Skip to contentSkip to conversation

Begin enabling Xe_HP SDV and DG2 platforms

PATCH Superseded Matt Roper 5 participants 33 messages Jul 29, 2021 – Aug 5, 2021

Subsystems (4)

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Matt Roper

author
#0 lore raw

v4 cover Begin enabling Xe_HP SDV and DG2 platforms

This series provides some of the initial enablement patches for two upcoming discrete GPUs:
* XeHP SDV: Xe_HP (version 12.50) graphics IP, no display IP
* DG2: Xe_HPG (version 12.55) graphics IP, Xe_LPD (version 13) display IP

Both platforms will need additional enablement patches beyond what's present in this series before they're truly usable, including various LMEM and GuC work that's already happening separately. The new features/functionality that these platforms bring (such as multi-tile support, dedicated compute engines, etc.) may be referenced in passing in some of these patches but will be fully enabled in future series.

v2:
- General rebase and incorporation of r-b's.
- Re-order intel_gt_info and intel_device_info structures to eliminate some unnecessary padding after the size change of intel_engine_mask_t. (Tvrtko)
- Use 'intel_step' mechanisms for revid->stepping mapping. (Jani)
- Drop the DSC patches for now; they need some rework. (Jani)

v3:
- About 20 of the patches have landed upstream now. Rebase and resend the rest. Some of these are already reviewed, but have dependencies on other unreviewed patches (e.g., the new engine definitions, the initial SNPS PHY support, etc.).

v4:
- Several more patches have landed upstream; rebase and re-send the rest. Some of the remaining patches are reviewed but still have dependencies on non-reviewed patches, so the order is shuffled this time to group patches by dependency rather than by xehp vs xehpsdv vs dg2.
- Minor cleanup to "drm/i915/xehp: handle new steering options" suggested by Caz.

Cc Rodrigo Vivi <[email protected]>
Cc Lucas De Marchi <[email protected]>
Cc James Ausmus <[email protected]>

Akeem G Abodunrin (1):
drm/i915/dg2: Add new LRI reg offsets

Ankit Nautiyal (1):
drm/i915/dg2: Configure PCON in DP pre-enable path

Daniele Ceraolo Spurio (1):
drm/i915/xehp: handle new steering options

Lucas De Marchi (2):
drm/i915/xehpsdv: Define MOCS table for XeHP SDV drm/i915/xehpsdv: factor out function to read RP_STATE_CAP

Matt Roper (11):
drm/i915/xehpsdv: Define steering tables drm/i915/dg2: Add forcewake table
drm/i915/dg2: Update LNCF steering ranges drm/i915/dg2: Add SQIDI steering
drm/i915/xehp: Loop over all gslices for INSTDONE processing drm/i915/dg2: Report INSTDONE_GEOM values in error state drm/i915/xehpsdv: Add maximum sseu limits drm/i915/dg2: DG2 uses the same sseu limits as XeHP SDV drm/i915/dg2: Define MOCS table for DG2 drm/i915/xehpsdv: Read correct RP_STATE_CAP register drm/i915/dg2: Maintain backward-compatible nested batch behavior

Matthew Auld (1):
drm/i915/xehp: Changes to ss/eu definitions

Stuart Summers (1):
drm/i915/xehpsdv: Add compute DSS type

 drivers/gpu/drm/i915/display/intel_ddi.c     |   3 +
 drivers/gpu/drm/i915/gt/debugfs_gt_pm.c      |   8 +-
 drivers/gpu/drm/i915/gt/intel_engine_cs.c    |  55 ++--
 drivers/gpu/drm/i915/gt/intel_engine_types.h |  15 +-
 drivers/gpu/drm/i915/gt/intel_gt.c           |  68 ++++-
 drivers/gpu/drm/i915/gt/intel_gt_types.h     |   7 +
 drivers/gpu/drm/i915/gt/intel_lrc.c          |  85 +++++-
 drivers/gpu/drm/i915/gt/intel_mocs.c         |  66 +++-
 drivers/gpu/drm/i915/gt/intel_region_lmem.c  |   1 +
 drivers/gpu/drm/i915/gt/intel_rps.c          |  19 +-
 drivers/gpu/drm/i915/gt/intel_rps.h          |   1 +
 drivers/gpu/drm/i915/gt/intel_sseu.c         | 116 +++++--
 drivers/gpu/drm/i915/gt/intel_sseu.h         |  20 +-
 drivers/gpu/drm/i915/gt/intel_sseu_debugfs.c |   2 +-
 drivers/gpu/drm/i915/gt/intel_workarounds.c  | 155 +++++++++-
 drivers/gpu/drm/i915/i915_debugfs.c          |   8 +-
 drivers/gpu/drm/i915/i915_drv.h              |   3 +
 drivers/gpu/drm/i915/i915_getparam.c         |   6 +-
 drivers/gpu/drm/i915/i915_gpu_error.c        |  36 ++-
 drivers/gpu/drm/i915/i915_pci.c              |   1 +
 drivers/gpu/drm/i915/i915_reg.h              |  15 +-
 drivers/gpu/drm/i915/intel_device_info.h     |   1 +
 drivers/gpu/drm/i915/intel_uncore.c          | 305 ++++++++++---------
 include/uapi/drm/i915_drm.h                  |   3 -
 24 files changed, 771 insertions(+), 228 deletions(-)
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#1 lore raw

PATCH v4 1/18 drm/i915/xehp: handle new steering options

From: Daniele Ceraolo Spurio <[email protected]>

Xe_HP is more modular than its predecessors and as a consequence it has more types of replicated registers. As with l3bank regions on previous platforms, we may need to explicitly re-steer accesses to these new types of ranges at runtime if we can't find a single default steering value that satisfies the fusing of all types.

v2:
- Add a local 'i915' variable to reduce gt->i915 usage. (Caz)
- Drop unused 'intel_gt_read_register' prototype. (Caz)

Bspec: 66534

---
 drivers/gpu/drm/i915/gt/intel_gt.c          | 42 +++++++++-
 drivers/gpu/drm/i915/gt/intel_gt_types.h    |  7 ++
 drivers/gpu/drm/i915/gt/intel_region_lmem.c |  1 +
 drivers/gpu/drm/i915/gt/intel_sseu.c        | 18 +++++
 drivers/gpu/drm/i915/gt/intel_sseu.h        |  6 ++
 drivers/gpu/drm/i915/gt/intel_workarounds.c | 89 +++++++++++++++++++--
 drivers/gpu/drm/i915/i915_drv.h             |  3 +
 drivers/gpu/drm/i915/i915_pci.c             |  1 +
 drivers/gpu/drm/i915/i915_reg.h             |  4 +
 drivers/gpu/drm/i915/intel_device_info.h    |  1 +
 10 files changed, 166 insertions(+), 6 deletions(-)
drivers/gpu/drm/i915/gt/intel_gt.c +41 -1
index a64aa43f7cd9..39b224600f38 100644
@@ -89,18 +89,40 @@ static const struct intel_mmio_range icl_l3bank_steering_table[] = {
 	{},
 };
 
+static u16 slicemask(struct intel_gt *gt, int count)
+{
+	u64 dss_mask = intel_sseu_get_subslices(&gt->info.sseu, 0);
+
+	return intel_slicemask_from_dssmask(dss_mask, count);
+}
+
 int intel_gt_init_mmio(struct intel_gt *gt)
 {
+	struct drm_i915_private *i915 = gt->i915;
+
 	intel_gt_init_clock_frequency(gt);
 
 	intel_uc_init_mmio(&gt->uc);
 	intel_sseu_info_init(gt);
 
-	if (GRAPHICS_VER(gt->i915) >= 11) {
+	/*
+	 * An mslice is unavailable only if both the meml3 for the slice is
+	 * disabled *and* all of the DSS in the slice (quadrant) are disabled.
+	 */
+	if (HAS_MSLICES(i915))
+		gt->info.mslice_mask =
+			slicemask(gt, GEN_DSS_PER_MSLICE) |
+			(intel_uncore_read(gt->uncore, GEN10_MIRROR_FUSE3) &
+			 GEN12_MEML3_EN_MASK);
+
+	if (GRAPHICS_VER(i915) >= 11 &&
+		   GRAPHICS_VER_FULL(i915) < IP_VER(12, 50)) {
 		gt->steering_table[L3BANK] = icl_l3bank_steering_table;
 		gt->info.l3bank_mask =
 			~intel_uncore_read(gt->uncore, GEN10_MIRROR_FUSE3) &
 			GEN10_L3BANK_MASK;
+	} else if (HAS_MSLICES(i915)) {
+		MISSING_CASE(INTEL_INFO(i915)->platform);
 	}
 
 	return intel_engines_init_mmio(gt);
@@ -787,6 +809,24 @@ static void intel_gt_get_valid_steering(struct intel_gt *gt,
 		*sliceid = 0;		/* unused */
 		*subsliceid = __ffs(gt->info.l3bank_mask);
 		break;
+	case MSLICE:
+		GEM_DEBUG_WARN_ON(!gt->info.mslice_mask); /* should be impossible! */
+
+		*sliceid = __ffs(gt->info.mslice_mask);
+		*subsliceid = 0;	/* unused */
+		break;
+	case LNCF:
+		GEM_DEBUG_WARN_ON(!gt->info.mslice_mask); /* should be impossible! */
+
+		/*
+		 * 0xFDC[29:28] selects the mslice to steer to and 0xFDC[27]
+		 * selects which LNCF within the mslice to steer to.  An LNCF
+		 * is always present if its mslice is present, so we can safely
+		 * just steer to LNCF 0 in all cases.
+		 */
+		*sliceid = __ffs(gt->info.mslice_mask) << 1;
+		*subsliceid = 0;	/* unused */
+		break;
 	default:
 		MISSING_CASE(type);
 		*sliceid = 0;
drivers/gpu/drm/i915/gt/intel_gt_types.h +7 -0
index 97a5075288d2..a81e21bf1bd1 100644
@@ -47,9 +47,14 @@ struct intel_mmio_range {
  * of multicast registers.  If another type of steering does not have any
  * overlap in valid steering targets with 'subslice' style registers, we will
  * need to explicitly re-steer reads of registers of the other type.
+ *
+ * Only the replication types that may need additional non-default steering
+ * are listed here.
  */
 enum intel_steering_type {
 	L3BANK,
+	MSLICE,
+	LNCF,
 
 	NUM_STEERING_TYPES
 };
@@ -184,6 +189,8 @@ struct intel_gt {
 
 		/* Slice/subslice/EU info */
 		struct sseu_dev_info sseu;
+
+		unsigned long mslice_mask;
 	} info;
 };
 
drivers/gpu/drm/i915/gt/intel_region_lmem.c +1 -0
index e3a2a2fa5f94..a74b72f50cc9 100644
@@ -10,6 +10,7 @@
 #include "gem/i915_gem_lmem.h"
 #include "gem/i915_gem_region.h"
 #include "gem/i915_gem_ttm.h"
+#include "gt/intel_gt.h"
 
 static int init_fake_lmem_bar(struct intel_memory_region *mem)
 {
drivers/gpu/drm/i915/gt/intel_sseu.c +18 -0
index 367fd44b81c8..bbed8e8625e1 100644
@@ -759,3 +759,21 @@ void intel_sseu_print_topology(const struct sseu_dev_info *sseu,
 		}
 	}
 }
+
+u16 intel_slicemask_from_dssmask(u64 dss_mask, int dss_per_slice)
+{
+	u16 slice_mask = 0;
+	int i;
+
+	WARN_ON(sizeof(dss_mask) * 8 / dss_per_slice > 8 * sizeof(slice_mask));
+
+	for (i = 0; dss_mask; i++) {
+		if (dss_mask & GENMASK(dss_per_slice - 1, 0))
+			slice_mask |= BIT(i);
+
+		dss_mask >>= dss_per_slice;
+	}
+
+	return slice_mask;
+}
+
drivers/gpu/drm/i915/gt/intel_sseu.h +6 -0
index 4cd1a8a7298a..1073471d1980 100644
@@ -22,6 +22,10 @@ struct drm_printer;
 #define GEN_MAX_EUS		(16) /* TGL upper bound */
 #define GEN_MAX_EU_STRIDE GEN_SSEU_STRIDE(GEN_MAX_EUS)
 
+#define GEN_DSS_PER_GSLICE	4
+#define GEN_DSS_PER_CSLICE	8
+#define GEN_DSS_PER_MSLICE	8
+
 struct sseu_dev_info {
 	u8 slice_mask;
 	u8 subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
@@ -104,4 +108,6 @@ void intel_sseu_dump(const struct sseu_dev_info *sseu, struct drm_printer *p);
 void intel_sseu_print_topology(const struct sseu_dev_info *sseu,
 			       struct drm_printer *p);
 
+u16 intel_slicemask_from_dssmask(u64 dss_mask, int dss_per_slice);
+
 #endif /* __INTEL_SSEU_H__ */
drivers/gpu/drm/i915/gt/intel_workarounds.c +84 -5

108 lines not on this page · raw message

drivers/gpu/drm/i915/i915_drv.h +3 -0
index 70bef688c6da..31b8bb43c581 100644
@@ -1694,6 +1694,9 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915,
 #define HAS_RUNTIME_PM(dev_priv) (INTEL_INFO(dev_priv)->has_runtime_pm)
 #define HAS_64BIT_RELOC(dev_priv) (INTEL_INFO(dev_priv)->has_64bit_reloc)
 
+#define HAS_MSLICES(dev_priv) \
+	(INTEL_INFO(dev_priv)->has_mslices)
+
 #define HAS_IPC(dev_priv)		 (INTEL_INFO(dev_priv)->display.has_ipc)
 
 #define HAS_REGION(i915, i) (INTEL_INFO(i915)->memory_regions & (i))
drivers/gpu/drm/i915/i915_pci.c +1 -0
index 08651ca03478..ed9edda727d2 100644
@@ -1007,6 +1007,7 @@ static const struct intel_device_info adl_p_info = {
 	.has_llc = 1, \
 	.has_logical_ring_contexts = 1, \
 	.has_logical_ring_elsq = 1, \
+	.has_mslices = 1, \
 	.has_rc6 = 1, \
 	.has_reset_engine = 1, \
 	.has_rps = 1, \
drivers/gpu/drm/i915/i915_reg.h +4 -0
index 1a164b90e9fc..f4113e7e8271 100644
@@ -2766,6 +2766,7 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define   GEN11_MCR_SLICE_MASK		GEN11_MCR_SLICE(0xf)
 #define   GEN11_MCR_SUBSLICE(subslice)	(((subslice) & 0x7) << 24)
 #define   GEN11_MCR_SUBSLICE_MASK	GEN11_MCR_SUBSLICE(0x7)
+#define   GEN11_MCR_MULTICAST		REG_BIT(31)
 #define RING_IPEIR(base)	_MMIO((base) + 0x64)
 #define RING_IPEHR(base)	_MMIO((base) + 0x68)
 #define RING_EIR(base)		_MMIO((base) + 0xb0)
@@ -3184,6 +3185,9 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define	GEN10_MIRROR_FUSE3		_MMIO(0x9118)
 #define GEN10_L3BANK_PAIR_COUNT     4
 #define GEN10_L3BANK_MASK   0x0F
+/* on Xe_HP the same fuses indicates mslices instead of L3 banks */
+#define GEN12_MAX_MSLICES 4
+#define GEN12_MEML3_EN_MASK 0x0F
 
 #define GEN8_EU_DISABLE0		_MMIO(0x9134)
 #define   GEN8_EU_DIS0_S0_MASK		0xffffff
drivers/gpu/drm/i915/intel_device_info.h +1 -0
index 121d6d9afd3a..2177372f9ac3 100644
@@ -133,6 +133,7 @@ enum intel_ppgtt_type {
 	func(has_llc); \
 	func(has_logical_ring_contexts); \
 	func(has_logical_ring_elsq); \
+	func(has_mslices); \
 	func(has_pooled_eu); \
 	func(has_rc6); \
 	func(has_rc6p); \
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v4 1/18
#2 lore raw

Re: [Intel-gfx] [PATCH v4 01/18] drm/i915/xehp: handle new steering options

Quoted 81 lines · Matt Roper

From: Daniele Ceraolo Spurio <[email protected]>

Xe_HP is more modular than its predecessors and as a consequence it has more types of replicated registers. As with l3bank regions on previous platforms, we may need to explicitly re-steer accesses to these new types of ranges at runtime if we can't find a single default steering value that satisfies the fusing of all types.

v2:
- Add a local 'i915' variable to reduce gt->i915 usage. (Caz)
- Drop unused 'intel_gt_read_register' prototype. (Caz)

Bspec: 66534

---

drivers/gpu/drm/i915/gt/intel_gt.c          | 42 +++++++++-
drivers/gpu/drm/i915/gt/intel_gt_types.h    |  7 ++
drivers/gpu/drm/i915/gt/intel_region_lmem.c |  1 +
drivers/gpu/drm/i915/gt/intel_sseu.c        | 18 +++++
drivers/gpu/drm/i915/gt/intel_sseu.h        |  6 ++
drivers/gpu/drm/i915/gt/intel_workarounds.c | 89 +++++++++++++++++++--
drivers/gpu/drm/i915/i915_drv.h             |  3 +
drivers/gpu/drm/i915/i915_pci.c             |  1 +
drivers/gpu/drm/i915/i915_reg.h             |  4 +
drivers/gpu/drm/i915/intel_device_info.h    |  1 +
10 files changed, 166 insertions(+), 6 deletions(-)
drivers/gpu/drm/i915/gt/intel_gt.c +24 -1
index a64aa43f7cd9..39b224600f38 100644
@@ -89,18 +89,40 @@ static const struct intel_mmio_range icl_l3bank_steering_table[] = {
 	{},
 };
 
+static u16 slicemask(struct intel_gt *gt, int count)
+{
+	u64 dss_mask = intel_sseu_get_subslices(&gt->info.sseu, 0);
+
+	return intel_slicemask_from_dssmask(dss_mask, count);
+}
+
 int intel_gt_init_mmio(struct intel_gt *gt)
 {
+	struct drm_i915_private *i915 = gt->i915;
+
 	intel_gt_init_clock_frequency(gt);
 
 	intel_uc_init_mmio(&gt->uc);
 	intel_sseu_info_init(gt);
 
-	if (GRAPHICS_VER(gt->i915) >= 11) {
+	/*
+	 * An mslice is unavailable only if both the meml3 for the slice is
+	 * disabled *and* all of the DSS in the slice (quadrant) are disabled.
+	 */
+	if (HAS_MSLICES(i915))
+		gt->info.mslice_mask =
+			slicemask(gt, GEN_DSS_PER_MSLICE) |
+			(intel_uncore_read(gt->uncore, GEN10_MIRROR_FUSE3) &
+			 GEN12_MEML3_EN_MASK);
+
+	if (GRAPHICS_VER(i915) >= 11 &&
+		   GRAPHICS_VER_FULL(i915) < IP_VER(12, 50)) {
 		gt->steering_table[L3BANK] = icl_l3bank_steering_table;
 		gt->info.l3bank_mask =
 			~intel_uncore_read(gt->uncore, GEN10_MIRROR_FUSE3) &
 			GEN10_L3BANK_MASK;
+	} else if (HAS_MSLICES(i915)) {
+		MISSING_CASE(INTEL_INFO(i915)->platform);
 	}
 
 	return intel_engines_init_mmio(gt);
@@ -787,6 +809,24 @@ static void intel_gt_get_valid_steering(struct intel_gt *gt,
 		*sliceid = 0;		/* unused */
 		*subsliceid = __ffs(gt->info.l3bank_mask);
 		break;
+	case MSLICE:

drivers/gpu/drm/i915/gt/intel_gt.c static void intel_gt_get_valid_steering(struct intel_gt *gt, · line 822

Quoted 10 lines
drivers/gpu/drm/i915/gt/intel_gt.c +10 -0
+		GEM_DEBUG_WARN_ON(!gt->info.mslice_mask); /* should be impossible! */
+
+		*sliceid = __ffs(gt->info.mslice_mask);
+		*subsliceid = 0;	/* unused */
+		break;
+	case LNCF:
+		GEM_DEBUG_WARN_ON(!gt->info.mslice_mask); /* should be impossible! */
+
+		/*
+		 * 0xFDC[29:28] selects the mslice to steer to and 0xFDC[27]

not sure if this clarifies, since this wording does not come from bspec... bspec rather says (0, 1) -> LNCFS in MSLICE0, (2, 3) -> LNCFS in MSLICE1, etc. Yes, they are equivalent, but I'd be ok with just the second part (below) of this comment.... Unless we have this wording somewhere in bspec I'm not seeing.

Quoted 243 lines
drivers/gpu/drm/i915/gt/intel_gt.c +7 -0
+		 * selects which LNCF within the mslice to steer to.  An LNCF
+		 * is always present if its mslice is present, so we can safely
+		 * just steer to LNCF 0 in all cases.
+		 */
+		*sliceid = __ffs(gt->info.mslice_mask) << 1;
+		*subsliceid = 0;	/* unused */
+		break;
 	default:
 		MISSING_CASE(type);
 		*sliceid = 0;
drivers/gpu/drm/i915/gt/intel_gt_types.h +7 -0
index 97a5075288d2..a81e21bf1bd1 100644
@@ -47,9 +47,14 @@ struct intel_mmio_range {
 * of multicast registers.  If another type of steering does not have any
 * overlap in valid steering targets with 'subslice' style registers, we will
 * need to explicitly re-steer reads of registers of the other type.
+ *
+ * Only the replication types that may need additional non-default steering
+ * are listed here.
 */
 enum intel_steering_type {
 	L3BANK,
+	MSLICE,
+	LNCF,
 
 	NUM_STEERING_TYPES
 };
@@ -184,6 +189,8 @@ struct intel_gt {
 
 		/* Slice/subslice/EU info */
 		struct sseu_dev_info sseu;
+
+		unsigned long mslice_mask;
 	} info;
 };
 
drivers/gpu/drm/i915/gt/intel_region_lmem.c +1 -0
index e3a2a2fa5f94..a74b72f50cc9 100644
@@ -10,6 +10,7 @@
 #include "gem/i915_gem_lmem.h"
 #include "gem/i915_gem_region.h"
 #include "gem/i915_gem_ttm.h"
+#include "gt/intel_gt.h"
 
 static int init_fake_lmem_bar(struct intel_memory_region *mem)
 {
drivers/gpu/drm/i915/gt/intel_sseu.c +18 -0
index 367fd44b81c8..bbed8e8625e1 100644
@@ -759,3 +759,21 @@ void intel_sseu_print_topology(const struct sseu_dev_info *sseu,
 		}
 	}
 }
+
+u16 intel_slicemask_from_dssmask(u64 dss_mask, int dss_per_slice)
+{
+	u16 slice_mask = 0;
+	int i;
+
+	WARN_ON(sizeof(dss_mask) * 8 / dss_per_slice > 8 * sizeof(slice_mask));
+
+	for (i = 0; dss_mask; i++) {
+		if (dss_mask & GENMASK(dss_per_slice - 1, 0))
+			slice_mask |= BIT(i);
+
+		dss_mask >>= dss_per_slice;
+	}
+
+	return slice_mask;
+}
+
drivers/gpu/drm/i915/gt/intel_sseu.h +6 -0
index 4cd1a8a7298a..1073471d1980 100644
@@ -22,6 +22,10 @@ struct drm_printer;
 #define GEN_MAX_EUS		(16) /* TGL upper bound */
 #define GEN_MAX_EU_STRIDE GEN_SSEU_STRIDE(GEN_MAX_EUS)
 
+#define GEN_DSS_PER_GSLICE	4
+#define GEN_DSS_PER_CSLICE	8
+#define GEN_DSS_PER_MSLICE	8
+
 struct sseu_dev_info {
 	u8 slice_mask;
 	u8 subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
@@ -104,4 +108,6 @@ void intel_sseu_dump(const struct sseu_dev_info *sseu, struct drm_printer *p);
 void intel_sseu_print_topology(const struct sseu_dev_info *sseu,
 			       struct drm_printer *p);
 
+u16 intel_slicemask_from_dssmask(u64 dss_mask, int dss_per_slice);
+
 #endif /* __INTEL_SSEU_H__ */
drivers/gpu/drm/i915/gt/intel_workarounds.c +13 -0
index 9173df59821a..f2ceabb0e2ea 100644
@@ -889,12 +889,24 @@ cfl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 		    GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
 }
 
+static void __add_mcr_wa(struct drm_i915_private *i915, struct i915_wa_list *wal,
+			 unsigned slice, unsigned subslice)
+{
+	u32 mcr, mcr_mask;
+
+	mcr = GEN11_MCR_SLICE(slice) | GEN11_MCR_SUBSLICE(subslice);
+	mcr_mask = GEN11_MCR_SLICE_MASK | GEN11_MCR_SUBSLICE_MASK;
+
+	drm_dbg(&i915->drm, "MCR slice/subslice = %x\n", mcr);
+
+	wa_write_clr_set(wal, GEN8_MCR_SELECTOR, mcr_mask, mcr);
+}
+

static void
icl_wa_init_mcr(struct drm_i915_private *i915, struct i915_wa_list *wal)

{
	const struct sseu_dev_info *sseu = &i915->gt.info.sseu;
	unsigned int slice, subslice;
drivers/gpu/drm/i915/gt/intel_workarounds.c +0 -1
-	u32 mcr, mcr_mask;
	GEM_BUG_ON(GRAPHICS_VER(i915) < 11);
	GEM_BUG_ON(hweight8(sseu->slice_mask) > 1);
drivers/gpu/drm/i915/gt/intel_workarounds.c +71 -4
@@ -919,12 +931,79 @@ icl_wa_init_mcr(struct drm_i915_private *i915, struct i915_wa_list *wal)
 	if (i915->gt.info.l3bank_mask & BIT(subslice))
 		i915->gt.steering_table[L3BANK] = NULL;
 
-	mcr = GEN11_MCR_SLICE(slice) | GEN11_MCR_SUBSLICE(subslice);
-	mcr_mask = GEN11_MCR_SLICE_MASK | GEN11_MCR_SUBSLICE_MASK;
+	__add_mcr_wa(i915, wal, slice, subslice);
+}
 
-	drm_dbg(&i915->drm, "MCR slice/subslice = %x\n", mcr);
+__maybe_unused
+static void
+xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal)
+{
+	struct drm_i915_private *i915 = gt->i915;
+	const struct sseu_dev_info *sseu = &gt->info.sseu;
+	unsigned long slice, subslice = 0, slice_mask = 0;
+	u64 dss_mask = 0;
+	u32 lncf_mask = 0;
+	int i;
 
