Taken from GCC6.1 manual and sorted. Whee.

-Waddress ENABLED TODAY --
-Waggregate-return (might be a good idea, but triggers today.)
-Wall ENABLED TODAY
-Warray-bounds ENABLED TODAY
-Warray-bounds=n (XXXX writeme)
-Wbad-function-cast ENABLED TODAY
-Wbool-compare ENABLED TODAY (by -Wall)
-Wc++-compat (I don't consider this a bug, or it would trigger way too often)
-Wc++11-compat ENABLED TODAY (by -Wall)
-Wc++14-compat ENABLED TODAY (by -Wall)
-Wc90-c99-compat (I don't consider this a bug, or it would trigger way too often)
-Wc99-c11-compat (could add trivially)
-Wcast-align (triggers today) GCC 3
-Wcast-qual (might be a good idea, but triggers today.) GCC 4.6
-Wchar-subscripts ENABLED TODAY
-Wclobbered ENABLED TODAY (by -Wextra)
-Wcomment ENABLED TODAY
-Wconditionally-supported (C++ only, irrelevant to tor.)
-Wconversion (might be a good idea, but triggers today.) GCC 4.6
-Wcoverage-mismatch ENABLED TODAY (by default)
-Wdate-time (could add trivially)
-Wdeclaration-after-statement (I don't consider this a bug, or it would trigger way too often)
-Wdelete-incomplete (C++ only, irrelevant to tor.)
-Wdisabled-optimization (I don't consider this a bug, or it would trigger way too often)
-Wdouble-promotion (might be a good idea, but triggers today.) GCC 4.6
-Wduplicated-cond (might be a good idea, but triggers today.) GCC 6.1
-Wempty-body ENABLED TODAY (by -Wextra)
-Wenum-compare ENABLED TODAY (by -Wall)
-Werror ENABLED TODAY
-Wextra ENABLED TODAY
-Wfloat-conversion (might be a good idea, but triggers today.) GCC 4.9
-Wfloat-equal ENABLED TODAY
-Wformat ENABLED TODAY (by -Wall)
-Wformat-nonliteral ENABLED TODAY (by -Wformat=2)
-Wformat-security ENABLED TODAY (by -Wformat=2)
-Wformat-signedness ENABLED TODAY (by -Wformat=2)
-Wformat-y2k ENABLED TODAY (by -Wformat=2)
-Wformat=2 ENABLED TODAY
-Wframe-address ENABLED TODAY (by -Wall)
-Wframe-larger-than=len (I don't consider this a bug, or it would trigger way too often)
-Whsa (irrelevant to Tor?)
-Wignored-attributes (could add trivially)
-Wignored-qualifiers ENABLED TODAY (by -Wextra)
-Wimplicit ENABLED TODAY (by -Wall)
-Wimplicit-function-declaration ENABLED TODAY (by -Wall)
-Wimplicit-int ENABLED TODAY (by -Wall)
-Wincompatible-pointer-types Should we try this one? it triggers a lot, I bet
Actually, this is on by default FWICT.
GCC 5.1
-Winit-self ENABLED TODAY
-Winline (I don't consider this a bug, or it would trigger way too often)
-Winvalid-memory-model ENABLED TODAY (by default)
-Winvalid-pch (I don't consider this a bug, or it would trigger way too often)
-Wjump-misses-init (might be a good idea, but triggers today.) GCC 4.6
-Wlarger-than=len (I don't consider this a bug, or it would trigger way too often)
-Wlogical-not-parentheses ENABLED TODAY (by -Wall)
-Wlogical-op ENABLED TODAY
-Wlong-long (I don't consider this a bug, or it would trigger way too often)
-Wmain ENABLED TODAY (by -Wall)
-Wmaybe-uninitialized ENABLED TODAY (by -Wall)
-Wmemset-transposed-args ENABLED TODAY (by -Wall)
-Wmisleading-indentation ENABLED TODAY (by -Wall)
-Wmissing-braces ENABLED TODAY (by -Wall)
-Wmissing-declarations ENABLED TODAY
-Wmissing-field-initializers ENABLED TODAY
-Wmissing-format-attribute (could add trivially)
-Wmissing-include-dirs (I don't consider this a bug, or it would trigger way too often)
-Wmissing-parameter-type ENABLED TODAY (by -Wextra)
-Wmissing-prototypes ENABLED TODAY
-Wnested-externs ENABLED TODAY
-Wnonnull ENABLED TODAY (by -Wall)
-Wnonnull-compare ENABLED TODAY (by -Wall)
-Wnormalized=[none|id|nfc|nfkc] We use -Wnormalized=id now; should we switch?
-Wnull-dereference (might be a good idea, but triggers today.) GCC 6.1
-Wodr (irrelevant to Tor?)
