harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alexey Varlamov" <alexey.v.varla...@gmail.com>
Subject Re: [classlib][nio] we may add Re: [jira] Commented: (HARMONY-3591) [classlib][nio] Can't run simple 3D HWA application using lwjgl library
Date Fri, 13 Apr 2007 09:01:18 GMT
Please go ahead.

2007/4/13, LvJimmy,Jing <firepure@gmail.com>:
> That's for sure :)
> My plan was:
> 1. create those DirectBufferAdapters
> 2. in the  BaseByteBuffer, return the wrapped DirectBufferAdapters if and only
> if the ByteBuffer itself is direct, otherwise return the original adapters.
> I'm still preparing for the patch, and will run regression test to
> make sure it works.
> Does it make sense?  Thanks :)
>
> 2007/4/13, Mikhail Fursov <mike.fursov@gmail.com>:
> > Jimmy,
> > Will the initial test work with your solution?
> >
> > On 4/13/07, LvJimmy,Jing <firepure@gmail.com> wrote:
> > >
> > > Hi,
> > >    I've just read the patch for this JIRA, I find it workable, but may
> > > need some enhancement for:
> > >    1. IntToByteBufferAdapter (and all related adapters) may not be
> > > DirectBuffer, let it implement DirectBuffer does not make sense.
> > >    2. all DirectBuffer methods implemented in these classes, e.g,
> > > getEffectiveAddress(), check if the ByteBuffer is DirectBuffer,
> > > otherwise return null . These do not make sense for those in-direct
> > > bytebuffer.
> > >    So I suggest we create other classes implements DirectBuffer as a
> > > adapter, e.g, names DirectIntToByteBufferAdapter, so that we can
> > > return this adapter as a DirectBuffer reasonably. What's more, we may
> > > do some performance tuning for DirectBuffer.
> > >    If no objection, I'd like to add a patch for this(raise another JIRA?).
> > >
> > > 2007/4/13, Mikhail Fursov (JIRA) <jira@apache.org>:
> > > >
> > > >     [
> > > https://issues.apache.org/jira/browse/HARMONY-3591?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12488465]
> > > >
> > > > Mikhail Fursov commented on HARMONY-3591:
> > > > -----------------------------------------
> > > >
> > > > >need to raise this question on dev-list@.
> > > > Yes, please, start the thread. I do not feel comfortable to do it by
> > > myself because I will limit my participation only with email posts while
> > > doing real tasks on JIT side.
> > > >
> > > > > [classlib][nio] Can't run simple 3D HWA application using lwjgl
> > > library
> > > > >
> > > -----------------------------------------------------------------------
> > > > >
> > > > >                 Key: HARMONY-3591
> > > > >                 URL:
> > > https://issues.apache.org/jira/browse/HARMONY-3591
> > > > >             Project: Harmony
> > > > >          Issue Type: Bug
> > > > >          Components: Classlib
> > > > >            Reporter: Mikhail Fursov
> > > > >         Assigned To: Alexey Varlamov
> > > > >         Attachments: lwjgl.zip, nio_view.diff
> > > > >
> > > > >
> > > > > I've tried to run simple 3D HWA application using popular openGL
Java
> > > wrapper: lwjgl library and found that Harmony crashes at a very early stage
> > > of lwjgl initialization.
> > > > > This crash prevents even simple scenarios to be run.
> > > > > I found that the crash is reproduced for every JIT or interpreter
mode
> > > we have.
> > > > > HereI think that the problem is with DLRVM threading support or with
> > > Classlibs NIO implementation.
