Comment # 3
on bug 12804
from Patrick
(In reply to Peter Wu from comment #2)
> I'm doubtful that this is caused by the SSL dissector, it is probably some
> tap callback that is invoked every 1 (or 3?) seconds.
>
> Can you reproduce this with no capture file open, or a different capture
> file? Can you reproduce this with the Wireshark GTK+ (Legacy) GUI?
Yeah, it might not be the dissector (I am not as familiar with WS under the
hood) but it definitely correlates with packets that contain that protocol.
It reproduces with normal, legacy, and tshark. When you open that packet,
memory use jumps by 200MB or so. My reproduction steps include the following
1. Open a resource monitor (I used the resource tab on task manager in windows
7)
2. Start up wireshark
3. Let wireshark settle its memory and note the amount there
4. Open up the packet attached to the case and observe the increase
The reason I found this was opening a 10MB pcap that included these packets I
experienced what appeared to be a runaway memory leak where the memory used by
wireshark increased quickly to an untenable point. My observation was that
wireshark would increase its memory allocated by 1GB a minute. I split the
packet capture up until I could determine which packets did and did not take
time to load and tracked memory usage.
You are receiving this mail because:
- You are watching all bug changes.
