Hi,
It looks like the syscall 403 for rv32 has not been implemented in gem5
yet. I believe it is the clock_gettime64 syscall returning the time in
64-bit integer [1]. It looks like this syscall will be preferred over the
32-bit version moving forward.
I think it should be straightforward to imple
Hi all,
It seems to be a bug in gem5.
You can run the gem5 binary without any arguments. The gem5 version is
within the first few lines of the output.
Regards,
Hoa Nguyen
On Sun, Mar 17, 2024, 14:31 yb_zhang--- via gem5-users
wrote:
> I tried it in my environment and got the same error as you
Hi Nick,
You can use the argument like this,
scons build/RISCV/gem5.opt –limit-ld-memory-usage -j`nproc`
I was able to compile gem5 with docker with 6GiB of RAM, and was not able
to do that with 4GiB of RAM.
I think you already hint at a better solution, which is to use build the
docker image o
Hi Nick,
I don't think there would be any problem with compiling gem5 on arm (or
riscv) architecture on Ubuntu. I've been compiling/using gem5 on an arm
system and it works for me.
However, please make sure that the system has enough memory to compile
gem5. Typically compiling gem5 takes a lot of
Hi,
You can find the precise tag name of every gem5 version here,
https://github.com/gem5/gem5/tags
Regards,
Hoa Nguyen
On Sat, Feb 24, 2024, 18:51 hu miao via gem5-users
wrote:
> HI:
> I need to use old version gem5 because I need to use
> configs/example/fs.py. However,after git clone
Hi,
The problem seems to be that the gem5 simulation and the m5term call are
run on different virtual machines.
If you're using docker then please make sure that you're calling m5term on
the same docker container as the one running gem5 simulation.
If that doesn't work then you can check the gue
Hi,
These are a few pointers on how to start modifying the RVV implementation.
- The previous gem5 bootcamp has details on how to implement instructions
in gem5, and examples of implementing RISCV P extension,
https://youtube.com/watch?v=Z5B02jkNpck&list=PL_hVbFs_loVSaSDPr1RJXP5RRFWjBMqq3&index=1
Hi Saras,
This is the methodology we used for most gem5-resources disk images
regarding automating the login process right after booting Ubuntu.
We have a systemd service file that will log you in as a root user.
https://github.com/gem5/gem5-resources/blob/stable/src/arm-ubuntu/disk-image/shared/
Hi Anamika,
Hey there. The se.py script has been depreciated in favor of the new gem5
standard library. You can find examples of using the standard library,
https://github.com/gem5/gem5/tree/stable/configs/example/gem5_library.
The internal of the standard library (how components are constructed
Ah, also, if modifying MAX_HARTS doesn't work then you can follow the
following PRs to use a newer bootloader, which works with 16 cores.
- Building the bootloader https://github.com/gem5/gem5-resources/pull/13
- Changing gem5 to use the new bootloader
https://github.com/gem5/gem5/pull/390#issuecom
Hi Nitish,
The number of CPUs might also be limited by the bootloader as well. You
might want to recompile the bootloader using the following documentation
https://github.com/gem5/gem5-resources/tree/stable/src/riscv-fs#bootloader-bbl
.
You might want to modify this parameter in the proxy kernel
Hi Kevin,
The error is quite odd, and it does look like it was hardware resource
related.
Can you try building gem5 from the official docker images here? The images
have newer versions of gcc and scons I believe. If the problem persists
then it's probably a bug in gem5.
https://www.gem5.org/docu
Hi all,
As mentioned above, the problem is likely due to using KVM then switch to
another gem5 core, which doesn't support all ISA extensions that the host
supports.
I wanted to add that, if updating the gem5 version doesn't solve the
problem, you can use AtomicSimpleCPU for fast-forwarding inste
Hi all,
It appears that you're using an older version of gem5 that does not have
the palignr instructions implemented. The current stable branch has those
instructions implemented.
