Re: [PATCH stable 6.6.y v3 2/4] bpf: Remove mark_precise_scalar_ids()

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Low] Leftover unused struct bpf_idset and union member idset_scratch.
- [Low] Multi-line comments do not follow the BPF subsystem modern comment style.
--

Note: Due to recitation policy restrictions, this report does not quote the
original patch code inline. Instead, it provides a free-form summary of the
findings with references to the modified files and functions.

commit b92dc3ff591ac22c49e14f36339cfd789c618695
Author: Eduard Zingerman <eddyz87@gmail.com>

bpf: Remove mark_precise_scalar_ids()

This commit removes the mark_precise_scalar_ids() function, as it is
superseded by jump history tracking and bt_sync_linked_regs(). Test cases
in progs/verifier_scalar_ids are updated to reflect the new verifier
behavior requiring conditional jumps.

Link: https://lore.kernel.org/bpf/20240718202357.1746514-3-eddyz87@gmail.com

File: include/linux/bpf_verifier.h

[Severity: Low]
Since the patch completely removes mark_precise_scalar_ids(), does this leave
struct bpf_idset and the idset_scratch member in struct bpf_verifier_env as
unused dead code?

It appears mark_precise_scalar_ids() was the only consumer of these
definitions. Removing them from include/linux/bpf_verifier.h might help keep
the verifier environment structure clean.

File: tools/testing/selftests/bpf/progs/verifier_scalar_ids.c

[Severity: Low]
This isn't a bug, but do the modified and newly added multi-line comments in
the file, including inside linked_regs_broken_link(), deviate from the
preferred BPF subsystem comment style?

The BPF guidelines request that multi-line comments have the opening
indicator on its own line, but several comments in this file keep the text
on the same line as the opening indicator.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/cover.1781194510.git.jt26wzz@gmail.com?part=2