cxp downconn firmware: GTX setup

testing: add loopmode mode & tsusrclk mmcm drp bitbang
testing: add IDLE word printout
downconn: add QPLL and GTX setup
downconn: add DRP to support all CXP linerate up to 12.5Gbps

src/gateware/cxp_downconn.py

@@ -1,5 +1,5 @@
 from migen import *
-from migen.genlib.cdc import MultiReg, PulseSynchronizer
+from migen.genlib.cdc import MultiReg
 from migen.genlib.resetsync import AsyncResetSynchronizer
 
 from misoc.cores.code_8b10b import Encoder, Decoder
@@ -33,10 +33,11 @@ class CXP_DownConn(Module, AutoCSR):
         # single & master tx_mode can lock with rx in loopback 
         self.submodules.gtx = gtx = GTX(self.qpll, pads, sys_clk_freq, tx_mode="single", rx_mode="single")
         
-        # TEST: txusrclk alignment
-        # 1) use GTREFCLK with TXSYSCLKSEL = 0b10 -> still inconsistant
-        # 2) tie qpllPDEN with ~qpll.reset, -> inconsistant
-        # 3) seems like tx_init gtxXreset fall too soon (cplllock hold hi too long) <--- this a cross clk domain issue :<
+        # NOTE: No need to connect cxp_gtx_tx, we don't use tx anyway (just for loopback)
+
+        # TODO: Connect slave cxp_gtx_rx clock tgt
+        # checkout channel interfaces & drtio_gtx
+        # checkout GTPTXPhaseAlignement for inspiration
 
         self.sync += [
             # PLL
@@ -156,18 +157,18 @@ class CXP_DownConn(Module, AutoCSR):
 
         # DEBUG: IO SMA & PMOD
         self.specials += [
-            Instance("OBUF", i_I=gtx.rxoutclk, o_O=debug_sma.p_tx),
+            Instance("OBUF", i_I=gtx.cd_cxp_gtx_rx.clk, o_O=debug_sma.p_tx),
             Instance("OBUF", i_I=gtx.cd_cxp_gtx_tx.clk, o_O=debug_sma.n_rx),
 
             # pmod 0-7 pin
-            Instance("OBUF", i_I=gtx.comma_det.word_aligned, o_O=pmod_pads[0]),
-            Instance("OBUF", i_I=gtx.comma_det.restart, o_O=pmod_pads[1]),
-            Instance("OBUF", i_I=gtx.comma_det.aligner_en_rxclk, o_O=pmod_pads[2]),
-            Instance("OBUF", i_I=gtx.comma_det.check_reset, o_O=pmod_pads[3]),
-            Instance("OBUF", i_I=gtx.comma_det.comma_aligned, o_O=pmod_pads[4]),
-            Instance("OBUF", i_I=gtx.comma_det.comma_seen, o_O=pmod_pads[5]),
-            Instance("OBUF", i_I=gtx.comma_det.has_error, o_O=pmod_pads[6]),
-            Instance("OBUF", i_I=gtx.comma_det.ready, o_O=pmod_pads[7]),
+            Instance("OBUF", i_I=gtx.comma_checker.comma_aligned, o_O=pmod_pads[0]),
+            Instance("OBUF", i_I=gtx.comma_checker.comma_det, o_O=pmod_pads[1]),
+            Instance("OBUF", i_I=gtx.comma_checker.restart_sys, o_O=pmod_pads[2]),
+            Instance("OBUF", i_I=gtx.comma_checker.aligner_en, o_O=pmod_pads[3]),
+            Instance("OBUF", i_I=gtx.comma_checker.check_reset, o_O=pmod_pads[4]),
+            Instance("OBUF", i_I=gtx.comma_checker.has_comma, o_O=pmod_pads[5]),
+            Instance("OBUF", i_I=gtx.comma_checker.has_error, o_O=pmod_pads[6]),
+            Instance("OBUF", i_I=gtx.comma_checker.ready_sys, o_O=pmod_pads[7]),
 
             # Instance("OBUF", i_I=gtx.dclk, o_O=pmod_pads[0]),
             # Instance("OBUF", i_I=gtx.den, o_O=pmod_pads[1]),
@@ -177,23 +178,6 @@ class CXP_DownConn(Module, AutoCSR):
 
