OpenSBI doesn't appear to be starting - no Output, no ASCII art of OpenSBI starting

[PietDelaney]

I got Wally Booting further now by uisng the 64GB SD card that Rose suggested instead of the 32GB code I had been using.
Now it's jumping info OpenSBI but isn't printing anything.

█▀█ █▀█ █▀█ █▀▀ █ █
█ █ █ █▄▀ █▄▄ ▄▄▄ █ █
█▄█ █▄█ █ █ █▄▄ ▀▄▀

---

\ \ / / / \ | | | | \ \ / /
\ \ __ / / / \ | | | | \ / /
\ / / / / /\ \ | | | | \ /
\ / / ____ \ | |___ | |___ | |
_/_/ // _|||| |__|

[0.05343] Initializing SPI Controller.
[0.05703] Initializing SD Card in SPI mode.
[0.06125] CMD0: Success, r = 0x01
[0.06436] CMD8: Success, 0x01000001AA
[0.06797] ACMD41: Success, r = 0x00
[0.30233] New clock frequency: 5000000
[0.30612] SD card is initialized.
[0.30928] Getting GPT information.
Blocks loaded: 1/1
[0.32199] Getting partition entries.
Blocks loaded: 1/1
[0.33485] Loading device tree at: 0x8F000000
Blocks loaded: 5/5
[0.35224] Loading OpenSBI at: 0x80000000
Blocks loaded: 271/271
[0.63531] Loading Linux Kernel at: 0x80200000
Blocks loaded: 161096/161096
[161.53362] Done! Flashing LEDs and jumping to OpenSBI...

The book shows it's suppose to print the ASC art OpenSBI:

OpenSBI v1.3___ _____ ____ / __ / | _ _ || | | | __ ___ _ __ | ( | |) || || | | | ' / _ ' \ | _ < | || || | |) | __/ | | |) | |) || |_/| ./ _|| ||/|/___|| |||Platform Name : wally-virt,qemuPlatform Features : medelegPlatform HART Count : 1Platform IPI Device : aclint-mswiPlatform Timer Device : aclint-mtimer @ 20000000HzPlatform Console Device : uart8250Platform HSM Device : ---Platform PMU Device : ---Platform Reboot Device : ---Platform Shutdown Device : ---Platform Suspend Dev

The OPenSBI docs show the same:

_bootloader version:210209-4547a8d
ddr 0x00000000, 1M test
ddr 0x00100000, 2M test
DDR clk 2133M,Version: 210302-5aea32f
0 crc flash: 7ebedaa2, crc ddr: 7ebedaa2
crc check PASSED

                                                    bootloader.

OpenSBI v0.9

---

/ __ \ / | _ _ |
| | | | __ ___ _ __ | ( | |) || |
| | | | '_ \ / _ \ '_ \ ___ | _ < | |
| || | |) | __/ | | |) | |) || |
_/| ./ _|| ||/|____/|
| |
|_|

Platform Name : StarFive VIC7100
Platform Features : timer,mfdeleg
Platform HART Count : 2
Firmware Base : 0x80000000
Firmware Size : 92 KB
Runtime SBI Version : 0.2_

[rosethompson]

I'm glad the different flash card fixed the CMD8 issue. I will see if I can get the same "problem" flash card and reproduce the issue you were having. We should be able to work with any standard flash card.

It is very strange you got part of the OpenSBI boot messages. The device parameters don't look right. For example the platform name should be @ 20000000Hz.

Just to double check the flash card is built correctly did you use the arty a7 dtb and not the dts?

The command to flash the card should look something like this.
./flash-sd.sh -b /opt/riscv/buildroot -d /opt/riscv/buildroot/output/images/wally-artya7.dtb /dev/sdb

[PietDelaney]

Hi Rose:

Here are the SD card adapters I used:
https://www.amazon.com/dp/B074N3GBYQ?ref=ppx_yo2ov_dt_b_fed_asin_title

Here are the 32GB SD cards I used:
https://www.amazon.com/dp/B0D5HD9R18?ref=ppx_yo2ov_dt_b_fed_asin_title

I was thinking we need to add a reference in the sd card driver that documents the meaning of the bits in the register
used int the CMD8 command. I think I may have see a driver but didn't see the CMD8 in that driver.

Re:
It is very strange you got part of the OpenSBI boot messages. The device parameters don't look right.
For example the platform name should be @ 20000000Hz.

The 20000000Hz is from documentation. OPenSBI didn't print anything on my console. :(

I'll try again flatining the dts to a dtb and rerunning the:
./flash-sd.sh -b /opt/riscv/buildroot -d /opt/riscv/buildroot/output/images/wally-artya7.dtb /dev/sdb

Last time I used:

./flash-sd.sh -d /home/piet/riscv/cvw/linux/devicetree/wally-artya7.dtb /dev/sdf

-piet

[rosethompson]

Hi Piet,

I just noticed at the start of your openSBI banner the version number is v0.9 which is rather old. Mine is currently on v1.3. I think this might be part of the problem. Can you verify you're using a recent git commit and rebuild the buildroot? You might consider making a backup of your exiting /opt/riscv/buildroot or just have the makefile only compile buildroot rather than copy to /opt/riscv/buildroot.

