JVM randomly crashes EXCEPTION_ACCESS_VIOLATION du

2019-08-19 01:59发布

We are developing multi-threaded java application using licensed Leadtools SDK 20 for converting large pdf documents to searchable pdf documents. During OCR process Java Virtual Machine randomly crashes - EXCEPTION_ACCESS_VIOLATION(Problematic frame: C [Ltocrx.dll+0x2af3a]).

I've noticed that it may happens if user try to cancel OCR process.

public class OcrProgressCallback implements OcrProgressListener {


    @Override
    public void onProgress(OcrProgressData ocrProgressData) {
        if (ocrProgressData.getStatus() != OcrProgressStatus.ABORT) {

            ocrDocument.ifPresent(doc -> {
                ProcessingStatus progressStatus = doc.getStatus();
                if (progressStatus == ProcessingStatus.CANCELED) {
                    ocrProgressData.setStatus(OcrProgressStatus.ABORT);
                    throw new RasterException(RasterExceptionCode.USER_ABORT);
                }
            });
        }
    }
}

Bellow attached log file:

#  EXCEPTION_ACCESS_VIOLATION

Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C  [Ltocrx.dll+0x2af3a]
C  [Ltocrx.dll+0x13d94]
C  0x00000000020c8c67

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j  leadtools.ocr.Ltocr.OcrPageDestroy(J)I+0
j  leadtools.ocr.OcrPage.dispose()V+52
j  leadtools.ocr.OcrPage.finalize()V+1
J 2364% C2 java.lang.ref.Finalizer$FinalizerThread.run()V (55 bytes) @ 0x0000000002738ac4 [0x0000000002738820+0x2a4]
v  ~StubRoutines::call_stub
        OcrEngine ocrEngine = OcrEngineManager.createEngine(OcrEngineType.LEAD);
        ocrEngine.startup(null, null, null, ocrRuntimePath);
        setInitialOcrConfiguration(processingDocument, ocrEngine);

        RasterCodecs rasterCodecs = ocrEngine.getRasterCodecsInstance();
        rasterCodecs.startOptimizedLoad();
        OcrDocument ocrDocument = ocrEngine.getDocumentManager()
                .createDocument(null, OcrCreateDocumentOptions.AUTO_DELETE_FILE.getValue());
        OcrProgressCallback ocrProgressCallback = new OcrProgressCallback(processingDocumentRepository, processingDocument);

        Path inputPath = ....
        Path outputPath = generateOutputDocumentPath();

        try {
            int pageCount = rasterCodecs.getTotalPages(inputPath.toString());
            IntStream.rangeClosed(1, pageCount)
                    .forEach(pageNumber -> {

                        RasterImage rasterImage = rasterCodecs.load(inputPath.toString(), pageNumber);
                        OcrPage ocrPage = ocrEngine.createPage(rasterImage, OcrImageSharingMode.AUTO_DISPOSE);

                        ocrPage.autoZone(null);
                        ocrPage.recognize(ocrProgressCallback);
                        ocrDocument.getPages().add(ocrPage);

                        rasterImage.dispose();
                        ocrPage.dispose();
                    });

            ocrDocument.save(outputPath, DocumentFormat.PDF, null);
        } finally {
            if(Objects.nonNull(rasterCodecs)) {
                rasterCodecs.detachOptimizedLoadData();
                rasterCodecs.stopOptimizedLoad();
            }
             if (Objects.nonNull(rasterCodecs)) {
               rasterCodecs.dispose();
             }

             if (Objects.nonNull(ocrDocument)) {
                ocrDocument.dispose();
             }

             ocrEngine.shutdown();
        }

0条回答
登录 后发表回答