         # DEBUG: datain
 
-        self.data_0 = CSRStorage(8)
-        self.data_1 = CSRStorage(8)
-        self.data_2 = CSRStorage(8)
-        self.data_3 = CSRStorage(8)
-        self.control_bit_0 = CSRStorage()
-        self.control_bit_1 = CSRStorage()
-        self.control_bit_2 = CSRStorage()
-        self.control_bit_3 = CSRStorage()
-        self.encoded_0 = CSRStatus(10)
-        self.encoded_1 = CSRStatus(10)
-
-        self.rxdata_0 = CSRStatus(10)
-        self.rxdata_1 = CSRStatus(10)
-        self.decoded_data_0 = CSRStatus(8)
-        self.decoded_data_1 = CSRStatus(8)
-        self.decoded_k_0 = CSRStatus()
-        self.decoded_k_1 = CSRStatus()
         
         self.sync.cxp_gtx_tx += [
             self.gtx.encoder.d[0].eq(0xBC),
@@ -204,15 +188,22 @@ class CXP_DownConn(Module, AutoCSR):
             self.gtx.encoder.k[2].eq(1),
             self.gtx.encoder.d[3].eq(0xB5),
             self.gtx.encoder.k[3].eq(0),
-
-            self.encoded_0.status.eq(self.gtx.encoder.output[0]),
-            self.encoded_1.status.eq(self.gtx.encoder.output[1]),
         ]
 
-        # keep it odd, so it will show data[n] where n is odd
-        stb_timer = Signal(reset=10, max=11)
+        self.rxdata_0 = CSRStatus(10)
+        self.rxdata_1 = CSRStatus(10)
+        self.rxdata_2 = CSRStatus(10)
+        self.rxdata_3 = CSRStatus(10)
+        self.decoded_data_0 = CSRStatus(8)
+        self.decoded_data_1 = CSRStatus(8)
+        self.decoded_data_2 = CSRStatus(8)
+        self.decoded_data_3 = CSRStatus(8)
+        self.decoded_k_0 = CSRStatus()
+        self.decoded_k_1 = CSRStatus()
+        self.decoded_k_2 = CSRStatus()
+        self.decoded_k_3 = CSRStatus()
+
         self.sync.cxp_gtx_rx += [
-            If(stb_timer == 0,
             self.rxdata_0.status.eq(self.gtx.decoders[0].input),
             self.decoded_data_0.status.eq(self.gtx.decoders[0].d),
             self.decoded_k_0.status.eq(self.gtx.decoders[0].k),
@@ -220,23 +211,17 @@ class CXP_DownConn(Module, AutoCSR):
             self.rxdata_1.status.eq(self.gtx.decoders[1].input),
             self.decoded_data_1.status.eq(self.gtx.decoders[1].d),
             self.decoded_k_1.status.eq(self.gtx.decoders[1].k),
-                stb_timer.eq(stb_timer.reset),
-            ).Else(
-                stb_timer.eq(stb_timer - 1),
-            )
+
+            self.rxdata_2.status.eq(self.gtx.decoders[2].input),
+            self.decoded_data_2.status.eq(self.gtx.decoders[2].d),
+            self.decoded_k_2.status.eq(self.gtx.decoders[2].k),
+
+            self.rxdata_3.status.eq(self.gtx.decoders[3].input),
+            self.decoded_data_3.status.eq(self.gtx.decoders[3].d),
+            self.decoded_k_3.status.eq(self.gtx.decoders[3].k),
         ]
 
 
-        # The TX phase alignment will fail with a wrong TXUSRCLK frequency
-
-
-
-        # NOTE: No need to connect cxp_gtx_tx, we don't use tx anyway (just for loopback)
-
-        # TODO: Connect slave cxp_gtx_rx clock tgt
-        # checkout channel interfaces & drtio_gtx
-        # checkout GTPTXPhaseAlignement for inspiration
-
 class QPLL(Module):
     def __init__(self, refclk, sys_clk_freq):
         self.clk = Signal()
@@ -255,23 +240,15 @@ class QPLL(Module):
         self.dready = Signal()
         # # #
 
-        # WARNING: VCO cannot do 12.5GHz on ZC706
-        # VCO freq = sys*qpll_fbdiv
-        # PLL output = VCO/2
-        qpll_fbdiv = 0b0100100000 # 80 div
+        # VCO @ 10GHz, linerate = 1.25Gbps
+        # feedback divider = 80
+        qpll_fbdiv = 0b0100100000
         qpll_fbdiv_ratio = 1
-
-        fbdiv_real = 80
         refclk_div = 1
         self.Xxout_div = 8
 
-        # qpll_fbdiv = 0b0101110000 # 100 div
-        # qpll_fbdiv_ratio = 1
-
-        # fbdiv_real = 100
-        # refclk_div = 2
-        # self.Xxout_div = 2
-
+        # DEBUG: txuserclk
+        fbdiv_real = 80
         self.tx_usrclk_freq = (sys_clk_freq*fbdiv_real/self.Xxout_div)/40
         
         self.specials += [
@@ -328,123 +305,123 @@ class QPLL(Module):
 
