[PATCH v2 00/11] Add Battlemage support

[PATCH v2 00/11] Add Battlemage support

[Balasubramani Vivekanandan]

Add basic support for Battlemage, which is a discrete graphics card
based on Xe2 architecture.
This series provides basic enabling support. This will be followed with
further patches series to enable display support and loading of required
firwares.

v2:
* Removed the BMG PCI IDs to stop driver from binding to it. Will be
  added later after enabling patches are merged.
* Function argument modified for function get_flat_ccs_offset
* Removed WA_16020183090 which is no more applicable.
  Extended WA_18033852989,WA_18034896535 to xe2hpg
* Switched to open upper bound for "Tuning: L3 Cache" setting
* improved commit description for
  "drm/xe/xe2hpg: Remove extra allocation of CCS pages for dgfx"
* Included the patch from [1] as the patch "Add initial GT workarounds"
  has a dependency on it
  [1]: https://patchwork.freedesktop.org/patch/587168/?series=131912&rev=3


Akshata Jahagirdar (1):
  drm/xe/xe2hpg: Remove extra allocation of CCS pages for dgfx

Balasubramani Vivekanandan (1):
  drm/xe/bmg: Add BMG mocs table

Bommu Krishnaiah (1):
  drm/xe/xe2: Add workaround 18034896535

Gustavo Sousa (1):
  drm/xe/xe2hpm: Add initial set of workarounds

Haridhar Kalvala (1):
  drm/xe/xe2hpg: Add initial GT workarounds

Himal Prasad Ghimiray (1):
  drm/xe/xe2hpg: Determine flat ccs offset for vram

Matt Roper (4):
  drm/xe/xe2: Recognize Xe2_HPG IP
  drm/xe/xe2: Recognize Xe2_HPM IP
  drm/xe/bmg: Add BMG platform definition
  drm/xe/bmg: Program an additional discrete-specific PAT setting

Shekhar Chauhan (1):
  drm/xe/xe2hpg: Introduce performance tuning changes for Xe2_HPG.

 drivers/gpu/drm/xe/regs/xe_gt_regs.h   |  14 ++++
 drivers/gpu/drm/xe/xe_bo.c             |   3 +
 drivers/gpu/drm/xe/xe_mmio.c           |  39 +++++++++-
 drivers/gpu/drm/xe/xe_mocs.c           |   1 +
 drivers/gpu/drm/xe/xe_pat.c            |   7 ++
 drivers/gpu/drm/xe/xe_pci.c            |  12 ++-
 drivers/gpu/drm/xe/xe_platform_types.h |   1 +
 drivers/gpu/drm/xe/xe_tuning.c         |   6 +-
 drivers/gpu/drm/xe/xe_wa.c             | 102 ++++++++++++++++++++++++-
 9 files changed, 177 insertions(+), 8 deletions(-)

-- 
2.25.1

[PATCH v2 01/11] drm/xe/xe2: Add workaround 18034896535

[Balasubramani Vivekanandan]

From: Bommu Krishnaiah <krishnaiah.bommu@intel.com>

Add 18034896535 as driver permanent workaround.

v2: 18034896535 and 16021540221 are two independent workarounds
that just happen to have the same implementation, hence keeping it.

Signed-off-by: Bommu Krishnaiah <krishnaiah.bommu@intel.com>
Reviewed-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Cc: Tejas Upadhyay <tejas.upadhyay@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
---
 drivers/gpu/drm/xe/xe_wa.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_wa.c b/drivers/gpu/drm/xe/xe_wa.c
index c904e55ced9c..43fac92e5d20 100644
--- a/drivers/gpu/drm/xe/xe_wa.c
+++ b/drivers/gpu/drm/xe/xe_wa.c
@@ -428,6 +428,11 @@ static const struct xe_rtp_entry_sr engine_was[] = {
 		       FUNC(xe_rtp_match_first_render_or_compute)),
 	  XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH))
 	},
+	{ XE_RTP_NAME("18034896535"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2004),
+		       FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH))
+	},
 	{ XE_RTP_NAME("14019322943"),
 	  XE_RTP_RULES(GRAPHICS_VERSION(2004), GRAPHICS_STEP(A0, B0),
 		       FUNC(xe_rtp_match_first_render_or_compute)),
-- 
2.25.1

[PATCH v2 01/11] drm/xe/xe2: Recognize Xe2_HPG IP

[Balasubramani Vivekanandan]

From: Matt Roper <matthew.d.roper@intel.com>

Xe2_HPG uses the same general feature flags as Xe2_LPG.  Xe2_HPG is
identified as version 20.01 in the GMD_ID register.

Bspec: 68090
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
---
 drivers/gpu/drm/xe/xe_pci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
index c47ab4b67467..6fd0f4258ba2 100644
--- a/drivers/gpu/drm/xe/xe_pci.c
+++ b/drivers/gpu/drm/xe/xe_pci.c
@@ -174,7 +174,7 @@ static const struct xe_graphics_desc graphics_xelpg = {
 		GENMASK(XE_HW_ENGINE_CCS3, XE_HW_ENGINE_CCS0)
 
 static const struct xe_graphics_desc graphics_xe2 = {
-	.name = "Xe2_LPG",
+	.name = "Xe2_LPG / Xe2_HPG",
 
 	XE2_GFX_FEATURES,
 };
@@ -345,6 +345,7 @@ static const struct gmdid_map graphics_ip_map[] = {
 	{ 1270, &graphics_xelpg },
 	{ 1271, &graphics_xelpg },
 	{ 1274, &graphics_xelpg },	/* Xe_LPG+ */
+	{ 2001, &graphics_xe2 },
 	{ 2004, &graphics_xe2 },
 };
 
-- 
2.25.1

[PATCH v2 02/11] drm/xe/xe2: Recognize Xe2_HPG IP

[Balasubramani Vivekanandan]

From: Matt Roper <matthew.d.roper@intel.com>

Xe2_HPG uses the same general feature flags as Xe2_LPG.  Xe2_HPG is
identified as version 20.01 in the GMD_ID register.

