Re: [PATCH 0/8] selftests/bpf: installation and out of tree build fixes

From: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
To: Andrii Nakryiko <andrii.nakryiko@gmail.com>
Cc: bpf <bpf@vger.kernel.org>, Jiri Benc <jbenc@redhat.com>,
	Jiri Olsa <jolsa@redhat.com>, Andrii Nakryiko <andriin@fb.com>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Shuah Khan <shuah@kernel.org>,
	"open list\:KERNEL SELFTEST FRAMEWORK" 
	<linux-kselftest@vger.kernel.org>
Subject: Re: [PATCH 0/8] selftests/bpf: installation and out of tree build fixes
Date: Wed, 27 May 2020 10:25:35 +0300	[thread overview]
Message-ID: <xunyy2pdon4g.fsf@redhat.com> (raw)
In-Reply-To: <CAEf4Bzau5MWYqP1XfZzYVj6tf7Y9fmJRgjbO00DzOmJ_iGschg@mail.gmail.com> (Andrii Nakryiko's message of "Tue, 26 May 2020 22:04:35 -0700")

Hi, Andrii!

>>>>> On Tue, 26 May 2020 22:04:35 -0700, Andrii Nakryiko  wrote:

 > On Tue, May 26, 2020 at 9:52 PM Yauheni Kaliuta
 > <yauheni.kaliuta@redhat.com> wrote:
 >> 
 >> Hi, Andrii!
 >> 
 >> >>>>> On Tue, 26 May 2020 15:32:10 -0700, Andrii Nakryiko  wrote:
 >> 
 >> > On Thu, May 21, 2020 at 11:41 PM Yauheni Kaliuta
 >> > <yauheni.kaliuta@redhat.com> wrote:
 >> >>
 >> >>
 >> >> Actually, a bit more needed :)
 >> 
 >> > From the other kselftest thread, it seems like selftests are not
 >> > supporting builds out-of-tree. With that, wouldn't it be simpler to
 >> > build in tree and then just copy selftests/bpf directory to wherever
 >> > you need to run tests from? It would be simple and reliable. Given I
 >> > and probably everyone else never build and run tests out-of-tree, it's
 >> > just too easy to break this and you'll be constantly chasing some
 >> > non-obvious breakages...
 >> 
 >> > Is there some problem with such approach?
 >> 
 >> This is `make install` ;).

 > So patch #2, #5, and #7 is solving just `make install` problem?..

No, they are fixing OOT build problems. I should have probably
split the series, oot and install fixes.

 > My point is that by building in tree and then just copying
 > everything under selftests/bpf directory to wherever you want
 > to "install" it would just work. And won't require
 > complicating already complicated Makefile. Any problem with
 > such approach?

I understand. I see only wasting of space as a problem, but
should check.

 >> 
 >> I personally do not need OOT build, but since it's in the code,
 >> I'd prefer either fix it or remove it, otherwise it's
 >> misleading. But I have not got reply from kselftest.
 >> 
 >> >>
 >> >> >>>>> On Fri, 22 May 2020 07:13:02 +0300, Yauheni Kaliuta  wrote:
 >> >>
 >> >> > I had a look, here are some fixes.
 >> >> > Yauheni Kaliuta (8):
 >> >> >   selftests/bpf: remove test_align from Makefile
 >> >> >   selftests/bpf: build bench.o for any $(OUTPUT)
 >> >> >   selftests/bpf: install btf .c files
 >> >> >   selftests/bpf: fix object files installation
 >> >> >   selftests/bpf: add output dir to include list
 >> >> >   selftests/bpf: fix urandom_read installation
 >> >> >   selftests/bpf: fix test.h placing for out of tree build
 >> >> >   selftests/bpf: factor out MKDIR rule
 >> >>
 >> >> >  tools/testing/selftests/bpf/Makefile | 77 ++++++++++++++++++++--------
 >> >> >  1 file changed, 55 insertions(+), 22 deletions(-)
 >> >>
 >> >> > --
 >> >> > 2.26.2
 >> >>
 >> >>
 >> >> --
 >> >> WBR,
 >> >> Yauheni Kaliuta
 >> >>
 >> 
 >> 
 >> --
 >> WBR,
 >> Yauheni Kaliuta
 >> 

-- 
WBR,
Yauheni Kaliuta