[Intel-gfx] [PATCH 3/3] drm/i915: Implement 16GB dimm wa for latency level-0
Kumar, Mahesh
mahesh1.kumar at intel.com
Wed Jul 18 12:34:13 UTC 2018
Hi,
On 7/18/2018 5:34 PM, Ville Syrjälä wrote:
> On Sat, Jul 14, 2018 at 07:40:43PM +0530, Kumar, Mahesh wrote:
>> Hi,
>>
>> Thanks for review.
>>
>>
>> On 7/13/2018 8:27 PM, Ville Syrjälä wrote:
>>> On Fri, Jul 13, 2018 at 07:41:24PM +0530, Mahesh Kumar wrote:
>>>> Memory with 16GB dimms require an increase of 1us in level-0 latency.
>>>> This patch implements the same.
>>>> Bspec: 4381
>>>>
>>>> Signed-off-by: Mahesh Kumar <mahesh1.kumar at intel.com>
>>>> ---
>>>> drivers/gpu/drm/i915/i915_drv.c | 35 +++++++++++++++++++++++++++++++++--
>>>> drivers/gpu/drm/i915/i915_drv.h | 2 ++
>>>> drivers/gpu/drm/i915/intel_pm.c | 13 +++++++++++++
>>>> 3 files changed, 48 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
>>>> index b93194cbd820..c6d30653d70c 100644
>>>> --- a/drivers/gpu/drm/i915/i915_drv.c
>>>> +++ b/drivers/gpu/drm/i915/i915_drv.c
>>>> @@ -1063,6 +1063,29 @@ static void intel_sanitize_options(struct drm_i915_private *dev_priv)
>>>> intel_gvt_sanitize_options(dev_priv);
>>>> }
>>>>
>>>> +static void
>>>> +intel_memdev_is_16gb_dimm(struct memdev_info *memdev_info,
>>>> + u8 rank, u8 size, u8 width)
>>>> +{
>>>> + bool found_16gb_dimm = false;
>>>> +
>>>> + if (size == 16 && width == SKL_DRAM_WIDTH_X8 &&
>>>> + rank == SKL_DRAM_RANK_SINGLE)
>>>> + found_16gb_dimm = true;
>>>> + else if (size == 32 && width == SKL_DRAM_WIDTH_X8 &&
>>>> + rank == SKL_DRAM_RANK_DUAL)
>>>> + found_16gb_dimm = true;
>>>> + else if (size == 8 && width == SKL_DRAM_WIDTH_X16 &&
>>>> + rank == SKL_DRAM_RANK_SINGLE)
>>>> + found_16gb_dimm = true;
>>>> + else if (size == 16 && width == SKL_DRAM_WIDTH_X16 &&
>>>> + rank == SKL_DRAM_RANK_DUAL)
>>>> + found_16gb_dimm = true;
>>>> +
>>>> + if (!memdev_info->is_16gb_dimm)
>>>> + memdev_info->is_16gb_dimm = found_16gb_dimm;
>>>> +}
>>> Pure functions are generally more fun. Would also make the function name
>>> match the implementation.
>> Are you suggesting to change the name to only "is_16gb_dimm" ?
> I'm suggesting making it pure.
ok sure, got it.
-Mahesh
>
More information about the Intel-gfx
mailing list