After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 384183 - Sometimes "hangs" on Send & Receive
Sometimes "hangs" on Send & Receive
Status: RESOLVED FIXED
Product: evolution
Classification: Applications
Component: Mailer
2.10.x (obsolete)
Other All
: Urgent critical
: ---
Assigned To: evolution-mail-maintainers
Evolution QA team
Depends on:
Blocks:
 
 
Reported: 2006-12-09 22:03 UTC by Mikael Nilsson
Modified: 2013-09-13 00:50 UTC
See Also:
GNOME target: ---
GNOME version: 2.17/2.18


Attachments
The dialog when "hung" (31.91 KB, image/png)
2006-12-09 22:04 UTC, Mikael Nilsson
  Details
Fix race condition in EMsgPort that can result in lost messages (1.23 KB, patch)
2007-01-10 14:26 UTC, Alexander Larsson
none Details | Review

Description Mikael Nilsson 2006-12-09 22:03:39 UTC
Please describe the problem:
Sometimes when pressing the Send & Receive button, the dialog will not go away.

I can close it [X] and continue working, but I cannot cancel it, and whenever I try again, it "hangs" in the same way again.

Steps to reproduce:
1. Press the Send & Receive button repeatedly




Actual results:
The dialog will not go away

Expected results:
The dialog should close normally, or at least be cancellable.

Does this happen every time?
No, I have to press the button repeatedly. But it's easy to reproduce

Other information:
I seems to hang on smtp... see attachment.

I'm using 2.9.3 on Ubuntu Feisty.
Comment 1 Mikael Nilsson 2006-12-09 22:04:11 UTC
Created attachment 78055 [details]
The dialog when "hung"
Comment 2 Alexander Larsson 2007-01-10 14:22:06 UTC
We're seeing this a lot in RHEL/FC6 too. Bug report at:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=217568
Comment 3 Alexander Larsson 2007-01-10 14:23:58 UTC
This is caused by a pretty bad race condition in the new e-msgport implementation, it can sometimes lose messages. Attaching patch to fix it.
Comment 4 Alexander Larsson 2007-01-10 14:26:34 UTC
Created attachment 79949 [details] [review]
Fix race condition in EMsgPort that can result in lost messages
Comment 5 Alexander Larsson 2007-01-10 14:40:02 UTC
(see rh bug for analysis)
Comment 6 Matthew Barnes 2007-01-10 14:51:51 UTC
I believe the patch in bug #359979 would also fix this.

It changes the e_msgport_wait() semantics from "block-and-peek" to "block-and-pop".
Comment 7 Matthew Barnes 2007-01-10 15:32:12 UTC
I strongly recommend Alex's patch for Evo 2.8.3 since it maintains semantic API compatibility.  CC'ing harish.
Comment 8 Veerapuram Varadhan 2007-02-07 15:11:35 UTC
Please commit the patch to stable branch.
Comment 9 Matthew Barnes 2007-02-07 18:29:22 UTC
Rats, I just realized that the stable branch does not use the new EMsgPort implementation from bug #348888 upon which this patch is based.  I remember now that we held off on committing the new implementation until 2.9 because it broke ABI.  (I did add it to Fedora Core 6 though, that's what was confusing me.)

I guess that means bug #348888 (massive file descriptor consumption) is still present in the stable branch!

For the development branch, the patch in bug #359979 is the preferred solution for this bug.  So I'm marking the patch here as obsolete and will close this bug once the other patch is committed.
Comment 10 Matthew Barnes 2007-02-07 22:01:39 UTC
Committed the patch in bug #359979, which fixes this bug.