2019-09-11 22:35:06

by Ralph Campbell

Subject: [PATCH 2/4] mm/hmm: allow snapshot of the special zero page

Allow hmm_range_fault() to return success (0) when the CPU pagetable
entry points to the special shared zero page.
The caller can then handle the zero page by possibly clearing device
private memory instead of DMAing a zero page.

Signed-off-by: Ralph Campbell <[email protected]>
Cc: "Jérôme Glisse" <[email protected]>
Cc: Jason Gunthorpe <[email protected]>
Cc: Christoph Hellwig <[email protected]>
---
mm/hmm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/hmm.c b/mm/hmm.c
index 06041d4399ff..7217912bef13 100644
--- a/mm/hmm.c
+++ b/mm/hmm.c
@@ -532,7 +532,7 @@ static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr,
return -EBUSY;
} else if (IS_ENABLED(CONFIG_ARCH_HAS_PTE_SPECIAL) && pte_special(pte)) {
*pfn = range->values[HMM_PFN_SPECIAL];
- return -EFAULT;
+ return is_zero_pfn(pte_pfn(pte)) ? 0 : -EFAULT;
}

*pfn = hmm_device_entry_from_pfn(range, pte_pfn(pte)) | cpu_flags;
--
2.20.1

2019-09-12 08:30:45

by Christoph Hellwig

Subject: Re: [PATCH 2/4] mm/hmm: allow snapshot of the special zero page

On Wed, Sep 11, 2019 at 03:28:27PM -0700, Ralph Campbell wrote:
> Allow hmm_range_fault() to return success (0) when the CPU pagetable
> entry points to the special shared zero page.
> The caller can then handle the zero page by possibly clearing device
> private memory instead of DMAing a zero page.
>
> Signed-off-by: Ralph Campbell <[email protected]>
> Cc: "J?r?me Glisse" <[email protected]>
> Cc: Jason Gunthorpe <[email protected]>
> Cc: Christoph Hellwig <[email protected]>
> ---
> mm/hmm.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/hmm.c b/mm/hmm.c
> index 06041d4399ff..7217912bef13 100644
> --- a/mm/hmm.c
> +++ b/mm/hmm.c
> @@ -532,7 +532,7 @@ static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr,
> return -EBUSY;
> } else if (IS_ENABLED(CONFIG_ARCH_HAS_PTE_SPECIAL) && pte_special(pte)) {
> *pfn = range->values[HMM_PFN_SPECIAL];
> - return -EFAULT;
> + return is_zero_pfn(pte_pfn(pte)) ? 0 : -EFAULT;

Any chance to just use a normal if here:

if (!is_zero_pfn(pte_pfn(pte)))
return -EFAULT;
return 0;

2019-09-12 19:07:34

by Ralph Campbell

Subject: Re: [PATCH 2/4] mm/hmm: allow snapshot of the special zero page

On 9/12/19 1:26 AM, Christoph Hellwig wrote:
> On Wed, Sep 11, 2019 at 03:28:27PM -0700, Ralph Campbell wrote:
>> Allow hmm_range_fault() to return success (0) when the CPU pagetable
>> entry points to the special shared zero page.
>> The caller can then handle the zero page by possibly clearing device
>> private memory instead of DMAing a zero page.
>>
>> Signed-off-by: Ralph Campbell <[email protected]>
>> Cc: "Jérôme Glisse" <[email protected]>
>> Cc: Jason Gunthorpe <[email protected]>
>> Cc: Christoph Hellwig <[email protected]>
>> ---
>> mm/hmm.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/mm/hmm.c b/mm/hmm.c
>> index 06041d4399ff..7217912bef13 100644
>> --- a/mm/hmm.c
>> +++ b/mm/hmm.c
>> @@ -532,7 +532,7 @@ static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr,
>> return -EBUSY;
>> } else if (IS_ENABLED(CONFIG_ARCH_HAS_PTE_SPECIAL) && pte_special(pte)) {
>> *pfn = range->values[HMM_PFN_SPECIAL];
>> - return -EFAULT;
>> + return is_zero_pfn(pte_pfn(pte)) ? 0 : -EFAULT;
>
> Any chance to just use a normal if here:
>
> if (!is_zero_pfn(pte_pfn(pte)))
> return -EFAULT;
> return 0;
>

Sure, no problem.