To rebuild buildroot
cd linux
make

To build without installing to /opt/riscv
cd linux
make clean download Image disassemble

Then reflash the card.

[PietDelaney]

OK I pulled 96af5d1 git commit.

[PietDelaney]

I just tried building wally from the git commit
96af5d1

but I got errors on my laptop. Are you sure this is a stable snapshot?

INFO | Running Build for Reference
INFO | Selecting Tests.
INFO | Running Tests on DUT.

ERROR | cc1: error: '-march=rv32ezbkc': unsupported ISA subset 'z'
terminate called after throwing an instance of 'std::runtime_error'
what(): could not open my.elf; searched paths:
. (current directory)
/opt/riscv/riscv64-unknown-elf/bin/ (based on configured --prefix and --with-target)
make[3]: *** [/media/verilog/885393a4-7d00-4efa-bc8b-eb28261181cd/tools/riscv/cvw/tests/riscof/riscof_work/arch32e/Makefile.DUT-spike:9: TARGET1] Error 255
make[3]: *** Waiting for unfinished jobs....
cc1: error: '-march=rv32ezbkc': unsupported ISA subset 'z'
terminate called after throwing an instance of 'std::runtime_error'
what(): could not open my.elf; searched paths:
. (current directory)
/opt/riscv/riscv64-unknown-elf/bin/ (based on configured --prefix and --with-target)
make[3]: *** [/media/verilog/885393a4-7d00-4efa-bc8b-eb28261181cd/tools/riscv/cvw/tests/riscof/riscof_work/arch32e/Makefile.DUT-spike:5: TARGET0] Error 255
INFO | Running Tests on Reference Model.
ERROR | cc1: error: '-march=rv32ezbkc': unsupported ISA subset 'z'
riscv64-unknown-elf-objdump: 'ref.elf': No such file
make[3]: *** [/media/verilog/885393a4-7d00-4efa-bc8b-eb28261181cd/tools/riscv/cvw/tests/riscof/riscof_work/arch32e/Makefile.Reference-sail_c_simulator:9: TARGET1] Error 1
make[3]: *** Waiting for unfinished jobs....
cc1: error: '-march=rv32ezbkc': unsupported ISA subset 'z'
riscv64-unknown-elf-objdump: 'ref.elf': No such file
make[3]: *** [/media/verilog/885393a4-7d00-4efa-bc8b-eb28261181cd/tools/riscv/cvw/tests/riscof/riscof_work/arch32e/Makefile.Reference-sail_c_simulator:5: TARGET0] Error 1
INFO | Initiating signature checking.
ERROR | Signature file : /media/verilog/885393a4-7d00-4efa-bc8b-eb28261181cd/tools/riscv/cvw/tests/riscof/riscof_work/arch32e/rv32e_m/B/src/clmul-01.S/dut/DUT-spike.signature does not exist
make[2]: *** [Makefile:27: arch32e] Error 1
rm config32e.ini
make[1]: *** [Makefile:16: riscoftests] Error 2
make: *** [Makefile:15: riscof] Error 2
(cvw) verilog@verilog-X555LAB:/tools/riscv/cvw$

[PietDelaney]

Make.log

Adding submodules to try to fix...

[jordancarlin]

I just tested with that commit checked out and didn't have any issues building the tests. I'd assume one of your submodules is out of date (try git submodule update). Only other thing I can think of is what version of the toolchain do you have installed? What is the output of riscv64-unknown-elf-gcc --version? It should be 14.2.0.

[PietDelaney]

$ riscv64-unknown-elf-gcc --version
riscv64-unknown-elf-gcc () 14.2.0
Copyright (C) 2024 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

[PietDelaney]

I rebuilt Buildroot, it pulled the latest bits from the git repo, and I'm now on the buildroot 2024.11.x,
I flattened the dts --> dtb and flashed the sdcard. The results appeared the same.

I just pulled the latest Wally bits and sub-repos:

git pull upstream main
git submodule update

I did a

make clean

and doing a fresh build of Wally.

The V0.9 was from a web document on OpenSBI, not my console.

[jordancarlin]

I hate to undo any work you've done, but we're aware of an issue with booting the newer version (2024.11) on the FPGA. For now I would recommend using the older buildroot configuration until we can get that sorted out. And just to confirm, you are building buildroot using the makefile in the cvw/linux directory?

[PietDelaney]

Are the errors in RED normal? I suspect they are.

