Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20221203093740.218935-1-liuxin350@huawei.com> <6ac9f767-e7f5-6603-6234-97126ea22005@iogearbox.net>
In-Reply-To: <6ac9f767-e7f5-6603-6234-97126ea22005@iogearbox.net>
From:   Andrii Nakryiko <andrii.nakryiko@gmail.com>
Date:   Tue, 6 Dec 2022 16:00:45 -0800
Message-ID: <CAEf4BzaC6hhNzKkzFa+s4bws7APWj-Nk8Uup+3J6avCXnMFziA@mail.gmail.com>
Subject: Re: [PATCH bpf-next] libbpf: Optimized return value in
 libbpf_strerror when errno is libbpf errno
To:     Daniel Borkmann <daniel@iogearbox.net>
Cc:     Xin Liu <liuxin350@huawei.com>, andrii@kernel.org, ast@kernel.org,
        martin.lau@linux.dev, song@kernel.org, yhs@fb.com,
        john.fastabend@gmail.com, kpsingh@kernel.org, sdf@google.com,
        haoluo@google.com, jolsa@kernel.org, bpf@vger.kernel.org,
        linux-kernel@vger.kernel.org, yanan@huawei.com,
        wuchangye@huawei.com, xiesongyang@huawei.com,
        kongweibin2@huawei.com, zhangmingyi5@huawei.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Dec 5, 2022 at 1:11 PM Daniel Borkmann <daniel@iogearbox.net> wrote:
>
> On 12/3/22 10:37 AM, Xin Liu wrote:
> > This is a small improvement in libbpf_strerror. When libbpf_strerror
> > is used to obtain the system error description, if the length of the
> > buf is insufficient, libbpf_sterror returns ERANGE and sets errno to
> > ERANGE.
> >
> > However, this processing is not performed when the error code
> > customized by libbpf is obtained. Make some minor improvements here,
> > return -ERANGE and set errno to ERANGE when buf is not enough for
> > custom description.
> >
> > Signed-off-by: Xin Liu <liuxin350@huawei.com>
> > ---
> >   tools/lib/bpf/libbpf_errno.c | 6 ++++++
> >   1 file changed, 6 insertions(+)
> >
> > diff --git a/tools/lib/bpf/libbpf_errno.c b/tools/lib/bpf/libbpf_errno.c
> > index 96f67a772a1b..48ce7d5b5bf9 100644
> > --- a/tools/lib/bpf/libbpf_errno.c
> > +++ b/tools/lib/bpf/libbpf_errno.c
> > @@ -54,10 +54,16 @@ int libbpf_strerror(int err, char *buf, size_t size)
> >
> >       if (err < __LIBBPF_ERRNO__END) {
> >               const char *msg;
> > +             size_t msg_size;
> >
> >               msg = libbpf_strerror_table[ERRNO_OFFSET(err)];
> >               snprintf(buf, size, "%s", msg);
> >               buf[size - 1] = '\0';
> > +
> > +             msg_size = strlen(msg);
> > +             if (msg_size >= size)
> > +                     return libbpf_err(-ERANGE);
>
> Given this is related to libbpf_strerror_table[] where the error strings are known
> lets do compile-time error instead. All callers should pass in a buffer of STRERR_BUFSIZE
> size in libbpf.

That sounds a bit too pessimistic?.. If the actual error message fits
in the buffer, why return -ERANGE just because theoretically some
error descriptions might fit?

But I don't think we need to calculate strlen(). snprintf above
returns the number of bytes required to print a full string, even if
it was truncated. So just comparing snprintf's result to size should
be enough.

>
> >               return 0;
> >       }
> >
> >
>