Per the comment added in the code, there are 2 definitions of
BPF_F_LINK in the kernel headers. Once is in an anonymous enum which
bindgen will constify, and once is via a #define macro. The values are
identical. The fix is to exclude BPF_F_LINK from the list of variables
in bindgen removing one of the duplicate definitions.
Signed-off-by: Dave Tucker <dave@dtucker.co.uk>
This will generate bindings for mips the next time that the codegen job
is run.
Signed-off-by: Ishan Jain <contact@ishanjain.me>
Co-authored-by: Dave Tucker <dave@dtucker.co.uk>
This commit adds the s390x and powerpc architectures
to codegen. This will enable an upcoming PR to add
support for aya to support theses architectures
in both aya and aya-ebpf.
Co-authored-by: Dave Tucker <dave@dtucker.co.uk>
Signed-off-by: Billy McFall <22157057+Billy99@users.noreply.github.com>
Currently, when running `cargo +nightly xtask codegen` locally and in
the codegen GHA workflow, an error occurs with only "Error: bindgen
failed" displayed.
This was due to a path using "bpf/..." instead of "ebpf/...". It is now
corrected to "ebpf", and bindgen fails should now display a more direct
message on why it failed.
Fixes: #914
Tamir Duberstein
Dave Tucker
Ishan Jain
astoycos
Billy McFall
tyrone-wu