PATCH: java/25330: A race condition in write_classfile

[Andrew Haley]

Tom Tromey writes:
 > >>>>> "H.J." == H J Lu <hjl@lucon.org> writes:
 > 
 > H.J.> With parallel build in libjava, several processes may want to generate
 > H.J.> the same class file.
 > 
 > When does this happen?
 > 
 > It really shouldn't.  gcj isn't eager like other java compilers -- it
 > only compiles what you ask it to.  We shouldn't have two invocations
 > working on the same .java file.

And, indeed, that doesn't seem to be happening in the log for this PR.
There may be more to this than is apparent.

Having said that, it seems to me hj's patch is correct.

Andrew.