All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Alexei Starovoitov" <alexei.starovoitov@gmail.com>
To: "KP Singh" <kpsingh@kernel.org>, <sashiko-reviews@lists.linux.dev>
Cc: <bpf@vger.kernel.org>
Subject: Re: [PATCH] bpf, libbpf: reject non-exclusive metadata maps in the signed loader
Date: Thu, 21 May 2026 03:17:03 -0700	[thread overview]
Message-ID: <DIO9ZMXP4AP8.2U06P4YWLPFV@gmail.com> (raw)
In-Reply-To: <CACYkzJ6rxGSrixr9sChFyXyu3JcjG3JmZGR5YgA0vV0WNJPs0g@mail.gmail.com>

On Tue May 19, 2026 at 8:00 PM PDT, KP Singh wrote:
> On Wed, May 20, 2026 at 1:17 AM <sashiko-bot@kernel.org> wrote:
>>
>> Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
>> - [High] The generated BPF loader program uses a hardcoded 8-byte read (BPF_DW) to access the `excl_prog_sha` pointer field, which causes the BPF verifier to reject the loader on 32-bit architectures and silently bypasses the exclusivity check.
>
> I plan to move this to the kernel with a kfunc anyways. Any
> suggestions for the 32 bit fix (apart from an ugly union in bpf_map).

Use _bpf_md_ptr(). It's a normal pattern.

>> - [High] Branch target offset calculation is off-by-one, causing the jump target to land past the beginning of the cleanup block.
>> --
>
> Yeah this is valid:
>
>  Compute the cleanup-label offset after emitting the MOV that sets the
>   error code in R7. Otherwise the JMP lands one insn past cleanup_label
>   and skips the first FD-close, matching the emit_check_err() pattern.
>
> diff --git a/tools/lib/bpf/gen_loader.c b/tools/lib/bpf/gen_loader.c
> index becfda5b9025..6950443ba698 100644
> --- a/tools/lib/bpf/gen_loader.c
> +++ b/tools/lib/bpf/gen_loader.c
> @@ -592,9 +592,9 @@ static void emit_signature_match(struct bpf_gen *gen)
>                 gen->hash_insn_offset[i] = gen->insn_cur - gen->insn_start;
>                 emit2(gen, BPF_LD_IMM64_RAW_FULL(BPF_REG_3, 0, 0, 0, 0, 0));
>
> -               off =  -(gen->insn_cur - gen->insn_start -
> gen->cleanup_label) / 8 - 1;
> +               emit(gen, BPF_MOV64_IMM(BPF_REG_7, -EINVAL));
> +               off = -(gen->insn_cur - gen->insn_start -
> gen->cleanup_label) / 8 - 1;
>                 if (is_simm16(off)) {
> -                       emit(gen, BPF_MOV64_IMM(BPF_REG_7, -EINVAL));

maybe fix the math instead of moving this emit ?
Also split the fix into pre-patch.

>                         emit(gen, BPF_JMP_REG(BPF_JNE, BPF_REG_2,
> BPF_REG_3, off));
>                 } else {
>                         gen->error = -ERANGE;

and while at it remove redundant 'emit(gen, BPF_JMP_IMM(BPF_JA, 0, 0, -1));'

pw-bot: cr


      reply	other threads:[~2026-05-21 10:17 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-19 21:53 [PATCH] bpf, libbpf: reject non-exclusive metadata maps in the signed loader KP Singh
2026-05-19 22:50 ` bot+bpf-ci
2026-05-19 23:17 ` sashiko-bot
2026-05-20  3:00   ` KP Singh
2026-05-21 10:17     ` Alexei Starovoitov [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=DIO9ZMXP4AP8.2U06P4YWLPFV@gmail.com \
    --to=alexei.starovoitov@gmail.com \
    --cc=bpf@vger.kernel.org \
    --cc=kpsingh@kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.