Skip to contentSkip to conversation

drm/nouveau: Various fixes for GP10B

PATCH Superseded Thierry Reding 5 participants 14 messages Nov 2, 2019 – Nov 29, 2019 2 messages not in the archive

Subsystems (5)
DRM drivers DRM misc Tegra architecture support IOMMU +1 more subsystems

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Thierry Reding

author
#0 lore raw

v2 cover drm/nouveau: Various fixes for GP10B

From: Thierry Reding <[email protected]>

Hi Ben,

here's a revised subset of the patches I had sent out a couple of weeks ago. I've reworked the BAR2 accesses in the way that you had suggested, which at least for GP10B turned out to be fairly trivial to do. I have not looked in detail at this for GV11B yet, but a cursory look showed that BAR2 is accessed in more places, so the equivalent for GV11B might be a bit more involved.

Other than that, not a lot has changed since then. I've added a couple of precursory patches to add IOMMU helper dummies for the case where IOMMU is disabled (as suggested by Ben Dooks).

Joerg, it'd be great if you could give an Acked-by on those two patches so that Ben can pick them both up into the Nouveau tree. Alternatively I can put them both into a stable branch and send a pull request to both of you. Or yet another alternative would be for Joerg to apply them now and Ben to wait for v5.5-rc1 until he picks up the rest. All of those work for me.

Thierry

Thierry Reding (9):
iommu: Document iommu_fwspec::flags field
iommu: Add dummy dev_iommu_fwspec_get() helper drm/nouveau: fault: Add support for GP10B drm/nouveau: tegra: Do not try to disable PCI device drm/nouveau: tegra: Avoid pulsing reset twice drm/nouveau: tegra: Set clock rate if not set drm/nouveau: secboot: Read WPR configuration from GPU registers drm/nouveau: gp10b: Add custom L2 cache implementation drm/nouveau: gp10b: Use correct copy engine

 .../drm/nouveau/include/nvkm/subdev/fault.h   |  1 +
 .../gpu/drm/nouveau/include/nvkm/subdev/ltc.h |  1 +
 drivers/gpu/drm/nouveau/nouveau_drm.c         |  3 +-
 .../gpu/drm/nouveau/nvkm/engine/device/base.c |  6 +-
 .../drm/nouveau/nvkm/engine/device/tegra.c    | 24 ++++--
 .../gpu/drm/nouveau/nvkm/subdev/fault/Kbuild  |  1 +
 .../gpu/drm/nouveau/nvkm/subdev/fault/base.c  |  2 +-
 .../gpu/drm/nouveau/nvkm/subdev/fault/gp100.c | 17 ++--
 .../gpu/drm/nouveau/nvkm/subdev/fault/gp10b.c | 53 ++++++++++++
 .../gpu/drm/nouveau/nvkm/subdev/fault/gv100.c |  1 +
 .../gpu/drm/nouveau/nvkm/subdev/fault/priv.h  | 10 +++
 .../gpu/drm/nouveau/nvkm/subdev/ltc/Kbuild    |  1 +
 .../gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c   | 65 +++++++++++++++
 .../gpu/drm/nouveau/nvkm/subdev/ltc/priv.h    |  2 +
 .../drm/nouveau/nvkm/subdev/secboot/gm200.h   |  2 +-
 .../drm/nouveau/nvkm/subdev/secboot/gm20b.c   | 81 ++++++++++++-------
 .../drm/nouveau/nvkm/subdev/secboot/gp10b.c   |  4 +-
 include/linux/iommu.h                         | 46 ++++++-----
 18 files changed, 249 insertions(+), 71 deletions(-)
 create mode 100644 drivers/gpu/drm/nouveau/nvkm/subdev/fault/gp10b.c
 create mode 100644 drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c
Signature · 6 lines
2.23.0

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

PATCH v2 1/9 iommu: Document iommu_fwspec::flags field

From: Thierry Reding <[email protected]>

When this field was added in commit 5702ee24182f ("ACPI/IORT: Check ATS capability in root complex nodes"), the kerneldoc comment wasn't updated at the same time.

