Recent gcc3.[01] Test Regressions for i686-pc-linux-gnu

[Jeffrey Oldham]

Last night's nightly i686-pc-linux-gnu builds yielded new test
regressions:

gcc 3.0:
    FAIL: final_inner -O output from bytecode->native test

gcc 3.1:
    FAIL: BlankFinal compilation from source
    FAIL: assignment_2 -O compilation from source
    FAIL: PR55 -O compilation from bytecode

Is anyone else seeing these regressions?  If so, can we reverse the
code change that caused them?

Thanks,
Jeffrey D. Oldham
oldham@codesourcery.com