Bspec: 68090
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
---
 drivers/gpu/drm/xe/xe_pci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
index c47ab4b67467..6fd0f4258ba2 100644
--- a/drivers/gpu/drm/xe/xe_pci.c
+++ b/drivers/gpu/drm/xe/xe_pci.c
@@ -174,7 +174,7 @@ static const struct xe_graphics_desc graphics_xelpg = {
 		GENMASK(XE_HW_ENGINE_CCS3, XE_HW_ENGINE_CCS0)
 
 static const struct xe_graphics_desc graphics_xe2 = {
-	.name = "Xe2_LPG",
+	.name = "Xe2_LPG / Xe2_HPG",
 
 	XE2_GFX_FEATURES,
 };
@@ -345,6 +345,7 @@ static const struct gmdid_map graphics_ip_map[] = {
 	{ 1270, &graphics_xelpg },
 	{ 1271, &graphics_xelpg },
 	{ 1274, &graphics_xelpg },	/* Xe_LPG+ */
+	{ 2001, &graphics_xe2 },
 	{ 2004, &graphics_xe2 },
 };
 
-- 
2.25.1

[PATCH v2 02/11] drm/xe/xe2: Recognize Xe2_HPM IP

[Balasubramani Vivekanandan]

From: Matt Roper <matthew.d.roper@intel.com>

Xe2_HPM uses the same general feature flags as Xe2_LPM.  Xe2_HPM is
identified as version 13.01 in the GMD_ID register.

Bspec: 68090, 67163
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
---
 drivers/gpu/drm/xe/xe_pci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
index 6fd0f4258ba2..2d3b18d6404e 100644
--- a/drivers/gpu/drm/xe/xe_pci.c
+++ b/drivers/gpu/drm/xe/xe_pci.c
@@ -207,7 +207,7 @@ static const struct xe_media_desc media_xelpmp = {
 };
 
 static const struct xe_media_desc media_xe2 = {
-	.name = "Xe2_LPM",
+	.name = "Xe2_LPM / Xe2_HPM",
 	.hw_engine_mask =
 		BIT(XE_HW_ENGINE_VCS0) | BIT(XE_HW_ENGINE_VECS0), /* TODO: GSC0 */
 };
@@ -352,6 +352,7 @@ static const struct gmdid_map graphics_ip_map[] = {
 /* Map of GMD_ID values to media IP */
 static const struct gmdid_map media_ip_map[] = {
 	{ 1300, &media_xelpmp },
+	{ 1301, &media_xe2 },
 	{ 2000, &media_xe2 },
 };
 
-- 
2.25.1

[PATCH v2 03/11] drm/xe/bmg: Add BMG platform definition

[Balasubramani Vivekanandan]

From: Matt Roper <matthew.d.roper@intel.com>

BMG is a discrete GPU based on the Xe2 architecture.
No device ids are bound to the BMG platform descriptor yet.
BMG device ids will be added once we have all the basic required
platform enabling patches landed.

v2: Removed device ids, deferring it to a later patch

Bspec: 68090
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
---
 drivers/gpu/drm/xe/xe_pci.c            | 6 ++++++
 drivers/gpu/drm/xe/xe_platform_types.h | 1 +
 2 files changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
index 2d3b18d6404e..3b30353dbc09 100644
--- a/drivers/gpu/drm/xe/xe_pci.c
+++ b/drivers/gpu/drm/xe/xe_pci.c
@@ -337,6 +337,12 @@ static const struct xe_device_desc lnl_desc = {
 	.require_force_probe = true,
 };
 
+static const struct xe_device_desc bmg_desc __maybe_unused = {
+	DGFX_FEATURES,
+	PLATFORM(XE_BATTLEMAGE),
+	.require_force_probe = true,
+};
+
 #undef PLATFORM
 __diag_pop();
 
diff --git a/drivers/gpu/drm/xe/xe_platform_types.h b/drivers/gpu/drm/xe/xe_platform_types.h
index 553f53dbd093..79b7042c4534 100644
--- a/drivers/gpu/drm/xe/xe_platform_types.h
+++ b/drivers/gpu/drm/xe/xe_platform_types.h
@@ -22,6 +22,7 @@ enum xe_platform {
 	XE_PVC,
 	XE_METEORLAKE,
 	XE_LUNARLAKE,
+	XE_BATTLEMAGE,
 };
 
 enum xe_subplatform {
-- 
2.25.1

[PATCH v2 03/11] drm/xe/xe2: Recognize Xe2_HPM IP

[Balasubramani Vivekanandan]

From: Matt Roper <matthew.d.roper@intel.com>

Xe2_HPM uses the same general feature flags as Xe2_LPM.  Xe2_HPM is
identified as version 13.01 in the GMD_ID register.

Bspec: 68090, 67163
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
---
 drivers/gpu/drm/xe/xe_pci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
index 6fd0f4258ba2..2d3b18d6404e 100644
--- a/drivers/gpu/drm/xe/xe_pci.c
+++ b/drivers/gpu/drm/xe/xe_pci.c
@@ -207,7 +207,7 @@ static const struct xe_media_desc media_xelpmp = {
 };
 
 static const struct xe_media_desc media_xe2 = {
-	.name = "Xe2_LPM",
+	.name = "Xe2_LPM / Xe2_HPM",
 	.hw_engine_mask =
 		BIT(XE_HW_ENGINE_VCS0) | BIT(XE_HW_ENGINE_VECS0), /* TODO: GSC0 */
 };
@@ -352,6 +352,7 @@ static const struct gmdid_map graphics_ip_map[] = {
 /* Map of GMD_ID values to media IP */
 static const struct gmdid_map media_ip_map[] = {
 	{ 1300, &media_xelpmp },
+	{ 1301, &media_xe2 },
 	{ 2000, &media_xe2 },
 };
 
-- 
2.25.1

[PATCH v2 04/11] drm/xe/bmg: Add BMG mocs table

[Balasubramani Vivekanandan]

BMG uses the same MOCS table as LNL.

Bpsec: 71582
CC: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
---
 drivers/gpu/drm/xe/xe_mocs.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/gpu/drm/xe/xe_mocs.c b/drivers/gpu/drm/xe/xe_mocs.c
index bff659d20062..4f836aeeabff 100644
--- a/drivers/gpu/drm/xe/xe_mocs.c
+++ b/drivers/gpu/drm/xe/xe_mocs.c
@@ -375,6 +375,7 @@ static unsigned int get_mocs_settings(struct xe_device *xe,
 
 	switch (xe->info.platform) {
 	case XE_LUNARLAKE:
+	case XE_BATTLEMAGE:
 		info->size = ARRAY_SIZE(xe2_mocs_table);
 		info->table = xe2_mocs_table;
 		info->n_entries = XE2_NUM_MOCS_ENTRIES;
-- 
2.25.1

[PATCH v2 04/11] drm/xe/bmg: Add BMG platform definition

[Balasubramani Vivekanandan]

From: Matt Roper <matthew.d.roper@intel.com>

BMG is a discrete GPU based on the Xe2 architecture.
No device ids are bound to the BMG platform descriptor yet.
BMG device ids will be added once we have all the basic required
platform enabling patches landed.

