Comment # 6
on bug 11205
from Guy Harris
(In reply to jgoff from comment #5)
> @Guy
>
> I don't know why we didn't implement radiotap (will try to find out) but the
> easiest win I see on our side is to inject either a padded fake FCS for
> these packets (that we can handle at the dissector) or the AP can add a
> dummy FCS calculation that is correct at the time.
>
> The packet-capture code on Aruba side has a number of fields which are dummy
> filled for the specific case of "the capturing AP is also transmitting",
> hence we can easily handle it at the AP. As you note there is no way to flag
> this, other than perhaps a dummy 0x00 * 4 FCS inserted.
Doug seems to say you're already doing that:
> Specific case need is packet capture by Aruba Access Points (AP). An AP
> that is capturing packets cannot actually capture its own transmitted
> frames, it has to fork off a copy of the frame on its way down the protocol
> stack for transmission. At the time the copy is made some of the fields
> are not filled in, including the FCS. Aruba could compute an FCS for the
> copy but that would eat up CPU cycles and it wouldn't represent what was
> transmitted anyway. So they pad four null bytes to occupy the FCS space.
> Of course Wireshark then sees this as a "malformed packet" because the
> FCS is incorrect.
So, *if* that's what you're doing, we could check the FCS if it's non-zero but
not if it's zero. That wouldn't require a preference.
You are receiving this mail because:
- You are watching all bug changes.
