Gabriel Somlo [Sun, 17 Nov 2019 15:08:50 +0000 (10:08 -0500)]
soc_sdram, bios/sdram: support sdram init for csr_data_width <= 32
Enable SDRAM to be initialized when csr_data_width > 8 bits.
Currently, csr_data_width up to 32 bits is supported.
Read leveling tested with csr_data_width [8, 16, 32] on the
ecp5-versa5g and trellisboard (using yosys/trellis/nextpnr),
and on the nexys4ddr (using Vivado).
Signed-off-by: Gabriel Somlo <gsomlo@gmail.com>
Florent Kermarrec [Sun, 17 Nov 2019 10:57:14 +0000 (11:57 +0100)]
soc/interconnect/packet/Depacketizer: cleanup "ALIGNED-DATA-COPY" state
Florent Kermarrec [Sun, 17 Nov 2019 10:50:09 +0000 (11:50 +0100)]
soc/interconnect/packet/Depacketizer: replace no_payload with sink_d.last
Florent Kermarrec [Sat, 16 Nov 2019 13:39:18 +0000 (14:39 +0100)]
test/test_packet: add randomness on ready output, fix corner-cases on Packetizer/Depacketizer
Florent Kermarrec [Sat, 16 Nov 2019 07:49:04 +0000 (08:49 +0100)]
test/test_packet: add randomness on valid input, fix corner-cases on Packetizer
enjoy-digital [Fri, 15 Nov 2019 17:17:35 +0000 (18:17 +0100)]
Merge pull request #307 from sergachev/master
change >512 B CSR memory exception to a warning
Florent Kermarrec [Fri, 15 Nov 2019 15:19:05 +0000 (16:19 +0100)]
soc/interconnect/packet: fix synthesis (synthesis tools can do all sort of optimizations, but we still need to provide valid verilog :))
Ilia Sergachev [Fri, 15 Nov 2019 14:34:12 +0000 (15:34 +0100)]
change >512 B CSR memory exception to a warning
Florent Kermarrec [Fri, 15 Nov 2019 13:57:31 +0000 (14:57 +0100)]
soc/interconnect/packet: connect error/last_be only present on both sink and source
Florent Kermarrec [Fri, 15 Nov 2019 13:34:56 +0000 (14:34 +0100)]
soc/interconnect/packet: simplify/refactor Packetizer/Depacketizer to keep it simple
To avoid complex FSMs, let the synthesis tool do the simplifications when the FSM states are not reachable.
Florent Kermarrec [Fri, 15 Nov 2019 10:36:52 +0000 (11:36 +0100)]
test/test_packet: add 32/64/128-bit loopback tests (passing :))
Florent Kermarrec [Fri, 15 Nov 2019 10:32:42 +0000 (11:32 +0100)]
test/test_packet: prepare for testing dw > 8-bit
Florent Kermarrec [Fri, 15 Nov 2019 10:25:38 +0000 (11:25 +0100)]
soc/interconnect/packet: update copyright
Vamsi K Vytla [Fri, 15 Nov 2019 10:22:49 +0000 (11:22 +0100)]
soc/interconnect/packet: add > 8-bit support to Packetizer/Depacketizer
With high speed link (10gbps XGMII ethernet for example), stream data_width is generally
> 8-bit which make header/data un-aligned on bytes boundaries. The change allows the
Packetizer/Depacketizer to work on stream with a data_width > 8-bit.
Florent Kermarrec [Fri, 15 Nov 2019 09:57:31 +0000 (10:57 +0100)]
build/sim: cleanup run_as_root
Vamsi K Vytla [Fri, 15 Nov 2019 09:47:13 +0000 (10:47 +0100)]
build/sim/modules: add XGMII 10Gbps ethernet module
Used to simulate SoCs with XGMII 10Gbps ethernet and to do LiteEth verification
Florent Kermarrec [Fri, 15 Nov 2019 09:39:49 +0000 (10:39 +0100)]
sim/ethernet: remove trailing whitespaces
Florent Kermarrec [Fri, 15 Nov 2019 09:29:39 +0000 (10:29 +0100)]
test: add initial test_packet
Use a header with 8,16,32,64,128-bit fields and test a Packetizer/Depacketizer loopback with random field values, random packet data & length.
