Commit Graph

8488 Commits

Author SHA1 Message Date
Joel Stanley 917c839c30 github: Update to Ubuntu 22.04
Move to the latest Ubuntu LTS for updated tooling. This requires us to
drop the zlibc package which no longer exists, and doesn't appear to be
required. Also update to the master branch of the GHDL action as this
contains 22.04 support.

Signed-off-by: Joel Stanley <joel@jms.id.au>
2022-11-15 23:08:17 +10:30
Florent Kermarrec a092927139 cores/cpu: Revert custom __init__ (Required by CPUNone designs). 2022-11-15 13:16:50 +01:00
Florent Kermarrec 4baeeed946 soc/cores/cpu: Add reset_address_check attribute and enable/disable methods and use it to disable CPU Reset Address check in Soc.
For designs willing to put the reset address of the CPU in a region not directly handled by LiteX,
self.cpu.disable_reset_address_check() can be used in the SoC to disable CPU reset address check.
2022-11-15 12:03:11 +01:00
enjoy-digital 703bd16a96
Merge pull request #1505 from antmicro/fix-wishbone-arbiter
Fix Wishbone arbiter
2022-11-15 10:11:59 +01:00
enjoy-digital 4afee8535b
Merge pull request #1504 from antmicro/msieron/fix-etherbone-timeouts
remote/comm_udp: Fix Etherbone timeout errors
2022-11-15 10:10:08 +01:00
enjoy-digital b90080ab74
Merge pull request #1503 from shenki/microwatt-ci
Add Microwatt CI
2022-11-15 10:06:22 +01:00
Maciej Dudek 07184d37df Fix Wishbone arbiter
Right now, when multiple masters want to access the bus,
access is granted to one of them, and is not revoked until
selected master has finished all of its transactions (cyc goes low).

This state causes master starvation if access is granted to high
bandwidth master, like cpu in busy loop.

This commit makes it so access to bus is revoked when pending transaction
is finished (ack and cyc are high) or when selected master is idle.

Signed-off-by: Maciej Dudek <mdudek@antmicro.com>
2022-11-14 21:31:24 +01:00
Michal Sieron 9ae0da667b remote/comm_udp: Fix Etherbone timeout errors
This commit fixes a bug where, retry after timeout would send another
request, but wouldn't discard a response that arrived after the timeout.

Retries worked, but only for dropped packets. If a response arrived, but
`socker.recvfrom` timed out, response would still be put in receive
queue. Later after sending another request, client would try to read
from the socket and would find a response. But this response would be
for the old request.
This way request/response pairs would get misaligned and stop working
properly.

This commit adds read numbering (writes do not have responses).
Numbering is achieved by utilizing the fact that responses to Etherbone
reads are actually writes to an address specified in a request.
This way, we don't need to extend Etherbone protocol, in fact we use it
as it is intended.

This numbering is then used to discard responses that don't match
current request.

I also cleaned setting of the timeout, as it was being set in multiple
places, sometimes to values so small that retry was bound to happen.

Signed-off-by: Michal Sieron <msieron@antmicro.com>
2022-11-14 21:24:37 +01:00
Joel Stanley b0b57491bb test_cpu: Add Microwatt to tested CPUs
Now that LiteX can convert from VHDL to Verilog using GHDL, and the
required dependencies are installed in the CI environment, start testing
Microwatt.

Signed-off-by: Joel Stanley <joel@jms.id.au>
2022-11-14 22:09:47 +10:30
Joel Stanley f89746af9f github: Add ppc64le toolchain
The toolchain is used by microwatt.

Signed-off-by: Joel Stanley <joel@jms.id.au>
2022-11-14 22:07:20 +10:30
Joel Stanley 25969237f5 github: Add GHDL to build environment
GHDL provides a github action that adds a nightly build to the CI
environment:

 https://github.com/ghdl/setup-ghdl-ci

Signed-off-by: Joel Stanley <joel@jms.id.au>
2022-11-14 22:07:20 +10:30
Joel Stanley b340b86975 test_cpu: Set number of verilator jobs
By default verilator will be built with -j with no arguments, spawning
many processors. This causes large designs to failure in CI (probably
due to exhausting the memory of the build box):

 Error: Process completed with exit code 143.

Set the number of jobs to the number of CPUs in the system. This allows
designs such as Microwatt to build in CI.

Signed-off-by: Joel Stanley <joel@jms.id.au>
2022-11-14 22:05:54 +10:30
Florent Kermarrec 05b0c59607 interconnect: For now remove the address_width checks; more verification will have to be done before enabling it to avoid regressions. 2022-11-14 10:34:48 +01:00
Florent Kermarrec 9115db5023 soc/cores/video/VideoS7GTPHDMIPHY: Fix typo (probably related to some refactoring). 2022-11-14 09:54:55 +01:00
Florent Kermarrec ec126f0e4d test/test_cpu: Move ibex to untested_cpus since seems to be broken since 2022.11.12. 2022-11-14 09:51:37 +01:00
Florent Kermarrec 269525862b interconnect/axi/axi_lite_to_simple: Avoid combinatorial loop on ax.valid/ax.ready. 2022-11-14 09:34:10 +01:00
Florent Kermarrec 82127043c3 interconnect: Add data_width/address_width checks on InterconnectShared/Crossbar and also propagate address_width. 2022-11-14 09:08:28 +01:00
Florent Kermarrec a3cc741d26 interconnect: Ensure data_width is propagated on all InterconnectShared/Crossbar modules. 2022-11-14 08:34:38 +01:00
enjoy-digital 68c34c64b0
Merge pull request #1501 from bunnie/axi-full-xbar-datawidth
AXICrossbar: absorb data width setting when building crossbar
2022-11-14 08:24:23 +01:00
bunnie d741346f63 AXICrossbar: absorb data width setting when building crossbar
This patch allows the AXICrossbar to absorb non-default datawidths
when the crossbar is built.