 # Warning: Xilinx transceivers are LSB first, and comma needs to be flipped
 # compared to the usual 8b10b binary representation.
-class Comma_Detector(Module):
-    def __init__(self, comma, check_period=1_000_000):
+class Comma_Checker(Module):
+    def __init__(self, comma, reset_period=10_000_000):
         self.data = Signal(20)
-        self.word_aligned = Signal()
+        self.comma_aligned = Signal()
+        self.comma_realigned = Signal()
+        self.comma_det = Signal()
 
-        self.aligner_en_rxclk = Signal()
-        self.ready = Signal()
-        self.restart = Signal()
+        self.aligner_en = Signal()
+        self.ready_sys = Signal()
+        self.restart_sys = Signal()
 
         # # #
 
 
+        # periodically reset rx until rx is connected and receiving valid data
+        # as after connecting RXP/RXN, the whole RX need to be reset 
         
-        # The built-in RXBYTEISALIGNED can be falsely asserted at linerate higher than 5Gbps
-        # - UG476 (v1.12.1) p.228
-
-        # The validity of data & comma are checked externally
-
-
-        check_counter = Signal(reset=check_period-1, max=check_period)
-        # check = Signal()
+        reset_counter = Signal(reset=reset_period-1, max=reset_period)
         self.sync += [
-            self.restart.eq(0),
-            # check.eq(0),
-            # check_reset.i.eq(0),
-            If(~self.ready,
-                If(check_counter == 0,
-                    check_counter.eq(check_counter.reset),
-                    # check.eq(1),
-                    self.restart.eq(1),
-                    # check_reset.i.eq(1),
+            self.restart_sys.eq(0),
+            If(~self.ready_sys,
+                If(reset_counter == 0,
+                    reset_counter.eq(reset_counter.reset),
+                    self.restart_sys.eq(1),
                 ).Else(
-                    check_counter.eq(check_counter - 1),
+                    reset_counter.eq(reset_counter - 1),
                 )
             )
         ]
 
-        # WIP: 
+
+        # Data and comma checker
+        # From UG476 (v1.12.1) p.228
+        # The built-in RXBYTEISALIGNED can be falsely asserted at linerate higher than 5Gbps
+        # The validity of data and comma needed to be checked externally
 
         comma_n = ~comma & 0b1111111111
 
-        has_error = Signal()
-        comma_aligned = Signal()
+        # DEBUG: remove after use
+        self.has_comma = Signal()
+        self.has_error = Signal()
+
         comma_seen = Signal()
         error_seen = Signal()
         one_counts = Signal(max=11)
 
-        counter_period = 5000
+        # From CXP-001-2021 section 9.2.5.1
+        # For high speed connection an IDLE word shall be transmitted at least once every 100 words
+        counter_period = 200
+
         counter = Signal(reset=counter_period-1, max=counter_period)
         check_reset = Signal()
         check = Signal()
+
         self.sync.cxp_gtx_rx += [
-            check.eq(0),
-            If(counter == 0,
+            If(check_reset,
                 counter.eq(counter.reset),   
+                check.eq(0),
+            ).Elif(counter == 0,
                 check.eq(1),
             ).Else(
                 counter.eq(counter - 1),   
-
             ),
 
-
-            has_error.eq(0),
-            one_counts.eq(reduce(add, [self.data[i] for i in range(10, 20)])),
-            If((one_counts != 4) & (one_counts != 5) & (one_counts != 6),
-                has_error.eq(1)
-            ),
-
-            comma_aligned.eq(0),
-            If((self.data[:10] == comma) | (self.data[:10] == comma_n),
-                comma_aligned.eq(1)
-            ),
-
-            # signal that need to be manually cleared
             If(check_reset,
                 comma_seen.eq(0),
             ).Elif((self.data[:10] == comma) | (self.data[:10] == comma_n),
                 comma_seen.eq(1)
             ),
 
-            one_counts.eq(reduce(add, [self.data[i] for i in range(10, 20)])),
+            one_counts.eq(reduce(add, [self.data[i] for i in range(10)])),
             If(check_reset,
                 error_seen.eq(0),
             ).Elif((one_counts != 4) & (one_counts != 5) & (one_counts != 6),
                 error_seen.eq(1),
             ),
+
+            # DEBUG: 
+            self.has_comma.eq(0),
+            If((self.data[:10] == comma) | (self.data[:10] == comma_n),
+               self.has_comma.eq(1),
+            ),
+
+            self.has_error.eq(0),
+            If((one_counts != 4) & (one_counts != 5) & (one_counts != 6),
+                self.has_error.eq(1),
+            ),
         ]
 
         # DEBUG: expose signal
         self.check_reset = Signal()
-        self.comma_aligned = Signal()
-        self.comma_seen = Signal()
-        self.has_error = Signal()
-        self.error_seen = Signal()
         self.comb +=[
             self.check_reset.eq(check_reset),
-            self.comma_aligned.eq(comma_aligned),
-            self.comma_seen.eq(comma_seen),
-            self.has_error.eq(has_error),
-            self.error_seen.eq(error_seen),
         ]
 