Signed-off-by Thierry Reding <[email protected]>
---
 include/linux/iommu.h | 1 +
 1 file changed, 1 insertion(+)
include/linux/iommu.h +1 -0
index e28e80dea141..7bf038b371b8 100644
@@ -570,6 +570,7 @@ struct iommu_group *fsl_mc_device_group(struct device *dev);
  * @ops: ops for this device's IOMMU
  * @iommu_fwnode: firmware handle for this device's IOMMU
  * @iommu_priv: IOMMU driver private data for this device
+ * @flags: IOMMU flags associated with this device
  * @num_ids: number of associated device IDs
  * @ids: IDs which this device may present to the IOMMU
  */
Signature · 6 lines
2.23.0

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

A message that is not in the archive (sent off-list or not yet received): 20191102175637.3065-1-thierry.reding-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

#3 lore raw

PATCH v2 2/9 iommu: Add dummy dev_iommu_fwspec_get() helper

From: Thierry Reding <[email protected]>

This dummy implementation is useful to avoid a dependency on the IOMMU_API Kconfig symbol in drivers that can optionally use the IOMMU API.

In order to fully use this, also move the struct iommu_fwspec definition out of the IOMMU_API protected region.

Suggested-by Ben Dooks <[email protected]>
Signed-off-by Thierry Reding <[email protected]>
---
 include/linux/iommu.h | 47 ++++++++++++++++++++++++-------------------
 1 file changed, 26 insertions(+), 21 deletions(-)
include/linux/iommu.h +26 -21
index 7bf038b371b8..b092e73b2c86 100644
@@ -190,6 +190,27 @@ struct iommu_sva_ops {
 	iommu_mm_exit_handler_t mm_exit;
 };
 
+/**
+ * struct iommu_fwspec - per-device IOMMU instance data
+ * @ops: ops for this device's IOMMU
+ * @iommu_fwnode: firmware handle for this device's IOMMU
+ * @iommu_priv: IOMMU driver private data for this device
+ * @flags: IOMMU flags associated with this device
+ * @num_ids: number of associated device IDs
+ * @ids: IDs which this device may present to the IOMMU
+ */
+struct iommu_fwspec {
+	const struct iommu_ops	*ops;
+	struct fwnode_handle	*iommu_fwnode;
+	void			*iommu_priv;
+	u32			flags;
+	unsigned int		num_ids;
+	u32			ids[1];
+};
+
+/* ATS is supported */
+#define IOMMU_FWSPEC_PCI_RC_ATS			(1 << 0)
+
 #ifdef CONFIG_IOMMU_API
 
 /**

@@ -565,27 +586,6 @@ extern struct iommu_group *generic_device_group(struct device *dev);
 /* FSL-MC device grouping function */
 struct iommu_group *fsl_mc_device_group(struct device *dev);
 
-/**
- * struct iommu_fwspec - per-device IOMMU instance data
- * @ops: ops for this device's IOMMU
- * @iommu_fwnode: firmware handle for this device's IOMMU
- * @iommu_priv: IOMMU driver private data for this device
- * @flags: IOMMU flags associated with this device
- * @num_ids: number of associated device IDs
- * @ids: IDs which this device may present to the IOMMU
- */
-struct iommu_fwspec {
-	const struct iommu_ops	*ops;
-	struct fwnode_handle	*iommu_fwnode;
-	void			*iommu_priv;
-	u32			flags;
-	unsigned int		num_ids;
-	u32			ids[1];
-};
-
-/* ATS is supported */
-#define IOMMU_FWSPEC_PCI_RC_ATS			(1 << 0)
-
 /**
  * struct iommu_sva - handle to a device-mm bond
  */

@@ -980,6 +980,11 @@ const struct iommu_ops *iommu_ops_from_fwnode(struct fwnode_handle *fwnode)
 	return NULL;
 }
 
+static inline struct iommu_fwspec *dev_iommu_fwspec_get(struct device *dev)
+{
+	return NULL;
+}
+
 static inline bool
 iommu_dev_has_feature(struct device *dev, enum iommu_dev_features feat)
 {
Signature · 6 lines
2.23.0

_______________________________________________
Nouveau mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/nouveau

A message that is not in the archive (sent off-list or not yet received): 20191102175637.3065-3-thierry.reding-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

build bot
#5 lore raw

Re: [PATCH v2 2/9] iommu: Add dummy dev_iommu_fwspec_get() helper

I love your patch! Yet something to improve:

