[Intel-gfx] [PATCH 1/2] iosys-map: Add a helper for pointer difference

Siva Mullati siva.mullati at intel.com
Mon Mar 14 05:40:56 UTC 2022


On 09/03/22 06:43, Lucas De Marchi wrote:
> On Tue, Mar 08, 2022 at 03:08:04PM +0530, Mullati Siva wrote:
>> From: Siva Mullati <siva.mullati at intel.com>
>>
>> iosys_map_ptrdiff to get the difference in address of
>> same memory type.
>>
>> Signed-off-by: Siva Mullati <siva.mullati at intel.com>
>> ---
>> include/linux/iosys-map.h | 21 +++++++++++++++++++++
>> 1 file changed, 21 insertions(+)
>>
>> diff --git a/include/linux/iosys-map.h b/include/linux/iosys-map.h
>> index e69a002d5aa4..8987f69ec1e9 100644
>> --- a/include/linux/iosys-map.h
>> +++ b/include/linux/iosys-map.h
>> @@ -8,6 +8,7 @@
>>
>> #include <linux/io.h>
>> #include <linux/string.h>
>> +#include <linux/types.h>
>>
>> /**
>>  * DOC: overview
>> @@ -208,6 +209,26 @@ static inline bool iosys_map_is_equal(const struct iosys_map *lhs,
>>         return lhs->vaddr == rhs->vaddr;
>> }
>>
>> +/**
>> + * iosys_map_ptrdiff - Difference of two iosys mapping addresses of same memory type
>> + * @lhs:       The iosys_map structure
>> + * @rhs:       A iosys_map structure to compare with
>> + *
>> + * Two iosys mapping structures of same memory type with the differences
>> + * in address within that memory.
>> + *
>> + * Returns:
>> + * Address difference of two memory locations with same memory type.
>> + */
>> +static inline ptrdiff_t iosys_map_ptrdiff(const struct iosys_map *lhs,
>> +                      const struct iosys_map *rhs)
>> +{
>> +    if (lhs->is_iomem)
>
> the other interfaces always check both arguments to make sure they are
> both iomem. So if we want this interface we will want to check like
> that.
>
Agreed, will use the other interface.
> I'm not sure this is really needed, but will depend on the secon patch,
> let's wait to settle this discussion there.
>
Since you have already covered the other patch, intel_guc_ct_enable() where this API has been used for calculating offset for cmds and desc. Do you this as good approach or is okay to directly access the ctb.send.cmds_map.vaddr?

>
> thanks
> Lucas De Marchi


More information about the Intel-gfx mailing list