-	wa_write_clr_set(wal, GEN8_MCR_SELECTOR, mcr_mask, mcr);
+	/*
+	 * On Xe_HP the steering increases in complexity. There are now several
+	 * more units that require steering and we're not guaranteed to be able
+	 * to find a common setting for all of them. These are:
+	 * - GSLICE (fusable)
+	 * - DSS (sub-unit within gslice; fusable)
+	 * - L3 Bank (fusable)
+	 * - MSLICE (fusable)
+	 * - LNCF (sub-unit within mslice; always present if mslice is present)
+	 * - SQIDI (always on)
+	 *
+	 * We'll do our default/implicit steering based on GSLICE (in the
+	 * sliceid field) and DSS (in the subsliceid field).  If we can
+	 * find overlap between the valid MSLICE and/or LNCF values with
+	 * a suitable GSLICE, then we can just re-use the default value and
+	 * skip and explicit steering at runtime.
+	 *
+	 * We only need to look for overlap between GSLICE/MSLICE/LNCF to find
+	 * a valid sliceid value.  DSS steering is the only type of steering
+	 * that utilizes the 'subsliceid' bits.
+	 *
+	 * Also note that, even though the steering domain is called "GSlice"
+	 * and it is encoded in the register using the gslice format, the spec
+	 * says that the combined (geometry | compute) fuse should be used to
+	 * select the steering.
+	 */
+
+	/* Find the potential gslice candidates */
+	dss_mask = intel_sseu_get_subslices(sseu, 0);
+	slice_mask = intel_slicemask_from_dssmask(dss_mask, GEN_DSS_PER_GSLICE);
+
+	/*
+	 * Find the potential LNCF candidates.  Either LNCF within a valid
+	 * mslice is fine.
+	 */
+	for_each_set_bit(i, &gt->info.mslice_mask, GEN12_MAX_MSLICES)
+		lncf_mask |= (0x3 << (i * 2));
+
+	/*
+	 * Are there any sliceid values that work for both GSLICE and LNCF
+	 * steering?
+	 */
+	if (slice_mask & lncf_mask) {
+		slice_mask &= lncf_mask;
+		gt->steering_table[LNCF] = NULL;
+	}
+
+	/* How about sliceid values that also work for MSLICE steering? */
+	if (slice_mask & gt->info.mslice_mask) {
+		slice_mask &= gt->info.mslice_mask;
+		gt->steering_table[MSLICE] = NULL;
+	}
+
+	slice = __ffs(slice_mask);
+	subslice = __ffs(dss_mask >> (slice * GEN_DSS_PER_GSLICE));
+	WARN_ON(subslice > GEN_DSS_PER_GSLICE);
+	WARN_ON(dss_mask >> (slice * GEN_DSS_PER_GSLICE) == 0);
+
+	__add_mcr_wa(i915, wal, slice, subslice);
 }
 

static void

drivers/gpu/drm/i915/i915_drv.h +3 -0
index 70bef688c6da..31b8bb43c581 100644
@@ -1694,6 +1694,9 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915,
 #define HAS_RUNTIME_PM(dev_priv) (INTEL_INFO(dev_priv)->has_runtime_pm)
 #define HAS_64BIT_RELOC(dev_priv) (INTEL_INFO(dev_priv)->has_64bit_reloc)
 
+#define HAS_MSLICES(dev_priv) \
+	(INTEL_INFO(dev_priv)->has_mslices)
+
 #define HAS_IPC(dev_priv)		 (INTEL_INFO(dev_priv)->display.has_ipc)
 
 #define HAS_REGION(i915, i) (INTEL_INFO(i915)->memory_regions & (i))
drivers/gpu/drm/i915/i915_pci.c +1 -0
index 08651ca03478..ed9edda727d2 100644
@@ -1007,6 +1007,7 @@ static const struct intel_device_info adl_p_info = {
 	.has_llc = 1, \
 	.has_logical_ring_contexts = 1, \
 	.has_logical_ring_elsq = 1, \
+	.has_mslices = 1, \
 	.has_rc6 = 1, \
 	.has_reset_engine = 1, \
 	.has_rps = 1, \
drivers/gpu/drm/i915/i915_reg.h +0 -0
index 1a164b90e9fc..f4113e7e8271 100644

drivers/gpu/drm/i915/i915_reg.h static inline bool i915_mmio_reg_valid(i915_reg_t reg) · line 2769

Quoted 4 lines
drivers/gpu/drm/i915/i915_reg.h +1 -0
@@ -2766,6 +2766,7 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define   GEN11_MCR_SLICE_MASK		GEN11_MCR_SLICE(0xf)
 #define   GEN11_MCR_SUBSLICE(subslice)	(((subslice) & 0x7) << 24)
 #define   GEN11_MCR_SUBSLICE_MASK	GEN11_MCR_SUBSLICE(0x7)
+#define   GEN11_MCR_MULTICAST		REG_BIT(31)

this is not used anywhere so we could just omit it. If we are adding to have the complete definition for the register, this needs to be before GEN11_MCR_SLICE() to adhere to the rules of definiting the most significant bit first.

other than that, patch looks fine to me. It took me some time to review and understand the table, the steering rules, etc. But AFAICS this seems to follow them

Lucas De Marchi

Quoted 32 lines
drivers/gpu/drm/i915/i915_reg.h +3 -0
 #define RING_IPEIR(base)	_MMIO((base) + 0x64)
 #define RING_IPEHR(base)	_MMIO((base) + 0x68)
 #define RING_EIR(base)		_MMIO((base) + 0xb0)
@@ -3184,6 +3185,9 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define	GEN10_MIRROR_FUSE3		_MMIO(0x9118)
 #define GEN10_L3BANK_PAIR_COUNT     4
 #define GEN10_L3BANK_MASK   0x0F
+/* on Xe_HP the same fuses indicates mslices instead of L3 banks */
+#define GEN12_MAX_MSLICES 4
+#define GEN12_MEML3_EN_MASK 0x0F
 
 #define GEN8_EU_DISABLE0		_MMIO(0x9134)
 #define   GEN8_EU_DIS0_S0_MASK		0xffffff
drivers/gpu/drm/i915/intel_device_info.h +1 -0
index 121d6d9afd3a..2177372f9ac3 100644
@@ -133,6 +133,7 @@ enum intel_ppgtt_type {
 	func(has_llc); \
 	func(has_logical_ring_contexts); \
 	func(has_logical_ring_elsq); \
+	func(has_mslices); \
 	func(has_pooled_eu); \
 	func(has_rc6); \
 	func(has_rc6p); \
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#3 lore raw

PATCH v4 2/18 drm/i915/xehpsdv: Define steering tables

In reply to Matt Roper

Define and initialize the MMIO ranges for which XeHP SDV requires MSLICE and LNCF steering.

Bspec: 66534

Cc Tvrtko Ursulin <[email protected]>
Cc Daniele Ceraolo Spurio <[email protected]>
---
 drivers/gpu/drm/i915/gt/intel_gt.c          | 19 ++++++++++++++++++-
 drivers/gpu/drm/i915/gt/intel_workarounds.c | 11 +++++++++--
 2 files changed, 27 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/gt/intel_gt.c +18 -1
index 39b224600f38..8e13bfc81634 100644
@@ -89,6 +89,20 @@ static const struct intel_mmio_range icl_l3bank_steering_table[] = {
 	{},
 };
 
+static const struct intel_mmio_range xehpsdv_mslice_steering_table[] = {
+	{ 0x004000, 0x004AFF },
+	{ 0x00C800, 0x00CFFF },
+	{ 0x00DD00, 0x00DDFF },
+	{ 0x00E900, 0x00FFFF }, /* 0xEA00 - OxEFFF is unused */
+	{},
+};
+
+static const struct intel_mmio_range xehpsdv_lncf_steering_table[] = {
+	{ 0x00B000, 0x00B0FF },
+	{ 0x00D800, 0x00D8FF },
+	{},
+};
+
 static u16 slicemask(struct intel_gt *gt, int count)
 {
 	u64 dss_mask = intel_sseu_get_subslices(&gt->info.sseu, 0);
@@ -115,7 +129,10 @@ int intel_gt_init_mmio(struct intel_gt *gt)
 			(intel_uncore_read(gt->uncore, GEN10_MIRROR_FUSE3) &
 			 GEN12_MEML3_EN_MASK);
 
-	if (GRAPHICS_VER(i915) >= 11 &&
+	if (IS_XEHPSDV(i915)) {
+		gt->steering_table[MSLICE] = xehpsdv_mslice_steering_table;
+		gt->steering_table[LNCF] = xehpsdv_lncf_steering_table;
+	} else if (GRAPHICS_VER(i915) >= 11 &&
 		   GRAPHICS_VER_FULL(i915) < IP_VER(12, 50)) {
 		gt->steering_table[L3BANK] = icl_l3bank_steering_table;
 		gt->info.l3bank_mask =
drivers/gpu/drm/i915/gt/intel_workarounds.c +9 -2
index f2ceabb0e2ea..8717337a6c81 100644
@@ -934,7 +934,6 @@ icl_wa_init_mcr(struct drm_i915_private *i915, struct i915_wa_list *wal)
 	__add_mcr_wa(i915, wal, slice, subslice);
 }
 
-__maybe_unused
 static void
 xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal)
 {
@@ -1136,10 +1135,18 @@ dg1_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 			    VSUNIT_CLKGATE_DIS_TGL);
 }
 
+static void
+xehpsdv_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
+{
+	xehp_init_mcr(&i915->gt, wal);
+}
+
 static void
 gt_init_workarounds(struct drm_i915_private *i915, struct i915_wa_list *wal)
 {
-	if (IS_DG1(i915))
+	if (IS_XEHPSDV(i915))
+		xehpsdv_gt_workarounds_init(i915, wal);
+	else if (IS_DG1(i915))
 		dg1_gt_workarounds_init(i915, wal);
 	else if (IS_TIGERLAKE(i915))
 		tgl_gt_workarounds_init(i915, wal);
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v4 2/18
#4 lore raw

Re: [Intel-gfx] [PATCH v4 02/18] drm/i915/xehpsdv: Define steering tables

Quoted 7 lines · Matt Roper

Define and initialize the MMIO ranges for which XeHP SDV requires MSLICE and LNCF steering.

Bspec: 66534

Cc Tvrtko Ursulin <[email protected]>
Cc Daniele Ceraolo Spurio <[email protected]>

Lucas De Marchi

Quoted 81 lines

---
drivers/gpu/drm/i915/gt/intel_gt.c | 19 ++++++++++++++++++-
drivers/gpu/drm/i915/gt/intel_workarounds.c | 11 +++++++++--

2 files changed, 27 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/gt/intel_gt.c +18 -1
index 39b224600f38..8e13bfc81634 100644
@@ -89,6 +89,20 @@ static const struct intel_mmio_range icl_l3bank_steering_table[] = {
 	{},
 };
 
+static const struct intel_mmio_range xehpsdv_mslice_steering_table[] = {
+	{ 0x004000, 0x004AFF },
+	{ 0x00C800, 0x00CFFF },
+	{ 0x00DD00, 0x00DDFF },
+	{ 0x00E900, 0x00FFFF }, /* 0xEA00 - OxEFFF is unused */
+	{},
+};
+
+static const struct intel_mmio_range xehpsdv_lncf_steering_table[] = {
+	{ 0x00B000, 0x00B0FF },
+	{ 0x00D800, 0x00D8FF },
+	{},
+};
+
 static u16 slicemask(struct intel_gt *gt, int count)
 {
 	u64 dss_mask = intel_sseu_get_subslices(&gt->info.sseu, 0);
@@ -115,7 +129,10 @@ int intel_gt_init_mmio(struct intel_gt *gt)
 			(intel_uncore_read(gt->uncore, GEN10_MIRROR_FUSE3) &
 			 GEN12_MEML3_EN_MASK);
 
-	if (GRAPHICS_VER(i915) >= 11 &&
+	if (IS_XEHPSDV(i915)) {
+		gt->steering_table[MSLICE] = xehpsdv_mslice_steering_table;
+		gt->steering_table[LNCF] = xehpsdv_lncf_steering_table;
+	} else if (GRAPHICS_VER(i915) >= 11 &&
 		   GRAPHICS_VER_FULL(i915) < IP_VER(12, 50)) {
 		gt->steering_table[L3BANK] = icl_l3bank_steering_table;
 		gt->info.l3bank_mask =
drivers/gpu/drm/i915/gt/intel_workarounds.c +0 -1
index f2ceabb0e2ea..8717337a6c81 100644
@@ -934,7 +934,6 @@ icl_wa_init_mcr(struct drm_i915_private *i915, struct i915_wa_list *wal)
 	__add_mcr_wa(i915, wal, slice, subslice);
 }
 
-__maybe_unused

static void
xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal) {

drivers/gpu/drm/i915/gt/intel_workarounds.c +6 -0
@@ -1136,10 +1135,18 @@ dg1_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 			    VSUNIT_CLKGATE_DIS_TGL);
 }
 
+static void
+xehpsdv_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
+{
+	xehp_init_mcr(&i915->gt, wal);
+}
+

static void
gt_init_workarounds(struct drm_i915_private *i915, struct i915_wa_list *wal) {

drivers/gpu/drm/i915/gt/intel_workarounds.c +3 -1
-	if (IS_DG1(i915))
+	if (IS_XEHPSDV(i915))
+		xehpsdv_gt_workarounds_init(i915, wal);
+	else if (IS_DG1(i915))
		dg1_gt_workarounds_init(i915, wal);
	else if (IS_TIGERLAKE(i915))
		tgl_gt_workarounds_init(i915, wal);
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#5 lore raw

PATCH v4 3/18 drm/i915/dg2: Add forcewake table

In reply to Matt Roper

The DG2 forcewake table is very similar to the one used by XeHP SDV (and both platforms are even presented as a single table in the bspec). For the most part DG2 starts using a few additional ranges that were 'reserved' on XeHP SDV and stops using some others. However there is a single range (0xd800-0xd87f) that needs to be handled differently between the two platforms (it needs GT wake on XeHP SDV, but render wake on DG2) so unless we want to wake both domains (which could waste power) or define new types of forcewake domains for this special case we need to have separate tables for the two platforms. Let's define the ranges for both platforms with a parameterized macro so that we don't actually need to duplicate everything in the code.

It should be fine for DG2 to re-use the Xe_HP shadow register list so we can continue to use the 'xehpsdv' MMIO write functions and don't need to spin up a separate DG2 instance.

Bspec: 66534

Cc Daniele Ceraolo Spurio <[email protected]>
---
 drivers/gpu/drm/i915/intel_uncore.c | 305 +++++++++++++++-------------
 1 file changed, 168 insertions(+), 137 deletions(-)
drivers/gpu/drm/i915/intel_uncore.c +168 -137
index 8cf53f54559d..6b38bc2811c1 100644
@@ -1317,143 +1317,170 @@ static const struct intel_forcewake_range __gen12_fw_ranges[] = {
 		0x1d3f00 - 0x1d3fff: VD2 */
 };
 
-/* *Must* be sorted by offset ranges! See intel_fw_table_check(). */
-static const struct intel_forcewake_range __xehp_fw_ranges[] = {
-	GEN_FW_RANGE(0x0, 0x1fff, 0), /*
-		  0x0 -  0xaff: reserved
-		0xb00 - 0x1fff: always on */
-	GEN_FW_RANGE(0x2000, 0x26ff, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x2700, 0x4aff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x4b00, 0x51ff, 0), /*
-		0x4b00 - 0x4fff: reserved
-		0x5000 - 0x51ff: always on */
-	GEN_FW_RANGE(0x5200, 0x7fff, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x8000, 0x813f, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x8140, 0x815f, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x8160, 0x81ff, 0), /*
-		0x8160 - 0x817f: reserved
-		0x8180 - 0x81ff: always on */
-	GEN_FW_RANGE(0x8200, 0x82ff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x8300, 0x84ff, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x8500, 0x94cf, FORCEWAKE_GT), /*
-		0x8500 - 0x87ff: gt
-		0x8800 - 0x8fff: reserved
-		0x9000 - 0x947f: gt
-		0x9480 - 0x94cf: reserved */
-	GEN_FW_RANGE(0x94d0, 0x955f, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x9560, 0x97ff, 0), /*
-		0x9560 - 0x95ff: always on
-		0x9600 - 0x97ff: reserved */
-	GEN_FW_RANGE(0x9800, 0xcfff, FORCEWAKE_GT), /*
-		0x9800 - 0xb4ff: gt
-		0xb500 - 0xbfff: reserved
-		0xc000 - 0xcfff: gt */
-	GEN_FW_RANGE(0xd000, 0xd7ff, 0),
-	GEN_FW_RANGE(0xd800, 0xdbff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0xdc00, 0xdcff, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0xdd00, 0xde7f, FORCEWAKE_GT), /*
-		0xdd00 - 0xddff: gt
-		0xde00 - 0xde7f: reserved */
-	GEN_FW_RANGE(0xde80, 0xe8ff, FORCEWAKE_RENDER), /*
-		0xde80 - 0xdfff: render
-		0xe000 - 0xe0ff: reserved
-		0xe100 - 0xe8ff: render */
-	GEN_FW_RANGE(0xe900, 0xffff, FORCEWAKE_GT), /*
-		0xe900 - 0xe9ff: gt
-		0xea00 - 0xefff: reserved
-		0xf000 - 0xffff: gt */
-	GEN_FW_RANGE(0x10000, 0x13fff, 0), /*
-		0x10000 - 0x11fff: reserved
-		0x12000 - 0x127ff: always on
-		0x12800 - 0x13fff: reserved */
-	GEN_FW_RANGE(0x14000, 0x141ff, FORCEWAKE_MEDIA_VDBOX0),
-	GEN_FW_RANGE(0x14200, 0x143ff, FORCEWAKE_MEDIA_VDBOX2),
-	GEN_FW_RANGE(0x14400, 0x145ff, FORCEWAKE_MEDIA_VDBOX4),
-	GEN_FW_RANGE(0x14600, 0x147ff, FORCEWAKE_MEDIA_VDBOX6),
-	GEN_FW_RANGE(0x14800, 0x1ffff, FORCEWAKE_RENDER), /*
-		0x14800 - 0x14fff: render
-		0x15000 - 0x16dff: reserved
-		0x16e00 - 0x1ffff: render */
-	GEN_FW_RANGE(0x20000, 0x21fff, FORCEWAKE_MEDIA_VDBOX0), /*
-		0x20000 - 0x20fff: VD0
-		0x21000 - 0x21fff: reserved */
-	GEN_FW_RANGE(0x22000, 0x23fff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x24000, 0x2417f, 0), /*
-		0x24000 - 0x2407f: always on
-		0x24080 - 0x2417f: reserved */
-	GEN_FW_RANGE(0x24180, 0x249ff, FORCEWAKE_GT), /*
-		0x24180 - 0x241ff: gt
-		0x24200 - 0x249ff: reserved */
-	GEN_FW_RANGE(0x24a00, 0x251ff, FORCEWAKE_RENDER), /*
-		0x24a00 - 0x24a7f: render
-		0x24a80 - 0x251ff: reserved */
-	GEN_FW_RANGE(0x25200, 0x25fff, FORCEWAKE_GT), /*
-		0x25200 - 0x252ff: gt
-		0x25300 - 0x25fff: reserved */
-	GEN_FW_RANGE(0x26000, 0x2ffff, FORCEWAKE_RENDER), /*
-		0x26000 - 0x27fff: render
-		0x28000 - 0x29fff: reserved
-		0x2a000 - 0x2ffff: undocumented */
-	GEN_FW_RANGE(0x30000, 0x3ffff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x40000, 0x1bffff, 0),
-	GEN_FW_RANGE(0x1c0000, 0x1c3fff, FORCEWAKE_MEDIA_VDBOX0), /*
-		0x1c0000 - 0x1c2bff: VD0
-		0x1c2c00 - 0x1c2cff: reserved
-		0x1c2d00 - 0x1c2dff: VD0
-		0x1c2e00 - 0x1c3eff: reserved
-		0x1c3f00 - 0x1c3fff: VD0 */
-	GEN_FW_RANGE(0x1c4000, 0x1c7fff, FORCEWAKE_MEDIA_VDBOX1), /*
-		0x1c4000 - 0x1c6bff: VD1
-		0x1c6c00 - 0x1c6cff: reserved
-		0x1c6d00 - 0x1c6dff: VD1
-		0x1c6e00 - 0x1c7fff: reserved */
-	GEN_FW_RANGE(0x1c8000, 0x1cbfff, FORCEWAKE_MEDIA_VEBOX0), /*
-		0x1c8000 - 0x1ca0ff: VE0
-		0x1ca100 - 0x1cbfff: reserved */
-	GEN_FW_RANGE(0x1cc000, 0x1ccfff, FORCEWAKE_MEDIA_VDBOX0),
-	GEN_FW_RANGE(0x1cd000, 0x1cdfff, FORCEWAKE_MEDIA_VDBOX2),
-	GEN_FW_RANGE(0x1ce000, 0x1cefff, FORCEWAKE_MEDIA_VDBOX4),
-	GEN_FW_RANGE(0x1cf000, 0x1cffff, FORCEWAKE_MEDIA_VDBOX6),
-	GEN_FW_RANGE(0x1d0000, 0x1d3fff, FORCEWAKE_MEDIA_VDBOX2), /*
-		0x1d0000 - 0x1d2bff: VD2
-		0x1d2c00 - 0x1d2cff: reserved
-		0x1d2d00 - 0x1d2dff: VD2
-		0x1d2e00 - 0x1d3eff: reserved
-		0x1d3f00 - 0x1d3fff: VD2 */
-	GEN_FW_RANGE(0x1d4000, 0x1d7fff, FORCEWAKE_MEDIA_VDBOX3), /*
-		0x1d4000 - 0x1d6bff: VD3
-		0x1d6c00 - 0x1d6cff: reserved
-		0x1d6d00 - 0x1d6dff: VD3
-		0x1d6e00 - 0x1d7fff: reserved */
-	GEN_FW_RANGE(0x1d8000, 0x1dffff, FORCEWAKE_MEDIA_VEBOX1), /*
-		0x1d8000 - 0x1da0ff: VE1
-		0x1da100 - 0x1dffff: reserved */
-	GEN_FW_RANGE(0x1e0000, 0x1e3fff, FORCEWAKE_MEDIA_VDBOX4), /*
-		0x1e0000 - 0x1e2bff: VD4
-		0x1e2c00 - 0x1e2cff: reserved
-		0x1e2d00 - 0x1e2dff: VD4
-		0x1e2e00 - 0x1e3eff: reserved
-		0x1e3f00 - 0x1e3fff: VD4 */
-	GEN_FW_RANGE(0x1e4000, 0x1e7fff, FORCEWAKE_MEDIA_VDBOX5), /*
-		0x1e4000 - 0x1e6bff: VD5
-		0x1e6c00 - 0x1e6cff: reserved
-		0x1e6d00 - 0x1e6dff: VD5
-		0x1e6e00 - 0x1e7fff: reserved */
-	GEN_FW_RANGE(0x1e8000, 0x1effff, FORCEWAKE_MEDIA_VEBOX2), /*
-		0x1e8000 - 0x1ea0ff: VE2
-		0x1ea100 - 0x1effff: reserved */
-	GEN_FW_RANGE(0x1f0000, 0x1f3fff, FORCEWAKE_MEDIA_VDBOX6), /*
-		0x1f0000 - 0x1f2bff: VD6
-		0x1f2c00 - 0x1f2cff: reserved
-		0x1f2d00 - 0x1f2dff: VD6
-		0x1f2e00 - 0x1f3eff: reserved
-		0x1f3f00 - 0x1f3fff: VD6 */
-	GEN_FW_RANGE(0x1f4000, 0x1f7fff, FORCEWAKE_MEDIA_VDBOX7), /*
-		0x1f4000 - 0x1f6bff: VD7
-		0x1f6c00 - 0x1f6cff: reserved
-		0x1f6d00 - 0x1f6dff: VD7
-		0x1f6e00 - 0x1f7fff: reserved */
+/*
+ * Graphics IP version 12.55 brings a slight change to the 0xd800 range,
+ * switching it from the GT domain to the render domain.
+ *
+ * *Must* be sorted by offset ranges! See intel_fw_table_check().
+ */
+#define XEHP_FWRANGES(FW_RANGE_D800)					\
+	GEN_FW_RANGE(0x0, 0x1fff, 0), /*					\
+		  0x0 -  0xaff: reserved					\
+		0xb00 - 0x1fff: always on */					\
+	GEN_FW_RANGE(0x2000, 0x26ff, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x2700, 0x4aff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x4b00, 0x51ff, 0), /*					\
+		0x4b00 - 0x4fff: reserved					\
+		0x5000 - 0x51ff: always on */					\
+	GEN_FW_RANGE(0x5200, 0x7fff, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x8000, 0x813f, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x8140, 0x815f, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x8160, 0x81ff, 0), /*					\
+		0x8160 - 0x817f: reserved					\
+		0x8180 - 0x81ff: always on */					\
+	GEN_FW_RANGE(0x8200, 0x82ff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x8300, 0x84ff, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x8500, 0x8cff, FORCEWAKE_GT), /*				\
+		0x8500 - 0x87ff: gt						\
+		0x8800 - 0x8c7f: reserved					\
+		0x8c80 - 0x8cff: gt (DG2 only) */				\
+	GEN_FW_RANGE(0x8d00, 0x8fff, FORCEWAKE_RENDER), /*			\
+		0x8d00 - 0x8dff: render (DG2 only)				\
+		0x8e00 - 0x8fff: reserved */					\
+	GEN_FW_RANGE(0x9000, 0x94cf, FORCEWAKE_GT), /*				\
+		0x9000 - 0x947f: gt						\
+		0x9480 - 0x94cf: reserved */					\
+	GEN_FW_RANGE(0x94d0, 0x955f, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x9560, 0x967f, 0), /*					\
+		0x9560 - 0x95ff: always on					\
+		0x9600 - 0x967f: reserved */					\
+	GEN_FW_RANGE(0x9680, 0x97ff, FORCEWAKE_RENDER), /*			\
+		0x9680 - 0x96ff: render (DG2 only)				\
+		0x9700 - 0x97ff: reserved */					\
+	GEN_FW_RANGE(0x9800, 0xcfff, FORCEWAKE_GT), /*				\
+		0x9800 - 0xb4ff: gt						\
+		0xb500 - 0xbfff: reserved					\
+		0xc000 - 0xcfff: gt */						\
+	GEN_FW_RANGE(0xd000, 0xd7ff, 0),					\
+	GEN_FW_RANGE(0xd800, 0xd87f, FW_RANGE_D800),			\
+	GEN_FW_RANGE(0xd880, 0xdbff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0xdc00, 0xdcff, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0xdd00, 0xde7f, FORCEWAKE_GT), /*				\
+		0xdd00 - 0xddff: gt						\
+		0xde00 - 0xde7f: reserved */					\
+	GEN_FW_RANGE(0xde80, 0xe8ff, FORCEWAKE_RENDER), /*			\
+		0xde80 - 0xdfff: render						\
+		0xe000 - 0xe0ff: reserved					\
+		0xe100 - 0xe8ff: render */					\
+	GEN_FW_RANGE(0xe900, 0xffff, FORCEWAKE_GT), /*				\
+		0xe900 - 0xe9ff: gt						\
+		0xea00 - 0xefff: reserved					\
+		0xf000 - 0xffff: gt */						\
+	GEN_FW_RANGE(0x10000, 0x12fff, 0), /*					\
+		0x10000 - 0x11fff: reserved					\
+		0x12000 - 0x127ff: always on					\
+		0x12800 - 0x12fff: reserved */					\
+	GEN_FW_RANGE(0x13000, 0x131ff, FORCEWAKE_MEDIA_VDBOX0), /* DG2 only */	\
+	GEN_FW_RANGE(0x13200, 0x13fff, FORCEWAKE_MEDIA_VDBOX2), /*		\
+		0x13200 - 0x133ff: VD2 (DG2 only)				\
+		0x13400 - 0x13fff: reserved */					\
+	GEN_FW_RANGE(0x14000, 0x141ff, FORCEWAKE_MEDIA_VDBOX0), /* XEHPSDV only */	\
+	GEN_FW_RANGE(0x14200, 0x143ff, FORCEWAKE_MEDIA_VDBOX2), /* XEHPSDV only */	\
+	GEN_FW_RANGE(0x14400, 0x145ff, FORCEWAKE_MEDIA_VDBOX4), /* XEHPSDV only */	\
+	GEN_FW_RANGE(0x14600, 0x147ff, FORCEWAKE_MEDIA_VDBOX6), /* XEHPSDV only */	\
+	GEN_FW_RANGE(0x14800, 0x14fff, FORCEWAKE_RENDER),			\
+	GEN_FW_RANGE(0x15000, 0x16dff, FORCEWAKE_GT), /*			\
+		0x15000 - 0x15fff: gt (DG2 only)				\
+		0x16000 - 0x16dff: reserved */					\
+	GEN_FW_RANGE(0x16e00, 0x1ffff, FORCEWAKE_RENDER),			\
+	GEN_FW_RANGE(0x20000, 0x21fff, FORCEWAKE_MEDIA_VDBOX0), /*		\
+		0x20000 - 0x20fff: VD0 (XEHPSDV only)				\
+		0x21000 - 0x21fff: reserved */					\
+	GEN_FW_RANGE(0x22000, 0x23fff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x24000, 0x2417f, 0), /*					\
+		0x24000 - 0x2407f: always on					\
+		0x24080 - 0x2417f: reserved */					\
+	GEN_FW_RANGE(0x24180, 0x249ff, FORCEWAKE_GT), /*			\
+		0x24180 - 0x241ff: gt						\
+		0x24200 - 0x249ff: reserved */					\
+	GEN_FW_RANGE(0x24a00, 0x251ff, FORCEWAKE_RENDER), /*			\
+		0x24a00 - 0x24a7f: render					\
+		0x24a80 - 0x251ff: reserved */					\
+	GEN_FW_RANGE(0x25200, 0x25fff, FORCEWAKE_GT), /*			\
+		0x25200 - 0x252ff: gt						\
+		0x25300 - 0x25fff: reserved */					\
+	GEN_FW_RANGE(0x26000, 0x2ffff, FORCEWAKE_RENDER), /*			\
+		0x26000 - 0x27fff: render					\
+		0x28000 - 0x29fff: reserved					\
+		0x2a000 - 0x2ffff: undocumented */				\
+	GEN_FW_RANGE(0x30000, 0x3ffff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x40000, 0x1bffff, 0),					\
+	GEN_FW_RANGE(0x1c0000, 0x1c3fff, FORCEWAKE_MEDIA_VDBOX0), /*		\
+		0x1c0000 - 0x1c2bff: VD0					\
+		0x1c2c00 - 0x1c2cff: reserved					\
+		0x1c2d00 - 0x1c2dff: VD0					\
+		0x1c2e00 - 0x1c3eff: VD0 (DG2 only)				\
+		0x1c3f00 - 0x1c3fff: VD0 */					\
+	GEN_FW_RANGE(0x1c4000, 0x1c7fff, FORCEWAKE_MEDIA_VDBOX1), /*		\
+		0x1c4000 - 0x1c6bff: VD1					\
+		0x1c6c00 - 0x1c6cff: reserved					\
+		0x1c6d00 - 0x1c6dff: VD1					\
+		0x1c6e00 - 0x1c7fff: reserved */				\
+	GEN_FW_RANGE(0x1c8000, 0x1cbfff, FORCEWAKE_MEDIA_VEBOX0), /*		\
+		0x1c8000 - 0x1ca0ff: VE0					\
+		0x1ca100 - 0x1cbfff: reserved */				\
+	GEN_FW_RANGE(0x1cc000, 0x1ccfff, FORCEWAKE_MEDIA_VDBOX0),		\
+	GEN_FW_RANGE(0x1cd000, 0x1cdfff, FORCEWAKE_MEDIA_VDBOX2),		\
+	GEN_FW_RANGE(0x1ce000, 0x1cefff, FORCEWAKE_MEDIA_VDBOX4),		\
+	GEN_FW_RANGE(0x1cf000, 0x1cffff, FORCEWAKE_MEDIA_VDBOX6),		\
+	GEN_FW_RANGE(0x1d0000, 0x1d3fff, FORCEWAKE_MEDIA_VDBOX2), /*		\
+		0x1d0000 - 0x1d2bff: VD2					\
+		0x1d2c00 - 0x1d2cff: reserved					\
+		0x1d2d00 - 0x1d2dff: VD2					\
+		0x1d2e00 - 0x1d3dff: VD2 (DG2 only)				\
+		0x1d3e00 - 0x1d3eff: reserved					\
+		0x1d3f00 - 0x1d3fff: VD2 */					\
+	GEN_FW_RANGE(0x1d4000, 0x1d7fff, FORCEWAKE_MEDIA_VDBOX3), /*		\
+		0x1d4000 - 0x1d6bff: VD3					\
+		0x1d6c00 - 0x1d6cff: reserved					\
+		0x1d6d00 - 0x1d6dff: VD3					\
+		0x1d6e00 - 0x1d7fff: reserved */				\
+	GEN_FW_RANGE(0x1d8000, 0x1dffff, FORCEWAKE_MEDIA_VEBOX1), /*		\
+		0x1d8000 - 0x1da0ff: VE1					\
+		0x1da100 - 0x1dffff: reserved */				\
+	GEN_FW_RANGE(0x1e0000, 0x1e3fff, FORCEWAKE_MEDIA_VDBOX4), /*		\
+		0x1e0000 - 0x1e2bff: VD4					\
+		0x1e2c00 - 0x1e2cff: reserved					\
+		0x1e2d00 - 0x1e2dff: VD4					\
+		0x1e2e00 - 0x1e3eff: reserved					\
+		0x1e3f00 - 0x1e3fff: VD4 */					\
+	GEN_FW_RANGE(0x1e4000, 0x1e7fff, FORCEWAKE_MEDIA_VDBOX5), /*		\
+		0x1e4000 - 0x1e6bff: VD5					\
+		0x1e6c00 - 0x1e6cff: reserved					\
+		0x1e6d00 - 0x1e6dff: VD5					\
+		0x1e6e00 - 0x1e7fff: reserved */				\
+	GEN_FW_RANGE(0x1e8000, 0x1effff, FORCEWAKE_MEDIA_VEBOX2), /*		\
+		0x1e8000 - 0x1ea0ff: VE2					\
+		0x1ea100 - 0x1effff: reserved */				\
+	GEN_FW_RANGE(0x1f0000, 0x1f3fff, FORCEWAKE_MEDIA_VDBOX6), /*		\
+		0x1f0000 - 0x1f2bff: VD6					\
+		0x1f2c00 - 0x1f2cff: reserved					\
+		0x1f2d00 - 0x1f2dff: VD6					\
+		0x1f2e00 - 0x1f3eff: reserved					\
+		0x1f3f00 - 0x1f3fff: VD6 */					\
+	GEN_FW_RANGE(0x1f4000, 0x1f7fff, FORCEWAKE_MEDIA_VDBOX7), /*		\
+		0x1f4000 - 0x1f6bff: VD7					\
+		0x1f6c00 - 0x1f6cff: reserved					\
+		0x1f6d00 - 0x1f6dff: VD7					\
+		0x1f6e00 - 0x1f7fff: reserved */				\
 	GEN_FW_RANGE(0x1f8000, 0x1fa0ff, FORCEWAKE_MEDIA_VEBOX3),
+
+static const struct intel_forcewake_range __xehp_fw_ranges[] = {
+	XEHP_FWRANGES(FORCEWAKE_GT)
+};
+
+static const struct intel_forcewake_range __dg2_fw_ranges[] = {
+	XEHP_FWRANGES(FORCEWAKE_RENDER)
 };
 
 static void
@@ -2084,7 +2111,11 @@ static int uncore_forcewake_init(struct intel_uncore *uncore)
 		return ret;
 	forcewake_early_sanitize(uncore, 0);
 
-	if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50)) {
+	if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 55)) {
+		ASSIGN_FW_DOMAINS_TABLE(uncore, __dg2_fw_ranges);
+		ASSIGN_WRITE_MMIO_VFUNCS(uncore, xehp_fwtable);
+		ASSIGN_READ_MMIO_VFUNCS(uncore, gen11_fwtable);
+	} else if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50)) {
 		ASSIGN_FW_DOMAINS_TABLE(uncore, __xehp_fw_ranges);
 		ASSIGN_WRITE_MMIO_VFUNCS(uncore, xehp_fwtable);
 		ASSIGN_READ_MMIO_VFUNCS(uncore, gen11_fwtable);
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v4 3/18
#6 lore raw

Re: [Intel-gfx] [PATCH v4 03/18] drm/i915/dg2: Add forcewake table

Quoted 12 lines · Matt Roper

The DG2 forcewake table is very similar to the one used by XeHP SDV (and both platforms are even presented as a single table in the bspec). For the most part DG2 starts using a few additional ranges that were 'reserved' on XeHP SDV and stops using some others. However there is a single range (0xd800-0xd87f) that needs to be handled differently between the two platforms (it needs GT wake on XeHP SDV, but render wake on DG2) so unless we want to wake both domains (which could waste power) or define new types of forcewake domains for this special case we need to have separate tables for the two platforms. Let's define the ranges for both platforms with a parameterized macro so that we don't actually need to duplicate everything in the code.

Quoted 3 lines

It should be fine for DG2 to re-use the Xe_HP shadow register list so we can continue to use the 'xehpsdv' MMIO write functions and don't need to spin up a separate DG2 instance.

Quoted 332 lines

Bspec: 66534

Cc Daniele Ceraolo Spurio <[email protected]>
---
 drivers/gpu/drm/i915/intel_uncore.c | 305 +++++++++++++++-------------
 1 file changed, 168 insertions(+), 137 deletions(-)
drivers/gpu/drm/i915/intel_uncore.c +168 -137
index 8cf53f54559d..6b38bc2811c1 100644
@@ -1317,143 +1317,170 @@ static const struct intel_forcewake_range __gen12_fw_ranges[] = {
 		0x1d3f00 - 0x1d3fff: VD2 */
 };
 
-/* *Must* be sorted by offset ranges! See intel_fw_table_check(). */
-static const struct intel_forcewake_range __xehp_fw_ranges[] = {
-	GEN_FW_RANGE(0x0, 0x1fff, 0), /*
-		  0x0 -  0xaff: reserved
-		0xb00 - 0x1fff: always on */
-	GEN_FW_RANGE(0x2000, 0x26ff, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x2700, 0x4aff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x4b00, 0x51ff, 0), /*
-		0x4b00 - 0x4fff: reserved
-		0x5000 - 0x51ff: always on */
-	GEN_FW_RANGE(0x5200, 0x7fff, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x8000, 0x813f, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x8140, 0x815f, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x8160, 0x81ff, 0), /*
-		0x8160 - 0x817f: reserved
-		0x8180 - 0x81ff: always on */
-	GEN_FW_RANGE(0x8200, 0x82ff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x8300, 0x84ff, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x8500, 0x94cf, FORCEWAKE_GT), /*
-		0x8500 - 0x87ff: gt
-		0x8800 - 0x8fff: reserved
-		0x9000 - 0x947f: gt
-		0x9480 - 0x94cf: reserved */
-	GEN_FW_RANGE(0x94d0, 0x955f, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0x9560, 0x97ff, 0), /*
-		0x9560 - 0x95ff: always on
-		0x9600 - 0x97ff: reserved */
-	GEN_FW_RANGE(0x9800, 0xcfff, FORCEWAKE_GT), /*
-		0x9800 - 0xb4ff: gt
-		0xb500 - 0xbfff: reserved
-		0xc000 - 0xcfff: gt */
-	GEN_FW_RANGE(0xd000, 0xd7ff, 0),
-	GEN_FW_RANGE(0xd800, 0xdbff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0xdc00, 0xdcff, FORCEWAKE_RENDER),
-	GEN_FW_RANGE(0xdd00, 0xde7f, FORCEWAKE_GT), /*
-		0xdd00 - 0xddff: gt
-		0xde00 - 0xde7f: reserved */
-	GEN_FW_RANGE(0xde80, 0xe8ff, FORCEWAKE_RENDER), /*
-		0xde80 - 0xdfff: render
-		0xe000 - 0xe0ff: reserved
-		0xe100 - 0xe8ff: render */
-	GEN_FW_RANGE(0xe900, 0xffff, FORCEWAKE_GT), /*
-		0xe900 - 0xe9ff: gt
-		0xea00 - 0xefff: reserved
-		0xf000 - 0xffff: gt */
-	GEN_FW_RANGE(0x10000, 0x13fff, 0), /*
-		0x10000 - 0x11fff: reserved
-		0x12000 - 0x127ff: always on
-		0x12800 - 0x13fff: reserved */
-	GEN_FW_RANGE(0x14000, 0x141ff, FORCEWAKE_MEDIA_VDBOX0),
-	GEN_FW_RANGE(0x14200, 0x143ff, FORCEWAKE_MEDIA_VDBOX2),
-	GEN_FW_RANGE(0x14400, 0x145ff, FORCEWAKE_MEDIA_VDBOX4),
-	GEN_FW_RANGE(0x14600, 0x147ff, FORCEWAKE_MEDIA_VDBOX6),
-	GEN_FW_RANGE(0x14800, 0x1ffff, FORCEWAKE_RENDER), /*
-		0x14800 - 0x14fff: render
-		0x15000 - 0x16dff: reserved
-		0x16e00 - 0x1ffff: render */
-	GEN_FW_RANGE(0x20000, 0x21fff, FORCEWAKE_MEDIA_VDBOX0), /*
-		0x20000 - 0x20fff: VD0
-		0x21000 - 0x21fff: reserved */
-	GEN_FW_RANGE(0x22000, 0x23fff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x24000, 0x2417f, 0), /*
-		0x24000 - 0x2407f: always on
-		0x24080 - 0x2417f: reserved */
-	GEN_FW_RANGE(0x24180, 0x249ff, FORCEWAKE_GT), /*
-		0x24180 - 0x241ff: gt
-		0x24200 - 0x249ff: reserved */
-	GEN_FW_RANGE(0x24a00, 0x251ff, FORCEWAKE_RENDER), /*
-		0x24a00 - 0x24a7f: render
-		0x24a80 - 0x251ff: reserved */
-	GEN_FW_RANGE(0x25200, 0x25fff, FORCEWAKE_GT), /*
-		0x25200 - 0x252ff: gt
-		0x25300 - 0x25fff: reserved */
-	GEN_FW_RANGE(0x26000, 0x2ffff, FORCEWAKE_RENDER), /*
-		0x26000 - 0x27fff: render
-		0x28000 - 0x29fff: reserved
-		0x2a000 - 0x2ffff: undocumented */
-	GEN_FW_RANGE(0x30000, 0x3ffff, FORCEWAKE_GT),
-	GEN_FW_RANGE(0x40000, 0x1bffff, 0),
-	GEN_FW_RANGE(0x1c0000, 0x1c3fff, FORCEWAKE_MEDIA_VDBOX0), /*
-		0x1c0000 - 0x1c2bff: VD0
-		0x1c2c00 - 0x1c2cff: reserved
-		0x1c2d00 - 0x1c2dff: VD0
-		0x1c2e00 - 0x1c3eff: reserved
-		0x1c3f00 - 0x1c3fff: VD0 */
-	GEN_FW_RANGE(0x1c4000, 0x1c7fff, FORCEWAKE_MEDIA_VDBOX1), /*
-		0x1c4000 - 0x1c6bff: VD1
-		0x1c6c00 - 0x1c6cff: reserved
-		0x1c6d00 - 0x1c6dff: VD1
-		0x1c6e00 - 0x1c7fff: reserved */
-	GEN_FW_RANGE(0x1c8000, 0x1cbfff, FORCEWAKE_MEDIA_VEBOX0), /*
-		0x1c8000 - 0x1ca0ff: VE0
-		0x1ca100 - 0x1cbfff: reserved */
-	GEN_FW_RANGE(0x1cc000, 0x1ccfff, FORCEWAKE_MEDIA_VDBOX0),
-	GEN_FW_RANGE(0x1cd000, 0x1cdfff, FORCEWAKE_MEDIA_VDBOX2),
-	GEN_FW_RANGE(0x1ce000, 0x1cefff, FORCEWAKE_MEDIA_VDBOX4),
-	GEN_FW_RANGE(0x1cf000, 0x1cffff, FORCEWAKE_MEDIA_VDBOX6),
-	GEN_FW_RANGE(0x1d0000, 0x1d3fff, FORCEWAKE_MEDIA_VDBOX2), /*
-		0x1d0000 - 0x1d2bff: VD2
-		0x1d2c00 - 0x1d2cff: reserved
-		0x1d2d00 - 0x1d2dff: VD2
-		0x1d2e00 - 0x1d3eff: reserved
-		0x1d3f00 - 0x1d3fff: VD2 */
-	GEN_FW_RANGE(0x1d4000, 0x1d7fff, FORCEWAKE_MEDIA_VDBOX3), /*
-		0x1d4000 - 0x1d6bff: VD3
-		0x1d6c00 - 0x1d6cff: reserved
-		0x1d6d00 - 0x1d6dff: VD3
-		0x1d6e00 - 0x1d7fff: reserved */
-	GEN_FW_RANGE(0x1d8000, 0x1dffff, FORCEWAKE_MEDIA_VEBOX1), /*
-		0x1d8000 - 0x1da0ff: VE1
-		0x1da100 - 0x1dffff: reserved */
-	GEN_FW_RANGE(0x1e0000, 0x1e3fff, FORCEWAKE_MEDIA_VDBOX4), /*
-		0x1e0000 - 0x1e2bff: VD4
-		0x1e2c00 - 0x1e2cff: reserved
-		0x1e2d00 - 0x1e2dff: VD4
-		0x1e2e00 - 0x1e3eff: reserved
-		0x1e3f00 - 0x1e3fff: VD4 */
-	GEN_FW_RANGE(0x1e4000, 0x1e7fff, FORCEWAKE_MEDIA_VDBOX5), /*
-		0x1e4000 - 0x1e6bff: VD5
-		0x1e6c00 - 0x1e6cff: reserved
-		0x1e6d00 - 0x1e6dff: VD5
-		0x1e6e00 - 0x1e7fff: reserved */
-	GEN_FW_RANGE(0x1e8000, 0x1effff, FORCEWAKE_MEDIA_VEBOX2), /*
-		0x1e8000 - 0x1ea0ff: VE2
-		0x1ea100 - 0x1effff: reserved */
-	GEN_FW_RANGE(0x1f0000, 0x1f3fff, FORCEWAKE_MEDIA_VDBOX6), /*
-		0x1f0000 - 0x1f2bff: VD6
-		0x1f2c00 - 0x1f2cff: reserved
-		0x1f2d00 - 0x1f2dff: VD6
-		0x1f2e00 - 0x1f3eff: reserved
-		0x1f3f00 - 0x1f3fff: VD6 */
-	GEN_FW_RANGE(0x1f4000, 0x1f7fff, FORCEWAKE_MEDIA_VDBOX7), /*
-		0x1f4000 - 0x1f6bff: VD7
-		0x1f6c00 - 0x1f6cff: reserved
-		0x1f6d00 - 0x1f6dff: VD7
-		0x1f6e00 - 0x1f7fff: reserved */
+/*
+ * Graphics IP version 12.55 brings a slight change to the 0xd800 range,
+ * switching it from the GT domain to the render domain.
+ *
+ * *Must* be sorted by offset ranges! See intel_fw_table_check().
+ */
+#define XEHP_FWRANGES(FW_RANGE_D800)					\
+	GEN_FW_RANGE(0x0, 0x1fff, 0), /*					\
+		  0x0 -  0xaff: reserved					\
+		0xb00 - 0x1fff: always on */					\
+	GEN_FW_RANGE(0x2000, 0x26ff, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x2700, 0x4aff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x4b00, 0x51ff, 0), /*					\
+		0x4b00 - 0x4fff: reserved					\
+		0x5000 - 0x51ff: always on */					\
+	GEN_FW_RANGE(0x5200, 0x7fff, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x8000, 0x813f, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x8140, 0x815f, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x8160, 0x81ff, 0), /*					\
+		0x8160 - 0x817f: reserved					\
+		0x8180 - 0x81ff: always on */					\
+	GEN_FW_RANGE(0x8200, 0x82ff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x8300, 0x84ff, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x8500, 0x8cff, FORCEWAKE_GT), /*				\
+		0x8500 - 0x87ff: gt						\
+		0x8800 - 0x8c7f: reserved					\
+		0x8c80 - 0x8cff: gt (DG2 only) */				\
+	GEN_FW_RANGE(0x8d00, 0x8fff, FORCEWAKE_RENDER), /*			\
+		0x8d00 - 0x8dff: render (DG2 only)				\
+		0x8e00 - 0x8fff: reserved */					\
+	GEN_FW_RANGE(0x9000, 0x94cf, FORCEWAKE_GT), /*				\
+		0x9000 - 0x947f: gt						\
+		0x9480 - 0x94cf: reserved */					\
+	GEN_FW_RANGE(0x94d0, 0x955f, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0x9560, 0x967f, 0), /*					\
+		0x9560 - 0x95ff: always on					\
+		0x9600 - 0x967f: reserved */					\
+	GEN_FW_RANGE(0x9680, 0x97ff, FORCEWAKE_RENDER), /*			\
+		0x9680 - 0x96ff: render (DG2 only)				\
+		0x9700 - 0x97ff: reserved */					\
+	GEN_FW_RANGE(0x9800, 0xcfff, FORCEWAKE_GT), /*				\
+		0x9800 - 0xb4ff: gt						\
+		0xb500 - 0xbfff: reserved					\
+		0xc000 - 0xcfff: gt */						\
+	GEN_FW_RANGE(0xd000, 0xd7ff, 0),					\
+	GEN_FW_RANGE(0xd800, 0xd87f, FW_RANGE_D800),			\
+	GEN_FW_RANGE(0xd880, 0xdbff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0xdc00, 0xdcff, FORCEWAKE_RENDER),				\
+	GEN_FW_RANGE(0xdd00, 0xde7f, FORCEWAKE_GT), /*				\
+		0xdd00 - 0xddff: gt						\
+		0xde00 - 0xde7f: reserved */					\
+	GEN_FW_RANGE(0xde80, 0xe8ff, FORCEWAKE_RENDER), /*			\
+		0xde80 - 0xdfff: render						\
+		0xe000 - 0xe0ff: reserved					\
+		0xe100 - 0xe8ff: render */					\
+	GEN_FW_RANGE(0xe900, 0xffff, FORCEWAKE_GT), /*				\
+		0xe900 - 0xe9ff: gt						\
+		0xea00 - 0xefff: reserved					\
+		0xf000 - 0xffff: gt */						\
+	GEN_FW_RANGE(0x10000, 0x12fff, 0), /*					\
+		0x10000 - 0x11fff: reserved					\
+		0x12000 - 0x127ff: always on					\
+		0x12800 - 0x12fff: reserved */					\
+	GEN_FW_RANGE(0x13000, 0x131ff, FORCEWAKE_MEDIA_VDBOX0), /* DG2 only */	\
+	GEN_FW_RANGE(0x13200, 0x13fff, FORCEWAKE_MEDIA_VDBOX2), /*		\
+		0x13200 - 0x133ff: VD2 (DG2 only)				\
+		0x13400 - 0x13fff: reserved */					\
+	GEN_FW_RANGE(0x14000, 0x141ff, FORCEWAKE_MEDIA_VDBOX0), /* XEHPSDV only */	\
+	GEN_FW_RANGE(0x14200, 0x143ff, FORCEWAKE_MEDIA_VDBOX2), /* XEHPSDV only */	\
+	GEN_FW_RANGE(0x14400, 0x145ff, FORCEWAKE_MEDIA_VDBOX4), /* XEHPSDV only */	\
+	GEN_FW_RANGE(0x14600, 0x147ff, FORCEWAKE_MEDIA_VDBOX6), /* XEHPSDV only */	\
+	GEN_FW_RANGE(0x14800, 0x14fff, FORCEWAKE_RENDER),			\
+	GEN_FW_RANGE(0x15000, 0x16dff, FORCEWAKE_GT), /*			\
+		0x15000 - 0x15fff: gt (DG2 only)				\
+		0x16000 - 0x16dff: reserved */					\
+	GEN_FW_RANGE(0x16e00, 0x1ffff, FORCEWAKE_RENDER),			\
+	GEN_FW_RANGE(0x20000, 0x21fff, FORCEWAKE_MEDIA_VDBOX0), /*		\
+		0x20000 - 0x20fff: VD0 (XEHPSDV only)				\
+		0x21000 - 0x21fff: reserved */					\
+	GEN_FW_RANGE(0x22000, 0x23fff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x24000, 0x2417f, 0), /*					\
+		0x24000 - 0x2407f: always on					\
+		0x24080 - 0x2417f: reserved */					\
+	GEN_FW_RANGE(0x24180, 0x249ff, FORCEWAKE_GT), /*			\
+		0x24180 - 0x241ff: gt						\
+		0x24200 - 0x249ff: reserved */					\
+	GEN_FW_RANGE(0x24a00, 0x251ff, FORCEWAKE_RENDER), /*			\
+		0x24a00 - 0x24a7f: render					\
+		0x24a80 - 0x251ff: reserved */					\
+	GEN_FW_RANGE(0x25200, 0x25fff, FORCEWAKE_GT), /*			\
+		0x25200 - 0x252ff: gt						\
+		0x25300 - 0x25fff: reserved */					\
+	GEN_FW_RANGE(0x26000, 0x2ffff, FORCEWAKE_RENDER), /*			\
+		0x26000 - 0x27fff: render					\
+		0x28000 - 0x29fff: reserved					\
+		0x2a000 - 0x2ffff: undocumented */				\
+	GEN_FW_RANGE(0x30000, 0x3ffff, FORCEWAKE_GT),				\
+	GEN_FW_RANGE(0x40000, 0x1bffff, 0),					\
+	GEN_FW_RANGE(0x1c0000, 0x1c3fff, FORCEWAKE_MEDIA_VDBOX0), /*		\
+		0x1c0000 - 0x1c2bff: VD0					\
+		0x1c2c00 - 0x1c2cff: reserved					\
+		0x1c2d00 - 0x1c2dff: VD0					\
+		0x1c2e00 - 0x1c3eff: VD0 (DG2 only)				\
+		0x1c3f00 - 0x1c3fff: VD0 */					\
+	GEN_FW_RANGE(0x1c4000, 0x1c7fff, FORCEWAKE_MEDIA_VDBOX1), /*		\
+		0x1c4000 - 0x1c6bff: VD1					\
+		0x1c6c00 - 0x1c6cff: reserved					\
+		0x1c6d00 - 0x1c6dff: VD1					\
+		0x1c6e00 - 0x1c7fff: reserved */				\
+	GEN_FW_RANGE(0x1c8000, 0x1cbfff, FORCEWAKE_MEDIA_VEBOX0), /*		\
+		0x1c8000 - 0x1ca0ff: VE0					\
+		0x1ca100 - 0x1cbfff: reserved */				\
+	GEN_FW_RANGE(0x1cc000, 0x1ccfff, FORCEWAKE_MEDIA_VDBOX0),		\
+	GEN_FW_RANGE(0x1cd000, 0x1cdfff, FORCEWAKE_MEDIA_VDBOX2),		\
+	GEN_FW_RANGE(0x1ce000, 0x1cefff, FORCEWAKE_MEDIA_VDBOX4),		\
+	GEN_FW_RANGE(0x1cf000, 0x1cffff, FORCEWAKE_MEDIA_VDBOX6),		\
+	GEN_FW_RANGE(0x1d0000, 0x1d3fff, FORCEWAKE_MEDIA_VDBOX2), /*		\
+		0x1d0000 - 0x1d2bff: VD2					\
+		0x1d2c00 - 0x1d2cff: reserved					\
+		0x1d2d00 - 0x1d2dff: VD2					\
+		0x1d2e00 - 0x1d3dff: VD2 (DG2 only)				\
+		0x1d3e00 - 0x1d3eff: reserved					\
+		0x1d3f00 - 0x1d3fff: VD2 */					\
+	GEN_FW_RANGE(0x1d4000, 0x1d7fff, FORCEWAKE_MEDIA_VDBOX3), /*		\
+		0x1d4000 - 0x1d6bff: VD3					\
+		0x1d6c00 - 0x1d6cff: reserved					\
+		0x1d6d00 - 0x1d6dff: VD3					\
+		0x1d6e00 - 0x1d7fff: reserved */				\
+	GEN_FW_RANGE(0x1d8000, 0x1dffff, FORCEWAKE_MEDIA_VEBOX1), /*		\
+		0x1d8000 - 0x1da0ff: VE1					\
+		0x1da100 - 0x1dffff: reserved */				\
+	GEN_FW_RANGE(0x1e0000, 0x1e3fff, FORCEWAKE_MEDIA_VDBOX4), /*		\
+		0x1e0000 - 0x1e2bff: VD4					\
+		0x1e2c00 - 0x1e2cff: reserved					\
+		0x1e2d00 - 0x1e2dff: VD4					\
+		0x1e2e00 - 0x1e3eff: reserved					\
+		0x1e3f00 - 0x1e3fff: VD4 */					\
+	GEN_FW_RANGE(0x1e4000, 0x1e7fff, FORCEWAKE_MEDIA_VDBOX5), /*		\
+		0x1e4000 - 0x1e6bff: VD5					\
+		0x1e6c00 - 0x1e6cff: reserved					\
+		0x1e6d00 - 0x1e6dff: VD5					\
+		0x1e6e00 - 0x1e7fff: reserved */				\
+	GEN_FW_RANGE(0x1e8000, 0x1effff, FORCEWAKE_MEDIA_VEBOX2), /*		\
+		0x1e8000 - 0x1ea0ff: VE2					\
+		0x1ea100 - 0x1effff: reserved */				\
+	GEN_FW_RANGE(0x1f0000, 0x1f3fff, FORCEWAKE_MEDIA_VDBOX6), /*		\
+		0x1f0000 - 0x1f2bff: VD6					\
+		0x1f2c00 - 0x1f2cff: reserved					\
+		0x1f2d00 - 0x1f2dff: VD6					\
+		0x1f2e00 - 0x1f3eff: reserved					\
+		0x1f3f00 - 0x1f3fff: VD6 */					\
+	GEN_FW_RANGE(0x1f4000, 0x1f7fff, FORCEWAKE_MEDIA_VDBOX7), /*		\
+		0x1f4000 - 0x1f6bff: VD7					\
+		0x1f6c00 - 0x1f6cff: reserved					\
+		0x1f6d00 - 0x1f6dff: VD7					\
+		0x1f6e00 - 0x1f7fff: reserved */				\
 	GEN_FW_RANGE(0x1f8000, 0x1fa0ff, FORCEWAKE_MEDIA_VEBOX3),
+
+static const struct intel_forcewake_range __xehp_fw_ranges[] = {
+	XEHP_FWRANGES(FORCEWAKE_GT)
+};
+
+static const struct intel_forcewake_range __dg2_fw_ranges[] = {
+	XEHP_FWRANGES(FORCEWAKE_RENDER)
 };
 
