Skip to contentSkip to conversation

Allwinner H6 Mali GPU support

PATCH Applied Clément Péron 5 participants 23 messages May 21, 2019 – Aug 28, 2019

Subsystems (7)
lore
Share & tools
lore
#0 lore raw

v6 cover Allwinner H6 Mali GPU support

Hi,

The Allwinner H6 has a Mali-T720 MP2 which should be supported by the new panfrost driver. This series fix two issues and introduce the dt-bindings but a simple benchmark show that it's still NOT WORKING.

I'm pushing it in case someone want to continue the work.

This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1].

One patch is from Icenowy Zheng where I changed the order as required by Rob Herring[2].

Thanks,
Clement

[1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32
[2] https://patchwork.kernel.org/patch/10699829/

[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002400400
[  345.209617] Reason: TODO
[  345.209617] raw fault status: 0x800002C1
[  345.209617] decoded fault status: SLAVE FAULT
[  345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  345.209617] access type 0x2: READ
[  345.209617] source id 0x8000
[  345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0,
status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9
[  346.055876] panfrost 1800000.gpu: mmu irq status=1
[  346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002C00A00
[  346.060680] Reason: TODO
[  346.060680] raw fault status: 0x810002C1
[  346.060680] decoded fault status: SLAVE FAULT
[  346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  346.060680] access type 0x2: READ
[  346.060680] source id 0x8100
[  346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1,
status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85
[  346.573913] panfrost 1800000.gpu: mmu irq status=1
[  346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002C00B80

Change in v5:
- Remove fix indent

Changes in v4:
- Add bus_clock probe
- Fix sanity check in io-pgtable
- Add vramp-delay
- Merge all boards into one patch
- Remove upstreamed Neil A. patch

Change in v3 (Thanks to Maxime Ripard):
- Reauthor Icenowy for her path

Changes in v2 (Thanks to Maxime Ripard):
- Drop GPU OPP Table
- Add clocks and clock-names in required

Clément Péron (5):
drm: panfrost: add optional bus_clock
iommu: io-pgtable: fix sanity check for non 48-bit mali iommu
dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible
arm64: dts: allwinner: Add ARM Mali GPU node for H6
arm64: dts: allwinner: Add mali GPU supply for H6 boards

Icenowy Zheng (1):
dt-bindings: gpu: add bus clock for Mali Midgard GPUs

 .../bindings/gpu/arm,mali-midgard.txt         | 15 ++++++++++++-
 .../dts/allwinner/sun50i-h6-beelink-gs1.dts   |  6 +++++
 .../dts/allwinner/sun50i-h6-orangepi-3.dts    |  6 +++++
 .../dts/allwinner/sun50i-h6-orangepi.dtsi     |  6 +++++
 .../boot/dts/allwinner/sun50i-h6-pine-h64.dts |  6 +++++
 arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi  | 14 ++++++++++++
 drivers/gpu/drm/panfrost/panfrost_device.c    | 22 +++++++++++++++++++
 drivers/gpu/drm/panfrost/panfrost_device.h    |  1 +
 drivers/iommu/io-pgtable-arm.c                |  2 +-
 9 files changed, 76 insertions(+), 2 deletions(-)
Signature · 6 lines
2.17.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
#1 lore raw

PATCH v6 1/6 drm: panfrost: add optional bus_clock

Allwinner H6 has an ARM Mali-T720 MP2 which required a bus_clock.

Add an optional bus_clock at the init of the panfrost driver.

---
 drivers/gpu/drm/panfrost/panfrost_device.c | 22 ++++++++++++++++++++++
 drivers/gpu/drm/panfrost/panfrost_device.h |  1 +
 2 files changed, 23 insertions(+)
drivers/gpu/drm/panfrost/panfrost_device.c +22 -0
index 3b2bced1b015..ccb8eb2a518c 100644
@@ -55,11 +55,33 @@ static int panfrost_clk_init(struct panfrost_device *pfdev)
 	if (err)
 		return err;
 
+	pfdev->bus_clock = devm_clk_get_optional(pfdev->dev, "bus");
+	if (IS_ERR(pfdev->bus_clock)) {
+		dev_err(pfdev->dev, "get bus_clock failed %ld\n",
+			PTR_ERR(pfdev->bus_clock));
+		return PTR_ERR(pfdev->bus_clock);
+	}
+
+	if (pfdev->bus_clock) {
+		rate = clk_get_rate(pfdev->bus_clock);
+		dev_info(pfdev->dev, "bus_clock rate = %lu\n", rate);
+
+		err = clk_prepare_enable(pfdev->bus_clock);
+		if (err)
+			goto disable_clock;
+	}
+
 	return 0;
+
+disable_clock:
+	clk_disable_unprepare(pfdev->clock);
+
+	return err;
 }
 
 static void panfrost_clk_fini(struct panfrost_device *pfdev)
 {
+	clk_disable_unprepare(pfdev->bus_clock);
 	clk_disable_unprepare(pfdev->clock);
 }
 
drivers/gpu/drm/panfrost/panfrost_device.h +1 -0
index 56f452dfb490..8074f221034b 100644
@@ -66,6 +66,7 @@ struct panfrost_device {
 
 	void __iomem *iomem;
 	struct clk *clock;
+	struct clk *bus_clock;
 	struct regulator *regulator;
 	struct reset_control *rstc;
 
Signature · 6 lines
2.17.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
#2 lore raw

PATCH v6 2/6 iommu: io-pgtable: fix sanity check for non 48-bit mali iommu

In reply to Clément Péron

Allwinner H6 SoC has a Mali T720MP2 with a 33-bit iommu which trig the sanity check during the alloc of the pgtable.

Change the sanity check to allow non 48-bit configuration.

---
 drivers/iommu/io-pgtable-arm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iommu/io-pgtable-arm.c +1 -1
index 4e21efbc4459..74f2ce802e6f 100644
@@ -1016,7 +1016,7 @@ arm_mali_lpae_alloc_pgtable(struct io_pgtable_cfg *cfg, void *cookie)
 {
 	struct io_pgtable *iop;
 
-	if (cfg->ias != 48 || cfg->oas > 40)
+	if (cfg->ias > 48 || cfg->oas > 40)
 		return NULL;
 
 	cfg->pgsize_bitmap &= (SZ_4K | SZ_2M | SZ_1G);
Signature · 1 line
2.17.1
#3 lore raw

PATCH v6 3/6 dt-bindings: gpu: add bus clock for Mali Midgard GPUs

In reply to Clément Péron

From: Icenowy Zheng <[email protected]>

Some SoCs adds a bus clock gate to the Mali Midgard GPU.

Add the binding for the bus clock.

---
 Documentation/devicetree/bindings/gpu/arm,mali-midgard.txt | 6 ++++++
 1 file changed, 6 insertions(+)
Documentation/devicetree/bindings/gpu/arm,mali-midgard.txt +6 -0
index 1b1a74129141..2e8bbce35695 100644
@@ -31,6 +31,12 @@ Optional properties:
 
 - clocks : Phandle to clock for the Mali Midgard device.
 
