[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAEf4BzaRZ60zajtDDkKA+0UwXCsT4HZpJ=f12_GzrEa+rkPfcQ@mail.gmail.com>
Date: Thu, 29 Oct 2020 15:24:50 -0700
From: Andrii Nakryiko <andrii.nakryiko@...il.com>
To: Ian Rogers <irogers@...gle.com>
Cc: Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Martin KaFai Lau <kafai@...com>,
Song Liu <songliubraving@...com>, Yonghong Song <yhs@...com>,
Andrii Nakryiko <andriin@...com>,
John Fastabend <john.fastabend@...il.com>,
KP Singh <kpsingh@...omium.org>,
Networking <netdev@...r.kernel.org>, bpf <bpf@...r.kernel.org>,
open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] libbpf hashmap: Avoid undefined behavior in hash_bits
On Thu, Oct 29, 2020 at 3:10 PM Ian Rogers <irogers@...gle.com> wrote:
>
> If bits is 0, the case when the map is empty, then the >> is the size of
> the register which is undefined behavior - on x86 it is the same as a
> shift by 0.
> Avoid calling hash_bits with bits == 0 by adding additional empty
> hashmap tests.
>
> Suggested-by: Andrii Nakryiko <andriin@...com>,
> Suggested-by: Song Liu <songliubraving@...com>
> Signed-off-by: Ian Rogers <irogers@...gle.com>
> ---
I didn't realize you'd need to add three extra checks. If that's the
case, let's just add `if (!bits) return 0;` to hash_bits() and be done
with it. Please keep
hashmap__for_each_key_entry_safe/hashmap__for_each_key_entry changes,
they are ok regardless.
> tools/lib/bpf/hashmap.c | 12 ++++++++++--
> tools/lib/bpf/hashmap.h | 12 ++++++------
> 2 files changed, 16 insertions(+), 8 deletions(-)
>
[...]
Powered by blists - more mailing lists