On Redhat Linux 6.1J with build 1.2.2-RC2-K

Baseline in TextField is different between English characters
and Japanese doble-bytes characters, Hiragana and Knaji.
The baseline for Japanese characters is lower than English ones.

To reproduce,

1. Compile and run the attached program. (tf_test.java)
2. Input Japanese characters, Hiragana or Kanji, and English characters.
Then, the baseline of Japanese characters is lower.