Using Chipyard's default configuration ReRoCCManyGemminiConfig, the ReRoCC build failed.

[silencerains]

Background Work

• Yes, I searched the mailing list
• Yes, I searched prior issues
• Yes, I searched the documentation

Chipyard Version and Hash

Release: 1.13.0

OS Setup

LSB Version: core-11.1.0ubuntu2-noarch
Distributor ID: Ubuntu
Description: Ubuntu 20.04.6 LTS
Release: 20.04
Codename: focal

SHELL=/bin/bash
LANG=en_US.UTF-8
PATH=path
CONDA_PREFIX=/home/user/workspace/.conda-env
RISCV=/home/user/workspace]/.conda-env/riscv-tools

Python: 3.10.14
GCC: 13.2.0
CMAKE: 3.26.3
JAVA: OpenJDK 20.0.2
Verilator: 5.022

Other Setup

Ex: Prior steps taken / Documentation Followed / etc...

Current Behavior

When building ReRoCC using the command: make run-binary-debug CONFIG=ReRoCCManyGemminiConfig BINARY=../../generators/rerocc/tests/rerocc.riscv -j 64
an error occurred. Additionally, I configured a custom SingleReRoCC configuration, but the same error persisted.
Technical details:
Command attempted: make run-binary-debug with ReRoCCManyGemminiConfig
Error message: Memory region 'ctrl' at List(AddressSet(0x20000, 0xfff)) only supports TransferSizes[1, 8] Get, but must support TransferSizes[1, 64]
Additional attempt: Custom SingleReRoCC configuration also failed with the same error
Issue: Transfer size mismatch in the control register region (supports 1-8 bytes but requires 1-64 bytes)
The specific error message is:
Exception in thread "main" java.lang.reflect.InvocationTargetException
at ... ()
at chipyard.stage.phases.PreElaboration.$anonfun$transform$1(PreElaboration.scala:35)
at chipyard.stage.phases.PreElaboration$$Lambda$498/485686996.apply(Unknown Source)
at chisel3.Module$.evaluate(Module.scala:88)
at chisel3.Module$.do_apply(Module.scala:35)
at chisel3.stage.phases.Elaborate.$anonfun$transform$2(Elaborate.scala:52)
at chisel3.stage.phases.Elaborate$$Lambda$512/405252209.apply(Unknown Source)
at chisel3.internal.Builder$.$anonfun$buildImpl$1(Builder.scala:1042)
at chisel3.internal.Builder$$$Lambda$515/1130551263.apply(Unknown Source)
at scala.util.DynamicVariable.withValue(DynamicVariable.scala:59)
at chisel3.internal.Builder$.buildImpl(Builder.scala:1032)
at chisel3.internal.Builder$.$anonfun$build$1(Builder.scala:1023)
at chisel3.internal.Builder$$$Lambda$513/1223050066.apply(Unknown Source)
at logger.Logger$.$anonfun$makeScope$4(Logger.scala:148)
at logger.Logger$$$Lambda$119/134310351.apply(Unknown Source)
at scala.util.DynamicVariable.withValue(DynamicVariable.scala:59)
at logger.Logger$.makeScope(Logger.scala:146)
at logger.Logger$.makeScope(Logger.scala:133)
at ... ()
at ... (Stack trace trimmed to user code only. Rerun with --full-stacktrace to see the full stack trace)
Caused by: java.lang.IllegalArgumentException: requirement failed: Memory region 'ctrl' at List(AddressSet(0x20000, 0xfff)) only supports TransferSizes[1, 8] Get, but must support TransferSizes[1, 64]

Expected Behavior

I have very limited knowledge about ReRoCC and I'm completely at a loss regarding this issue. Interestingly, when I used RocketGemminiConfig, no such error occurred.

Other Information

No response

[jerryz123]

Could not reproduce. Please use latest main, not releases.

[silencerains]

Could not reproduce. Please use latest main, not releases.

Jerry, thank you so much for your reply. I followed your suggestion and switched Chipyard to the main branch. However, I encountered some minor issues with the build environment. For instance, during step 3, I received the following error:

========== BEGINNING STEP 3: Building toolchain collateral ==========
Installing extra toolchain utilities/tests to /home/cp.conda-env/riscv-tools
==>  Installing Spike
=>  Starting riscv-isa-sim build
==>  Initializing riscv-isa-sim submodule
+ git submodule update --init --recursive /home/toolchains/riscv-tools/riscv-isa-sim
Cloning into '/home/cp/toolchains/riscv-tools/riscv-isa-sim'...
Submodule path 'toolchains/riscv-tools/riscv-isa-sim': checked out 'b81b5a068636829eb7b299e63a32800b5dd047a0'
==>  cd /home/cp/toolchains/riscv-tools/riscv-isa-sim
+ mkdir -p build
==>  Configuring riscv-isa-sim
+ cd build
+ ../configure --prefix=/home/cp/.conda-env/riscv-tools --with-boost=no --with-boost-asio=no --with-boost-regex=no
checking build system type... x86_64-pc-linux-gnu
checking host system type... x86_64-pc-linux-gnu
checking for gcc... gcc
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables... 
checking whether we are cross compiling... configure: error: in `/home/cp/toolchains/riscv-tools/riscv-isa-sim/build':
configure: error: cannot run C compiled programs.
If you meant to cross compile, use `--host'.
See `config.log' for more details

When I tried using the --host option, although it compiles, things didn't improve. During the simulator build, it appears that the system cannot find the corresponding glibc.
Could you suggest how I should address this problem?

[adamsir32]

hello bro，i have the same question ，did you solve this problem？

[Fi50]

Hello! Did you try to rerun CY build-setup when you switched to the latest version and source the new environment?
If you are still having this issue, could you clarify what command you're running this time, output you get during although it compiles, things didn't improve and the system cannot find the corresponding glibc? You might also want to take a look at the config.log and post anything relevant to the error.

[Fi50]

You might be facing the same issue as #2241 ? They're also on Ubuntu 20. If so, it's best to move the conversation there as the original issue in this thread was resolved. Feel free to reopen this if needed though!

[suhwanhong999]

I am having the same issues as well. Using some of the latest commits to master, I am able to get past the setup step 3 issue for most of them but still run into the original problem when trying to make CONFIG=ReRoCCManyGemminiConfig. This issue doesn't happen for CONFIG=GemminiRocketConfig, for example.

[suhwanhong999]

@jerryz123 As seen in #2241, I tried to use the latest version possible that would run on my stem (Linux), which was commit seen here (#2166). So, I am currently on b48e992, merged 1/10/25. While the build-setup.sh runs fine, I still run into the same issue originally described. Not sure why this issue persists