v2: Removed device ids, deferring it to a later patch

Bspec: 68090
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
---
 drivers/gpu/drm/xe/xe_pci.c            | 6 ++++++
 drivers/gpu/drm/xe/xe_platform_types.h | 1 +
 2 files changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
index 2d3b18d6404e..3b30353dbc09 100644
--- a/drivers/gpu/drm/xe/xe_pci.c
+++ b/drivers/gpu/drm/xe/xe_pci.c
@@ -337,6 +337,12 @@ static const struct xe_device_desc lnl_desc = {
 	.require_force_probe = true,
 };
 
+static const struct xe_device_desc bmg_desc __maybe_unused = {
+	DGFX_FEATURES,
+	PLATFORM(XE_BATTLEMAGE),
+	.require_force_probe = true,
+};
+
 #undef PLATFORM
 __diag_pop();
 
diff --git a/drivers/gpu/drm/xe/xe_platform_types.h b/drivers/gpu/drm/xe/xe_platform_types.h
index 553f53dbd093..79b7042c4534 100644
--- a/drivers/gpu/drm/xe/xe_platform_types.h
+++ b/drivers/gpu/drm/xe/xe_platform_types.h
@@ -22,6 +22,7 @@ enum xe_platform {
 	XE_PVC,
 	XE_METEORLAKE,
 	XE_LUNARLAKE,
+	XE_BATTLEMAGE,
 };
 
 enum xe_subplatform {
-- 
2.25.1

[PATCH v2 05/11] drm/xe/bmg: Add BMG mocs table

[Balasubramani Vivekanandan]

BMG uses the same MOCS table as LNL.

Bpsec: 71582
CC: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
---
 drivers/gpu/drm/xe/xe_mocs.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/gpu/drm/xe/xe_mocs.c b/drivers/gpu/drm/xe/xe_mocs.c
index bff659d20062..4f836aeeabff 100644
--- a/drivers/gpu/drm/xe/xe_mocs.c
+++ b/drivers/gpu/drm/xe/xe_mocs.c
@@ -375,6 +375,7 @@ static unsigned int get_mocs_settings(struct xe_device *xe,
 
 	switch (xe->info.platform) {
 	case XE_LUNARLAKE:
+	case XE_BATTLEMAGE:
 		info->size = ARRAY_SIZE(xe2_mocs_table);
 		info->table = xe2_mocs_table;
 		info->n_entries = XE2_NUM_MOCS_ENTRIES;
-- 
2.25.1

[PATCH v2 05/11] drm/xe/bmg: Program an additional discrete-specific PAT setting

[Balasubramani Vivekanandan]

From: Matt Roper <matthew.d.roper@intel.com>

Discrete Xe2 platforms require programming of one additional row of PAT
settings which controls the access characteristics for PPGTT and LMTT
page tables.  Integrated GPUs do not need this programming and will
leave the register at its hardware default value.

Bspec:  71582
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
---
 drivers/gpu/drm/xe/xe_pat.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_pat.c b/drivers/gpu/drm/xe/xe_pat.c
index 66d8e3dd8237..13812042177d 100644
--- a/drivers/gpu/drm/xe/xe_pat.c
+++ b/drivers/gpu/drm/xe/xe_pat.c
@@ -142,6 +142,7 @@ static const struct xe_pat_table_entry xe2_pat_table[] = {
 
 /* Special PAT values programmed outside the main table */
 static const struct xe_pat_table_entry xe2_pat_ats = XE2_PAT( 0, 0, 0, 0, 3, 3 );
+static const struct xe_pat_table_entry xe2_pat_pta = XE2_PAT( 0, 0, 0, 0, 3, 0 );
 
 u16 xe_pat_index_get_coh_mode(struct xe_device *xe, u16 pat_index)
 {
@@ -302,6 +303,9 @@ static void xe2lpg_program_pat(struct xe_gt *gt, const struct xe_pat_table_entry
 {
 	program_pat_mcr(gt, table, n_entries);
 	xe_gt_mcr_multicast_write(gt, XE_REG_MCR(_PAT_ATS), xe2_pat_ats.value);
+
+	if (IS_DGFX(gt_to_xe(gt)))
+		xe_gt_mcr_multicast_write(gt, XE_REG_MCR(_PAT_PTA), xe2_pat_pta.value);
 }
 
 static void xe2lpm_program_pat(struct xe_gt *gt, const struct xe_pat_table_entry table[],
@@ -309,6 +313,9 @@ static void xe2lpm_program_pat(struct xe_gt *gt, const struct xe_pat_table_entry
 {
 	program_pat(gt, table, n_entries);
 	xe_mmio_write32(gt, XE_REG(_PAT_ATS), xe2_pat_ats.value);
+
+	if (IS_DGFX(gt_to_xe(gt)))
+		xe_mmio_write32(gt, XE_REG(_PAT_PTA), xe2_pat_pta.value);
 }
 
 static void xe2_dump(struct xe_gt *gt, struct drm_printer *p)
-- 
2.25.1

[PATCH v2 06/11] drm/xe/bmg: Program an additional discrete-specific PAT setting

[Balasubramani Vivekanandan]

From: Matt Roper <matthew.d.roper@intel.com>

Discrete Xe2 platforms require programming of one additional row of PAT
settings which controls the access characteristics for PPGTT and LMTT
page tables.  Integrated GPUs do not need this programming and will
leave the register at its hardware default value.

Bspec:  71582
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
---
 drivers/gpu/drm/xe/xe_pat.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_pat.c b/drivers/gpu/drm/xe/xe_pat.c
index 66d8e3dd8237..13812042177d 100644
--- a/drivers/gpu/drm/xe/xe_pat.c
+++ b/drivers/gpu/drm/xe/xe_pat.c
@@ -142,6 +142,7 @@ static const struct xe_pat_table_entry xe2_pat_table[] = {
 
