Hi All;
Although we've not been posting much to the list, there actually has been a lot
of activity lately... Here's a ruthless summary (please follow-up if I've
missed anything):
Progress:
(1) Todd implemented: synchronized methods in the JVM; virtual consoles in a
java daemon.
(2) Thomas Bocek found a bug in two of the JVM's opcodes -- irem, and idiv.
Either Todd or jm will integrate his fix with the other active mods to the JVM.
Will probably be committed to CVS within a day or two.
(3) CVS pserver has been acting up again. Todd was kind enough to restart the
server again.
(4) jm implemented an unbelievably slow (but working), malloc-free memory
manager. Implements hybrid best/first-fit, with sentinel words to try and
detect writing over the end of blocks, and with status bits to detect
multiple-freeing. Todd and jm used it to detect and fix two memory-management
bugs in the JVM. This capability and the bug fixes have not yet been committed
to CVS because of a problem with the i386 build.
Ongoing, not-yet-complete efforts:
(1) Todd has begun to implement throws/exceptions in the JVM.
(2) The combined jjos+decaf+GC dies shortly after the first GC. George Marrows
and jm are looking into this. (Fixing the memory management bugs mentioned
above unfortunately had no effect on this problem.) Scheduling constraints will
likely preclude any resolution before late this weekend.
(3) jm ran into a problem with the i386 build while verifying code changes prior
to CVS commitment. jm's P5 box complains about not having CPUID implemented.
Weird, because that's the box the CPUID code was debugged upon. (Anybody else
seen this?) Resolution no earlier than Thursday due to schedule constraints.
(4) Todd and jm are looking into integrating GNU classpath (so we can have
sources to help us debug the low-level classes -- right now we have no idea
what's going to be called except through time-consuming trial-and-error with
Sun's JDK -- plus we can't redistribute it. This is also preventing us from
making a key, "Right Thing To Do" change to the JVM to radically speed-up Java
method calls.). jm has already contacted key classpath developers via email and
gotten their go-ahead.
As usual, please reply to me and/or the list with any errors of either omission
or commission.
-jm
--
==== John Morrison ==== MaK Technologies, Inc.
==== Chief Technology Officer ==== 185 Alewife Brook Pkwy, Cambridge, MA 02138
==== [EMAIL PROTECTED] ==== http://www.mak.com/welcome.html
==== vox:617-876-8085 x115 ==== fax:617-876-9208
_______________________________________________
Kernel maillist - [EMAIL PROTECTED]
http://jos.org/mailman/listinfo/kernel