+- clock-names : Specify the names of the clocks specified in clocks
+  when multiple clocks are present.
+    * core: clock driving the GPU itself (When only one clock is present,
+      assume it's this clock.)
+    * bus: bus clock for the GPU
+
 - mali-supply : Phandle to regulator for the Mali device. Refer to
   Documentation/devicetree/bindings/regulator/regulator.txt for details.
 
Signature · 6 lines
2.17.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
#4 lore raw

PATCH v6 4/6 dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible

In reply to Clément Péron

This add the H6 mali compatible in the dt-bindings to later support specific implementation.

---
 .../devicetree/bindings/gpu/arm,mali-midgard.txt         | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
Documentation/devicetree/bindings/gpu/arm,mali-midgard.txt +8 -1
index 2e8bbce35695..4bf17e1cf555 100644
@@ -15,6 +15,7 @@ Required properties:
     + "arm,mali-t860"
     + "arm,mali-t880"
   * which must be preceded by one of the following vendor specifics:
+    + "allwinner,sun50i-h6-mali"
     + "amlogic,meson-gxm-mali"
     + "rockchip,rk3288-mali"
     + "rockchip,rk3399-mali"
@@ -49,9 +50,15 @@ Vendor-specific bindings
 ------------------------
 
 The Mali GPU is integrated very differently from one SoC to
-another. In order to accomodate those differences, you have the option
+another. In order to accommodate those differences, you have the option
 to specify one more vendor-specific compatible, among:
 
+- "allwinner,sun50i-h6-mali"
+  Required properties:
+  - clocks : phandles to core and bus clocks
+  - clock-names : must contain "core" and "bus"
+  - resets: phandle to GPU reset line
+
 - "amlogic,meson-gxm-mali"
   Required properties:
   - resets : Should contain phandles of :
Signature · 1 line
2.17.1
#5 lore raw

PATCH v6 5/6 arm64: dts: allwinner: Add ARM Mali GPU node for H6

In reply to Clément Péron

Add the mali gpu node to the H6 device-tree.

---
 arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi +14 -0
index 16c5c3d0fd81..6aad06095c40 100644
@@ -157,6 +157,20 @@
 			allwinner,sram = <&ve_sram 1>;
 		};
 
+		gpu: gpu@1800000 {
+			compatible = "allwinner,sun50i-h6-mali",
+				     "arm,mali-t720";
+			reg = <0x01800000 0x4000>;
+			interrupts = <GIC_SPI 84 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 85 IRQ_TYPE_LEVEL_HIGH>,
+				     <GIC_SPI 83 IRQ_TYPE_LEVEL_HIGH>;
+			interrupt-names = "job", "mmu", "gpu";
+			clocks = <&ccu CLK_GPU>, <&ccu CLK_BUS_GPU>;
+			clock-names = "core", "bus";
+			resets = <&ccu RST_BUS_GPU>;
+			status = "disabled";
+		};
+
 		syscon: syscon@3000000 {
 			compatible = "allwinner,sun50i-h6-system-control",
 				     "allwinner,sun50i-a64-system-control";
Signature · 1 line
2.17.1
#6 lore raw

PATCH v6 6/6 arm64: dts: allwinner: Add mali GPU supply for H6 boards

In reply to Clément Péron

Enable and add supply to the Mali GPU node on all the H6 boards.

Regarding the datasheet the maximum time for supply to reach its voltage is 32ms.

---
 arch/arm64/boot/dts/allwinner/sun50i-h6-beelink-gs1.dts | 6 ++++++
 arch/arm64/boot/dts/allwinner/sun50i-h6-orangepi-3.dts  | 6 ++++++
 arch/arm64/boot/dts/allwinner/sun50i-h6-orangepi.dtsi   | 6 ++++++
 arch/arm64/boot/dts/allwinner/sun50i-h6-pine-h64.dts    | 6 ++++++
 4 files changed, 24 insertions(+)
arch/arm64/boot/dts/allwinner/sun50i-h6-beelink-gs1.dts +6 -0
index 0dc33c90dd60..fe36c6588d8e 100644
@@ -70,6 +70,11 @@
 	status = "okay";
 };
 
+&gpu {
+	mali-supply = <&reg_dcdcc>;
+	status = "okay";
+};
+
 &hdmi {
 	status = "okay";
 };
@@ -206,6 +211,7 @@
 			};
 
 			reg_dcdcc: dcdcc {
+				regulator-enable-ramp-delay = <32000>;
 				regulator-min-microvolt = <810000>;
 				regulator-max-microvolt = <1080000>;
 				regulator-name = "vdd-gpu";
arch/arm64/boot/dts/allwinner/sun50i-h6-orangepi-3.dts +6 -0
index 17d496990108..ea4866b0fa7a 100644
@@ -58,6 +58,11 @@
 	status = "okay";
 };
 
+&gpu {
+	mali-supply = <&reg_dcdcc>;
+	status = "okay";
+};
+
 &mmc0 {
 	vmmc-supply = <&reg_cldo1>;
 	cd-gpios = <&pio 5 6 GPIO_ACTIVE_LOW>; /* PF6 */
@@ -165,6 +170,7 @@
 			};
 
 			reg_dcdcc: dcdcc {
+				regulator-enable-ramp-delay = <32000>;
 				regulator-min-microvolt = <810000>;
 				regulator-max-microvolt = <1080000>;
 				regulator-name = "vdd-gpu";
arch/arm64/boot/dts/allwinner/sun50i-h6-orangepi.dtsi +6 -0
index 62e27948a3fa..ec770f07aa82 100644
@@ -55,6 +55,11 @@
 	status = "okay";
 };
 
+&gpu {
+	mali-supply = <&reg_dcdcc>;
+	status = "okay";
+};
+
 &mmc0 {
 	vmmc-supply = <&reg_cldo1>;
 	cd-gpios = <&pio 5 6 GPIO_ACTIVE_LOW>;
@@ -163,6 +168,7 @@
 			};
 
 			reg_dcdcc: dcdcc {
+				regulator-enable-ramp-delay = <32000>;
 				regulator-min-microvolt = <810000>;
 				regulator-max-microvolt = <1080000>;
 				regulator-name = "vdd-gpu";
arch/arm64/boot/dts/allwinner/sun50i-h6-pine-h64.dts +6 -0
index 4802902e128f..625a29a25c52 100644
@@ -85,6 +85,11 @@
 	status = "okay";
 };
 
+&gpu {
+	mali-supply = <&reg_dcdcc>;
+	status = "okay";
+};
+
 &hdmi {
 	status = "okay";
 };
@@ -215,6 +220,7 @@
 			};
 
 			reg_dcdcc: dcdcc {
+				regulator-enable-ramp-delay = <32000>;
 				regulator-min-microvolt = <810000>;
 				regulator-max-microvolt = <1080000>;
 				regulator-name = "vdd-gpu";
Signature · 1 line
2.17.1
Applied
#7 lore raw
In reply to Clément Péron
Quoted 71 lines · Clément Péron

Hi,

The Allwinner H6 has a Mali-T720 MP2 which should be supported by the new panfrost driver. This series fix two issues and introduce the dt-bindings but a simple benchmark show that it's still NOT WORKING.

I'm pushing it in case someone want to continue the work.

This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1].

One patch is from Icenowy Zheng where I changed the order as required by Rob Herring[2].