Hi Thierry,

I love your patch! Yet something to improve:

[auto build test ERROR on linus/master]
[also build test ERROR on v5.4-rc5 next-20191031]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system. BTW, we also suggest to use '--base' option to specify the base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url:    https://github.com/0day-ci/linux/commits/Thierry-Reding/drm-nouveau-Various-fixes-for-GP10B/20191103-125101
base:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git 56cfd2507d3e720f4b1dbf9513e00680516a0826
config: i386-tinyconfig (attached as .config)
compiler: gcc-7 (Debian 7.4.0-14) 7.4.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386

If you fix the issue, kindly add following tag

Reported-by kbuild test robot <[email protected]>

All errors (new ones prefixed by >>):

In file included from arch/x86/kernel/pci-dma.c:4:0:

>> include/linux/iommu.h:615:8: error: redefinition of 'struct iommu_fwspec'
    struct iommu_fwspec {};
           ^~~~~~~~~~~~
   In file included from include/linux/dma-mapping.h:7:0,
                    from include/linux/dma-direct.h:5,
                    from arch/x86/kernel/pci-dma.c:2:
   include/linux/device.h:43:8: note: originally defined here
    struct iommu_fwspec;
           ^~~~~~~~~~~~

vim +615 include/linux/iommu.h

4a77a6cf6d9bf9 Joerg Roedel    2008-11-26  612  
39d4ebb9592504 Joerg Roedel    2011-09-06  613  struct iommu_ops {};
d72e31c9374627 Alex Williamson 2012-05-30  614  struct iommu_group {};
57f98d2f61e191 Robin Murphy    2016-09-13 @615  struct iommu_fwspec {};
b0119e870837dc Joerg Roedel    2017-02-01  616  struct iommu_device {};
4e32348ba5269a Jacob Pan       2019-06-03  617  struct iommu_fault_param {};
a7d20dc19d9ea7 Will Deacon     2019-07-02  618  struct iommu_iotlb_gather {};
4a77a6cf6d9bf9 Joerg Roedel    2008-11-26  619

:::::: The code at line 615 was first introduced by commit
:::::: 57f98d2f61e191ef9d06863c9ce3f8621f3671ef iommu: Introduce iommu_fwspec

:::::: TO: Robin Murphy <[email protected]> :::::: CC: Will Deacon <[email protected]>

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

_______________________________________________
Nouveau mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/nouveau

#6 lore raw

PATCH v2 3/9 drm/nouveau: fault: Add support for GP10B

In reply to a message not in the archive

From: Thierry Reding <[email protected]>

There is no BAR2 on GP10B and there is no need to map through BAR2 because all memory is shared between the GPU and the CPU. Add a custom implementation of the fault sub-device that uses nvkm_memory_addr() instead of nvkm_memory_bar2() to return the address of a pinned fault buffer.

Signed-off-by Thierry Reding <[email protected]>
---
 .../drm/nouveau/include/nvkm/subdev/fault.h   |  1 +
 .../gpu/drm/nouveau/nvkm/engine/device/base.c |  2 +-
 .../gpu/drm/nouveau/nvkm/subdev/fault/Kbuild  |  1 +
 .../gpu/drm/nouveau/nvkm/subdev/fault/base.c  |  2 +-
 .../gpu/drm/nouveau/nvkm/subdev/fault/gp100.c | 17 ++++--
 .../gpu/drm/nouveau/nvkm/subdev/fault/gp10b.c | 53 +++++++++++++++++++
 .../gpu/drm/nouveau/nvkm/subdev/fault/gv100.c |  1 +
 .../gpu/drm/nouveau/nvkm/subdev/fault/priv.h  | 10 ++++
 8 files changed, 80 insertions(+), 7 deletions(-)
 create mode 100644 drivers/gpu/drm/nouveau/nvkm/subdev/fault/gp10b.c
drivers/gpu/drm/nouveau/include/nvkm/subdev/fault.h +1 -0
index 97322f95b3ee..a513c16ab105 100644
@@ -31,6 +31,7 @@ struct nvkm_fault_data {
 };
 
