreau <w@....eu>
> > > Link: https://lore.kernel.org/lkml/ZK0AB1OXH1s2xYsh@1wt.eu/
> > > Signed-off-by: Zhangjin Wu <falcon@...ylab.org>
> > > ---
> > >  tools/testing/selftests/nolibc/Makefile | 7 +++++--
> > >  1 file changed, 5 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/tools/testing/selftests/nolibc/Makefile b/tools/testing/selftests/nolibc/Makefile
> > > index 51fef5e6a152..af590aee063a 100644
> > > --- a/tools/testing/selftests/nolibc/Makefile
> > > +++ b/tools/testing/selftests/nolibc/Makefile
> > > @@ -9,6 +9,9 @@ ifeq ($(srctree),)
> > >  srctree := $(patsubst %/tools/testing/selftests/,%,$(dir $(CURDIR)))
> > >  endif
> > >  
> > > +# add objtree for O= argument, required by IMAGE and .config
> > > +objtree ?= $(srctree)
> > 
> > Isn't this already set by the included tools/scripts/Makefile.include?
> >
> 
> Good question, but it is empty if no O= specified, checked it several
> times before ;-)

For me it is not empty when I am in tools/testing/selftests/nolibc/.

> > Also I'm not entirely if O= works as intended currently.
> > When using O=foo in the $LINUX/tools/testing/selftests/nolibc directory
> 
> It does work, I have used it to test all of the run targets of the
> tinyconfig patches like this:
> 
>     $ make run O=kernel-$arch RUN_OUT=run.$arch.out ...
> 
> Everything about the kernel will be built in $(O).

It does not work for me:

$ cd $HOME/src/linux/tools/testing/selftests/nolibc
$ make O=out kernel
../../../scripts/Makefile.include:4: *** O=out does not exist.  Stop.
$ mkdir out
$ make O=out kernel
... Stuff happens
... all the output is in $HOME/src/linux/out not in
    tools/testing/selftests/nolibc/out where I would expect it.
    The new out directory for which an error was reported before is
    completely empty.

> Just rechecked the O variable in top-level Makefile, selftests/nolibc Makefile
> and tools/nolibc Makefile, all of them get the right O value from command line.

But the one from the command line is a relative path. And it seems to be
resolved from the kernel source tree instead of from where make is
executed.

> From my Makefile experience, an option from command line will be passed
> to every sub Makefile via $(-*-command-variables-*-), it has the highest
> priority then the others, except when we use 'override' keyword
> internally.
> 
> > the build instead is happening in $LINUX/foo. But the Makefile first
> > validates that $LINUX/tools/testing/selftests/nolibc/foo exists.
> >
> 
> Sorry, I didn't get your meaning above?
> 
> Do you mean this line:
> 
>     srctree := $(patsubst %/tools/testing/selftests/,%,$(dir $(CURDIR)))
> 
> by removing tools/testing/selftests/ of curdir's dir (means no nolibc itself),
> srctree above is just the top-level kernel source code tree.

No, srctree should be fine.

> > It seems we need to pass $(COMMAND_O) to the recursive calls to $(MAKE),
> > too?
> >
> 
> So, no need to pass O or COMMAND_O.

As above, it doesn't work for me.

If I add COMMAND_O to all the $(MAKE) commands that are executed
explicitly, how it is also done in Makefile.include itself, things start
to work as I would expect:
All the kernel build output is in out/ in the current directory.

> Thanks,
> Zhangjin
> 
> > > +
> > >  ifeq ($(ARCH),)
> > >  include $(srctree)/scripts/subarch.include
> > >  ARCH = $(SUBARCH)
> > > @@ -217,12 +220,12 @@ kernel: initramfs
> > >  
> > >  # run the tests after building the kernel
> > >  run: kernel
> > > -	$(Q)qemu-system-$(QEMU_ARCH) -display none -no-reboot -kernel "$(srctree)/$(IMAGE)" -serial stdio $(QEMU_ARGS) > "$(CURDIR)/run.out"
> > > +	$(Q)qemu-system-$(QEMU_ARCH) -display none -no-reboot -kernel "$(objtree)/$(IMAGE)" -serial stdio $(QEMU_ARGS) > "$(CURDIR)/run.out"
> > >  	$(Q)$(REPORT) $(CURDIR)/run.out
> > >  
> > >  # re-run the tests from an existing kernel
> > >  rerun:
> > > -	$(Q)qemu-system-$(QEMU_ARCH) -display none -no-reboot -kernel "$(srctree)/$(IMAGE)" -serial stdio $(QEMU_ARGS) > "$(CURDIR)/run.out"
> > > +	$(Q)qemu-system-$(QEMU_ARCH) -display none -no-reboot -kernel "$(objtree)/$(IMAGE)" -serial stdio $(QEMU_ARGS) > "$(CURDIR)/run.out"
> > >  	$(Q)$(REPORT) $(CURDIR)/run.out
> > >  
> > >  # report with existing test log
> > > -- 
> > > 2.25.1
> > >