 static void
@@ -2084,7 +2111,11 @@ static int uncore_forcewake_init(struct intel_uncore *uncore)
 		return ret;
 	forcewake_early_sanitize(uncore, 0);
 
-	if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50)) {
+	if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 55)) {
+		ASSIGN_FW_DOMAINS_TABLE(uncore, __dg2_fw_ranges);
+		ASSIGN_WRITE_MMIO_VFUNCS(uncore, xehp_fwtable);
+		ASSIGN_READ_MMIO_VFUNCS(uncore, gen11_fwtable);
+	} else if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50)) {
 		ASSIGN_FW_DOMAINS_TABLE(uncore, __xehp_fw_ranges);
 		ASSIGN_WRITE_MMIO_VFUNCS(uncore, xehp_fwtable);
 		ASSIGN_READ_MMIO_VFUNCS(uncore, gen11_fwtable);
author
#7 lore raw

PATCH v4 4/18 drm/i915/dg2: Update LNCF steering ranges

In reply to Matt Roper

DG2's replicated register ranges are almost the same at XeHP SDV with the exception of one LNCF sub-range that switches to gslice steering. We can re-use the XeHP SDV mslice steering table and just provide a DG2-specific LNCF steering table.

Bspec: 66534

Cc Daniele Ceraolo Spurio <[email protected]>
---
 drivers/gpu/drm/i915/gt/intel_gt.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_gt.c +10 -1
index 8e13bfc81634..1971e34da254 100644
@@ -103,6 +103,12 @@ static const struct intel_mmio_range xehpsdv_lncf_steering_table[] = {
 	{},
 };
 
+static const struct intel_mmio_range dg2_lncf_steering_table[] = {
+	{ 0x00B000, 0x00B0FF },
+	{ 0x00D880, 0x00D8FF },
+	{},
+};
+
 static u16 slicemask(struct intel_gt *gt, int count)
 {
 	u64 dss_mask = intel_sseu_get_subslices(&gt->info.sseu, 0);
@@ -129,7 +135,10 @@ int intel_gt_init_mmio(struct intel_gt *gt)
 			(intel_uncore_read(gt->uncore, GEN10_MIRROR_FUSE3) &
 			 GEN12_MEML3_EN_MASK);
 
-	if (IS_XEHPSDV(i915)) {
+	if (IS_DG2(i915)) {
+		gt->steering_table[MSLICE] = xehpsdv_mslice_steering_table;
+		gt->steering_table[LNCF] = dg2_lncf_steering_table;
+	} else if (IS_XEHPSDV(i915)) {
 		gt->steering_table[MSLICE] = xehpsdv_mslice_steering_table;
 		gt->steering_table[LNCF] = xehpsdv_lncf_steering_table;
 	} else if (GRAPHICS_VER(i915) >= 11 &&
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v4 4/18
#8 lore raw

Re: [Intel-gfx] [PATCH v4 04/18] drm/i915/dg2: Update LNCF steering ranges

Quoted 30 lines · Matt Roper

DG2's replicated register ranges are almost the same at XeHP SDV with the exception of one LNCF sub-range that switches to gslice steering. We can re-use the XeHP SDV mslice steering table and just provide a DG2-specific LNCF steering table.

Bspec: 66534

Cc Daniele Ceraolo Spurio <[email protected]>

---
drivers/gpu/drm/i915/gt/intel_gt.c | 11 ++++++++++-

1 file changed, 10 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_gt.c +6 -0
index 8e13bfc81634..1971e34da254 100644
@@ -103,6 +103,12 @@ static const struct intel_mmio_range xehpsdv_lncf_steering_table[] = {
 	{},
 };
 
+static const struct intel_mmio_range dg2_lncf_steering_table[] = {
+	{ 0x00B000, 0x00B0FF },
+	{ 0x00D880, 0x00D8FF },
+	{},
+};
+
 static u16 slicemask(struct intel_gt *gt, int count)
 {
 	u64 dss_mask = intel_sseu_get_subslices(&gt->info.sseu, 0);

drivers/gpu/drm/i915/gt/intel_gt.c int intel_gt_init_mmio(struct intel_gt *gt) · line 139

Quoted 6 lines
drivers/gpu/drm/i915/gt/intel_gt.c +2 -1
@@ -129,7 +135,10 @@ int intel_gt_init_mmio(struct intel_gt *gt)
 			(intel_uncore_read(gt->uncore, GEN10_MIRROR_FUSE3) &
 			 GEN12_MEML3_EN_MASK);
 
-	if (IS_XEHPSDV(i915)) {
+	if (IS_DG2(i915)) {
+		gt->steering_table[MSLICE] = xehpsdv_mslice_steering_table;

a title "Update steering tables" would be more appropriate as this is also setting the mslice table, even if just re-using from xehpsdv.

Lucas De Marchi

Quoted 12 lines
drivers/gpu/drm/i915/gt/intel_gt.c +2 -0
+		gt->steering_table[LNCF] = dg2_lncf_steering_table;
+	} else if (IS_XEHPSDV(i915)) {
 		gt->steering_table[MSLICE] = xehpsdv_mslice_steering_table;
 		gt->steering_table[LNCF] = xehpsdv_lncf_steering_table;
 	} else if (GRAPHICS_VER(i915) >= 11 &&
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#9 lore raw

PATCH v4 5/18 drm/i915/dg2: Add SQIDI steering

In reply to Matt Roper

Although DG2_G10 platforms will always have all SQIDI's present and don't need steering for registers in a SQIDI MMIO range, this isn't true for DG2_G11 platforms; only SQIDI's 2 and 3 can be used on those.

We handle SQIDI ranges a bit differently from other types of explicit steering. The SQIDI ranges belong to either the MCFG unit or the SF unit, both of which have their own dedicated steering registers and do not use the typical 0xFDC steering control that all other types of ranges use. Thus we only need to worry about picking a valid initial value for the MCFG and SF steering registers (0xFD0 and 0xFD8 resepectively) at driver init; they won't change after we set them up so we don't need to worry about re-steering them explicitly at runtime.

Given that any SQIDI value should work fine for DG2-G10 and XeHP SDV, while only values of 2 and 3 are valid for DG2-G11, we'll just initialize the MCFG and SF steering registers to a constant value of "2" for all XeHP-based platforms for simplicity --- that will work in all cases.

Bspec: 66534

Cc Radhakrishna Sripada <[email protected]>
---
 drivers/gpu/drm/i915/gt/intel_workarounds.c | 28 +++++++++++++++++----
 drivers/gpu/drm/i915/i915_reg.h             |  2 ++
 2 files changed, 25 insertions(+), 5 deletions(-)
drivers/gpu/drm/i915/gt/intel_workarounds.c +23 -5
index 8717337a6c81..6895b083523d 100644
@@ -889,17 +889,24 @@ cfl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 		    GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
 }
 
-static void __add_mcr_wa(struct drm_i915_private *i915, struct i915_wa_list *wal,
-			 unsigned slice, unsigned subslice)
+static void __set_mcr_steering(struct i915_wa_list *wal,
+			       i915_reg_t steering_reg,
+			       unsigned int slice, unsigned int subslice)
 {
 	u32 mcr, mcr_mask;
 
 	mcr = GEN11_MCR_SLICE(slice) | GEN11_MCR_SUBSLICE(subslice);
 	mcr_mask = GEN11_MCR_SLICE_MASK | GEN11_MCR_SUBSLICE_MASK;
 
-	drm_dbg(&i915->drm, "MCR slice/subslice = %x\n", mcr);
+	wa_write_clr_set(wal, steering_reg, mcr_mask, mcr);
+}
+
+static void __add_mcr_wa(struct drm_i915_private *i915, struct i915_wa_list *wal,
+			 unsigned int slice, unsigned int subslice)
+{
+	drm_dbg(&i915->drm, "MCR slice=0x%x, subslice=0x%x\n", slice, subslice);
 
-	wa_write_clr_set(wal, GEN8_MCR_SELECTOR, mcr_mask, mcr);
+	__set_mcr_steering(wal, GEN8_MCR_SELECTOR, slice, subslice);
 }
 
 static void
@@ -953,7 +960,6 @@ xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal)
 	 * - L3 Bank (fusable)
 	 * - MSLICE (fusable)
 	 * - LNCF (sub-unit within mslice; always present if mslice is present)
-	 * - SQIDI (always on)
 	 *
 	 * We'll do our default/implicit steering based on GSLICE (in the
 	 * sliceid field) and DSS (in the subsliceid field).  If we can
@@ -1003,6 +1009,18 @@ xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal)
 	WARN_ON(dss_mask >> (slice * GEN_DSS_PER_GSLICE) == 0);
 
 	__add_mcr_wa(i915, wal, slice, subslice);
+
+	/*
+	 * SQIDI ranges are special because they use different steering
+	 * registers than everything else we work with.  On XeHP SDV and
+	 * DG2-G10, any value in the steering registers will work fine since
+	 * all instances are present, but DG2-G11 only has SQIDI instances at
+	 * ID's 2 and 3, so we need to steer to one of those.  For simplicity
+	 * we'll just steer to a hardcoded "2" since that value will work
+	 * everywhere.
+	 */
+	__set_mcr_steering(wal, MCFG_MCR_SELECTOR, 0, 2);
+	__set_mcr_steering(wal, SF_MCR_SELECTOR, 0, 2);
 }
 
 static void
drivers/gpu/drm/i915/i915_reg.h +2 -0
index f4113e7e8271..39ce6befff52 100644
@@ -2757,6 +2757,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define GEN12_SC_INSTDONE_EXTRA2	_MMIO(0x7108)
 #define GEN7_SAMPLER_INSTDONE	_MMIO(0xe160)
 #define GEN7_ROW_INSTDONE	_MMIO(0xe164)
+#define MCFG_MCR_SELECTOR		_MMIO(0xfd0)
+#define SF_MCR_SELECTOR			_MMIO(0xfd8)
 #define GEN8_MCR_SELECTOR		_MMIO(0xfdc)
 #define   GEN8_MCR_SLICE(slice)		(((slice) & 3) << 26)
 #define   GEN8_MCR_SLICE_MASK		GEN8_MCR_SLICE(3)
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v4 5/18
#10 lore raw

Re: [Intel-gfx] [PATCH v4 05/18] drm/i915/dg2: Add SQIDI steering

Quoted 11 lines · Matt Roper

Although DG2_G10 platforms will always have all SQIDI's present and don't need steering for registers in a SQIDI MMIO range, this isn't true for DG2_G11 platforms; only SQIDI's 2 and 3 can be used on those.

We handle SQIDI ranges a bit differently from other types of explicit steering. The SQIDI ranges belong to either the MCFG unit or the SF unit, both of which have their own dedicated steering registers and do not use the typical 0xFDC steering control that all other types of ranges use. Thus we only need to worry about picking a valid initial value for the MCFG and SF steering registers (0xFD0 and 0xFD8 resepectively) at driver init; they won't change after we set them up so

respectively

Quoted 33 lines

we don't need to worry about re-steering them explicitly at runtime.

Given that any SQIDI value should work fine for DG2-G10 and XeHP SDV, while only values of 2 and 3 are valid for DG2-G11, we'll just initialize the MCFG and SF steering registers to a constant value of "2" for all XeHP-based platforms for simplicity --- that will work in all cases.

Bspec: 66534

Cc Radhakrishna Sripada <[email protected]>

---
drivers/gpu/drm/i915/gt/intel_workarounds.c | 28 +++++++++++++++++----
drivers/gpu/drm/i915/i915_reg.h | 2 ++

2 files changed, 25 insertions(+), 5 deletions(-)
drivers/gpu/drm/i915/gt/intel_workarounds.c +3 -2
index 8717337a6c81..6895b083523d 100644
@@ -889,17 +889,24 @@ cfl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 		    GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
 }
 
-static void __add_mcr_wa(struct drm_i915_private *i915, struct i915_wa_list *wal,
-			 unsigned slice, unsigned subslice)
+static void __set_mcr_steering(struct i915_wa_list *wal,
+			       i915_reg_t steering_reg,
+			       unsigned int slice, unsigned int subslice)
 {
 	u32 mcr, mcr_mask;
 
 	mcr = GEN11_MCR_SLICE(slice) | GEN11_MCR_SUBSLICE(subslice);

drivers/gpu/drm/i915/gt/intel_workarounds.c cfl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) · line 907

Quoted 10 lines
drivers/gpu/drm/i915/gt/intel_workarounds.c +7 -1
 	mcr_mask = GEN11_MCR_SLICE_MASK | GEN11_MCR_SUBSLICE_MASK;
 
-	drm_dbg(&i915->drm, "MCR slice/subslice = %x\n", mcr);
+	wa_write_clr_set(wal, steering_reg, mcr_mask, mcr);
+}
+
+static void __add_mcr_wa(struct drm_i915_private *i915, struct i915_wa_list *wal,
+			 unsigned int slice, unsigned int subslice)
+{
+	drm_dbg(&i915->drm, "MCR slice=0x%x, subslice=0x%x\n", slice, subslice);

maybe we could leave the debug message in __set_mcr_steering() and add what steering register we are setting? Up to you.

Quoted 53 lines
drivers/gpu/drm/i915/gt/intel_workarounds.c +1 -1
-	wa_write_clr_set(wal, GEN8_MCR_SELECTOR, mcr_mask, mcr);
+	__set_mcr_steering(wal, GEN8_MCR_SELECTOR, slice, subslice);
 }
 

static void

drivers/gpu/drm/i915/gt/intel_workarounds.c +12 -1
@@ -953,7 +960,6 @@ xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal)
 	 * - L3 Bank (fusable)
 	 * - MSLICE (fusable)
 	 * - LNCF (sub-unit within mslice; always present if mslice is present)
-	 * - SQIDI (always on)
 	 *
 	 * We'll do our default/implicit steering based on GSLICE (in the
 	 * sliceid field) and DSS (in the subsliceid field).  If we can
@@ -1003,6 +1009,18 @@ xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal)
 	WARN_ON(dss_mask >> (slice * GEN_DSS_PER_GSLICE) == 0);
 
 	__add_mcr_wa(i915, wal, slice, subslice);
+
+	/*
+	 * SQIDI ranges are special because they use different steering
+	 * registers than everything else we work with.  On XeHP SDV and
+	 * DG2-G10, any value in the steering registers will work fine since
+	 * all instances are present, but DG2-G11 only has SQIDI instances at
+	 * ID's 2 and 3, so we need to steer to one of those.  For simplicity
+	 * we'll just steer to a hardcoded "2" since that value will work
+	 * everywhere.
+	 */
+	__set_mcr_steering(wal, MCFG_MCR_SELECTOR, 0, 2);
+	__set_mcr_steering(wal, SF_MCR_SELECTOR, 0, 2);
 }
 

static void

drivers/gpu/drm/i915/i915_reg.h +2 -0
index f4113e7e8271..39ce6befff52 100644
@@ -2757,6 +2757,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define GEN12_SC_INSTDONE_EXTRA2	_MMIO(0x7108)
 #define GEN7_SAMPLER_INSTDONE	_MMIO(0xe160)
 #define GEN7_ROW_INSTDONE	_MMIO(0xe164)
+#define MCFG_MCR_SELECTOR		_MMIO(0xfd0)
+#define SF_MCR_SELECTOR			_MMIO(0xfd8)
 #define GEN8_MCR_SELECTOR		_MMIO(0xfdc)
 #define   GEN8_MCR_SLICE(slice)		(((slice) & 3) << 26)
 #define   GEN8_MCR_SLICE_MASK		GEN8_MCR_SLICE(3)
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#11 lore raw

Re: [Intel-gfx] [PATCH v4 05/18] drm/i915/dg2: Add SQIDI steering

Quoted 50 lines · Lucas De Marchi
Quoted 11 lines · Matt Roper

Although DG2_G10 platforms will always have all SQIDI's present and don't need steering for registers in a SQIDI MMIO range, this isn't true for DG2_G11 platforms; only SQIDI's 2 and 3 can be used on those.

We handle SQIDI ranges a bit differently from other types of explicit steering. The SQIDI ranges belong to either the MCFG unit or the SF unit, both of which have their own dedicated steering registers and do not use the typical 0xFDC steering control that all other types of ranges use. Thus we only need to worry about picking a valid initial value for the MCFG and SF steering registers (0xFD0 and 0xFD8 resepectively) at driver init; they won't change after we set them up so

respectively

Quoted 35 lines

we don't need to worry about re-steering them explicitly at runtime.

Given that any SQIDI value should work fine for DG2-G10 and XeHP SDV, while only values of 2 and 3 are valid for DG2-G11, we'll just initialize the MCFG and SF steering registers to a constant value of "2" for all XeHP-based platforms for simplicity --- that will work in all cases.

Bspec: 66534

Cc Radhakrishna Sripada <[email protected]>

---
drivers/gpu/drm/i915/gt/intel_workarounds.c | 28 +++++++++++++++++----
drivers/gpu/drm/i915/i915_reg.h | 2 ++

2 files changed, 25 insertions(+), 5 deletions(-)
drivers/gpu/drm/i915/gt/intel_workarounds.c +3 -2
index 8717337a6c81..6895b083523d 100644
@@ -889,17 +889,24 @@ cfl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal)
 		    GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
 }
 
-static void __add_mcr_wa(struct drm_i915_private *i915, struct i915_wa_list *wal,
-			 unsigned slice, unsigned subslice)
+static void __set_mcr_steering(struct i915_wa_list *wal,
+			       i915_reg_t steering_reg,
+			       unsigned int slice, unsigned int subslice)
 {
 	u32 mcr, mcr_mask;
 
 	mcr = GEN11_MCR_SLICE(slice) | GEN11_MCR_SUBSLICE(subslice);
 	mcr_mask = GEN11_MCR_SLICE_MASK | GEN11_MCR_SUBSLICE_MASK;
 
Quoted 12 lines

drivers/gpu/drm/i915/gt/intel_workarounds.c cfl_gt_workarounds_init(struct drm_i915_private *i915, struct i915_wa_list *wal) · line 907

Quoted 8 lines
drivers/gpu/drm/i915/gt/intel_workarounds.c +7 -1
-	drm_dbg(&i915->drm, "MCR slice/subslice = %x\n", mcr);
+	wa_write_clr_set(wal, steering_reg, mcr_mask, mcr);
+}
+
+static void __add_mcr_wa(struct drm_i915_private *i915, struct i915_wa_list *wal,
+			 unsigned int slice, unsigned int subslice)
+{
+	drm_dbg(&i915->drm, "MCR slice=0x%x, subslice=0x%x\n", slice, subslice);

maybe we could leave the debug message in __set_mcr_steering() and add what steering register we are setting? Up to you.

I've got a separate patch that adds more clear steering debug information via a drm_printer and then prints it both in the dmesg log and in a new debugfs node. The patch depends on some debugfs changes that haven't shown up yet so I didn't include it here, but I'll rebase and send it soon if the debugfs changes don't happen first.

Matt

Quoted 57 lines
Quoted 53 lines
drivers/gpu/drm/i915/gt/intel_workarounds.c +1 -1
-	wa_write_clr_set(wal, GEN8_MCR_SELECTOR, mcr_mask, mcr);
+	__set_mcr_steering(wal, GEN8_MCR_SELECTOR, slice, subslice);
 }
 

static void

drivers/gpu/drm/i915/gt/intel_workarounds.c +12 -1
@@ -953,7 +960,6 @@ xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal)
 	 * - L3 Bank (fusable)
 	 * - MSLICE (fusable)
 	 * - LNCF (sub-unit within mslice; always present if mslice is present)
-	 * - SQIDI (always on)
 	 *
 	 * We'll do our default/implicit steering based on GSLICE (in the
 	 * sliceid field) and DSS (in the subsliceid field).  If we can
@@ -1003,6 +1009,18 @@ xehp_init_mcr(struct intel_gt *gt, struct i915_wa_list *wal)
 	WARN_ON(dss_mask >> (slice * GEN_DSS_PER_GSLICE) == 0);
 
 	__add_mcr_wa(i915, wal, slice, subslice);
+
+	/*
+	 * SQIDI ranges are special because they use different steering
+	 * registers than everything else we work with.  On XeHP SDV and
+	 * DG2-G10, any value in the steering registers will work fine since
+	 * all instances are present, but DG2-G11 only has SQIDI instances at
+	 * ID's 2 and 3, so we need to steer to one of those.  For simplicity
+	 * we'll just steer to a hardcoded "2" since that value will work
+	 * everywhere.
+	 */
+	__set_mcr_steering(wal, MCFG_MCR_SELECTOR, 0, 2);
+	__set_mcr_steering(wal, SF_MCR_SELECTOR, 0, 2);
 }
 

static void

drivers/gpu/drm/i915/i915_reg.h +2 -0
index f4113e7e8271..39ce6befff52 100644
@@ -2757,6 +2757,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define GEN12_SC_INSTDONE_EXTRA2	_MMIO(0x7108)
 #define GEN7_SAMPLER_INSTDONE	_MMIO(0xe160)
 #define GEN7_ROW_INSTDONE	_MMIO(0xe164)
+#define MCFG_MCR_SELECTOR		_MMIO(0xfd0)
+#define SF_MCR_SELECTOR			_MMIO(0xfd8)
 #define GEN8_MCR_SELECTOR		_MMIO(0xfdc)
 #define   GEN8_MCR_SLICE(slice)		(((slice) & 3) << 26)
 #define   GEN8_MCR_SLICE_MASK		GEN8_MCR_SLICE(3)
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 5 lines
Matt Roper
Graphics Software Engineer
VTT-OSGC Platform Enablement
Intel Corporation
(916) 356-2795
author
#12 lore raw

PATCH v4 6/18 drm/i915/xehp: Loop over all gslices for INSTDONE processing

In reply to Matt Roper

We no longer have traditional slices on Xe_HP platforms, but the INSTDONE registers are replicated according to gslice representation which is similar. We can mostly re-use the existing instdone code with just a few modifications:

* Create an alternate instdone loop macro that will iterate over the flat DSS space, but still provide the gslice/dss steering values for compatibility with the legacy code.

