Re: [PATCH] mm/huge_memory: Fix vmf_insert_pfn_{pmd, pud}() crash, handle unaligned addresses

[Matthew Wilcox]

On Thu, May 09, 2019 at 09:31:41AM -0700, Dan Williams wrote:
> diff --git a/drivers/dax/device.c b/drivers/dax/device.c
> index e428468ab661..996d68ff992a 100644
> --- a/drivers/dax/device.c
> +++ b/drivers/dax/device.c
> @@ -184,8 +184,7 @@ static vm_fault_t __dev_dax_pmd_fault(struct dev_dax *dev_dax,
> +	return vmf_insert_pfn_pmd(vmf, *pfn, vmf->flags & FAULT_FLAG_WRITE);

I think we can ditch the third parameter too.  Going through the callers ...

> @@ -235,8 +234,7 @@ static vm_fault_t __dev_dax_pud_fault(struct dev_dax *dev_dax,
> +	return vmf_insert_pfn_pud(vmf, *pfn, vmf->flags & FAULT_FLAG_WRITE);

> @@ -1575,8 +1575,7 @@ static vm_fault_t dax_iomap_pmd_fault(struct vm_fault *vmf, pfn_t *pfnp,
> +		result = vmf_insert_pfn_pmd(vmf, pfn, write);

This 'write' parameter came earlier from:

        bool write = vmf->flags & FAULT_FLAG_WRITE;

and it is not modified subsequently.

> @@ -1686,8 +1685,7 @@ dax_insert_pfn_mkwrite(struct vm_fault *vmf, pfn_t pfn, unsigned int order)
> +		ret = vmf_insert_pfn_pmd(vmf, pfn, FAULT_FLAG_WRITE);

If FAULT_FLAG_WRITE is not set in a mkwrite handler, I don't know
what's gone wrong with the world.

Even without these changes,

Reviewed-by: Matthew Wilcox <willy@xxxxxxxxxxxxx>