public inbox for intel-gfx@lists.freedesktop.org
 help / color / mirror / Atom feed
* [PATCH] fixing eu_mask value for BXT due to wrong bit mapping
@ 2015-09-11 22:42 Dongwon Kim
  2015-09-11 23:31 ` [PATCH] drm/i915: fixing eu_mask value for BXT due to wrong bit mapping (v2) Dongwon Kim
  0 siblings, 1 reply; 3+ messages in thread
From: Dongwon Kim @ 2015-09-11 22:42 UTC (permalink / raw)
  To: intel-gfx; +Cc: dw kim

From: dw kim <dongwon.kim@intel.com>

Correct bit mapping of EUs in 'eu_disable' fuse register
for each subslice is (each word represents one subslice),

bit # 7   6   5   4   3   2   1   0
EU # 11  10   9   8   3   2   1   0

In BXT, each subslice has 6 EUs at most, which are EU0~EU2
and EU8~EU10 (EU11 and EU3 don't exist). Therefore, correct
eu_mask should be 0x77 to cover all EUs in each subslice that
can be possibly enabled.

Cc: Matthew D Roper <matthew.d.roper@intel.com>
Signed-off-by: dw kim <dongwon.kim@intel.com>
---
 drivers/gpu/drm/i915/i915_dma.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 066a0ef..810f148 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -633,13 +633,21 @@ static void gen9_sseu_info_init(struct drm_device *dev)
 
 	/*
 	 * BXT has a single slice. BXT also has at most 6 EU per subslice,
-	 * and therefore only the lowest 6 bits of the 8-bit EU disable
-	 * fields are valid.
+	 * and bit mappings vs EU # in eu_disable fuse value in one subslice
+         * is like this;
+         *
+         *  Bit # 7   6   5   4   3   2   1   0
+         *  EU # 11  10   9   8   3   2   1   0
+         *
+         * Since only EUs with number "0, 1, 2, 8, 9, 10" are available in
+         * each subslice in BXT, bits in eu_mask should also be set according
+         * to this mapping, therefore, eu_mask = 0x77.
 	*/
+
 	if (IS_BROXTON(dev)) {
 		s_max = 1;
 		eu_max = 6;
-		eu_mask = 0x3f;
+		eu_mask = 0x77;
 	}
 
 	info = (struct intel_device_info *)&dev_priv->info;
-- 
1.9.1

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* [PATCH] drm/i915: fixing eu_mask value for BXT due to wrong bit mapping (v2)
  2015-09-11 22:42 [PATCH] fixing eu_mask value for BXT due to wrong bit mapping Dongwon Kim
@ 2015-09-11 23:31 ` Dongwon Kim
  2015-09-14 11:06   ` Imre Deak
  0 siblings, 1 reply; 3+ messages in thread
From: Dongwon Kim @ 2015-09-11 23:31 UTC (permalink / raw)
  To: intel-gfx; +Cc: dw kim

From: dw kim <dongwon.kim@intel.com>

Correct bit mapping of EUs in 'eu_disable' fuse register
for each subslice is (each word represents one subslice),

bit # 7   6   5   4   3   2   1   0
EU # 11  10   9   8   3   2   1   0

In BXT, each subslice has 6 EUs at most, which are EU0~EU2
and EU8~EU10 (EU11 and EU3 don't exist). Therefore, correct
eu_mask should be 0x77 to cover all EUs in each subslice that
can be possibly enabled.

v2: fixed some whitespaces and added "drm/i915:" to the title

Cc: Matthew D Roper <matthew.d.roper@intel.com>
Signed-off-by: dw kim <dongwon.kim@intel.com>
---
 drivers/gpu/drm/i915/i915_dma.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 066a0ef..3f34dd3 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -633,13 +633,21 @@ static void gen9_sseu_info_init(struct drm_device *dev)
 
