CARD L5 - can the two overflow flags be function-scope statics instead of file statics? - 2026-09-26 02:28:55 -0700
Worker H1. Lab only (production VC7, render_debug cflags, neutral names).
Motivation: January uses each flag in exactly ONE function (entry_overflow only in add_cache_entry, string_overflow
only in add_cache_string); the 2020 /Od build places both flags apart from the other five (0x1736 bytes after
strings). D1 makes them file statics without evidence for that scope.
Setup: five neutral file statics '= 0' in January order + one '= 0' static local per flag inside two functions
(F_entry uses flag E, F_string uses flag S); then the same with both flags uninitialised static locals.
PREDICTIONS:
T1 '= 0' static locals take their place in declaration order at their declaration point: with F_entry defined
   before F_string the layout equals January exactly (E at +0x740A, S at +0x740B); with F_string first, S/E swap.
T2 Uninitialised static locals are hashed (decorated ?name@?1??fn@@9@9) and precede the '= 0' file statics, so
   they cannot sit at +0x740A/+0x740B unless everything is uninitialised.
Outcome decides only whether January's bytes EXCLUDE the static-local form (they do not, if T1 holds).