> > > > > Here is the list of the last methods compiled before crash:
> > > > > EM: compile start:[CS_OPT n=980]
> > > java/nio/DirectByteBuffer::get([BII)Ljava/nio/ByteBuffer;
> > > > > EM: compile done:[CS_OPT n=980: OK]
> > > java/nio/DirectByteBuffer::get([BII)Ljava/nio/ByteBuffer;
> > > > > EM: compile start:[CS_OPT n=981]
> > > org/apache/harmony/luni/platform/PlatformAddress::getByteArray(I[BII)V
> > > > > EM: compile done:[CS_OPT n=981: OK]
> > > org/apache/harmony/luni/platform/PlatformAddress::getByteArray(I[BII)V
> > > > > EM: compile start:[CS_OPT n=982]
> > > org/lwjgl/opengl/WindowsDisplay::setIcon([Ljava/nio/ByteBuffer;)I
> > > > > EM: compile done:[CS_OPT n=982: OK]
> > > org/lwjgl/opengl/WindowsDisplay::setIcon([Ljava/nio/ByteBuffer;)I
> > > > > Here is stacktrace I've got on  Windows:
> > > > > >     lwjgl.dll!01865f90()
> > > > >       lwjgl.dll!0186620e()
> > > > >       harmonyvm.dll!m2n_free_local_handles()  Line 267 + 0xc  C++
> > > > >       harmonyvm.dll!get_vm_thread(HyThread * thr=0x011fdfa0)  Line
190
> > > + 0xb  C++
> > > > >       011fe010()
> > > > >       harmonyvm.dll!vm_invoke_native_array_stub(unsigned int *
> > > args=0x0013f794, int sz=1, void * f=0x01800010)  Line 77       C++
> > > > >       harmonyvm.dll!JIT_execute_method_default(void * jit=0x00000000,
> > > _jmethodID * methodID=0x03105890, jvalue * return_value=0x00000000, jvalue
*
> > > args=0x02ffa948)  Line 200 C++
> > > > >       em.dll!DrlEMImpl::executeMethod(_jmethodID * meth=0x03105890,
> > > jvalue * return_value=0x00000000, jvalue * args=0x02ffa948)  Line 539 +
> > > 0x16      C++
> > > > >       em.dll!ExecuteMethod(_jmethodID * meth=0x03105890, jvalue *
> > > return_value=0x00000000, jvalue * args=0x02ffa948)  Line 43 + 0x21  C++
> > > > >       harmonyvm.dll!vm_execute_java_method_array(_jmethodID *
> > > method=0x03105890, jvalue * result=0x00000000, jvalue *
> > > args=0x02ffa948)  Line 51 + 0x1b        C++
> > > > >       harmonyvm.dll!call_static_method_no_ref_result(JNIEnv_External
*
> > > jni_env=0x011fe050, _jobject * clazz=0x03103ee0, _jmethodID *
> > > methodID=0x03105890, jvalue * args=0x02ffa948, jvalue *
> > > result=0x00000000)  Line 1154 + 0x11     C++
> > > > >       harmonyvm.dll!CallStaticVoidMethodA(JNIEnv_External *
> > > jni_env=0x011fe050, _jobject * clazz=0x03103ee0, _jmethodID *
> > > methodID=0x03105890, jvalue * args=0x02ffa948)  Line 1562 + 0x17    C++
> > > > >       harmonyvm.dll!CallStaticVoidMethodV(JNIEnv_External *
> > > jni_env=0x011fe050, _jobject * clazz=0x03103ee0, _jmethodID *
> > > methodID=0x03105890, char * args=0x0013fa18)  Line 1547     C++
> > > > >       harmonyvm.dll!CallStaticVoidMethod(JNIEnv_External *
> > > jni_env=0x011fe050, _jobject * clazz=0x03103ee0, _jmethodID *
> > > methodID=0x03105890, ...)  Line 1537 C++
> > > > >       java.exe!main_runJavaMain(const JNINativeInterface_ * *
> > > env=0x011fe050, char * mainClassName=0x00366f79, int nameIsUTF=0, int
> > > java_argc=0, char * * java_argv=0x0036247c, HyPortLibrary *
> > > portLibrary=0x0013fbb8)  Line 1311 + 0x1b     C
> > > > >       java.exe!invocation(HyPortLibrary * portLibrary=0x0013fbb8,
int
> > > argc=7, char * * argv=0x00362460, unsigned int handle=5308416, int
> > > version=65540, unsigned char ignoreUnrecognized='', char *
> > > mainClass=0x00366f79, unsigned int classArg=6, char *
> > > propertiesFileName=0x0015db60, int isStandaloneJar=0, char *
> > > vmdllsubdir=0x0013fb44)  Line 732 + 0x2c       C
> > > > >       java.exe!gpProtectedMain(haCmdlineOptions *
> > > args=0x0013fb90)  Line 381 + 0x33   C
> > > > >       java.exe!main(int argc=7, char * * argv=0x00362460, char *
*
> > > envp=0x00364180)  Line 147 + 0xc   C
> > > > >       java.exe!mainCRTStartup()  Line 398 + 0xe       C
> > > > > I checked the same test with SUN and BEA VMs.
> > > > > Sun1.4/1.5/1.6 and BEA1.4 runs OK.
> > > > > BEA 1.5 crashes.
> > > >
> > > > --
> > > > This message is automatically generated by JIRA.
> > > > -
> > > > You can reply to this email to add a comment to the issue online.
> > > >
> > > >
> > >
> > >
> > > --
> > >
> > > Best Regards!
> > >
> > > Jimmy, Jing Lv
> > > China Software Development Lab, IBM
> > >
> >
> >
> >
> > --
> > Mikhail Fursov
> >
>
>
> --
>
> Best Regards!
>
> Jimmy, Jing Lv
> China Software Development Lab, IBM
>

Mime
View raw message