Wireshark-bugs: [Wireshark-bugs] [Bug 6690] Wireshark quite slow displaying frames with many det
Date: Sun, 24 Jun 2012 09:45:09 -0700 (PDT)
https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=6690

--- Comment #12 from Jim Young <[email protected]> 2012-06-24 09:45:08 PDT ---
Hello Jakub,

bv07.patch looks good, real good!  

Of course the extra whitespace is gone, but I think you achieved the
performance goal that triggered this bug report.

Here's some observations regarding Wireshark performance BEFORE the bv07.patch
is applied...

Without bv07.patch applied I replicated the performance problem Bill reported
in Comment #1.  I tested with Attachment #5769 ("non-ssl file showing the same
behavior") from Bug #3737. 

Selecting Frame #4763 from this trace file triggers the CPU-intensive
performance issue.

On my MacBook Pro it consistently took about 37 seconds before the transient
"Processing: Packet Details" window closed.  Interestingly during most of this
time there were no updates to the transient "Processing: Packet Details"
window, it just stalled with "00:00" until the last second or so where it
reported an Elapsed Time: field changed from "00:00" to "00:35".

Selecting the "Reassembled TCP" or "De-chunked body" tabs and then clicking on
various locations within the hexpane will again cause the transient
"Processing: Packet details" window to appear for various times from a few
seconds to 30 or more seconds.   

>From within the packet-details pane as I selected new detail items each would
each trigger the transient "Processing: Packet details" window. 

Interestingly when quitting Wireshark it would sometimes stall 7 to 10 seconds
(or more) before the window closes.  Sometimes on exit Wireshark would trigger
the following message.

> bash-3.2$ ./wireshark &
> [1] 36235
> bash-3.2$ 
> (wireshark:36235): GLib-CRITICAL **: g_array_free: assertion `array' failed
> 
> [1]+  Done                    ./wireshark
> bash-3.2$ 

In repeated attempts I could not trigger the above "g_array_free: assertion"
message if I did NOT first select Frame #4673 from Attachment #5769.

Here's come observations regarding Wireshark AFTER applying the bv07.patch...

With bv07.patch applied I could not replicate the performance problem Bill
reported in Comment #1.  Again I tested using Attachment #5769 ("non-ssl file
showing the same behavior") from Bug #3737. And again I used Frame #4763 from
Attachment #5769 to test the new ui/gtk/bytes_view.c behavior.

On my MacBook Pro selecting Frame #4763 displayed the packet immediately with
no transient "Processing: Packet details" window displayed.

Selecting the "Reassembed TCP" or "De-chunked body" tabs and then clicking on
various locations within the hexpane immediately highlighted the selected bytes
again with no transient "Processing: Packet" details window.

>From within the packet-details pane as I selected new detail items each item
would be highlighted virtually immediately with no transient "Processing:
Packet details" window. 

With repeated attempts on program exit I could not replicate the "g_array_free:
assertion" problem intermittently seen without bv07.patch applied after having
selected frame #4673 from Attachment #5769.

-- 
Configure bugmail: https://bugs.wireshark.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching all bug changes.