 /* Special PAT values programmed outside the main table */
 static const struct xe_pat_table_entry xe2_pat_ats = XE2_PAT( 0, 0, 0, 0, 3, 3 );
+static const struct xe_pat_table_entry xe2_pat_pta = XE2_PAT( 0, 0, 0, 0, 3, 0 );
 
 u16 xe_pat_index_get_coh_mode(struct xe_device *xe, u16 pat_index)
 {
@@ -302,6 +303,9 @@ static void xe2lpg_program_pat(struct xe_gt *gt, const struct xe_pat_table_entry
 {
 	program_pat_mcr(gt, table, n_entries);
 	xe_gt_mcr_multicast_write(gt, XE_REG_MCR(_PAT_ATS), xe2_pat_ats.value);
+
+	if (IS_DGFX(gt_to_xe(gt)))
+		xe_gt_mcr_multicast_write(gt, XE_REG_MCR(_PAT_PTA), xe2_pat_pta.value);
 }
 
 static void xe2lpm_program_pat(struct xe_gt *gt, const struct xe_pat_table_entry table[],
@@ -309,6 +313,9 @@ static void xe2lpm_program_pat(struct xe_gt *gt, const struct xe_pat_table_entry
 {
 	program_pat(gt, table, n_entries);
 	xe_mmio_write32(gt, XE_REG(_PAT_ATS), xe2_pat_ats.value);
+
+	if (IS_DGFX(gt_to_xe(gt)))
+		xe_mmio_write32(gt, XE_REG(_PAT_PTA), xe2_pat_pta.value);
 }
 
 static void xe2_dump(struct xe_gt *gt, struct drm_printer *p)
-- 
2.25.1

[PATCH v2 06/11] drm/xe/xe2hpg: Determine flat ccs offset for vram

[Balasubramani Vivekanandan]

From: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>

on Xe2 dgfx platform determine the offset using Flat CCS size
bitfield of XE2_FLAT_CCS_BASE_RANGE_[UPPER/LOWER] mcr registers.

Bspec: 68023

v2: function argument tile_size changed from pass by reference to pass
by value

Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Signed-off-by: Akshata Jahagirdar <akshata.jahagirdar@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
---
 drivers/gpu/drm/xe/regs/xe_gt_regs.h |  5 ++++
 drivers/gpu/drm/xe/xe_mmio.c         | 39 ++++++++++++++++++++++++++--
 2 files changed, 42 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/xe/regs/xe_gt_regs.h b/drivers/gpu/drm/xe/regs/xe_gt_regs.h
index 6617c86a096b..d404f211bc36 100644
--- a/drivers/gpu/drm/xe/regs/xe_gt_regs.h
+++ b/drivers/gpu/drm/xe/regs/xe_gt_regs.h
@@ -69,6 +69,7 @@
 
 #define XEHP_TILE_ADDR_RANGE(_idx)		XE_REG_MCR(0x4900 + (_idx) * 4)
 #define XEHP_FLAT_CCS_BASE_ADDR			XE_REG_MCR(0x4910)
+#define XEHP_FLAT_CCS_PTR			REG_GENMASK(31, 8)
 
 #define WM_CHICKEN3				XE_REG_MCR(0x5588, XE_REG_OPTION_MASKED)
 #define   HIZ_PLANE_COMPRESSION_DIS		REG_BIT(10)
@@ -142,6 +143,10 @@
 
 #define XE2_FLAT_CCS_BASE_RANGE_LOWER		XE_REG_MCR(0x8800)
 #define   XE2_FLAT_CCS_ENABLE			REG_BIT(0)
+#define XE2_FLAT_CCS_BASE_LOWER_ADDR_MASK	REG_GENMASK(31, 6)
+
+#define XE2_FLAT_CCS_BASE_RANGE_UPPER		XE_REG_MCR(0x8804)
+#define XE2_FLAT_CCS_BASE_UPPER_ADDR_MASK	REG_GENMASK(7, 0)
 
 #define GSCPSMI_BASE				XE_REG(0x880c)
 
diff --git a/drivers/gpu/drm/xe/xe_mmio.c b/drivers/gpu/drm/xe/xe_mmio.c
index 5d13fc7cb9d2..d66da1a9f165 100644
--- a/drivers/gpu/drm/xe/xe_mmio.c
+++ b/drivers/gpu/drm/xe/xe_mmio.c
@@ -163,6 +163,42 @@ static int xe_determine_lmem_bar_size(struct xe_device *xe)
 	return 0;
 }
 
+static inline u64 get_flat_ccs_offset(struct xe_gt *gt, u64 tile_size)
+{
+	struct xe_device *xe = gt_to_xe(gt);
+	u64 offset;
+	u32 reg;
+
+	if (GRAPHICS_VER(xe) >= 20) {
+		u64 ccs_size = tile_size / 512;
+		u64 offset_hi, offset_lo;
+		u32 nodes, num_enabled;
+
+		reg = xe_mmio_read32(gt, MIRROR_FUSE3);
+		nodes = REG_FIELD_GET(XE2_NODE_ENABLE_MASK, reg);
+		num_enabled = hweight32(nodes); /* Number of enabled l3 nodes */
+
+		reg = xe_gt_mcr_unicast_read_any(gt, XE2_FLAT_CCS_BASE_RANGE_LOWER);
+		offset_lo = REG_FIELD_GET(XE2_FLAT_CCS_BASE_LOWER_ADDR_MASK, reg);
+
+		reg = xe_gt_mcr_unicast_read_any(gt, XE2_FLAT_CCS_BASE_RANGE_UPPER);
+		offset_hi = REG_FIELD_GET(XE2_FLAT_CCS_BASE_UPPER_ADDR_MASK, reg);
+
+		offset = offset_hi << 32; /* HW view bits 39:32 */
+		offset |= offset_lo << 6; /* HW view bits 31:6 */
+		offset *= num_enabled; /* convert to SW view */
+
+		/* We don't expect any holes */
+		xe_assert_msg(xe, offset == (xe_mmio_read64_2x32(gt, GSMBASE) - ccs_size),
+			      "Hole between CCS and GSM.\n");
+	} else {
+		reg = xe_gt_mcr_unicast_read_any(gt, XEHP_FLAT_CCS_BASE_ADDR);
+		offset = (u64)REG_FIELD_GET(XEHP_FLAT_CCS_PTR, reg) * SZ_64K;
+	}
+
+	return offset;
+}
+
 /**
  * xe_mmio_tile_vram_size() - Collect vram size and offset information
  * @tile: tile to get info for
@@ -207,8 +243,7 @@ static int xe_mmio_tile_vram_size(struct xe_tile *tile, u64 *vram_size,
 
 	/* minus device usage */
 	if (xe->info.has_flat_ccs) {
-		reg = xe_gt_mcr_unicast_read_any(gt, XEHP_FLAT_CCS_BASE_ADDR);
-		offset = (u64)REG_FIELD_GET(GENMASK(31, 8), reg) * SZ_64K;
+		offset = get_flat_ccs_offset(gt, *tile_size);
 	} else {
 		offset = xe_mmio_read64_2x32(gt, GSMBASE);
 	}
-- 
2.25.1

[PATCH v2 07/11] drm/xe/xe2hpg: Determine flat ccs offset for vram

[Balasubramani Vivekanandan]

From: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>

on Xe2 dgfx platform determine the offset using Flat CCS size
bitfield of XE2_FLAT_CCS_BASE_RANGE_[UPPER/LOWER] mcr registers.

Bspec: 68023

v2: function argument tile_size changed from pass by reference to pass
by value

Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Signed-off-by: Akshata Jahagirdar <akshata.jahagirdar@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
---
 drivers/gpu/drm/xe/regs/xe_gt_regs.h |  5 ++++
 drivers/gpu/drm/xe/xe_mmio.c         | 39 ++++++++++++++++++++++++++--
 2 files changed, 42 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/xe/regs/xe_gt_regs.h b/drivers/gpu/drm/xe/regs/xe_gt_regs.h
index 6617c86a096b..d404f211bc36 100644
--- a/drivers/gpu/drm/xe/regs/xe_gt_regs.h
+++ b/drivers/gpu/drm/xe/regs/xe_gt_regs.h
@@ -69,6 +69,7 @@
 
 #define XEHP_TILE_ADDR_RANGE(_idx)		XE_REG_MCR(0x4900 + (_idx) * 4)
 #define XEHP_FLAT_CCS_BASE_ADDR			XE_REG_MCR(0x4910)
+#define XEHP_FLAT_CCS_PTR			REG_GENMASK(31, 8)
 
 #define WM_CHICKEN3				XE_REG_MCR(0x5588, XE_REG_OPTION_MASKED)
 #define   HIZ_PLANE_COMPRESSION_DIS		REG_BIT(10)
@@ -142,6 +143,10 @@
 
 #define XE2_FLAT_CCS_BASE_RANGE_LOWER		XE_REG_MCR(0x8800)
 #define   XE2_FLAT_CCS_ENABLE			REG_BIT(0)
+#define XE2_FLAT_CCS_BASE_LOWER_ADDR_MASK	REG_GENMASK(31, 6)
+
+#define XE2_FLAT_CCS_BASE_RANGE_UPPER		XE_REG_MCR(0x8804)
+#define XE2_FLAT_CCS_BASE_UPPER_ADDR_MASK	REG_GENMASK(7, 0)
 
 #define GSCPSMI_BASE				XE_REG(0x880c)
 
diff --git a/drivers/gpu/drm/xe/xe_mmio.c b/drivers/gpu/drm/xe/xe_mmio.c
index 5d13fc7cb9d2..d66da1a9f165 100644
--- a/drivers/gpu/drm/xe/xe_mmio.c
+++ b/drivers/gpu/drm/xe/xe_mmio.c
@@ -163,6 +163,42 @@ static int xe_determine_lmem_bar_size(struct xe_device *xe)
 	return 0;
 }
 
+static inline u64 get_flat_ccs_offset(struct xe_gt *gt, u64 tile_size)
+{
+	struct xe_device *xe = gt_to_xe(gt);
+	u64 offset;
+	u32 reg;
+
+	if (GRAPHICS_VER(xe) >= 20) {
+		u64 ccs_size = tile_size / 512;
+		u64 offset_hi, offset_lo;
+		u32 nodes, num_enabled;
+
+		reg = xe_mmio_read32(gt, MIRROR_FUSE3);
+		nodes = REG_FIELD_GET(XE2_NODE_ENABLE_MASK, reg);
+		num_enabled = hweight32(nodes); /* Number of enabled l3 nodes */
+
+		reg = xe_gt_mcr_unicast_read_any(gt, XE2_FLAT_CCS_BASE_RANGE_LOWER);
+		offset_lo = REG_FIELD_GET(XE2_FLAT_CCS_BASE_LOWER_ADDR_MASK, reg);
+
+		reg = xe_gt_mcr_unicast_read_any(gt, XE2_FLAT_CCS_BASE_RANGE_UPPER);
+		offset_hi = REG_FIELD_GET(XE2_FLAT_CCS_BASE_UPPER_ADDR_MASK, reg);
+
+		offset = offset_hi << 32; /* HW view bits 39:32 */
+		offset |= offset_lo << 6; /* HW view bits 31:6 */
+		offset *= num_enabled; /* convert to SW view */
+
+		/* We don't expect any holes */
+		xe_assert_msg(xe, offset == (xe_mmio_read64_2x32(gt, GSMBASE) - ccs_size),
+			      "Hole between CCS and GSM.\n");
+	} else {
+		reg = xe_gt_mcr_unicast_read_any(gt, XEHP_FLAT_CCS_BASE_ADDR);
+		offset = (u64)REG_FIELD_GET(XEHP_FLAT_CCS_PTR, reg) * SZ_64K;
+	}
+
+	return offset;
+}
+
 /**
  * xe_mmio_tile_vram_size() - Collect vram size and offset information
  * @tile: tile to get info for
@@ -207,8 +243,7 @@ static int xe_mmio_tile_vram_size(struct xe_tile *tile, u64 *vram_size,
 
 	/* minus device usage */
 	if (xe->info.has_flat_ccs) {
-		reg = xe_gt_mcr_unicast_read_any(gt, XEHP_FLAT_CCS_BASE_ADDR);
-		offset = (u64)REG_FIELD_GET(GENMASK(31, 8), reg) * SZ_64K;
+		offset = get_flat_ccs_offset(gt, *tile_size);
 	} else {
 		offset = xe_mmio_read64_2x32(gt, GSMBASE);
 	}
-- 
2.25.1

[PATCH v2 07/11] drm/xe/xe2hpg: Remove extra allocation of CCS pages for dgfx

[Balasubramani Vivekanandan]

From: Akshata Jahagirdar <akshata.jahagirdar@intel.com>

On Xe2 dGPU, compression is only supported with VRAM. When copying from
VRAM -> system memory the KMD uses mapping with uncompressed PAT
so the copy in system memory is guaranteed to be uncompressed.
When restoring such buffers from system memory -> VRAM the KMD can't
easily know which pages were originally compressed, so we always use
uncompressed -> uncompressed here.
so this means that there's no need for extra CCS storage on such
platforms.

v2: More description added to commit message

Signed-off-by: Akshata Jahagirdar <akshata.jahagirdar@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
---
 drivers/gpu/drm/xe/xe_bo.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c
index 6166bc715656..fdeb3691d3f6 100644
--- a/drivers/gpu/drm/xe/xe_bo.c
+++ b/drivers/gpu/drm/xe/xe_bo.c
@@ -2201,6 +2201,9 @@ bool xe_bo_needs_ccs_pages(struct xe_bo *bo)
 {
 	struct xe_device *xe = xe_bo_device(bo);
 
+	if (GRAPHICS_VER(xe) >= 20 && IS_DGFX(xe))
+		return false;
+
 	if (!xe_device_has_flat_ccs(xe) || bo->ttm.type != ttm_bo_type_device)
 		return false;
 
-- 
2.25.1

Re: [PATCH v2 07/11] drm/xe/xe2hpg: Remove extra allocation of CCS pages for dgfx

[Ghimiray, Himal Prasad]

On 08-04-2024 21:59, Balasubramani Vivekanandan wrote:
> From: Akshata Jahagirdar <akshata.jahagirdar@intel.com>
>
> On Xe2 dGPU, compression is only supported with VRAM. When copying from
> VRAM -> system memory the KMD uses mapping with uncompressed PAT
> so the copy in system memory is guaranteed to be uncompressed.
> When restoring such buffers from system memory -> VRAM the KMD can't
> easily know which pages were originally compressed, so we always use
> uncompressed -> uncompressed here.
> so this means that there's no need for extra CCS storage on such
> platforms.
>
> v2: More description added to commit message
>
> Signed-off-by: Akshata Jahagirdar <akshata.jahagirdar@intel.com>
> Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
> ---
>   drivers/gpu/drm/xe/xe_bo.c | 3 +++
>   1 file changed, 3 insertions(+)
>
> diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c
> index 6166bc715656..fdeb3691d3f6 100644
> --- a/drivers/gpu/drm/xe/xe_bo.c
> +++ b/drivers/gpu/drm/xe/xe_bo.c
> @@ -2201,6 +2201,9 @@ bool xe_bo_needs_ccs_pages(struct xe_bo *bo)
>   {
>   	struct xe_device *xe = xe_bo_device(bo);
>   
> +	if (GRAPHICS_VER(xe) >= 20 && IS_DGFX(xe))
> +		return false;
> +

LGTM.

Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>

>   	if (!xe_device_has_flat_ccs(xe) || bo->ttm.type != ttm_bo_type_device)
>   		return false;
>   

[PATCH v2 08/11] drm/xe/xe2: Add workaround 18034896535

[Balasubramani Vivekanandan]

From: Bommu Krishnaiah <krishnaiah.bommu@intel.com>

Add 18034896535 as driver permanent workaround.

v2: 18034896535 and 16021540221 are two independent workarounds
that just happen to have the same implementation, hence keeping it.

Signed-off-by: Bommu Krishnaiah <krishnaiah.bommu@intel.com>
Reviewed-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Cc: Tejas Upadhyay <tejas.upadhyay@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
---
 drivers/gpu/drm/xe/xe_wa.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_wa.c b/drivers/gpu/drm/xe/xe_wa.c
index c904e55ced9c..43fac92e5d20 100644
--- a/drivers/gpu/drm/xe/xe_wa.c
+++ b/drivers/gpu/drm/xe/xe_wa.c
@@ -428,6 +428,11 @@ static const struct xe_rtp_entry_sr engine_was[] = {
 		       FUNC(xe_rtp_match_first_render_or_compute)),
 	  XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH))
 	},
+	{ XE_RTP_NAME("18034896535"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2004),
+		       FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH))
+	},
 	{ XE_RTP_NAME("14019322943"),
 	  XE_RTP_RULES(GRAPHICS_VERSION(2004), GRAPHICS_STEP(A0, B0),
 		       FUNC(xe_rtp_match_first_render_or_compute)),
-- 
2.25.1

[PATCH v2 08/11] drm/xe/xe2hpg: Remove extra allocation of CCS pages for dgfx

[Balasubramani Vivekanandan]

From: Akshata Jahagirdar <akshata.jahagirdar@intel.com>

On Xe2 dGPU, compression is only supported with VRAM. When copying from
VRAM -> system memory the KMD uses mapping with uncompressed PAT
so the copy in system memory is guaranteed to be uncompressed.
When restoring such buffers from system memory -> VRAM the KMD can't
easily know which pages were originally compressed, so we always use
uncompressed -> uncompressed here.
so this means that there's no need for extra CCS storage on such
platforms.

v2: More description added to commit message

Signed-off-by: Akshata Jahagirdar <akshata.jahagirdar@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
---
 drivers/gpu/drm/xe/xe_bo.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c
index 6166bc715656..fdeb3691d3f6 100644
--- a/drivers/gpu/drm/xe/xe_bo.c
+++ b/drivers/gpu/drm/xe/xe_bo.c
@@ -2201,6 +2201,9 @@ bool xe_bo_needs_ccs_pages(struct xe_bo *bo)
 {
 	struct xe_device *xe = xe_bo_device(bo);
 
+	if (GRAPHICS_VER(xe) >= 20 && IS_DGFX(xe))
+		return false;
+
 	if (!xe_device_has_flat_ccs(xe) || bo->ttm.type != ttm_bo_type_device)
 		return false;
 
-- 
2.25.1

[PATCH v2 09/11] drm/xe/xe2hpg: Add initial GT workarounds

[Balasubramani Vivekanandan]

From: Haridhar Kalvala <haridhar.kalvala@intel.com>

Add the initial set of Xe2_HPG gt/engine/lrc workarounds.

v2: Removed WA_16020183090 which is no more applicable
    Extended WA_18033852989,18034896535 also to xe2hpg

Signed-off-by: Haridhar Kalvala <haridhar.kalvala@intel.com>
Signed-off-by: Clint Taylor <clinton.a.taylor@intel.com>
Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Dnyaneshar Bhadane <dnyaneshwar.bhadane@intel.com>
Signed-off-by: Shekhar Chauhan <shekhar.chauhan@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
---
 drivers/gpu/drm/xe/regs/xe_gt_regs.h |  5 +++
 drivers/gpu/drm/xe/xe_wa.c           | 67 +++++++++++++++++++++++++++-
 2 files changed, 70 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/xe/regs/xe_gt_regs.h b/drivers/gpu/drm/xe/regs/xe_gt_regs.h
index d404f211bc36..0ce79ba19bda 100644
--- a/drivers/gpu/drm/xe/regs/xe_gt_regs.h
+++ b/drivers/gpu/drm/xe/regs/xe_gt_regs.h
@@ -74,6 +74,9 @@
 #define WM_CHICKEN3				XE_REG_MCR(0x5588, XE_REG_OPTION_MASKED)
 #define   HIZ_PLANE_COMPRESSION_DIS		REG_BIT(10)
 