Florent Kermarrec [Thu, 14 Nov 2019 10:19:07 +0000 (11:19 +0100)]
tools/litex_sim: cleanup/update (no functional change)
Florent Kermarrec [Mon, 11 Nov 2019 17:38:10 +0000 (18:38 +0100)]
tools/litex_term: remove automatic reboot when flashing and clear mem_regions to avoid re-flashing on next reboot(s)
Florent Kermarrec [Fri, 8 Nov 2019 23:00:55 +0000 (00:00 +0100)]
bios/flash: minor cleanup on serialboot flashing, add flash address support
enjoy-digital [Fri, 8 Nov 2019 22:51:49 +0000 (23:51 +0100)]
Merge pull request #305 from FrankBuss/master
adding support to flash an FBI image
Florent Kermarrec [Fri, 8 Nov 2019 22:27:58 +0000 (23:27 +0100)]
soc_core: add integrated-rom-file parameter to allow initializing rom from command line
Florent Kermarrec [Fri, 8 Nov 2019 18:43:01 +0000 (19:43 +0100)]
cores/code_8b10b/Decoder: add basic invalid symbols detection
Check that we have 4,5 or 6 ones in the symbol. This does not report all
invalid symbols but still allow detecting issues with the link.
fb@frank-buss.de [Fri, 8 Nov 2019 16:16:28 +0000 (17:16 +0100)]
adding support to flash an FBI image
Florent Kermarrec [Fri, 8 Nov 2019 12:14:21 +0000 (13:14 +0100)]
software/bios: rename ef command to fe (for consistency)
Florent Kermarrec [Fri, 8 Nov 2019 12:13:54 +0000 (13:13 +0100)]
software/libbase/spiflash: rename CHIP_ERASE_CMD to CE_CMD (for consistency)
enjoy-digital [Fri, 8 Nov 2019 12:04:33 +0000 (13:04 +0100)]
Merge pull request #302 from FrankBuss/master
erase flash command added
Florent Kermarrec [Fri, 8 Nov 2019 11:55:29 +0000 (12:55 +0100)]
soc_core: remove add_cpu method (when no real CPU but only wishbone masters, self.cpu is declared as CPUNone)
fb@frank-buss.de [Thu, 7 Nov 2019 18:19:54 +0000 (19:19 +0100)]
erase flash command added
Florent Kermarrec [Thu, 7 Nov 2019 08:02:31 +0000 (09:02 +0100)]
integration/export: do not include soc.h in csr.h when with_access_functions=False
Idealy we should have another parameter for that.
Florent Kermarrec [Thu, 7 Nov 2019 08:00:54 +0000 (09:00 +0100)]
soc_sdram/kcu105: add optional main_ram_size_limit and use it on KCU105 to limit to 1GB instead of 2GB.
CSR map will need to be updated to support the 2GB.
Florent Kermarrec [Thu, 7 Nov 2019 07:56:52 +0000 (08:56 +0100)]
soc_sdram: remove use_full_memory_we parameter (always used as True)
Florent Kermarrec [Thu, 7 Nov 2019 07:44:34 +0000 (08:44 +0100)]
soc_sdram: update copyrights
enjoy-digital [Thu, 7 Nov 2019 07:40:30 +0000 (08:40 +0100)]
Merge pull request #300 from gsomlo/gls-rocket-axi
RFC: Direct link between Rocket/mem_axi <--> LiteDRAM dataport
Gabriel Somlo [Fri, 1 Nov 2019 12:45:23 +0000 (08:45 -0400)]
cpu/rocket: parameterize axi interface data width
Rocket variants can be configured with axi port data widths that
are multiples of the native word size (64 bits in our case). In
the future, we will add variants with mem_axi data width > 64 bit,
to match the native data width of the LiteDRAM controller on
various development boards (e.g., 128 bits on the ecp5versa, and
256 bits on the trellisboard).
Signed-off-by: Gabriel Somlo <gsomlo@gmail.com>
Gabriel Somlo [Thu, 31 Oct 2019 20:23:36 +0000 (16:23 -0400)]
soc_sdram: remove upper limit on usable main RAM
Revert commit #
68a503174.
Gabriel Somlo [Wed, 30 Oct 2019 14:37:17 +0000 (10:37 -0400)]
cpu/rocket, soc_sdram: Connect mem_axi to LiteDRAM, bypass WB bus
Connect Rocket's dedicated port for cached RAM accesses (mem_axi)
directly to the LiteDRAM data port, bypassing the shared LiteX
(Wishbone) bus.
When both Rocket's mem_axi and LiteDRAM's port have the same data
width, use a native point-to-point AXI connection.
Otherwise, convert both ends to Wishbone, and use the Wishbone
data width converter to bridge the gap.
