From: Cyrus H. <ch...@bo...> - 2006-02-21 05:08:44
|
Well, after a day of chasing my tail, I finally figured out that I had screwed up interrupt handling and things are back in order with a slightly cleaned-up purify to boot. http://cyrusharmon.org/cl/static/patches/sbcl-gencgc-ppc- candidate-4.patch post-merge clean-up items include: * fixing the offset fixup in ppc-vm.lisp * some of the x86[_64] ifdefs should probably be C_STACK_IS_CONTROL_STACK or whatever it is * do stack allocation for dx objects in the allocation macro instead of doing them inline in the vops * clean up the p_a macros with arch_ versions and either unify with or rename existing arch_p_a_xxx functions that expect an interrupt context * MORE GC VERIFICATION! It would be nice to be able to do some gc- verification in, say, the tests without having to recompile. * figure out what's going on with fpu_save/restore on ppc and determine if we really need to save/restore all of those fp registers. * rename ALLOCATION_POINTER * consider cleaning up the GC interfaces such that we can reduce some of the cut-n-paste ooao-violating nightmares in purify.c/gencgc.c/ cheneygc.c * cleanup the #ifdef x86/gencgc/allocation_pointer stuff in globals.c/h But I think all of that can wait. Cyrus |