resources: update gapbs to use new 24.04 images

[Harshil2107]

No description provided.

[erin-le]

Add link

[erin-le]

Add link

[Harshil2107]

If everything looks good then I will upload the disks and add the links.

[erin-le]

Elaborate more on exit events in the base disk image and/or add link to base image documentation.

I think you should also say "Look at the Boot Sequences section for information on what exit events will take place", then tell users where they can find configuration scripts with that exit event setup. Since the PR to update the config scripts is still pending, maybe you could point users to the "usage" tab of one of the gem5 resources with the fleshed out exit event handler setup (such as this one: https://resources.gem5.org/resources/x86-ubuntu-24.04-img/usage?database=gem5-resources&version=2.0.0). Or even just put it in a code block somewhere in this document.

[mbabaie]

I agree with more explanations about two things:

1. what the path to the programs' executable is in the disk image. This is probably required to setup the command in the run script. I saw in the examples Erin provided below, she has not specified that, and has used them directly as ./cc for this purpose.
2. number of exit events, their orders, and where they are in the main program.

[Harshil2107]

1. I have added the path to the executable, I will also add the links to the workloads on gem5 resources when the changes looks good, and the disks have been uploads. There will be a workload for each executable benchmark in 3 sizes and with two real graphs. Is there any other information that I should add?

2. The boot sequences section in the readme explains, where the exits will occur. will it be beneficial to mention exact files?

[mbabaie]

1. Thanks for adding this. Looks good to me.
2. That also looks good with the stars marked to see how many exit events must be handled in the runs scripts.

[erin-le]

It could be good to add an overview of the benchmarks available on the built disk, or point toward the GAPBS documentation

[Harshil2107]

I will add a section with all the benchmark workloads, when we upload the disk to the cloud and I can push the JSONs to MongoDB.

[erin-le]

Add copyright (?)

[hnpl]

This line has a problem when the .gz file already exists. Can you do one of the following?

1. If the .gz file already exists, wget will download it to .gz.1 or so. Before downloading the .gz file, make sure you remove the old download of .gz file.
2. If the .gz file already exists, checksum the file. If the checksum does not match, remove the old .gz file and re-download the .gz file.

[hnpl]

This also applies to similar wget commands below.

[erin-le]

I think it would also be good to add examples of how to set the workload directly into the README. It was something that I struggled with when testing your other disk image refactors. Examples for Arm, RISCV, and X86 are included below:

# for RISCV
board.set_kernel_disk_workload(
    kernel=obtain_resource(
        "riscv-linux-6.6.33-kernel", resource_version="1.0.0"
    ),
    disk_image=DiskImageResource("/path/to/gem5-resources/src/gapbs/riscv-disk-image-ubuntu-24-04/riscv-ubuntu", root_partition="1"),
    bootloader=obtain_resource("riscv-bootloader-opensbi-1.3.1"), 
    readfile_contents="./cc -g 21\n",
)

# for x86
board.set_kernel_disk_workload(
    kernel=obtain_resource(
        "x86-linux-kernel-5.4.0-105-generic", resource_version="1.0.0"
    ),
    disk_image=DiskImageResource("/path/to/gem5-resources/src/gapbs/disk-image-ubuntu-24-04/x86-ubuntu-24-04-gapbs"),
    readfile_contents="./cc -g 23\n",
    kernel_args=[
        "earlyprintk=ttyS0",
        "console=ttyS0",
        "lpj=7999923",
        "root=/dev/sda2",
        "no_systemd=true"
    ]

)

# for arm
disk_img = DiskImageResource("/path/to/gem5-resources/src/gapbs/disk-image-arm-npb/arm-ubuntu", root_partition="2")
board.set_kernel_disk_workload(
        disk_image=disk_img, 
        bootloader=obtain_resource("arm64-bootloader-foundation"), 
        kernel=obtain_resource("arm64-linux-kernel-5.15.36"),
        readfile_contents="./cc -g 21\n",
        )

[mbabaie]

Related to the examples Erin has provided, I think it's good to have a full example of readfile_contents that has everything in it (whether it logs in with sudo, the full path to the binary, etc.)

[Harshil2107]

Related to the examples Erin has provided, I think it's good to have a full example of readfile_contents that has everything in it (whether it logs in with sudo, the full path to the binary, etc.)

I am thinking of adding a base example scripts to run the gapbs workloads like we have right now for x86 but for all the ISAs and then point them in the readme. I think those should be good examples? is there anything else specific that I should add?

[mbabaie]

That would be very good and helpful.
Another thing to add is to make sure we remind the users it should be run with sudo in the examples.
I don't have anything else to add at the moment :)

[mbabaie]

This is what I meant (figured it out with trial&error) that worked perfectly:
"echo 12345 | sudo -S ./gapbs/bfs -g 22"