You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Andy Cutright <An...@borland.com> on 2005/04/05 20:15:30 UTC

javahl & AIX

hi all, 

is anyone actively testing javahl & the svnClientAdapter on AIX? i'm
getting a core dump due to a SIGILL and i'm wondering if this is
something i'm doing (oh so very likely) or a known issue. i don't see
anything mentioned in the issue tracker. i've been able to get this code
compiled & running on solaris/ hp-ux/ win XP. 

1XHFULLVERSION J2RE 1.4.1 IBM AIX build ca1411ifx-20041029a
(141SR3+78993)
NULL           
1XHCURRENTTHD  Current Thread Details
NULL           ----------------------
2XHCURRSYSTHD      "main" sys_thread_t:0x3000D928
3XHNATIVESTACK       Native Stack
NULL                 ------------
3XHSTACKLINE         at 0xD3DAB9B0 in JNIInit__7JNIUtilFP7JNIEnv_
3XHSTACKLINE         at 0xD3DC1E90 in
__ct__15JNIStackElementFP7JNIEnv_PCcPCcP8_jobject
3XHSTACKLINE         at 0xD3DE8B1C in
Java_org_tigris_subversion_javahl_SVNClient_ctNative
3XHSTACKLINE         at 0xD39FE914 in mmipInvoke_V_J
3XHSTACKLINE         at 0xD39FA210 in mmipInvokeLazyJniMethod
3XHSTACKLINE         at 0xD39C6370 in mmipExecuteJava
3XHSTACKLINE         at 0xD39BE92C in xeRunJniMethod
3XHSTACKLINE         at 0xD39F6F58 in invokeJniMethod
3XHSTACKLINE         at 0xD39F0D4C in jni_CallStaticVoidMethod
1XHOPENV       Operating Environment
NULL           ---------------------
2XHOSLEVEL     OS Level         : AIX 5.2.0.0
2XHCPUS        Processors -
3XHCPUARCH       Architecture   : POWER_PC (impl: POWER_630, ver:
PV_630)
3XHNUMCPUS       How Many       : 4
3XHCPUSENABLED   Enabled        : 4


i'm using the IBM xl* compilers, version 6. i can get subversion itself
running properly, but that's straight C code. there doesn't appear to be
much support in the javahl configuration for the xl* compilers, so i've
hacked up a makefile myself. it's quite possible my linkage is wrong,
and i'm exploring that. 

i'm a bit curious about the javahl implementation. i see a C++ object
being cast as a long & passed to the VM. does passing a C++ object this
way make an assumption about the underlying VM implementation? 

cheers,
andy 


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org


Re: javahl & AIX

Posted by Patrick Mayweg <ma...@qint.de>.
Hi Andy,

Andy Cutright wrote:

>hi all, 
>
>is anyone actively testing javahl & the svnClientAdapter on AIX? i'm
>getting a core dump due to a SIGILL and i'm wondering if this is
>something i'm doing (oh so very likely) or a known issue. i don't see
>anything mentioned in the issue tracker. i've been able to get this code
>compiled & running on solaris/ hp-ux/ win XP. 
>
>1XHFULLVERSION J2RE 1.4.1 IBM AIX build ca1411ifx-20041029a
>(141SR3+78993)
>NULL           
>1XHCURRENTTHD  Current Thread Details
>NULL           ----------------------
>2XHCURRSYSTHD      "main" sys_thread_t:0x3000D928
>3XHNATIVESTACK       Native Stack
>NULL                 ------------
>3XHSTACKLINE         at 0xD3DAB9B0 in JNIInit__7JNIUtilFP7JNIEnv_
>3XHSTACKLINE         at 0xD3DC1E90 in
>__ct__15JNIStackElementFP7JNIEnv_PCcPCcP8_jobject
>3XHSTACKLINE         at 0xD3DE8B1C in
>Java_org_tigris_subversion_javahl_SVNClient_ctNative
>3XHSTACKLINE         at 0xD39FE914 in mmipInvoke_V_J
>3XHSTACKLINE         at 0xD39FA210 in mmipInvokeLazyJniMethod
>3XHSTACKLINE         at 0xD39C6370 in mmipExecuteJava
>3XHSTACKLINE         at 0xD39BE92C in xeRunJniMethod
>3XHSTACKLINE         at 0xD39F6F58 in invokeJniMethod
>3XHSTACKLINE         at 0xD39F0D4C in jni_CallStaticVoidMethod
>1XHOPENV       Operating Environment
>NULL           ---------------------
>2XHOSLEVEL     OS Level         : AIX 5.2.0.0
>2XHCPUS        Processors -
>3XHCPUARCH       Architecture   : POWER_PC (impl: POWER_630, ver:
>PV_630)
>3XHNUMCPUS       How Many       : 4
>3XHCPUSENABLED   Enabled        : 4
>
>
>i'm using the IBM xl* compilers, version 6. i can get subversion itself
>running properly, but that's straight C code. there doesn't appear to be
>much support in the javahl configuration for the xl* compilers, so i've
>hacked up a makefile myself. it's quite possible my linkage is wrong,
>and i'm exploring that. 
>
>i'm a bit curious about the javahl implementation. i see a C++ object
>being cast as a long & passed to the VM. does passing a C++ object this
>way make an assumption about the underlying VM implementation? 
>  
>
It should not. I cast the pointer/address of the C++ object to a jlong, 
which is 64 bits long. That should be long enough even for a 64 bit 
machine. I am reversing the cast (jlong to SVNClient*) later to get the 
address of the C++ object back. Your crash is not in this area. It 
chrashes during JNIUtil::JNIInit. If you could build javahl with 
debugging information, maybe the stack trace will reveal the exact point 
of the crash. It is a time, since I last worked on AIX, so I am not 
current on any specifics.

>cheers,
>andy 
>
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
>For additional commands, e-mail: dev-help@subversion.tigris.org
>
>  
>
Patrick

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org