Databrowser example for SWT crash on Ubuntu 13.10

[berryma4]

There seems to be a reproducible way to kill CS-Studio without an exception in the linux 64bit version 3.2.x (haven't tested others yet)

1. start with a new workspace
2. start cs-studio
3. click the new databrowser button
4. right click graph add pv
5. right click traces view add second pv
6. cs-studio dies

This is true with the FRIB and NSLSII 3.2.x products.

[kasemir]

Cannot reproduce here. Works fine with SNS product, 3.2.x branch, RedHat EL 6, adding "sim://sine" first, then "sim://ramp" via right-click, add in the traces view.

[berryma4]

Martin Konrad is going to further test (original issue on Ubuntu 13.10)

[mark0n]

I tried to reproduce it in a VM:

Debian Wheezy with LXDE: problem does not occur
Debian Wheezy with KDE: problem does not occur
Ubuntu 13.10 with KDE: CSS hangs when clicking data-browser button (cannot test)

But on my host system (Ubuntu 13.10 with KDE) it crashes after adding sim://sine and sim://ramp as well (no need for a real IOC).

[kasemir]

As for 'hang', send a "kill -QUIT" to the java process to get a stack trace.

Could it be that this is all releated to your archive setup, reading from xnds://...?

[mark0n]

I opened another ticked for the "hang" issue: #356

[berryma4]

Reminder, we were able to reproduce this with the nslsII product also.
On Jan 27, 2014 1:46 PM, "Martin Konrad" notifications@github.com wrote:

I opened another ticked for the "hang" issue: #356#356

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/350#issuecomment-33406812
.

[kasemir]

I've downloaded CSS-NSLSII-linux-3.2.13a.zip. Runs fine, just many warnings with -consoleLog because http://..sourceforce../updates has no repo, cannot connect to xnds://ca01...ArchiveDataServer.cgi.

Can your hangups be related to the printer hangup, i.e. you're on a computer with iffy network settings?
Or could it be an SWT issue: The way you add the second PV is just so that a tool-tip appears for some cell in the Data Browser Properties Traces table? Tool tips on Linux have caused crashes in the past, which usually show up in a hs_err_pid… file.

[mark0n]

As far as I can see the network settings on my host (where CS-Studio crashes) should be fine:

$ ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
link/ether 28:d2:44:0c:2b:ec brd ff:ff:ff:ff:ff:ff
inet 35.9.58.51/24 brd 35.9.58.255 scope global eth0
valid_lft forever preferred_lft forever
inet6 fe80::2ad2:44ff:fe0c:2bec/64 scope link
valid_lft forever preferred_lft forever
3: wlan0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN qlen 1000
link/ether 3c:a9:f4:26:d3:ac brd ff:ff:ff:ff:ff:ff
inet6 fe80::3ea9:f4ff:fe26:d3ac/64 scope link
valid_lft forever preferred_lft forever
9: vboxnet0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
link/ether 0a:00:27:00:00:00 brd ff:ff:ff:ff:ff:ff
10: vboxnet1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
link/ether 0a:00:27:00:00:01 brd ff:ff:ff:ff:ff:ff
11: vboxnet2: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
link/ether 0a:00:27:00:00:02 brd ff:ff:ff:ff:ff:ff
12: vboxnet3: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
link/ether 0a:00:27:00:00:03 brd ff:ff:ff:ff:ff:ff
13: vboxnet4: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN qlen 1000
link/ether 0a:00:27:00:00:04 brd ff:ff:ff:ff:ff:ff

$ ip route
default via 35.9.58.1 dev eth0 proto static
35.9.58.0/24 dev eth0 proto kernel scope link src 35.9.58.51 metric 1

[mark0n]

I tried again with the NSLS-II product. Here is some detailed information: https://gist.github.com/mark0n/2d3065ae23a6616eab9e

[mark0n]

Regarding the tool tips: I tried to add a single PV and tried the tool tips: They are working properly - no crash. Here is what I do to crash it:

right click into the traces area to add a first PV sim://ramp
right click into the traces area to add a second PV sim://sine (CS-Studio crashes immediately when I hit Enter/Ok)

[kasemir]

Good that you found a hs_err_pid file. It shows a stack trace ending in "org.eclipse.swt.internal.gtk.OS._g_object_get_qdata()".
There are reports about Ubuntu problems with SWT:
https://bugs.kde.org/show_bug.cgi?id=324438
https://bugs.launchpad.net/ubuntu/+source/unity/+bug/1241101

Is your problem limited to Ubuntu?

[mark0n]

I tried to reproduce this issue in a VM - without success. So for now it's limited to my Ubuntu 13.10 system with KDE (no Unity, no Gnome). It also crashes with

UBUNTU_MENUPROXY= /opt/css-nsls2/css-nsls2 -vmargs -Dorg.eclipse.swt.internal.gtk.disablePrinting

But setting System Settings -> Application Appearance -> GTK -> Select a GTK2 Theme makes a difference:
"Raleigh" -> works
"oxygen-gtk" -> crashes

[mark0n]

As a work around

GTK2_RC_FILES=/usr/share/themes/Raleigh/gtk-2.0/gtkrc /opt/css-nsls2/css-nsls2 -vmargs -Dorg.eclipse.swt.internal.gtk.disablePrinting

works.

[kasemir]

Very good that you can limit it to a specific GTK theme on Ubuntu!

https://bugs.kde.org/show_bug.cgi?id=324438 mentions getting a similar crash in swt.internal.gtk.OS::_g_object_get_qdata (native) by invoking a file-search in eclipse. The problem started after upgrading to Kubuntu 13.10, i.e. the same release that you mention.

Can you reproduce the crash by invoking the menu Search, File, …, "Search" from CSS, or - even better - the basic Eclipse IDE?