Thanks for bringing attention to the issue. This is an existing-----Original Message-----
From: Shuah Khan
Fix memfd to support relocatable build (O=objdir). This calls out
source files necessary to build tests and simplfies the dependency
enforcement.
Tested the following:
Note that cross-build for fuse_mnt has dependency on -lfuse.
make all
make clean
make kselftest-install O=/arm64_build/ ARCH=arm64 HOSTCC=gcc \
CROSS_COMPILE=aarch64-linux-gnu- TARGETS=memfd
Signed-off-by: Shuah Khan <skhan@xxxxxxxxxxxxxxxxxxx>
---
tools/testing/selftests/memfd/Makefile | 9 ++++-----
1 file changed, 4 insertions(+), 5 deletions(-)
diff --git a/tools/testing/selftests/memfd/Makefile b/tools/testing/selftests/memfd/Makefile
index 53a848109f7b..0a15f9e23431 100644
--- a/tools/testing/selftests/memfd/Makefile
+++ b/tools/testing/selftests/memfd/Makefile
@@ -4,9 +4,8 @@ CFLAGS += -I../../../../include/uapi/
CFLAGS += -I../../../../include/
CFLAGS += -I../../../../usr/include/
-TEST_GEN_PROGS := memfd_test
+TEST_GEN_PROGS := memfd_test fuse_test fuse_mnt
TEST_PROGS := run_fuse_test.sh run_hugetlbfs_test.sh
-TEST_GEN_FILES := fuse_mnt fuse_test
fuse_mnt.o: CFLAGS += $(shell pkg-config fuse --cflags)
@@ -14,7 +13,7 @@ include ../lib.mk
$(OUTPUT)/fuse_mnt: LDLIBS += $(shell pkg-config fuse --libs)
Sorry to interject here. This comment doesn't have to do with this patch,
but the above line is problematical for cross-compilation.
$(shell pkg-config fuse --libs) will use pkg-config from the host
machine, and may produce something different from what's
installed on the target. For most enterprise and desktop systems
library paths for are mostly standardized. The effect is that if
you're doing 'simple' embedded, the pkg-config output on the
host will match that of the target board. However, in some
embedded products the partition and filesystem setups are
often quite complicated (e.g. I recently worked on a TV set with 14
partitions) and things are not in the locations preferred by the
Filesystem Hierarchy Standard.
When cross-compiling for our products, we tend to use the pkg-config
from the filesystem of the target (a copy of which is on the host), rather
than the host's pkg-config.
I don't have a board in my test lab where this shows up, nor a
proposed solution at the moment, but I'm just mentioning it as a
potential host-environment build contamination issue. Maybe I
should start keeping a list, and start thinking of possible solutions for
more of these types of issues.
P.S. I'd ignore this issue for approving this patch. I'm just bringing
it up as a red flag for future work.