KDE-Forum.org

I am running Kmail 1.8.2 (KDE 3.4.2 Level b SUSE 10.0). I have three accounts configured in Kmail, and whenever I try to fetch mail from my server from one of those accounts a get a dialog box with the heading "Invalid Response From Server", and the error "Unable to complete LIST operation." At that point I click "OK". Then when I exit Kmail I get a crash (SIGSEV).

I found a bug report from 2005 (99589) with status unconfirmed that is an exact duplicate of my issue, and that bug was a duplicate of 3 others that were unconfirmed. I updated that bug with new information 2/28/07.

To investigate I tried fetching from that account's server using a different mail client which works. I also snooped the packets exchanged between my laptop and the pop3 server, and the packets were valid.

I tried to upgrade my version of KDE to 3.5.6 to see if I could reproduce it there, but I ran into several problems with the upgrade (see thread from me in Installation forum), so I uninstalled kdepim, rebooted and reinstalled from my distibution disks. The problem is persistent.

I believe the crash is due to an uninitialized variable in the error path of the dialog result because the original bug identified problems when a packet was malformed (not true in my case).

Since I have deleted and reinstalled the application, I believe the root of my problem is a cache kept on my machine. I tried deleting and recreating that particular account, but the problem still persists.

My question is what files might I try clearing or rebuilding to fix that account? As I mentioned the "List.." error only happens on one account, and the others work fine, so the files must be specific to that account. I suspect it is one of the files in my .kde/socket-<host> directory. Does anyone know how those files are established and what they do?

Well I selectively cleaned out the files in .kde/socket-<host> directory, and that doesn't clear up the problem. I've looked into the .kderc file, but there is no obvious error there either. I have also verified that if I kill the kmail process after the "List..." error (which never completes) there is no seg fault (duh!) I guess I'll go back to the sniffer and see if I can find any difference in the protocol exchanges between the working accounts and the faulted one.

David

Gustl,

Thanks for the suggestion. I went back and tried to find something like you described, but could not (both find -follow and grep -r have no problems). I suspect there may be multiple root causes to this same symptom. I did run Kmail again, and noticed that the kio_pop3 process seems to be the one that generates the "List Error" dialog. Also, when the List error comes up it looks like the thread is hung on a read_nocancel which is consistent because if I try to cancel the operation it doesn't really change state. Without looking into it more deeply it seems like it could be a race condition that is exposed by a cached value. Your relocation of the directory may have removed that, or it may have been the circular link.

I have since changed to evolution for my mail application because I didn't have the time to spend on KDE to really debug it and because of the 3.5 build issues I had. I may pick up 3.5 with my next distro (looking at etch). If so I'll rebuild it then so that if the error comes up again (and given that it's been around in one form or another for several years that seems like a good bet) I'll have source to trace.

David