-Wold-style-declaration ENABLED TODAY (by -Wextra)
-Wold-style-definition ENABLED TODAY
-Wopenmp-simd ENABLED TODAY (by -Wall)
-Woverlength-strings (might be a good idea, but triggers today.) GCC 4.6
-Woverride-init-side-effects ENABLED TODAY (by default)
-Wpacked (could add trivially)
-Wpacked-bitfield-compat (I don't consider this a bug, or it would trigger way too often)
-Wpadded (might be a good idea, but triggers today.) GCC 3
-Wparentheses ENABLED TODAY (by -Wall)
-Wpedantic (I don't consider this a bug, or it would trigger way too often)
-Wplacement-new (C++ only, irrelevant to tor.)
-Wplacement-new=n (C++ only, irrelevant to tor.)
-Wpointer-arith ENABLED TODAY
-Wpointer-sign ENABLED TODAY (by -Wall)
-Wredundant-decls ENABLED TODAY
-Wreturn-type ENABLED TODAY (by -Wall)
-Wsequence-point ENABLED TODAY (by -Wall)
-Wshadow We have a ticket to turn this one on. Good idea? GCC 3
-Wshift-count-negative (could add trivially)
-Wshift-count-overflow (could add trivially)
-Wshift-negative-value (could add trivially)
-Wshift-overflow ENABLED TODAY (by default)
-Wshift-overflow=n Easy, when n==2.
-Wsign-compare ENABLED TODAY (by -Wextra)
-Wsign-conversion (might be a good idea, but triggers today.) GCC 4.6
-Wsizeof-array-argument (could add trivially)
-Wsizeof-pointer-memaccess ENABLED TODAY (by -Wall)
-Wstack-protector ENABLED TODAY
-Wstack-usage=len Would trigger way too much.
-Wstrict-aliasing ENABLED TODAY (by -Wall)
-Wstrict-aliasing=n Already on; could raise n to 5? GCC 3?
-Wstrict-overflow (might be a good idea, but triggers today.) GCC 4.2
-Wstrict-overflow=n (might be a good idea, but triggers today.) GCC 4.2
-Wstrict-prototypes ENABLED TODAY
-Wsubobject-linkage (C++ only, irrelevant to tor.)
-Wsuggest-attribute=pure Triggers a lot on our current code GCC 4.6
-Wsuggest-attribute=const Triggers a lot on our current code GCC 4.6
-Wsuggest-attribute=noreturn (No problem to add)
-Wsuggest-attribute=format (no problem to add)
-Wsuggest-final-methods (C++ only, irrelevant to tor.)
-Wsuggest-final-types (C++ only, irrelevant to tor.)
-Wsuggest-override (C++ only, irrelevant to tor.)
-Wswitch ENABLED TODAY (by -Wall)
-Wswitch-bool (could add trivially)
-Wswitch-default (might be a good idea, but triggers today.) GCC 3
-Wswitch-enum ENABLED TODAY
-Wsync-nand (could add trivially)
-Wsystem-headers (I don't consider this a bug, or it would trigger way too often)
-Wtautological-compare ENABLED TODAY (by -Wall)
-Wtraditional (I don't consider this a bug, or it would trigger way too often)
-Wtraditional-conversion (I don't consider this a bug, or it would trigger way too often)
-Wtrampolines (could add trivially)
-Wtrigraphs ENABLED TODAY (by -Wall)
-Wtype-limits ENABLED TODAY (by -Wextra)
-Wundef ENABLED TODAY
-Wuninitialized ENABLED TODAY (by -Wall)
-Wunknown-pragmas ENABLED TODAY (by -Wall)
-Wunsafe-loop-optimizations (might be a good idea, but triggers today.) GCC 4.1
-Wunsuffixed-float-constants (might be a good idea, but triggers today.) GCC 4.6
-Wunused (could add trivially)
-Wunused-but-set-parameter (could add trivially)
-Wunused-but-set-variable (could add trivially)
-Wunused-const-variable (might be a good idea, but triggers today.) GCC 6.1
-Wunused-const-variable=n (Triggers on our current code.) GCC 6.1
-Wunused-function ENABLED TODAY (by -Wall)
-Wunused-label ENABLED TODAY (by -Wall)
-Wunused-local-typedefs (could add trivially)
-Wunused-parameter (could add trivially)
-Wunused-value ENABLED TODAY (by -Wall)
-Wunused-variable ENABLED TODAY (by -Wall)
-Wuseless-cast (C++ only, irrelevant to tor.)
-Wvariadic-macros (could add trivially)
-Wvector-operation-performance (irrelevant to Tor?)
-Wvla (I don't consider this a bug, or it would trigger way too often)
-Wvolatile-register-var ENABLED TODAY (by -Wall)
-Wwrite-strings ENABLED TODAY
-Wzero-as-null-pointer-constant (C++ only, irrelevant to tor.)
-pedantic-errors (I don't consider this a bug, or it would trigger way too often)