Thanks,
Clement

[1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32
[2] https://patchwork.kernel.org/patch/10699829/

[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002400400
[  345.209617] Reason: TODO
[  345.209617] raw fault status: 0x800002C1
[  345.209617] decoded fault status: SLAVE FAULT
[  345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  345.209617] access type 0x2: READ
[  345.209617] source id 0x8000
[  345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0,
status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9
[  346.055876] panfrost 1800000.gpu: mmu irq status=1
[  346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002C00A00
[  346.060680] Reason: TODO
[  346.060680] raw fault status: 0x810002C1
[  346.060680] decoded fault status: SLAVE FAULT
[  346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  346.060680] access type 0x2: READ
[  346.060680] source id 0x8100
[  346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1,
status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85
[  346.573913] panfrost 1800000.gpu: mmu irq status=1
[  346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002C00B80

Change in v5:
- Remove fix indent

Changes in v4:
- Add bus_clock probe
- Fix sanity check in io-pgtable
- Add vramp-delay
- Merge all boards into one patch
- Remove upstreamed Neil A. patch

Change in v3 (Thanks to Maxime Ripard):
- Reauthor Icenowy for her path

Changes in v2 (Thanks to Maxime Ripard):
- Drop GPU OPP Table
- Add clocks and clock-names in required

Clément Péron (5):
drm: panfrost: add optional bus_clock
iommu: io-pgtable: fix sanity check for non 48-bit mali iommu
dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible
arm64: dts: allwinner: Add ARM Mali GPU node for H6
arm64: dts: allwinner: Add mali GPU supply for H6 boards

Quoted 2 lines

Icenowy Zheng (1):
dt-bindings: gpu: add bus clock for Mali Midgard GPUs

I've applied patches 1 and 3 to drm-misc. I was going to do patch 4 too, but it doesn't apply.

Patch 2 can go in via the iommu tree and the rest via the allwinner tree.

Rob
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu

#8 lore raw

Hi Rob,

Quoted 73 lines · Rob Herring
Quoted 71 lines · Clément Péron

Hi,

The Allwinner H6 has a Mali-T720 MP2 which should be supported by the new panfrost driver. This series fix two issues and introduce the dt-bindings but a simple benchmark show that it's still NOT WORKING.

I'm pushing it in case someone want to continue the work.

This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1].

One patch is from Icenowy Zheng where I changed the order as required by Rob Herring[2].

Thanks,
Clement

[1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32
[2] https://patchwork.kernel.org/patch/10699829/

[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002400400
[  345.209617] Reason: TODO
[  345.209617] raw fault status: 0x800002C1
[  345.209617] decoded fault status: SLAVE FAULT
[  345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  345.209617] access type 0x2: READ
[  345.209617] source id 0x8000
[  345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0,
status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9
[  346.055876] panfrost 1800000.gpu: mmu irq status=1
[  346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002C00A00
[  346.060680] Reason: TODO
[  346.060680] raw fault status: 0x810002C1
[  346.060680] decoded fault status: SLAVE FAULT
[  346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  346.060680] access type 0x2: READ
[  346.060680] source id 0x8100
[  346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1,
status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85
[  346.573913] panfrost 1800000.gpu: mmu irq status=1
[  346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002C00B80

Change in v5:
- Remove fix indent

Changes in v4:
- Add bus_clock probe
- Fix sanity check in io-pgtable
- Add vramp-delay
- Merge all boards into one patch
- Remove upstreamed Neil A. patch

Change in v3 (Thanks to Maxime Ripard):
- Reauthor Icenowy for her path

Changes in v2 (Thanks to Maxime Ripard):
- Drop GPU OPP Table
- Add clocks and clock-names in required

Clément Péron (5):
drm: panfrost: add optional bus_clock
iommu: io-pgtable: fix sanity check for non 48-bit mali iommu
dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible
arm64: dts: allwinner: Add ARM Mali GPU node for H6
arm64: dts: allwinner: Add mali GPU supply for H6 boards

Quoted 5 lines
Quoted 2 lines

Icenowy Zheng (1):
dt-bindings: gpu: add bus clock for Mali Midgard GPUs

I've applied patches 1 and 3 to drm-misc. I was going to do patch 4 too, but it doesn't apply.

Thanks,

I have tried to applied on drm-misc/for-linux-next but it doesn't apply too. It looks like commit d5ff1adb3809e2f74a3b57cea2e57c8e37d693c4 is missing on drm-misc ?
https://github.com/torvalds/linux/commit/d5ff1adb3809e2f74a3b57cea2e57c8e37d693c4#diff-c3172f5d421d492ff91d7bb44dd44917

Clément

Quoted 4 lines

Patch 2 can go in via the iommu tree and the rest via the allwinner tree.

Rob

#9 lore raw
Quoted 83 lines · Clément Péron

Hi Rob,

Quoted 73 lines · Rob Herring
Quoted 71 lines · Clément Péron

Hi,

The Allwinner H6 has a Mali-T720 MP2 which should be supported by the new panfrost driver. This series fix two issues and introduce the dt-bindings but a simple benchmark show that it's still NOT WORKING.

I'm pushing it in case someone want to continue the work.

This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1].

One patch is from Icenowy Zheng where I changed the order as required by Rob Herring[2].

Thanks,
Clement

[1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32
[2] https://patchwork.kernel.org/patch/10699829/

[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002400400
[  345.209617] Reason: TODO
[  345.209617] raw fault status: 0x800002C1
[  345.209617] decoded fault status: SLAVE FAULT
[  345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  345.209617] access type 0x2: READ
[  345.209617] source id 0x8000
[  345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0,
status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9
[  346.055876] panfrost 1800000.gpu: mmu irq status=1
[  346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002C00A00
[  346.060680] Reason: TODO
[  346.060680] raw fault status: 0x810002C1
[  346.060680] decoded fault status: SLAVE FAULT
[  346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  346.060680] access type 0x2: READ
[  346.060680] source id 0x8100
[  346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1,
status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85
[  346.573913] panfrost 1800000.gpu: mmu irq status=1
[  346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002C00B80

Change in v5:
- Remove fix indent

Changes in v4:
- Add bus_clock probe
- Fix sanity check in io-pgtable
- Add vramp-delay
- Merge all boards into one patch
- Remove upstreamed Neil A. patch

Change in v3 (Thanks to Maxime Ripard):
- Reauthor Icenowy for her path

Changes in v2 (Thanks to Maxime Ripard):
- Drop GPU OPP Table
- Add clocks and clock-names in required

Clément Péron (5):
drm: panfrost: add optional bus_clock
iommu: io-pgtable: fix sanity check for non 48-bit mali iommu
dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible
arm64: dts: allwinner: Add ARM Mali GPU node for H6
arm64: dts: allwinner: Add mali GPU supply for H6 boards

Quoted 5 lines
Quoted 2 lines

Icenowy Zheng (1):
dt-bindings: gpu: add bus clock for Mali Midgard GPUs

I've applied patches 1 and 3 to drm-misc. I was going to do patch 4 too, but it doesn't apply.

