patch: hash synchronization for mingw/cygwin

Adam Megacz gcj@lists.megacz.com
Mon Oct 14 09:46:00 GMT 2002


"Boehm, Hans" <hans_boehm@hp.com> writes:
> 1) I would add a comment near the first use of
> JV_LINKER_CANNOT_8BYTE_ALIGN_STATICS that this only addresses
> misalignment of statics, not heap objects.  And it works for statics
> only because registering it for finalization is a noop, no matter
> what the least significant bits are.  I think the code is correct,
> but the reasons are more subtle than I would like to see without
> documentation.

Gotcha.

> 2) If you haven't done so, it would be good to verify that
> GetCurrentThreadId is not a bottleneck.

By 'not a bottleneck', I assume you mean "runs quickly"?  If so, I'll
run GetCurrentThreadId() in a tight loop and see how many invocations
per second I can get.

To the rest of the maintainers, pending the above two changes, is it
ok to commit?

Also, does anybody know how to use autoconf to test if the *linker
invoked by gcc* is GNU ld >= 2.3.x?  I don't believe it's possible to
test for this, but if it were, I could undef
JV_LINKER_CANNOT_8BYTE_ALIGN_STATICS for people who have a recent
linker.

I'll also add a comment indicating that
JV_LINKER_CANNOT_8BYTE_ALIGN_STATICS is not required for GNU ld >=
2.3.x.

  - a

-- 
"Through your rags I see your vanity"  -- Socrates



More information about the Java mailing list