-        self.submodules.rxfsm = rxfsm = ClockDomainsRenamer("cxp_gtx_rx")(FSM(reset_state="ALIGNING"))
+        self.submodules.rxfsm = rxfsm = ClockDomainsRenamer("cxp_gtx_rx")(FSM(reset_state="WAIT_COMMA"))
 
-        # the data from gtxe2 is delayed and checking at the output may not reflect the alignment inside the aligner
-        # thus, failing to alignment on high linerate >5Gbps is common due to the aligner_en being asserted longer than necessary and lead to a lose of lock
-        # a timer is used to wait till the "aligned" data to arrive and do a system check like the datasheet suggested
-        self.submodules.timer = timer = ClockDomainsRenamer("cxp_gtx_rx")(WaitTimer(5000))
+        rxfsm.act("WAIT_COMMA",
+            If(self.comma_det,
+                # # start aligner early, so word aligned will fall
+                # self.aligner_en_rxclk.eq(1),
+                NextState("ALIGNING"),
+            )
+        )
 
         rxfsm.act("ALIGNING",
-            self.aligner_en_rxclk.eq(1),
-            If(self.word_aligned,
-                check_reset.eq(1),
+            If(self.comma_aligned & (~self.comma_realigned),
                 NextState("WAIT_ALIGNED_DATA"),               
+            ).Else(
+                self.aligner_en.eq(1),
             )
         )
 
+        # wait for the aligned data to arrive at the FPGA RX interface 
+        # as there is a delay before the data is avaiable after RXBYTEISALIGNED is asserted
+        self.submodules.timer = timer = ClockDomainsRenamer("cxp_gtx_rx")(WaitTimer(10_000))
+
         rxfsm.act("WAIT_ALIGNED_DATA",
             timer.wait.eq(1),
             If(timer.done,
@@ -459,19 +436,19 @@ class Comma_Detector(Module):
                 If(comma_seen & (~error_seen),
                     NextState("READY"),
                 ).Else(
-                    NextState("ALIGNING")
+                    NextState("WAIT_COMMA")
                 )
             )
         )
 
         ready = Signal()
-        self.specials += MultiReg(ready, self.ready)
+        self.specials += MultiReg(ready, self.ready_sys)
         rxfsm.act("READY",
             ready.eq(1),
             If(check,
                 check_reset.eq(1),
                 If(~(comma_seen & (~error_seen)),
-                    NextState("ALIGNING"),
+                    NextState("WAIT_COMMA"),
                 )
             )
         )
@@ -539,13 +516,15 @@ class GTX(Module):
         txdata = Signal(40)
         rxdata = Signal(40)
 
-        word_aligned = Signal()
+        comma_aligned = Signal()
+        comma_realigned = Signal()
+        comma_det = Signal()
         comma_aligner_en = Signal()
         # Note: the following parameters were set after consulting AR45360
         self.specials += \
             Instance("GTXE2_CHANNEL",
                 # PMA Attributes
-                p_PMA_RSV=0x00018480,
+                p_PMA_RSV=0x001E7080,
                 p_PMA_RSV2=0x2050,              # PMA_RSV2[5] = 0: Eye scan feature disabled
                 p_PMA_RSV3=0,
                 p_PMA_RSV4=1,                   # PMA_RSV[4],RX_CM_TRIM[2:0] = 0b1010: Common mode 800mV
@@ -599,7 +578,7 @@ class GTX(Module):
 
                 # TX data
                 p_TX_DATA_WIDTH=40,
-                p_TX_INT_DATAWIDTH=1,
+                p_TX_INT_DATAWIDTH=1, # 1 if a line rate is greater than 6.6 Gbps
                 i_TXCHARDISPMODE=Cat(txdata[9], txdata[19], txdata[29], txdata[39]),
                 i_TXCHARDISPVAL=Cat(txdata[8], txdata[18], txdata[28], txdata[38]),
                 i_TXDATA=Cat(txdata[:8], txdata[10:18], txdata[20:28], txdata[30:38]),
@@ -632,7 +611,10 @@ class GTX(Module):
                 i_RXDFEXYDEN=1,
                 i_RXDFEXYDHOLD=0,
                 i_RXDFEXYDOVRDEN=0,
-                i_RXLPMEN=0,                    # RXLPMEN = 0: DFE mode is enabled for non scramble data
+                i_RXLPMEN=1,                    # RXLPMEN = 1: LPM mode is enable for non scramble 8b10b data
+                p_RXLPM_HF_CFG=0b00000011110000,
+                p_RXLPM_LF_CFG=0b00000011110000,
+
                 p_RX_DFE_GAIN_CFG=0x0207EA,
                 p_RX_DFE_VP_CFG=0b00011111100000011,
                 p_RX_DFE_UT_CFG=0b10001000000000000,
@@ -662,7 +644,7 @@ class GTX(Module):
 