+#define CHICKEN_RASTER_1			XE_REG_MCR(0x6204, XE_REG_OPTION_MASKED)
+#define   DIS_SF_ROUND_NEAREST_EVEN		REG_BIT(8)
+
 #define CHICKEN_RASTER_2			XE_REG_MCR(0x6208, XE_REG_OPTION_MASKED)
 #define   TBIMR_FAST_CLIP			REG_BIT(5)
 
@@ -355,6 +358,7 @@
 #define   THREAD_EX_ARB_MODE_RR_AFTER_DEP	REG_FIELD_PREP(THREAD_EX_ARB_MODE, 0x2)
 
 #define ROW_CHICKEN3				XE_REG_MCR(0xe49c, XE_REG_OPTION_MASKED)
+#define   XE2_EUPEND_CHK_FLUSH_DIS		REG_BIT(14)
 #define   DIS_FIX_EOT1_FLUSH			REG_BIT(9)
 
 #define TDL_TSL_CHICKEN				XE_REG_MCR(0xe4c4, XE_REG_OPTION_MASKED)
@@ -385,6 +389,7 @@
 
 #define LSC_CHICKEN_BIT_0			XE_REG_MCR(0xe7c8)
 #define   DISABLE_D8_D16_COASLESCE		REG_BIT(30)
+#define   WR_REQ_CHAINING_DIS			REG_BIT(26)
 #define   TGM_WRITE_EOM_FORCE			REG_BIT(17)
 #define   FORCE_1_SUB_MESSAGE_PER_FRAGMENT	REG_BIT(15)
 #define   SEQUENTIAL_ACCESS_UPGRADE_DISABLE	REG_BIT(13)
diff --git a/drivers/gpu/drm/xe/xe_wa.c b/drivers/gpu/drm/xe/xe_wa.c
index 43fac92e5d20..014d27c126ae 100644
--- a/drivers/gpu/drm/xe/xe_wa.c
+++ b/drivers/gpu/drm/xe/xe_wa.c
@@ -429,7 +429,7 @@ static const struct xe_rtp_entry_sr engine_was[] = {
 	  XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH))
 	},
 	{ XE_RTP_NAME("18034896535"),
-	  XE_RTP_RULES(GRAPHICS_VERSION(2004),
+	  XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2001, 2004),
 		       FUNC(xe_rtp_match_first_render_or_compute)),
 	  XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH))
 	},
@@ -464,6 +464,55 @@ static const struct xe_rtp_entry_sr engine_was[] = {
 	  XE_RTP_RULES(GRAPHICS_VERSION(2004), FUNC(xe_rtp_match_first_render_or_compute)),
 	  XE_RTP_ACTIONS(SET(TDL_TSL_CHICKEN, SLM_WMTP_RESTORE))
 	},
+
+	/* Xe2_HPG */
+
+	{ XE_RTP_NAME("16018712365"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(LSC_CHICKEN_BIT_0_UDW, XE2_ALLOC_DPA_STARVE_FIX_DIS))
+	},
+	{ XE_RTP_NAME("16018737384"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(ROW_CHICKEN, EARLY_EOT_DIS))
+	},
+	{ XE_RTP_NAME("14019988906"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(XEHP_PSS_CHICKEN, FLSH_IGNORES_PSD))
+	},
+	{ XE_RTP_NAME("14019877138"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(XEHP_PSS_CHICKEN, FD_END_COLLECT))
+	},
+	{ XE_RTP_NAME("14020338487"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(ROW_CHICKEN3, XE2_EUPEND_CHK_FLUSH_DIS))
+	},
+	{ XE_RTP_NAME("18032247524"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(LSC_CHICKEN_BIT_0, SEQUENTIAL_ACCESS_UPGRADE_DISABLE))
+	},
+	{ XE_RTP_NAME("14018471104"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(LSC_CHICKEN_BIT_0_UDW, ENABLE_SMP_LD_RENDER_SURFACE_CONTROL))
+	},
+	/*
+	 * Although this workaround isn't required for the RCS, disabling these
+	 * reports has no impact for our driver or the GuC, so we go ahead and
+	 * apply this to all engines for simplicity.
+	 */
+	{ XE_RTP_NAME("16021639441"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001)),
+	  XE_RTP_ACTIONS(SET(CSFE_CHICKEN1(0),
+			     GHWSP_CSB_REPORT_DIS |
+			     PPHWSP_CSB_AND_TIMESTAMP_REPORT_DIS,
+			     XE_RTP_ACTION_FLAG(ENGINE_BASE)))
+	},
+	{ XE_RTP_NAME("14019811474"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001),
+		       FUNC(xe_rtp_match_first_render_or_compute)),
+	  XE_RTP_ACTIONS(SET(LSC_CHICKEN_BIT_0, WR_REQ_CHAINING_DIS))
+	},
+
 	{}
 };
 
@@ -585,10 +634,24 @@ static const struct xe_rtp_entry_sr lrc_was[] = {
 	  XE_RTP_ACTIONS(SET(INSTPM(RENDER_RING_BASE), ENABLE_SEMAPHORE_POLL_BIT))
 	},
 	{ XE_RTP_NAME("18033852989"),
-	  XE_RTP_RULES(GRAPHICS_VERSION(2004), ENGINE_CLASS(RENDER)),
+	  XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2001, 2004), ENGINE_CLASS(RENDER)),
 	  XE_RTP_ACTIONS(SET(COMMON_SLICE_CHICKEN1, DISABLE_BOTTOM_CLIP_RECTANGLE_TEST))
 	},
 
+	/* Xe2_HPG */
+	{ XE_RTP_NAME("15010599737"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), ENGINE_CLASS(RENDER)),
+	  XE_RTP_ACTIONS(SET(CHICKEN_RASTER_1, DIS_SF_ROUND_NEAREST_EVEN))
+	},
+	{ XE_RTP_NAME("14019386621"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), ENGINE_CLASS(RENDER)),
+	  XE_RTP_ACTIONS(SET(VF_SCRATCHPAD, XE2_VFG_TED_CREDIT_INTERFACE_DISABLE))
+	},
+	{ XE_RTP_NAME("14020756599"),
+	  XE_RTP_RULES(GRAPHICS_VERSION(2001), ENGINE_CLASS(RENDER)),
+	  XE_RTP_ACTIONS(SET(WM_CHICKEN3, HIZ_PLANE_COMPRESSION_DIS))
+	},
+
 	{}
 };
 
-- 
2.25.1

[PATCH v2 10/11] drm/xe/xe2hpg: Introduce performance tuning changes for Xe2_HPG.

[Balasubramani Vivekanandan]

From: Shekhar Chauhan <shekhar.chauhan@intel.com>

Introduces performance tuning guide changes for Xe_HPG.