 int gp100_fault_new(struct nvkm_device *, int, struct nvkm_fault **);
+int gp10b_fault_new(struct nvkm_device *, int, struct nvkm_fault **);
 int gv100_fault_new(struct nvkm_device *, int, struct nvkm_fault **);
 int tu102_fault_new(struct nvkm_device *, int, struct nvkm_fault **);
 #endif
drivers/gpu/drm/nouveau/nvkm/engine/device/base.c +1 -1
index c3c7159f3411..b061df138142 100644
@@ -2375,7 +2375,7 @@ nv13b_chipset = {
 	.name = "GP10B",
 	.bar = gm20b_bar_new,
 	.bus = gf100_bus_new,
-	.fault = gp100_fault_new,
+	.fault = gp10b_fault_new,
 	.fb = gp10b_fb_new,
 	.fuse = gm107_fuse_new,
 	.ibus = gp10b_ibus_new,
drivers/gpu/drm/nouveau/nvkm/subdev/fault/Kbuild +1 -0
index 53b9d638f2c8..d65ec719f153 100644
@@ -2,5 +2,6 @@
 nvkm-y += nvkm/subdev/fault/base.o
 nvkm-y += nvkm/subdev/fault/user.o
 nvkm-y += nvkm/subdev/fault/gp100.o
+nvkm-y += nvkm/subdev/fault/gp10b.o
 nvkm-y += nvkm/subdev/fault/gv100.o
 nvkm-y += nvkm/subdev/fault/tu102.o
drivers/gpu/drm/nouveau/nvkm/subdev/fault/base.c +1 -1
index ca251560d3e0..1c4b852b26c3 100644
@@ -108,7 +108,7 @@ nvkm_fault_oneinit_buffer(struct nvkm_fault *fault, int id)
 		return ret;
 
 	/* Pin fault buffer in BAR2. */
-	buffer->addr = nvkm_memory_bar2(buffer->mem);
+	buffer->addr = fault->func->buffer.pin(buffer);
 	if (buffer->addr == ~0ULL)
 		return -EFAULT;
drivers/gpu/drm/nouveau/nvkm/subdev/fault/gp100.c +12 -5
index 4f3c4e091117..f6b189cc4330 100644
@@ -21,25 +21,26 @@
  */
 #include "priv.h"
 
+#include <core/memory.h>
 #include <subdev/mc.h>
 
 #include <nvif/class.h>
 
-static void
+void
 gp100_fault_buffer_intr(struct nvkm_fault_buffer *buffer, bool enable)
 {
 	struct nvkm_device *device = buffer->fault->subdev.device;
 	nvkm_mc_intr_mask(device, NVKM_SUBDEV_FAULT, enable);
 }
 
-static void
+void
 gp100_fault_buffer_fini(struct nvkm_fault_buffer *buffer)
 {
 	struct nvkm_device *device = buffer->fault->subdev.device;
 	nvkm_mask(device, 0x002a70, 0x00000001, 0x00000000);
 }
 
-static void
+void
 gp100_fault_buffer_init(struct nvkm_fault_buffer *buffer)
 {
 	struct nvkm_device *device = buffer->fault->subdev.device;

@@ -48,7 +49,12 @@ gp100_fault_buffer_init(struct nvkm_fault_buffer *buffer)
 	nvkm_mask(device, 0x002a70, 0x00000001, 0x00000001);
 }
 
-static void
+u64 gp100_fault_buffer_pin(struct nvkm_fault_buffer *buffer)
+{
+	return nvkm_memory_bar2(buffer->mem);
+}
+
+void
 gp100_fault_buffer_info(struct nvkm_fault_buffer *buffer)
 {
 	buffer->entries = nvkm_rd32(buffer->fault->subdev.device, 0x002a78);

@@ -56,7 +62,7 @@ gp100_fault_buffer_info(struct nvkm_fault_buffer *buffer)
 	buffer->put = 0x002a80;
 }
 
-static void
+void
 gp100_fault_intr(struct nvkm_fault *fault)
 {
 	nvkm_event_send(&fault->event, 1, 0, NULL, 0);

@@ -68,6 +74,7 @@ gp100_fault = {
 	.buffer.nr = 1,
 	.buffer.entry_size = 32,
 	.buffer.info = gp100_fault_buffer_info,
+	.buffer.pin = gp100_fault_buffer_pin,
 	.buffer.init = gp100_fault_buffer_init,
 	.buffer.fini = gp100_fault_buffer_fini,
 	.buffer.intr = gp100_fault_buffer_intr,
drivers/gpu/drm/nouveau/nvkm/subdev/fault/gp10b.c new file +53 -0

53 lines not on this page · raw message

drivers/gpu/drm/nouveau/nvkm/subdev/fault/gv100.c +1 -0
index 6747f09c2dc3..2707be4ffabc 100644
@@ -214,6 +214,7 @@ gv100_fault = {
 	.buffer.nr = 2,
 	.buffer.entry_size = 32,
 	.buffer.info = gv100_fault_buffer_info,
+	.buffer.pin = gp100_fault_buffer_pin,
 	.buffer.init = gv100_fault_buffer_init,
 	.buffer.fini = gv100_fault_buffer_fini,
 	.buffer.intr = gv100_fault_buffer_intr,
drivers/gpu/drm/nouveau/nvkm/subdev/fault/priv.h +10 -0
index 975e66ac6344..f6f1dd7eee1f 100644
@@ -30,6 +30,7 @@ struct nvkm_fault_func {
 		int nr;
 		u32 entry_size;
 		void (*info)(struct nvkm_fault_buffer *);
+		u64 (*pin)(struct nvkm_fault_buffer *);
 		void (*init)(struct nvkm_fault_buffer *);
 		void (*fini)(struct nvkm_fault_buffer *);
 		void (*intr)(struct nvkm_fault_buffer *, bool enable);

@@ -40,6 +41,15 @@ struct nvkm_fault_func {
 	} user;
 };
 
+void gp100_fault_buffer_intr(struct nvkm_fault_buffer *, bool enable);
+void gp100_fault_buffer_fini(struct nvkm_fault_buffer *);
+void gp100_fault_buffer_init(struct nvkm_fault_buffer *);
+u64 gp100_fault_buffer_pin(struct nvkm_fault_buffer *);
+void gp100_fault_buffer_info(struct nvkm_fault_buffer *);
+void gp100_fault_intr(struct nvkm_fault *);
+
+u64 gp10b_fault_buffer_pin(struct nvkm_fault_buffer *);
+
 int gv100_fault_oneinit(struct nvkm_fault *);
 
 int nvkm_ufault_new(struct nvkm_device *, const struct nvkm_oclass *,
Signature · 6 lines
2.23.0

_______________________________________________
Nouveau mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/nouveau
#7 lore raw

PATCH v2 5/9 drm/nouveau: tegra: Avoid pulsing reset twice

In reply to a message not in the archive

From: Thierry Reding <[email protected]>

When the GPU powergate is controlled by a generic power domain provider, the reset will automatically be asserted and deasserted as part of the power-ungating procedure.

On some Jetson TX2 boards, doing an additional assert and deassert of the GPU outside of the power-ungate procedure can cause the GPU to go into a bad state where the memory interface can no longer access system memory.

Signed-off-by Thierry Reding <[email protected]>
---
 drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c +6 -6
index 0e372a190d3f..747a775121cf 100644
@@ -52,18 +52,18 @@ nvkm_device_tegra_power_up(struct nvkm_device_tegra *tdev)
 	clk_set_rate(tdev->clk_pwr, 204000000);
 	udelay(10);
 
-	reset_control_assert(tdev->rst);
-	udelay(10);
-
 	if (!tdev->pdev->dev.pm_domain) {
+		reset_control_assert(tdev->rst);
+		udelay(10);
+
 		ret = tegra_powergate_remove_clamping(TEGRA_POWERGATE_3D);
 		if (ret)
 			goto err_clamp;
 		udelay(10);
-	}
 
-	reset_control_deassert(tdev->rst);
-	udelay(10);
+		reset_control_deassert(tdev->rst);
+		udelay(10);
+	}
 
 	return 0;
Signature · 6 lines
2.23.0

_______________________________________________
Nouveau mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/nouveau
#8 lore raw

PATCH v2 6/9 drm/nouveau: tegra: Set clock rate if not set

In reply to a message not in the archive

From: Thierry Reding <[email protected]>

If the GPU clock has not had a rate set, initialize it to the maximum clock rate to make sure it does run.

Signed-off-by Thierry Reding <[email protected]>
---
 drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)
drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c +12 -0
index 747a775121cf..d0d52c1d4aee 100644
@@ -279,6 +279,7 @@ nvkm_device_tegra_new(const struct nvkm_device_tegra_func *func,
 		      struct nvkm_device **pdevice)
 {
 	struct nvkm_device_tegra *tdev;
+	unsigned long rate;
 	int ret;
 
 	if (!(tdev = kzalloc(sizeof(*tdev), GFP_KERNEL)))

@@ -307,6 +308,17 @@ nvkm_device_tegra_new(const struct nvkm_device_tegra_func *func,
 		goto free;
 	}
 
+	rate = clk_get_rate(tdev->clk);
+	if (rate == 0) {
+		ret = clk_set_rate(tdev->clk, ULONG_MAX);
+		if (ret < 0)
+			goto free;
+
+		rate = clk_get_rate(tdev->clk);
+
+		dev_dbg(&pdev->dev, "GPU clock set to %lu\n", rate);
+	}
+
 	if (func->require_ref_clk)
 		tdev->clk_ref = devm_clk_get(&pdev->dev, "ref");
 	if (IS_ERR(tdev->clk_ref)) {
Signature · 6 lines
2.23.0

_______________________________________________
Nouveau mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/nouveau
#9 lore raw

PATCH v2 9/9 drm/nouveau: gp10b: Use correct copy engine

In reply to a message not in the archive

From: Thierry Reding <[email protected]>

gp10b uses the new engine enumeration mechanism introduced in the Pascal architecture. As a result, the copy engine, which used to be at index 2 for prior Tegra GPU instantiations, has now moved to index 0. Fix up the index and also use the gp100 variant of the copy engine class because on gp10b the PASCAL_DMA_COPY_B class is not supported.

Signed-off-by Thierry Reding <[email protected]>
---
 drivers/gpu/drm/nouveau/nvkm/engine/device/base.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nvkm/engine/device/base.c +1 -1
index 231ec0073af3..eba450e689b2 100644
@@ -2387,7 +2387,7 @@ nv13b_chipset = {
 	.pmu = gm20b_pmu_new,
 	.timer = gk20a_timer_new,
 	.top = gk104_top_new,
-	.ce[2] = gp102_ce_new,
+	.ce[0] = gp100_ce_new,
 	.dma = gf119_dma_new,
 	.fifo = gp10b_fifo_new,
 	.gr = gp10b_gr_new,
Signature · 6 lines
2.23.0

_______________________________________________
Nouveau mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/nouveau
#10 lore raw
In reply to a message not in the archive
Quoted 15 lines · Thierry Reding

From: Thierry Reding <[email protected]>

Hi Ben,

here's a revised subset of the patches I had sent out a couple of weeks ago. I've reworked the BAR2 accesses in the way that you had suggested, which at least for GP10B turned out to be fairly trivial to do. I have not looked in detail at this for GV11B yet, but a cursory look showed that BAR2 is accessed in more places, so the equivalent for GV11B might be a bit more involved.

Other than that, not a lot has changed since then. I've added a couple of precursory patches to add IOMMU helper dummies for the case where IOMMU is disabled (as suggested by Ben Dooks).

Quoted 6 lines

Joerg, it'd be great if you could give an Acked-by on those two patches so that Ben can pick them both up into the Nouveau tree. Alternatively I can put them both into a stable branch and send a pull request to both of you. Or yet another alternative would be for Joerg to apply them now and Ben to wait for v5.5-rc1 until he picks up the rest. All of those work for me.

Hi Joerg, Ben,

do you guys have any further comments on this series? I've got an updated patch to silence the warning that the kbuild robot flagged, so if there are no other comments I can send a final v3 of the series.

Thierry

_______________________________________________
Nouveau mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/nouveau

#11 lore raw
Quoted 24 lines · Thierry Reding
Quoted 15 lines · Thierry Reding

From: Thierry Reding <[email protected]>

Hi Ben,

here's a revised subset of the patches I had sent out a couple of weeks ago. I've reworked the BAR2 accesses in the way that you had suggested, which at least for GP10B turned out to be fairly trivial to do. I have not looked in detail at this for GV11B yet, but a cursory look showed that BAR2 is accessed in more places, so the equivalent for GV11B might be a bit more involved.

