[
https://issues.apache.org/jira/browse/PDFBOX-2901?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15628279#comment-15628279
]
Evan Sayer commented on PDFBOX-2901:
------------------------------------
Here's a stack-trace from a thread-dump of the stuck JVM:
{code}
"pf_test-emitter-0" #323 daemon prio=5 os_prio=0 tid=0x00007fc09839a000
nid=0x32a6 runnable [0x00007fbfad6bc000]
java.lang.Thread.State: RUNNABLE
at java.lang.System.identityHashCode(Native Method)
at java.util.HashMap$TreeNode.tieBreakOrder(HashMap.java:1896)
at java.util.HashMap$TreeNode.putTreeVal(HashMap.java:1992)
at java.util.HashMap.putVal(HashMap.java:637)
at java.util.HashMap.put(HashMap.java:611)
at
org.apache.pdfbox.pdmodel.graphics.shading.TriangleBasedShadingContext.calcPixelTable(TriangleBasedShadingContext.java:121)
at
org.apache.pdfbox.pdmodel.graphics.shading.GouraudShadingContext.calcPixelTable(GouraudShadingContext.java:111)
at
org.apache.pdfbox.pdmodel.graphics.shading.TriangleBasedShadingContext.createPixelTable(TriangleBasedShadingContext.java:80)
at
org.apache.pdfbox.pdmodel.graphics.shading.Type4ShadingContext.<init>(Type4ShadingContext.java:65)
at
org.apache.pdfbox.pdmodel.graphics.shading.Type4ShadingPaint.createContext(Type4ShadingPaint.java:66)
at sun.java2d.pipe.AlphaPaintPipe.startSequence(AlphaPaintPipe.java:84)
at sun.java2d.pipe.AAShapePipe.renderTiles(AAShapePipe.java:168)
at sun.java2d.pipe.AAShapePipe.renderPath(AAShapePipe.java:159)
at sun.java2d.pipe.AAShapePipe.fill(AAShapePipe.java:68)
at
sun.java2d.pipe.PixelToParallelogramConverter.fill(PixelToParallelogramConverter.java:164)
at sun.java2d.pipe.ValidatePipe.fill(ValidatePipe.java:160)
at sun.java2d.SunGraphics2D.fill(SunGraphics2D.java:2527)
at
org.apache.pdfbox.rendering.PageDrawer.shadingFill(PageDrawer.java:921)
at
org.apache.pdfbox.contentstream.operator.graphics.ShadingFill.process(ShadingFill.java:41)
at
org.apache.pdfbox.contentstream.PDFStreamEngine.processOperator(PDFStreamEngine.java:815)
at
org.apache.pdfbox.contentstream.PDFStreamEngine.processStreamOperators(PDFStreamEngine.java:472)
at
org.apache.pdfbox.contentstream.PDFStreamEngine.processTilingPattern(PDFStreamEngine.java:371)
at
org.apache.pdfbox.rendering.PageDrawer.drawTilingPattern(PageDrawer.java:222)
at
org.apache.pdfbox.rendering.TilingPaint.getImage(TilingPaint.java:171)
at org.apache.pdfbox.rendering.TilingPaint.<init>(TilingPaint.java:69)
at org.apache.pdfbox.rendering.PageDrawer.getPaint(PageDrawer.java:251)
at
org.apache.pdfbox.rendering.PageDrawer.getNonStrokingPaint(PageDrawer.java:529)
at org.apache.pdfbox.rendering.PageDrawer.fillPath(PageDrawer.java:588)
at
org.apache.pdfbox.contentstream.operator.graphics.FillEvenOddRule.process(FillEvenOddRule.java:36)
at
org.apache.pdfbox.contentstream.PDFStreamEngine.processOperator(PDFStreamEngine.java:815)
at
org.apache.pdfbox.contentstream.PDFStreamEngine.processStreamOperators(PDFStreamEngine.java:472)
at
org.apache.pdfbox.contentstream.PDFStreamEngine.processStream(PDFStreamEngine.java:446)
at
org.apache.pdfbox.contentstream.PDFStreamEngine.processPage(PDFStreamEngine.java:149)
at org.apache.pdfbox.rendering.PageDrawer.drawPage(PageDrawer.java:189)
at
org.apache.pdfbox.rendering.PDFRenderer.renderPage(PDFRenderer.java:208)
at
org.apache.pdfbox.rendering.PDFRenderer.renderImage(PDFRenderer.java:139)
at
org.apache.pdfbox.rendering.PDFRenderer.renderImageWithDPI(PDFRenderer.java:94)
{code}
{{PageDrawer.getNonStrokingPaint()}} can lead to that same shading logic it
seems, but that call happens before the patch here is applied to clip the
shaded area. Or that's my understanding, any assistance/advice is much
appreciated. :-)
> High CPU load and OutOfMemoryError when rendering shading
> ---------------------------------------------------------
>
> Key: PDFBOX-2901
> URL: https://issues.apache.org/jira/browse/PDFBOX-2901
> Project: PDFBox
> Issue Type: Bug
> Components: Rendering
> Affects Versions: 1.8.9, 1.8.10, 1.8.11, 2.0.0
> Reporter: Tilman Hausherr
> Assignee: Tilman Hausherr
> Fix For: 1.8.11, 2.0.0
>
> Attachments: PDFBOX-2901-outofmemory.pdf
>
>
> By Frank D. from the mailing list:
> {quote}
> When we try to convert the attached pdf, the CPU load of tomcat is raising
> and it seems, that the process hangs. The tomcat process is no more
> responsive and after a long while, we get an memory overflow. Also the server
> load is very high meanwhile.