I'm not sure if this is a problem in your setup or related to this problem,
but I wanted to add that I was not able
Hi,
This patch might help.
https://gem5-review.googlesource.com/c/public/gem5/+/70017
Regards,
Hoa Nguyen
On Sat, May 6, 2023, 00:19 zhangcongwu via gem5-users
wrote:
> Dear gem5 community,
>
> I am running `configs/example/gem5_library/arm-ubuntu-run.py` to simulate
> a multi-core architectu
Hi Zhang,
I wanted to add that you can check the Linux kernel booting log to see
whether the kernel recognized the maximum vector length of SVE correctly.
Note that even if the SVE vlen is correctly recognized by the kernel, it
doesn't mean the whole vlen is available to userspace. Though, I think
Hi Karim,
I'm not very familiar with the Mesh_XY topology, but here is my attempt to
explain this.
I think the layout of the topology looks like this,
2 (0,1) <-> 3 (1,1)
^ ^
| |
v v
0 (0,0) <-> 1 (1,0)
It's straightforward to send a packet from
Hi Alex,
There used to be a beginner friendly gem5 + ARM SVE tutorial from HiPEAC
2021 that could be easily found on the internet. I wasn't able to find that
this time, however.
You can start with the resources here
https://github.com/arm-university/arm-gem5-rsk
Without modifying anything, the s
Hi Alex,
The answer to your question depends on quite a few factors.
One of them is what are the ISAs do you plan to investigate. As far as I
know, x86 in gem5 does not support a lot of SIMD extensions. I believe arm
NEON is pretty well supported in gem5.
In terms of vector length agnostic instr
Hi,
I believe you can compile build/X86/gem5.debug to have all the symbols
available for debugging.
I'm not sure about the error, but generally, when debugging an instruction,
I would try to find the internal name of the instruction in gem5, and look
at gem5/build/X86/arch/x86/arch/generated/ to
Hi,
It seems that it's down to how se.py parses the arguments. Specifically,
the string passed to "-o" will always be splitted with the space character
as the delimiter as shown here,
https://gem5.googlesource.com/public/gem5/+/refs/heads/stable/configs/example/se.py#98
You can change the parsing
Hi,
I wanted to add that, if nested virtualization is not an option, you can
remove the "accelerator": "kvm" line in the json file. Without kvm, the
process might take a few hours.
https://gem5.googlesource.com/public/gem5-resources/+/refs/heads/stable/src/x86-ubuntu/disk-image/x86-ubuntu/x86-ubu
Hi Nikos,
The problem you ran into with the FS mode seems to be the same problem
described here [1] [2].
Can you try downloading the changes and let me know if it fixes the problem?
[1] https://gem5-review.googlesource.com/c/public/gem5/+/65272
[2] https://gem5-review.googlesource.com/c/public/ge
Hi all,
I also ran into the same problem using another benchmark. I want to note
that this problem also appears when using the AtomicCPU.
Regards,
Hoa Nguyen
On Tue, Nov 1, 2022 at 3:02 AM Νικόλαος Ταμπουρατζής via gem5-users <
gem5-users@gem5.org> wrote:
> Dear Boddy,
>
> Thank you for the upd
Hi Chrysanthos,
As indicated by this line,
> ---[ end Kernel panic - not syncing: VFS: Unable to mount root fs on
unknown-block(3,0) ]---
the problem seems to be that the root partition was identified incorrectly.
You can run the `fdisk` command to find the location of the root partition.
For exa
Platform Name : riscv-virtio,qemu
> >>> >> >>>> Platform Features : timer,mfdeleg
> >>> >> >>>> Platform HART Count : 8
> >>> >> >>>> Firmware Base : 0x8000
> >>> >> >>>> Firmware Size : 156 KB
> >>> &
Hi,
It also took 6+ hours to boot Linux using that disk image on my end, so I
think it's systemd being very slow.
By profiling systemd, I find that systemd took a lot of time to bring up
the networking service, and especially the "motd" (message of the day?)
service, which queries for updates fro
Hi Andreas,
My guess is that for Method 1, the pointer is of a virtual address so
there's a page fault there.
I'm not sure why the write() function wasn't invoked on Method 2. I got
into the same problem recently where I used mmap() to write to a physical
address, which should be handled by a Pio
Hi,
It's not clear from the error log where the problem is, but it does
indicate that qemu failed to start. My suggestion is to check the path of
the qemu binary (i.e. this line
https://gem5.googlesource.com/public/gem5-resources/+/refs/heads/develop/src/parsec/disk-image/parsec/parsec.json#44).
I
Hi Fengze,
The command to compile gem5 as a library is,
scons build/X86/libgem5_opt.so
The ISA and the variant (e.g. opt, debug, fast) options are the same as
those of compiling gem5 as an executable.
Regards,
Hoa Nguyen
On Thu, Oct 7, 2021, 3:01 AM Fengze Yu via gem5-users
wrote:
>
> Hi
>
>
Hi Reiko,
It's not immediately clear to me what caused the problem.
Though, I do think it was because the host was running out of memory.
I think you can resolve that by modifying the "vm_memory" parameter of
the disk-image/spec-2017/spec-2017.json file.
Regards,
Hoa Nguyen
On 9/24/21, Reiko Ma
Hi Reiko,
Can you point to the tutorial and the launch experiment script that you
used?
I think it's possible that the
file disk-image/spec2017/spec2017-image/spec2017 didn't exist, or the path
to that file was incorrect.
Regards,
Hoa Nguyen
On Wed, Sep 22, 2021, 12:24 PM Reiko Matsuda-dunn via
Hi,
riscv-tests is currently not well-supported as there are several
outstanding issues mentioned here:
https://gem5.atlassian.net/browse/GEM5-938
Regards,
Hoa Nguyen
On 7/29/21, Boya Chen via gem5-users wrote:
> Hi, all
>
> I use the latest v21.1 gem5 to run RISCV simulation, using the workloa
Hi,
Can you try changing the "--root-device" option in the command to run
gem5 to one of "/dev/vda1", "/dev/sda1", "/dev/hda", "/dev/vda",
"/dev/sda"?
Regards,
Hoa Nguyen
On 7/2/21, Eliot Moss via gem5-users wrote:
> I've ried my best to follow the instruction on how to build a disk image
> usi
Hi Rajesh,
I assume that you're on the develop branch of gem5. You can remove the line
self.intrctrl = IntrControl()
to fix the config file.
Regards,
Hoa Nguyen
On Wed, Jun 9, 2021, 10:26 PM Rajesh Shashi Kumar via gem5-users <
gem5-users@gem5.org> wrote:
> Hi Ayaz,
>
> Thank you very muc
Hi Ange-Thierry,
The logs and the packer script seem fine to me.
The installation process might get stuck somewhere and it doesn't show here.
You can use VNC to inspect the process as detailed here:
https://gem5art.readthedocs.io/en/latest/main-doc/disks.html#ii-inspect-the-building-process
The
gt; terminal, I can clearly see that it is executed.
>>> >
>>> > qemu: build-essential is already the newest version (12.4ubuntu1).
>>> >> qemu: gfortran is already the newest version (4:7.4.0-1ubuntu2.3).
>>> >> qemu: 0 upgraded, 0 newly instal
Hi,
The log seems to indicate that QEMU cannot use KVM on your machine.
To resolve this problem, you can either enable KVM, or choose not to
use KVM for creating the disk image.
To not to use KVM, you'll need to remove the line:
"accelerator": "kvm"
in this file:
https://gem5.googlesource.com
mage only works with X86.
>
> Regards,
> Hoa Nguyen
>
> On 5/20/21, Liyichao wrote:
>> Hi Hoa:
>>Is the spec-2017 img just for X86?
>>
>> Does it support for AARCH64?Does it support for running with KVM+O3?
>>
>>
>> -邮件
>> Hi Hoa:
>> Is the spec-2017 img just for X86?
>>
>> Does it support for AARCH64?Does it support for running with KVM+O3?
>>
>>
>> -邮件原件-
>> 发件人: Hoa Nguyen via gem5-users [mailto:gem5-users@gem5.org]
>> 发送时间: 20
Hi Liyichao,
Currently, the spec-2017 image only works with X86.
Regards,
Hoa Nguyen
On 5/20/21, Liyichao wrote:
> Hi Hoa:
> Is the spec-2017 img just for X86?
>
> Does it support for AARCH64?Does it support for running with KVM+O3?
>
>
> -----邮件原件-----
>
> seeing the exact same files as if I'm mounting the SPEC2017 iso.
>
> Thanks
>
> On Wed, 19 May 2021 at 07:09, Hoa Nguyen via gem5-users <
> gem5-users@gem5.org> wrote:
>
>> Hi Victor,
>>
>> I'm not sure what caused the errors of building
Hi Victor,
I'm not sure what caused the errors of building the spec benchmarks.
Also, I'm not sure how to prevent the vm from being closed after the
builds fail. However, there are a few steps can be done to exterminate
the disk image after the builds fail:
- In the file src/spec-2017/disk-image/
Hi,
Can you be more specific about the command line and the gem5 version
that you used?
>From the screenshot, it seems to be a problem with LTO. You can
compile gem5 with --no-lto flag to not to use LTO for compiling.
Regards,
Hoa Nguyen
On 4/20/21, VAIDYA ROHINI VILAS wrote:
> Hello,
> I am t
Hi,
This discussion
(https://www.mail-archive.com/gem5-users@gem5.org/msg19294.html) might
have the information to make compiling process faster.
Regards,
Hoa Nguyen
On 4/19/21, VAIDYA ROHINI VILAS via gem5-users wrote:
> Hi,
> No error message is there.
> process just stops at [ LINK ] ->
Hi,
The message is not an error. It means that if you press enter, the git
commit hook will be installed and then gem5 will be compiled; while
ctrl+c will exit the compiling program.
Regards,
Hoa Nguyen
On 4/19/21, VAIDYA ROHINI VILAS wrote:
> Hello ,
> I am trying to download gem5 in my new sy
Hi,
Thank you for reporting this bug. The links have been updated!
Regards,
Hoa Nguyen
On 4/6/21, Ahmad SB via gem5-users wrote:
> Hi
> Links on https://www.gem5.org/documentation/learning_gem5 are not valid
> anymore (e.g.
> gem5.org/dist/current/gem5/cpu_tests/benchmarks/bin/arm/Bubblesort )
Hi,
final_tick and sim_insts are RootStats (they are in src/sim/root.cc)
and they won't be reset by calling reset stats.
Regards,
Hoa Nguyen
On 4/11/21, kong han via gem5-users wrote:
> Hi all,
> Now I using the KVM CPU to run fs mode, and I try two ways to reset the
> m5out/stats.txt but all c
Hi,
Thanks for reporting this bug. It is indeed a python3 compatibility
issue, which is is addressed here
https://gem5-review.googlesource.com/c/public/gem5/+/7 and will be
merged to the develop branch soon.
Regards,
Hoa Nguyen
On 4/12/21, Gabe Black via gem5-users wrote:
> Maybe a python 2
Hi all,
This pull request (https://github.com/takekoputa/gem5/pull/2/files)
contains an example of how to add DRAMsim3 to the
configs/common/MemConfigs.py file. Please replace the hardcoded path
to an .ini with a path in your system.
There were mem_ranges having overlapping ranges because
create_
Hi,
Can you be elaborate more about the issue? Such as what is the command line
and what binaries are you using to run fs.py.
It is possible that the guest program is waiting for some input from
users/stdin. Also, you can run fs.py with faster cpu models to verify
whether the guest program got st
Hi Duc Anh,
I'm not sure the setup laid out in the tutorial would work for arm
system as we only tested it on x86 arch. However, I believe there are
forks of packer that would work with arm arch
(https://www.packer.io/community-tools), and there are also tutorials
for using packer to build disk im
roperly when using the fs.py config script
> but i don't get a terminal to interact with the simulated system using
> telnet. Is that expected because i need to use the --script parameter or
> am I missing something?
>
> Greetings,
> Miguel antonio Avargues Gutiérrez.
>
>
Hi Miguel,
For X86 full system simulation, specifically for creating custom disk
image that runs with newer Linux kernels, I believe gem5-resources has
the resources for that. More details can be found here,
- General introduction and artifacts:
https://www.gem5.org/documentation/general_docs/gem5
Hi,
There is an whitespace between `gem5` and `2` in the path to the disk
image, so the parser thinks that
`2/full_system_images/aarch64-ubuntu-trusty-headless.img` is argument.
Regards,
Hoa Nguyen
On 2/23/21, VAIDYA ROHINI VILAS via gem5-users wrote:
> Hello,
> I am trying to run gem5 in fs mo
Hi Krishnan,
I'm not sure if this would address the problem, but those are some
pointers to adding new stats.
The main data structure holding Stats object in gem5 is Stats::Group
implemented here
https://gem5.googlesource.com/public/gem5/+/e0441fa7a7c858d7e472fbb1e03089a9a11b8462/src/base/stats/g
Hi,
I believe this is an error produced by a Python script. If the
configs/example/fs.py file has been modified, please check the
indentation of the modified part of the script for indentation
inconsistencies (for example, space indentations vs tab indentations).
Regards,
Hoa Nguyen
On 1/19/21,
Hi,
Can you be a bit more specific about the error that you saw and the
version of gem5 that you're using?
I tried something similar and it worked for me,
build/ARM/gem5.opt configs/example/se.py --stats-root
'system.cpu[:].mmu.itb' --caches -c
tests/test-progs/hello/bin/arm/linux/hello --ruby -n
HI Duc Anh,
It seems that you're using the default Linux kernel config, which
won't work with gem5. There are a few Linux kernel configs of older
versions that are known to work with gem5, and we can use them to
generate new ones that might work with gem5. You can find the known
working ones here
I'm glad it helped :D
Regards,
Hoa Nguyen
On Mon, Nov 23, 2020, 12:22 AM sujayyadalam--- via gem5-users <
gem5-users@gem5.org> wrote:
> So I was able to solve this by creating a new config from scratch.
>
> Previously, I was using a modified version of the fs.py.
> Now, I used the run_npb.py in
Hi Sujay,
I'm not sure about the problem with the gem5 config.
You might want to take a look at gem5 resources
(https://gem5.googlesource.com/public/gem5-resources/).
The gem5 resources repo has the resources (including gem5 configs) for
running a workload in FS mode by booting with KVM and swit
Hi Tracy,
I think you can put scheduleInstStop() and m5.simulate() in a loop to do that.
Regards,
Hoa Nguyen
On 11/10/20, Tracy Mac via gem5-users wrote:
> Hi All!
> I am trying to limit instructions number repeatedly,stop the simulation
> when the instruction is limited, do other operations, a
Hi,
I assume that you want to keep track of a piece of data at a specific
virtual/physical address. I don't know a definite answer to this, but
one thing you can do is to use the gem5 logging system (i.e. DPRINTF
and the like) and figure out what flags you should enable to keep
track of transactio
Hi Duc Anh,
Can you please confirm that this change,
https://gem5-review.googlesource.com/c/public/gem5/+/36296, fixes the
stats error?
Regards,
Hoa Nguyen
On 11/12/20, Hoa Nguyen wrote:
> Hi Duc Anh,
>
> I think the stats error is related to this change,
> https://gem5-review.googlesource.com/
Hi Duc Anh,
To answer the question about simulation speed, I believe the
difference between SE mode and FS mode is minimal. In addition to
that, kernel interactions, like process scheduler and page table
walks, have significant impact on system performance, so in most
cases, I believe it is more a
Hi Duc Anh,
I think the stats error is related to this change,
https://gem5-review.googlesource.com/c/public/gem5/+/36296.
The stats bug should be resolved when you cherry-pick that change.
Hopefully we can add the SMT feature and fix the multicore DerivO3 bug by
the next release.
Regarding the
Hi Siqing,
You'll need to call "m5 readfile" in the simulated system (i.e., the
guest OS must call "m5 readfile" at some point), and the command will
output the content of the script that you pass from the host.
gem5 resources has some example of calling "m5 readfile" in the guest OS.
For exampl
Hi Duc Anh,
I assume you want to enable SMT for X86. SMT for X86 has yet to be
supported, but there is progress on this recently. You can keep track
of the development via JIRA ticket
(https://gem5.atlassian.net/browse/GEM5-332) and via this relation
change (https://gem5-review.googlesource.com/c/
Hi,
It should be `/sbin/m5 exit` instead of `/sbin/m5`.
Regards,
Hoa Nguyen
On 11/7/20, Hoa Nguyen wrote:
> Hi,
>
> One way to do it is to use the `--debug-start TICK` option, which will
> start logging after `TICK` ticks.
>
> To figure out the number of simulated ticks, you can run `/sbin/m5`
Hi,
One way to do it is to use the `--debug-start TICK` option, which will
start logging after `TICK` ticks.
To figure out the number of simulated ticks, you can run `/sbin/m5` in
the guest right before the ROI, then use `m5.curTick()` in the gem5
system config. The output would be useful if the
Hi,
Can you check whether the "cannot allocate memory" occured in the host
system or the simulated system?
If it occurs on the simulated system then one thing you can do is to
increase the memory size in the gem5 system configuration.
Regards,
Hoa Nguyen
On Tue, Nov 3, 2020, 10:20 PM Hasan, S M
Hi Siqing,
You can find the PARSEC ROI example mentioned in the webpage is here,
https://github.com/darchr/parsec-benchmark/commits/gem5-20-annotations
gem5-resources (https://gem5.googlesource.com/public/gem5-resources/)
has the documentation of setting up full system gem5 experiments with
sever
Hi Siqing,
It should be "x86" with a lowercase x rather than "X86".
Regards,
Hoa Nguyen
On Tue, Nov 3, 2020, 12:52 AM -17 via gem5-users
wrote:
> Hi all:
>
> About a few months ago, I followed the guidelines of this link to
> successfully build M5ops:
>
> https://www.gem5.org/documentation
Hi,
Though I'm not sure how to add the cache clearing instruction specifically,
those are a few examples that new instructions are added,
https://gem5-review.googlesource.com/c/public/gem5/+/26123
https://gem5-review.googlesource.com/c/public/gem5/+/23262
https://gem5-review.googlesource.com/c/
> On Sun, Oct 25, 2020 at 9:31 PM Hoa Nguyen via gem5-users <
> gem5-users@gem5.org> wrote:
>
>> Hi Wenqi,
>>
>> We have some documentation about the new m5 utility here:
>> https://www.gem5.org/documentation/general_docs/m5ops/
>>
>> The following link i
Hi Wenqi,
We have some documentation about the new m5 utility here:
https://www.gem5.org/documentation/general_docs/m5ops/
The following link is an example of annotating PARSEC:
https://github.com/darchr/parsec-benchmark/commits/gem5-20-annotations
Regards,
Hoa Nguyen
On 10/25/20, wqyin--- via
Hi Tracy,
The example of PARSEC benchmark mentioned in the documentation page
can be found here,
https://github.com/darchr/parsec-benchmark/commits/gem5-20-annotations
Regards,
Hoa Nguyen
On 10/13/20, Tracy Mac via gem5-users wrote:
> Hi ALL!
>
> I try to run spec2006 in gem5 full system mode
Hi Jinpeng,
The error seems to cause by 'sysinfo', which is called by SPEC 2017
before it starts running a workload. SPEC 2017 documents how to turn
it off here: https://www.spec.org/cpu2017/Docs/utility.html#sysinfo.
Basically, the fix is to add this line:
sysinfo_program =
to the SPEC config
78 matches
Mail list logo