Re: Serious GC bug

[Tom Tromey <tromey at redhat dot com>]

>> It turns out the hash synchronization bug I was chasing was
>> actually a serious GC bug that apparently was there all along.
>> GC_gcj_malloc apparently released the allocator lock twice, in most
>> cases.

Bryce> I've checked this in. Interestingly, your test case ran fine
Bryce> for me with or without the patch, however it runs measurably
Bryce> faster with the patch applied.

Thanks Bryce.

Hans -- FYI you still have cvs write access to the tree -- everybody
who had write access to the old /cvs/java tree got access to the gcc
tree.

Here's some notes on checkinsthat I'm including for completeness.
Please accept my apologies if you already know it.  For the
`gcc/boehm-gc' directory you can commit whatever you like; for other
directories there is a process to go through (for non-Java directories
it is especially important).  Weirdly, even though it is your code,
there are a couple caveats to checking things in (I'd like a ChangeLog
entry for local changes and I'd like to see the patch sent to
java-patches).  If you ever wanted to import a new version of the GC,
that would be fine too, though doing this requires some cvs magic
which can be daunting unless you've used cvs for a while.

BTW I'm really looking forward to the hash synchronization work.  This
has been on my wish list basically since I first wrote the
synchronization code way back.

Tom