Re: Problems with libffi on PowerPC64

Re: Problems with libffi on PowerPC64

[Daniela Quaresima]

Thanks for answering so quickly. 
I tried to update libffi with the latest version from CVS and I made the tests again, but nothing changed.
If the result of the tests isn't so bad, does it mean that the problem could be somewhere else? 
One of the errors with the testsuite of JNA occurs when the function ffi_call is invoked and the expected result type is an int (smaller than 64 bit): the value returned is 0 instead of whatever it should be. 

----- Messaggio originale -----
Da: Andrew Haley <aph@redhat.com>
A: Daniela Quaresima <daniela_quaresima@yahoo.it>
Cc: libffi-discuss@sourceware.org
Inviato: Martedì 16 giugno 2009, 12:15:41
Oggetto: Re: Problems with libffi on PowerPC64

Daniela Quaresima wrote:
> 
> Hi! I'm having difficulties in using JNA on IBM (x series) ppc64 architecture with Linux SUSE SLES 9.
> It was suggested that maybe the problem was in the libffi used.
> I tried to run libffi testsuite as follows:
> -starting in the lidffi directory inside jna/native
> mkdir build; cd build; ../configure; make; make check
> 
> Some tests failed and so I wanted to know how I could resolve my problem and (hopefully) make JNA work.
> I also tried to install libffi (version 3.0.8) as a standalone, but the testsuite failed even in that case.
> 
> This is the result of the tests for the jna/native/libffi:

These results are pretty good.  We'd need to see the actual test log
to know the cause of the failures.

Andrew.

Re: Problems with libffi on PowerPC64

[Andrew Haley]

Daniela Quaresima wrote:
> Thanks for answering so quickly.

> I tried to update libffi with the latest version from CVS and I made
> the tests again, but nothing changed.

> If the result of the tests isn't so bad, does it mean that the
> problem could be somewhere else?

I think so.  You're only seeing a failure with long doubles (which
Java doesn't support), with a huge struct, and some sort of C++
exception compilation problem.  I would not expect any of these to
cause JNA failures.

> One of the errors with the testsuite of JNA occurs when the function
> ffi_call is invoked and the expected result type is an int (smaller
> than 64 bit): the value returned is 0 instead of whatever it should
> be.

I think the bug may be inside JNA itself.  The libffi test results
show no problems in this area.  I think you've got to start debugging.

Andrew.

Re: Problems with libffi on PowerPC64

[Daniela Quaresima]

I'm not really an expert on JNA, so I couldn't tell you if it uses closures or not. I tried to re-run the libffi testsuite after having applied the patch, but I got the same results as before. 
However probably my problems stem from JNA and not libffi, so I'm trying to analize the report of that testsuite to find the bug(s).

Thanks for your help all the same.
Daniela



----- Messaggio originale -----
Da: Anthony Green <green@redhat.com>
A: Daniela Quaresima <daniela_quaresima@yahoo.it>
Cc: libffi-discuss@sourceware.org; jakub@redhat.com
Inviato: Martedì 16 giugno 2009, 16:10:11
Oggetto: Re: Problems with libffi on PowerPC64

Daniela Quaresima wrote:
> Hi! I'm having difficulties in using JNA on IBM (x series) ppc64 architecture with Linux SUSE SLES 9.
> It was suggested that maybe the problem was in the libffi used.
>  
Does JNA use closures?  There's an outstanding PPC64 patch here:
http://sourceware.org/ml/libffi-discuss/2008/msg00112.html

Could you please try with this patch applied?

Jakub - I think you were the last to do work on the PPC64 port.  Could 
you please look at this patch?

Thanks,

AG

Re: Problems with libffi on PowerPC64

[Anthony Green]

Daniela Quaresima wrote:
> Hi! I'm having difficulties in using JNA on IBM (x series) ppc64 architecture with Linux SUSE SLES 9.
> It was suggested that maybe the problem was in the libffi used.
>   
Does JNA use closures?  There's an outstanding PPC64 patch here:
 http://sourceware.org/ml/libffi-discuss/2008/msg00112.html

Could you please try with this patch applied?

Jakub - I think you were the last to do work on the PPC64 port.  Could 
you please look at this patch?

