Re: [GIT PULL] Kselftest update for Linux 5.2-rc1

[Alexei Starovoitov <alexei.starovoitov@gmail.com>]

On Mon, May 06, 2019 at 10:56:56AM -0600, Shuah Khan wrote:
> Hi Linus,
> 
> Please pull the following Kselftest update for Linux 5.2-rc1
> 
> This Kselftest update for Linux 5.2-rc1 consists of
> 
> - fixes to seccomp test, and kselftest framework
> - cleanups to remove duplicate header defines
> - fixes to efivarfs "make clean" target
> - cgroup cleanup path
> - Moving the IMA kexec_load selftest to selftests/kexec work from
>   Mimi Johar and Petr Vorel
> - A framework to kselftest for writing kernel test modules addition
>   from Tobin C. Harding
> 
> diff is attached.
> 
> thanks,
> -- Shuah
> 
> 
> ----------------------------------------------------------------
> The following changes since commit 15ade5d2e7775667cf191cf2f94327a4889f8b9d:
> 
>   Linux 5.1-rc4 (2019-04-07 14:09:59 -1000)
> 
> are available in the Git repository at:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest
> tags/linux-kselftest-5.2-rc1
> 
> for you to fetch changes up to d917fb876f6eaeeea8a2b620d2a266ce26372f4d:
> 
>   selftests: build and run gpio when output directory is the src dir
> (2019-04-22 17:02:26 -0600)
> 
> ----------------------------------------------------------------
> linux-kselftest-5.2-rc1
> 
> This Kselftest update for Linux 5.2-rc1 consists of
> 
> - fixes to seccomp test, and kselftest framework
> - cleanups to remove duplicate header defines
> - fixes to efivarfs "make clean" target
> - cgroup cleanup path
> - Moving the IMA kexec_load selftest to selftests/kexec work from
>   Mimi Johar and Petr Vorel
> - A framework to kselftest for writing kernel test modules addition
>   from Tobin C. Harding
> 
> ----------------------------------------------------------------
> Kees Cook (3):
>       selftests/seccomp: Handle namespace failures gracefully
>       selftests/harness: Add 30 second timeout per test
>       selftests/ipc: Fix msgque compiler warnings
> 
> Mathieu Desnoyers (1):
>       rseq/selftests: Adapt number of threads to the number of detected cpus
> 
> Mimi Zohar (9):
>       selftests/kexec: move the IMA kexec_load selftest to selftests/kexec
>       selftests/kexec: cleanup the kexec selftest
>       selftests/kexec: define a set of common functions
>       selftests/kexec: define common logging functions
>       selftests/kexec: define "require_root_privileges"
>       selftests/kexec: kexec_file_load syscall test
>       selftests/kexec: check kexec_load and kexec_file_load are enabled
>       selftests/kexec: make kexec_load test independent of IMA being enabled
>       selftests/kexec: update get_secureboot_mode
> 
> Petr Vorel (1):
>       selftests/kexec: Add missing '=y' to config options
> 
> Po-Hsu Lin (1):
>       selftests/efivarfs: clean up test files from test_create*()
> 
> Roman Gushchin (1):
>       selftests: cgroup: fix cleanup path in test_memcg_subtree_control()
> 
> Sabyasachi Gupta (4):
>       selftest/x86/mpx-dig.c: Remove duplicate header
>       selftest/timers: Remove duplicate header
>       selftest/rseq: Remove duplicate header
>       selftest/gpio: Remove duplicate header
> 
> Shuah Khan (2):
>       selftests: fix headers_install circular dependency

Shuah,

the commit 8ce72dc32578 ("selftests: fix headers_install circular dependency")
broke our build/test workflow, since it added:
  ifneq ($(KBUILD_OUTPUT),)
          OUTPUT := $(KBUILD_OUTPUT)
  else

which means that all of selftests/bpf artifacts are now going into
main build directory cluttering it with all sorts of .o, generated files
and executables.
The end result is humans and scripts can no longer find tests.

For now I hacked it as:
diff --git a/tools/testing/selftests/lib.mk b/tools/testing/selftests/lib.mk
index 5979fdc4f36c..caecec7aebde 100644
--- a/tools/testing/selftests/lib.mk
+++ b/tools/testing/selftests/lib.mk
@@ -6,12 +6,8 @@ ifeq (0,$(MAKELEVEL))
     ifneq ($(O),)
        OUTPUT := $(O)
     else
-       ifneq ($(KBUILD_OUTPUT),)
-               OUTPUT := $(KBUILD_OUTPUT)
-       else
-               OUTPUT := $(shell pwd)
-               DEFAULT_INSTALL_HDR_PATH := 1
-       endif
+       OUTPUT := $(shell pwd)
+       DEFAULT_INSTALL_HDR_PATH := 1
     endif
 endif

bpf developers are doing "cd tools/testing/selftests/bpf; make; ./test_verifier; ..."
while KBUILD_OUTPUT is also set.
I don't quite get this 'circular dependency' issue that your commit suppose to address
but please fix it differently, so bpf developer's workflow is restored and buildbots work again.
People and scripts depend on it.
It's even described in Documentation/bpf/bpf_devel_QA.rst

Thanks