On Fri, 2011-04-01 at 20:11 -0700, Guy Harris wrote:
> On Apr 1, 2011, at 6:03 PM, Rick Jones wrote:
> 
> > tcpdump 4.1.1, and 4.3.0-PRE-GIT_2011_04_01 prints just one expanded
> > counter sample per captured PDU because it mistakenly skips forward
> > sflow_sample_len when it has already adjusted tprt and tlen while it was
> > printing the sample contents. This then leaves it confused about what it
> > is seeing. Shifting the adjustment to the "default sample" case where
> > the sample wasn't printed appears to fix this, though there is still
> > some question as to whether it should advance by sflow_sample_len or
> > some adjustment thereof.
> 
> Actually, it should probably be checking whether sflow_sample_len is
>  too small for the sample; if it decrements sflow_sample_len as it
>  goes, after doing that check, that should also fix the same problem.
> 
> Does this do it?  (It also makes some white-space changes to make the
>  "tptr += ..." and "tlen -= ..." stuff consistent, and makes various
>  counts, types, and lengths unsigned.)

It seems to do it.

rick

-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.

Reply via email to