FIXME: In the future, this part should be replaced with a native
AXI data width converter!
Signed-off-by: Gabriel Somlo <gsomlo@gmail.com>
Florent Kermarrec [Fri, 1 Nov 2019 10:33:43 +0000 (11:33 +0100)]
interconnect/csr_bus/SRAM: add mem_size check
Memory size is limited to 512 bytes:
- CSR region size is 0x800 (4096)
- default csr_data_width is 8
maximum size = 4096/8 = 512 bytes.
Florent Kermarrec [Fri, 1 Nov 2019 10:30:50 +0000 (11:30 +0100)]
soc_core/soc_core_args: specify default cpu (vexriscv)
Florent Kermarrec [Fri, 1 Nov 2019 09:11:12 +0000 (10:11 +0100)]
lattice/diamond/tcl: always use / separators, even on windows
Florent Kermarrec [Fri, 1 Nov 2019 08:25:02 +0000 (09:25 +0100)]
cpu/minerva: elaborate minerva verilog to build directory
Florent Kermarrec [Fri, 1 Nov 2019 08:23:42 +0000 (09:23 +0100)]
soc/integration/builder: pass output_dir to platform, make sure gateware/software directory are created before finalizing
Florent Kermarrec [Thu, 31 Oct 2019 20:15:12 +0000 (21:15 +0100)]
cpu/minerva: generate minerva.v near core.py not in submodule
Florent Kermarrec [Thu, 31 Oct 2019 07:52:04 +0000 (08:52 +0100)]
cpu/minverva: give more explicit error message when not able to elaborate cpu
Tim Ansell [Thu, 31 Oct 2019 03:49:27 +0000 (20:49 -0700)]
Merge pull request #297 from mithro/mem-region-pp
Improve the error message on memory region conflict.
Tim 'mithro' Ansell [Thu, 31 Oct 2019 02:32:20 +0000 (19:32 -0700)]
Improve the error message on memory region conflict.
Before;
```
ValueError: Memory region conflict between rom and main_ram
```
After;
```
ValueError: Memory region conflict between rom (<SoCMemRegion 0x10000000 0x10000 cached>) and main_ram (<SoCMemRegion 0x0 0x20000000 cached>)
```
Fixes #296.
Tim Ansell [Wed, 30 Oct 2019 21:28:48 +0000 (14:28 -0700)]
Merge pull request #293 from mithro/mor1kx-fix
Fix file names for the mor1kx processor.
Tim 'mithro' Ansell [Wed, 30 Oct 2019 20:49:24 +0000 (13:49 -0700)]
Fix file names for the mor1kx processor.
Fixes #292.
Florent Kermarrec [Wed, 30 Oct 2019 15:33:40 +0000 (16:33 +0100)]
targets: use type="io" instead of io_region=True
Florent Kermarrec [Wed, 30 Oct 2019 15:31:27 +0000 (16:31 +0100)]
integration/SoCMemRegion: use type instead of io_region/linker_region and export type to csv/json
Supported types: "cached", "io", "cached+linker", "io+linker", default="cached"
Florent Kermarrec [Mon, 28 Oct 2019 17:32:28 +0000 (18:32 +0100)]
soc_core: add check_regions_overlap method, add linker_region support (overlap is not checked on linker_regions)
Florent Kermarrec [Mon, 28 Oct 2019 16:07:37 +0000 (17:07 +0100)]
soc_core/add_memory_region: fix memory overlap detection
Florent Kermarrec [Mon, 28 Oct 2019 09:59:43 +0000 (10:59 +0100)]
test/test_targets: skip Minerva test on Travis-CI, remove commented tests
Florent Kermarrec [Mon, 28 Oct 2019 09:22:17 +0000 (10:22 +0100)]
cpu/minerva: elaborate from nmigen sources during build, enable hardware multiplier
enjoy-digital [Fri, 25 Oct 2019 10:28:29 +0000 (12:28 +0200)]
Merge pull request #286 from gsomlo/gls-timingstrict
build/lattice/trellis: optionally allow failure if p&r timing not met
Gabriel Somlo [Thu, 24 Oct 2019 17:56:20 +0000 (13:56 -0400)]
build/lattice/trellis: optionally allow failure if p&r timing not met
When timing requirements are strict, allow the build process to fail upon
failure to meet timing. This facilitates running the build process from a
loop, repeatedly, until a "lucky" p&r solution is found, e.g.:
while true; do
litex/boards/targets/versa_ecp5.py --gateware-toolchain trellis \
--sys-clk-freq=60e06 --cpu-type rocket --cpu-variant linux \
--with-ethernet --yosys-nowidelut \
--nextpnr-timingstrict
if [ "$?" == "0" ]; then
echo "Success" | mail -s "Build Succeeded" your@email.here
break
fi
done
This augments commit #
683e0668, which unconditionally forced p&r to
succeed, regardless of whether timing was met, via '--timing-allow-fail'.