Other than that, not a lot has changed since then. I've added a couple of precursory patches to add IOMMU helper dummies for the case where IOMMU is disabled (as suggested by Ben Dooks).

Quoted 6 lines

Joerg, it'd be great if you could give an Acked-by on those two patches so that Ben can pick them both up into the Nouveau tree. Alternatively I can put them both into a stable branch and send a pull request to both of you. Or yet another alternative would be for Joerg to apply them now and Ben to wait for v5.5-rc1 until he picks up the rest. All of those work for me.

Quoted 5 lines

Hi Joerg, Ben,

do you guys have any further comments on this series? I've got an updated patch to silence the warning that the kbuild robot flagged, so if there are no other comments I can send a final v3 of the series.

I'm on leave at the moment. But the nouveau fixes look fine to me and I'm happy to pick them up once you send a v3, and allow Jeorg to apply the rest through his tree.

Thanks,
Ben.

Quoted 2 lines

Thierry

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

author
#13 lore raw

PATCH v2 4/9 drm/nouveau: tegra: Do not try to disable PCI device

In reply to Thierry Reding

From: Thierry Reding <[email protected]>

When Nouveau is instantiated on top of a platform device, the dev->pdev field will be NULL and calling pci_disable_device() will crash. Move the PCI disabling code to the PCI specific driver removal code.

Reviewed-by Lyude Paul <[email protected]>
Signed-off-by Thierry Reding <[email protected]>
---
 drivers/gpu/drm/nouveau/nouveau_drm.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/nouveau/nouveau_drm.c +1 -2
index 2cd83849600f..b65ae817eabf 100644
@@ -715,7 +715,6 @@ static int nouveau_drm_probe(struct pci_dev *pdev,
 void
 nouveau_drm_device_remove(struct drm_device *dev)
 {
-	struct pci_dev *pdev = dev->pdev;
 	struct nouveau_drm *drm = nouveau_drm(dev);
 	struct nvkm_client *client;
 	struct nvkm_device *device;

@@ -727,7 +726,6 @@ nouveau_drm_device_remove(struct drm_device *dev)
 	device = nvkm_device_find(client->device);
 
 	nouveau_drm_device_fini(dev);
-	pci_disable_device(pdev);
 	drm_dev_put(dev);
 	nvkm_device_del(&device);
 }

@@ -738,6 +736,7 @@ nouveau_drm_remove(struct pci_dev *pdev)
 	struct drm_device *dev = pci_get_drvdata(pdev);
 
 	nouveau_drm_device_remove(dev);
+	pci_disable_device(pdev);
 }
 
 static int
Signature · 6 lines
2.23.0

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

PATCH v2 7/9 drm/nouveau: secboot: Read WPR configuration from GPU registers

In reply to Thierry Reding

From: Thierry Reding <[email protected]>

The GPUs found on Tegra SoCs have registers that can be used to read the WPR configuration. Use these registers instead of reaching into the memory controller's register space to read the same information.

Signed-off-by Thierry Reding <[email protected]>
---
 .../drm/nouveau/nvkm/subdev/secboot/gm200.h   |  2 +-
 .../drm/nouveau/nvkm/subdev/secboot/gm20b.c   | 81 ++++++++++++-------
 .../drm/nouveau/nvkm/subdev/secboot/gp10b.c   |  4 +-
 3 files changed, 53 insertions(+), 34 deletions(-)
drivers/gpu/drm/nouveau/nvkm/subdev/secboot/gm200.h +1 -1
index 62c5e162099a..280b1448df88 100644
@@ -41,6 +41,6 @@ int gm200_secboot_run_blob(struct nvkm_secboot *, struct nvkm_gpuobj *,
 			   struct nvkm_falcon *);
 
 /* Tegra-only */
-int gm20b_secboot_tegra_read_wpr(struct gm200_secboot *, u32);
+int gm20b_secboot_tegra_read_wpr(struct gm200_secboot *);
 
 #endif
drivers/gpu/drm/nouveau/nvkm/subdev/secboot/gm20b.c +51 -30

108 lines not on this page · raw message

drivers/gpu/drm/nouveau/nvkm/subdev/secboot/gp10b.c +1 -3
index 28ca29d0eeee..d84e85825995 100644
@@ -23,15 +23,13 @@
 #include "acr.h"
 #include "gm200.h"
 
