Re: [PATCH 04/31] mm/pgtable: allow pte_offset_map[_lock]() to fail

["Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com>]

Hi Hugh,

Sorry for not checking about this before. I am looking at a kernel
crash (BUG_ON()) on ppc64 with 4K page size. The reason we hit
BUG_ON() is beause we have pmd_same calling BUG_ON on 4K with hash
translation. We don't support THP with 4k page size and hash
translation.

Hugh Dickins <hughd@google.com> writes:

....

 +
> +pte_t *pte_offset_map_nolock(struct mm_struct *mm, pmd_t *pmd,
> +			     unsigned long addr, spinlock_t **ptlp)
> +{
> +	pmd_t pmdval;
> +	pte_t *pte;
> +
> +	pte = __pte_offset_map(pmd, addr, &pmdval);
> +	if (likely(pte))
> +		*ptlp = pte_lockptr(mm, &pmdval);
> +	return pte;
> +}
> +
> +pte_t *__pte_offset_map_lock(struct mm_struct *mm, pmd_t *pmd,
> +			     unsigned long addr, spinlock_t **ptlp)
> +{
> +	spinlock_t *ptl;
> +	pmd_t pmdval;
> +	pte_t *pte;
> +again:
> +	pte = __pte_offset_map(pmd, addr, &pmdval);
> +	if (unlikely(!pte))
> +		return pte;
> +	ptl = pte_lockptr(mm, &pmdval);
> +	spin_lock(ptl);
> +	if (likely(pmd_same(pmdval, pmdp_get_lockless(pmd)))) {
> +		*ptlp = ptl;
> +		return pte;
> +	}
> +	pte_unmap_unlock(pte, ptl);
> +	goto again;
> +}

What is expected by that pmd_same check? We are holding pte lock
and not pmd lock. So contents of pmd can change.

-aneesh