Signed-off-by: Gabriel Somlo <gsomlo@gmail.com>
enjoy-digital [Sun, 20 Oct 2019 13:30:22 +0000 (15:30 +0200)]
Merge pull request #283 from kbeckmann/kbeckmann/bios_increment_address
bios: Increment address when writing to flash
Konrad Beckmann [Sat, 19 Oct 2019 20:58:24 +0000 (22:58 +0200)]
bios: Increment address when writing to flash
Florent Kermarrec [Fri, 18 Oct 2019 12:12:01 +0000 (14:12 +0200)]
build/lattice/trellis: use --timing-allow-fail to allow generating bistream when timings are not met
This is the default behaviour of the others tools and allow testing designs on hardware with small violations.
Florent Kermarrec [Fri, 18 Oct 2019 08:26:47 +0000 (10:26 +0200)]
soc/cores/icap/ICAPBitstream: always keep fifo.source.ready to 1
enjoy-digital [Fri, 18 Oct 2019 08:24:20 +0000 (10:24 +0200)]
Merge pull request #282 from antmicro/icapbitstream_fixes
Fix ICAPBitstream
Jan Kowalewski [Fri, 18 Oct 2019 07:33:31 +0000 (09:33 +0200)]
cores/icap/ICAPBitstream: add source ready signal.
Florent Kermarrec [Thu, 17 Oct 2019 10:44:16 +0000 (12:44 +0200)]
soc/integration/__init__: remove imports (not used and causing issues
Florent Kermarrec [Thu, 17 Oct 2019 10:13:06 +0000 (12:13 +0200)]
build: always use platform.add_source and avoid manipulate platform.sources directly
Florent Kermarrec [Thu, 17 Oct 2019 07:52:31 +0000 (09:52 +0200)]
build/generic_platform: replace set with list for sources/verilog_include_paths
Python does not have native OrderedSet and we need to be able to preserve the order of the sources
for some backends (Verilator for instance), so use list instead of set.
Florent Kermarrec [Wed, 16 Oct 2019 12:56:17 +0000 (14:56 +0200)]
cores/icap/ICAPBitstream: simplify, add icap_clk_div parameter, describe how to use it.
Florent Kermarrec [Tue, 15 Oct 2019 10:13:42 +0000 (12:13 +0200)]
build/generic_platform: keep language to None if None after tools.language_by_filename
Florent Kermarrec [Mon, 14 Oct 2019 20:15:02 +0000 (22:15 +0200)]
soc_core: fix default --uart_name
Florent Kermarrec [Mon, 14 Oct 2019 07:12:25 +0000 (09:12 +0200)]
integration/soc_core: expose more SoC parameters
Tim Ansell [Sun, 13 Oct 2019 18:29:46 +0000 (11:29 -0700)]
Merge pull request #280 from kbeckmann/picorv32_typo
picorv32: Fix minimal variant params
Konrad Beckmann [Sun, 13 Oct 2019 10:56:55 +0000 (12:56 +0200)]
picorv32: Fix minimal variant params
The param p_ENABLE_COUNTERS was misspelled.
Florent Kermarrec [Sat, 12 Oct 2019 21:05:53 +0000 (23:05 +0200)]
soc_core: fix soc_core_argdict
Florent Kermarrec [Sat, 12 Oct 2019 17:20:50 +0000 (19:20 +0200)]
cpu/lm32: add missing buses
Florent Kermarrec [Sat, 12 Oct 2019 17:18:57 +0000 (19:18 +0200)]
soc_core/soc_core_argdict: use inspect to get all parameters and simplify
Florent Kermarrec [Fri, 11 Oct 2019 19:55:26 +0000 (21:55 +0200)]
integration: simplify cpu buses auto-conversion (always use Converter, thanks gsomlo)
Florent Kermarrec [Fri, 11 Oct 2019 19:49:11 +0000 (21:49 +0200)]
interconnect/wishbone: fix Converter case when buses are identical
Florent Kermarrec [Fri, 11 Oct 2019 12:28:29 +0000 (14:28 +0200)]
platforms/versa_ecp5: add serdes refclk/sma
Florent Kermarrec [Fri, 11 Oct 2019 07:01:50 +0000 (09:01 +0200)]
cpu/rocket: expose 64-bit buses (use automatic down-conversion of SoCCore)
Florent Kermarrec [Fri, 11 Oct 2019 06:59:25 +0000 (08:59 +0200)]
soc_core: add automatic down-conversion of CPU buses to 32-bit (if needed)
Florent Kermarrec [Fri, 11 Oct 2019 06:41:05 +0000 (08:41 +0200)]
soc/interconnect/axi: re-align to improve readability
Florent Kermarrec [Fri, 11 Oct 2019 06:38:12 +0000 (08:38 +0200)]
software/bios: simplify banners
Florent Kermarrec [Thu, 10 Oct 2019 20:29:54 +0000 (22:29 +0200)]
cpu/picorv32: remove obsolete comment
Florent Kermarrec [Thu, 10 Oct 2019 20:02:04 +0000 (22:02 +0200)]
cpu/picorv32: use a single idbus
Florent Kermarrec [Thu, 10 Oct 2019 19:52:09 +0000 (21:52 +0200)]
cpu: cleanup/re-align
Florent Kermarrec [Thu, 10 Oct 2019 19:40:29 +0000 (21:40 +0200)]
cpu/rocket: rename ibus/dbus to mem_wb/mmio_wb and add size suffix
Florent Kermarrec [Thu, 10 Oct 2019 19:35:06 +0000 (21:35 +0200)]
cpu: add buses list and use it in soc_core to add bus masters
Florent Kermarrec [Thu, 10 Oct 2019 19:15:49 +0000 (21:15 +0200)]
integration: move soc constants to soc.h of csr.h
software retro-compat with soc.h included in csr.h
Florent Kermarrec [Thu, 10 Oct 2019 17:39:33 +0000 (19:39 +0200)]
build/generic_platform: only add sources if language is not None
Florent Kermarrec [Thu, 10 Oct 2019 17:36:17 +0000 (19:36 +0200)]
xilinx/vivado: replace "xy" == language with language == "xy"
enjoy-digital [Thu, 10 Oct 2019 17:31:09 +0000 (19:31 +0200)]
Merge pull request #277 from railnova/feature/vivado_sysverilog_support
[feature] Add SystemVerilog support for the Vivado builder
Florent Kermarrec [Thu, 10 Oct 2019 17:23:01 +0000 (19:23 +0200)]
integration/soc_zynq: shadow_base no longer recommended (replace with io_regions)
Florent Kermarrec [Thu, 10 Oct 2019 17:21:32 +0000 (19:21 +0200)]
bios/main: use same banner than README (MiSoC cited in README/LICENSE)
Florent Kermarrec [Thu, 10 Oct 2019 17:18:28 +0000 (19:18 +0200)]
software/bios: don't show peripherals init banner if nothing to init, add Ethernet init printf
Martin Cornil [Thu, 10 Oct 2019 12:06:37 +0000 (14:06 +0200)]
Add system Verilog support for the Vivado builder
enjoy-digital [Wed, 9 Oct 2019 19:25:57 +0000 (21:25 +0200)]
Merge pull request #276 from gsomlo/gls-rocket-map
cpu/rocket: swap main_mem and io regions
Gabriel Somlo [Wed, 9 Oct 2019 18:25:41 +0000 (14:25 -0400)]
cpu/rocket: swap main_mem and io regions
The total size of RAM (main_mem) can be expected to vary significantly,
and often exceed the size needed for MMIO allocations by a large margin.
As such, place Rocket's MMIO (io regions) below 0x8000_0000, and start
the RAM (main_mem) at 0x8000_0000, with nothing above it to limit its
future growth.
Also, bump the pre-built Rocket verilog submodule to an updated version,
which also comes with matching changes to the way MMIO and RAM accesses
are mapped and routed to their respective AXI interfaces.
Signed-off-by: Gabriel Somlo <gsomlo@gmail.com>
Florent Kermarrec [Wed, 9 Oct 2019 10:05:10 +0000 (12:05 +0200)]
cpu: add default io_regions to CPUNone (all address range can be used as IO)
enjoy-digital [Wed, 9 Oct 2019 09:20:50 +0000 (11:20 +0200)]
Merge pull request #275 from pcotret/patch-1
Update README (related to issue #273)
Florent Kermarrec [Wed, 9 Oct 2019 08:47:19 +0000 (10:47 +0200)]
soc_core: improve check_io_region error message