Re: [PATCH bpf-next v2] selftests/bpf: Include runner extras for install target
From: Björn Töpel
Date: Sat Jan 27 2024 - 04:03:40 EST
Andrii Nakryiko <andrii.nakryiko@xxxxxxxxx> writes:
> On Thu, Jan 25, 2024 at 8:05 AM Björn Töpel <bjorn@xxxxxxxxxx> wrote:
>>
>> From: Björn Töpel <bjorn@xxxxxxxxxxxx>
>>
>> When using the "install" or targets depending on install, e.g.
>> "gen_tar", the "runner extras" weren't included for the BPF machine
>> flavors.
>>
>> Make sure the necessary helper scripts/tools are added to
>> corresponding BPF machine flavor.
>>
>> Signed-off-by: Björn Töpel <bjorn@xxxxxxxxxxxx>
>> ---
>> v2: Added btf_dump_test_case files
>> ---
>> tools/testing/selftests/bpf/Makefile | 18 +++++++++++++++---
>> 1 file changed, 15 insertions(+), 3 deletions(-)
>>
>> diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
>> index fd15017ed3b1..d5cff32997b3 100644
>> --- a/tools/testing/selftests/bpf/Makefile
>> +++ b/tools/testing/selftests/bpf/Makefile
>> @@ -744,8 +744,20 @@ EXTRA_CLEAN := $(TEST_CUSTOM_PROGS) $(SCRATCH_DIR) $(HOST_SCRATCH_DIR) \
>> DEFAULT_INSTALL_RULE := $(INSTALL_RULE)
>> override define INSTALL_RULE
>> $(DEFAULT_INSTALL_RULE)
>> - @for DIR in $(TEST_INST_SUBDIRS); do \
>> - mkdir -p $(INSTALL_PATH)/$$DIR; \
>> - rsync -a $(OUTPUT)/$$DIR/*.bpf.o $(INSTALL_PATH)/$$DIR;\
>> + @for DIR in $(TEST_INST_SUBDIRS); do \
>> + mkdir -p $(INSTALL_PATH)/$$DIR; \
>> + rsync -a --copy-unsafe-links \
>> + $(OUTPUT)/$$DIR/bpf_testmod.ko \
>> + $(OUTPUT)/$$DIR/bpftool \
>> + $(OUTPUT)/$$DIR/ima_setup.sh \
>> + $(OUTPUT)/$$DIR/liburandom_read.so \
>> + $(OUTPUT)/$$DIR/sign-file \
>> + $(OUTPUT)/$$DIR/uprobe_multi \
>> + $(OUTPUT)/$$DIR/urandom_read \
>> + $(OUTPUT)/$$DIR/verify_sig_setup.sh \
>> + $(OUTPUT)/$$DIR/xdp_synproxy \
>> + $(OUTPUT)/$$DIR/btf_dump_test_case_*.c \
>> + $(OUTPUT)/$$DIR/*.bpf.o \
>> + $(INSTALL_PATH)/$$DIR; \
>
> My concern is that this will get out of sync and will go unnoticed
> next time we add another "extra" file. We have TRUNNER_EXTRA_FILES,
> should we use that list to keep these extras in fewer places?
Yeah, you're completely right -- this was a lazy approach. I'll spin a
more robust v3.
Cheers,
Björn