Wireshark-bugs: [Wireshark-bugs] [Bug 8515] Global variables in X.400 P1 dissector considered ha
Evan Huus
changed
bug 8515
What |
Removed |
Added |
CC |
|
eapache@gmail.com
|
Comment # 4
on bug 8515
from Evan Huus
(In reply to comment #3)
> Created attachment 11002 [details]
> Remove p1 global variables
>
> The attached patch removes all global varibles from the P1 dissector. It
> appears that all had "at most" a packet scope. Those that had "packet
> scope", I used p_get_proto_data() to keep the data. Those that had
> "sequence scope" (consecutive fields in a packet that needed to relay their
> information), I used the value_ptr.
> Not sure of the performance ramifications off all of the p_get_proto_data()
> calls, but it seemed the most "complete" solution. Some of the
> p_get_proto_data calls could be replaced by assigning the data to
> _asn1_ctx_t->private_data, if for instance, they had "sequence scope". I'll
> leave that to those more familiar with the dissector.
Performance of p_get_proto_data() gets worse with the number of protocols using
it in a frame. For the usual 5-10 protocols per frame it should be more than
fast enough.
> As a side note, I used x400-ping-refuse.pcap in my testing, and it has
> drastically different dissection results using "default" preferences between
> 1.8.x, 1.10.x, and the SVN. I reverted my changes on the SVN because I
> thought I has really messed something up, but even without my changes, the
> protocols detected vary per version.
That's odd - the files don't seem to have changed that much since 1.10...
You are receiving this mail because:
- You are watching all bug changes.