Hi All;
(1) Thomas Bocek found and fixed a bug in idiv (operands transposed).
jm fixed isomorphic bugs in fdiv and ldiv (but curiously not
ddiv).
(2) Thomas also found and fixed a similar operand transposition error
in irem. jm fixed isomorphic bugs in drem, frem, and lrem.
(3) George Marrows found a use of "interior pointers" to C++ objects,
wherein decaf retained a pointer to the "middle" of an object (as
opposed to a pointer to its reasonably easy-to-identify head). He
updated his gc code to cope with most of these problems. jm
incorporated George's fixes in the host build, which ran overnight
through 179 garbage collections before jm terminated it this
morning.
(4) However (and this is a big "however"), the i386 build (as built by
jm from the same sources as the host build) does NOT seem to
survive (much after?) the first gc. If the "vga.test()" method is
run, the program seems to hang for a bit, but then eventually
resets the box after drawing about 400 horizontal lines (anybody
else seen this?). If repeated testVM() (and other methods) are
run, then the program runs a long time (hours), but it is not
clear, as of this writing, whether a successful GC has occurred
prior to crashing. jm is continuing to work the issue this
weekend.
(5) jm's sources are pretty much up to date with respect to the CVS
repository (only init.java and vga.java need updating, and jm has
extensively modified init.java). jm has modified 8 files in
decaf, and 3 files in jjos.
My question, pursuant to (5) -- the conservative GC is a Big Win for
those doing host development. It's not necessarily a Big Win for
those doing i386 development. Would those people doing i386
development PLEASE let me know whether I should commit the sources to
CVS, which include the fixes in items (1) through (3)??? I don't want
to commit sources which contain known bugs (e.g., (5) above), even if
it fixes some problems, unless people tell me it's OK to do so.
Thanks, and sorry for the slow progress.
-jm
--
==== John Morrison
==== MaK Technologies Inc.
==== 185 Alewife Brook Parkway, Cambridge, MA 02138
==== http://www.mak.com/
==== vox:617-876-8085 x115
==== fax:617-876-9208
==== [EMAIL PROTECTED]
_______________________________________________
Kernel maillist - [EMAIL PROTECTED]
http://jos.org/mailman/listinfo/kernel