* We should allocate INSTDONE storage space according to the maximum number of gslices rather than the maximum number of legacy slices to ensure we have enough storage space to hold all of the values. XeHP design has 8 gslices, whereas older platforms never had more than 3 slices.

---
 drivers/gpu/drm/i915/gt/intel_engine_cs.c    | 48 +++++++++++---------
 drivers/gpu/drm/i915/gt/intel_engine_types.h | 12 ++++-
 drivers/gpu/drm/i915/gt/intel_sseu.h         |  7 +++
 drivers/gpu/drm/i915/i915_gpu_error.c        | 32 +++++++++----
 4 files changed, 66 insertions(+), 33 deletions(-)
drivers/gpu/drm/i915/gt/intel_engine_cs.c +26 -22
index dea0e522c5c7..06733e86e88b 100644
@@ -1166,16 +1166,16 @@ void intel_engine_get_instdone(const struct intel_engine_cs *engine,
 	u32 mmio_base = engine->mmio_base;
 	int slice;
 	int subslice;
+	int iter;
 
 	memset(instdone, 0, sizeof(*instdone));
 
-	switch (GRAPHICS_VER(i915)) {
-	default:
+	if (GRAPHICS_VER(i915) >= 8) {
 		instdone->instdone =
 			intel_uncore_read(uncore, RING_INSTDONE(mmio_base));
 
 		if (engine->id != RCS0)
-			break;
+			return;
 
 		instdone->slice_common =
 			intel_uncore_read(uncore, GEN7_SC_INSTDONE);
@@ -1185,21 +1185,32 @@ void intel_engine_get_instdone(const struct intel_engine_cs *engine,
 			instdone->slice_common_extra[1] =
 				intel_uncore_read(uncore, GEN12_SC_INSTDONE_EXTRA2);
 		}
-		for_each_instdone_slice_subslice(i915, sseu, slice, subslice) {
-			instdone->sampler[slice][subslice] =
-				read_subslice_reg(engine, slice, subslice,
-						  GEN7_SAMPLER_INSTDONE);
-			instdone->row[slice][subslice] =
-				read_subslice_reg(engine, slice, subslice,
-						  GEN7_ROW_INSTDONE);
+
+		if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50)) {
+			for_each_instdone_gslice_dss_xehp(i915, sseu, iter, slice, subslice) {
+				instdone->sampler[slice][subslice] =
+					read_subslice_reg(engine, slice, subslice,
+							  GEN7_SAMPLER_INSTDONE);
+				instdone->row[slice][subslice] =
+					read_subslice_reg(engine, slice, subslice,
+							  GEN7_ROW_INSTDONE);
+			}
+		} else {
+			for_each_instdone_slice_subslice(i915, sseu, slice, subslice) {
+				instdone->sampler[slice][subslice] =
+					read_subslice_reg(engine, slice, subslice,
+							  GEN7_SAMPLER_INSTDONE);
+				instdone->row[slice][subslice] =
+					read_subslice_reg(engine, slice, subslice,
+							  GEN7_ROW_INSTDONE);
+			}
 		}
-		break;
-	case 7:
+	} else if (GRAPHICS_VER(i915) >= 7) {
 		instdone->instdone =
 			intel_uncore_read(uncore, RING_INSTDONE(mmio_base));
 
 		if (engine->id != RCS0)
-			break;
+			return;
 
 		instdone->slice_common =
 			intel_uncore_read(uncore, GEN7_SC_INSTDONE);
@@ -1207,22 +1218,15 @@ void intel_engine_get_instdone(const struct intel_engine_cs *engine,
 			intel_uncore_read(uncore, GEN7_SAMPLER_INSTDONE);
 		instdone->row[0][0] =
 			intel_uncore_read(uncore, GEN7_ROW_INSTDONE);
-
-		break;
-	case 6:
-	case 5:
-	case 4:
+	} else if (GRAPHICS_VER(i915) >= 4) {
 		instdone->instdone =
 			intel_uncore_read(uncore, RING_INSTDONE(mmio_base));
 		if (engine->id == RCS0)
 			/* HACK: Using the wrong struct member */
 			instdone->slice_common =
 				intel_uncore_read(uncore, GEN4_INSTDONE1);
-		break;
-	case 3:
-	case 2:
+	} else {
 		instdone->instdone = intel_uncore_read(uncore, GEN2_INSTDONE);
-		break;
 	}
 }
 
drivers/gpu/drm/i915/gt/intel_engine_types.h +10 -2
index ed91bcff20eb..0b4846b01626 100644
@@ -67,8 +67,8 @@ struct intel_instdone {
 	/* The following exist only in the RCS engine */
 	u32 slice_common;
 	u32 slice_common_extra[2];
-	u32 sampler[I915_MAX_SLICES][I915_MAX_SUBSLICES];
-	u32 row[I915_MAX_SLICES][I915_MAX_SUBSLICES];
+	u32 sampler[GEN_MAX_GSLICES][I915_MAX_SUBSLICES];
+	u32 row[GEN_MAX_GSLICES][I915_MAX_SUBSLICES];
 };
 
 /*
@@ -578,4 +578,12 @@ intel_engine_has_relative_mmio(const struct intel_engine_cs * const engine)
 		for_each_if((instdone_has_slice(dev_priv_, sseu_, slice_)) && \
 			    (instdone_has_subslice(dev_priv_, sseu_, slice_, \
 						    subslice_)))
+
+#define for_each_instdone_gslice_dss_xehp(dev_priv_, sseu_, iter_, gslice_, dss_) \
+	for ((iter_) = 0, (gslice_) = 0, (dss_) = 0; \
+	     (iter_) < GEN_MAX_SUBSLICES; \
+	     (iter_)++, (gslice_) = (iter_) / GEN_DSS_PER_GSLICE, \
+	     (dss_) = (iter_) % GEN_DSS_PER_GSLICE) \
+		for_each_if(intel_sseu_has_subslice((sseu_), 0, (iter_)))
+
 #endif /* __INTEL_ENGINE_TYPES_H__ */
drivers/gpu/drm/i915/gt/intel_sseu.h +7 -0
index 1073471d1980..74487650b08f 100644
@@ -26,6 +26,9 @@ struct drm_printer;
 #define GEN_DSS_PER_CSLICE	8
 #define GEN_DSS_PER_MSLICE	8
 
+#define GEN_MAX_GSLICES		(GEN_MAX_SUBSLICES / GEN_DSS_PER_GSLICE)
+#define GEN_MAX_CSLICES		(GEN_MAX_SUBSLICES / GEN_DSS_PER_CSLICE)
+
 struct sseu_dev_info {
 	u8 slice_mask;
 	u8 subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
@@ -78,6 +81,10 @@ intel_sseu_has_subslice(const struct sseu_dev_info *sseu, int slice,
 	u8 mask;
 	int ss_idx = subslice / BITS_PER_BYTE;
 
+	if (slice >= sseu->max_slices ||
+	    subslice >= sseu->max_subslices)
+		return false;
+
 	GEM_BUG_ON(ss_idx >= sseu->ss_stride);
 
 	mask = sseu->subslice_mask[slice * sseu->ss_stride + ss_idx];
drivers/gpu/drm/i915/i915_gpu_error.c +23 -9
index 0f08bcfbe964..8230bc3ac8a9 100644
@@ -444,15 +444,29 @@ static void error_print_instdone(struct drm_i915_error_state_buf *m,
 	if (GRAPHICS_VER(m->i915) <= 6)
 		return;
 
-	for_each_instdone_slice_subslice(m->i915, sseu, slice, subslice)
-		err_printf(m, "  SAMPLER_INSTDONE[%d][%d]: 0x%08x\n",
-			   slice, subslice,
-			   ee->instdone.sampler[slice][subslice]);
-
-	for_each_instdone_slice_subslice(m->i915, sseu, slice, subslice)
-		err_printf(m, "  ROW_INSTDONE[%d][%d]: 0x%08x\n",
-			   slice, subslice,
-			   ee->instdone.row[slice][subslice]);
+	if (GRAPHICS_VER_FULL(m->i915) >= IP_VER(12, 50)) {
+		int iter;
+
+		for_each_instdone_gslice_dss_xehp(m->i915, sseu, iter, slice, subslice)
+			err_printf(m, "  SAMPLER_INSTDONE[%d][%d]: 0x%08x\n",
+				   slice, subslice,
+				   ee->instdone.sampler[slice][subslice]);
+
+		for_each_instdone_gslice_dss_xehp(m->i915, sseu, iter, slice, subslice)
+			err_printf(m, "  ROW_INSTDONE[%d][%d]: 0x%08x\n",
+				   slice, subslice,
+				   ee->instdone.row[slice][subslice]);
+	} else {
+		for_each_instdone_slice_subslice(m->i915, sseu, slice, subslice)
+			err_printf(m, "  SAMPLER_INSTDONE[%d][%d]: 0x%08x\n",
+				   slice, subslice,
+				   ee->instdone.sampler[slice][subslice]);
+
+		for_each_instdone_slice_subslice(m->i915, sseu, slice, subslice)
+			err_printf(m, "  ROW_INSTDONE[%d][%d]: 0x%08x\n",
+				   slice, subslice,
+				   ee->instdone.row[slice][subslice]);
+	}
 
 	if (GRAPHICS_VER(m->i915) < 12)
 		return;
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#13 lore raw

PATCH v4 7/18 drm/i915/dg2: Report INSTDONE_GEOM values in error state

In reply to Matt Roper

Xe_HPG adds some additional INSTDONE_GEOM debug registers; the Mesa team has indicated that having these reported in the error state would be useful for debugging GPU hangs. These registers are replicated per-DSS with gslice steering.

---
 drivers/gpu/drm/i915/gt/intel_engine_cs.c    |  7 +++++++
 drivers/gpu/drm/i915/gt/intel_engine_types.h |  3 +++
 drivers/gpu/drm/i915/i915_gpu_error.c        | 10 ++++++++--
 drivers/gpu/drm/i915/i915_reg.h              |  1 +
 4 files changed, 19 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/gt/intel_engine_cs.c +7 -0
index 06733e86e88b..0a70a3baef9d 100644
@@ -1205,6 +1205,13 @@ void intel_engine_get_instdone(const struct intel_engine_cs *engine,
 							  GEN7_ROW_INSTDONE);
 			}
 		}
+
+		if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 55)) {
+			for_each_instdone_gslice_dss_xehp(i915, sseu, iter, slice, subslice)
+				instdone->geom_svg[slice][subslice] =
+					read_subslice_reg(engine, slice, subslice,
+							  XEHPG_INSTDONE_GEOM_SVG);
+		}
 	} else if (GRAPHICS_VER(i915) >= 7) {
 		instdone->instdone =
 			intel_uncore_read(uncore, RING_INSTDONE(mmio_base));
drivers/gpu/drm/i915/gt/intel_engine_types.h +3 -0
index 0b4846b01626..bfbfe53c23dd 100644
@@ -69,6 +69,9 @@ struct intel_instdone {
 	u32 slice_common_extra[2];
 	u32 sampler[GEN_MAX_GSLICES][I915_MAX_SUBSLICES];
 	u32 row[GEN_MAX_GSLICES][I915_MAX_SUBSLICES];
+
+	/* Added in XeHPG */
+	u32 geom_svg[GEN_MAX_GSLICES][I915_MAX_SUBSLICES];
 };
 
 /*
drivers/gpu/drm/i915/i915_gpu_error.c +8 -2
index 8230bc3ac8a9..91d5da7b0a2b 100644
@@ -431,6 +431,7 @@ static void error_print_instdone(struct drm_i915_error_state_buf *m,
 	const struct sseu_dev_info *sseu = &ee->engine->gt->info.sseu;
 	int slice;
 	int subslice;
+	int iter;
 
 	err_printf(m, "  INSTDONE: 0x%08x\n",
 		   ee->instdone.instdone);
@@ -445,8 +446,6 @@ static void error_print_instdone(struct drm_i915_error_state_buf *m,
 		return;
 
 	if (GRAPHICS_VER_FULL(m->i915) >= IP_VER(12, 50)) {
-		int iter;
-
 		for_each_instdone_gslice_dss_xehp(m->i915, sseu, iter, slice, subslice)
 			err_printf(m, "  SAMPLER_INSTDONE[%d][%d]: 0x%08x\n",
 				   slice, subslice,
@@ -471,6 +470,13 @@ static void error_print_instdone(struct drm_i915_error_state_buf *m,
 	if (GRAPHICS_VER(m->i915) < 12)
 		return;
 
+	if (GRAPHICS_VER_FULL(m->i915) >= IP_VER(12, 55)) {
+		for_each_instdone_gslice_dss_xehp(m->i915, sseu, iter, slice, subslice)
+			err_printf(m, "  GEOM_SVGUNIT_INSTDONE[%d][%d]: 0x%08x\n",
+				   slice, subslice,
+				   ee->instdone.geom_svg[slice][subslice]);
+	}
+
 	err_printf(m, "  SC_INSTDONE_EXTRA: 0x%08x\n",
 		   ee->instdone.slice_common_extra[0]);
 	err_printf(m, "  SC_INSTDONE_EXTRA2: 0x%08x\n",
drivers/gpu/drm/i915/i915_reg.h +1 -0
index 39ce6befff52..62af453c8c54 100644
@@ -2757,6 +2757,7 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define GEN12_SC_INSTDONE_EXTRA2	_MMIO(0x7108)
 #define GEN7_SAMPLER_INSTDONE	_MMIO(0xe160)
 #define GEN7_ROW_INSTDONE	_MMIO(0xe164)
+#define XEHPG_INSTDONE_GEOM_SVG		_MMIO(0x666c)
 #define MCFG_MCR_SELECTOR		_MMIO(0xfd0)
 #define SF_MCR_SELECTOR			_MMIO(0xfd8)
 #define GEN8_MCR_SELECTOR		_MMIO(0xfdc)
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#14 lore raw

PATCH v4 8/18 drm/i915/xehp: Changes to ss/eu definitions

In reply to Matt Roper

From: Matthew Auld <[email protected]>

Xe_HP no longer has "slices" in the same way that old platforms did. There are new concepts (gslices, cslices, mslices) that apply in various contexts, but for the purposes of fusing slices no longer exist and we just have one large pool of dual-subslices (DSS) to work with. Furthermore, the meaning of the DSS fuse is inverted compared to past platforms --- it now specifies which DSS are enabled rather than which ones are disabled.

---
 drivers/gpu/drm/i915/gt/intel_sseu.c | 24 ++++++++++++++++++++----
 drivers/gpu/drm/i915/i915_getparam.c |  6 ++++--
 drivers/gpu/drm/i915/i915_reg.h      |  3 +++
 3 files changed, 27 insertions(+), 6 deletions(-)
drivers/gpu/drm/i915/gt/intel_sseu.c +20 -4
index bbed8e8625e1..5d1b7d06c96b 100644
@@ -139,17 +139,33 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 	 * Gen12 has Dual-Subslices, which behave similarly to 2 gen11 SS.
 	 * Instead of splitting these, provide userspace with an array
 	 * of DSS to more closely represent the hardware resource.
+	 *
+	 * In addition, the concept of slice has been removed in Xe_HP.
+	 * To be compatible with prior generations, assume a single slice
+	 * across the entire device. Then calculate out the DSS for each
+	 * workload type within that software slice.
 	 */
 	intel_sseu_set_info(sseu, 1, 6, 16);
 
-	s_en = intel_uncore_read(uncore, GEN11_GT_SLICE_ENABLE) &
-		GEN11_GT_S_ENA_MASK;
+	/*
+	 * As mentioned above, Xe_HP does not have the concept of a slice.
+	 * Enable one for software backwards compatibility.
+	 */
+	if (GRAPHICS_VER_FULL(gt->i915) >= IP_VER(12, 50))
+		s_en = 0x1;
+	else
+		s_en = intel_uncore_read(uncore, GEN11_GT_SLICE_ENABLE) &
+		       GEN11_GT_S_ENA_MASK;
 
 	dss_en = intel_uncore_read(uncore, GEN12_GT_DSS_ENABLE);
 
 	/* one bit per pair of EUs */
-	eu_en_fuse = ~(intel_uncore_read(uncore, GEN11_EU_DISABLE) &
-		       GEN11_EU_DIS_MASK);
+	if (GRAPHICS_VER_FULL(gt->i915) >= IP_VER(12, 50))
+		eu_en_fuse = intel_uncore_read(uncore, XEHP_EU_ENABLE) & XEHP_EU_ENA_MASK;
+	else
+		eu_en_fuse = ~(intel_uncore_read(uncore, GEN11_EU_DISABLE) &
+			       GEN11_EU_DIS_MASK);
+
 	for (eu = 0; eu < sseu->max_eus_per_subslice / 2; eu++)
 		if (eu_en_fuse & BIT(eu))
 			eu_en |= BIT(eu * 2) | BIT(eu * 2 + 1);
drivers/gpu/drm/i915/i915_getparam.c +4 -2
index 24e18219eb50..e289397d9178 100644
@@ -15,7 +15,7 @@ int i915_getparam_ioctl(struct drm_device *dev, void *data,
 	struct pci_dev *pdev = to_pci_dev(dev->dev);
 	const struct sseu_dev_info *sseu = &i915->gt.info.sseu;
 	drm_i915_getparam_t *param = data;
-	int value;
+	int value = 0;
 
 	switch (param->param) {
 	case I915_PARAM_IRQ_ACTIVE:
@@ -150,7 +150,9 @@ int i915_getparam_ioctl(struct drm_device *dev, void *data,
 			return -ENODEV;
 		break;
 	case I915_PARAM_SUBSLICE_MASK:
-		value = sseu->subslice_mask[0];
+		/* Only copy bits from the first slice */
+		memcpy(&value, sseu->subslice_mask,
+		       min(sseu->ss_stride, (u8)sizeof(value)));
 		if (!value)
 			return -ENODEV;
 		break;
drivers/gpu/drm/i915/i915_reg.h +3 -0
index 62af453c8c54..99858bc593f0 100644
@@ -3225,6 +3225,9 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 
 #define GEN12_GT_DSS_ENABLE _MMIO(0x913C)
 
+#define XEHP_EU_ENABLE			_MMIO(0x9134)
+#define XEHP_EU_ENA_MASK		0xFF
+
 #define GEN6_BSD_SLEEP_PSMI_CONTROL	_MMIO(0x12050)
 #define   GEN6_BSD_SLEEP_MSG_DISABLE	(1 << 0)
 #define   GEN6_BSD_SLEEP_FLUSH_DISABLE	(1 << 2)
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v4 8/18
#15 lore raw

Re: [Intel-gfx] [PATCH v4 08/18] drm/i915/xehp: Changes to ss/eu definitions

Quoted 9 lines · Matt Roper

From: Matthew Auld <[email protected]>

Xe_HP no longer has "slices" in the same way that old platforms did. There are new concepts (gslices, cslices, mslices) that apply in various contexts, but for the purposes of fusing slices no longer exist and we just have one large pool of dual-subslices (DSS) to work with. Furthermore, the meaning of the DSS fuse is inverted compared to past platforms --- it now specifies which DSS are enabled rather than which ones are disabled.

Quoted 96 lines
---
 drivers/gpu/drm/i915/gt/intel_sseu.c | 24 ++++++++++++++++++++----
 drivers/gpu/drm/i915/i915_getparam.c |  6 ++++--
 drivers/gpu/drm/i915/i915_reg.h      |  3 +++
 3 files changed, 27 insertions(+), 6 deletions(-)
drivers/gpu/drm/i915/gt/intel_sseu.c +20 -4
index bbed8e8625e1..5d1b7d06c96b 100644
@@ -139,17 +139,33 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 	 * Gen12 has Dual-Subslices, which behave similarly to 2 gen11 SS.
 	 * Instead of splitting these, provide userspace with an array
 	 * of DSS to more closely represent the hardware resource.
+	 *
+	 * In addition, the concept of slice has been removed in Xe_HP.
+	 * To be compatible with prior generations, assume a single slice
+	 * across the entire device. Then calculate out the DSS for each
+	 * workload type within that software slice.
 	 */
 	intel_sseu_set_info(sseu, 1, 6, 16);
 
-	s_en = intel_uncore_read(uncore, GEN11_GT_SLICE_ENABLE) &
-		GEN11_GT_S_ENA_MASK;
+	/*
+	 * As mentioned above, Xe_HP does not have the concept of a slice.
+	 * Enable one for software backwards compatibility.
+	 */
+	if (GRAPHICS_VER_FULL(gt->i915) >= IP_VER(12, 50))
+		s_en = 0x1;
+	else
+		s_en = intel_uncore_read(uncore, GEN11_GT_SLICE_ENABLE) &
+		       GEN11_GT_S_ENA_MASK;
 
 	dss_en = intel_uncore_read(uncore, GEN12_GT_DSS_ENABLE);
 
 	/* one bit per pair of EUs */
-	eu_en_fuse = ~(intel_uncore_read(uncore, GEN11_EU_DISABLE) &
-		       GEN11_EU_DIS_MASK);
+	if (GRAPHICS_VER_FULL(gt->i915) >= IP_VER(12, 50))
+		eu_en_fuse = intel_uncore_read(uncore, XEHP_EU_ENABLE) & XEHP_EU_ENA_MASK;
+	else
+		eu_en_fuse = ~(intel_uncore_read(uncore, GEN11_EU_DISABLE) &
+			       GEN11_EU_DIS_MASK);
+
 	for (eu = 0; eu < sseu->max_eus_per_subslice / 2; eu++)
 		if (eu_en_fuse & BIT(eu))
 			eu_en |= BIT(eu * 2) | BIT(eu * 2 + 1);
drivers/gpu/drm/i915/i915_getparam.c +4 -2
index 24e18219eb50..e289397d9178 100644
@@ -15,7 +15,7 @@ int i915_getparam_ioctl(struct drm_device *dev, void *data,
 	struct pci_dev *pdev = to_pci_dev(dev->dev);
 	const struct sseu_dev_info *sseu = &i915->gt.info.sseu;
 	drm_i915_getparam_t *param = data;
-	int value;
+	int value = 0;
 
 	switch (param->param) {
 	case I915_PARAM_IRQ_ACTIVE:
@@ -150,7 +150,9 @@ int i915_getparam_ioctl(struct drm_device *dev, void *data,
 			return -ENODEV;
 		break;
 	case I915_PARAM_SUBSLICE_MASK:
-		value = sseu->subslice_mask[0];
+		/* Only copy bits from the first slice */
+		memcpy(&value, sseu->subslice_mask,
+		       min(sseu->ss_stride, (u8)sizeof(value)));
 		if (!value)
 			return -ENODEV;
 		break;
drivers/gpu/drm/i915/i915_reg.h +3 -0
index 62af453c8c54..99858bc593f0 100644
@@ -3225,6 +3225,9 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 
 #define GEN12_GT_DSS_ENABLE _MMIO(0x913C)
 
+#define XEHP_EU_ENABLE			_MMIO(0x9134)
+#define XEHP_EU_ENA_MASK		0xFF
+
 #define GEN6_BSD_SLEEP_PSMI_CONTROL	_MMIO(0x12050)
 #define   GEN6_BSD_SLEEP_MSG_DISABLE	(1 << 0)
 #define   GEN6_BSD_SLEEP_FLUSH_DISABLE	(1 << 2)
author
#16 lore raw

PATCH v4 9/18 drm/i915/xehpsdv: Add maximum sseu limits

In reply to Matt Roper

Due to the removal of legacy slices and the transition to a gslice/cslice/mslice/etc. design, we'll internally store all DSS under "slice0."

---
 drivers/gpu/drm/i915/gt/intel_sseu.c         | 5 ++++-
 drivers/gpu/drm/i915/gt/intel_sseu.h         | 2 +-
 drivers/gpu/drm/i915/gt/intel_sseu_debugfs.c | 2 +-
 3 files changed, 6 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/gt/intel_sseu.c +4 -1
index 5d1b7d06c96b..16c0552fcd1d 100644
@@ -145,7 +145,10 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 	 * across the entire device. Then calculate out the DSS for each
 	 * workload type within that software slice.
 	 */
-	intel_sseu_set_info(sseu, 1, 6, 16);
+	if (IS_XEHPSDV(gt->i915))
+		intel_sseu_set_info(sseu, 1, 32, 16);
+	else
+		intel_sseu_set_info(sseu, 1, 6, 16);
 
 	/*
 	 * As mentioned above, Xe_HP does not have the concept of a slice.
drivers/gpu/drm/i915/gt/intel_sseu.h +1 -1
index 74487650b08f..204ea6709460 100644
@@ -16,7 +16,7 @@ struct intel_gt;
 struct drm_printer;
 
 #define GEN_MAX_SLICES		(6) /* CNL upper bound */
-#define GEN_MAX_SUBSLICES	(8) /* ICL upper bound */
+#define GEN_MAX_SUBSLICES	(32) /* XEHPSDV upper bound */
 #define GEN_SSEU_STRIDE(max_entries) DIV_ROUND_UP(max_entries, BITS_PER_BYTE)
 #define GEN_MAX_SUBSLICE_STRIDE GEN_SSEU_STRIDE(GEN_MAX_SUBSLICES)
 #define GEN_MAX_EUS		(16) /* TGL upper bound */
drivers/gpu/drm/i915/gt/intel_sseu_debugfs.c +1 -1
index 714fe8495775..a424150b052e 100644
@@ -53,7 +53,7 @@ static void cherryview_sseu_device_status(struct intel_gt *gt,
 static void gen10_sseu_device_status(struct intel_gt *gt,
 				     struct sseu_dev_info *sseu)
 {
-#define SS_MAX 6
+#define SS_MAX 8
 	struct intel_uncore *uncore = gt->uncore;
 	const struct intel_gt_info *info = &gt->info;
 	u32 s_reg[SS_MAX], eu_reg[2 * SS_MAX], eu_mask[2];
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Reviewed-by on v4 9/18
#17 lore raw

Re: [Intel-gfx] [PATCH v4 09/18] drm/i915/xehpsdv: Add maximum sseu limits

Quoted 33 lines · Matt Roper

Due to the removal of legacy slices and the transition to a gslice/cslice/mslice/etc. design, we'll internally store all DSS under "slice0."

