Community
Participate
Working Groups
Created attachment 257811 [details] Full log of stdout/stderr with kill -3 stack trace Latest Mars 4.5.1 from eclipse.org Fedora Linux 64-bit 21, 22, & 23 Oracle Java 1.8.0_66-b17 and OpenJDK 1.7, and OpenJDK 1.8 This only happens on one of my Fedora machines (HP Compaq 6710b with Intel i915 GPU), I have another machine, a Dell, with the same version of Fedora and Eclipse, where this problem does not occur. But on the machine it does occur on, it has been happening through several versions of Fedora and Eclipse. It also only happens after the first run. When Eclipse is installed, the installer process automatically starts Eclipse. During that run of Eclipse, it works fine. But, all subsequent runs after the first show this problem. The 'ps -mo pid lwp pcpu' command identifies the thread that is consuming the CPU. With that I can find the Java thread in the output of jstack <pid>. It is the main #1 thread and it is doing something with a native SWT function: "main" #1 prio=6 os_prio=0 tid=0x00007f04b0009800 nid=0x3514 runnable [0x00007f04b979c000] java.lang.Thread.State: RUNNABLE at org.eclipse.swt.internal.gtk.OS._gdk_window_get_origin(Native Method) at org.eclipse.swt.internal.gtk.OS.gdk_window_get_origin(OS.java:5814) at org.eclipse.swt.widgets.Control.getWindowOrigin(Control.java:5603) at org.eclipse.swt.widgets.Display.map(Display.java:3066) at org.eclipse.swt.widgets.Control.drawBackground(Control.java:132) at org.eclipse.swt.widgets.Control.windowProc(Control.java:5573) at org.eclipse.swt.widgets.Display.windowProc(Display.java:4717) at org.eclipse.swt.internal.gtk.OS._gtk_main_do_event(Native Method) at org.eclipse.swt.internal.gtk.OS.gtk_main_do_event(OS.java:9279) at org.eclipse.swt.widgets.Display.eventProc(Display.java:1225) at org.eclipse.swt.internal.gtk.OS._g_main_context_iteration(Native Method) (more in attached file)
Created attachment 257812 [details] Output of strace on the java/eclipse thread that is consuming the CPU This shows that it is polling something in a tight loop, each invocation results in an EAGAIN error, which means that whatever it's polling is not ready to send any data.
The Eclipse main window is displayed, but it is unusable because it is slow and some operations hang for minutes of indefinitely.
I believe that this also occurred under CentOS 7 on this same machine. I don't have Red Hat, but I could reinstall CentOS if that is closer to a "supported" platform than Fedora. My guess is that there is an issue between SWT and the Linux driver for the i915 graphics card.
Created attachment 258260 [details] output for #jstack -F 3351 Output form #jstack -F 3351
Comment on attachment 258260 [details] output for #jstack -F 3351 I added a file with the output. This is my kernel version: 4.2.0-1-amd64 #1 SMP Debian 4.2.5-1 (2015-10-27) x86_64 GNU/Linux I' running debian testing and I have this output: *-display description: VGA compatible controller product: Broadwell-U Integrated Graphics [8086:1616] vendor: Intel Corporation [8086] physical id: 2 bus info: pci@0000:00:02.0 version: 09 width: 64 bits clock: 33MHz capabilities: msi pm vga_controller bus_master cap_list rom configuration: driver=i915 latency=0 resources: irq:50 memory:c0000000-c0ffffff memory:b0000000-bfffffff ioport:4000(size=64) for #lshw -numeric -C display
Created attachment 260149 [details] output of jstack -F (multiple times) I am adding my output of jstack -F (run multiple times, just the relevant thread) My machine: Fedora 23 gtk3 v 3.18.7-2.fc23 java version "1.8.0_65" Java(TM) SE Runtime Environment (build 1.8.0_65-b17) Java HotSpot(TM) 64-Bit Server VM (build 25.65-b01, mixed mode) kernel 4.4.3-300.fc23.x86_64 $ sudo lshw -numeric -C display *-display description: VGA compatible controller product: Haswell-ULT Integrated Graphics Controller [8086:A16] vendor: Intel Corporation [8086] physical id: 2 bus info: pci@0000:00:02.0 version: 0b width: 64 bits clock: 33MHz capabilities: msi pm vga_controller bus_master cap_list rom configuration: driver=i915 latency=0 resources: irq:46 memory:e0000000-e03fffff memory:c0000000-dfffffff ioport:3000(size=64)
I just want to point out that the i915 graphics driver is common to all 3 cases being reported here by different people.
(In reply to Bill Chatfield from comment #7) > I just want to point out that the i915 graphics driver is common to all 3 > cases being reported here by different people. Bill I also have i915. And the same symptoms. I tried to localise the bug by some minimal example, w/o success.
Same issue here with Kubuntu 16.04 and HP 6710b.
Hi. I've just upgraded from Ubuntu 15.10 (Kernel 4.2.0-35) to Ubuntu 16.04 LTS (Kernel 4.4.0-21) and this bug has APPEARED. Eclipse is constantly having a "progress bar" for background jobs ins the lower right corner (but details about operation are absent) and it though does not prevent but greatly slows down all the work. And yes, I have i815. Eclipse 20160218-0600 If some details are needed - I will gladly post them - just ask.
Would you please try with lastest Neon I-build? There are significant changes in SWT so let's check whether it's reproducible with latest code.
I have just tried with neon M4 and it seems that it works well. Thanks!
I meant Neon M6.
Version: Neon Milestone 6 (4.6.0M6) Build id: 20160329-0659 Bug does not appear so far.
*** This bug has been marked as a duplicate of bug 478962 ***