Quoted 6 lines

Thanks,

I have tried to applied on drm-misc/for-linux-next but it doesn't apply too. It looks like commit d5ff1adb3809e2f74a3b57cea2e57c8e37d693c4 is missing on drm-misc ?
https://github.com/torvalds/linux/commit/d5ff1adb3809e2f74a3b57cea2e57c8e37d693c4#diff-c3172f5d421d492ff91d7bb44dd44917

5.2-rc1 is merged in now and I've applied patch 4.

Rob

reviewer
#11 lore raw
In reply to Clément Péron
Quoted 44 lines · Clément Péron

Hi,

The Allwinner H6 has a Mali-T720 MP2 which should be supported by the new panfrost driver. This series fix two issues and introduce the dt-bindings but a simple benchmark show that it's still NOT WORKING.

I'm pushing it in case someone want to continue the work.

This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1].

One patch is from Icenowy Zheng where I changed the order as required by Rob Herring[2].

Thanks,
Clement

[1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32
[2] https://patchwork.kernel.org/patch/10699829/

[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002400400
[  345.209617] Reason: TODO
[  345.209617] raw fault status: 0x800002C1
[  345.209617] decoded fault status: SLAVE FAULT
[  345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  345.209617] access type 0x2: READ
[  345.209617] source id 0x8000
[  345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0,
status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9
[  346.055876] panfrost 1800000.gpu: mmu irq status=1
[  346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002C00A00
[  346.060680] Reason: TODO
[  346.060680] raw fault status: 0x810002C1
[  346.060680] decoded fault status: SLAVE FAULT
[  346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  346.060680] access type 0x2: READ
[  346.060680] source id 0x8100
[  346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1,
status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85
[  346.573913] panfrost 1800000.gpu: mmu irq status=1
[  346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
Quoted 1 line

0x0000000002C00B80

FWIW I seem to have reproduced the same behaviour on a different T720 setup, so this may well be more about the GPU than the platform. There doesn't look to be anything obviously wrong with the pagetables, but if I can find some more free time I may have a bit more of a poke around.

Robin.

Quoted 28 lines

Change in v5:
- Remove fix indent

Changes in v4:
- Add bus_clock probe
- Fix sanity check in io-pgtable
- Add vramp-delay
- Merge all boards into one patch
- Remove upstreamed Neil A. patch

Change in v3 (Thanks to Maxime Ripard):
- Reauthor Icenowy for her path

Changes in v2 (Thanks to Maxime Ripard):
- Drop GPU OPP Table
- Add clocks and clock-names in required

Clément Péron (5):
drm: panfrost: add optional bus_clock
iommu: io-pgtable: fix sanity check for non 48-bit mali iommu
dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible
arm64: dts: allwinner: Add ARM Mali GPU node for H6
arm64: dts: allwinner: Add mali GPU supply for H6 boards

Icenowy Zheng (1):
dt-bindings: gpu: add bus clock for Mali Midgard GPUs

Quoted 11 lines
  .../bindings/gpu/arm,mali-midgard.txt         | 15 ++++++++++++-
  .../dts/allwinner/sun50i-h6-beelink-gs1.dts   |  6 +++++
  .../dts/allwinner/sun50i-h6-orangepi-3.dts    |  6 +++++
  .../dts/allwinner/sun50i-h6-orangepi.dtsi     |  6 +++++
  .../boot/dts/allwinner/sun50i-h6-pine-h64.dts |  6 +++++
  arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi  | 14 ++++++++++++
  drivers/gpu/drm/panfrost/panfrost_device.c    | 22 +++++++++++++++++++
  drivers/gpu/drm/panfrost/panfrost_device.h    |  1 +
  drivers/iommu/io-pgtable-arm.c                |  2 +-
  9 files changed, 76 insertions(+), 2 deletions(-)

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu

#12 lore raw
In reply to Clément Péron
Quoted 1 line · Clément Péron

[snip]

Quoted 5 lines
[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002400400

From what I can see here, 0x0000000002400400 points to the first byte

of the first submitted job descriptor.

So mapping buffers for the GPU doesn't seem to be working at all on 64-bit T-760.

Steven, Robin, do you have any idea of why this could be?

Thanks,

Tomeu

#20 lore raw

PATCH 1/1 arm64: dts: allwinner: Add GPU operating points for H6

In reply to Clément Péron

The GPU driver needs them to change the clock frequency and regulator voltage depending on the load.

Cc Clément Péron <[email protected]>

---

Feel free to pick up this patch if you are going to keep pushing this series forward.

Thanks,

Tomeu

---
 arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi | 66 ++++++++++++++++++++
 1 file changed, 66 insertions(+)
arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi +66 -0
index 6aad06095c40..decf7b56e2df 100644
@@ -157,6 +157,71 @@
 			allwinner,sram = <&ve_sram 1>;
 		};
 
+		gpu_opp_table: opp-table2 {
+			compatible = "operating-points-v2";
+
+			opp00 {
+				opp-hz = /bits/ 64 <756000000>;
+				opp-microvolt = <1040000>;
+			};
+			opp01 {
+				opp-hz = /bits/ 64 <624000000>;
+				opp-microvolt = <950000>;
+			};
+			opp02 {
+				opp-hz = /bits/ 64 <576000000>;
+				opp-microvolt = <930000>;
+			};
+			opp03 {
+				opp-hz = /bits/ 64 <540000000>;
+				opp-microvolt = <910000>;
+			};
+			opp04 {
+				opp-hz = /bits/ 64 <504000000>;
+				opp-microvolt = <890000>;
+			};
+			opp05 {
+				opp-hz = /bits/ 64 <456000000>;
+				opp-microvolt = <870000>;
+			};
+			opp06 {
+				opp-hz = /bits/ 64 <432000000>;
+				opp-microvolt = <860000>;
+			};
+			opp07 {
+				opp-hz = /bits/ 64 <420000000>;
+				opp-microvolt = <850000>;
+			};
+			opp08 {
+				opp-hz = /bits/ 64 <408000000>;
+				opp-microvolt = <840000>;
+			};
+			opp09 {
+				opp-hz = /bits/ 64 <384000000>;
+				opp-microvolt = <830000>;
+			};
+			opp10 {
+				opp-hz = /bits/ 64 <360000000>;
+				opp-microvolt = <820000>;
+			};
+			opp11 {
+				opp-hz = /bits/ 64 <336000000>;
+				opp-microvolt = <810000>;
+			};
+			opp12 {
+				opp-hz = /bits/ 64 <312000000>;
+				opp-microvolt = <810000>;
+			};
+			opp13 {
+				opp-hz = /bits/ 64 <264000000>;
+				opp-microvolt = <810000>;
+			};
+			opp14 {
+				opp-hz = /bits/ 64 <216000000>;
+				opp-microvolt = <810000>;
+			};
+		};
+
 		gpu: gpu@1800000 {
 			compatible = "allwinner,sun50i-h6-mali",
 				     "arm,mali-t720";
@@ -168,6 +233,7 @@
 			clocks = <&ccu CLK_GPU>, <&ccu CLK_BUS_GPU>;
 			clock-names = "core", "bus";
 			resets = <&ccu RST_BUS_GPU>;
+			operating-points-v2 = <&gpu_opp_table>;
 			status = "disabled";
 		};
 
Signature · 1 line
2.20.1
#21 lore raw

Re: [PATCH] arm64: dts: allwinner: Add GPU operating points for H6

Hi Tomeu,

Quoted 3 lines · Tomeu Vizoso

The GPU driver needs them to change the clock frequency and regulator voltage depending on the load.

