Re: [PATCH] tools/mm: Add madvise tool

[Mike Rapoport <rppt@kernel.org>]

On Fri, Sep 05, 2025 at 02:36:36PM +0100, Kiryl Shutsemau wrote:
> On Fri, Sep 05, 2025 at 01:59:10PM +0300, Mike Rapoport wrote:
> > On Thu, Sep 04, 2025 at 06:57:29PM +0100, kirill@shutemov.name wrote:
> > > From: Kiryl Shutsemau <kas@kernel.org>
> > > 
> > > Add a simple tool that allows to issue an advice on a process or a file.
> > > 
> > > It can be useful to experiment with effects of an advice on a workload
> > > without modifying the workload itself.
> > > 
> > > Only supports advices available for process_madvise().
> > > 
> > > Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
> > > ---
> > >  tools/mm/.gitignore |   4 +-
> > >  tools/mm/Makefile   |   2 +-
> > >  tools/mm/madvise.c  | 170 ++++++++++++++++++++++++++++++++++++++++++++
> > >  3 files changed, 174 insertions(+), 2 deletions(-)
> > >  create mode 100644 tools/mm/madvise.c
> > > 
> > 
> > ...
> > 
> > > +		unsigned long addr, hpage_pmd_size;
> > > +		void *p;
> > > +		int ret;
> > > +
> > > +		hpage_pmd_size = read_pmd_pagesize();
> > > +		if (!hpage_pmd_size) {
> > > +			printf("Reading PMD pagesize failed");
> > > +			exit(-1);
> > 
> > Shouldn't it fail only for collapse? Other advices don't depend on THP.
> 
> True. But aligning everybody wouldn't hurt.

Right, but this make the tool unusable for kernels built with !THP. 
 
> I would argue MADV_COLLAPSE shouldn't require this either. You don't
> need PMD to collapse page cache folio.
> 
> -- 
>   Kiryl Shutsemau / Kirill A. Shutemov

-- 
Sincerely yours,
Mike.