The way the test target was defined before, when building with clang we
get a command line like this:
clang -Wall -Werror -g -I../../../../usr/include/ \
regression_enomem.c ../pidfd/pidfd.h -o regression_enomem
This yields an error, because clang thinks we want to produce both a *.o
file, as well as a precompiled header:
clang: error: cannot specify -o when generating multiple output files
gcc, for whatever reason, doesn't exhibit the same behavior which I
suspect is why the problem wasn't noticed before.
This can be fixed simply by using the LOCAL_HDRS infrastructure the
selftests lib.mk provides. It does the right think and marks the target
as depending on the header (so if the header changes, we rebuild), but
it filters the header out of the compiler command line, so we don't get
the error described above.
Signed-off-by: Axel Rasmussen <[email protected]>
---
tools/testing/selftests/pid_namespace/Makefile | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/tools/testing/selftests/pid_namespace/Makefile b/tools/testing/selftests/pid_namespace/Makefile
index dcaefa224ca0..edafaca1aeb3 100644
--- a/tools/testing/selftests/pid_namespace/Makefile
+++ b/tools/testing/selftests/pid_namespace/Makefile
@@ -1,8 +1,8 @@
# SPDX-License-Identifier: GPL-2.0
CFLAGS += -g -I../../../../usr/include/
-TEST_GEN_PROGS := regression_enomem
+TEST_GEN_PROGS = regression_enomem
-include ../lib.mk
+LOCAL_HDRS += $(selfdir)/pidfd/pidfd.h
-$(OUTPUT)/regression_enomem: regression_enomem.c ../pidfd/pidfd.h
+include ../lib.mk
--
2.35.1.1021.g381101b075-goog
On 3/24/22 4:39 PM, Axel Rasmussen wrote:
> The way the test target was defined before, when building with clang we
> get a command line like this:
>
> clang -Wall -Werror -g -I../../../../usr/include/ \
> regression_enomem.c ../pidfd/pidfd.h -o regression_enomem
>
> This yields an error, because clang thinks we want to produce both a *.o
> file, as well as a precompiled header:
>
> clang: error: cannot specify -o when generating multiple output files
>
> gcc, for whatever reason, doesn't exhibit the same behavior which I
> suspect is why the problem wasn't noticed before.
>
Thank you fixing this.
> This can be fixed simply by using the LOCAL_HDRS infrastructure the
> selftests lib.mk provides. It does the right think and marks the target
> as depending on the header (so if the header changes, we rebuild), but
> it filters the header out of the compiler command line, so we don't get
> the error described above.
>
> Signed-off-by: Axel Rasmussen <[email protected]>
> ---
> tools/testing/selftests/pid_namespace/Makefile | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/tools/testing/selftests/pid_namespace/Makefile b/tools/testing/selftests/pid_namespace/Makefile
> index dcaefa224ca0..edafaca1aeb3 100644
> --- a/tools/testing/selftests/pid_namespace/Makefile
> +++ b/tools/testing/selftests/pid_namespace/Makefile
> @@ -1,8 +1,8 @@
> # SPDX-License-Identifier: GPL-2.0
> CFLAGS += -g -I../../../../usr/include/
>
> -TEST_GEN_PROGS := regression_enomem
> +TEST_GEN_PROGS = regression_enomem
>
> -include ../lib.mk
> +LOCAL_HDRS += $(selfdir)/pidfd/pidfd.h
>
> -$(OUTPUT)/regression_enomem: regression_enomem.c ../pidfd/pidfd.h
> +include ../lib.mk
>
Will apply this for Linux 5.18-rc2
thanks,
-- Shuah