Guy Harris
changed
bug 8629
| What |
Removed |
Added |
| Hardware |
x86
|
All
|
| OS |
Windows 7
|
All
|
Comment # 6
on bug 8629
from Guy Harris
We already have a preference to disable TCP checksum validation.
Is there a compelling reason not to use that to handle TCP checksum offloading,
i.e. if you have a capture with TCP checksums offloaded, so that (at least on
outgoing packets) the TCP checksum doesn't reflect what went out on the wire,
should that be handled by disabling TCP checksum offloading?
(In the long term, we should have direction information in packets, and "is
this interface doing {IPv4, IPv6, TCP, UDP, etc.} checksum offloading?"
information in pcap-NG, at least, and if that information is all present, we
could disable checksum checking automatically for the appropriate packets.
That requires OS changes as well as pcap changes - and pcap-ng changes to add
the offloading information - so that's not going to happen soon, and it still
doesn't handle captures *not* done on platforms where we can do that.)
Or, alternatively, have that preference have three values, one of which is "do
checksum validation but, if the checksum is zero, assume it was offloaded", and
the other two are the existing ones ("always validate" and "never validate").
You are receiving this mail because:
- You are watching all bug changes.