                 # RX data
                 p_RX_DATA_WIDTH=40,
-                p_RX_INT_DATAWIDTH=1,
+                p_RX_INT_DATAWIDTH=1,   # 1 if a line rate is greater than 6.6 Gbps
                 o_RXDISPERR=Cat(rxdata[9], rxdata[19], rxdata[29], rxdata[39]),
                 o_RXCHARISK=Cat(rxdata[8], rxdata[18], rxdata[28], rxdata[38]),
                 o_RXDATA=Cat(rxdata[:8], rxdata[10:18], rxdata[20:28], rxdata[30:38]),
@@ -675,7 +657,7 @@ class GTX(Module):
                 p_ALIGN_MCOMMA_VALUE=0b1010000011,
                 p_ALIGN_PCOMMA_DET="TRUE",
                 p_ALIGN_PCOMMA_VALUE=0b0101111100,
-                p_SHOW_REALIGN_COMMA="TRUE",
+                p_SHOW_REALIGN_COMMA="FALSE",
                 p_RXSLIDE_AUTO_WAIT=7,
                 p_RXSLIDE_MODE="OFF",
                 p_RX_SIG_VALID_DLY=10,
@@ -683,7 +665,9 @@ class GTX(Module):
                 i_RXMCOMMAALIGNEN=comma_aligner_en,
                 i_RXCOMMADETEN=1,
                 i_RXSLIDE=0,
-                o_RXBYTEISALIGNED=word_aligned,
+                o_RXBYTEISALIGNED=comma_aligned,
+                o_RXBYTEREALIGN=comma_realigned,
+                o_RXCOMMADET=comma_det,
 
                 # RX 8B/10B Decoder Attributes
                 p_RX_DISPERR_SEQ_MATCH="FALSE",
@@ -716,8 +700,8 @@ class GTX(Module):
                 p_RX_DEFER_RESET_BUF_EN="TRUE",
 
                 # CDR Attributes
-                p_RXCDR_CFG=0x03_0000_23FF_1040_0020,   # DFE @ <= 6.6Gb/s, 8B/10B encoded data, CDR setting < +/- 200ppm
-                                                        # (See UG476 (v1.12.1), p.205)
+                p_RXCDR_CFG=0x03_0000_23FF_1008_0020,   # LPM @ 0.5G-1.5625G , 8B/10B encoded data, CDR setting < +/- 200ppm
+                                                        # (See UG476 (v1.12.1), p.206)
                 p_RXCDR_FR_RESET_ON_EIDLE=0b0,
                 p_RXCDR_HOLD_DURING_EIDLE=0b0,
                 p_RXCDR_PH_RESET_ON_EIDLE=0b0,
@@ -822,13 +806,15 @@ class GTX(Module):
         ]
 
 
-        self.submodules.comma_det = comma_det = Comma_Detector(0b0101111100)
+        self.submodules.comma_checker = comma_checker = Comma_Checker(0b0101111100)
         self.comb += [
-            comma_det.data.eq(rxdata),
-            comma_det.word_aligned.eq(word_aligned),
-            comma_aligner_en.eq(comma_det.aligner_en_rxclk),
-            self.rx_ready.eq(comma_det.ready),
+            comma_checker.data.eq(rxdata),
+            comma_checker.comma_aligned.eq(comma_aligned),
+            comma_checker.comma_realigned.eq(comma_realigned),
+            comma_checker.comma_det.eq(comma_det),
+            comma_aligner_en.eq(comma_checker.aligner_en),
+            self.rx_ready.eq(comma_checker.ready_sys),
 
-            rx_init.restart.eq(self.rx_restart | comma_det.restart),
+            rx_init.restart.eq(self.rx_restart | comma_checker.restart_sys),
             tx_init.restart.eq(self.tx_restart),
         ]

src/gateware/zc706.py

@@ -691,6 +691,11 @@ class CXP_FMC():
         )
         self.csr_devices.append("cxp")
 