As requested by Maxime, I have dropped these OPP table as It's taken from vendor and untested with Panfrost.

https://lore.kernel.org/patchwork/patch/1060374/

Regards,
Clément

Quoted 103 lines
Cc Clément Péron <[email protected]>

---

Feel free to pick up this patch if you are going to keep pushing this series forward.

Thanks,

Tomeu

---
 arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi | 66 ++++++++++++++++++++
 1 file changed, 66 insertions(+)
arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi +66 -1
index 6aad06095c40..decf7b56e2df 100644
@@ -157,6 +157,71 @@
                        allwinner,sram = <&ve_sram 1>;
                };
 
+               gpu_opp_table: opp-table2 {
+                       compatible = "operating-points-v2";
+
+                       opp00 {
+                               opp-hz = /bits/ 64 <756000000>;
+                               opp-microvolt = <1040000>;
+                       };
+                       opp01 {
+                               opp-hz = /bits/ 64 <624000000>;
+                               opp-microvolt = <950000>;
+                       };
+                       opp02 {
+                               opp-hz = /bits/ 64 <576000000>;
+                               opp-microvolt = <930000>;
+                       };
+                       opp03 {
+                               opp-hz = /bits/ 64 <540000000>;
+                               opp-microvolt = <910000>;
+                       };
+                       opp04 {
+                               opp-hz = /bits/ 64 <504000000>;
+                               opp-microvolt = <890000>;
+                       };
+                       opp05 {
+                               opp-hz = /bits/ 64 <456000000>;
+                               opp-microvolt = <870000>;
+                       };
+                       opp06 {
+                               opp-hz = /bits/ 64 <432000000>;
+                               opp-microvolt = <860000>;
+                       };
+                       opp07 {
+                               opp-hz = /bits/ 64 <420000000>;
+                               opp-microvolt = <850000>;
+                       };
+                       opp08 {
+                               opp-hz = /bits/ 64 <408000000>;
+                               opp-microvolt = <840000>;
+                       };
+                       opp09 {
+                               opp-hz = /bits/ 64 <384000000>;
+                               opp-microvolt = <830000>;
+                       };
+                       opp10 {
+                               opp-hz = /bits/ 64 <360000000>;
+                               opp-microvolt = <820000>;
+                       };
+                       opp11 {
+                               opp-hz = /bits/ 64 <336000000>;
+                               opp-microvolt = <810000>;
+                       };
+                       opp12 {
+                               opp-hz = /bits/ 64 <312000000>;
+                               opp-microvolt = <810000>;
+                       };
+                       opp13 {
+                               opp-hz = /bits/ 64 <264000000>;
+                               opp-microvolt = <810000>;
+                       };
+                       opp14 {
+                               opp-hz = /bits/ 64 <216000000>;
+                               opp-microvolt = <810000>;
+                       };
+               };
+
                gpu: gpu@1800000 {
                        compatible = "allwinner,sun50i-h6-mali",
                                     "arm,mali-t720";
@@ -168,6 +233,7 @@
                        clocks = <&ccu CLK_GPU>, <&ccu CLK_BUS_GPU>;
                        clock-names = "core", "bus";
                        resets = <&ccu RST_BUS_GPU>;
+                       operating-points-v2 = <&gpu_opp_table>;
                        status = "disabled";
                };
 
--

2.20.1

#22 lore raw

Re: [PATCH] arm64: dts: allwinner: Add GPU operating points for H6

Quoted 5 lines · Clément Péron

Hi Tomeu,

Quoted 1 line · Tomeu Vizoso
Quoted 7 lines
Quoted 2 lines

The GPU driver needs them to change the clock frequency and regulator voltage depending on the load.

As requested by Maxime, I have dropped these OPP table as It's taken from vendor and untested with Panfrost.

https://lore.kernel.org/patchwork/patch/1060374/

Ok, guess this series should wait then until we can run Panfrost on it and check how DVFS is working.

Thanks,

Tomeu

Quoted 106 lines

Regards,
Clément

Quoted 103 lines
Cc Clément Péron <[email protected]>

---

Feel free to pick up this patch if you are going to keep pushing this series forward.

Thanks,

Tomeu

---
 arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi | 66 ++++++++++++++++++++
 1 file changed, 66 insertions(+)
arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi +66 -1
index 6aad06095c40..decf7b56e2df 100644
@@ -157,6 +157,71 @@
                        allwinner,sram = <&ve_sram 1>;
                };
 
+               gpu_opp_table: opp-table2 {
+                       compatible = "operating-points-v2";
+
+                       opp00 {
+                               opp-hz = /bits/ 64 <756000000>;
+                               opp-microvolt = <1040000>;
+                       };
+                       opp01 {
+                               opp-hz = /bits/ 64 <624000000>;
+                               opp-microvolt = <950000>;
+                       };
+                       opp02 {
+                               opp-hz = /bits/ 64 <576000000>;
+                               opp-microvolt = <930000>;
+                       };
+                       opp03 {
+                               opp-hz = /bits/ 64 <540000000>;
+                               opp-microvolt = <910000>;
+                       };
+                       opp04 {
+                               opp-hz = /bits/ 64 <504000000>;
+                               opp-microvolt = <890000>;
+                       };
+                       opp05 {
+                               opp-hz = /bits/ 64 <456000000>;
+                               opp-microvolt = <870000>;
+                       };
+                       opp06 {
+                               opp-hz = /bits/ 64 <432000000>;
+                               opp-microvolt = <860000>;
+                       };
+                       opp07 {
+                               opp-hz = /bits/ 64 <420000000>;
+                               opp-microvolt = <850000>;
+                       };
+                       opp08 {
+                               opp-hz = /bits/ 64 <408000000>;
+                               opp-microvolt = <840000>;
+                       };
+                       opp09 {
+                               opp-hz = /bits/ 64 <384000000>;
+                               opp-microvolt = <830000>;
+                       };
+                       opp10 {
+                               opp-hz = /bits/ 64 <360000000>;
+                               opp-microvolt = <820000>;
+                       };
+                       opp11 {
+                               opp-hz = /bits/ 64 <336000000>;
+                               opp-microvolt = <810000>;
+                       };
+                       opp12 {
+                               opp-hz = /bits/ 64 <312000000>;
+                               opp-microvolt = <810000>;
+                       };
+                       opp13 {
+                               opp-hz = /bits/ 64 <264000000>;
+                               opp-microvolt = <810000>;
+                       };
+                       opp14 {
+                               opp-hz = /bits/ 64 <216000000>;
+                               opp-microvolt = <810000>;
+                       };
+               };
+
                gpu: gpu@1800000 {
                        compatible = "allwinner,sun50i-h6-mali",
                                     "arm,mali-t720";
@@ -168,6 +233,7 @@
                        clocks = <&ccu CLK_GPU>, <&ccu CLK_BUS_GPU>;
                        clock-names = "core", "bus";
                        resets = <&ccu RST_BUS_GPU>;
+                       operating-points-v2 = <&gpu_opp_table>;
                        status = "disabled";
                };
 
--

2.20.1

reviewer
#13 lore raw
In reply to Tomeu Vizoso
Quoted 7 lines · Tomeu Vizoso
Quoted 1 line · Clément Péron

[snip]

Quoted 3 lines
[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002400400

Quoted 7 lines

From what I can see here, 0x0000000002400400 points to the first byte
of the first submitted job descriptor.

