[issues] Re: Help , test iconvdata/tst-iconv7 failed on mips simulator

["Amker.Cheng" <amker.cheng@xxxxxxxxx>]

On Fri, Sep 11, 2009 at 7:11 PM, Maxim Kuvyrkov <maxim@xxxxxxxxxxxxxxxx> wrote:
> Amker.Cheng wrote:
>>
>> Hi :
>>   Don't know if it's ok to send this mail to this list, But I can not
>> find other lists more appropriate than this one.
>> Please inform me if it should not be post here.
>
> The correct list would be issues@xxxxxxxxxx, forwarding it there.
>
>> I have set up the cross test environment for mips on qemu simulator
>> and now can run test suite normally.
>> But test process breaks when running iconvdata/tst-iconv7.
>> the dump file tst-iconv7.out says that "wrong first conversion".
>> I debugged tst-iconv7.c and found the function call
>> "size_t r = iconv (cd, &inptr, &inlen, &outptr, &outlen);"  (around line
>> 26)
>> failed with errno string as "Invalid or incomplete multibyte or wide
>> character"
>>
>> I am not sure what's going wrong, Please help me out, any tips will be
>> highly appreciated.
>>
>> following is dump result, hope it will be help.
>> <--------------------start
>> before iconv call
>>        inptr len = 7
>>        outptr =
>>        outptr len = 32
>> after iconv call
>>        return value = -1
>>        inptr len = 7
>>        outptr =
>>        outptr len = 32
>>        errno : Invalid or incomplete multibyte or wide character
>>
>> wrong first conversion
>> -------------------->end
>
>

Sorry for mis-listing, It has been solved. Thanks.
As long as eglibc test is concerned, I have tested eglibc-2.8 released
in Codesourcery's  "Sourcery G++ Lite 4.3-154"
on qemu based simulated mips-linux(version 2.6.27.28). Both glibc and
linux are configured using soft float.
Following are the results:
-------------------------------------start here
math/test-fenv.out] Error 1
math/test-misc.out] Error 1
math/bug-nextafter.out] Error 16
math/bug-nexttoward.out] Error 16
math/test-float.out] Error 1
math/test-double.out] Error 1
math/test-ifloat.out] Error 1

stdio-common/test-vfprintf.out] Error 1

make[2]: *** [/home/amker/nfs-share/eglibc-build-check/obj/eglibc/timezone/testdata/America/New_York]
Error 127
make[2]: *** [/home/amker/nfs-share/eglibc-build-check/obj/eglibc/timezone/testdata/Etc/UTC]
Error 127
make[2]: *** [/home/amker/nfs-share/eglibc-build-check/obj/eglibc/timezone/testdata/Europe/Berlin]
Error 127
make[2]: *** [/home/amker/nfs-share/eglibc-build-check/obj/eglibc/timezone/testdata/Australia/Melbourne]
Error 127
make[2]: *** [/home/amker/nfs-share/eglibc-build-check/obj/eglibc/timezone/testdata/America/Sao_Paulo]
Error 127
make[2]: *** [/home/amker/nfs-share/eglibc-build-check/obj/eglibc/timezone/testdata/Asia/Tokyo]
Error 127


elf/tst-audit1.out] Error 127
elf/tst-audit2.out] Error 127
elf/check-localplt.out] Error 1
---------------------------------------------end here

I am puzzled with the test results :
1, I don't know why there are so many math tests failed, does it matter?
2, I don't know why there are errors complain can not found the
listing timezone data files in build directory.
3, as "stdio-common/test-vfprintf.out", There maybe a minor bug in
vfprintf.c leading to overflow error.
4, "elf/tst-audit1.out" and "elf/tst-audit2.out" are very suspicious.
It complains "/home/amker/nfs-share/eglibc-build-check/obj/eglibc/elf/tst-audit1:
no PLTREL found in object
/home/amker/nfs-share/eglibc-build-check/obj/eglibc/libc.so.6" and
aborts.
I checked the source code of dynamic linker and found it may be caused
by the plt extension  for mips abi merged in eglibc-2.8
released in "Sourcery G++ Lite 4.3-154".

So, does anyone have any idea about the cross test results? I am some
kind of lost, and any tip will be highly appreciated.

Thanks.



-- 
Best Regards.