harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Spark Shen" <smallsmallor...@gmail.com>
Subject Re: [testing][classlib] classlib test org.apache.harmony.nio.tests.java.nio.channels.FileChannelTest failed now on x86_64
Date Tue, 13 Mar 2007 08:59:53 GMT
After consideration, IMO, writing macros for this problem would be tedious.
Some OS defines off_t as long, some as int, endian further complicates the
cast problem. If use macro way, there may be at least [off_t size
array].length * [endian].types different macro, which is confusing.

Cast jlong to off_t would be simpler. Compiler automatically does the
conversion for us. This conversion is safe, since the passed in offset is
guaranteed in integer scope in java layer, and off_t >= int. I have provided
another patch. Would any committer take a look?

Best regards.

2007/3/13, Yang Paulex <paulex.yang@gmail.com>:
>
> 2007/3/13, Stepan Mishura <stepan.mishura@gmail.com>:
> >
> > On 3/12/07, Vladimir Ivanov wrote:
> >
> > > If changes to the file OSFileSystemLinux32.c were reverted and
> > > classlib+drlvm were rebuilt this test passed.
> > > Command to revert:
> > > classlib/trunk> svn up -r517104
> > > modules/luni/src/main/native/luni/unix/OSFileSystemLinux32.c
> >
> >
> > Yes, I can reproduce the failure on x86_64 on Linux and if I revert
> update
> > the test passes.
> >
> > Paulex, could you look into your commit r517145?:
> > patch for HARMONY-3334([classlib][luni-native] Native Method
> > XXX_transferImpl in luni/src/main/native/luni/unix/OSFileSystemLinux32.c
> > did
> > not handle endian properly)
>
>
> Patch reverted at revision r517562, thank you to point it out and sorry
> for
> inconvenience, instead of the tricky and error prone way, seems defining
> some macros for big endian architecture is the right thing to do :-(.
>
> Thanks,
> > Stepan.
> >
> >
> >
> >
> > > thanks, Vladimir
> > >
> > > On 3/12/07, Vladimir Ivanov <ivavladimir@gmail.com> wrote:
> > > > This test passed on r517014 and failed starting from r517145.
> > > >
> > > >  thanks, Vladimir
> > > >
> > > > On 3/12/07, Vladimir Ivanov <ivavladimir@gmail.com> wrote:
> > > > > Hello everyone,
> > > > > the classlib test
> > > > >
> org.apache.harmony.nio.tests.java.nio.channels.FileChannelTestfailed
> > > > > now on the SUSE 9 Linux on x86_64 platform. This failure is
> stable.
> > I
> > > > > try to identify the guilty commit.
> > > > >
> > > > >  Thanks, Vladimir
> > > > >
> > > > > Execution log:
> > > > > Test: test_transferToJJLWritableByteChannel_SocketChannelClass:
> > > > > org.apache.harmony.nio.tests.java.nio.channels.FileChannelTest
> > > > > java.io.IOException
> > > > >        at org.apache.harmony.luni.platform.OSFileSystem.transfer(
> > > OSFileSystem.java:238)
> > > > >        at
> > > org.apache.harmony.nio.internal.FileChannelImpl.kernelTransfer(
> > > FileChannelImpl.java:455)
> > > > >        at
> org.apache.harmony.nio.internal.FileChannelImpl.transferTo
> > (
> > > FileChannelImpl.java:443)
> > > > >        at
> > >
> >
> org.apache.harmony.nio.tests.java.nio.channels.FileChannelTest.test_transferToJJLWritableByteChannel_SocketChannel
> > > (FileChannelTest.java:3058)
> > > > >        at java.lang.reflect.VMReflection.invokeMethod(
> > > VMReflection.java)
> > > > >
> > > >
> > >
> >
> >
> >
> > --
> > Stepan Mishura
> > Intel Enterprise Solutions Software Division
> >
>
>
>
> --
> Paulex Yang
> China Software Development laboratory
> IBM
>



-- 
Spark Shen
China Software Development Lab, IBM

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message