 	/*
 	 * BXT has a single slice. BXT also has at most 6 EU per subslice,
-	 * and therefore only the lowest 6 bits of the 8-bit EU disable
-	 * fields are valid.
-	*/
+	 * and bit mappings vs EU # in eu_disable fuse value in one subslice
+	 * is like this;
+	 *
+	 *  Bit # 7   6   5   4   3   2   1   0
+	 *  EU # 11  10   9   8   3   2   1   0
+	 *
+	 * Since only EUs with number "0, 1, 2, 8, 9, 10" are available in
+	 * each subslice in BXT, bits in eu_mask should also be set according
+	 * to this mapping, therefore, eu_mask = 0x77.
+	 */
+
 	if (IS_BROXTON(dev)) {
 		s_max = 1;
 		eu_max = 6;
-		eu_mask = 0x3f;
+		eu_mask = 0x77;
 	}
 
 	info = (struct intel_device_info *)&dev_priv->info;
-- 
1.9.1

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH] drm/i915: fixing eu_mask value for BXT due to wrong bit mapping (v2)
  2015-09-11 23:31 ` [PATCH] drm/i915: fixing eu_mask value for BXT due to wrong bit mapping (v2) Dongwon Kim
@ 2015-09-14 11:06   ` Imre Deak
  0 siblings, 0 replies; 3+ messages in thread
From: Imre Deak @ 2015-09-14 11:06 UTC (permalink / raw)
  To: Dongwon Kim; +Cc: intel-gfx

On pe, 2015-09-11 at 16:31 -0700, Dongwon Kim wrote:
> From: dw kim <dongwon.kim@intel.com>
> 
> Correct bit mapping of EUs in 'eu_disable' fuse register
> for each subslice is (each word represents one subslice),
> 
> bit # 7   6   5   4   3   2   1   0
> EU # 11  10   9   8   3   2   1   0
> 
> In BXT, each subslice has 6 EUs at most, which are EU0~EU2
> and EU8~EU10 (EU11 and EU3 don't exist). Therefore, correct
> eu_mask should be 0x77 to cover all EUs in each subslice that
> can be possibly enabled.
> 
> v2: fixed some whitespaces and added "drm/i915:" to the title
> 
> Cc: Matthew D Roper <matthew.d.roper@intel.com>
> Signed-off-by: dw kim <dongwon.kim@intel.com>
> ---
>  drivers/gpu/drm/i915/i915_dma.c | 16 ++++++++++++----
>  1 file changed, 12 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
> index 066a0ef..3f34dd3 100644
> --- a/drivers/gpu/drm/i915/i915_dma.c
> +++ b/drivers/gpu/drm/i915/i915_dma.c
> @@ -633,13 +633,21 @@ static void gen9_sseu_info_init(struct drm_device *dev)
>  
>  	/*
>  	 * BXT has a single slice. BXT also has at most 6 EU per subslice,
> -	 * and therefore only the lowest 6 bits of the 8-bit EU disable
> -	 * fields are valid.
> -	*/
> +	 * and bit mappings vs EU # in eu_disable fuse value in one subslice
> +	 * is like this;
> +	 *
> +	 *  Bit # 7   6   5   4   3   2   1   0
> +	 *  EU # 11  10   9   8   3   2   1   0
> +	 *
> +	 * Since only EUs with number "0, 1, 2, 8, 9, 10" are available in
> +	 * each subslice in BXT, bits in eu_mask should also be set according
> +	 * to this mapping, therefore, eu_mask = 0x77.
> +	 */
> +
>  	if (IS_BROXTON(dev)) {
>  		s_max = 1;
>  		eu_max = 6;
> -		eu_mask = 0x3f;
> +		eu_mask = 0x77;
>  	}

Yes, this matches bspec (and is a review fail on my side). I would
prefer though fixing it by simply removing the above BXT specific
max/mask values. Both GEN8_FUSE2 and GEN9_EU_DISABLE(0) contain the
proper values for each platform/SKU, so that the generic parsing will
work everywhere without the need to differentiate between platforms.

>  
>  	info = (struct intel_device_info *)&dev_priv->info;


_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2015-09-14 11:06 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-09-11 22:42 [PATCH] fixing eu_mask value for BXT due to wrong bit mapping Dongwon Kim
2015-09-11 23:31 ` [PATCH] drm/i915: fixing eu_mask value for BXT due to wrong bit mapping (v2) Dongwon Kim
2015-09-14 11:06   ` Imre Deak

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox