Re: [PATCH v9 0/3] Enhance sysfs handling for memory hotplug in weighted interleave

[Andrew Morton]

On Thu, 17 Apr 2025 15:41:30 -0700 Dan Williams <dan.j.williams@xxxxxxxxx> wrote:

> Andrew Morton wrote:
> > On Thu, 17 Apr 2025 17:10:08 +0900 Rakie Kim <rakie.kim@xxxxxx> wrote:
> > 
> > > I sincerely apologize for causing repeated inconvenience. The series of
> > > patches version v8 that was merged into -mm, mm-new today needs
> > > additional corrections.
> > > Link: https://lore.kernel.org/all/6800742de6315_130fd2949c@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx.notmuch/
> > > Therefore, I have updated a new version v9, in which the problems have
> > > been addressed.
> > 
> > No probs, this is why mm.git workflow (mm-new -> mm-unstable ->
> > mm-stable -> mainline) operates as it does - to easily permit revisions
> > and replacements as patches move towards their final state.
> > 
> > Please note that I added a cc:stable to your [1/N] patch - sysfs leaks
> > should be fixed in earlier kernels.  I considered this to be low
> > priority - if it's higher priority than this patch should best have
> > been separated from the series, so it can take a different merge path
> > from the other patches.  
> 
> The risk of leak is low because it only appears to trigger if setup
> fails. Setup only fails due to -ENOMEM which is unlikely to happen from
> a late_initcall() when memory pressure is low.

Oh, OK, thanks.  I added the above paragraph to the changelog and
removed the cc:stable.

Generally, we assume that -ENOMEM doesn't happen in __init code.  If it
does, the kernel is totally messed up anyway)