Optimization in 'markold'

OLD1 objects can be potentially anywhere in the 'allgc' list (up
to 'reallyold'), but frequently they are all after 'old1' (natural
evolution of survivals) or do not exist at all (when all objects die
young). So, instead of 'markold' starts looking for them always
from the start of 'allgc', the collector keeps an extra pointer,
'firstold1', that points to the first OLD1 object in the 'allgc' list,
or is NULL if there are no OLD1 objects in that list.

testes/gengc.lua

@@ -37,6 +37,22 @@ do
 end
 
 
+do
+  -- ensure that 'firstold1' is corrected when object is removed from
+  -- the 'allgc' list
+  local function foo () end
+  local old = {10}
+  collectgarbage()    -- make 'old' old
+  assert(not T or T.gcage(old) == "old")
+  setmetatable(old, {})    -- new table becomes OLD0 (barrier)
+  assert(not T or T.gcage(getmetatable(old)) == "old0")
+  collectgarbage("step", 0)   -- new table becomes OLD1 and firstold1
+  assert(not T or T.gcage(getmetatable(old)) == "old1")
+  setmetatable(getmetatable(old), {__gc = foo})  -- get it out of allgc list
+  collectgarbage("step", 0)   -- should not seg. fault
+end
+
+
 do   -- bug in 5.4.0
 -- When an object aged OLD1 is finalized, it is moved from the list
 -- 'finobj' to the *beginning* of the list 'allgc', but that part of the