Thread (5 messages) 5 messages, 3 authors, 2024-08-20

Re: [PATCH bpf-next] bpf: use kfunc hooks instead of program types

From: Alexei Starovoitov <hidden>
Date: 2024-08-20 06:34:36
Also in: bpf, cgroups, linux-doc, linux-fsdevel, linux-kselftest, linux-trace-kernel, lkml, netdev, netfilter-devel

On Tue, Aug 20, 2024 at 2:03 AM Matteo Croce [off-list ref] wrote:
From: Matteo Croce <redacted>

Pass to register_btf_kfunc_id_set() a btf_kfunc_hook directly, instead
of a bpf_prog_type.
Many program types share the same kfunc hook, so some calls to
register_btf_kfunc_id_set() can be removed.

Tested compiling the kernel with -Werror=enum-conversion to catch all
the occourrences.

Signed-off-by: Matteo Croce <redacted>
I think it's too soon to do this kind of cleanup.
We need to refactor this logic to accommodate sched-ext
allow/deny pattern for kfunc per struct-ops hook.
Let's keep this code as-is for now.
pw-bot: cr
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help