+        # max freq of cxp_gtx_rx = linerate/internal_datawidth = 12.5Gbps/40 = 312.5MHz
+        platform.add_period_constraint(self.cxp.downconn.gtx.cd_cxp_gtx_tx.clk, 3.2)
+        platform.add_period_constraint(self.cxp.downconn.gtx.cd_cxp_gtx_rx.clk, 3.2)
+        platform.add_false_path_constraints(self.cxp.downconn.gtx.cd_cxp_gtx_tx.clk, self.cxp.downconn.gtx.cd_cxp_gtx_rx.clk)
+
         rtio_channels = []
         # FIXME remove this placeholder RTIO channel
         # There are too few RTIO channels and cannot be compiled (adr width issue of the lane distributor)

src/libboard_artiq/src/cxp_downconn.rs

@@ -22,38 +22,11 @@ pub enum CXP_SPEED {
     CXP_12,
 }
 
-fn loopback_testing(timer: &mut GlobalTimer, data: u8, control_bit: u8) {
+pub fn loopback_testing(timer: &mut GlobalTimer, speed: CXP_SPEED) {
+    println!("==============================================================================");
+    CXP_GTX::change_linerate(timer, speed);
+
     unsafe {
-        // send K28_5 for CDR to align
-        const K28_5: u8 = 0xBC;
-        const K28_1: u8 = 0x3C;
-        const D21_5: u8 = 21 | 5 << 5;
-
-        // NOTE: testing with IDLE word + manual comma alignment
-        // 62.5MHz   OK
-        // 125MHz    OK
-        // 156.25MHz
-        // 250MHz    OK
-        // 312.5MHz
-        // 500MHz    ?? havn't change CDR config yet
-        const LEN: usize = 4;
-        const DATA: [[u8; LEN]; 2] = [
-            // [K28_5, K28_1, K28_5, K28_1],
-            // [1, 1, 1, 1],
-            [K28_5, K28_1, K28_1, D21_5],
-            [1, 1, 1, 0],
-        ];
-
-        csr::cxp::downconn_data_0_write(DATA[0][0]);
-        csr::cxp::downconn_data_1_write(DATA[0][1]);
-        csr::cxp::downconn_data_2_write(DATA[0][2]);
-        csr::cxp::downconn_data_3_write(DATA[0][3]);
-
-        csr::cxp::downconn_control_bit_0_write(DATA[1][0]);
-        csr::cxp::downconn_control_bit_1_write(DATA[1][1]);
-        csr::cxp::downconn_control_bit_2_write(DATA[1][2]);
-        csr::cxp::downconn_control_bit_3_write(DATA[1][3]);
-
         info!("waiting for tx&rx setup...");
         timer.delay_us(50_000);
         info!(
@@ -66,19 +39,16 @@ fn loopback_testing(timer: &mut GlobalTimer, data: u8, control_bit: u8) {
         csr::cxp::downconn_txenable_write(1);
 
         info!("waiting for rx to align...");
-        // timer.delay_us(50_000);
-        // while csr::cxp::downconn_rx_ready_read() != 1 {}
-        // info!("rx ready!");
+        while csr::cxp::downconn_rx_ready_read() != 1 {}
+        info!("rx ready!");
 
-        // println!("0xA8 = {:#06x}", read(0x62));
-        // write(0x62, 0x001A);
-        // println!("0xA8 = {:#06x}", read(0x62));
-
-        // for _ in 0..20 {
-        loop {
-            // NOTE: raw data
-            let data0 = csr::cxp::downconn_rxdata_0_read();
-            let data1 = csr::cxp::downconn_rxdata_1_read();
+        // loop {
+        for _ in 0..20 {
+            // NOTE: raw bits
+            // let data0 = csr::cxp::downconn_rxdata_0_read();
+            // let data1 = csr::cxp::downconn_rxdata_1_read();
+            // let data2 = csr::cxp::downconn_rxdata_2_read();
+            // let data3 = csr::cxp::downconn_rxdata_3_read();
             // let rxready = csr::cxp::downconn_rx_ready_read();
             // timer.delay_us(100);
             // if data0 == 0b0101111100 || data0 == 0b1010000011 {
@@ -105,14 +75,30 @@ fn loopback_testing(timer: &mut GlobalTimer, data: u8, control_bit: u8) {
             //     timer.delay_us(1_000_000);
             // }
 
-            println!("0b{:010b}{:010b}", data0, data1);
-            timer.delay_us(1_000_000);
+            // timer.delay_us(1_000_000);
+            // NOTE: raw bits
+            // let data0 = csr::cxp::downconn_rxdata_0_read();
+            // let data1 = csr::cxp::downconn_rxdata_1_read();
+            // let data2 = csr::cxp::downconn_rxdata_2_read();
+            // let data3 = csr::cxp::downconn_rxdata_3_read();
+            // println!(
+            //     "0b{:010b} {:010b} {:010b} {:010b}",
+            //     data0, data1, data2, data3
+            // );
 
