As I am working on the next functionalities of HID-BPF, I realized that I had a few issues while preloading the skeleton at boot.
None of the errors are terrible as they are not inducing a kernel crash, so it's not super urgent IMO.
Regarding the last one, I'm not sure what makes RHEL behave slightly different than upstream. But I am not sure also that the code matches upstream everywhere, so lazy loading it seems like a sensible idea.
Furthermore, that also means that the code will not be available until requested by user space, which fits well in the whole idea of HID-BPF: if the user doesn't want it, then it shouldn't be it.
Signed-off-by: Benjamin Tissoires bentiss@kernel.org --- Benjamin Tissoires (3): HID: bpf: fix a comment in a define HID: bpf: fix return value of entrypoints_*__attach() HID: bpf: lazy load the hid_tail_call entrypoint
drivers/hid/bpf/hid_bpf_dispatch.c | 6 ------ drivers/hid/bpf/hid_bpf_jmp_table.c | 17 ++++++++++++----- 2 files changed, 12 insertions(+), 11 deletions(-) --- base-commit: b912cf042072e12e93faa874265b30cc0aa521b9 change-id: 20240419-hid_bpf_lazy_skel-ab0d674cb49b
Best regards,