---

drivers/gpu/drm/i915/gt/intel_sseu.c         | 5 ++++-
drivers/gpu/drm/i915/gt/intel_sseu.h         | 2 +-
drivers/gpu/drm/i915/gt/intel_sseu_debugfs.c | 2 +-
3 files changed, 6 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/gt/intel_sseu.c +4 -1
index 5d1b7d06c96b..16c0552fcd1d 100644
@@ -145,7 +145,10 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 	 * across the entire device. Then calculate out the DSS for each
 	 * workload type within that software slice.
 	 */
-	intel_sseu_set_info(sseu, 1, 6, 16);
+	if (IS_XEHPSDV(gt->i915))
+		intel_sseu_set_info(sseu, 1, 32, 16);
+	else
+		intel_sseu_set_info(sseu, 1, 6, 16);
 
 	/*
 	 * As mentioned above, Xe_HP does not have the concept of a slice.
drivers/gpu/drm/i915/gt/intel_sseu.h +0 -0
index 74487650b08f..204ea6709460 100644

drivers/gpu/drm/i915/gt/intel_sseu.h struct intel_gt; · line 18

Quoted 3 lines
drivers/gpu/drm/i915/gt/intel_sseu.h +0 -0
@@ -16,7 +16,7 @@ struct intel_gt;
 struct drm_printer;
 
 #define GEN_MAX_SLICES		(6) /* CNL upper bound */

needs a rebase :-/

Lucas De Marchi

Quoted 25 lines
drivers/gpu/drm/i915/gt/intel_sseu.h +1 -1
-#define GEN_MAX_SUBSLICES	(8) /* ICL upper bound */
+#define GEN_MAX_SUBSLICES	(32) /* XEHPSDV upper bound */
 #define GEN_SSEU_STRIDE(max_entries) DIV_ROUND_UP(max_entries, BITS_PER_BYTE)
 #define GEN_MAX_SUBSLICE_STRIDE GEN_SSEU_STRIDE(GEN_MAX_SUBSLICES)
 #define GEN_MAX_EUS		(16) /* TGL upper bound */
drivers/gpu/drm/i915/gt/intel_sseu_debugfs.c +1 -1
index 714fe8495775..a424150b052e 100644
@@ -53,7 +53,7 @@ static void cherryview_sseu_device_status(struct intel_gt *gt,
 static void gen10_sseu_device_status(struct intel_gt *gt,
 				     struct sseu_dev_info *sseu)
 {
-#define SS_MAX 6
+#define SS_MAX 8
 	struct intel_uncore *uncore = gt->uncore;
 	const struct intel_gt_info *info = &gt->info;
 	u32 s_reg[SS_MAX], eu_reg[2 * SS_MAX], eu_mask[2];
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#18 lore raw

PATCH v4 10/18 drm/i915/xehpsdv: Add compute DSS type

In reply to Matt Roper

From: Stuart Summers <[email protected]>

Starting in XeHP, the concept of slice has been removed in favor of DSS (Dual-Subslice) masks for various workload types. These workloads have been divided into those enabled for geometry and those enabled for compute.

i915 currently maintains a single set of S/SS/EU masks for the device. The goal of this patch set is to minimize the amount of impact to prior generations while still giving the user maximum flexibility.

Bspec: 33117, 33118, 20376

---
 drivers/gpu/drm/i915/gt/intel_sseu.c | 73 ++++++++++++++++++++--------
 drivers/gpu/drm/i915/gt/intel_sseu.h |  5 +-
 drivers/gpu/drm/i915/i915_reg.h      |  3 +-
 include/uapi/drm/i915_drm.h          |  3 --
 4 files changed, 59 insertions(+), 25 deletions(-)
drivers/gpu/drm/i915/gt/intel_sseu.c +53 -20
index 16c0552fcd1d..5d3b8dff464c 100644
@@ -46,11 +46,11 @@ u32 intel_sseu_get_subslices(const struct sseu_dev_info *sseu, u8 slice)
 }
 
 void intel_sseu_set_subslices(struct sseu_dev_info *sseu, int slice,
-			      u32 ss_mask)
+			      u8 *subslice_mask, u32 ss_mask)
 {
 	int offset = slice * sseu->ss_stride;
 
-	memcpy(&sseu->subslice_mask[offset], &ss_mask, sseu->ss_stride);
+	memcpy(&subslice_mask[offset], &ss_mask, sseu->ss_stride);
 }
 
 unsigned int
@@ -100,14 +100,24 @@ static u16 compute_eu_total(const struct sseu_dev_info *sseu)
 	return total;
 }
 
-static void gen11_compute_sseu_info(struct sseu_dev_info *sseu,
-				    u8 s_en, u32 ss_en, u16 eu_en)
+static u32 get_ss_stride_mask(struct sseu_dev_info *sseu, u8 s, u32 ss_en)
+{
+	u32 ss_mask;
+
+	ss_mask = ss_en >> (s * sseu->max_subslices);
+	ss_mask &= GENMASK(sseu->max_subslices - 1, 0);
+
+	return ss_mask;
+}
+
+static void gen11_compute_sseu_info(struct sseu_dev_info *sseu, u8 s_en,
+				    u32 g_ss_en, u32 c_ss_en, u16 eu_en)
 {
 	int s, ss;
 
-	/* ss_en represents entire subslice mask across all slices */
+	/* g_ss_en/c_ss_en represent entire subslice mask across all slices */
 	GEM_BUG_ON(sseu->max_slices * sseu->max_subslices >
-		   sizeof(ss_en) * BITS_PER_BYTE);
+		   sizeof(g_ss_en) * BITS_PER_BYTE);
 
 	for (s = 0; s < sseu->max_slices; s++) {
 		if ((s_en & BIT(s)) == 0)
@@ -115,7 +125,23 @@ static void gen11_compute_sseu_info(struct sseu_dev_info *sseu,
 
 		sseu->slice_mask |= BIT(s);
 
-		intel_sseu_set_subslices(sseu, s, ss_en);
+		/*
+		 * XeHP introduces the concept of compute vs
+		 * geometry DSS. To reduce variation between GENs
+		 * around subslice usage, store a mask for both the
+		 * geometry and compute enabled masks, to provide
+		 * to user space later in QUERY_TOPOLOGY_INFO, and
+		 * compute a total enabled subslice count for the
+		 * purposes of selecting subslices to use in a
+		 * particular GEM context.
+		 */
+		intel_sseu_set_subslices(sseu, s, sseu->compute_subslice_mask,
+					 get_ss_stride_mask(sseu, s, c_ss_en));
+		intel_sseu_set_subslices(sseu, s, sseu->geometry_subslice_mask,
+					 get_ss_stride_mask(sseu, s, g_ss_en));
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 get_ss_stride_mask(sseu, s,
+							    g_ss_en | c_ss_en));
 
 		for (ss = 0; ss < sseu->max_subslices; ss++)
 			if (intel_sseu_has_subslice(sseu, s, ss))
@@ -129,7 +155,7 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 {
 	struct sseu_dev_info *sseu = &gt->info.sseu;
 	struct intel_uncore *uncore = gt->uncore;
-	u32 dss_en;
+	u32 g_dss_en, c_dss_en = 0;
 	u16 eu_en = 0;
 	u8 eu_en_fuse;
 	u8 s_en;
@@ -145,10 +171,12 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 	 * across the entire device. Then calculate out the DSS for each
 	 * workload type within that software slice.
 	 */
-	if (IS_XEHPSDV(gt->i915))
+	if (IS_XEHPSDV(gt->i915)) {
 		intel_sseu_set_info(sseu, 1, 32, 16);
-	else
+		sseu->has_compute_dss = 1;
+	} else {
 		intel_sseu_set_info(sseu, 1, 6, 16);
+	}
 
 	/*
 	 * As mentioned above, Xe_HP does not have the concept of a slice.
@@ -160,7 +188,9 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 		s_en = intel_uncore_read(uncore, GEN11_GT_SLICE_ENABLE) &
 		       GEN11_GT_S_ENA_MASK;
 
-	dss_en = intel_uncore_read(uncore, GEN12_GT_DSS_ENABLE);
+	g_dss_en = intel_uncore_read(uncore, GEN12_GT_GEOMETRY_DSS_ENABLE);
+	if (sseu->has_compute_dss)
+		c_dss_en = intel_uncore_read(uncore, GEN12_GT_COMPUTE_DSS_ENABLE);
 
 	/* one bit per pair of EUs */
 	if (GRAPHICS_VER_FULL(gt->i915) >= IP_VER(12, 50))
@@ -173,7 +203,7 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 		if (eu_en_fuse & BIT(eu))
 			eu_en |= BIT(eu * 2) | BIT(eu * 2 + 1);
 
-	gen11_compute_sseu_info(sseu, s_en, dss_en, eu_en);
+	gen11_compute_sseu_info(sseu, s_en, g_dss_en, c_dss_en, eu_en);
 
 	/* TGL only supports slice-level power gating */
 	sseu->has_slice_pg = 1;
@@ -199,7 +229,7 @@ static void gen11_sseu_info_init(struct intel_gt *gt)
 	eu_en = ~(intel_uncore_read(uncore, GEN11_EU_DISABLE) &
 		  GEN11_EU_DIS_MASK);
 
-	gen11_compute_sseu_info(sseu, s_en, ss_en, eu_en);
+	gen11_compute_sseu_info(sseu, s_en, ss_en, 0, eu_en);
 
 	/* ICL has no power gating restrictions. */
 	sseu->has_slice_pg = 1;
@@ -260,9 +290,9 @@ static void gen10_sseu_info_init(struct intel_gt *gt)
 		 * Slice0 can have up to 3 subslices, but there are only 2 in
 		 * slice1/2.
 		 */
-		intel_sseu_set_subslices(sseu, s, s == 0 ?
-					 subslice_mask_with_eus :
-					 subslice_mask_with_eus & 0x3);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 s == 0 ? subslice_mask_with_eus :
+						  subslice_mask_with_eus & 0x3);
 	}
 
 	sseu->eu_total = compute_eu_total(sseu);
@@ -317,7 +347,7 @@ static void cherryview_sseu_info_init(struct intel_gt *gt)
 		sseu_set_eus(sseu, 0, 1, ~disabled_mask);
 	}
 
-	intel_sseu_set_subslices(sseu, 0, subslice_mask);
+	intel_sseu_set_subslices(sseu, 0, sseu->subslice_mask, subslice_mask);
 
 	sseu->eu_total = compute_eu_total(sseu);
 
@@ -373,7 +403,8 @@ static void gen9_sseu_info_init(struct intel_gt *gt)
 			/* skip disabled slice */
 			continue;
 
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		eu_disable = intel_uncore_read(uncore, GEN9_EU_DISABLE(s));
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
@@ -485,7 +516,8 @@ static void bdw_sseu_info_init(struct intel_gt *gt)
 			/* skip disabled slice */
 			continue;
 
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
 			u8 eu_disabled_mask;
@@ -583,7 +615,8 @@ static void hsw_sseu_info_init(struct intel_gt *gt)
 			    sseu->eu_per_subslice);
 
 	for (s = 0; s < sseu->max_slices; s++) {
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
 			sseu_set_eus(sseu, s, ss,
drivers/gpu/drm/i915/gt/intel_sseu.h +4 -1
index 204ea6709460..b383e7d97554 100644
@@ -32,6 +32,8 @@ struct drm_printer;
 struct sseu_dev_info {
 	u8 slice_mask;
 	u8 subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
+	u8 geometry_subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
+	u8 compute_subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
 	u8 eu_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICES * GEN_MAX_EU_STRIDE];
 	u16 eu_total;
 	u8 eu_per_subslice;
@@ -41,6 +43,7 @@ struct sseu_dev_info {
 	u8 has_slice_pg:1;
 	u8 has_subslice_pg:1;
 	u8 has_eu_pg:1;
+	u8 has_compute_dss:1;
 
 	/* Topology fields */
 	u8 max_slices;
@@ -104,7 +107,7 @@ intel_sseu_subslices_per_slice(const struct sseu_dev_info *sseu, u8 slice);
 u32  intel_sseu_get_subslices(const struct sseu_dev_info *sseu, u8 slice);
 
 void intel_sseu_set_subslices(struct sseu_dev_info *sseu, int slice,
-			      u32 ss_mask);
+			      u8 *subslice_mask, u32 ss_mask);
 
 void intel_sseu_info_init(struct intel_gt *gt);
 
drivers/gpu/drm/i915/i915_reg.h +2 -1
index 99858bc593f0..d7e4418955f7 100644
@@ -3223,7 +3223,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 
 #define GEN11_GT_SUBSLICE_DISABLE _MMIO(0x913C)
 
-#define GEN12_GT_DSS_ENABLE _MMIO(0x913C)
+#define GEN12_GT_GEOMETRY_DSS_ENABLE _MMIO(0x913C)
+#define GEN12_GT_COMPUTE_DSS_ENABLE _MMIO(0x9144)
 
 #define XEHP_EU_ENABLE			_MMIO(0x9134)
 #define XEHP_EU_ENA_MASK		0xFF
include/uapi/drm/i915_drm.h +0 -3
index 7f13d241417f..aef15542a95b 100644
@@ -2589,9 +2589,6 @@ struct drm_i915_query {
  *                 Z / 8] >> (Z % 8)) & 1
  */
 struct drm_i915_query_topology_info {
-	/*
-	 * Unused for now. Must be cleared to zero.
-	 */
 	__u16 flags;
 
 	__u16 max_slices;
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#19 lore raw

Re: [Intel-gfx] [PATCH v4 10/18] drm/i915/xehpsdv: Add compute DSS type

Quoted 71 lines · Matt Roper

From: Stuart Summers <[email protected]>

Starting in XeHP, the concept of slice has been removed in favor of DSS (Dual-Subslice) masks for various workload types. These workloads have been divided into those enabled for geometry and those enabled for compute.

i915 currently maintains a single set of S/SS/EU masks for the device. The goal of this patch set is to minimize the amount of impact to prior generations while still giving the user maximum flexibility.

Bspec: 33117, 33118, 20376

---

drivers/gpu/drm/i915/gt/intel_sseu.c | 73 ++++++++++++++++++++--------
drivers/gpu/drm/i915/gt/intel_sseu.h |  5 +-
drivers/gpu/drm/i915/i915_reg.h      |  3 +-
include/uapi/drm/i915_drm.h          |  3 --
4 files changed, 59 insertions(+), 25 deletions(-)
drivers/gpu/drm/i915/gt/intel_sseu.c +2 -2
index 16c0552fcd1d..5d3b8dff464c 100644
@@ -46,11 +46,11 @@ u32 intel_sseu_get_subslices(const struct sseu_dev_info *sseu, u8 slice)
 }
 
 void intel_sseu_set_subslices(struct sseu_dev_info *sseu, int slice,
-			      u32 ss_mask)
+			      u8 *subslice_mask, u32 ss_mask)
 {
 	int offset = slice * sseu->ss_stride;
 
-	memcpy(&sseu->subslice_mask[offset], &ss_mask, sseu->ss_stride);
+	memcpy(&subslice_mask[offset], &ss_mask, sseu->ss_stride);
 }
 

unsigned int

drivers/gpu/drm/i915/gt/intel_sseu.c +14 -4
@@ -100,14 +100,24 @@ static u16 compute_eu_total(const struct sseu_dev_info *sseu)
 	return total;
 }
 
-static void gen11_compute_sseu_info(struct sseu_dev_info *sseu,
-				    u8 s_en, u32 ss_en, u16 eu_en)
+static u32 get_ss_stride_mask(struct sseu_dev_info *sseu, u8 s, u32 ss_en)
+{
+	u32 ss_mask;
+
+	ss_mask = ss_en >> (s * sseu->max_subslices);
+	ss_mask &= GENMASK(sseu->max_subslices - 1, 0);
+
+	return ss_mask;
+}
+
+static void gen11_compute_sseu_info(struct sseu_dev_info *sseu, u8 s_en,
+				    u32 g_ss_en, u32 c_ss_en, u16 eu_en)
 {
 	int s, ss;
 
-	/* ss_en represents entire subslice mask across all slices */
+	/* g_ss_en/c_ss_en represent entire subslice mask across all slices */
 	GEM_BUG_ON(sseu->max_slices * sseu->max_subslices >
-		   sizeof(ss_en) * BITS_PER_BYTE);
+		   sizeof(g_ss_en) * BITS_PER_BYTE);
 
 	for (s = 0; s < sseu->max_slices; s++) {
 		if ((s_en & BIT(s)) == 0)

drivers/gpu/drm/i915/gt/intel_sseu.c static void gen11_compute_sseu_info(struct sseu_dev_info *sseu, · line 133

Quoted 10 lines
drivers/gpu/drm/i915/gt/intel_sseu.c +6 -1
@@ -115,7 +125,23 @@ static void gen11_compute_sseu_info(struct sseu_dev_info *sseu,
 
 		sseu->slice_mask |= BIT(s);
 
-		intel_sseu_set_subslices(sseu, s, ss_en);
+		/*
+		 * XeHP introduces the concept of compute vs
+		 * geometry DSS. To reduce variation between GENs
+		 * around subslice usage, store a mask for both the
+		 * geometry and compute enabled masks, to provide
+		 * to user space later in QUERY_TOPOLOGY_INFO, and

this is not true anymore... I think when this was written the idea was to use QUERY_TOPOLOGY_INFO's flags field to differentiate between compute/geometry. However looking at next patches it seems this is not there anymore: we still check for

        if (query_item->flags != 0)
                return -EINVAL;

Cc'ing Stuart and Matt Atwood

Also, it would be good to have the patches adding the query as the very next patch.

Lucas De Marchi

Quoted 184 lines
drivers/gpu/drm/i915/gt/intel_sseu.c +31 -13
+		 * compute a total enabled subslice count for the
+		 * purposes of selecting subslices to use in a
+		 * particular GEM context.
+		 */
+		intel_sseu_set_subslices(sseu, s, sseu->compute_subslice_mask,
+					 get_ss_stride_mask(sseu, s, c_ss_en));
+		intel_sseu_set_subslices(sseu, s, sseu->geometry_subslice_mask,
+					 get_ss_stride_mask(sseu, s, g_ss_en));
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 get_ss_stride_mask(sseu, s,
+							    g_ss_en | c_ss_en));
 
 		for (ss = 0; ss < sseu->max_subslices; ss++)
 			if (intel_sseu_has_subslice(sseu, s, ss))
@@ -129,7 +155,7 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 {
 	struct sseu_dev_info *sseu = &gt->info.sseu;
 	struct intel_uncore *uncore = gt->uncore;
-	u32 dss_en;
+	u32 g_dss_en, c_dss_en = 0;
 	u16 eu_en = 0;
 	u8 eu_en_fuse;
 	u8 s_en;
@@ -145,10 +171,12 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 	 * across the entire device. Then calculate out the DSS for each
 	 * workload type within that software slice.
 	 */
-	if (IS_XEHPSDV(gt->i915))
+	if (IS_XEHPSDV(gt->i915)) {
 		intel_sseu_set_info(sseu, 1, 32, 16);
-	else
+		sseu->has_compute_dss = 1;
+	} else {
 		intel_sseu_set_info(sseu, 1, 6, 16);
+	}
 
 	/*
 	 * As mentioned above, Xe_HP does not have the concept of a slice.
@@ -160,7 +188,9 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 		s_en = intel_uncore_read(uncore, GEN11_GT_SLICE_ENABLE) &
 		       GEN11_GT_S_ENA_MASK;
 
-	dss_en = intel_uncore_read(uncore, GEN12_GT_DSS_ENABLE);
+	g_dss_en = intel_uncore_read(uncore, GEN12_GT_GEOMETRY_DSS_ENABLE);
+	if (sseu->has_compute_dss)
+		c_dss_en = intel_uncore_read(uncore, GEN12_GT_COMPUTE_DSS_ENABLE);
 
 	/* one bit per pair of EUs */
 	if (GRAPHICS_VER_FULL(gt->i915) >= IP_VER(12, 50))
@@ -173,7 +203,7 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 		if (eu_en_fuse & BIT(eu))
 			eu_en |= BIT(eu * 2) | BIT(eu * 2 + 1);
 
-	gen11_compute_sseu_info(sseu, s_en, dss_en, eu_en);
+	gen11_compute_sseu_info(sseu, s_en, g_dss_en, c_dss_en, eu_en);
 
 	/* TGL only supports slice-level power gating */
 	sseu->has_slice_pg = 1;
@@ -199,7 +229,7 @@ static void gen11_sseu_info_init(struct intel_gt *gt)
 	eu_en = ~(intel_uncore_read(uncore, GEN11_EU_DISABLE) &
 		  GEN11_EU_DIS_MASK);
 
-	gen11_compute_sseu_info(sseu, s_en, ss_en, eu_en);
+	gen11_compute_sseu_info(sseu, s_en, ss_en, 0, eu_en);
 
 	/* ICL has no power gating restrictions. */
 	sseu->has_slice_pg = 1;
@@ -260,9 +290,9 @@ static void gen10_sseu_info_init(struct intel_gt *gt)
 		 * Slice0 can have up to 3 subslices, but there are only 2 in
 		 * slice1/2.
 		 */
-		intel_sseu_set_subslices(sseu, s, s == 0 ?
-					 subslice_mask_with_eus :
-					 subslice_mask_with_eus & 0x3);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 s == 0 ? subslice_mask_with_eus :
+						  subslice_mask_with_eus & 0x3);
 	}
 
 	sseu->eu_total = compute_eu_total(sseu);
@@ -317,7 +347,7 @@ static void cherryview_sseu_info_init(struct intel_gt *gt)
 		sseu_set_eus(sseu, 0, 1, ~disabled_mask);
 	}
 
-	intel_sseu_set_subslices(sseu, 0, subslice_mask);
+	intel_sseu_set_subslices(sseu, 0, sseu->subslice_mask, subslice_mask);
 
 	sseu->eu_total = compute_eu_total(sseu);
 
@@ -373,7 +403,8 @@ static void gen9_sseu_info_init(struct intel_gt *gt)
 			/* skip disabled slice */
 			continue;
 
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		eu_disable = intel_uncore_read(uncore, GEN9_EU_DISABLE(s));
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
@@ -485,7 +516,8 @@ static void bdw_sseu_info_init(struct intel_gt *gt)
 			/* skip disabled slice */
 			continue;
 
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
 			u8 eu_disabled_mask;
@@ -583,7 +615,8 @@ static void hsw_sseu_info_init(struct intel_gt *gt)
 			    sseu->eu_per_subslice);
 
 	for (s = 0; s < sseu->max_slices; s++) {
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
 			sseu_set_eus(sseu, s, ss,
drivers/gpu/drm/i915/gt/intel_sseu.h +4 -1
index 204ea6709460..b383e7d97554 100644
@@ -32,6 +32,8 @@ struct drm_printer;
 struct sseu_dev_info {
 	u8 slice_mask;
 	u8 subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
+	u8 geometry_subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
+	u8 compute_subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
 	u8 eu_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICES * GEN_MAX_EU_STRIDE];
 	u16 eu_total;
 	u8 eu_per_subslice;
@@ -41,6 +43,7 @@ struct sseu_dev_info {
 	u8 has_slice_pg:1;
 	u8 has_subslice_pg:1;
 	u8 has_eu_pg:1;
+	u8 has_compute_dss:1;
 
 	/* Topology fields */
 	u8 max_slices;
@@ -104,7 +107,7 @@ intel_sseu_subslices_per_slice(const struct sseu_dev_info *sseu, u8 slice);
 u32  intel_sseu_get_subslices(const struct sseu_dev_info *sseu, u8 slice);
 
 void intel_sseu_set_subslices(struct sseu_dev_info *sseu, int slice,
-			      u32 ss_mask);
+			      u8 *subslice_mask, u32 ss_mask);
 
 void intel_sseu_info_init(struct intel_gt *gt);
 
drivers/gpu/drm/i915/i915_reg.h +2 -1
index 99858bc593f0..d7e4418955f7 100644
@@ -3223,7 +3223,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 
 #define GEN11_GT_SUBSLICE_DISABLE _MMIO(0x913C)
 
-#define GEN12_GT_DSS_ENABLE _MMIO(0x913C)
+#define GEN12_GT_GEOMETRY_DSS_ENABLE _MMIO(0x913C)
+#define GEN12_GT_COMPUTE_DSS_ENABLE _MMIO(0x9144)
 
 #define XEHP_EU_ENABLE			_MMIO(0x9134)
 #define XEHP_EU_ENA_MASK		0xFF
include/uapi/drm/i915_drm.h +0 -3
index 7f13d241417f..aef15542a95b 100644
@@ -2589,9 +2589,6 @@ struct drm_i915_query {
 *                 Z / 8] >> (Z % 8)) & 1
 */
 struct drm_i915_query_topology_info {
-	/*
-	 * Unused for now. Must be cleared to zero.
-	 */
 	__u16 flags;
 
 	__u16 max_slices;
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#20 lore raw

Re: [Intel-gfx] [PATCH v4 10/18] drm/i915/xehpsdv: Add compute DSS type

Quoted 89 lines · Lucas De Marchi
Quoted 71 lines · Matt Roper

From: Stuart Summers <[email protected]>

Starting in XeHP, the concept of slice has been removed in favor of DSS (Dual-Subslice) masks for various workload types. These workloads have been divided into those enabled for geometry and those enabled for compute.

i915 currently maintains a single set of S/SS/EU masks for the device. The goal of this patch set is to minimize the amount of impact to prior generations while still giving the user maximum flexibility.

