libbpf: Remove mode check in libbpf_set_strict_mode()
authorMauricio Vásquez <mauricio@kinvolk.io>
Mon, 7 Feb 2022 14:50:50 +0000 (09:50 -0500)
committerAndrii Nakryiko <andrii@kernel.org>
Mon, 7 Feb 2022 20:12:22 +0000 (12:12 -0800)
commite4e835c87bb5ef7e7923e72da57c923bfcade418
tree136eac0689057d673d5cb7a2c85ccf6c6de1c7c6
parente91d280c840f133560072f246321f9a4d1f4eb14
libbpf: Remove mode check in libbpf_set_strict_mode()

libbpf_set_strict_mode() checks that the passed mode doesn't contain
extra bits for LIBBPF_STRICT_* flags that don't exist yet.

It makes it difficult for applications to disable some strict flags as
something like "LIBBPF_STRICT_ALL & ~LIBBPF_STRICT_MAP_DEFINITIONS"
is rejected by this check and they have to use a rather complicated
formula to calculate it.[0]

One possibility is to change LIBBPF_STRICT_ALL to only contain the bits
of all existing LIBBPF_STRICT_* flags instead of 0xffffffff. However
it's not possible because the idea is that applications compiled against
older libbpf_legacy.h would still be opting into latest
LIBBPF_STRICT_ALL features.[1]

The other possibility is to remove that check so something like
"LIBBPF_STRICT_ALL & ~LIBBPF_STRICT_MAP_DEFINITIONS" is allowed. It's
what this commit does.

[0]: https://lore.kernel.org/bpf/20220204220435.301896-1-mauricio@kinvolk.io/
[1]: https://lore.kernel.org/bpf/CAEf4BzaTWa9fELJLh+bxnOb0P1EMQmaRbJVG0L+nXZdy0b8G3Q@mail.gmail.com/

Fixes: 93b8952d223a ("libbpf: deprecate legacy BPF map definitions")
Signed-off-by: Mauricio Vásquez <mauricio@kinvolk.io>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20220207145052.124421-2-mauricio@kinvolk.io
tools/lib/bpf/libbpf.c