2015-02-24 19:54:19

by Danesh Petigara

Subject: [PATCH] mm: cma: fix CMA aligned offset calculation

The CMA aligned offset calculation is incorrect for
non-zero order_per_bit values.

For example, if cma->order_per_bit=1, cma->base_pfn=
0x2f800000 and align_order=12, the function returns
a value of 0x17c00 instead of 0x400.

This patch fixes the CMA aligned offset calculation.

Cc: [email protected]
Signed-off-by: Danesh Petigara <[email protected]>
Reviewed-by: Gregory Fong <[email protected]>
---
mm/cma.c | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/mm/cma.c b/mm/cma.c
index 75016fd..58f37bd 100644
--- a/mm/cma.c
+++ b/mm/cma.c
@@ -70,9 +70,13 @@ static unsigned long cma_bitmap_aligned_offset(struct cma *cma, int align_order)

if (align_order <= cma->order_per_bit)
return 0;
- alignment = 1UL << (align_order - cma->order_per_bit);
- return ALIGN(cma->base_pfn, alignment) -
- (cma->base_pfn >> cma->order_per_bit);
+
+ /*
+ * Find a PFN aligned to the specified order and return
+ * an offset represented in order_per_bits.
+ */
+ return (ALIGN(cma->base_pfn, (1UL << align_order))
+ - cma->base_pfn) >> cma->order_per_bit;
}

static unsigned long cma_bitmap_maxno(struct cma *cma)
--
1.9.1

2015-02-24 20:55:11

by Gregory Fong

Subject: Re: [PATCH] mm: cma: fix CMA aligned offset calculation

[removed stable from cc]

On Tue, Feb 24, 2015 at 11:55 AM, Danesh Petigara
<[email protected]> wrote:
> diff --git a/mm/cma.c b/mm/cma.c
> index 75016fd..58f37bd 100644
> --- a/mm/cma.c
> +++ b/mm/cma.c
> @@ -70,9 +70,13 @@ static unsigned long cma_bitmap_aligned_offset(struct cma *cma, int align_order)
>
> if (align_order <= cma->order_per_bit)
> return 0;
> - alignment = 1UL << (align_order - cma->order_per_bit);
> - return ALIGN(cma->base_pfn, alignment) -
> - (cma->base_pfn >> cma->order_per_bit);
> +
> + /*
> + * Find a PFN aligned to the specified order and return
> + * an offset represented in order_per_bits.
> + */
> + return (ALIGN(cma->base_pfn, (1UL << align_order))
> + - cma->base_pfn) >> cma->order_per_bit;

It was noticed that this doesn't remove the now-unused 'alignment'
variable, so I think Danesh is planning to submit an updated patch.

Best regards,
Gregory

2015-02-24 21:10:36

by Michal Nazarewicz

Subject: Re: [PATCH] mm: cma: fix CMA aligned offset calculation

On Tue, Feb 24 2015, Danesh Petigara <[email protected]> wrote:
> The CMA aligned offset calculation is incorrect for
> non-zero order_per_bit values.
>
> For example, if cma->order_per_bit=1, cma->base_pfn=
> 0x2f800000 and align_order=12, the function returns
> a value of 0x17c00 instead of 0x400.
>
> This patch fixes the CMA aligned offset calculation.
>
> Cc: [email protected]
> Signed-off-by: Danesh Petigara <[email protected]>
> Reviewed-by: Gregory Fong <[email protected]>

Acked-by: Michal Nazarewicz <[email protected]>

> ---
> mm/cma.c | 10 +++++++---
> 1 file changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/mm/cma.c b/mm/cma.c
> index 75016fd..58f37bd 100644
> --- a/mm/cma.c
> +++ b/mm/cma.c
> @@ -70,9 +70,13 @@ static unsigned long cma_bitmap_aligned_offset(struct cma *cma, int align_order)
>
> if (align_order <= cma->order_per_bit)
> return 0;
> - alignment = 1UL << (align_order - cma->order_per_bit);
> - return ALIGN(cma->base_pfn, alignment) -
> - (cma->base_pfn >> cma->order_per_bit);
> +
> + /*
> + * Find a PFN aligned to the specified order and return
> + * an offset represented in order_per_bits.
> + */

It probably makes sense to move this comment outside of the function as
function documentation.

> + return (ALIGN(cma->base_pfn, (1UL << align_order))
> + - cma->base_pfn) >> cma->order_per_bit;
> }
>
> static unsigned long cma_bitmap_maxno(struct cma *cma)
> --
> 1.9.1
>

--
Best regards, _ _
.o. | Liege of Serenely Enlightened Majesty of o' \,=./ `o
..o | Computer Science, Michał “mina86” Nazarewicz (o o)
ooo +--<[email protected]>--<xmpp:[email protected]>--ooO--(_)--Ooo--

2015-02-24 21:43:54

by Danesh Petigara

Subject: Re: [PATCH] mm: cma: fix CMA aligned offset calculation

On 2/24/2015 1:10 PM, Michal Nazarewicz wrote:
> On Tue, Feb 24 2015, Danesh Petigara <[email protected]> wrote:
>> The CMA aligned offset calculation is incorrect for
>> non-zero order_per_bit values.
>>
>> For example, if cma->order_per_bit=1, cma->base_pfn=
>> 0x2f800000 and align_order=12, the function returns
>> a value of 0x17c00 instead of 0x400.
>>
>> This patch fixes the CMA aligned offset calculation.
>>
>> Cc: [email protected]
>> Signed-off-by: Danesh Petigara <[email protected]>
>> Reviewed-by: Gregory Fong <[email protected]>
>
> Acked-by: Michal Nazarewicz <[email protected]>
>
>> ---
>> mm/cma.c | 10 +++++++---
>> 1 file changed, 7 insertions(+), 3 deletions(-)
>>
>> diff --git a/mm/cma.c b/mm/cma.c
>> index 75016fd..58f37bd 100644
>> --- a/mm/cma.c
>> +++ b/mm/cma.c
>> @@ -70,9 +70,13 @@ static unsigned long cma_bitmap_aligned_offset(struct cma *cma, int align_order)
>>
>> if (align_order <= cma->order_per_bit)
>> return 0;
>> - alignment = 1UL << (align_order - cma->order_per_bit);
>> - return ALIGN(cma->base_pfn, alignment) -
>> - (cma->base_pfn >> cma->order_per_bit);
>> +
>> + /*
>> + * Find a PFN aligned to the specified order and return
>> + * an offset represented in order_per_bits.
>> + */
>
> It probably makes sense to move this comment outside of the function as
> function documentation.
>

Thanks for the feedback. Will send out patch v2 with the comment moved
outside the function and also remove the unused 'alignment' variable.

>> + return (ALIGN(cma->base_pfn, (1UL << align_order))
>> + - cma->base_pfn) >> cma->order_per_bit;
>> }
>>
>> static unsigned long cma_bitmap_maxno(struct cma *cma)
>> --
>> 1.9.1
>>
>

Best regards,
Danesh Petigara