So mapping buffers for the GPU doesn't seem to be working at all on 64-bit T-760.

Steven, Robin, do you have any idea of why this could be?

I tried rolling back to the old panfrost/nondrm shim, and it works fine with kbase, and I also found that T-820 falls over in the exact same manner, so the fact that it seemed to be common to the smaller 33-bit designs rather than anything to do with the other job_descriptor_size/v4/v5 complication turned out to be telling.

[ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or is it just that nobody's yet observed a 64-bit blob driving one? ]

Long story short, it appears that 'Mali LPAE' is also lacking the start level notion of VMSA, and expects a full 4-level table even for <40 bits when level 0 effectively redundant. Thus walking the 3-level table that io-pgtable comes back with ends up going wildly wrong. The hack below seems to do the job for me; if Clément can confirm (on T-720 you'll still need the userspace hack to force 32-bit jobs as well) then I think I'll cook up a proper refactoring of the allocator to put things right.

Robin.

----->8-----

drivers/iommu/io-pgtable-arm.c +2 -0
index 546968d8a349..f29da6e8dc08 100644
@@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct io_pgtable_cfg *cfg, void *cookie)
 	iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie);
 	if (iop) {
 		u64 mair, ttbr;
+		struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops);
 
+		data->levels = 4;
 		/* Copy values as union fields overlap */
 		mair = cfg->arm_lpae_s1_cfg.mair[0];
 		ttbr = cfg->arm_lpae_s1_cfg.ttbr[0];
#14 lore raw
Quoted 15 lines · Robin Murphy
Quoted 13 lines · Tomeu Vizoso
Quoted 1 line · Clément Péron

[snip]

Quoted 3 lines
[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002400400

From what I can see here, 0x0000000002400400 points to the first byte
of the first submitted job descriptor.

So mapping buffers for the GPU doesn't seem to be working at all on 64-bit T-760.

Quoted 7 lines
Quoted 1 line

Steven, Robin, do you have any idea of why this could be?

I tried rolling back to the old panfrost/nondrm shim, and it works fine with kbase, and I also found that T-820 falls over in the exact same manner, so the fact that it seemed to be common to the smaller 33-bit designs rather than anything to do with the other job_descriptor_size/v4/v5 complication turned out to be telling.

Is this complication something you can explain? I don't know what v4 and v5 are meant here.

Quoted 2 lines

[ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or is it just that nobody's yet observed a 64-bit blob driving one? ]

I'm looking right now at getting Panfrost working on T720 with 64-bit descriptors, with the ultimate goal of making Panfrost 64-bit-descriptor only so we can have a single build of Mesa in distros.

Quoted 7 lines

Long story short, it appears that 'Mali LPAE' is also lacking the start level notion of VMSA, and expects a full 4-level table even for <40 bits when level 0 effectively redundant. Thus walking the 3-level table that io-pgtable comes back with ends up going wildly wrong. The hack below seems to do the job for me; if Clément can confirm (on T-720 you'll still need the userspace hack to force 32-bit jobs as well) then I think I'll cook up a proper refactoring of the allocator to put things right.

Mmaps seem to work with this patch, thanks.

The main complication I'm facing right now seems to be that the SFBD descriptor on T720 seems to be different from the one we already had (tested on T6xx?).

Cheers,

Tomeu

Quoted 24 lines

Robin.

----->8-----

drivers/iommu/io-pgtable-arm.c +2 -0
index 546968d8a349..f29da6e8dc08 100644
@@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct
 io_pgtable_cfg *cfg, void *cookie)
        iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie);
        if (iop) {
                u64 mair, ttbr;
+               struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops);
 
+               data->levels = 4;
                /* Copy values as union fields overlap */
                mair = cfg->arm_lpae_s1_cfg.mair[0];
                ttbr = cfg->arm_lpae_s1_cfg.ttbr[0];
 

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

reviewer
#15 lore raw
Quoted 18 lines · Tomeu Vizoso
Quoted 17 lines · Robin Murphy
Quoted 14 lines · Tomeu Vizoso
Quoted 1 line · Clément Péron

[snip]

Quoted 3 lines
[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002400400

From what I can see here, 0x0000000002400400 points to the first byte
of the first submitted job descriptor.

So mapping buffers for the GPU doesn't seem to be working at all on 64-bit T-760.

Steven, Robin, do you have any idea of why this could be?

Quoted 8 lines
Quoted 5 lines

I tried rolling back to the old panfrost/nondrm shim, and it works fine with kbase, and I also found that T-820 falls over in the exact same manner, so the fact that it seemed to be common to the smaller 33-bit designs rather than anything to do with the other job_descriptor_size/v4/v5 complication turned out to be telling.

Is this complication something you can explain? I don't know what v4 and v5 are meant here.

I was alluding to BASE_HW_FEATURE_V4, which I believe refers to the Midgard architecture version - the older versions implemented by T6xx and T720 seem to be collectively treated as "v4", while T760 and T8xx would effectively be "v5".

Quoted 7 lines
Quoted 2 lines

[ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or is it just that nobody's yet observed a 64-bit blob driving one? ]

I'm looking right now at getting Panfrost working on T720 with 64-bit descriptors, with the ultimate goal of making Panfrost 64-bit-descriptor only so we can have a single build of Mesa in distros.

Cool, I'll keep an eye out, and hope that it might be enough for T620 on Juno, too :)

Quoted 8 lines
Quoted 7 lines

Long story short, it appears that 'Mali LPAE' is also lacking the start level notion of VMSA, and expects a full 4-level table even for <40 bits when level 0 effectively redundant. Thus walking the 3-level table that io-pgtable comes back with ends up going wildly wrong. The hack below seems to do the job for me; if Clément can confirm (on T-720 you'll still need the userspace hack to force 32-bit jobs as well) then I think I'll cook up a proper refactoring of the allocator to put things right.

Quoted 5 lines

Mmaps seem to work with this patch, thanks.

The main complication I'm facing right now seems to be that the SFBD descriptor on T720 seems to be different from the one we already had (tested on T6xx?).

OK - with the 32-bit hack pointed to up-thread, a quick kmscube test gave me the impression that T720 works fine, but on closer inspection some parts of glmark2 do seem to go a bit wonky (although I suspect at least some of it is just down to the FPGA setup being both very slow and lacking in memory bandwidth), and the "nv12-1img" mode of kmscube turns out to render in some delightfully wrong colours.

I'll try to get a 'proper' version of the io-pgtable patch posted soon.

Thanks,
Robin.

Quoted 29 lines

Cheers,

Tomeu

Quoted 24 lines

Robin.