BSpec: 72161

v2: Switched to open upper bound for "Tuning: L3 Cache" setting.

Signed-off-by: Shekhar Chauhan <shekhar.chauhan@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
---
 drivers/gpu/drm/xe/xe_tuning.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_tuning.c b/drivers/gpu/drm/xe/xe_tuning.c
index bb6db2817ada..d4e6fa918942 100644
--- a/drivers/gpu/drm/xe/xe_tuning.c
+++ b/drivers/gpu/drm/xe/xe_tuning.c
@@ -28,7 +28,7 @@ static const struct xe_rtp_entry_sr gt_tunings[] = {
 	/* Xe2 */
 
 	{ XE_RTP_NAME("Tuning: L3 cache"),
-	  XE_RTP_RULES(GRAPHICS_VERSION(2004)),
+	  XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2001, XE_RTP_END_VERSION_UNDEFINED)),
 	  XE_RTP_ACTIONS(FIELD_SET(XEHP_L3SQCREG5, L3_PWM_TIMER_INIT_VAL_MASK,
 				   REG_FIELD_PREP(L3_PWM_TIMER_INIT_VAL_MASK, 0x7f)))
 	},
@@ -38,11 +38,11 @@ static const struct xe_rtp_entry_sr gt_tunings[] = {
 				   REG_FIELD_PREP(L3_PWM_TIMER_INIT_VAL_MASK, 0x7f)))
 	},
 	{ XE_RTP_NAME("Tuning: Compression Overfetch"),
-	  XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2004, XE_RTP_END_VERSION_UNDEFINED)),
+	  XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2001, XE_RTP_END_VERSION_UNDEFINED)),
 	  XE_RTP_ACTIONS(CLR(CCCHKNREG1, ENCOMPPERFFIX)),
 	},
 	{ XE_RTP_NAME("Tuning: Enable compressible partial write overfetch in L3"),
-	  XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2004, XE_RTP_END_VERSION_UNDEFINED)),
+	  XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2001, XE_RTP_END_VERSION_UNDEFINED)),
 	  XE_RTP_ACTIONS(SET(L3SQCREG3, COMPPWOVERFETCHEN))
 	},
 	{}
-- 
2.25.1

[PATCH v2 11/11] drm/xe/xe2hpm: Add initial set of workarounds

[Balasubramani Vivekanandan]

From: Gustavo Sousa <gustavo.sousa@intel.com>

Define the initial set of workarounds for Xe2_HPM.

Signed-off-by: Gustavo Sousa <gustavo.sousa@intel.com>
Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
---
 drivers/gpu/drm/xe/regs/xe_gt_regs.h |  4 ++++
 drivers/gpu/drm/xe/xe_wa.c           | 32 ++++++++++++++++++++++++++++
 2 files changed, 36 insertions(+)

diff --git a/drivers/gpu/drm/xe/regs/xe_gt_regs.h b/drivers/gpu/drm/xe/regs/xe_gt_regs.h
index 0ce79ba19bda..8fe811ea404a 100644
--- a/drivers/gpu/drm/xe/regs/xe_gt_regs.h
+++ b/drivers/gpu/drm/xe/regs/xe_gt_regs.h
@@ -280,6 +280,10 @@
 #define FORCEWAKE_GT				XE_REG(0xa188)
 
 #define PG_ENABLE				XE_REG(0xa210)
+#define   VD2_MFXVDENC_POWERGATE_ENABLE		REG_BIT(8)
+#define   VD2_HCP_POWERGATE_ENABLE		REG_BIT(7)
+#define   VD0_MFXVDENC_POWERGATE_ENABLE		REG_BIT(4)
+#define   VD0_HCP_POWERGATE_ENABLE		REG_BIT(3)
 
 #define CTC_MODE				XE_REG(0xa26c)
 #define   CTC_SHIFT_PARAMETER_MASK		REG_GENMASK(2, 1)
diff --git a/drivers/gpu/drm/xe/xe_wa.c b/drivers/gpu/drm/xe/xe_wa.c
index 014d27c126ae..632bd9066f8d 100644
--- a/drivers/gpu/drm/xe/xe_wa.c
+++ b/drivers/gpu/drm/xe/xe_wa.c
@@ -228,6 +228,28 @@ static const struct xe_rtp_entry_sr gt_was[] = {
 	  XE_RTP_ENTRY_FLAG(FOREACH_ENGINE),
 	},
 
+	/* Xe2_HPM */
+
+	{ XE_RTP_NAME("16021867713"),
+	  XE_RTP_RULES(MEDIA_VERSION(1301),
+		       ENGINE_CLASS(VIDEO_DECODE)),
+	  XE_RTP_ACTIONS(SET(VDBOX_CGCTL3F1C(0), MFXPIPE_CLKGATE_DIS)),
+	  XE_RTP_ENTRY_FLAG(FOREACH_ENGINE),
+	},
+	{ XE_RTP_NAME("14020316580"),
+	  XE_RTP_RULES(MEDIA_VERSION(1301)),
+	  XE_RTP_ACTIONS(CLR(PG_ENABLE,
+			     VD0_HCP_POWERGATE_ENABLE |
+			     VD0_MFXVDENC_POWERGATE_ENABLE |
+			     VD2_HCP_POWERGATE_ENABLE |
+			     VD2_MFXVDENC_POWERGATE_ENABLE)),
+	},
+	{ XE_RTP_NAME("14019449301"),
+	  XE_RTP_RULES(MEDIA_VERSION(1301), ENGINE_CLASS(VIDEO_DECODE)),
+	  XE_RTP_ACTIONS(SET(VDBOX_CGCTL3F08(0), CG3DDISHRS_CLKGATE_DIS)),
+	  XE_RTP_ENTRY_FLAG(FOREACH_ENGINE),
+	},
+
 	{}
 };
 
@@ -513,6 +535,16 @@ static const struct xe_rtp_entry_sr engine_was[] = {
 	  XE_RTP_ACTIONS(SET(LSC_CHICKEN_BIT_0, WR_REQ_CHAINING_DIS))
 	},
 
+	/* Xe2_HPM */
+
+	{ XE_RTP_NAME("16021639441"),
+	  XE_RTP_RULES(MEDIA_VERSION(1301)),
+	  XE_RTP_ACTIONS(SET(CSFE_CHICKEN1(0),
+			     GHWSP_CSB_REPORT_DIS |
+			     PPHWSP_CSB_AND_TIMESTAMP_REPORT_DIS,
+			     XE_RTP_ACTION_FLAG(ENGINE_BASE)))
+	},
+
 	{}
 };
 
-- 
2.25.1