e.g. in the case of a 64-bit AXICrossbar, without this patch the
crossbar was only connecting the bottom 32 bits.
2022-11-12 18:29:51 +08:00
enjoy-digital a7475d7f96
Merge pull request #1500 from cklarhorst/cd_fix
core/naxriscv: Don't use os.system to execute sbt
2022-11-12 11:00:36 +01:00
Christian Klarhorst c1c4910d67 core/naxriscv: Don't use os.system to execute sbt
The use of os.system together with cd is a problem because it changes the
CWD for the whole python process. This breaks for example --csr-csv.
2022-11-11 14:14:47 +01:00
Florent Kermarrec c1885b333f build/altera/platform: Don't set keep attribute on clk signal when using add_period_constraints.
This was preventing the build with Quartus.
2022-11-11 10:06:20 +01:00
Florent Kermarrec 62e869296f build/generic_toolchain: Make adding keep attribute to clk signals optionals in add_period_constraint/add_false_path_constraint.
Keep it enabled by default.
2022-11-11 10:05:30 +01:00
Florent Kermarrec 1a66f4a6ad soc: Only do logging.BasicConfig when not already configured by top level script.
Allow having a default logging and overriding it in user scripts.
2022-11-11 09:32:56 +01:00
Florent Kermarrec d738eacf3d build/parser: Add logging_group to configure logging (filename and level for now). 2022-11-11 09:32:48 +01:00
Florent Kermarrec 01b9ae7894 integration/soc: Convert sys_clk_freq to int in SoC to allow passing float to SoC. 2022-11-10 10:08:41 +01:00
Florent Kermarrec 1e2ad2250d compat/soc_core: Fix register_mem/rom missing imports. 2022-11-09 19:11:15 +01:00
enjoy-digital 7157b4c5e8
Merge pull request #1496 from MateuszKarlic/json2renode-update
json2renode: Multicore configuration support
2022-11-09 15:52:31 +01:00
Florent Kermarrec 877dff8a09 soc/compat: Fix add_wb_slave compatibility that was no longer working correclty since finalization order changes.
We should really remove this compatibility layer, but let's wait a bit to make
sure all designs are converted.
2022-11-09 15:40:09 +01:00
Florent Kermarrec 4a740651f0 litex_sim: Simplify configuration by creating a temporary config_soc that is then used for the configuration.
This avoid several workarounds for CPU endianness, Bus data-width, RAM boot offset.
2022-11-09 09:24:30 +01:00
Florent Kermarrec 89afed5970 litex_sim: Switch to new LiteXArgumentParser and let it handle verilator build args. 2022-11-09 08:45:48 +01:00
Florent Kermarrec c39d35de83 build/sim/platform: Add fill_args/get_argdict methods. 2022-11-09 08:45:12 +01:00
Florent Kermarrec a2aa891baa build/paltform: Minor cleanup on supported_toolchains/toolchain_group. 2022-11-09 08:44:49 +01:00
Florent Kermarrec a2b5bb0db2 litex_sim: Switch from self.submodules to self. 2022-11-09 08:21:53 +01:00
enjoy-digital d5df6e23f4
Merge pull request #1490 from Icenowy/litex-sim-membase-hack
tools/litex_sim: hack to allow memory base other than 0x40000000
2022-11-09 08:18:50 +01:00
Mateusz Karlic cd90e2623a json2renode: cpu: Overhaul generate_cpu 2022-11-08 15:32:09 +01:00
Mateusz Karlic eccb26874e json2renode: cpu: Extract minor common logic 2022-11-08 15:32:09 +01:00
Mateusz Karlic 97772eb4bd json2renode: IRQ: Fallback to cpu0 if plic is unavailable 2022-11-08 15:32:09 +01:00
Mateusz Karlic 4c959740dd json2renode: Use cpu_count as local variable 2022-11-08 15:32:09 +01:00
Mateusz Karlic b4bddc68e7 json2renode: Add LiteX MMCM 2022-11-08 15:32:09 +01:00
Mateusz Karlic 65964692dd json2renode: Add auto-align hint 2022-11-08 15:32:09 +01:00
Mateusz Karlic 8b9ca8e9e9 json2renode: Silence false warnings about unsupported peripherals 2022-11-08 15:32:09 +01:00
Mateusz Karlic 283a237876 json2renode: Fix registration of GPIO peripherals 2022-11-08 15:32:09 +01:00
Mateusz Karlic a9caeda30e json2renode: Fix typo 2022-11-08 15:32:09 +01:00
Mateusz Karlic 55dcee16c3 json2renode: Use spiflash->base instead of flash_boot_address 2022-11-08 15:32:09 +01:00
Mateusz Karlic 860ca8673e json2renode: Use opensbi->base for bios binary 2022-11-08 15:32:09 +01:00
Mateusz Karlic ca63d12509 json2renode: Add support for multicore builds 2022-11-08 15:32:09 +01:00
Florent Kermarrec 240b24b7ff gen/fhdl/hierarchy: Use [] for BlackBoxes. 2022-11-08 15:08:12 +01:00
Florent Kermarrec b32d694ae4 interconnect/axi: Do not expose dest on AXI-Full (Only present for on AXI-Stream).
We are using AXI-Stream for AXI-Full channels, so do an exception for dest signal.
2022-11-08 14:52:05 +01:00