Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 328343

Summary:

[contentassist] 100% cpu on large javadoc

Product:

[Eclipse Project] Platform

Reporter:

P.V. Goddijn <pvgoddijn>

Component:

SWT

Assignee:

Platform-SWT-Inbox <platform-swt-inbox>

Status:

CLOSED WORKSFORME

QA Contact:

Severity:

normal

Priority:

CC:

contact, dominik.stadler, ericwill, kman, loskutov, tparker

Version:

3.6

Keywords:

triaged

Target Milestone:

---

Hardware:

OS:

Linux

Whiteboard:

Attachments:

Description	Flags
stack trace at beginning of problem (cpu @100%)	none
stack trace 2 minutes later (cpu @100%)	none
offending java file (generated by Jaxb)	none
new stacktrace for eclipse 3.7.0	none
Stacktrace when Eclipse does not respond	none
Javascript File which is minified (i.e. has all newlines removed)	none

Description P.V. Goddijn

2010-10-21 08:44:28 EDT

Build Identifier: 20100617-1415

when the contentassist tries to display an item with a very large javadoc,
CPU usage goes to 100% for several minutes.

eventualy the the porcess finishes and the tooltip is shown.
the tooltip is regular size and the contains the large amount of javadoc

jstack shows the offending process is:
org.eclipse.swt.internal.gtk.OS._pango_layout_get_line_count(Native Method)


Reproducible: Always

Steps to Reproduce:
1. Type name of offending class(non-unique)
2. use arrow key to scroll down to the offending class
3. when offending class is highlighted cpu --> 100%

Comment 1 P.V. Goddijn

2010-10-21 08:47:15 EDT

Created attachment 181386 [details]
stack trace at beginning of problem (cpu @100%)

Comment 2 P.V. Goddijn

2010-10-21 08:47:56 EDT

Created attachment 181387 [details]
stack trace 2 minutes later (cpu @100%)

Comment 3 P.V. Goddijn

2010-10-21 08:50:39 EDT

Created attachment 181388 [details]
offending java file (generated by Jaxb)

Comment 4 Terry Parker

2011-09-08 22:34:19 EDT

FYI

This sounds like a dup of 329288, but the attached stack traces don't line up with that.  Is this fixed in Eclpse 3.6.2 or 3.7.0?

Comment 5 P.V. Goddijn

2011-09-09 07:53:14 EDT

this is still present in eclipse 3.7.0

i'll upload a stacktrace for 3.7.0.

Comment 6 P.V. Goddijn

2011-09-09 07:54:09 EDT

Created attachment 203051 [details]
new stacktrace for eclipse 3.7.0

Comment 7 Dominik Stadler

2011-10-12 13:00:39 EDT

I have the same behavior with a minified .js file that has one very long line, stacktrace and js-file attached. This one takes half a minute on my machine to display and as soon as you do the slightest thing in Eclipse hangs again until the offending file is closed.

Comment 8 Dominik Stadler

2011-10-12 13:01:15 EDT

Created attachment 205051 [details]
Stacktrace when Eclipse does not respond

Comment 9 Dominik Stadler

2011-10-12 13:02:08 EDT

Created attachment 205052 [details]
Javascript File which is minified (i.e. has all newlines removed)

Comment 10 Alexander Kurtakov

2017-01-05 17:40:09 EST

Is this bug still reproducible for you? I can't reproduce here.

Comment 11 Andrey Loskutov

2017-05-16 02:53:27 EDT

*** Bug 385980 has been marked as a duplicate of this bug. ***

Comment 12 Andrey Loskutov

2017-05-16 02:59:40 EDT

(In reply to Alexander Kurtakov from comment #10)
> Is this bug still reproducible for you? I can't reproduce here.

I've just seen exact this on 4.6.3 RHEL 7.2 GTK3.14, 100% screen scaling. Eclipse froze forever with 100% CPU load (I've killed it after 5 minutes) trying to display a huge man page entry in opened shell script file editor.

jstack:
"main" #1 prio=6 os_prio=0 tid=0x00007f702800b000 nid=0xa7f runnable [0x00007f7030c65000]
   java.lang.Thread.State: RUNNABLE
	at org.eclipse.swt.internal.gtk.OS._pango_layout_get_line_count(Native Method)
	at org.eclipse.swt.internal.gtk.OS.pango_layout_get_line_count(OS.java:16636)
	at org.eclipse.swt.graphics.TextLayout.getLineBoundsInPixels(TextLayout.java:1036)
	at org.eclipse.swt.graphics.TextLayout.getLineBounds(TextLayout.java:1031)
	at org.eclipse.swt.graphics.TextLayout.getBounds(TextLayout.java:803)
	at org.eclipse.jface.internal.text.html.BrowserInformationControl.computeSizeHint(BrowserInformationControl.java:446)
	at org.eclipse.jface.text.AbstractInformationControlManager.internalShowInformationControl(AbstractInformationControlManager.java:1217)
	at org.eclipse.jface.text.AbstractInformationControlManager.presentInformation(AbstractInformationControlManager.java:1163)
	at org.eclipse.jface.text.AbstractHoverInformationControlManager.presentInformation(AbstractHoverInformationControlManager.java:894)
	at org.eclipse.jface.text.TextViewerHoverManager.doPresentInformation(TextViewerHoverManager.java:244)
	at org.eclipse.jface.text.TextViewerHoverManager$5.run(TextViewerHoverManager.java:234)
	at org.eclipse.swt.widgets.RunnableLock.run(RunnableLock.java:35)
	at org.eclipse.swt.widgets.Synchronizer.runAsyncMessages(Synchronizer.java:182)
	- locked <0x000000079c2c42d0> (a org.eclipse.swt.widgets.RunnableLock)
	at org.eclipse.swt.widgets.Display.runAsyncMessages(Display.java:4536)
	at org.eclipse.swt.widgets.Display.readAndDispatch(Display.java:4154)

ptack:
#0  0x00007f700d8074be in g_slice_alloc () at /lib64/libglib-2.0.so.0
#1  0x00007f700d80799e in g_slice_alloc0 () at /lib64/libglib-2.0.so.0
#2  0x00007f700df7ce14 in itemize_state_process_run () at /lib64/libpango-1.0.so.0
#3  0x00007f700df7e2c8 in pango_itemize_with_base_dir () at /lib64/libpango-1.0.so.0
#4  0x00007f700df8546a in pango_layout_check_lines () at /lib64/libpango-1.0.so.0
#5  0x00007f700df8689e in pango_layout_get_line_count () at /lib64/libpango-1.0.so.0
#6  0x00007f6ff195ab84 in Java_org_eclipse_swt_internal_gtk_OS__1pango_1layout_1get_1line_1count () at /home/aloskuto/.eclipse_ide/advantest_ide_RH7-463/eclipse/configuration/org.eclipse.osgi/404/0/.cp/libswt-pi3-gtk-4629.so
#7  0x00007f7018c20971 in  ()
#8  0x00007f7030c655e0 in  ()
#9  0x0000000000000000 in  ()

Comment 13 Eric Williams

2018-06-15 15:44:11 EDT

Sounds like a webkit1 hang. Is this still reproducible on GTK3.22 and 4.8?

Comment 14 Eric Williams

2018-07-04 11:15:07 EDT

(In reply to Eric Williams from comment #13)
> Sounds like a webkit1 hang. Is this still reproducible on GTK3.22 and 4.8?

Closing on account of this. Andrey if you still experience this issue on GTK3.22 and 4.8, please reopen this ticket.