[gem5-users] Re: KVM patch for FS mode and PARSEC on gem5-21

[Hoa Nguyen via gem5-users]

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 much. That was indeed the issue. I also noticed that the
> following was missing in configs/system/system.py. After adding this line,
> I'm able to get FS,PARSEC simulations running on gem5-v21
>
>*self.intrctrl = IntrControl()*
>
> Error message:
> ```
> Error in unproxying param 'intrctrl' of system.pc
> Traceback (most recent call last):
>   File "", line 1, in 
>   File "build/X86/python/m5/main.py", line 455, in main
>   File "/home/rajesh/work/parsec-tests/configs/run_parsec.py", line 102,
> in 
> m5.instantiate()
>   File "build/X86/python/m5/simulate.py", line 88, in instantiate
>   File "build/X86/python/m5/SimObject.py", line 1599, in unproxyParams
>   File "build/X86/python/m5/proxy.py", line 134, in unproxy
> AttributeError: Can't resolve proxy 'any' of type 'IntrControl' from
> 'system.pc'
> ```
>
> Thank you for your time.
>
> -- Rajesh Shashi Kumar
>
>
> On Thu, Jun 10, 2021 at 3:20 AM Ayaz Akram  wrote:
>
>> Hi Rajesh,
>>
>> I think the error you are seeing is because tlbs are exposed through an
>> mmu unit in gem5 now. I guess changing line 117 in caches.py as in this
>> file (
>> https://gem5.googlesource.com/public/gem5-resources/+/refs/heads/stable/src/boot-exit/configs/system/caches.py)
>> should work for you.
>>
>> -Ayaz
>>
>> On Wed, Jun 9, 2021 at 2:04 PM Rajesh Shashi Kumar via gem5-users <
>> gem5-users@gem5.org> wrote:
>>
>>> Thank you for the quick response.
>>>
>>>
>>>- Thank you for confirming that the KVM fix has been integrated into
>>>the stable branch (pointing to gem5-v21.0). At this time, I have not
>>>managed to get PARSEC working in FS mode on the stable branch (v21.0) 
>>> and I
>>>am debugging the issue described below
>>>- For context, I encountered the following error on gem5-v21.0 while
>>>running PARSEC which made me explore reverting back to v20.1 (and the KVM
>>>patch) as described in the gem5art documentation. I'm using the latest
>>>PARSEC config run scripts from the gem5-resources repository
>>>
>>> 
>>>
>>> gem5 version 21.0.0.0
>>> gem5 compiled Jun 10 2021 01:34:21
>>> gem5 started Jun 10 2021 01:57:17
>>> gem5 executing on lenovo, pid 20958
>>> command line: gem5/build/X86/gem5.opt 
>>> /home/rajesh/tests/configs/run_parsec.py 
>>> /home/rajesh/tests/linux-stable/vmlinux-4.19.83 
>>> /home/rajesh/tests/disk-image/parsec/image/parsec.img kvm streamcluster 
>>> simsmall 1
>>>
>>> Traceback (most recent call last):
>>>   File "", line 1, in 
>>>   File "build/X86/python/m5/main.py", line 455, in main
>>>   File "/home/rajesh/tests/configs/run_parsec.py", line 79, in 
>>> system = MySystem(kernel, disk, cpu, int(num_cpus))
>>>   File "/home/rajesh/tests/configs/system/system.py", line 78, in __init__
>>> self.createCacheHierarchy()
>>>   File "/home/rajesh/tests/configs/system/system.py", line 158, in 
>>> createCacheHierarchy
>>> cpu.mmucache.connectCPU(cpu)
>>>   File "/home/rajesh/tests/configs/system/caches.py", line 117, in 
>>> connectCPU
>>> for tlb in [cpu.itb, cpu.dtb]:
>>>   File "build/X86/python/m5/SimObject.py", line 1379, in __getattr__
>>> AttributeError: object 'X86KvmCPU' has no attribute 'itb'
>>>   (C++ object is not yet constructed, so wrapped C++ methods are 
>>> unavailable.)
>>>
>>> Thank you for your time.
>>>
>>> -- Rajesh Shashi Kumar
>>>
>>> On Thu, Jun 10, 2021 at 12:51 AM Bobby Bruce  wrote:
>>>
 The "Fix KVM on Intel platforms" patch is on stable, it was added here:
 https://gem5-review.googlesource.com/c/public/gem5/+/12278

 The other patch your cherry-picking I'm less sure of, but I dont think
 it's a relevant problem anymore from what I can ascertain.

 I apologize that the gem5art documentation is using such an old version
 of gem5. Have you managed to get this working with the stable branch
 (version 21.0)? I'm not aware of any reason why it shouldn't work.

 --
 Dr. Bobby R. Bruce
 Room 3050,
 Kemper Hall, UC Davis
 Davis,
 CA, 95616

 web: https://www.bobbybruce.net


 On Wed, Jun 9, 2021 at 7:06 AM Rajesh Shashi Kumar via gem5-users <
 gem5-users@gem5.org> wrote:

> Hi,
>
> I'm attempting to run PARSEC benchmarks in FS mode on gem5-v21.0. I
> have encountered an issue with KVM on Intel machine. I believe the same 
> was
> addressed by this patch
>  recommended
> earlier in the mailing list
> 
> .
>
>

[gem5-users] Re: KVM patch for FS mode and PARSEC on gem5-21

[Rajesh Shashi Kumar via gem5-users]

Hi Ayaz,

Thank you very much. That was indeed the issue. I also noticed that the
following was missing in configs/system/system.py. After adding this line,
I'm able to get FS,PARSEC simulations running on gem5-v21

   *self.intrctrl = IntrControl()*

Error message:
```
Error in unproxying param 'intrctrl' of system.pc
Traceback (most recent call last):
  File "", line 1, in 
  File "build/X86/python/m5/main.py", line 455, in main
  File "/home/rajesh/work/parsec-tests/configs/run_parsec.py", line 102, in

m5.instantiate()
  File "build/X86/python/m5/simulate.py", line 88, in instantiate
  File "build/X86/python/m5/SimObject.py", line 1599, in unproxyParams
  File "build/X86/python/m5/proxy.py", line 134, in unproxy
AttributeError: Can't resolve proxy 'any' of type 'IntrControl' from
'system.pc'
```

Thank you for your time.

-- Rajesh Shashi Kumar


On Thu, Jun 10, 2021 at 3:20 AM Ayaz Akram  wrote:

> Hi Rajesh,
>
> I think the error you are seeing is because tlbs are exposed through an
> mmu unit in gem5 now. I guess changing line 117 in caches.py as in this
> file (
> https://gem5.googlesource.com/public/gem5-resources/+/refs/heads/stable/src/boot-exit/configs/system/caches.py)
> should work for you.
>
> -Ayaz
>
> On Wed, Jun 9, 2021 at 2:04 PM Rajesh Shashi Kumar via gem5-users <
> gem5-users@gem5.org> wrote:
>
>> Thank you for the quick response.
>>
>>
>>- Thank you for confirming that the KVM fix has been integrated into
>>the stable branch (pointing to gem5-v21.0). At this time, I have not
>>managed to get PARSEC working in FS mode on the stable branch (v21.0) and 
>> I
>>am debugging the issue described below
>>- For context, I encountered the following error on gem5-v21.0 while
>>running PARSEC which made me explore reverting back to v20.1 (and the KVM
>>patch) as described in the gem5art documentation. I'm using the latest
>>PARSEC config run scripts from the gem5-resources repository
>>
>> 
>>
>> gem5 version 21.0.0.0
>> gem5 compiled Jun 10 2021 01:34:21
>> gem5 started Jun 10 2021 01:57:17
>> gem5 executing on lenovo, pid 20958
>> command line: gem5/build/X86/gem5.opt 
>> /home/rajesh/tests/configs/run_parsec.py 
>> /home/rajesh/tests/linux-stable/vmlinux-4.19.83 
>> /home/rajesh/tests/disk-image/parsec/image/parsec.img kvm streamcluster 
>> simsmall 1
>>
>> Traceback (most recent call last):
>>   File "", line 1, in 
>>   File "build/X86/python/m5/main.py", line 455, in main
>>   File "/home/rajesh/tests/configs/run_parsec.py", line 79, in 
>> system = MySystem(kernel, disk, cpu, int(num_cpus))
>>   File "/home/rajesh/tests/configs/system/system.py", line 78, in __init__
>> self.createCacheHierarchy()
>>   File "/home/rajesh/tests/configs/system/system.py", line 158, in 
>> createCacheHierarchy
>> cpu.mmucache.connectCPU(cpu)
>>   File "/home/rajesh/tests/configs/system/caches.py", line 117, in connectCPU
>> for tlb in [cpu.itb, cpu.dtb]:
>>   File "build/X86/python/m5/SimObject.py", line 1379, in __getattr__
>> AttributeError: object 'X86KvmCPU' has no attribute 'itb'
>>   (C++ object is not yet constructed, so wrapped C++ methods are 
>> unavailable.)
>>
>> Thank you for your time.
>>
>> -- Rajesh Shashi Kumar
>>
>> On Thu, Jun 10, 2021 at 12:51 AM Bobby Bruce  wrote:
>>
>>> The "Fix KVM on Intel platforms" patch is on stable, it was added here:
>>> https://gem5-review.googlesource.com/c/public/gem5/+/12278
>>>
>>> The other patch your cherry-picking I'm less sure of, but I dont think
>>> it's a relevant problem anymore from what I can ascertain.
>>>
>>> I apologize that the gem5art documentation is using such an old version
>>> of gem5. Have you managed to get this working with the stable branch
>>> (version 21.0)? I'm not aware of any reason why it shouldn't work.
>>>
>>> --
>>> Dr. Bobby R. Bruce
>>> Room 3050,
>>> Kemper Hall, UC Davis
>>> Davis,
>>> CA, 95616
>>>
>>> web: https://www.bobbybruce.net
>>>
>>>
>>> On Wed, Jun 9, 2021 at 7:06 AM Rajesh Shashi Kumar via gem5-users <
>>> gem5-users@gem5.org> wrote:
>>>
 Hi,

 I'm attempting to run PARSEC benchmarks in FS mode on gem5-v21.0. I
 have encountered an issue with KVM on Intel machine. I believe the same was
 addressed by this patch
  recommended
 earlier in the mailing list
 
 .

 panic: KVM: Failed to enter virtualized mode (hw reason:
 0x8021)

 Most of the tutorials on gem5art
  in
 gem5 documentation recommend using gem5-v20.1, but I noticed that applying
 the patch on 20.1 or 21 is not a trivial merge. For the time being, I've
 been able to use this specific commit tag described in
 

[gem5-users] Re: KVM patch for FS mode and PARSEC on gem5-21

[Ayaz Akram via gem5-users]

Hi Rajesh,

I think the error you are seeing is because tlbs are exposed through an mmu
unit in gem5 now. I guess changing line 117 in caches.py as in this file (
https://gem5.googlesource.com/public/gem5-resources/+/refs/heads/stable/src/boot-exit/configs/system/caches.py)
should work for you.

-Ayaz

On Wed, Jun 9, 2021 at 2:04 PM Rajesh Shashi Kumar via gem5-users <
gem5-users@gem5.org> wrote:

> Thank you for the quick response.
>
>
>- Thank you for confirming that the KVM fix has been integrated into
>the stable branch (pointing to gem5-v21.0). At this time, I have not
>managed to get PARSEC working in FS mode on the stable branch (v21.0) and I
>am debugging the issue described below
>- For context, I encountered the following error on gem5-v21.0 while
>running PARSEC which made me explore reverting back to v20.1 (and the KVM
>patch) as described in the gem5art documentation. I'm using the latest
>PARSEC config run scripts from the gem5-resources repository
>
> 
>
> gem5 version 21.0.0.0
> gem5 compiled Jun 10 2021 01:34:21
> gem5 started Jun 10 2021 01:57:17
> gem5 executing on lenovo, pid 20958
> command line: gem5/build/X86/gem5.opt 
> /home/rajesh/tests/configs/run_parsec.py 
> /home/rajesh/tests/linux-stable/vmlinux-4.19.83 
> /home/rajesh/tests/disk-image/parsec/image/parsec.img kvm streamcluster 
> simsmall 1
>
> Traceback (most recent call last):
>   File "", line 1, in 
>   File "build/X86/python/m5/main.py", line 455, in main
>   File "/home/rajesh/tests/configs/run_parsec.py", line 79, in 
> system = MySystem(kernel, disk, cpu, int(num_cpus))
>   File "/home/rajesh/tests/configs/system/system.py", line 78, in __init__
> self.createCacheHierarchy()
>   File "/home/rajesh/tests/configs/system/system.py", line 158, in 
> createCacheHierarchy
> cpu.mmucache.connectCPU(cpu)
>   File "/home/rajesh/tests/configs/system/caches.py", line 117, in connectCPU
> for tlb in [cpu.itb, cpu.dtb]:
>   File "build/X86/python/m5/SimObject.py", line 1379, in __getattr__
> AttributeError: object 'X86KvmCPU' has no attribute 'itb'
>   (C++ object is not yet constructed, so wrapped C++ methods are unavailable.)
>
> Thank you for your time.
>
> -- Rajesh Shashi Kumar
>
> On Thu, Jun 10, 2021 at 12:51 AM Bobby Bruce  wrote:
>
>> The "Fix KVM on Intel platforms" patch is on stable, it was added here:
>> https://gem5-review.googlesource.com/c/public/gem5/+/12278
>>
>> The other patch your cherry-picking I'm less sure of, but I dont think
>> it's a relevant problem anymore from what I can ascertain.
>>
>> I apologize that the gem5art documentation is using such an old version
>> of gem5. Have you managed to get this working with the stable branch
>> (version 21.0)? I'm not aware of any reason why it shouldn't work.
>>
>> --
>> Dr. Bobby R. Bruce
>> Room 3050,
>> Kemper Hall, UC Davis
>> Davis,
>> CA, 95616
>>
>> web: https://www.bobbybruce.net
>>
>>
>> On Wed, Jun 9, 2021 at 7:06 AM Rajesh Shashi Kumar via gem5-users <
>> gem5-users@gem5.org> wrote:
>>
>>> Hi,
>>>
>>> I'm attempting to run PARSEC benchmarks in FS mode on gem5-v21.0. I have
>>> encountered an issue with KVM on Intel machine. I believe the same was
>>> addressed by this patch
>>>  recommended
>>> earlier in the mailing list
>>> 
>>> .
>>>
>>> panic: KVM: Failed to enter virtualized mode (hw reason: 0x8021)
>>>
>>> Most of the tutorials on gem5art
>>>  in
>>> gem5 documentation recommend using gem5-v20.1, but I noticed that applying
>>> the patch on 20.1 or 21 is not a trivial merge. For the time being, I've
>>> been able to use this specific commit tag described in
>>> gem5art-experiments .
>>>
>>> git clone https://gem5.googlesource.com/public/gem5cd gem5
>>> git checkout *d40f0bc579fb8b10da7181*
>>> git remote add darchr https://github.com/darchr/gem5
>>> git fetch darchr*git cherry-pick 
>>> 6450aaa7ca9e3040fb9eecf69c51a01884ac370c**git cherry-pick 
>>> 3403665994b55f664f4edfc9074650aaa7ddcd2c*
>>> scons build/X86/gem5.opt -j8
>>>
>>> At this point, I have two questions:
>>>
>>>1. Is the KVM patch necessary on gem5-v21?
>>>2. In case someone has encountered a similar issue previously, were
>>>you able to find a way to merge the patch on newer gem5 releases
>>>
>>> Thank you for your time.
>>>
>>> Regards,
>>> Rajesh Shashi Kumar
>>> ___
>>> gem5-users mailing list -- gem5-users@gem5.org
>>> To unsubscribe send an email to gem5-users-le...@gem5.org
>>> %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
>>
>> ___
> gem5-users mailing list -- 

[gem5-users] Re: KVM patch for FS mode and PARSEC on gem5-21

[Rajesh Shashi Kumar via gem5-users]

Thank you for the quick response.


   - Thank you for confirming that the KVM fix has been integrated into the
   stable branch (pointing to gem5-v21.0). At this time, I have not managed to
   get PARSEC working in FS mode on the stable branch (v21.0) and I am
   debugging the issue described below
   - For context, I encountered the following error on gem5-v21.0 while
   running PARSEC which made me explore reverting back to v20.1 (and the KVM
   patch) as described in the gem5art documentation. I'm using the latest
   PARSEC config run scripts from the gem5-resources repository
   


gem5 version 21.0.0.0
gem5 compiled Jun 10 2021 01:34:21
gem5 started Jun 10 2021 01:57:17
gem5 executing on lenovo, pid 20958
command line: gem5/build/X86/gem5.opt
/home/rajesh/tests/configs/run_parsec.py
/home/rajesh/tests/linux-stable/vmlinux-4.19.83
/home/rajesh/tests/disk-image/parsec/image/parsec.img kvm
streamcluster simsmall 1

Traceback (most recent call last):
  File "", line 1, in 
  File "build/X86/python/m5/main.py", line 455, in main
  File "/home/rajesh/tests/configs/run_parsec.py", line 79, in 
system = MySystem(kernel, disk, cpu, int(num_cpus))
  File "/home/rajesh/tests/configs/system/system.py", line 78, in __init__
self.createCacheHierarchy()
  File "/home/rajesh/tests/configs/system/system.py", line 158, in
createCacheHierarchy
cpu.mmucache.connectCPU(cpu)
  File "/home/rajesh/tests/configs/system/caches.py", line 117, in connectCPU
for tlb in [cpu.itb, cpu.dtb]:
  File "build/X86/python/m5/SimObject.py", line 1379, in __getattr__
AttributeError: object 'X86KvmCPU' has no attribute 'itb'
  (C++ object is not yet constructed, so wrapped C++ methods are unavailable.)

Thank you for your time.

-- Rajesh Shashi Kumar

On Thu, Jun 10, 2021 at 12:51 AM Bobby Bruce  wrote:

> The "Fix KVM on Intel platforms" patch is on stable, it was added here:
> https://gem5-review.googlesource.com/c/public/gem5/+/12278
>
> The other patch your cherry-picking I'm less sure of, but I dont think
> it's a relevant problem anymore from what I can ascertain.
>
> I apologize that the gem5art documentation is using such an old version of
> gem5. Have you managed to get this working with the stable branch (version
> 21.0)? I'm not aware of any reason why it shouldn't work.
>
> --
> Dr. Bobby R. Bruce
> Room 3050,
> Kemper Hall, UC Davis
> Davis,
> CA, 95616
>
> web: https://www.bobbybruce.net
>
>
> On Wed, Jun 9, 2021 at 7:06 AM Rajesh Shashi Kumar via gem5-users <
> gem5-users@gem5.org> wrote:
>
>> Hi,
>>
>> I'm attempting to run PARSEC benchmarks in FS mode on gem5-v21.0. I have
>> encountered an issue with KVM on Intel machine. I believe the same was
>> addressed by this patch
>>  recommended
>> earlier in the mailing list
>> 
>> .
>>
>> panic: KVM: Failed to enter virtualized mode (hw reason: 0x8021)
>>
>> Most of the tutorials on gem5art
>>  in
>> gem5 documentation recommend using gem5-v20.1, but I noticed that applying
>> the patch on 20.1 or 21 is not a trivial merge. For the time being, I've
>> been able to use this specific commit tag described in
>> gem5art-experiments .
>>
>> git clone https://gem5.googlesource.com/public/gem5cd gem5
>> git checkout *d40f0bc579fb8b10da7181*
>> git remote add darchr https://github.com/darchr/gem5
>> git fetch darchr*git cherry-pick 
>> 6450aaa7ca9e3040fb9eecf69c51a01884ac370c**git cherry-pick 
>> 3403665994b55f664f4edfc9074650aaa7ddcd2c*
>> scons build/X86/gem5.opt -j8
>>
>> At this point, I have two questions:
>>
>>1. Is the KVM patch necessary on gem5-v21?
>>2. In case someone has encountered a similar issue previously, were
>>you able to find a way to merge the patch on newer gem5 releases
>>
>> Thank you for your time.
>>
>> Regards,
>> Rajesh Shashi Kumar
>> ___
>> gem5-users mailing list -- gem5-users@gem5.org
>> To unsubscribe send an email to gem5-users-le...@gem5.org
>> %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
>
>
___
gem5-users mailing list -- gem5-users@gem5.org
To unsubscribe send an email to gem5-users-le...@gem5.org
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s

[gem5-users] Re: KVM patch for FS mode and PARSEC on gem5-21

[Bobby Bruce via gem5-users]

The "Fix KVM on Intel platforms" patch is on stable, it was added here:
https://gem5-review.googlesource.com/c/public/gem5/+/12278

The other patch your cherry-picking I'm less sure of, but I dont think it's
a relevant problem anymore from what I can ascertain.

I apologize that the gem5art documentation is using such an old version of
gem5. Have you managed to get this working with the stable branch (version
21.0)? I'm not aware of any reason why it shouldn't work.

--
Dr. Bobby R. Bruce
Room 3050,
Kemper Hall, UC Davis
Davis,
CA, 95616

web: https://www.bobbybruce.net


On Wed, Jun 9, 2021 at 7:06 AM Rajesh Shashi Kumar via gem5-users <
gem5-users@gem5.org> wrote:

> Hi,
>
> I'm attempting to run PARSEC benchmarks in FS mode on gem5-v21.0. I have
> encountered an issue with KVM on Intel machine. I believe the same was
> addressed by this patch
>  recommended
> earlier in the mailing list
> 
> .
>
> panic: KVM: Failed to enter virtualized mode (hw reason: 0x8021)
>
> Most of the tutorials on gem5art
>  in
> gem5 documentation recommend using gem5-v20.1, but I noticed that applying
> the patch on 20.1 or 21 is not a trivial merge. For the time being, I've
> been able to use this specific commit tag described in gem5art-experiments
> .
>
> git clone https://gem5.googlesource.com/public/gem5cd gem5
> git checkout *d40f0bc579fb8b10da7181*
> git remote add darchr https://github.com/darchr/gem5
> git fetch darchr*git cherry-pick 
> 6450aaa7ca9e3040fb9eecf69c51a01884ac370c**git cherry-pick 
> 3403665994b55f664f4edfc9074650aaa7ddcd2c*
> scons build/X86/gem5.opt -j8
>
> At this point, I have two questions:
>
>1. Is the KVM patch necessary on gem5-v21?
>2. In case someone has encountered a similar issue previously, were
>you able to find a way to merge the patch on newer gem5 releases
>
> Thank you for your time.
>
> Regards,
> Rajesh Shashi Kumar
> ___
> gem5-users mailing list -- gem5-users@gem5.org
> To unsubscribe send an email to gem5-users-le...@gem5.org
> %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
___
gem5-users mailing list -- gem5-users@gem5.org
To unsubscribe send an email to gem5-users-le...@gem5.org
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s