Re: [PATCH 07/13] mm/gup: Fix the lockless PMD access

[Hugh Dickins]

On Sat, 22 Oct 2022, Peter Zijlstra wrote:

> On architectures where the PTE/PMD is larger than the native word size
> (i386-PAE for example), READ_ONCE() can do the wrong thing. Use
> pmdp_get_lockless() just like we use ptep_get_lockless().

I thought that was something Will Deacon put a lot of effort
into handling around 5.8 and 5.9: see "strong prevailing wind" in
include/asm-generic/rwonce.h, formerly in include/linux/compiler.h.

Was it too optimistic?  Did the wind drop?

I'm interested in the answer, but I've certainly no objection
to making this all more obviously robust - thanks.

Hugh

> 
> Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> ---
>  kernel/events/core.c |    2 +-
>  mm/gup.c             |    2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
> 
> --- a/kernel/events/core.c
> +++ b/kernel/events/core.c
> @@ -7186,7 +7186,7 @@ static u64 perf_get_pgtable_size(struct
>  		return pud_leaf_size(pud);
>  
>  	pmdp = pmd_offset_lockless(pudp, pud, addr);
> -	pmd = READ_ONCE(*pmdp);
> +	pmd = pmdp_get_lockless(pmdp);
>  	if (!pmd_present(pmd))
>  		return 0;
>  
> --- a/mm/gup.c
> +++ b/mm/gup.c
> @@ -2507,7 +2507,7 @@ static int gup_pmd_range(pud_t *pudp, pu
>  
>  	pmdp = pmd_offset_lockless(pudp, pud, addr);
>  	do {
> -		pmd_t pmd = READ_ONCE(*pmdp);
> +		pmd_t pmd = pmdp_get_lockless(pmdp);
>  
>  		next = pmd_addr_end(addr, end);
>  		if (!pmd_present(pmd))