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 <[email protected]>
---
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,
--
Benjamin Tissoires <[email protected]>