-#define TEGRA186_MC_BASE			0x02c10000
-
 static int
 gp10b_secboot_oneinit(struct nvkm_secboot *sb)
 {
 	struct gm200_secboot *gsb = gm200_secboot(sb);
 	int ret;
 
-	ret = gm20b_secboot_tegra_read_wpr(gsb, TEGRA186_MC_BASE);
+	ret = gm20b_secboot_tegra_read_wpr(gsb);
 	if (ret)
 		return ret;
Signature · 6 lines
2.23.0

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author
#15 lore raw

PATCH v2 8/9 drm/nouveau: gp10b: Add custom L2 cache implementation

In reply to Thierry Reding

From: Thierry Reding <[email protected]>

There are extra registers that need to be programmed to make the level 2 cache work on GP10B, such as the stream ID register that is used when an SMMU is used to translate memory addresses.

Signed-off-by Thierry Reding <[email protected]>

---
Changes in v2:
- remove IOMMU_API protection to increase compile coverage
- relies on dummy dev_iommu_fwspec_get() helper

 .../gpu/drm/nouveau/include/nvkm/subdev/ltc.h |  1 +
 .../gpu/drm/nouveau/nvkm/engine/device/base.c |  2 +-
 .../gpu/drm/nouveau/nvkm/subdev/ltc/Kbuild    |  1 +
 .../gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c   | 65 +++++++++++++++++++
 .../gpu/drm/nouveau/nvkm/subdev/ltc/priv.h    |  2 +
 5 files changed, 70 insertions(+), 1 deletion(-)
 create mode 100644 drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c
drivers/gpu/drm/nouveau/include/nvkm/subdev/ltc.h +1 -0
index 644d527c3b96..d76f60d7d29a 100644
@@ -40,4 +40,5 @@ int gm107_ltc_new(struct nvkm_device *, int, struct nvkm_ltc **);
 int gm200_ltc_new(struct nvkm_device *, int, struct nvkm_ltc **);
 int gp100_ltc_new(struct nvkm_device *, int, struct nvkm_ltc **);
 int gp102_ltc_new(struct nvkm_device *, int, struct nvkm_ltc **);
+int gp10b_ltc_new(struct nvkm_device *, int, struct nvkm_ltc **);
 #endif
drivers/gpu/drm/nouveau/nvkm/engine/device/base.c +1 -1
index b061df138142..231ec0073af3 100644
@@ -2380,7 +2380,7 @@ nv13b_chipset = {
 	.fuse = gm107_fuse_new,
 	.ibus = gp10b_ibus_new,
 	.imem = gk20a_instmem_new,
-	.ltc = gp102_ltc_new,
+	.ltc = gp10b_ltc_new,
 	.mc = gp10b_mc_new,
 	.mmu = gp10b_mmu_new,
 	.secboot = gp10b_secboot_new,
drivers/gpu/drm/nouveau/nvkm/subdev/ltc/Kbuild +1 -0
index 2b6d36ea7067..728d75010847 100644
@@ -6,3 +6,4 @@ nvkm-y += nvkm/subdev/ltc/gm107.o
 nvkm-y += nvkm/subdev/ltc/gm200.o
 nvkm-y += nvkm/subdev/ltc/gp100.o
 nvkm-y += nvkm/subdev/ltc/gp102.o
+nvkm-y += nvkm/subdev/ltc/gp10b.o
drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c new file +65 -0

65 lines not on this page · raw message

drivers/gpu/drm/nouveau/nvkm/subdev/ltc/priv.h +2 -0
index 2fcf18e46ce3..eca5a711b1b8 100644
@@ -46,4 +46,6 @@ void gm107_ltc_zbc_clear_depth(struct nvkm_ltc *, int, const u32);
 int gp100_ltc_oneinit(struct nvkm_ltc *);
 void gp100_ltc_init(struct nvkm_ltc *);
 void gp100_ltc_intr(struct nvkm_ltc *);
+
+void gp102_ltc_zbc_clear_stencil(struct nvkm_ltc *, int, const u32);
 #endif
Signature · 6 lines
2.23.0

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