Re: [PATCH] rteval: kcompile.py: Relax the requirement for kernel tarball filename

[John Kacur <jkacur@redhat.com>]

On Wed, 22 Sep 2021, venkata.pyla@toshiba-tsip.com wrote:

> From: venkata pyla <venkata.pyla@toshiba-tsip.com>
> 
> The kcompile module includes kernel version when looking for tarballs
> or source folders. This necessitates periodically bumping the
> kernel_prefix but also requires the user to provide particular version
> of the kernel sources when using this workload. As the intent is to
> generate a compile workload the source version used shouldn't matter.
> 
> To make life easier both for users and developers, relax the
> requirement for specific version of kernel sources when using
> kcompile.
> 
> Signed-off-by: venkata pyla <venkata.pyla@toshiba-tsip.com>
> ---
>  rteval/modules/loads/kcompile.py | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/rteval/modules/loads/kcompile.py b/rteval/modules/loads/kcompile.py
> index 8d08a3d..be40a62 100644
> --- a/rteval/modules/loads/kcompile.py
> +++ b/rteval/modules/loads/kcompile.py
> @@ -35,7 +35,7 @@ from rteval.Log import Log
>  from rteval.misc import expand_cpulist, compress_cpulist
>  from rteval.systopology import SysTopology
>  
> -kernel_prefix = "linux-5.13"
> +kernel_prefix = "linux"
>  
>  class KBuildJob:
>      '''Class to manage a build job bound to a particular node'''
> -- 
> 2.20.1
> 
> 
> 

It shouldn't be onerous for a developer to grab the right version of the 
kernel from kernel.org and save it in loadsource

As for users, they are probably getting rteval from a distribution.
In Fedora and rhel we split it out into two packages.
rteval and rteval-loads. rteval-loads has the correct kernel version.

The idea behind this is that rteval changes more often than the version of 
the kernel we compile, so the user doesn't have to download a new kernel 
everytime they download a new version of rteval.

You make a good point though that the purpose is to provide a load.
We have had some discussions in the past about whether making sure 
everyone is using the same load so that we have apples to apples 
comparisons is necessary or not. I might revisit this in the future with 
some innovations to kcompile, but for now this is a NAK.

Thank You

John