[PATCH] mm/nvdimm: Use correct alignment when looking at first pfn from a region

[PATCH] mm/nvdimm: Use correct alignment when looking at first pfn from a region

[Aneesh Kumar K.V]

We already add the start_pad to the resource->start but fails to section
align the start. This make sure with altmap we compute the right first
pfn when start_pad is zero and we are doing an align down of start address.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
---
 kernel/memremap.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/memremap.c b/kernel/memremap.c
index a856cb5ff192..23d77b60e728 100644
--- a/kernel/memremap.c
+++ b/kernel/memremap.c
@@ -59,9 +59,9 @@ static unsigned long pfn_first(struct dev_pagemap *pgmap)
 {
 	const struct resource *res = &pgmap->res;
 	struct vmem_altmap *altmap = &pgmap->altmap;
-	unsigned long pfn;
+	unsigned long pfn = PHYS_PFN(res->start);
 
-	pfn = res->start >> PAGE_SHIFT;
+	pfn = SECTION_ALIGN_DOWN(pfn);
 	if (pgmap->altmap_valid)
 		pfn += vmem_altmap_offset(altmap);
 	return pfn;
-- 
2.21.0

Re: [PATCH] mm/nvdimm: Use correct alignment when looking at first pfn from a region

[Dan Williams]

On Mon, May 13, 2019 at 7:55 PM Aneesh Kumar K.V
<aneesh.kumar@linux.ibm.com> wrote:
>
> We already add the start_pad to the resource->start but fails to section
> align the start. This make sure with altmap we compute the right first
> pfn when start_pad is zero and we are doing an align down of start address.
>
> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
> ---
>  kernel/memremap.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/memremap.c b/kernel/memremap.c
> index a856cb5ff192..23d77b60e728 100644
> --- a/kernel/memremap.c
> +++ b/kernel/memremap.c
> @@ -59,9 +59,9 @@ static unsigned long pfn_first(struct dev_pagemap *pgmap)
>  {
>         const struct resource *res = &pgmap->res;
>         struct vmem_altmap *altmap = &pgmap->altmap;
> -       unsigned long pfn;
> +       unsigned long pfn = PHYS_PFN(res->start);
>
> -       pfn = res->start >> PAGE_SHIFT;
> +       pfn = SECTION_ALIGN_DOWN(pfn);

This does not seem right to me it breaks the assumptions of where the
first expected valid pfn occurs in the passed in range.

Re: [PATCH] mm/nvdimm: Use correct alignment when looking at first pfn from a region

[Aneesh Kumar K.V]

On 5/14/19 9:59 AM, Dan Williams wrote:
> On Mon, May 13, 2019 at 7:55 PM Aneesh Kumar K.V
> <aneesh.kumar@linux.ibm.com> wrote:
>>
>> We already add the start_pad to the resource->start but fails to section
>> align the start. This make sure with altmap we compute the right first
>> pfn when start_pad is zero and we are doing an align down of start address.
>>
>> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
>> ---
>>   kernel/memremap.c | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/kernel/memremap.c b/kernel/memremap.c
>> index a856cb5ff192..23d77b60e728 100644
>> --- a/kernel/memremap.c
>> +++ b/kernel/memremap.c
>> @@ -59,9 +59,9 @@ static unsigned long pfn_first(struct dev_pagemap *pgmap)
>>   {
>>          const struct resource *res = &pgmap->res;
>>          struct vmem_altmap *altmap = &pgmap->altmap;
>> -       unsigned long pfn;
>> +       unsigned long pfn = PHYS_PFN(res->start);
>>
>> -       pfn = res->start >> PAGE_SHIFT;
>> +       pfn = SECTION_ALIGN_DOWN(pfn);
> 
> This does not seem right to me it breaks the assumptions of where the
> first expected valid pfn occurs in the passed in range.
> 

How do we define the first valid pfn? Isn't that at pfn_sb->dataoff ?

-aneesh

Re: [PATCH] mm/nvdimm: Use correct alignment when looking at first pfn from a region

[Aneesh Kumar K.V]

"Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com> writes:

> On 5/14/19 9:59 AM, Dan Williams wrote:
>> On Mon, May 13, 2019 at 7:55 PM Aneesh Kumar K.V
>> <aneesh.kumar@linux.ibm.com> wrote:
>>>
>>> We already add the start_pad to the resource->start but fails to section
>>> align the start. This make sure with altmap we compute the right first
>>> pfn when start_pad is zero and we are doing an align down of start address.
>>>
>>> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
>>> ---
>>>   kernel/memremap.c | 4 ++--
>>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/kernel/memremap.c b/kernel/memremap.c
>>> index a856cb5ff192..23d77b60e728 100644
>>> --- a/kernel/memremap.c
>>> +++ b/kernel/memremap.c
>>> @@ -59,9 +59,9 @@ static unsigned long pfn_first(struct dev_pagemap *pgmap)
>>>   {
>>>          const struct resource *res = &pgmap->res;
>>>          struct vmem_altmap *altmap = &pgmap->altmap;
>>> -       unsigned long pfn;
>>> +       unsigned long pfn = PHYS_PFN(res->start);
>>>
>>> -       pfn = res->start >> PAGE_SHIFT;
>>> +       pfn = SECTION_ALIGN_DOWN(pfn);
>> 
>> This does not seem right to me it breaks the assumptions of where the
>> first expected valid pfn occurs in the passed in range.
>> 
>
> How do we define the first valid pfn? Isn't that at pfn_sb->dataoff ?

for altmap the pfn_first should be

pfn_first = altmap->base_pfn + vmem_altmap_offset(altmap);

?

-aneesh