diff --git a/milkymist/m1crg/__init__.py b/milkymist/m1crg/__init__.py
index 0d20fd8f9..6938e98df 100644
--- a/milkymist/m1crg/__init__.py
+++ b/milkymist/m1crg/__init__.py
@@ -3,8 +3,9 @@ from fractions import Fraction
 from migen.fhdl.structure import *
 from migen.fhdl.specials import Instance
 from migen.fhdl.module import Module
+from migen.bank.description import *
 
-class M1CRG(Module):
+class M1CRG(Module, AutoReg):
 	def __init__(self, pads, outfreq1x):
 		self.clock_domains.cd_sys = ClockDomain()
 		self.clock_domains.cd_sys2x_270 = ClockDomain()
@@ -17,12 +18,22 @@ class M1CRG(Module):
 		self.clk4x_wr_strb = Signal()
 		self.clk4x_rd_strb = Signal()
 
+		self._r_cmd_data = RegisterField(10)
+		self._r_send_cmd_data = RegisterRaw()
+		self._r_send_go = RegisterRaw()
+		self._r_status = RegisterField(3, READ_ONLY, WRITE_ONLY)
+
 		###
 		
 		infreq = 50*1000000
 		ratio = Fraction(outfreq1x)/Fraction(infreq)
 		in_period = float(Fraction(1000000000)/Fraction(infreq))
 
+		vga_progdata = Signal()
+		vga_progen = Signal()
+		vga_progdone = Signal()
+		vga_locked = Signal()
+
 		self.specials += Instance("m1crg",
 			Instance.Parameter("in_period", in_period),
 			Instance.Parameter("f_mult", ratio.numerator),
@@ -48,4 +59,40 @@ class M1CRG(Module):
 			Instance.Output("ddr_clk_pad_p", pads.ddr_clk_p),
 			Instance.Output("ddr_clk_pad_n", pads.ddr_clk_n),
 			Instance.Output("eth_phy_clk_pad", pads.eth_phy_clk),
-			Instance.Output("vga_clk_pad", pads.vga_clk))
+			Instance.Output("vga_clk_pad", pads.vga_clk),
+
+			Instance.Input("vga_progclk", ClockSignal()),
+			Instance.Input("vga_progdata", vga_progdata),
+			Instance.Input("vga_progen", vga_progen),
+			Instance.Output("vga_progdone", vga_progdone),
+			Instance.Output("vga_locked", vga_locked))
+
+		remaining_bits = Signal(max=11)
+		transmitting = Signal()
+		self.comb += transmitting.eq(remaining_bits != 0)
+		sr = Signal(10)
+		self.sync += [
+			If(self._r_send_cmd_data.re,
+				remaining_bits.eq(10),
+				sr.eq(self._r_cmd_data.field.r)
+			).Elif(transmitting,
+				remaining_bits.eq(remaining_bits - 1),
+				sr.eq(sr[1:])
+			)
+		]
+		self.comb += [
+			vga_progdata.eq(transmitting & sr[0]),
+			vga_progen.eq(transmitting | self._r_send_go.re)
+		]
+
+		# enforce gap between commands
+		busy_counter = Signal(max=14)
+		busy = Signal()
+		self.comb += busy.eq(busy_counter != 0)
+		self.sync += If(self._r_send_cmd_data.re,
+				busy_counter.eq(13)
+			).Elif(busy,
+				busy_counter.eq(busy_counter - 1)
+			)
+
+		self.comb += self._r_status.field.w.eq(Cat(busy, vga_progdone, vga_locked))
diff --git a/software/include/hw/flags.h b/software/include/hw/flags.h
index 64f6a7edb..759246a69 100644
--- a/software/include/hw/flags.h
+++ b/software/include/hw/flags.h
@@ -18,4 +18,8 @@
 #define MINIMAC_EV_RX1	0x2
 #define MINIMAC_EV_TX	0x4
 
+#define CLKGEN_STATUS_BUSY		0x1
+#define CLKGEN_STATUS_PROGDONE	0x2
+#define CLKGEN_STATUS_LOCKED	0x4
+
 #endif /* __HW_FLAGS_H */
diff --git a/top.py b/top.py
index 96d1787b6..e5b111783 100644
--- a/top.py
+++ b/top.py
@@ -65,17 +65,18 @@ class M1ClockPads:
 class SoC(Module):
 	csr_base = 0xe0000000
 	csr_map = {
-		"uart":					0,
-		"dfii":					1,
-		"identifier":			2,
-		"timer0":				3,
-		"minimac":				4,
-		"fb":					5,
-		"asmiprobe":			6,
-		"dvisampler0":			7,
-		"dvisampler0_edid_mem":	8,
-		"dvisampler1":			9,
-		"dvisampler1_edid_mem":	10,
+		"crg":					0,
+		"uart":					1,
+		"dfii":					2,
+		"identifier":			3,
+		"timer0":				4,
+		"minimac":				5,
+		"fb":					6,
+		"asmiprobe":			7,
+		"dvisampler0":			8,
+		"dvisampler0_edid_mem":	9,
+		"dvisampler1":			10,
+		"dvisampler1_edid_mem":	11,
 	}
 
 	interrupt_map = {
@@ -134,6 +135,7 @@ class SoC(Module):
 		#
 		# CSR
 		#
+		self.submodules.crg = m1crg.M1CRG(M1ClockPads(platform), clk_freq)
 		self.submodules.uart = uart.UART(platform.request("serial"), clk_freq, baud=115200)
 		self.submodules.identifier = identifier.Identifier(0x4D31, version, int(clk_freq))
 		self.submodules.timer0 = timer.Timer()
@@ -151,11 +153,10 @@ class SoC(Module):
 		#
 		for k, v in sorted(self.interrupt_map.items(), key=itemgetter(1)):
 			self.comb += self.cpu.interrupt[v].eq(getattr(self, k).ev.irq)
-		
+
 		#
 		# Clocking
 		#
-		self.submodules.crg = m1crg.M1CRG(M1ClockPads(platform), clk_freq)
 		self.comb += [
 			self.ddrphy.clk4x_wr_strb.eq(self.crg.clk4x_wr_strb),
 			self.ddrphy.clk4x_rd_strb.eq(self.crg.clk4x_rd_strb)
diff --git a/verilog/m1crg/m1crg.v b/verilog/m1crg/m1crg.v
index d5582d405..bf91d2be9 100644
--- a/verilog/m1crg/m1crg.v
+++ b/verilog/m1crg/m1crg.v
@@ -33,7 +33,14 @@ module m1crg #(
 	
 	/* VGA clock */
 	output vga_clk,		/* < buffered, to internal clock network */
-	output vga_clk_pad	/* < forwarded through ODDR2, to I/O */
+	output vga_clk_pad,	/* < forwarded through ODDR2, to I/O */
+
+	/* VGA clock control */
+	input vga_progclk,
+	input vga_progdata,
+	input vga_progen,
+	output vga_progdone,
+	output vga_locked
 );
 
 /*
@@ -257,7 +264,6 @@ assign eth_tx_clk = eth_tx_clk_pad;
  * VGA clock
  */
 
-// TODO: hook up the reprogramming interface
 DCM_CLKGEN #(
 	.CLKFXDV_DIVIDE(2),
 	.CLKFX_DIVIDE(4),
@@ -270,15 +276,15 @@ DCM_CLKGEN #(
 	.CLKFX(vga_clk),
 	.CLKFX180(),
 	.CLKFXDV(),
-	.LOCKED(),
-	.PROGDONE(),
 	.STATUS(),
 	.CLKIN(pllout4),
 	.FREEZEDCM(1'b0),
-	.PROGCLK(1'b0),
-	.PROGDATA(),
-	.PROGEN(1'b0),
-	.RST(1'b0)
+	.PROGCLK(vga_progclk),
+	.PROGDATA(vga_progdata),
+	.PROGEN(vga_progen),
+	.PROGDONE(vga_progdone),
+	.LOCKED(vga_locked),
+	.RST(~pll_lckd)
 );
 
 ODDR2 #(