             // NOTE:decode data
-            // let data0_decoded = csr::cxp::downconn_decoded_data_0_read();
             // let data0_k = csr::cxp::downconn_decoded_k_0_read();
-            // let data1_decoded = csr::cxp::downconn_decoded_data_1_read();
             // let data1_k = csr::cxp::downconn_decoded_k_1_read();
+            // let data2_k = csr::cxp::downconn_decoded_k_2_read();
+            // let data3_k = csr::cxp::downconn_decoded_k_3_read();
+            let data0_decoded = csr::cxp::downconn_decoded_data_0_read();
+            let data1_decoded = csr::cxp::downconn_decoded_data_1_read();
+            let data2_decoded = csr::cxp::downconn_decoded_data_2_read();
+            let data3_decoded = csr::cxp::downconn_decoded_data_3_read();
+            println!(
+                "{:#04x} {:#04x} {:#04x} {:#04x}",
+                data0_decoded, data1_decoded, data2_decoded, data3_decoded,
+            );
             // println!(
             //     "decoded_data[0] = {:#04x} decoded_k[0] = {:#b} decoded_data[1] = {:#04x} decoded_k[1] = {:#b}",
             //     data0_decoded,
@@ -120,12 +106,11 @@ fn loopback_testing(timer: &mut GlobalTimer, data: u8, control_bit: u8) {
             //     data1_decoded,
             //     data1_k,
             // );
-            // timer.delay_us(1_000_000);
         }
     }
 }
 
-pub fn setup(timer: &mut GlobalTimer, speed: CXP_SPEED) {
+pub fn setup(timer: &mut GlobalTimer) {
     unsafe {
         info!("turning on pmc loopback mode...");
         csr::cxp::downconn_loopback_mode_write(0b010); // Near-End PMA Loopback
@@ -147,12 +132,9 @@ pub fn setup(timer: &mut GlobalTimer, speed: CXP_SPEED) {
             csr::cxp::downconn_txinit_phaligndone_read(),
             csr::cxp::downconn_rxinit_phaligndone_read(),
         );
-        println!("==============================================================================");
     }
 
-    CXP_GTX::change_linerate(timer, speed);
-
-    loopback_testing(timer, 0x00, 0);
+    CXP_GTX::change_linerate(timer, CXP_SPEED::CXP_1);
 }
 
 pub mod CXP_GTX {
@@ -189,70 +171,32 @@ pub mod CXP_GTX {
     }
 
     fn change_qpll_settings(speed: CXP_SPEED) {
-        match speed {
-            CXP_SPEED::CXP_12 => {
-                panic!("CXP 12.5Gbps is not supported on zc706");
-            }
-            _ => {}
-        }
-
-        // this switches between High and Low band VCO
-        // NOT needed if VCO can do 12.5GHz
-        let qpll_cfg_reg0 = match speed {
-            // NOTE: for ZC706 QPLL VCO that cannot go up to 12.5GHz
-            CXP_SPEED::CXP_1 | CXP_SPEED::CXP_2 | CXP_SPEED::CXP_5 | CXP_SPEED::CXP_10 => 0x0181,
-            CXP_SPEED::CXP_3 | CXP_SPEED::CXP_6 | CXP_SPEED::CXP_12 => 0x01C1,
-        };
-
-        // Change QPLL_REFCLK_DIV
-        let qpll_div_reg0 = match speed {
-            // NOTE: for ZC706 QPLL VCO that cannot go up to 12.5GHz
-            CXP_SPEED::CXP_1 | CXP_SPEED::CXP_2 | CXP_SPEED::CXP_5 | CXP_SPEED::CXP_10 => 0x8068,
-            CXP_SPEED::CXP_3 | CXP_SPEED::CXP_6 | CXP_SPEED::CXP_12 => 0x0068,
-        };
-
         // Change QPLL_FBDIV
-        let qpll_div_reg1 = match speed {
-            CXP_SPEED::CXP_1 | CXP_SPEED::CXP_2 | CXP_SPEED::CXP_5 | CXP_SPEED::CXP_10 => 0x0120,
-            CXP_SPEED::CXP_3 | CXP_SPEED::CXP_6 | CXP_SPEED::CXP_12 => 0x0170,
+        let qpll_div_reg = match speed {
+            CXP_SPEED::CXP_1 | CXP_SPEED::CXP_2 | CXP_SPEED::CXP_5 | CXP_SPEED::CXP_10 => 0x0120, // FB_Divider = 80
+            CXP_SPEED::CXP_3 | CXP_SPEED::CXP_6 | CXP_SPEED::CXP_12 => 0x0170,                    // FB_Divider = 100
         };
 
-        println!("0x32 = {:#018b}", qpll_read(0x32));
-        qpll_write(0x32, qpll_cfg_reg0);
-        println!("0x32 = {:#018b}", qpll_read(0x32));
-
-        println!("0x33 = {:#018b}", qpll_read(0x33));
-        qpll_write(0x33, qpll_div_reg0);
-        println!("0x33 = {:#018b}", qpll_read(0x33));
-
         println!("0x36 = {:#018b}", qpll_read(0x36));
-        qpll_write(0x36, qpll_div_reg1);
+        qpll_write(0x36, qpll_div_reg);
         println!("0x36 = {:#018b}", qpll_read(0x36));
 