----->8-----

drivers/iommu/io-pgtable-arm.c +2 -0
index 546968d8a349..f29da6e8dc08 100644
@@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct
 io_pgtable_cfg *cfg, void *cookie)
         iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie);
         if (iop) {
                 u64 mair, ttbr;
+               struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops);
 
+               data->levels = 4;
                 /* Copy values as union fields overlap */
                 mair = cfg->arm_lpae_s1_cfg.mair[0];
                 ttbr = cfg->arm_lpae_s1_cfg.ttbr[0];
 

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#10 lore raw
In reply to Rob Herring

Hi Maxime, Joerg,

Quoted 73 lines · Rob Herring
Quoted 71 lines · Clément Péron

Hi,

The Allwinner H6 has a Mali-T720 MP2 which should be supported by the new panfrost driver. This series fix two issues and introduce the dt-bindings but a simple benchmark show that it's still NOT WORKING.

I'm pushing it in case someone want to continue the work.

This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1].

One patch is from Icenowy Zheng where I changed the order as required by Rob Herring[2].

Thanks,
Clement

[1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32
[2] https://patchwork.kernel.org/patch/10699829/

[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002400400
[  345.209617] Reason: TODO
[  345.209617] raw fault status: 0x800002C1
[  345.209617] decoded fault status: SLAVE FAULT
[  345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  345.209617] access type 0x2: READ
[  345.209617] source id 0x8000
[  345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0,
status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9
[  346.055876] panfrost 1800000.gpu: mmu irq status=1
[  346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA
0x0000000002C00A00
[  346.060680] Reason: TODO
[  346.060680] raw fault status: 0x810002C1
[  346.060680] decoded fault status: SLAVE FAULT
[  346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1
[  346.060680] access type 0x2: READ
[  346.060680] source id 0x8100
[  346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1,
status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85
[  346.573913] panfrost 1800000.gpu: mmu irq status=1
[  346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002C00B80

Change in v5:
- Remove fix indent

Changes in v4:
- Add bus_clock probe
- Fix sanity check in io-pgtable
- Add vramp-delay
- Merge all boards into one patch
- Remove upstreamed Neil A. patch

Change in v3 (Thanks to Maxime Ripard):
- Reauthor Icenowy for her path

Changes in v2 (Thanks to Maxime Ripard):
- Drop GPU OPP Table
- Add clocks and clock-names in required

Clément Péron (5):
drm: panfrost: add optional bus_clock
iommu: io-pgtable: fix sanity check for non 48-bit mali iommu
dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible
arm64: dts: allwinner: Add ARM Mali GPU node for H6
arm64: dts: allwinner: Add mali GPU supply for H6 boards

Quoted 7 lines
Quoted 2 lines

Icenowy Zheng (1):
dt-bindings: gpu: add bus clock for Mali Midgard GPUs

I've applied patches 1 and 3 to drm-misc. I was going to do patch 4 too, but it doesn't apply.

Patch 2 can go in via the iommu tree and the rest via the allwinner tree.

Is this OK for you to pick up this series?

Thanks,
Clément

Quoted 2 lines

Rob

#19 lore raw
In reply to Robin Murphy
Quoted 17 lines · Robin Murphy
Quoted 7 lines · Tomeu Vizoso
Quoted 1 line · Clément Péron

[snip]

Quoted 3 lines
[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA

0x0000000002400400

Quoted 7 lines

From what I can see here, 0x0000000002400400 points to the first byte
of the first submitted job descriptor.

So mapping buffers for the GPU doesn't seem to be working at all on 64-bit T-760.

Steven, Robin, do you have any idea of why this could be?

Quoted 8 lines

I tried rolling back to the old panfrost/nondrm shim, and it works fine with kbase, and I also found that T-820 falls over in the exact same manner, so the fact that it seemed to be common to the smaller 33-bit designs rather than anything to do with the other job_descriptor_size/v4/v5 complication turned out to be telling.

[ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or is it just that nobody's yet observed a 64-bit blob driving one? ]

Do you know if 64-bit descriptors work on v4 GPUs with our kernel driver but with the DDK?

Wonder if there something else to be fixed in the kernel for that scenario.

Thanks,

Tomeu

Quoted 32 lines

Long story short, it appears that 'Mali LPAE' is also lacking the start level notion of VMSA, and expects a full 4-level table even for <40 bits when level 0 effectively redundant. Thus walking the 3-level table that io-pgtable comes back with ends up going wildly wrong. The hack below seems to do the job for me; if Clément can confirm (on T-720 you'll still need the userspace hack to force 32-bit jobs as well) then I think I'll cook up a proper refactoring of the allocator to put things right.

Robin.

----->8-----

drivers/iommu/io-pgtable-arm.c +2 -0
index 546968d8a349..f29da6e8dc08 100644
@@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct
 io_pgtable_cfg *cfg, void *cookie)
        iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie);
        if (iop) {
                u64 mair, ttbr;
+               struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops);
 
+               data->levels = 4;
                /* Copy values as union fields overlap */
                mair = cfg->arm_lpae_s1_cfg.mair[0];
                ttbr = cfg->arm_lpae_s1_cfg.ttbr[0];
 

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#16 lore raw
In reply to Robin Murphy

Hi Robyn,

Quoted 50 lines · Robin Murphy
Quoted 18 lines · Tomeu Vizoso
Quoted 17 lines · Robin Murphy
Quoted 14 lines · Tomeu Vizoso
Quoted 1 line · Clément Péron

[snip]

Quoted 3 lines

[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA 0x0000000002400400

  From what I can see here, 0x0000000002400400 points to the first byte of the first submitted job descriptor.

So mapping buffers for the GPU doesn't seem to be working at all on 64-bit T-760.

Steven, Robin, do you have any idea of why this could be?

Quoted 8 lines
Quoted 5 lines

I tried rolling back to the old panfrost/nondrm shim, and it works fine with kbase, and I also found that T-820 falls over in the exact same manner, so the fact that it seemed to be common to the smaller 33-bit designs rather than anything to do with the other job_descriptor_size/v4/v5 complication turned out to be telling.

Is this complication something you can explain? I don't know what v4 and v5 are meant here.

I was alluding to BASE_HW_FEATURE_V4, which I believe refers to the Midgard architecture version - the older versions implemented by T6xx and T720 seem to be collectively treated as "v4", while T760 and T8xx would effectively be "v5".

Quoted 7 lines
Quoted 2 lines

[ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or is it just that nobody's yet observed a 64-bit blob driving one? ]

I'm looking right now at getting Panfrost working on T720 with 64-bit descriptors, with the ultimate goal of making Panfrost 64-bit-descriptor only so we can have a single build of Mesa in distros.

Cool, I'll keep an eye out, and hope that it might be enough for T620 on Juno, too :)

Quoted 10 lines
Quoted 7 lines

Long story short, it appears that 'Mali LPAE' is also lacking the start level notion of VMSA, and expects a full 4-level table even for <40 bits when level 0 effectively redundant. Thus walking the 3-level table that io-pgtable comes back with ends up going wildly wrong. The hack below seems to do the job for me; if Clément can confirm (on T-720 you'll still need the userspace hack to force 32-bit jobs as well) then I think I'll cook up a proper refactoring of the allocator to put things right.

Mmaps seem to work with this patch, thanks.

Quoted 7 lines
Quoted 3 lines

The main complication I'm facing right now seems to be that the SFBD descriptor on T720 seems to be different from the one we already had (tested on T6xx?).

OK - with the 32-bit hack pointed to up-thread, a quick kmscube test gave me the impression that T720 works fine, but on closer inspection some parts of glmark2 do seem to go a bit wonky (although I suspect at least some of it is just down to the FPGA setup being both very slow and lacking in memory bandwidth), and the "nv12-1img" mode of kmscube turns out to render in some delightfully wrong colours.

I'll try to get a 'proper' version of the io-pgtable patch posted soon.

I'm trying to collect all the fixes needed to make T820 work again, and I was wondering if you finally have a proper patch for this and "cfg->ias > 48" hack ? Or one I can test ?

Thanks,
Neil

Quoted 29 lines

Thanks,
Robin.

Quoted 25 lines

Cheers,

Tomeu

Quoted 20 lines

Robin.

----->8-----

drivers/iommu/io-pgtable-arm.c +2 -0
index 546968d8a349..f29da6e8dc08 100644
@@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct
 io_pgtable_cfg *cfg, void *cookie)
          iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie);
          if (iop) {
                  u64 mair, ttbr;
+               struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops);
 
+               data->levels = 4;
                  /* Copy values as union fields overlap */
                  mair = cfg->arm_lpae_s1_cfg.mair[0];
                  ttbr = cfg->arm_lpae_s1_cfg.ttbr[0];
 
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

reviewer
#17 lore raw

Hi Neil,

Quoted 57 lines · Neil Armstrong

Hi Robin,

Quoted 54 lines · Robin Murphy
Quoted 18 lines · Tomeu Vizoso
Quoted 17 lines · Robin Murphy
Quoted 14 lines · Tomeu Vizoso
Quoted 1 line · Clément Péron

[snip]

Quoted 3 lines

[  345.204813] panfrost 1800000.gpu: mmu irq status=1
[  345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA 0x0000000002400400

  From what I can see here, 0x0000000002400400 points to the first byte
of the first submitted job descriptor.

So mapping buffers for the GPU doesn't seem to be working at all on 64-bit T-760.

Steven, Robin, do you have any idea of why this could be?

Quoted 8 lines
Quoted 5 lines

I tried rolling back to the old panfrost/nondrm shim, and it works fine with kbase, and I also found that T-820 falls over in the exact same manner, so the fact that it seemed to be common to the smaller 33-bit designs rather than anything to do with the other job_descriptor_size/v4/v5 complication turned out to be telling.

Is this complication something you can explain? I don't know what v4 and v5 are meant here.

I was alluding to BASE_HW_FEATURE_V4, which I believe refers to the Midgard architecture version - the older versions implemented by T6xx and T720 seem to be collectively treated as "v4", while T760 and T8xx would effectively be "v5".

Quoted 7 lines
Quoted 2 lines

[ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or is it just that nobody's yet observed a 64-bit blob driving one? ]

I'm looking right now at getting Panfrost working on T720 with 64-bit descriptors, with the ultimate goal of making Panfrost 64-bit-descriptor only so we can have a single build of Mesa in distros.

Cool, I'll keep an eye out, and hope that it might be enough for T620 on Juno, too :)

Quoted 13 lines
Quoted 7 lines

Long story short, it appears that 'Mali LPAE' is also lacking the start level notion of VMSA, and expects a full 4-level table even for <40 bits when level 0 effectively redundant. Thus walking the 3-level table that io-pgtable comes back with ends up going wildly wrong. The hack below seems to do the job for me; if Clément can confirm (on T-720 you'll still need the userspace hack to force 32-bit jobs as well) then I think I'll cook up a proper refactoring of the allocator to put things right.

Mmaps seem to work with this patch, thanks.

The main complication I'm facing right now seems to be that the SFBD descriptor on T720 seems to be different from the one we already had (tested on T6xx?).

Quoted 7 lines
Quoted 3 lines

OK - with the 32-bit hack pointed to up-thread, a quick kmscube test gave me the impression that T720 works fine, but on closer inspection some parts of glmark2 do seem to go a bit wonky (although I suspect at least some of it is just down to the FPGA setup being both very slow and lacking in memory bandwidth), and the "nv12-1img" mode of kmscube turns out to render in some delightfully wrong colours.

I'll try to get a 'proper' version of the io-pgtable patch posted soon.

I'm trying to collect all the fixes needed to make T820 work again, and I was wondering if you finally have a proper patch for this and "cfg->ias > 48" hack ? Or one I can test ?

I do have a handful of io-pgtable patches written up and ready to go, I'm just treading carefully and waiting for the internal approval box to be ticked before I share anything :(

Robin.

Quoted 38 lines

Thanks,
Neil

Quoted 33 lines

Thanks,
Robin.

Quoted 29 lines

Cheers,

Tomeu

Quoted 24 lines

Robin.

----->8-----

drivers/iommu/io-pgtable-arm.c +2 -0
index 546968d8a349..f29da6e8dc08 100644
@@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct
 io_pgtable_cfg *cfg, void *cookie)
          iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie);
          if (iop) {
                  u64 mair, ttbr;
+               struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops);
 
+               data->levels = 4;
                  /* Copy values as union fields overlap */
                  mair = cfg->arm_lpae_s1_cfg.mair[0];
                  ttbr = cfg->arm_lpae_s1_cfg.ttbr[0];
 

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#18 lore raw
Quoted 5 lines · Robin Murphy

Hi Neil,

Quoted 2 lines · Neil Armstrong

Hi Robin,

[...]

Quoted 10 lines
Quoted 8 lines
Quoted 4 lines

OK - with the 32-bit hack pointed to up-thread, a quick kmscube test gave me the impression that T720 works fine, but on closer inspection some parts of glmark2 do seem to go a bit wonky (although I suspect at least some of it is just down to the FPGA setup being both very slow and lacking in memory bandwidth), and the "nv12-1img" mode of kmscube turns out to render in some delightfully wrong colours.

I'll try to get a 'proper' version of the io-pgtable patch posted soon.

I'm trying to collect all the fixes needed to make T820 work again, and I was wondering if you finally have a proper patch for this and "cfg->ias > 48" hack ? Or one I can test ?

I do have a handful of io-pgtable patches written up and ready to go, I'm just treading carefully and waiting for the internal approval box to be ticked before I share anything :(

Great !

No problem, it can totally wait until approval,

Thanks,
Neil

Quoted 41 lines

Robin.

Quoted 38 lines

Thanks,
Neil

Quoted 33 lines

Thanks,
Robin.

Quoted 29 lines

Cheers,

Tomeu

Quoted 24 lines

Robin.

----->8-----

drivers/iommu/io-pgtable-arm.c +2 -0
index 546968d8a349..f29da6e8dc08 100644
@@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct
 io_pgtable_cfg *cfg, void *cookie)
           iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie);
           if (iop) {
                   u64 mair, ttbr;
+               struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops);
 
+               data->levels = 4;
                   /* Copy values as union fields overlap */
                   mair = cfg->arm_lpae_s1_cfg.mair[0];
                   ttbr = cfg->arm_lpae_s1_cfg.ttbr[0];
 

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel