Hi again,
I now inserted printf-statements to see the value of the parameters
before and after passing them:
In X11TextRenderer: numGlpyhs:10, usePositions:0,
subpixPos:0, rgbOrder:0, lcdc:140, glypx:100.500000,
glyphy:400.500000, Images: -1376605040, NULL
In X11TextRenderer_md: numGlpyhs:10, usePositions:0, subpixPos:0,
rgbOrder:0, lcdc:140, glypx:0.000000, glyphy:3.392578, Images: 0
Positions:1081673728
So somehow it looks like passing the jfloats causes troubles, after
commenting them out, everything works as expected:
In X11TextRenderer: numGlpyhs:10, usePositions:0, subpixPos:0,
rgbOrder:0, lcdc:140, glypx:100.500000, glyphy:400.50000, Images:
-1376605792
In _md: numGlpyhs:10, usePositions:0, subpixPos:0, rgbOrder:0,
lcdc:140, glypx:-0.000000, glyphy:-0.000000, Images: -1376605792,
Positions:0
Any idea why passing the jfloats fail?
Thank you in advance, lg Clemens
2008/6/3 Clemens Eisserer <[EMAIL PROTECTED]>:
> Hi Dmitri,
>
>> Did you try to run it with -Xcheck:jni (preferably on a fastdebug
>> build)? What does it say?
> Thanks for the hint, it cleans the array-handle is not valid.
>
> I added printf-statements and Hotspot is of course right, the
> "original" object handle was != NULL, but the one I passed is zero.
> The strange thing is if I call a method in the same shared library
> (fontmanager.so) the handle is passed correctly, but when calling into
> libmawt.so the array-handle is NULL.
> I pass the array-handles by value, is this ok?
>
> For now I could work-arround that by simply calling
> GetPrimitiveArrayCritical in the first method, however thats somehow
> strange and I would like to understand where my mistake is to learn of
> my faults ;)
>
> Thanks a lot for your patience, Clemens
>
> One thing that also puzzles me is how the compiler knows about
> AWTXRDrawGlyphList?
> There's no header-file which specifies it, does the compiler guess?
> This is the how the code looks like:
>
> 1.) In libmawt.so the method which is called and crashes:
> void AWTXRDrawGlyphList
> (JNIEnv *env, jobject self,
> jlong dstData, jint numGlyphs, jboolean usePositions,
> jboolean subPixPos, jboolean rgbOrder, jint lcdContrast,
> jfloat glyphListOrigX, jfloat glyphListOrigY,
> jlongArray imgArray, jfloatArray posArray)
> {
> images = (jlong *) (*env)->GetPrimitiveArrayCritical(env, imgArray, NULL);
> }
>
> 2.)The method called from JNI in libfontmanager.so and a test-dummy method:
> JNIEXPORT void JNICALL Java_sun_font_X11TextRenderer_doDrawGlyphList
> //Method called from JNI
> (JNIEnv *env, jobject xtr,
> jlong dstData, jint totalGlyphs, jboolean usePositions,
> jboolean subPixPos, jboolean rgbOrder, jint lcdContrast,
> jfloat glyphListOrigX, jfloat glyphListOrigY,
> jlongArray images, jfloatArray positions)
> {
> TESTDrawGlyphList(env, xtr, dstData, totalGlyphs, usePositions,
> //Does not crash
> subPixPos, rgbOrder, lcdContrast, glyphListOrigX,
> glyphListOrigY, images, positions);
>
> AWTXRDrawGlyphList(env, xtr, dstData, totalGlyphs, usePositions,
> //Does crash, because images == NULL
> subPixPos, rgbOrder, lcdContrast, glyphListOrigX,
> glyphListOrigY, images, positions);
> }
>
> JNIEXPORT void TESTDrawGlyphList //DummyTestMethod
> (JNIEnv *env, jobject self,
> jlong dstData, jint numGlyphs, jboolean usePositions,
> jboolean subPixPos, jboolean rgbOrder, jint lcdContrast,
> jfloat glyphListOrigX, jfloat glyphListOrigY,
> jlongArray imgArray, jfloatArray posArray)
> {
> jlong *images;
> images = (jlong *) (*env)->GetPrimitiveArrayCritical(env, imgArray, NULL);
> }
>
> These are the outputs when I run the code with Xcheck:jni:
> Pointer: -1388390664 <- Value of imgArray-handle before passed to the
> 2nd method in libmawt.so
> Pointer: 0 <- Passed value
> FATAL ERROR in native method: Non-array passed to JNI array operations
>
> and this is without XCheck:jni:
> V [libjvm.so+0x24252b]
> C [libmawt.so+0x1fc81] AWTXRDrawGlyphList+0xa1
> C [libfontmanager.so+0xb393]
> Java_sun_font_X11TextRenderer_doDrawGlyphList+0x113
> j sun.font.X11TextRenderer.doDrawGlyphList(JIZZZIFF[J[F)V+0
>
> So the first call succeed (which basically does exectly the same, its
> just in another shared library), but the second time the parameter is
> not passed :-/
>