Bspec: 33117, 33118, 20376

---

drivers/gpu/drm/i915/gt/intel_sseu.c | 73 ++++++++++++++++++++--------
drivers/gpu/drm/i915/gt/intel_sseu.h |  5 +-
drivers/gpu/drm/i915/i915_reg.h      |  3 +-
include/uapi/drm/i915_drm.h          |  3 --
4 files changed, 59 insertions(+), 25 deletions(-)
drivers/gpu/drm/i915/gt/intel_sseu.c +2 -2
index 16c0552fcd1d..5d3b8dff464c 100644
@@ -46,11 +46,11 @@ u32 intel_sseu_get_subslices(const struct sseu_dev_info *sseu, u8 slice)
 }
 
 void intel_sseu_set_subslices(struct sseu_dev_info *sseu, int slice,
-			      u32 ss_mask)
+			      u8 *subslice_mask, u32 ss_mask)
 {
 	int offset = slice * sseu->ss_stride;
 
-	memcpy(&sseu->subslice_mask[offset], &ss_mask, sseu->ss_stride);
+	memcpy(&subslice_mask[offset], &ss_mask, sseu->ss_stride);
 }
 

unsigned int

drivers/gpu/drm/i915/gt/intel_sseu.c +14 -4
@@ -100,14 +100,24 @@ static u16 compute_eu_total(const struct sseu_dev_info *sseu)
 	return total;
 }
 
-static void gen11_compute_sseu_info(struct sseu_dev_info *sseu,
-				    u8 s_en, u32 ss_en, u16 eu_en)
+static u32 get_ss_stride_mask(struct sseu_dev_info *sseu, u8 s, u32 ss_en)
+{
+	u32 ss_mask;
+
+	ss_mask = ss_en >> (s * sseu->max_subslices);
+	ss_mask &= GENMASK(sseu->max_subslices - 1, 0);
+
+	return ss_mask;
+}
+
+static void gen11_compute_sseu_info(struct sseu_dev_info *sseu, u8 s_en,
+				    u32 g_ss_en, u32 c_ss_en, u16 eu_en)
 {
 	int s, ss;
 
-	/* ss_en represents entire subslice mask across all slices */
+	/* g_ss_en/c_ss_en represent entire subslice mask across all slices */
 	GEM_BUG_ON(sseu->max_slices * sseu->max_subslices >
-		   sizeof(ss_en) * BITS_PER_BYTE);
+		   sizeof(g_ss_en) * BITS_PER_BYTE);
 
 	for (s = 0; s < sseu->max_slices; s++) {
 		if ((s_en & BIT(s)) == 0)

drivers/gpu/drm/i915/gt/intel_sseu.c static void gen11_compute_sseu_info(struct sseu_dev_info *sseu, · line 133

Quoted 10 lines
drivers/gpu/drm/i915/gt/intel_sseu.c +6 -1
@@ -115,7 +125,23 @@ static void gen11_compute_sseu_info(struct sseu_dev_info *sseu,
 
 		sseu->slice_mask |= BIT(s);
 
-		intel_sseu_set_subslices(sseu, s, ss_en);
+		/*
+		 * XeHP introduces the concept of compute vs
+		 * geometry DSS. To reduce variation between GENs
+		 * around subslice usage, store a mask for both the
+		 * geometry and compute enabled masks, to provide
+		 * to user space later in QUERY_TOPOLOGY_INFO, and

this is not true anymore... I think when this was written the idea was to use QUERY_TOPOLOGY_INFO's flags field to differentiate between compute/geometry. However looking at next patches it seems this is not there anymore: we still check for

Quoted 8 lines
        if (query_item->flags != 0)
                return -EINVAL;

Cc'ing Stuart and Matt Atwood

Also, it would be good to have the patches adding the query as the very next patch.

I held off on including any of the UAPI changes in this series since those will require UMD availability too. I know the OCL driver has started landing some support for XeHP SDV, but from a quick skim I didn't see the uapi-related code in what they've made available so far.

Matt

Quoted 187 lines

Lucas De Marchi

Quoted 184 lines
drivers/gpu/drm/i915/gt/intel_sseu.c +31 -13
+		 * compute a total enabled subslice count for the
+		 * purposes of selecting subslices to use in a
+		 * particular GEM context.
+		 */
+		intel_sseu_set_subslices(sseu, s, sseu->compute_subslice_mask,
+					 get_ss_stride_mask(sseu, s, c_ss_en));
+		intel_sseu_set_subslices(sseu, s, sseu->geometry_subslice_mask,
+					 get_ss_stride_mask(sseu, s, g_ss_en));
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 get_ss_stride_mask(sseu, s,
+							    g_ss_en | c_ss_en));
 
 		for (ss = 0; ss < sseu->max_subslices; ss++)
 			if (intel_sseu_has_subslice(sseu, s, ss))
@@ -129,7 +155,7 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 {
 	struct sseu_dev_info *sseu = &gt->info.sseu;
 	struct intel_uncore *uncore = gt->uncore;
-	u32 dss_en;
+	u32 g_dss_en, c_dss_en = 0;
 	u16 eu_en = 0;
 	u8 eu_en_fuse;
 	u8 s_en;
@@ -145,10 +171,12 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 	 * across the entire device. Then calculate out the DSS for each
 	 * workload type within that software slice.
 	 */
-	if (IS_XEHPSDV(gt->i915))
+	if (IS_XEHPSDV(gt->i915)) {
 		intel_sseu_set_info(sseu, 1, 32, 16);
-	else
+		sseu->has_compute_dss = 1;
+	} else {
 		intel_sseu_set_info(sseu, 1, 6, 16);
+	}
 
 	/*
 	 * As mentioned above, Xe_HP does not have the concept of a slice.
@@ -160,7 +188,9 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 		s_en = intel_uncore_read(uncore, GEN11_GT_SLICE_ENABLE) &
 		       GEN11_GT_S_ENA_MASK;
 
-	dss_en = intel_uncore_read(uncore, GEN12_GT_DSS_ENABLE);
+	g_dss_en = intel_uncore_read(uncore, GEN12_GT_GEOMETRY_DSS_ENABLE);
+	if (sseu->has_compute_dss)
+		c_dss_en = intel_uncore_read(uncore, GEN12_GT_COMPUTE_DSS_ENABLE);
 
 	/* one bit per pair of EUs */
 	if (GRAPHICS_VER_FULL(gt->i915) >= IP_VER(12, 50))
@@ -173,7 +203,7 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 		if (eu_en_fuse & BIT(eu))
 			eu_en |= BIT(eu * 2) | BIT(eu * 2 + 1);
 
-	gen11_compute_sseu_info(sseu, s_en, dss_en, eu_en);
+	gen11_compute_sseu_info(sseu, s_en, g_dss_en, c_dss_en, eu_en);
 
 	/* TGL only supports slice-level power gating */
 	sseu->has_slice_pg = 1;
@@ -199,7 +229,7 @@ static void gen11_sseu_info_init(struct intel_gt *gt)
 	eu_en = ~(intel_uncore_read(uncore, GEN11_EU_DISABLE) &
 		  GEN11_EU_DIS_MASK);
 
-	gen11_compute_sseu_info(sseu, s_en, ss_en, eu_en);
+	gen11_compute_sseu_info(sseu, s_en, ss_en, 0, eu_en);
 
 	/* ICL has no power gating restrictions. */
 	sseu->has_slice_pg = 1;
@@ -260,9 +290,9 @@ static void gen10_sseu_info_init(struct intel_gt *gt)
 		 * Slice0 can have up to 3 subslices, but there are only 2 in
 		 * slice1/2.
 		 */
-		intel_sseu_set_subslices(sseu, s, s == 0 ?
-					 subslice_mask_with_eus :
-					 subslice_mask_with_eus & 0x3);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 s == 0 ? subslice_mask_with_eus :
+						  subslice_mask_with_eus & 0x3);
 	}
 
 	sseu->eu_total = compute_eu_total(sseu);
@@ -317,7 +347,7 @@ static void cherryview_sseu_info_init(struct intel_gt *gt)
 		sseu_set_eus(sseu, 0, 1, ~disabled_mask);
 	}
 
-	intel_sseu_set_subslices(sseu, 0, subslice_mask);
+	intel_sseu_set_subslices(sseu, 0, sseu->subslice_mask, subslice_mask);
 
 	sseu->eu_total = compute_eu_total(sseu);
 
@@ -373,7 +403,8 @@ static void gen9_sseu_info_init(struct intel_gt *gt)
 			/* skip disabled slice */
 			continue;
 
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		eu_disable = intel_uncore_read(uncore, GEN9_EU_DISABLE(s));
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
@@ -485,7 +516,8 @@ static void bdw_sseu_info_init(struct intel_gt *gt)
 			/* skip disabled slice */
 			continue;
 
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
 			u8 eu_disabled_mask;
@@ -583,7 +615,8 @@ static void hsw_sseu_info_init(struct intel_gt *gt)
 			    sseu->eu_per_subslice);
 
 	for (s = 0; s < sseu->max_slices; s++) {
-		intel_sseu_set_subslices(sseu, s, subslice_mask);
+		intel_sseu_set_subslices(sseu, s, sseu->subslice_mask,
+					 subslice_mask);
 
 		for (ss = 0; ss < sseu->max_subslices; ss++) {
 			sseu_set_eus(sseu, s, ss,
drivers/gpu/drm/i915/gt/intel_sseu.h +4 -1
index 204ea6709460..b383e7d97554 100644
@@ -32,6 +32,8 @@ struct drm_printer;
 struct sseu_dev_info {
 	u8 slice_mask;
 	u8 subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
+	u8 geometry_subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
+	u8 compute_subslice_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICE_STRIDE];
 	u8 eu_mask[GEN_MAX_SLICES * GEN_MAX_SUBSLICES * GEN_MAX_EU_STRIDE];
 	u16 eu_total;
 	u8 eu_per_subslice;
@@ -41,6 +43,7 @@ struct sseu_dev_info {
 	u8 has_slice_pg:1;
 	u8 has_subslice_pg:1;
 	u8 has_eu_pg:1;
+	u8 has_compute_dss:1;
 
 	/* Topology fields */
 	u8 max_slices;
@@ -104,7 +107,7 @@ intel_sseu_subslices_per_slice(const struct sseu_dev_info *sseu, u8 slice);
 u32  intel_sseu_get_subslices(const struct sseu_dev_info *sseu, u8 slice);
 
 void intel_sseu_set_subslices(struct sseu_dev_info *sseu, int slice,
-			      u32 ss_mask);
+			      u8 *subslice_mask, u32 ss_mask);
 
 void intel_sseu_info_init(struct intel_gt *gt);
 
drivers/gpu/drm/i915/i915_reg.h +2 -1
index 99858bc593f0..d7e4418955f7 100644
@@ -3223,7 +3223,8 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 
 #define GEN11_GT_SUBSLICE_DISABLE _MMIO(0x913C)
 
-#define GEN12_GT_DSS_ENABLE _MMIO(0x913C)
+#define GEN12_GT_GEOMETRY_DSS_ENABLE _MMIO(0x913C)
+#define GEN12_GT_COMPUTE_DSS_ENABLE _MMIO(0x9144)
 
 #define XEHP_EU_ENABLE			_MMIO(0x9134)
 #define XEHP_EU_ENA_MASK		0xFF
include/uapi/drm/i915_drm.h +0 -3
index 7f13d241417f..aef15542a95b 100644
@@ -2589,9 +2589,6 @@ struct drm_i915_query {
 *                 Z / 8] >> (Z % 8)) & 1
 */
 struct drm_i915_query_topology_info {
-	/*
-	 * Unused for now. Must be cleared to zero.
-	 */
 	__u16 flags;
 
 	__u16 max_slices;
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 5 lines
Matt Roper
Graphics Software Engineer
VTT-OSGC Platform Enablement
Intel Corporation
(916) 356-2795
author
#21 lore raw

PATCH v4 11/18 drm/i915/dg2: DG2 uses the same sseu limits as XeHP SDV

In reply to Matt Roper

DG2 supports compute DSS and has the same maximum number of DSS and EU as XeHP SDV.

---
 drivers/gpu/drm/i915/gt/intel_sseu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_sseu.c +1 -1
index 5d3b8dff464c..eaff221db5b0 100644
@@ -171,7 +171,7 @@ static void gen12_sseu_info_init(struct intel_gt *gt)
 	 * across the entire device. Then calculate out the DSS for each
 	 * workload type within that software slice.
 	 */
-	if (IS_XEHPSDV(gt->i915)) {
+	if (IS_DG2(gt->i915) || IS_XEHPSDV(gt->i915)) {
 		intel_sseu_set_info(sseu, 1, 32, 16);
 		sseu->has_compute_dss = 1;
 	} else {
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#22 lore raw

PATCH v4 12/18 drm/i915/xehpsdv: Define MOCS table for XeHP SDV

In reply to Matt Roper

From: Lucas De Marchi <[email protected]>

Like DG1, XeHP SDV doesn't have LLC/eDRAM control values due to being a dgfx card. XeHP SDV adds 2 more bits: L3_GLBGO to "push the Go point to memory for L3 destined transaction" and L3_LKP to "enable Lookup for uncacheable accesses".

Bspec: 45101

---
 drivers/gpu/drm/i915/gt/intel_mocs.c | 33 +++++++++++++++++++++++++++-
 1 file changed, 32 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_mocs.c +32 -1
index 17848807f111..0c9d0b936c20 100644
@@ -40,6 +40,8 @@ struct drm_i915_mocs_table {
 #define L3_ESC(value)		((value) << 0)
 #define L3_SCC(value)		((value) << 1)
 #define _L3_CACHEABILITY(value)	((value) << 4)
+#define L3_GLBGO(value)		((value) << 6)
+#define L3_LKUP(value)		((value) << 7)
 
 /* Helper defines */
 #define GEN9_NUM_MOCS_ENTRIES	64  /* 63-64 are reserved, but configured. */
@@ -314,6 +316,31 @@ static const struct drm_i915_mocs_entry dg1_mocs_table[] = {
 	MOCS_ENTRY(63, 0, L3_1_UC),
 };
 
+static const struct drm_i915_mocs_entry xehpsdv_mocs_table[] = {
+	/* wa_1608975824 */
+	MOCS_ENTRY(0, 0, L3_3_WB | L3_LKUP(1)),
+
+	/* UC - Coherent; GO:L3 */
+	MOCS_ENTRY(1, 0, L3_1_UC | L3_LKUP(1)),
+	/* UC - Coherent; GO:Memory */
+	MOCS_ENTRY(2, 0, L3_1_UC | L3_GLBGO(1) | L3_LKUP(1)),
+	/* UC - Non-Coherent; GO:Memory */
+	MOCS_ENTRY(3, 0, L3_1_UC | L3_GLBGO(1)),
+	/* UC - Non-Coherent; GO:L3 */
+	MOCS_ENTRY(4, 0, L3_1_UC),
+
+	/* WB */
+	MOCS_ENTRY(5, 0, L3_3_WB | L3_LKUP(1)),
+
+	/* HW Reserved - SW program but never use. */
+	MOCS_ENTRY(48, 0, L3_3_WB | L3_LKUP(1)),
+	MOCS_ENTRY(49, 0, L3_1_UC | L3_LKUP(1)),
+	MOCS_ENTRY(60, 0, L3_1_UC),
+	MOCS_ENTRY(61, 0, L3_1_UC),
+	MOCS_ENTRY(62, 0, L3_1_UC),
+	MOCS_ENTRY(63, 0, L3_1_UC),
+};
+
 enum {
 	HAS_GLOBAL_MOCS = BIT(0),
 	HAS_ENGINE_MOCS = BIT(1),
@@ -340,7 +367,11 @@ static unsigned int get_mocs_settings(const struct drm_i915_private *i915,
 {
 	unsigned int flags;
 
-	if (IS_DG1(i915)) {
+	if (IS_XEHPSDV(i915)) {
+		table->size = ARRAY_SIZE(xehpsdv_mocs_table);
+		table->table = xehpsdv_mocs_table;
+		table->n_entries = GEN9_NUM_MOCS_ENTRIES;
+	} else if (IS_DG1(i915)) {
 		table->size = ARRAY_SIZE(dg1_mocs_table);
 		table->table = dg1_mocs_table;
 		table->n_entries = GEN9_NUM_MOCS_ENTRIES;
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#23 lore raw

Re: [Intel-gfx] [PATCH v4 12/18] drm/i915/xehpsdv: Define MOCS table for XeHP SDV

Quoted 7 lines · Matt Roper

From: Lucas De Marchi <[email protected]>

Like DG1, XeHP SDV doesn't have LLC/eDRAM control values due to being a dgfx card. XeHP SDV adds 2 more bits: L3_GLBGO to "push the Go point to memory for L3 destined transaction" and L3_LKP to "enable Lookup for uncacheable accesses".

Quoted 5 lines

Bspec: 45101

+Ayaz, +Daniel

I think this can't land as is since we risk forgetting additional changes that we will have to do. We already made the mistake once of forgetting MOCS changes.

There are some patches to initialize unused MOCS entries and similar that should have been sent already to upstream. Ayaz, what's the state of those patches?

Lucas De Marchi

Quoted 69 lines

---
drivers/gpu/drm/i915/gt/intel_mocs.c | 33 +++++++++++++++++++++++++++-

1 file changed, 32 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_mocs.c +32 -1
index 17848807f111..0c9d0b936c20 100644
@@ -40,6 +40,8 @@ struct drm_i915_mocs_table {
 #define L3_ESC(value)		((value) << 0)
 #define L3_SCC(value)		((value) << 1)
 #define _L3_CACHEABILITY(value)	((value) << 4)
+#define L3_GLBGO(value)		((value) << 6)
+#define L3_LKUP(value)		((value) << 7)
 
 /* Helper defines */
 #define GEN9_NUM_MOCS_ENTRIES	64  /* 63-64 are reserved, but configured. */
@@ -314,6 +316,31 @@ static const struct drm_i915_mocs_entry dg1_mocs_table[] = {
 	MOCS_ENTRY(63, 0, L3_1_UC),
 };
 
+static const struct drm_i915_mocs_entry xehpsdv_mocs_table[] = {
+	/* wa_1608975824 */
+	MOCS_ENTRY(0, 0, L3_3_WB | L3_LKUP(1)),
+
+	/* UC - Coherent; GO:L3 */
+	MOCS_ENTRY(1, 0, L3_1_UC | L3_LKUP(1)),
+	/* UC - Coherent; GO:Memory */
+	MOCS_ENTRY(2, 0, L3_1_UC | L3_GLBGO(1) | L3_LKUP(1)),
+	/* UC - Non-Coherent; GO:Memory */
+	MOCS_ENTRY(3, 0, L3_1_UC | L3_GLBGO(1)),
+	/* UC - Non-Coherent; GO:L3 */
+	MOCS_ENTRY(4, 0, L3_1_UC),
+
+	/* WB */
+	MOCS_ENTRY(5, 0, L3_3_WB | L3_LKUP(1)),
+
+	/* HW Reserved - SW program but never use. */
+	MOCS_ENTRY(48, 0, L3_3_WB | L3_LKUP(1)),
+	MOCS_ENTRY(49, 0, L3_1_UC | L3_LKUP(1)),
+	MOCS_ENTRY(60, 0, L3_1_UC),
+	MOCS_ENTRY(61, 0, L3_1_UC),
+	MOCS_ENTRY(62, 0, L3_1_UC),
+	MOCS_ENTRY(63, 0, L3_1_UC),
+};
+
 enum {
 	HAS_GLOBAL_MOCS = BIT(0),
 	HAS_ENGINE_MOCS = BIT(1),
@@ -340,7 +367,11 @@ static unsigned int get_mocs_settings(const struct drm_i915_private *i915,
 {
 	unsigned int flags;
 
-	if (IS_DG1(i915)) {
+	if (IS_XEHPSDV(i915)) {
+		table->size = ARRAY_SIZE(xehpsdv_mocs_table);
+		table->table = xehpsdv_mocs_table;
+		table->n_entries = GEN9_NUM_MOCS_ENTRIES;
+	} else if (IS_DG1(i915)) {
 		table->size = ARRAY_SIZE(dg1_mocs_table);
 		table->table = dg1_mocs_table;
 		table->n_entries = GEN9_NUM_MOCS_ENTRIES;
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#24 lore raw

Re: [Intel-gfx] [PATCH v4 12/18] drm/i915/xehpsdv: Define MOCS table for XeHP SDV

Quoted 27 lines

-----Original Message-----
From: De Marchi, Lucas <[email protected]>
Sent: Thursday, July 29, 2021 11:01 PM
To: Roper, Matthew D <[email protected]>
Cc: [email protected]; Siddiqui, Ayaz A <[email protected]>; Daniel Vetter <[email protected]>
Subject: Re: [Intel-gfx] [PATCH v4 12/18] drm/i915/xehpsdv: Define MOCS table for XeHP SDV

Quoted 7 lines · Matt Roper

From: Lucas De Marchi <[email protected]>

Like DG1, XeHP SDV doesn't have LLC/eDRAM control values due to being a dgfx card. XeHP SDV adds 2 more bits: L3_GLBGO to "push the Go point to memory for L3 destined transaction" and L3_LKP to "enable Lookup for uncacheable accesses".

Quoted 5 lines

Bspec: 45101

+Ayaz, +Daniel

Quoted 7 lines

I think this can't land as is since we risk forgetting additional changes that we will have to do. We already made the mistake once of forgetting MOCS changes.

There are some patches to initialize unused MOCS entries and similar that should have been sent already to upstream. Ayaz, what's the state of those patches?

There are two section of patches .
1. Programming to CMD_CCTL for with UC this patch should sent for upstream. I'll share the patches for CMD_CCTL programming.
2. Programming of unused/unspecified MOCS index to WB, we observed some regression on media use cases. So we decided to delay upstreaming of those patches till MOCS programming are fixed in UMDs.
Regards
-Ayaz

Quoted 72 lines

Lucas De Marchi

Quoted 2 lines

---
drivers/gpu/drm/i915/gt/intel_mocs.c | 33

+++++++++++++++++++++++++++-

Quoted 8 lines
1 file changed, 32 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_mocs.c +0 -0
index 17848807f111..0c9d0b936c20 100644

drivers/gpu/drm/i915/gt/intel_mocs.c struct drm_i915_mocs_table { · line 47

Quoted 8 lines
drivers/gpu/drm/i915/gt/intel_mocs.c +2 -0
@@ -40,6 +40,8 @@ struct drm_i915_mocs_table {
 #define L3_ESC(value)		((value) << 0)
 #define L3_SCC(value)		((value) << 1)
 #define _L3_CACHEABILITY(value)	((value) << 4)
+#define L3_GLBGO(value)		((value) << 6)
+#define L3_LKUP(value)		((value) << 7)
 
 /* Helper defines */
 #define GEN9_NUM_MOCS_ENTRIES	64  /* 63-64 are reserved, but

configured. */

drivers/gpu/drm/i915/gt/intel_mocs.c static const struct drm_i915_mocs_entry

Quoted 1 line
drivers/gpu/drm/i915/gt/intel_mocs.c +0 -0
@@ -314,6 +316,31 @@ static const struct drm_i915_mocs_entry

dg1_mocs_table[] = {

Quoted 47 lines
drivers/gpu/drm/i915/gt/intel_mocs.c +30 -2
 	MOCS_ENTRY(63, 0, L3_1_UC),
 };
 
+static const struct drm_i915_mocs_entry xehpsdv_mocs_table[] = {
+	/* wa_1608975824 */
+	MOCS_ENTRY(0, 0, L3_3_WB | L3_LKUP(1)),
+
+	/* UC - Coherent; GO:L3 */
+	MOCS_ENTRY(1, 0, L3_1_UC | L3_LKUP(1)),
+	/* UC - Coherent; GO:Memory */
+	MOCS_ENTRY(2, 0, L3_1_UC | L3_GLBGO(1) | L3_LKUP(1)),
+	/* UC - Non-Coherent; GO:Memory */
+	MOCS_ENTRY(3, 0, L3_1_UC | L3_GLBGO(1)),
+	/* UC - Non-Coherent; GO:L3 */
+	MOCS_ENTRY(4, 0, L3_1_UC),
+
+	/* WB */
+	MOCS_ENTRY(5, 0, L3_3_WB | L3_LKUP(1)),
+
+	/* HW Reserved - SW program but never use. */
+	MOCS_ENTRY(48, 0, L3_3_WB | L3_LKUP(1)),
+	MOCS_ENTRY(49, 0, L3_1_UC | L3_LKUP(1)),
+	MOCS_ENTRY(60, 0, L3_1_UC),
+	MOCS_ENTRY(61, 0, L3_1_UC),
+	MOCS_ENTRY(62, 0, L3_1_UC),
+	MOCS_ENTRY(63, 0, L3_1_UC),
+};
+
 enum {
 	HAS_GLOBAL_MOCS = BIT(0),
 	HAS_ENGINE_MOCS = BIT(1),
@@ -340,7 +367,11 @@ static unsigned int get_mocs_settings(const struct
 drm_i915_private *i915,  {
 	unsigned int flags;
 
-	if (IS_DG1(i915)) {
+	if (IS_XEHPSDV(i915)) {
+		table->size = ARRAY_SIZE(xehpsdv_mocs_table);
+		table->table = xehpsdv_mocs_table;
+		table->n_entries = GEN9_NUM_MOCS_ENTRIES;
+	} else if (IS_DG1(i915)) {
 		table->size = ARRAY_SIZE(dg1_mocs_table);
 		table->table = dg1_mocs_table;
 		table->n_entries = GEN9_NUM_MOCS_ENTRIES;
--

2.25.4

Quoted 4 lines
Quoted 4 lines

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#25 lore raw

Re: [Intel-gfx] [PATCH v4 12/18] drm/i915/xehpsdv: Define MOCS table for XeHP SDV

Quoted 34 lines · Ayaz A
Quoted 27 lines

-----Original Message-----
From: De Marchi, Lucas <[email protected]>
Sent: Thursday, July 29, 2021 11:01 PM
To: Roper, Matthew D <[email protected]>
Cc: [email protected]; Siddiqui, Ayaz A <[email protected]>; Daniel Vetter <[email protected]>
Subject: Re: [Intel-gfx] [PATCH v4 12/18] drm/i915/xehpsdv: Define MOCS table for XeHP SDV

Quoted 7 lines · Matt Roper

From: Lucas De Marchi <[email protected]>

Like DG1, XeHP SDV doesn't have LLC/eDRAM control values due to being a dgfx card. XeHP SDV adds 2 more bits: L3_GLBGO to "push the Go point to memory for L3 destined transaction" and L3_LKP to "enable Lookup for uncacheable accesses".

Quoted 5 lines

Bspec: 45101

+Ayaz, +Daniel

Quoted 7 lines

I think this can't land as is since we risk forgetting additional changes that we will have to do. We already made the mistake once of forgetting MOCS changes.

There are some patches to initialize unused MOCS entries and similar that should have been sent already to upstream. Ayaz, what's the state of those patches?

Quoted 6 lines

There are two section of patches .
1. Programming to CMD_CCTL for with UC this patch should sent for upstream. I'll share the patches for CMD_CCTL programming.
2. Programming of unused/unspecified MOCS index to WB, we observed some regression on media use cases. So we decided to delay upstreaming of those patches till MOCS programming are fixed in UMDs.

I don't follow on #2. At the moment we don't have a complete XeHP SDV driver upstream (the device IDs aren't even there and won't be for a while) so there's nothing that can regress yet from an upstream point of view --- you can't run a UMD at all today. If internal builds of the UMDs run against our internal staging trees are still buggy and won't work properly with the proper MOCS table settings, then that just makes it more important to upstream these patches now so that the UMD's don't come to rely on the incorrect table values. If we wait until later after the driver is "complete" and we've already enabled first contact of the UMDs in general, then we won't be able to update the MOCS programming to the proper values anymore without breaking ABI.

Unless you're talking about regressing older platforms? In that case we should be able to restrict the WB programming to just Xe_HP SDV for now; ultimately we'll probably need to figure out some kind of opt-in mechanism for the older platforms, but that can be separate from doing it unconditionally on Xe_HP SDV and beyond I think.

Matt

Quoted 79 lines

Regards
-Ayaz

Quoted 76 lines

Lucas De Marchi

Quoted 2 lines

---
drivers/gpu/drm/i915/gt/intel_mocs.c | 33

+++++++++++++++++++++++++++-

Quoted 8 lines
1 file changed, 32 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_mocs.c +0 -0
index 17848807f111..0c9d0b936c20 100644

drivers/gpu/drm/i915/gt/intel_mocs.c struct drm_i915_mocs_table { · line 47

Quoted 8 lines
drivers/gpu/drm/i915/gt/intel_mocs.c +2 -0
@@ -40,6 +40,8 @@ struct drm_i915_mocs_table {
 #define L3_ESC(value)               ((value) << 0)
 #define L3_SCC(value)               ((value) << 1)
 #define _L3_CACHEABILITY(value)     ((value) << 4)
+#define L3_GLBGO(value)             ((value) << 6)
+#define L3_LKUP(value)              ((value) << 7)
 
 /* Helper defines */
 #define GEN9_NUM_MOCS_ENTRIES       64  /* 63-64 are reserved, but

configured. */

drivers/gpu/drm/i915/gt/intel_mocs.c static const struct drm_i915_mocs_entry

Quoted 1 line
drivers/gpu/drm/i915/gt/intel_mocs.c +0 -0
@@ -314,6 +316,31 @@ static const struct drm_i915_mocs_entry

dg1_mocs_table[] = {

Quoted 51 lines
drivers/gpu/drm/i915/gt/intel_mocs.c +30 -2
    MOCS_ENTRY(63, 0, L3_1_UC),
 };
 
+static const struct drm_i915_mocs_entry xehpsdv_mocs_table[] = {
+    /* wa_1608975824 */
+    MOCS_ENTRY(0, 0, L3_3_WB | L3_LKUP(1)),
+
+    /* UC - Coherent; GO:L3 */
+    MOCS_ENTRY(1, 0, L3_1_UC | L3_LKUP(1)),
+    /* UC - Coherent; GO:Memory */
+    MOCS_ENTRY(2, 0, L3_1_UC | L3_GLBGO(1) | L3_LKUP(1)),
+    /* UC - Non-Coherent; GO:Memory */
+    MOCS_ENTRY(3, 0, L3_1_UC | L3_GLBGO(1)),
+    /* UC - Non-Coherent; GO:L3 */
+    MOCS_ENTRY(4, 0, L3_1_UC),
+
+    /* WB */
+    MOCS_ENTRY(5, 0, L3_3_WB | L3_LKUP(1)),
+
+    /* HW Reserved - SW program but never use. */
+    MOCS_ENTRY(48, 0, L3_3_WB | L3_LKUP(1)),
+    MOCS_ENTRY(49, 0, L3_1_UC | L3_LKUP(1)),
+    MOCS_ENTRY(60, 0, L3_1_UC),
+    MOCS_ENTRY(61, 0, L3_1_UC),
+    MOCS_ENTRY(62, 0, L3_1_UC),
+    MOCS_ENTRY(63, 0, L3_1_UC),
+};
+
 enum {
    HAS_GLOBAL_MOCS = BIT(0),
    HAS_ENGINE_MOCS = BIT(1),
@@ -340,7 +367,11 @@ static unsigned int get_mocs_settings(const struct
 drm_i915_private *i915,  {
    unsigned int flags;
 
-    if (IS_DG1(i915)) {
+    if (IS_XEHPSDV(i915)) {
+            table->size = ARRAY_SIZE(xehpsdv_mocs_table);
+            table->table = xehpsdv_mocs_table;
+            table->n_entries = GEN9_NUM_MOCS_ENTRIES;
+    } else if (IS_DG1(i915)) {
            table->size = ARRAY_SIZE(dg1_mocs_table);
            table->table = dg1_mocs_table;
            table->n_entries = GEN9_NUM_MOCS_ENTRIES;
--

2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Signature · 5 lines
Matt Roper
Graphics Software Engineer
VTT-OSGC Platform Enablement
Intel Corporation
(916) 356-2795
author
#26 lore raw

PATCH v4 13/18 drm/i915/dg2: Define MOCS table for DG2

In reply to Matt Roper

Bspec: 45101, 45427

---
 drivers/gpu/drm/i915/gt/intel_mocs.c | 35 +++++++++++++++++++++++++++-
 1 file changed, 34 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_mocs.c +34 -1
index 0c9d0b936c20..ad021337225d 100644
@@ -341,6 +341,30 @@ static const struct drm_i915_mocs_entry xehpsdv_mocs_table[] = {
 	MOCS_ENTRY(63, 0, L3_1_UC),
 };
 
+static const struct drm_i915_mocs_entry dg2_mocs_table[] = {
+	/* UC - Coherent; GO:L3 */
+	MOCS_ENTRY(0, 0, L3_1_UC | L3_LKUP(1)),
+	/* UC - Coherent; GO:Memory */
+	MOCS_ENTRY(1, 0, L3_1_UC | L3_GLBGO(1) | L3_LKUP(1)),
+	/* UC - Non-Coherent; GO:Memory */
+	MOCS_ENTRY(2, 0, L3_1_UC | L3_GLBGO(1)),
+
+	/* WB - LC */
+	MOCS_ENTRY(3, 0, L3_3_WB | L3_LKUP(1)),
+};
+
+static const struct drm_i915_mocs_entry dg2_mocs_table_g10_ax[] = {
+	/* Wa_14011441408: Set Go to Memory for MOCS#0 */
+	MOCS_ENTRY(0, 0, L3_1_UC | L3_GLBGO(1) | L3_LKUP(1)),
+	/* UC - Coherent; GO:Memory */
+	MOCS_ENTRY(1, 0, L3_1_UC | L3_GLBGO(1) | L3_LKUP(1)),
+	/* UC - Non-Coherent; GO:Memory */
+	MOCS_ENTRY(2, 0, L3_1_UC | L3_GLBGO(1)),
+
+	/* WB - LC */
+	MOCS_ENTRY(3, 0, L3_3_WB | L3_LKUP(1)),
+};
+
 enum {
 	HAS_GLOBAL_MOCS = BIT(0),
 	HAS_ENGINE_MOCS = BIT(1),
@@ -367,7 +391,16 @@ static unsigned int get_mocs_settings(const struct drm_i915_private *i915,
 {
 	unsigned int flags;
 
-	if (IS_XEHPSDV(i915)) {
+	if (IS_DG2(i915)) {
+		if (IS_DG2_GT_STEP(i915, G10, STEP_A0, STEP_B0)) {
+			table->size = ARRAY_SIZE(dg2_mocs_table_g10_ax);
+			table->table = dg2_mocs_table_g10_ax;
+		} else {
+			table->size = ARRAY_SIZE(dg2_mocs_table);
+			table->table = dg2_mocs_table;
+		}
+		table->n_entries = GEN9_NUM_MOCS_ENTRIES;
+	} else if (IS_XEHPSDV(i915)) {
 		table->size = ARRAY_SIZE(xehpsdv_mocs_table);
 		table->table = xehpsdv_mocs_table;
 		table->n_entries = GEN9_NUM_MOCS_ENTRIES;
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#27 lore raw

PATCH v4 14/18 drm/i915/xehpsdv: factor out function to read RP_STATE_CAP

In reply to Matt Roper

From: Lucas De Marchi <[email protected]>

Instead of maintaining the same if ladder in 3 different places, add a function to read RP_STATE_CAP.

---
 drivers/gpu/drm/i915/gt/debugfs_gt_pm.c |  8 +++-----
 drivers/gpu/drm/i915/gt/intel_rps.c     | 17 ++++++++++++-----
 drivers/gpu/drm/i915/gt/intel_rps.h     |  1 +
 drivers/gpu/drm/i915/i915_debugfs.c     |  8 +++-----
 4 files changed, 19 insertions(+), 15 deletions(-)
drivers/gpu/drm/i915/gt/debugfs_gt_pm.c +3 -5
index 4270b5a34a83..1061a62bdfce 100644
@@ -309,13 +309,11 @@ static int frequency_show(struct seq_file *m, void *unused)
 		int max_freq;
 
 		rp_state_limits = intel_uncore_read(uncore, GEN6_RP_STATE_LIMITS);
-		if (IS_GEN9_LP(i915)) {
-			rp_state_cap = intel_uncore_read(uncore, BXT_RP_STATE_CAP);
+		rp_state_cap = intel_rps_read_state_cap(rps);
+		if (IS_GEN9_LP(i915))
 			gt_perf_status = intel_uncore_read(uncore, BXT_GT_PERF_STATUS);
-		} else {
-			rp_state_cap = intel_uncore_read(uncore, GEN6_RP_STATE_CAP);
+		else
 			gt_perf_status = intel_uncore_read(uncore, GEN6_GT_PERF_STATUS);
-		}
 
 		/* RPSTAT1 is in the GT power well */
 		intel_uncore_forcewake_get(uncore, FORCEWAKE_ALL);
drivers/gpu/drm/i915/gt/intel_rps.c +12 -5
index 0c8e7f2b06f0..8b0f429ba5be 100644
@@ -975,20 +975,16 @@ int intel_rps_set(struct intel_rps *rps, u8 val)
 static void gen6_rps_init(struct intel_rps *rps)
 {
 	struct drm_i915_private *i915 = rps_to_i915(rps);
-	struct intel_uncore *uncore = rps_to_uncore(rps);
+	u32 rp_state_cap = intel_rps_read_state_cap(rps);
 
 	/* All of these values are in units of 50MHz */
 
 	/* static values from HW: RP0 > RP1 > RPn (min_freq) */
 	if (IS_GEN9_LP(i915)) {
-		u32 rp_state_cap = intel_uncore_read(uncore, BXT_RP_STATE_CAP);
-
 		rps->rp0_freq = (rp_state_cap >> 16) & 0xff;
 		rps->rp1_freq = (rp_state_cap >>  8) & 0xff;
 		rps->min_freq = (rp_state_cap >>  0) & 0xff;
 	} else {
-		u32 rp_state_cap = intel_uncore_read(uncore, GEN6_RP_STATE_CAP);
-
 		rps->rp0_freq = (rp_state_cap >>  0) & 0xff;
 		rps->rp1_freq = (rp_state_cap >>  8) & 0xff;
 		rps->min_freq = (rp_state_cap >> 16) & 0xff;
@@ -1940,6 +1936,17 @@ u32 intel_rps_read_actual_frequency(struct intel_rps *rps)
 	return freq;
 }
 
+u32 intel_rps_read_state_cap(struct intel_rps *rps)
+{
+	struct drm_i915_private *i915 = rps_to_i915(rps);
+	struct intel_uncore *uncore = rps_to_uncore(rps);
+
+	if (IS_GEN9_LP(i915))
+		return intel_uncore_read(uncore, BXT_RP_STATE_CAP);
+	else
+		return intel_uncore_read(uncore, GEN6_RP_STATE_CAP);
+}
+
 /* External interface for intel_ips.ko */
 
 static struct drm_i915_private __rcu *ips_mchdev;
drivers/gpu/drm/i915/gt/intel_rps.h +1 -0
index 1d2cfc98b510..6e06dd61f818 100644
@@ -31,6 +31,7 @@ int intel_gpu_freq(struct intel_rps *rps, int val);
 int intel_freq_opcode(struct intel_rps *rps, int val);
 u32 intel_rps_get_cagf(struct intel_rps *rps, u32 rpstat1);
 u32 intel_rps_read_actual_frequency(struct intel_rps *rps);
+u32 intel_rps_read_state_cap(struct intel_rps *rps);
 
 void gen5_rps_irq_handler(struct intel_rps *rps);
 void gen6_rps_irq_handler(struct intel_rps *rps, u32 pm_iir);
drivers/gpu/drm/i915/i915_debugfs.c +3 -5
index 0529576f069c..37056b2c044a 100644
@@ -420,13 +420,11 @@ static int i915_frequency_info(struct seq_file *m, void *unused)
 		int max_freq;
 
 		rp_state_limits = intel_uncore_read(&dev_priv->uncore, GEN6_RP_STATE_LIMITS);
-		if (IS_GEN9_LP(dev_priv)) {
-			rp_state_cap = intel_uncore_read(&dev_priv->uncore, BXT_RP_STATE_CAP);
+		rp_state_cap = intel_rps_read_state_cap(rps);
+		if (IS_GEN9_LP(dev_priv))
 			gt_perf_status = intel_uncore_read(&dev_priv->uncore, BXT_GT_PERF_STATUS);
-		} else {
-			rp_state_cap = intel_uncore_read(&dev_priv->uncore, GEN6_RP_STATE_CAP);
+		else
 			gt_perf_status = intel_uncore_read(&dev_priv->uncore, GEN6_GT_PERF_STATUS);
-		}
 
 		/* RPSTAT1 is in the GT power well */
 		intel_uncore_forcewake_get(&dev_priv->uncore, FORCEWAKE_ALL);
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#28 lore raw

PATCH v4 15/18 drm/i915/xehpsdv: Read correct RP_STATE_CAP register

In reply to Matt Roper

The RP_STATE_CAP register is no longer part of the MCHBAR on XEHPSDV; this register is now a per-tile register at GTTMMADDR offset 0x250014.

---
 drivers/gpu/drm/i915/gt/intel_rps.c | 4 +++-
 drivers/gpu/drm/i915/i915_reg.h     | 1 +
 2 files changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/gt/intel_rps.c +3 -1
index 8b0f429ba5be..10bb172a3826 100644
@@ -1941,7 +1941,9 @@ u32 intel_rps_read_state_cap(struct intel_rps *rps)
 	struct drm_i915_private *i915 = rps_to_i915(rps);
 	struct intel_uncore *uncore = rps_to_uncore(rps);
 
-	if (IS_GEN9_LP(i915))
+	if (IS_XEHPSDV(i915))
+		return intel_uncore_read(uncore, XEHPSDV_RP_STATE_CAP);
+	else if (IS_GEN9_LP(i915))
 		return intel_uncore_read(uncore, BXT_RP_STATE_CAP);
 	else
 		return intel_uncore_read(uncore, GEN6_RP_STATE_CAP);
drivers/gpu/drm/i915/i915_reg.h +1 -0
index d7e4418955f7..ed85f3ec1727 100644
@@ -4184,6 +4184,7 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define GEN6_RP_STATE_CAP	_MMIO(MCHBAR_MIRROR_BASE_SNB + 0x5998)
 #define BXT_RP_STATE_CAP        _MMIO(0x138170)
 #define GEN9_RP_STATE_LIMITS	_MMIO(0x138148)
+#define XEHPSDV_RP_STATE_CAP	_MMIO(0x250014)
 
 /*
  * Logical Context regs
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#29 lore raw

PATCH v4 16/18 drm/i915/dg2: Add new LRI reg offsets

In reply to Matt Roper

From: Akeem G Abodunrin <[email protected]>

New LRI register offsets were introduced for DG2, this patch adds those extra registers, and create new register table for setting offsets to compare with HW generated context image - especially for gt_lrc test. Also updates general purpose register with scratch offset for DG2, in order to use it for live_lrc_fixed selftest.

Cc Chris P Wilson <[email protected]>
Cc Prathap Kumar Valsan <[email protected]>
---
 drivers/gpu/drm/i915/gt/intel_lrc.c | 85 ++++++++++++++++++++++++++++-
 1 file changed, 83 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/gt/intel_lrc.c +83 -2
index c3f5bec8ae15..1b7e75e4c011 100644
@@ -226,6 +226,40 @@ static const u8 gen12_xcs_offsets[] = {
 	END
 };
 
+static const u8 dg2_xcs_offsets[] = {
+	NOP(1),
+	LRI(15, POSTED),
+	REG16(0x244),
+	REG(0x034),
+	REG(0x030),
+	REG(0x038),
+	REG(0x03c),
+	REG(0x168),
+	REG(0x140),
+	REG(0x110),
+	REG(0x1c0),
+	REG(0x1c4),
+	REG(0x1c8),
+	REG(0x180),
+	REG16(0x2b4),
+	REG(0x120),
+	REG(0x124),
+
+	NOP(1),
+	LRI(9, POSTED),
+	REG16(0x3a8),
+	REG16(0x28c),
+	REG16(0x288),
+	REG16(0x284),
+	REG16(0x280),
+	REG16(0x27c),
+	REG16(0x278),
+	REG16(0x274),
+	REG16(0x270),
+
+	END
+};
+
 static const u8 gen8_rcs_offsets[] = {
 	NOP(1),
 	LRI(14, POSTED),
@@ -525,6 +559,49 @@ static const u8 xehp_rcs_offsets[] = {
 	END
 };
 
+static const u8 dg2_rcs_offsets[] = {
+	NOP(1),
+	LRI(15, POSTED),
+	REG16(0x244),
+	REG(0x034),
+	REG(0x030),
+	REG(0x038),
+	REG(0x03c),
+	REG(0x168),
+	REG(0x140),
+	REG(0x110),
+	REG(0x1c0),
+	REG(0x1c4),
+	REG(0x1c8),
+	REG(0x180),
+	REG16(0x2b4),
+	REG(0x120),
+	REG(0x124),
+
+	NOP(1),
+	LRI(9, POSTED),
+	REG16(0x3a8),
+	REG16(0x28c),
+	REG16(0x288),
+	REG16(0x284),
+	REG16(0x280),
+	REG16(0x27c),
+	REG16(0x278),
+	REG16(0x274),
+	REG16(0x270),
+
+	LRI(3, POSTED),
+	REG(0x1b0),
+	REG16(0x5a8),
+	REG16(0x5ac),
+
+	NOP(6),
+	LRI(1, 0),
+	REG(0x0c8),
+
+	END
+};
+
 #undef END
 #undef REG16
 #undef REG
@@ -543,7 +620,9 @@ static const u8 *reg_offsets(const struct intel_engine_cs *engine)
 		   !intel_engine_has_relative_mmio(engine));
 
 	if (engine->class == RENDER_CLASS) {
-		if (GRAPHICS_VER_FULL(engine->i915) >= IP_VER(12, 50))
+		if (GRAPHICS_VER_FULL(engine->i915) >= IP_VER(12, 55))
+			return dg2_rcs_offsets;
+		else if (GRAPHICS_VER_FULL(engine->i915) >= IP_VER(12, 50))
 			return xehp_rcs_offsets;
 		else if (GRAPHICS_VER(engine->i915) >= 12)
 			return gen12_rcs_offsets;
@@ -554,7 +633,9 @@ static const u8 *reg_offsets(const struct intel_engine_cs *engine)
 		else
 			return gen8_rcs_offsets;
 	} else {
-		if (GRAPHICS_VER(engine->i915) >= 12)
+		if (GRAPHICS_VER_FULL(engine->i915) >= IP_VER(12, 55))
+			return dg2_xcs_offsets;
+		else if (GRAPHICS_VER(engine->i915) >= 12)
 			return gen12_xcs_offsets;
 		else if (GRAPHICS_VER(engine->i915) >= 9)
 			return gen9_xcs_offsets;
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#30 lore raw

PATCH v4 17/18 drm/i915/dg2: Maintain backward-compatible nested batch behavior

In reply to Matt Roper

For tgl+, the per-context setting of MI_MODE[12] determines whether the bits of a nested MI_BATCH_BUFFER_START instruction should be interpreted in the traditional manner or whether they should instead use a new tgl+ meaning that breaks backward compatibility, but allows nesting into 3rd-level batchbuffers. For previous platforms, the hardware default for this register bit is to maintain backward-compatible behavior unless a context intentionally opts into the new behavior; however Xe_HPG flips the hardware default behavior.

From a SW perspective, we want to maintain the backward-compatible behavior for userspace, so we'll apply a fake workaround to set it back to the legacy behavior on platforms where the hardware default is to break compatibility. At the moment there is no Linux userspace that utilizes third-level batchbuffers, so this will avoid userspace from needing to make any changes. using the legacy meaning is the correct thing to do. If/when we have userspace consumers that want to utilize third-level batch nesting, we can provide a context parameter to allow them to opt-in.

Bspec: 45974, 45718

Cc John Harrison <[email protected]>
---
 drivers/gpu/drm/i915/gt/intel_workarounds.c | 39 +++++++++++++++++++--
 drivers/gpu/drm/i915/i915_reg.h             |  1 +
 2 files changed, 38 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/gt/intel_workarounds.c +37 -2
index 6895b083523d..3e756b761526 100644
@@ -644,6 +644,37 @@ static void dg1_ctx_workarounds_init(struct intel_engine_cs *engine,
 		     DG1_HZ_READ_SUPPRESSION_OPTIMIZATION_DISABLE);
 }
 
+static void fakewa_disable_nestedbb_mode(struct intel_engine_cs *engine,
+					 struct i915_wa_list *wal)
+{
+	/*
+	 * This is a "fake" workaround defined by software to ensure we
+	 * maintain reliable, backward-compatible behavior for userspace with
+	 * regards to how nested MI_BATCH_BUFFER_START commands are handled.
+	 *
+	 * The per-context setting of MI_MODE[12] determines whether the bits
+	 * of a nested MI_BATCH_BUFFER_START instruction should be interpreted
+	 * in the traditional manner or whether they should instead use a new
+	 * tgl+ meaning that breaks backward compatibility, but allows nesting
+	 * into 3rd-level batchbuffers.  When this new capability was first
+	 * added in TGL, it remained off by default unless a context
+	 * intentionally opted in to the new behavior.  However Xe_HPG now
+	 * flips this on by default and requires that we explicitly opt out if
+	 * we don't want the new behavior.
+	 *
+	 * From a SW perspective, we want to maintain the backward-compatible
+	 * behavior for userspace, so we'll apply a fake workaround to set it
+	 * back to the legacy behavior on platforms where the hardware default
+	 * is to break compatibility.  At the moment there is no Linux
+	 * userspace that utilizes third-level batchbuffers, so this will avoid
+	 * userspace from needing to make any changes.  using the legacy
+	 * meaning is the correct thing to do.  If/when we have userspace
+	 * consumers that want to utilize third-level batch nesting, we can
+	 * provide a context parameter to allow them to opt-in.
+	 */
+	wa_masked_dis(wal, RING_MI_MODE(engine->mmio_base), TGL_NESTED_BB_EN);
+}
+
 static void
 __intel_engine_init_ctx_wa(struct intel_engine_cs *engine,
 			   struct i915_wa_list *wal,
@@ -651,11 +682,15 @@ __intel_engine_init_ctx_wa(struct intel_engine_cs *engine,
 {
 	struct drm_i915_private *i915 = engine->i915;
 
+	wa_init_start(wal, name, engine->name);
+
+	/* Applies to all engines */
+	if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 55))
+		fakewa_disable_nestedbb_mode(engine, wal);
+
 	if (engine->class != RENDER_CLASS)
 		return;
 
-	wa_init_start(wal, name, engine->name);
-
 	if (IS_DG1(i915))
 		dg1_ctx_workarounds_init(engine, wal);
 	else if (GRAPHICS_VER(i915) == 12)
drivers/gpu/drm/i915/i915_reg.h +1 -0
index ed85f3ec1727..c192c1e096d7 100644
@@ -2893,6 +2893,7 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define MI_MODE		_MMIO(0x209c)
 # define VS_TIMER_DISPATCH				(1 << 6)
 # define MI_FLUSH_ENABLE				(1 << 12)
+# define TGL_NESTED_BB_EN				(1 << 12)
 # define ASYNC_FLIP_PERF_DISABLE			(1 << 14)
 # define MODE_IDLE					(1 << 9)
 # define STOP_RING					(1 << 8)
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#31 lore raw

PATCH v4 18/18 drm/i915/dg2: Configure PCON in DP pre-enable path

In reply to Matt Roper

From: Ankit Nautiyal <[email protected]>

Add the functions to configure HDMI2.1 pcon for DG2, before DP link training.

---
 drivers/gpu/drm/i915/display/intel_ddi.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/i915/display/intel_ddi.c +3 -0
index 85856d040637..233f02c617a4 100644
@@ -2576,6 +2576,7 @@ static void dg2_ddi_pre_enable_dp(struct intel_atomic_state *state,
 	if (!is_mst)
 		intel_dp_set_power(intel_dp, DP_SET_POWER_D0);
 
+	intel_dp_configure_protocol_converter(intel_dp, crtc_state);
 	intel_dp_sink_set_decompression_state(intel_dp, crtc_state, true);
 	/*
 	 * DDI FEC: "anticipates enabling FEC encoding sets the FEC_READY bit
@@ -2583,6 +2584,8 @@ static void dg2_ddi_pre_enable_dp(struct intel_atomic_state *state,
 	 * training
 	 */
 	intel_dp_sink_set_fec_ready(intel_dp, crtc_state);
+	intel_dp_check_frl_training(intel_dp);
+	intel_dp_pcon_dsc_configure(intel_dp, crtc_state);
 
 	/*
 	 * 5.h Follow DisplayPort specification training sequence (see notes for
Signature · 6 lines
2.25.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#32 lore raw

[Intel-gfx] ✗ Fi.CI.BUILD: failure for Begin enabling Xe_HP SDV and DG2 platforms (rev8)

In reply to Matt Roper
== Series Details ==

== Series Details ==

Series: Begin enabling Xe_HP SDV and DG2 platforms (rev8)
URL : https://patchwork.freedesktop.org/series/92135/
State : failure

== Summary ==

Applying: drm/i915/xehp: handle new steering options
Applying: drm/i915/xehpsdv: Define steering tables
Applying: drm/i915/dg2: Add forcewake table
Applying: drm/i915/dg2: Update LNCF steering ranges
Applying: drm/i915/dg2: Add SQIDI steering
Applying: drm/i915/xehp: Loop over all gslices for INSTDONE processing
Applying: drm/i915/dg2: Report INSTDONE_GEOM values in error state
Applying: drm/i915/xehp: Changes to ss/eu definitions
Applying: drm/i915/xehpsdv: Add maximum sseu limits
Using index info to reconstruct a base tree...
M drivers/gpu/drm/i915/gt/intel_sseu.c
M drivers/gpu/drm/i915/gt/intel_sseu.h
M drivers/gpu/drm/i915/gt/intel_sseu_debugfs.c
Falling back to patching base and 3-way merge...
Auto-merging drivers/gpu/drm/i915/gt/intel_sseu_debugfs.c Auto-merging drivers/gpu/drm/i915/gt/intel_sseu.h
CONFLICT (content): Merge conflict in drivers/gpu/drm/i915/gt/intel_sseu.h Auto-merging drivers/gpu/drm/i915/gt/intel_sseu.c
error: Failed to merge in the changes.
hint: Use 'git am --show-current-patch=diff' to see the failed patch Patch failed at 0009 drm/i915/xehpsdv: Add maximum sseu limits When you have resolved this problem, run "git am --continue". If you prefer to skip this patch, run "git am --skip" instead. To restore the original branch and stop patching, run "git am --abort".

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx