Re: [PATCH v6 14/18] fs/resctrl: Control visibility of global per-CPU kernel mode group files

[Babu Moger]

Hi Reinette,

On 10/19/26 17:00, Reinette Chatre wrote:
> Hi Babu,
>
> On 10/19/26 9:31 AM, Babu Moger wrote:
> > Hi Reinette,
> >
> > On 10/15/26 10:46, Reinette Chatre wrote:
> > > Hi Babu,
> > >
> > > On 10/1/26 8:10 AM, Babu Moger wrote:
> > > > The kmode_cpus and kmode_cpus_list files are specific to the
> > > > RESCTRL_GLOBAL_ENABLE_PER_CPU kernel mode. These files are created in a
> > > > hidden state when an rdtgroup is created.
> > > >
> > > > Introduce resctrl_kmode_global_files_show() to manage the visibility of
> > > > these files based on whether the global per-CPU kernel mode is enabled or
> > > > disabled.
> > >
> > > This does not sound right and does not match the patch. The visibility of files
> > > is per group, not globally whether the mode is enabled or not, right?
> > >
> > > It is ok to have cryptic changelogs when changes are small and simple but these
> > > changes are not small and simple.
> >
> > "The kmode_cpus and kmode_cpus_list files are specific to the RESCTRL_GLOBAL_ENABLE_PER_CPU kernel mode. These files are created in a hidden state when an rdtgroup is created.
>
> Could the context please help the reader a bit more? Some items that could help:
> - *why* are these files specific to global per-CPU kernel mode? (mention that it is used by user space
>     to do the per-CPU enable/disable of the mode)
> - *why* is it created in a hidden state? (mention that a resource group can only be assigned
>    to the global per-CPU kernel mode after it has been created via a separate resctrl file)

Sure. Will add that.

> > > Introduce resctrl_kmode_global_files_show() to manage the visibility
> > of these files based on whether a resource group backs the global
>
> hmmm ... resctrl_kmode_global_files_show() does not actually do any
> checks to ensure the resource group backs the global per-CPU kernel mode.

Yes. Will fix this.

> > per-CPU kernel mode. Hide the files when the group is no longer
> > associated with the mode.
> Could the patches be organized so that this becomes a symmetrical change
> that highlights when files become visible and when they become invisible to support
> the description within this changelog?

I didn't quite understand this comment. Could you please elaborate?

Thanks,
Babu