Re: [Xenomai] [PATCH v3] ipipe x86 mm: handle huge pages in memory pinning

[Henning Schild <henning.schild@siemens.com>]

Hey,

i discussed the issue with the guy that introduced huge ioremap. He
agreed that mainline also needs to handle huge pages in vmalloc_fault
and came up with a very similar patch. (the discussion can partially be
found in this lists archive)
I assume it will soon hit mainline and will get backported to 4.1. So i
guess we can just wait for that and use "nohugeiomap" until the issue
is fixed mainline and merged in our 4.1.y branch.

Henning

On Wed, 3 Feb 2016 13:59:25 +0100
Henning Schild <henning.schild@siemens.com> wrote:

> In 4.1 huge page mapping of io memory was introduced, enable ipipe to
> handle that when pinning kernel memory.
> 
> change that introduced the feature
> 0f616be120c632c818faaea9adcb8f05a7a8601f
> 
> Signed-off-by: Henning Schild <henning.schild@siemens.com>
> ---
>  arch/x86/mm/fault.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> This version has the 32bit case changed according to the problems
> found in the review. That patch finally should keep the old
> behaviour. But we should still discuss whether the old behaviour was
> correct and should be kept. If the pinning it not required it should
> be removed instead of beeing made more complex.
> 
> diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
> index fd5bbcc..e78bd09 100644
> --- a/arch/x86/mm/fault.c
> +++ b/arch/x86/mm/fault.c
> @@ -211,6 +211,8 @@ static inline pmd_t *vmalloc_sync_one(pgd_t *pgd,
> unsigned long address) pud_k = pud_offset(pgd_k, address);
>  	if (!pud_present(*pud_k))
>  		return NULL;
> +	if (pud_large(*pud))
> +		return NULL;
>  
>  	pmd = pmd_offset(pud, address);
>  	pmd_k = pmd_offset(pud_k, address);
> @@ -400,6 +402,8 @@ static inline int vmalloc_sync_one(pgd_t *pgd,
> unsigned long address) 
>  	if (pud_none(*pud) || pud_page_vaddr(*pud) !=
> pud_page_vaddr(*pud_ref)) BUG();
> +	if (pud_large(*pud))
> +		return 0;
>  
>  	pmd = pmd_offset(pud, address);
>  	pmd_ref = pmd_offset(pud_ref, address);
> @@ -408,6 +412,8 @@ static inline int vmalloc_sync_one(pgd_t *pgd,
> unsigned long address) 
>  	if (pmd_none(*pmd) || pmd_page(*pmd) != pmd_page(*pmd_ref))
>  		BUG();
> +	if (pmd_large(*pmd))
> +		return 0;
>  
>  	pte_ref = pte_offset_kernel(pmd_ref, address);
>  	if (!pte_present(*pte_ref))