Here are the first cut of suggested changes to allow java jDK1.0 src to
compile and run on ppc and x86, as discussed.
The following testing was performed on the changes.
1) jdk1.0 was built with itself on x86/ppc Solaris 2.4/2.5/2.5.1/2.6.
2) jdk1.0 was built with itself on sparc Solaris 2.5.
3) an older version of HotJava was added to the the jdk1.0 release, and
generally exercised.
The following known problems exist with this version:
1) x86: bin/hotjava_g crashes. It appears there's still an unresolved
context switching problem with the x86 version (see removal of
setedi/getedi hack), or perhaps this is another problem altogether.
I'd really like to have the get/setedi hack removed, the overhead to
block signals is unacceptable.
2) sparc: _brk_unlocked() override for 2.6 is missing. To make java run
on 2.6, this system call should be overridden (see changes to the now
misnamed edi.s).
It's probably best to file separate bugs to track these.
compile and run on ppc and x86, as discussed.
The following testing was performed on the changes.
1) jdk1.0 was built with itself on x86/ppc Solaris 2.4/2.5/2.5.1/2.6.
2) jdk1.0 was built with itself on sparc Solaris 2.5.
3) an older version of HotJava was added to the the jdk1.0 release, and
generally exercised.
The following known problems exist with this version:
1) x86: bin/hotjava_g crashes. It appears there's still an unresolved
context switching problem with the x86 version (see removal of
setedi/getedi hack), or perhaps this is another problem altogether.
I'd really like to have the get/setedi hack removed, the overhead to
block signals is unacceptable.
2) sparc: _brk_unlocked() override for 2.6 is missing. To make java run
on 2.6, this system call should be overridden (see changes to the now
misnamed edi.s).
It's probably best to file separate bugs to track these.