-        let divider = match speed {
-            // NOTE: for ZC706 QPLL VCO that cannot go up to 12.5GHz
-            CXP_SPEED::CXP_1 => 0b100,                     // Divided by 8
-            CXP_SPEED::CXP_2 => 0b011,                     // Divided by 4
-            CXP_SPEED::CXP_5 | CXP_SPEED::CXP_3 => 0b010,  // Divided by 2
-            CXP_SPEED::CXP_10 | CXP_SPEED::CXP_6 => 0b001, // Divided by 1
-            CXP_SPEED::CXP_12 => 0b000,
-            // NOTE: for ZC706 QPLL VCO that go up to 12.5GHz
-            // CXP_SPEED::CXP_1 => 0b100,                      // Divided by 8
-            // CXP_SPEED::CXP_2 | CXP_SPEED::CXP_3 => 0b011,   // Divided by 4
-            // CXP_SPEED::CXP_5 | CXP_SPEED::CXP_6 => 0b010,   // Divided by 2
-            // CXP_SPEED::CXP_10 | CXP_SPEED::CXP_12 => 0b001, // Divided by 1
+        let rxout_div = match speed {
+            CXP_SPEED::CXP_1 => 0b100,                      // 8
+            CXP_SPEED::CXP_2 | CXP_SPEED::CXP_3 => 0b011,   // 4
+            CXP_SPEED::CXP_5 | CXP_SPEED::CXP_6 => 0b010,   // 2
+            CXP_SPEED::CXP_10 | CXP_SPEED::CXP_12 => 0b001, // 1
         };
 
         unsafe {
-            csr::cxp::downconn_rx_div_write(divider);
-            csr::cxp::downconn_tx_div_write(divider);
+            csr::cxp::downconn_rx_div_write(rxout_div);
+            csr::cxp::downconn_tx_div_write(rxout_div);
         }
     }
 
     fn change_cdr_cfg(speed: CXP_SPEED) {
-        // NOTE: for ZC706 QPLL VCO that cannot go up to 12.5GHz
         let cdr_cfg = match speed {
-            // Divided by 8
+            // rxout_div = 8
             CXP_SPEED::CXP_1 => {
                 RX_CDR_CFG {
                     cfg_reg0: 0x0020, //0x0A8
@@ -262,8 +206,8 @@ pub mod CXP_GTX {
                     cfg_reg4: 0x0003, //0x0AC
                 }
             }
-            // Divided by 4
-            CXP_SPEED::CXP_2 => {
+            // rxout_div = 4
+            CXP_SPEED::CXP_2 | CXP_SPEED::CXP_5 => {
                 RX_CDR_CFG {
                     cfg_reg0: 0x0020, //0x0A8
                     cfg_reg1: 0x1010, //0x0A9
@@ -272,8 +216,8 @@ pub mod CXP_GTX {
                     cfg_reg4: 0x0003, //0x0AC
                 }
             }
-            // Divided by 2
-            CXP_SPEED::CXP_3 | CXP_SPEED::CXP_5 => {
+            // rxout_div = 2
+            CXP_SPEED::CXP_3 | CXP_SPEED::CXP_6 => {
                 RX_CDR_CFG {
                     cfg_reg0: 0x0020, //0x0A8
                     cfg_reg1: 0x1020, //0x0A9
@@ -282,17 +226,17 @@ pub mod CXP_GTX {
                     cfg_reg4: 0x0003, //0x0AC
                 }
             }
-            // Divided by 1
-            CXP_SPEED::CXP_6 => {
-                RX_CDR_CFG {
-                    cfg_reg0: 0x0020, //0x0A8
-                    cfg_reg1: 0x1040, //0x0A9
-                    cfg_reg2: 0x23FF, //0x0AA
-                    cfg_reg3: 0x0000, //0x0AB
-                    cfg_reg4: 0x0003, //0x0AC
-                }
-            }
-            // Divided by 1
+            // // Divided by 1
+            // CXP_SPEED::CXP_6 => {
+            //     RX_CDR_CFG {
+            //         cfg_reg0: 0x0020, //0x0A8
+            //         cfg_reg1: 0x1040, //0x0A9
+            //         cfg_reg2: 0x23FF, //0x0AA
+            //         cfg_reg3: 0x0000, //0x0AB
+            //         cfg_reg4: 0x0003, //0x0AC
+            //     }
+            // }
+            // rxout_div = 1
             CXP_SPEED::CXP_10 | CXP_SPEED::CXP_12 => {
                 RX_CDR_CFG {
                     cfg_reg0: 0x0020, //0x0A8