Thanks,

AG




> I tried to run libffi testsuite as follows:
> -starting in the lidffi directory inside jna/native
> mkdir build; cd build; ../configure; make; make check
>
> Some tests failed and so I wanted to know how I could resolve my problem and (hopefully) make JNA work.
> I also tried to install libffi (version 3.0.8) as a standalone, but the testsuite failed even in that case.
>
> This is the result of the tests for the jna/native/libffi:
>
> Native configuration is powerpc64-unknown-linux-gnu
>
>                 === libffi tests ===
>
> Schedule of variations:
>     unix
>
> Running target unix
> Using /usr/local/share/dejagnu/baseboards/unix.exp as board description file for target.
> Using /usr/local/share/dejagnu/config/unix.exp as generic interface file for target.
> Using /home/vts/JNA_v3.10_modif_daniela/native/libffi/testsuite/config/default.exp as tool-and-target-specific interface file.
> Running /home/vts/JNA_v3.10_modif_daniela/native/libffi/testsuite/libffi.call/call.exp ...
> FAIL: libffi.call/cls_align_longdouble_split.c -O0 -W -Wall (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split.c -O0 -W -Wall compilation failed to produce executable
> FAIL: libffi.call/cls_align_longdouble_split2.c -O0 -W -Wall (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split2.c -O0 -W -Wall compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble.c -O0 -W -Wall (test for excess errors)
> WARNING: libffi.call/cls_longdouble.c -O0 -W -Wall compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble_va.c -O0 -W -Wall output pattern test, is %.1f
> res: 5
> %.1f
> res: 5
> ? should match 7.0
> ?es: 4
> ?.0
> res: 4
> FAIL: libffi.call/err_bad_abi.c -O0 -W -Wall execution test
> XPASS: libffi.call/float2.c -O0 -W -Wall (test for excess errors)
> WARNING: libffi.call/huge_struct.c -O0 -W -Wall compilation failed to produce executable
> FAIL: libffi.call/return_uc.c -O0 -W -Wall (test for excess errors)
> FAIL: libffi.call/cls_align_longdouble_split.c -O2 (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split.c -O2 compilation failed to produce executable
> FAIL: libffi.call/cls_align_longdouble_split2.c -O2 (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split2.c -O2 compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble.c -O2 (test for excess errors)
> WARNING: libffi.call/cls_longdouble.c -O2 compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble_va.c -O2 output pattern test, is %.1f
> res: 5
> %.1f
> res: 5
> ? should match 7.0
> ?es: 4
> ?.0
> res: 4
> FAIL: libffi.call/err_bad_abi.c -O2 execution test
> XPASS: libffi.call/float2.c -O2 (test for excess errors)
> WARNING: libffi.call/huge_struct.c -O2 compilation failed to produce executable
> FAIL: libffi.call/cls_align_longdouble_split.c -O3 (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split.c -O3 compilation failed to produce executable
> FAIL: libffi.call/cls_align_longdouble_split2.c -O3 (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split2.c -O3 compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble.c -O3 (test for excess errors)
> WARNING: libffi.call/cls_longdouble.c -O3 compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble_va.c -O3 output pattern test, is %.1f
> res: 5
> %.1f
> res: 5
> ? should match 7.0
> ?es: 4
> ?.0
> res: 4
> FAIL: libffi.call/err_bad_abi.c -O3 execution test
> XPASS: libffi.call/float2.c -O3 (test for excess errors)
> WARNING: libffi.call/huge_struct.c -O3 compilation failed to produce executable
> FAIL: libffi.call/cls_align_longdouble_split.c -Os (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split.c -Os compilation failed to produce executable
> FAIL: libffi.call/cls_align_longdouble_split2.c -Os (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split2.c -Os compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble.c -Os (test for excess errors)
> WARNING: libffi.call/cls_longdouble.c -Os compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble_va.c -Os output pattern test, is %.1f
> res: 5
> %.1f
> res: 5
> ? should match 7.0
> ?es: 4
> ?.0
> res: 4
> FAIL: libffi.call/err_bad_abi.c -Os execution test
> XPASS: libffi.call/float2.c -Os (test for excess errors)
> WARNING: libffi.call/huge_struct.c -Os compilation failed to produce executable
> FAIL: libffi.call/cls_align_longdouble_split.c -O2 -fomit-frame-pointer (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split.c -O2 -fomit-frame-pointer compilation failed to produce executable
> FAIL: libffi.call/cls_align_longdouble_split2.c -O2 -fomit-frame-pointer (test for excess errors)
> WARNING: libffi.call/cls_align_longdouble_split2.c -O2 -fomit-frame-pointer compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble.c -O2 -fomit-frame-pointer (test for excess errors)
> WARNING: libffi.call/cls_longdouble.c -O2 -fomit-frame-pointer compilation failed to produce executable
> FAIL: libffi.call/cls_longdouble_va.c -O2 -fomit-frame-pointer output pattern test, is %.1f
> res: 5
> %.1f
> res: 5
> ? should match 7.0
> ?es: 4
> ?.0
> res: 4
> FAIL: libffi.call/err_bad_abi.c -O2 -fomit-frame-pointer execution test
> XPASS: libffi.call/float2.c -O2 -fomit-frame-pointer (test for excess errors)
> WARNING: libffi.call/huge_struct.c -O2 -fomit-frame-pointer compilation failed to produce executable
> Running /home/vts/JNA_v3.10_modif_daniela/native/libffi/testsuite/libffi.special/special.exp ...
> FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ (test for excess errors)
> WARNING: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
> FAIL: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ (test for excess errors)
> WARNING: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
> FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ (test for excess errors)
> WARNING: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
> FAIL: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ (test for excess errors)
> WARNING: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
> FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ (test for excess errors)
> WARNING: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
> FAIL: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ (test for excess errors)
> WARNING: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
> FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ (test for excess errors)
> WARNING: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
> FAIL: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ (test for excess errors)
> WARNING: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
>
>                 === libffi Summary ===
>
> # of expected passes            1519
> # of unexpected failures        34
> # of unexpected successes       5
> # of expected failures          5
> # of unsupported tests          15 
>
> Can someone tell me what could be the problem and what steps should I take to resolve it?
>
> Thanks,
> Daniela Quaresima.
>
>
>
>   