> {quote}
> Some debugging at 50 dpi:
> {code}
> PDFOperator{cs}: [COSName{CS0}]
> PDFOperator{scn}: [COSName{P0}]
> PDFOperator{gs}: [COSName{GS0}]
> PDFOperator{m}: [COSFloat{-10551.00391}, COSFloat{667.6051}]
> PDFOperator{l}: [COSFloat{-10551.00391}, COSFloat{1447.09143}]
> PDFOperator{l}: [COSFloat{9338.57129}, COSFloat{1447.09143}]
> PDFOperator{l}: [COSFloat{9338.57129}, COSFloat{667.6051}]
> PDFOperator{f}: []
> f = Fill path using non zero winding rule
> (...wait...)
> PDFOperator{cs}: [COSName{CS0}]
> PDFOperator{scn}: [COSName{P0}]
> PDFOperator{gs}: [COSName{GS0}]
> PDFOperator{m}: [COSFloat{-18801.79492}, COSFloat{667.6051}]
> PDFOperator{l}: [COSFloat{-18801.79492}, COSFloat{1447.09143}]
> PDFOperator{l}: [COSFloat{1087.7804}, COSFloat{1447.09143}]
> PDFOperator{l}: [COSFloat{1087.7804}, COSFloat{667.6051}]
> PDFOperator{f}: []
> (...wait...)
> PDFOperator{cs}: [COSName{CS0}]
> PDFOperator{scn}: [COSName{P0}]
> PDFOperator{gs}: [COSName{GS0}]
> PDFOperator{m}: [COSFloat{-18801.79492}, COSFloat{11.15869}]
> PDFOperator{l}: [COSFloat{-18801.79492}, COSFloat{790.64502}]
> PDFOperator{l}: [COSFloat{1087.7804}, COSFloat{790.64502}]
> PDFOperator{l}: [COSFloat{1087.7804}, COSFloat{11.15869}]
> PDFOperator{f}: []
> (...wait...)
> (...wait...)
> (...wait...)
> PDFOperator{cs}: [COSName{CS0}]
> PDFOperator{scn}: [COSName{P0}]
> PDFOperator{gs}: [COSName{GS0}]
> PDFOperator{m}: [COSFloat{-10551.00391}, COSFloat{667.6051}]
> PDFOperator{l}: [COSFloat{-10551.00391}, COSFloat{1447.09143}]
> PDFOperator{l}: [COSFloat{9338.57129}, COSFloat{1447.09143}]
> PDFOperator{l}: [COSFloat{9338.57129}, COSFloat{667.6051}]
> PDFOperator{f}: []
> Exception in thread "main" java.lang.OutOfMemoryError: GC overhead limit
> exceeded
> at sun.java2d.cmm.kcms.CMMImageLayout.<init>(Unknown Source)
> at sun.java2d.cmm.kcms.ICC_Transform.colorConvert(Unknown Source)
> at java.awt.color.ICC_ColorSpace.toRGB(Unknown Source)
> at
> org.apache.pdfbox.pdmodel.graphics.color.PDDeviceRGB.toRGB(PDDeviceRGB.java:82)
> at
> org.apache.pdfbox.pdmodel.graphics.shading.ShadingContext.convertToRGB(ShadingContext.java:142)
> at
> org.apache.pdfbox.pdmodel.graphics.shading.TriangleBasedShadingContext.evalFunctionAndConvertToRGB(TriangleBasedShadingContext.java:167)
> at
> org.apache.pdfbox.pdmodel.graphics.shading.TriangleBasedShadingContext.calcPixelTable(TriangleBasedShadingContext.java:121)
> at
> org.apache.pdfbox.pdmodel.graphics.shading.GouraudShadingContext.calcPixelTable(GouraudShadingContext.java:111)
> at
> org.apache.pdfbox.pdmodel.graphics.shading.TriangleBasedShadingContext.createPixelTable(TriangleBasedShadingContext.java:80)
> at
> org.apache.pdfbox.pdmodel.graphics.shading.Type4ShadingContext.<init>(Type4ShadingContext.java:65)
> at
> org.apache.pdfbox.pdmodel.graphics.shading.Type4ShadingPaint.createContext(Type4ShadingPaint.java:66)
> at sun.java2d.pipe.AlphaPaintPipe.startSequence(Unknown Source)
> at sun.java2d.pipe.SpanShapeRenderer$Composite.startSequence(Unknown
> Source)
> at sun.java2d.pipe.SpanShapeRenderer.renderSpans(Unknown Source)
> at sun.java2d.pipe.SpanShapeRenderer.fill(Unknown Source)
> at sun.java2d.pipe.ValidatePipe.fill(Unknown Source)
> at sun.java2d.SunGraphics2D.fill(Unknown Source)
> at
> org.apache.pdfbox.rendering.PageDrawer.fillPath(PageDrawer.java:611)
> {code}
> The problem is that a type 4 shading ("gouraud shading") is called with huge
> triangles (not shown here but I saw it in debugging) and huge device bounds
> (e.g. x=-10551.004,y=667.6051,w=19889.574,h=779.4863). The shading routines
> of types 4-7 create a mapping table for the color of every point in a shading
> triangle. Because these triangles and the device bounds are so huge, the
> table is huge and takes a long time to be calculated, uses a lot of memory,
> with the result above.
> IMHO it is a java bug, it should consider the clipping when calculating the
> device bounds when calling Paint.createContext() and not pass device bounds
> outside of the clipping region. The solution will be (once again :-( ) to do
> ourselves what java doesn't.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