[jordancarlin]

Yes. That is a known issue upstream with riscv-arch-test. It was just resolved and the changes will be incorporated into wally with PR #1238.

[PietDelaney]

ROSE: After building Wally I tried simulation of buildroot but wsim fails not finding vsim:

(cvw) piet@piet-Precision-7920-Tower:~/riscv/cvw$ wsim buildroot buildroot --args +INSTR_LIMIT=600000000 --lockstep
Config=buildroot tests=buildroot sim=questa gui=False args='+INSTR_LIMIT=600000000' params='' define=''

Running Questa with command: cd $WALLY/sim/questa; MTI_VCO_MODE=64 IMPERAS_TOOLS=/home/piet/riscv/cvw/config/deriv/buildroot/imperas.ic vsim -c -do "do wally.do buildroot buildroot testbench --args "+INSTR_LIMIT=600000000" --define "+define+USE_IMPERAS_DV" --lockstep"
sh: 1: vsim: not found
(cvw) piet@piet-Precision-7920-Tower:~/riscv/cvw$

[jordancarlin]

The default for wsim is to use questa. If you add the --sim verilator flag you should be good. Verilator does not support lockstep though, so make sure to remove that flag.

[PietDelaney]

Thanks, Jordan, adding --sim verilator got it starting the simulation of buildroot stuff. Ill see if it finishes later...

fpga is stll not working. I often end up with it asking where the bitstream is.

OpenSBI is printing on the simulation (But not on fpga):

OpenSBI v1.6

---

/ __ \ / | _ _ |
| | | | __ ___ _ __ | ( | |) || |
| | | | '_ \ / _ \ '_ \ ___ | _ < | |
| || | |) | __/ | | |) | |) || |
_/| ./ _|| ||/|____/|
| |
|_|

Reached 1600000 instructions
Reached 1700000 instructions
Reached 1800000 instructions
Reached 1900000 instructions
.
.
.
Reached 18100000 instructions
Reached 18100000 instructions
Reached 18100000 instructions
Reached 18100000 instructions
Reached 18100000 instructions
Reached 18100000 instructions
Reached 18100000 instructions
Reached 18100000 instructions
.
.
.

[jordancarlin]

Glad it is working in simulation. Note that OpenSBI 1.6 is currently only working in simulation and currently fails on the FPGA. You will need to downgrade back to the previous version for now if you want it to work on the FPGA. @JacobPease is working on debugging this.

Seeing the specific error messages that you're getting with the FPGA would be helpful for debugging.

[stineje]

Pete, Just curious - did you set up site-setup.sh with your location of Questa. Questa is the best of the simulators to use as we originally use this tool. However, we did set up access for Verilator and Synopsys VCS. Once you install the tool chain, you should modify site-setup.sh or site-setup.csh, depending on your SHELL, to point to your tool base. You may have done this already, but just a friendly reminder, if you did not. Also, Jordan gave the command to change the simulator for wsim at the command-line too. Take care. All my best, James

…

On Jan 19, 2025, at 1:06 AM, Pete Delaney ***@***.***> wrote: ROSE: After building Wally I tried simulation of buildroot but wsim fails not finding vsim: (cvw) ***@***.***:~/riscv/cvw$ wsim buildroot buildroot --args +INSTR_LIMIT=600000000 --lockstep Config=buildroot tests=buildroot sim=questa gui=False args='+INSTR_LIMIT=600000000' params='' define='' Running Questa with command: cd $WALLY/sim/questa; MTI_VCO_MODE=64 IMPERAS_TOOLS=/home/piet/riscv/cvw/config/deriv/buildroot/imperas.ic vsim -c -do "do wally.do buildroot buildroot testbench --args "+INSTR_LIMIT=600000000" --define "+define+USE_IMPERAS_DV" --lockstep" sh: 1: vsim: not found (cvw) ***@***.***:~/riscv/cvw$ — Reply to this email directly, view it on GitHub <https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fopenhwgroup%2Fcvw%2Fdiscussions%2F1233%23discussioncomment-11879901&data=05%7C02%7Cjames.stine%40okstate.edu%7C8887213120eb439f493708dd3857d61d%7C2a69c91de8494e34a230cdf8b27e1964%7C0%7C0%7C638728672110364528%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=eOLib%2FjZF1BmiUDBYnkSnJfgi1sys6UNmQ4mEPEyXAQ%3D&reserved=0>, or unsubscribe <https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAFSIHPCFO4MDGEBPHSPZVKD2LNFIJAVCNFSM6AAAAABVIQ7QBSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCOBXHE4TAMI&data=05%7C02%7Cjames.stine%40okstate.edu%7C8887213120eb439f493708dd3857d61d%7C2a69c91de8494e34a230cdf8b27e1964%7C0%7C0%7C638728672110388080%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=9UrCnIHK9HgwPHjlVlvMfUpwiUfcOU2KWWXK1qTL65M%3D&reserved=0>. You are receiving this because you are subscribed to this thread.

