Re: [PATCH] mm/ptdump: Replace READ_ONCE() with standard page table accessors

[David Hildenbrand]

On 30.09.25 06:37, Dev Jain wrote:
> On 30/09/25 8:22 am, Anshuman Khandual wrote:
> > Replace READ_ONCE() with standard page table accessors i.e pxdp_get() which
> > anyways default into READ_ONCE() in cases where platform does not override.
> >
> > Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> > Cc: David Hildenbrand <david@xxxxxxxxxx>
> > Cc: linux-mm@xxxxxxxxx
> > Cc: linux-kernel@xxxxxxxxxxxxxxx
> > Signed-off-by: Anshuman Khandual <anshuman.khandual@xxxxxxx>
> > ---
> >    mm/ptdump.c | 8 ++++----
> >    1 file changed, 4 insertions(+), 4 deletions(-)
> >
> > diff --git a/mm/ptdump.c b/mm/ptdump.c
> > index b600c7f864b8..18861501b533 100644
> > --- a/mm/ptdump.c
> > +++ b/mm/ptdump.c
> > @@ -31,7 +31,7 @@ static int ptdump_pgd_entry(pgd_t *pgd, unsigned long addr,
> >    			    unsigned long next, struct mm_walk *walk)
> >    {
> >    	struct ptdump_state *st = walk->private;
> > -	pgd_t val = READ_ONCE(*pgd);
> > +	pgd_t val = pgdp_get(pgd);
> >    
> >    #if CONFIG_PGTABLE_LEVELS > 4 && \
> >    		(defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS))
> > @@ -54,7 +54,7 @@ static int ptdump_p4d_entry(p4d_t *p4d, unsigned long addr,
> >    			    unsigned long next, struct mm_walk *walk)
> >    {
> >    	struct ptdump_state *st = walk->private;
> > -	p4d_t val = READ_ONCE(*p4d);
> > +	p4d_t val = p4dp_get(p4d);
> >    
> >    #if CONFIG_PGTABLE_LEVELS > 3 && \
> >    		(defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS))
> > @@ -77,7 +77,7 @@ static int ptdump_pud_entry(pud_t *pud, unsigned long addr,
> >    			    unsigned long next, struct mm_walk *walk)
> >    {
> >    	struct ptdump_state *st = walk->private;
> > -	pud_t val = READ_ONCE(*pud);
> > +	pud_t val = pudp_get(pud);
> >    
> >    #if CONFIG_PGTABLE_LEVELS > 2 && \
> >    		(defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS))
> > @@ -100,7 +100,7 @@ static int ptdump_pmd_entry(pmd_t *pmd, unsigned long addr,
> >    			    unsigned long next, struct mm_walk *walk)
> >    {
> >    	struct ptdump_state *st = walk->private;
> > -	pmd_t val = READ_ONCE(*pmd);
> > +	pmd_t val = pmdp_get(pmd);
>
> I believe this should go through pmdp_get_lockless(). I can see in pgtable.h that
> some magic is required on some arches to decode the pmd correctly in case walking
> without locks.

pmdp_get_lockless() is a nasty thingy to handle selected 32bit 
architectures.

But given that we're using ptep_get_lockless() in ptdump_pmd_entry() it 
probably wouldn't hurt to use pmdp_get_lockless() here.

Staring at ARCH_HAS_PTDUMP, I don't think any 32bit arch would actually 
end up compiling ptdump.c.

E.g., on x86 only X86_64 ends up selecting ARCH_HAS_PTDUMP.

--
Cheers

David / dhildenb