Re: [PATCH v3 01/13] net: eth: Add a helper to pad a short ethernet frame

From: Bin Meng <bmeng.cn@gmail.com>
To: "Philippe Mathieu-Daudé" <philmd@redhat.com>
Cc: Peter Maydell <peter.maydell@linaro.org>,
	Jason Wang <jasowang@redhat.com>,
	"qemu-devel@nongnu.org Developers" <qemu-devel@nongnu.org>
Subject: Re: [PATCH v3 01/13] net: eth: Add a helper to pad a short ethernet frame
Date: Tue, 16 Mar 2021 16:58:33 +0800	[thread overview]
Message-ID: <CAEUhbmUNN_s9kGvgn6gnFP191J9iu7y7Kq3Mwzgg=XCzxm0nuA@mail.gmail.com> (raw)
In-Reply-To: <cea98a59-0327-0a80-8a9f-1f4a8008cd4b@redhat.com>

On Tue, Mar 16, 2021 at 4:49 PM Philippe Mathieu-Daudé
<philmd@redhat.com> wrote:
>
> On 3/16/21 9:12 AM, Bin Meng wrote:
> > Add a helper to pad a short ethernet frame to the minimum required
> > length, which can be used by backend codes.
> >
> > Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
> >
> > ---
> >
> > Changes in v3:
> > - use 'without' instead of 'sans'
> > - add a helper to pad short frames
> >
> >  include/net/eth.h | 25 +++++++++++++++++++++++++
> >  1 file changed, 25 insertions(+)
> >
> > diff --git a/include/net/eth.h b/include/net/eth.h
> > index 0671be6916..bc064f8e52 100644
> > --- a/include/net/eth.h
> > +++ b/include/net/eth.h
> > @@ -31,6 +31,31 @@
> >
> >  #define ETH_ALEN 6
> >  #define ETH_HLEN 14
> > +#define ETH_ZLEN 60     /* Min. octets in frame without FCS */
> > +
> > +/**
> > + * pad_short_frame - pad a short frame to the minimum ethernet frame length
> > + *
> > + * If the ethernet frame size is shorter than 60 bytes, it will be padded to
> > + * 60 bytes at the address @min_pkt.
> > + *
> > + * @min_pkt: buffer address to hold the padded frame
> > + * @pkt: address to hold the original ethernet frame
> > + * @size: size of the original ethernet frame
> > + * @return true if the frame is padded, otherwise false
> > + */
> > +static inline bool pad_short_frame(uint8_t *min_pkt, const uint8_t *pkt,
> > +                                   int size)
> > +{
> > +    if (size < ETH_ZLEN) {
> > +        /* pad to minimum ethernet frame length */
> > +        memcpy(min_pkt, pkt, size);
> > +        memset(&min_pkt[size], 0, ETH_ZLEN - size);
> > +        return true;
> > +    }
> > +
> > +    return false;
> > +}
>
> I don't want to be too nitpicky but since I'm Cc'ed...
>
> - 'ethernet' -> 'Ethernet'
>
> - I'm not sure inlining is justified
>
> - The same function is used for 2 different operations,
>   . check if padding is required
>   . do the padding
>
> - If we provide a function a buffer to fill, we need to check the
>   buffer size is big enough to avoid overflow
>
> What about something like:
>
>   bool pad_short_frame(char *padded_pkt, size_t *padded_buflen,
>                        const void *pkt, size_t pkt_size);
>   {
>       assert(padded_buflen && *padded_buflen >= ETH_ZLEN);
>       if (src_size >= ETH_ZLEN) {
>           return false;
>       }
>       /* pad to minimum ethernet frame length */
>       memcpy(padded_pkt, pkt, pkt_size);
>       memset(&padded_pkt[pkt_size], 0, ETH_ZLEN - padded_buflen);
>       *padded_buflen = ETH_ZLEN;
>
>       return true;
>   }
>
> What do you think?

Looks good to me. Will update in the next version.

Regards,
Bin