Re: Problems with libffi on PowerPC64

[Andrew Haley]

Daniela Quaresima wrote:
> 
> Hi! I'm having difficulties in using JNA on IBM (x series) ppc64 architecture with Linux SUSE SLES 9.
> It was suggested that maybe the problem was in the libffi used.
> I tried to run libffi testsuite as follows:
> -starting in the lidffi directory inside jna/native
> mkdir build; cd build; ../configure; make; make check
> 
> Some tests failed and so I wanted to know how I could resolve my problem and (hopefully) make JNA work.
> I also tried to install libffi (version 3.0.8) as a standalone, but the testsuite failed even in that case.
> 
> This is the result of the tests for the jna/native/libffi:

These results are pretty good.  We'd need to see the actual test log
to know the cause of the failures.

Andrew.

Problems with libffi on PowerPC64

[Daniela Quaresima]

Hi! I'm having difficulties in using JNA on IBM (x series) ppc64 architecture with Linux SUSE SLES 9.
It was suggested that maybe the problem was in the libffi used.
I tried to run libffi testsuite as follows:
-starting in the lidffi directory inside jna/native
mkdir build; cd build; ../configure; make; make check

Some tests failed and so I wanted to know how I could resolve my problem and (hopefully) make JNA work.
I also tried to install libffi (version 3.0.8) as a standalone, but the testsuite failed even in that case.

This is the result of the tests for the jna/native/libffi:

Native configuration is powerpc64-unknown-linux-gnu

                === libffi tests ===

Schedule of variations:
    unix

Running target unix
Using /usr/local/share/dejagnu/baseboards/unix.exp as board description file for target.
Using /usr/local/share/dejagnu/config/unix.exp as generic interface file for target.
Using /home/vts/JNA_v3.10_modif_daniela/native/libffi/testsuite/config/default.exp as tool-and-target-specific interface file.
Running /home/vts/JNA_v3.10_modif_daniela/native/libffi/testsuite/libffi.call/call.exp ...
FAIL: libffi.call/cls_align_longdouble_split.c -O0 -W -Wall (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split.c -O0 -W -Wall compilation failed to produce executable
FAIL: libffi.call/cls_align_longdouble_split2.c -O0 -W -Wall (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split2.c -O0 -W -Wall compilation failed to produce executable
FAIL: libffi.call/cls_longdouble.c -O0 -W -Wall (test for excess errors)
WARNING: libffi.call/cls_longdouble.c -O0 -W -Wall compilation failed to produce executable
FAIL: libffi.call/cls_longdouble_va.c -O0 -W -Wall output pattern test, is %.1f
res: 5
%.1f
res: 5
? should match 7.0
?es: 4
?.0
res: 4
FAIL: libffi.call/err_bad_abi.c -O0 -W -Wall execution test
XPASS: libffi.call/float2.c -O0 -W -Wall (test for excess errors)
WARNING: libffi.call/huge_struct.c -O0 -W -Wall compilation failed to produce executable
FAIL: libffi.call/return_uc.c -O0 -W -Wall (test for excess errors)
FAIL: libffi.call/cls_align_longdouble_split.c -O2 (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split.c -O2 compilation failed to produce executable
FAIL: libffi.call/cls_align_longdouble_split2.c -O2 (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split2.c -O2 compilation failed to produce executable
FAIL: libffi.call/cls_longdouble.c -O2 (test for excess errors)
WARNING: libffi.call/cls_longdouble.c -O2 compilation failed to produce executable
FAIL: libffi.call/cls_longdouble_va.c -O2 output pattern test, is %.1f
res: 5
%.1f
res: 5
? should match 7.0
?es: 4
?.0
res: 4
FAIL: libffi.call/err_bad_abi.c -O2 execution test
XPASS: libffi.call/float2.c -O2 (test for excess errors)
WARNING: libffi.call/huge_struct.c -O2 compilation failed to produce executable
FAIL: libffi.call/cls_align_longdouble_split.c -O3 (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split.c -O3 compilation failed to produce executable
FAIL: libffi.call/cls_align_longdouble_split2.c -O3 (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split2.c -O3 compilation failed to produce executable
FAIL: libffi.call/cls_longdouble.c -O3 (test for excess errors)
WARNING: libffi.call/cls_longdouble.c -O3 compilation failed to produce executable
FAIL: libffi.call/cls_longdouble_va.c -O3 output pattern test, is %.1f
res: 5
%.1f
res: 5
? should match 7.0
?es: 4
?.0
res: 4
FAIL: libffi.call/err_bad_abi.c -O3 execution test
XPASS: libffi.call/float2.c -O3 (test for excess errors)
WARNING: libffi.call/huge_struct.c -O3 compilation failed to produce executable
FAIL: libffi.call/cls_align_longdouble_split.c -Os (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split.c -Os compilation failed to produce executable
FAIL: libffi.call/cls_align_longdouble_split2.c -Os (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split2.c -Os compilation failed to produce executable
FAIL: libffi.call/cls_longdouble.c -Os (test for excess errors)
WARNING: libffi.call/cls_longdouble.c -Os compilation failed to produce executable
FAIL: libffi.call/cls_longdouble_va.c -Os output pattern test, is %.1f
res: 5
%.1f
res: 5
? should match 7.0
?es: 4
?.0
res: 4
FAIL: libffi.call/err_bad_abi.c -Os execution test
XPASS: libffi.call/float2.c -Os (test for excess errors)
WARNING: libffi.call/huge_struct.c -Os compilation failed to produce executable
FAIL: libffi.call/cls_align_longdouble_split.c -O2 -fomit-frame-pointer (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split.c -O2 -fomit-frame-pointer compilation failed to produce executable
FAIL: libffi.call/cls_align_longdouble_split2.c -O2 -fomit-frame-pointer (test for excess errors)
WARNING: libffi.call/cls_align_longdouble_split2.c -O2 -fomit-frame-pointer compilation failed to produce executable
FAIL: libffi.call/cls_longdouble.c -O2 -fomit-frame-pointer (test for excess errors)
WARNING: libffi.call/cls_longdouble.c -O2 -fomit-frame-pointer compilation failed to produce executable
FAIL: libffi.call/cls_longdouble_va.c -O2 -fomit-frame-pointer output pattern test, is %.1f
res: 5
%.1f
res: 5
? should match 7.0
?es: 4
?.0
res: 4
FAIL: libffi.call/err_bad_abi.c -O2 -fomit-frame-pointer execution test
XPASS: libffi.call/float2.c -O2 -fomit-frame-pointer (test for excess errors)
WARNING: libffi.call/huge_struct.c -O2 -fomit-frame-pointer compilation failed to produce executable
Running /home/vts/JNA_v3.10_modif_daniela/native/libffi/testsuite/libffi.special/special.exp ...
FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ (test for excess errors)
WARNING: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
FAIL: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ (test for excess errors)
WARNING: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ (test for excess errors)
WARNING: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
FAIL: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ (test for excess errors)
WARNING: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ (test for excess errors)
WARNING: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
FAIL: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ (test for excess errors)
WARNING: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ (test for excess errors)
WARNING: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ compilation failed to produce executable
FAIL: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ (test for excess errors)
WARNING: libffi.special/unwindtest_ffi_call.cc  -shared-libgcc -lstdc++ compilation failed to produce executable

                === libffi Summary ===

# of expected passes            1519
# of unexpected failures        34
# of unexpected successes       5
# of expected failures          5
# of unsupported tests          15 

Can someone tell me what could be the problem and what steps should I take to resolve it?

Thanks,
Daniela Quaresima.