Comment # 14
on bug 8629
from Pascal Quantin
(In reply to comment #12)
> Comment on attachment 10719 [details]
> Patch to indicate 0x0000 as a bad checksum but still allow reassembly.
>
> So is the original patch good enough then? This displays the computed
> checksum in the event that the checksum is 0x0000, but does not otherwise
> change the current desegmentation behavior.
It could be a good tradeoff. But the more I think about it, the more I believe
we should prevent reassembly in this case also (as CRC is invalid) and change
the option description like I suggested in comment #11.
But I might be wrong. Anyone else having an opinion?
(In reply to comment #13)
> I don't know enough about TCO and all of its implementations, but if TCO is
> enabled, is the TCP checksum that Wireshark sees always 0x0000? I doesn't
> seem so, based on the current comment in the Wireshark code. Assuming that
> it's not always going to be 0x0000, then I still think it could be useful to
> provide a way to allow reassembly to be attempted even if checksums are
> computed by Wireshark to be invalid. I don't think that disabling TCO would
> always be an option.
But in that case you can deactivate checksum verification, and in this case
reassembly will always be performed (that's the default configuration).
You are receiving this mail because:
- You are watching all bug changes.
