On Thu, 25 Mar 2004, Andreas Fuchs wrote:
> this patch brings, I'm going to commit it this weekend. We'll always
> be able to undo it when linkage tables are done (-:
I'll take this as an opportunity for a quick progress report: the problem
that stumped me in December has been solved with kind help from Raymond
Toy, and the build gets significantly farther into warm init -- actually
as far as building and loading pcl, but goes *bang* due to what looks like
heap corruption to me.
The exact place where this happens seems to depend on having sb-show or
not, but the symptom so far is constant: a lowtag turns to #b10. If
someone wants to play with this I can upload the patch somewhere.
I'm hoping this to be the last major hurdle, after which it's time for
cleanups and seeing if dlhandle logic could be handed of to the OS. IIRC
Tim Moore said something to the effect that this caused problems on
FreeBSD, but since approximate half of the C-code is due to this I want to
check if something can be done.
However, I'm planning to finish package-locks first, and the season for
exams is fast approching, so I expect it to be early summer before
significant progress happens with linkage-table.