From 060dbcc70d8ced1532f9c20d24a084086cd1fdbe Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?J=C4=99drzej=20Boczar?= Date: Thu, 27 May 2021 14:20:47 +0200 Subject: [PATCH] phy/lpddr4: extract command serialization logic into separate class --- litedram/phy/lpddr4/basephy.py | 64 +++++++---------------------- litedram/phy/utils.py | 75 ++++++++++++++++++++++++++++++++++ 2 files changed, 89 insertions(+), 50 deletions(-) diff --git a/litedram/phy/lpddr4/basephy.py b/litedram/phy/lpddr4/basephy.py index f3961f3..1940071 100644 --- a/litedram/phy/lpddr4/basephy.py +++ b/litedram/phy/lpddr4/basephy.py @@ -6,7 +6,6 @@ from operator import or_ from functools import reduce -from collections import defaultdict from migen import * @@ -15,8 +14,8 @@ from litex.soc.interconnect.csr import * from litedram.common import * from litedram.phy.dfi import * -from litedram.phy.utils import ( - bitpattern, delayed, ConstBitSlip, DQSPattern, Serializer, Deserializer, Latency) +from litedram.phy.utils import (bitpattern, delayed, DQSPattern, Serializer, Deserializer, Latency, + CommandsPipeline) from litedram.phy.lpddr4.commands import DFIPhaseAdapter @@ -210,55 +209,20 @@ class LPDDR4PHY(Module, AutoCSR): ] # LPDDR4 Commands -------------------------------------------------------------------------- - # Each LPDDR4 command can span several phases (2 or 4), so the commands cannot overlap. - # This should be guaranteed by the controller based on module timings, but we also include - # an overlaps check in PHY logic. - # Basic check will make sure that no command will be sent to DRAM if there was any command - # sent by the controller on DFI during 3 previous cycles. The extended version will instead - # make sure no command is sent to DRAM if there was any command _actually sent to DRAM_ - # during 3 previous cycles. This is more expensive in terms of resources and generally not - # needed. + # Each LPDDR4 command can span several phases (2 or 4), so in theory the commands could + # overlap. No overlap should be guaranteed by the controller based on module timings, but + # we also include an overlaps check in PHY logic. + self.submodules.commands = CommandsPipeline(adapters, + cs_ser_width = len(self.out.cs), + ca_ser_width = len(self.out.ca[0]), + ca_nbits = len(self.out.ca), + cmd_nphases_span = 4, + extended_overlaps_check = extended_overlaps_check + ) - # Create a history of valid adapters used for masking overlapping ones - valids = ConstBitSlip(dw=nphases, cycles=1, slp=0) - self.submodules += valids - self.comb += valids.i.eq(Cat(a.valid for a in adapters)) - valids_hist = valids.r - if extended_overlaps_check: - valids_hist = Signal.like(valids.r) - for i in range(len(valids_hist)): - was_valid_before = reduce(or_, valids_hist[max(0, i-3):i], 0) - self.comb += valids_hist[i].eq(valids.r[i] & ~was_valid_before) - - cs_per_adapter = [] - ca_per_adapter = defaultdict(list) - for phase, adapter in enumerate(adapters): - # The signals from an adapter can be used if there were no commands on 3 previous cycles - allowed = ~reduce(or_, valids_hist[nphases+phase - 3:nphases+phase]) - - # Use CS and CA of given adapter slipped by `phase` bits - cs_bs = ConstBitSlip(dw=nphases, cycles=1, slp=phase) - self.submodules += cs_bs - self.comb += cs_bs.i.eq(Cat(adapter.cs)), - cs_mask = Replicate(allowed, len(cs_bs.o)) - cs = cs_bs.o & cs_mask - cs_per_adapter.append(cs) - - # For CA we need to do the same for each bit - ca_bits = [] - for bit in range(6): - ca_bs = ConstBitSlip(dw=nphases, cycles=1, slp=phase) - self.submodules += ca_bs - ca_bit_hist = [adapter.ca[i][bit] for i in range(4)] - self.comb += ca_bs.i.eq(Cat(*ca_bit_hist)), - ca_mask = Replicate(allowed, len(ca_bs.o)) - ca = ca_bs.o & ca_mask - ca_per_adapter[bit].append(ca) - - # OR all the masked signals - self.comb += self.out.cs.eq(reduce(or_, cs_per_adapter)) + self.comb += self.out.cs.eq(self.commands.cs) for bit in range(6): - self.comb += self.out.ca[bit].eq(reduce(or_, ca_per_adapter[bit])) + self.comb += self.out.ca[bit].eq(self.commands.ca[bit]) # DQ --------------------------------------------------------------------------------------- dq_oe = Signal() diff --git a/litedram/phy/utils.py b/litedram/phy/utils.py index 96a92f9..0a80071 100644 --- a/litedram/phy/utils.py +++ b/litedram/phy/utils.py @@ -8,6 +8,7 @@ import re from fractions import Fraction from functools import reduce from operator import or_ +from collections import defaultdict from migen import * @@ -120,6 +121,80 @@ class Latency: return "Latency({} sys clk)".format(self._sys) +class CommandsPipeline(Module): + """Commands pipeline logic for LPDDR4/LPDDR5 + + Single DFI command may require more than one LPDDR4/LPDDR5 command, effectively + spanning multiple DFI phases. This module given a list of DFI phase adapters + will use them to translate DFI commands to data on the CS/CA lines and will + handle any possible overlaps. + + Basic check will make sure that no command will be sent to DRAM if there was any command + sent by the controller on DFI during previous phases. The extended version will instead + make sure no command is sent to DRAM if there was any command _actually sent to DRAM_ + during the previous phasess. This is more expensive in terms of resources and generally + not needed. + + Adapters have to provide the following fields: valid, cs, ca. + """ + def __init__(self, adapters, *, + cs_ser_width, # n bits serialized in controller cycle (depends on CS being SDR/DDR) + ca_ser_width, # n bits serialized in controller cycle (depends on CA being SDR/DDR) + ca_nbits, # number of CA lines (LPDDR4/5 -> 6/7) + cmd_nphases_span, # at most how many phases can a command span + extended_overlaps_check=False): + nphases = len(adapters) + self.cs = Signal(cs_ser_width) + self.ca = [Signal(ca_ser_width) for _ in range(ca_nbits)] + + # # # + + # Number of phases (before the current one) we need to check for overlaps + n_previous = cmd_nphases_span - 1 + + # Create a history of valid adapters used for masking overlapping ones + valids = ConstBitSlip(dw=nphases, cycles=1, slp=0) + self.submodules += valids + self.comb += valids.i.eq(Cat(a.valid for a in adapters)) + valids_hist = valids.r + if extended_overlaps_check: + valids_hist = Signal.like(valids.r) + for i in range(len(valids_hist)): + hist_before = valids_hist[max(0, i-n_previous):i] + was_valid_before = reduce(or_, hist_before, 0) + self.comb += valids_hist[i].eq(valids.r[i] & ~was_valid_before) + + cs_per_adapter = [] + ca_per_adapter = defaultdict(list) + for phase, adapter in enumerate(adapters): + # The signals from an adapter can be used if there were no commands on previous cycles + allowed = ~reduce(or_, valids_hist[nphases+phase - n_previous:nphases+phase]) + + # Use CS and CA of given adapter slipped by `phase` bits + cs_bs = ConstBitSlip(dw=cs_ser_width, cycles=1, slp=phase) + self.submodules += cs_bs + self.comb += cs_bs.i.eq(Cat(adapter.cs)), + cs_mask = Replicate(allowed, len(cs_bs.o)) + cs = cs_bs.o & cs_mask + cs_per_adapter.append(cs) + + # For CA we need to do the same for each bit + ca_bits = [] + for bit in range(ca_nbits): + ca_bs = ConstBitSlip(dw=ca_ser_width, cycles=1, slp=phase) + self.submodules += ca_bs + ca_bit_hist = [adapter.ca[i][bit] for i in range(cmd_nphases_span)] + self.comb += ca_bs.i.eq(Cat(*ca_bit_hist)), + ca_mask = Replicate(allowed, len(ca_bs.o)) + ca = ca_bs.o & ca_mask + ca_per_adapter[bit].append(ca) + + # OR all the masked signals + self.comb += self.cs.eq(reduce(or_, cs_per_adapter)) + for bit in range(ca_nbits): + self.comb += self.ca[bit].eq(reduce(or_, ca_per_adapter[bit])) + + class Serializer(Module): """Serialize given input signal