dwt: reverse the dwt_prf challenge; remote poke/jump now works

The firmware transforms the reply (16-round Feistel per 8-byte record) and
compares it with the host-order copy of the challenge, so the answer is
swap4(inverse_transform(swap4(challenge))).  The font stages are memcpy in
this path because dwt_prf initialises only mode 1.

challenge.py implements the transform and its inverse using the firmware's
fixed key/SP table; solve.py answers the challenge over TCP using the
unmodified exploit client.  320 forward vectors, the firmware's own inverse
and three live sessions validate it.  A live run with no breakpoints poked
the 82-byte hello-world payload and ran it: MAGIC DEAD/BEEF/0400, VRAM
1024/1024.  Notes and evidence under dwt/notes and dwt/evidence.

dwt/evidence/scripts/live_verify.lua

@@ -0,0 +1,72 @@
+-- live_verify.lua -- passive verification of the remote challenge bypass.
+--
+-- No breakpoint here writes a register or memory: it only prints.  In
+-- particular the memcmp at 3F5A0920 is *observed*, never forced.
+local machine = manager.machine
+local cpu = machine.devices[':maincpu']
+local dbg = machine.debugger
+local space = cpu.spaces['program']
+if not dbg then machine:exit(); return end
+dbg.visible_cpu = cpu
+
+local D = "/home/arti/code/k-space/vt1200/dwt/evidence/live-run"
+local MAGIC1 = 0x01810000
+local MAGIC2 = 0x01810010
+local COUNT  = 0x01810020
+local VRAM   = 0x04019040
+local f = 0
+local armed = false
+local saw_challenge = false
+local saw_memcmp = false
+local done_at = nil
+
+local function report_magic(tag)
+	print(string.format("LIVE2 %s: MAGIC1=%04X MAGIC2=%04X COUNT=%04X",
+		tag, space:read_u16(MAGIC1), space:read_u16(MAGIC2), space:read_u16(COUNT)))
+	local filled, holes = 0, 0
+	for r = 0, 63 do
+		for c = 0, 15 do
+			if space:read_u16(VRAM + r * 1024 + c * 16) == 0xFFFF then
+				filled = filled + 1
+			else
+				holes = holes + 1
+			end
+		end
+	end
+	print(string.format("LIVE2 %s: VRAM block %d/1024 words FFFF, %d holes", tag, filled, holes))
+end
+
+emu.register_frame_done(function()
+	f = f + 1
+	if f == 1 and not armed then
+		armed = true
+		dbg:command('bpset 3F58AD60,,{ printf "LIVE2 accept_debug_conn entry pc=%08X\\n", pc ; g }')
+		dbg:command('bpset 3F59F320,,{ printf "LIVE2 dispatcher entry pc=%08X\\n", pc ; g }')
+		dbg:command('bpset 3F59FDC0,,{ printf "LIVE2 dwt_prf challenge entry pc=%08X\\n", pc ; g }')
+		dbg:command('bpset 3F5A0200,,{ printf "LIVE2 challenge sent a13=%08X wire=%08X.%08X.%08X.%08X host=%08X.%08X.%08X.%08X\\n", a13, d@(a13), d@(a13+0x20), d@(a13+0x40), d@(a13+0x60), d@(a13+0x100), d@(a13+0x120), d@(a13+0x140), d@(a13+0x160) ; g }')
+		dbg:command('bpset 3F5A0740,,{ printf "LIVE2 pre-transform reply=%08X.%08X.%08X.%08X\\n", d@(a13+0x80), d@(a13+0xa0), d@(a13+0xc0), d@(a13+0xe0) ; g }')
+		dbg:command('bpset 3F5A08E0,,{ printf "LIVE2 memcmp-args expecting=%08X.%08X.%08X.%08X got=%08X.%08X.%08X.%08X\\n", d@(a9), d@(a9+0x20), d@(a9+0x40), d@(a9+0x60), d@(a11), d@(a11+0x20), d@(a11+0x40), d@(a11+0x60) ; g }')
+		-- OBSERVE ONLY: print the comparison result; never assign to a8.
+		dbg:command('bpset 3F5A0920,,{ printf "LIVE2 challenge memcmp a8=%08X (0=equal)\\n", a8 ; g }')
+		dbg:command('bpset 3F5A1B50,,{ printf "LIVE2 rdebug_poke entry pc=%08X\\n", pc ; g }')
+		dbg:command('bpset 3F5A3890,,{ printf "LIVE2 rdebug_jump entry pc=%08X\\n", pc ; g }')
+		print("LIVE2 armed read-only breakpoints")
+	end
+	if f % 500 == 0 then
+		print(string.format("LIVE2 frame %5d word@0xAECD0=%04X field2..3=%d",
+			f, space:read_u16(0xaecd0), (space:read_u16(0xaecd0) >> 2) & 3))
+	end
+	if done_at == nil and space:read_u16(MAGIC1) == 0xDEAD then
+		done_at = f
+		report_magic("payload-ran")
+	end
+	if done_at and f >= done_at + 120 then
+		print(string.format("LIVE2 exiting at frame %d (payload seen at %d)", f, done_at))
+		machine:exit()
+	end
+	if f >= 9000 then
+		print("LIVE2 timeout: no payload markers")
+		machine:exit()
+	end
+end, "vt1200-live-verify")
+dbg.execution_state = 'run'