[PietDelaney]

ROSE: How to pull the WALLY environment with a working FPGA build. Your using openSBI ver 1.3 and the wally scripts for copying the sdcard are changing. It would be conventional to have a git TAG for stable releases as the main branch is being used for active development. IE: more like the way the Linux and Android AOSP git repos are maintained.

-pete

[stineje]

Pete, You can always re-install the tools with Jordan’s awesome script that will install only what is needed. Take care. All my best, James

…

On Jan 19, 2025, at 4:49 PM, Jordan Carlin ***@***.***> wrote: I just tested with that commit checked out and didn't have any issues building the tests. I'd assume one of your submodules is out of date (try git submodule update). Only other thing I can think of is what version of the toolchain do you have installed? What is the output of riscv64-unknown-elf-gcc --version? It should be 14.2.0. — Reply to this email directly, view it on GitHub <https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fopenhwgroup%2Fcvw%2Fdiscussions%2F1233%23discussioncomment-11884558&data=05%7C02%7Cjames.stine%40okstate.edu%7C32e79d88ea5345ff64b608dd38db94ae%7C2a69c91de8494e34a230cdf8b27e1964%7C0%7C0%7C638729237939525126%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=S9GyZN700Ta1upHnAUvqM8GSnpOUJlmWKLI6fY701AA%3D&reserved=0>, or unsubscribe <https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAFSIHPAA5JUQG3KUVVICAJT2LQTYZAVCNFSM6AAAAABVIQ7QBSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCOBYGQ2TKOA&data=05%7C02%7Cjames.stine%40okstate.edu%7C32e79d88ea5345ff64b608dd38db94ae%7C2a69c91de8494e34a230cdf8b27e1964%7C0%7C0%7C638729237939545465%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=8y4AavIYf2TYZ%2F0wTG0s5u27W8sdgbxyKz35YhA5T04%3D&reserved=0>. You are receiving this because you commented.

[davidharrishmc]

This is turning into a huge effort and maybe not a great use of everyone’s time. Jordan, I know you are transitioning buildroot versions. Is there a cvw commit that Pete could check out that just works for us, or should he wait until things stabilize and then check out a known working version from the head of the main branch? Pete, would it be possible for you to install exactly the same Ubuntu that we are using? Even if it has to be on a virtual machine? Then could you check out and reproduce exactly what works for Rose and Jordan? Once you can reproduce their work, you can change one variable at a time until it works in your preferred distribution. David On Sun, Jan 19, 2025 at 3:17 PM James E. Stine ***@***.***> wrote:

…

Pete, You can always re-install the tools with Jordan’s awesome script that will install only what is needed. Take care. All my best, James > On Jan 19, 2025, at 4:49 PM, Jordan Carlin ***@***.***> wrote: > > > I just tested with that commit checked out and didn't have any issues building the tests. I'd assume one of your submodules is out of date (try git submodule update). Only other thing I can think of is what version of the toolchain do you have installed? What is the output of riscv64-unknown-elf-gcc --version? It should be 14.2.0. > > — > Reply to this email directly, view it on GitHub < https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fopenhwgroup%2Fcvw%2Fdiscussions%2F1233%23discussioncomment-11884558&data=05%7C02%7Cjames.stine%40okstate.edu%7C32e79d88ea5345ff64b608dd38db94ae%7C2a69c91de8494e34a230cdf8b27e1964%7C0%7C0%7C638729237939525126%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=S9GyZN700Ta1upHnAUvqM8GSnpOUJlmWKLI6fY701AA%3D&reserved=0>, or unsubscribe < https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAFSIHPAA5JUQG3KUVVICAJT2LQTYZAVCNFSM6AAAAABVIQ7QBSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCOBYGQ2TKOA&data=05%7C02%7Cjames.stine%40okstate.edu%7C32e79d88ea5345ff64b608dd38db94ae%7C2a69c91de8494e34a230cdf8b27e1964%7C0%7C0%7C638729237939545465%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=8y4AavIYf2TYZ%2F0wTG0s5u27W8sdgbxyKz35YhA5T04%3D&reserved=0>. > You are receiving this because you commented. > — Reply to this email directly, view it on GitHub <#1233 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AR4AA33YJ3LGY667WE6DDKD2LQW7HAVCNFSM6AAAAABVIQ7QBSVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCOBYGQ3DMMA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[PietDelaney]

David: Sure I'll switch from my Linux mint workstation to my Ubuntu 24.04.1 workstation.
I'm building Wally using the default build tree on Ubuntu at main branch for now. It was last updated on Jan 17.

[rosethompson]

I've been busy the last few weeks so this fell off my radar. Were you able to get past the scrambled OpenSBI boot?