Sync'ing up jdk7 at b50

[Alan Bateman]

Andrew John Hughes wrote:
> :
> It only happened last week :)
> We already have one portability issue you are probably aware of.
> Thanks to testing from Matthias Klose, we've found that the build
> fails on armel:
>
> ../../../src/solaris/native/sun/nio/fs/UnixNativeDispatcher.c: In
> function 'Java_sun_nio_fs_UnixNativeDispatcher_initIDs':
> ../../../src/solaris/native/sun/nio/fs/UnixNativeDispatcher.c:203:
> error: 'fstatat64_wrapper' undeclared (first use in this function)
> ../../../src/solaris/native/sun/nio/fs/UnixNativeDispatcher.c:203:
> error: (Each undeclared identifier is reported only once
> ../../../src/solaris/native/sun/nio/fs/UnixNativeDispatcher.c:203:
> error: for each function it appears in.)
> make[6]: *** [/build/buildd/openjdk-6-6b14/openjdk/control/build/linux-arm/tmp/java/java.nio/nio/obj/UnixNativeDispatcher.o]
> Error 1
>
> http://launchpadlibrarian.net/22537446/buildlog_ubuntu-jaunty-armel.openjdk-6_6b14-0ubuntu12_FAILEDTOBUILD.txt.gz
>
> Strangely, it passes on SPARC and PowerPC (both GNU/Linux).  This
> function is available in glibc 2.9 so the conditional to just turn it
> off for GNU/Linux platforms is wrong.
>   
Yes, I guessed this would be a problem but I thought I had changed it so 
that the fallback to use syscall was only compiled in on x86/x64. I see 
this was in b99 (your first mail mentioned b98) so that might explain it.

-Alan.