Skip to contentSkip to conversation

iommu/arm-smmu: Allow client devices to select direct mapping

PATCH Superseded Sai Prakash Ranjan 7 participants 16 messages Jan 22, 2020 – Apr 16, 2020

Subsystems (2)

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

cover iommu/arm-smmu: Allow client devices to select direct mapping

This series allows drm devices to set a default identity mapping using iommu_request_dm_for_dev(). First patch is a cleanup to support other SoCs to call into QCOM specific implementation and preparation for second patch. Second patch sets the default identity domain for drm devices.

Jordan Crouse (1):
iommu/arm-smmu: Allow client devices to select direct mapping

Sai Prakash Ranjan (1):
iommu: arm-smmu-impl: Convert to a generic reset implementation

 drivers/iommu/arm-smmu-impl.c |  8 +++--
 drivers/iommu/arm-smmu-qcom.c | 55 +++++++++++++++++++++++++++++++++--
 drivers/iommu/arm-smmu.c      |  3 ++
 drivers/iommu/arm-smmu.h      |  5 ++++
 4 files changed, 65 insertions(+), 6 deletions(-)
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
#1 lore raw

PATCH 1/2 iommu: arm-smmu-impl: Convert to a generic reset implementation

Currently the QCOM specific smmu reset implementation is very specific to SDM845 SoC and has a wait-for-safe logic which may not be required for other SoCs. So move the SDM845 specific logic to its specific reset function. Also add SC7180 SMMU compatible for calling into QCOM specific implementation.

---
 drivers/iommu/arm-smmu-impl.c |  8 +++++---
 drivers/iommu/arm-smmu-qcom.c | 16 +++++++++++++---
 2 files changed, 18 insertions(+), 6 deletions(-)
drivers/iommu/arm-smmu-impl.c +5 -3
index 74d97a886e93..c75b9d957b70 100644
@@ -150,6 +150,8 @@ static const struct arm_smmu_impl arm_mmu500_impl = {
 
 struct arm_smmu_device *arm_smmu_impl_init(struct arm_smmu_device *smmu)
 {
+	const struct device_node *np = smmu->dev->of_node;
+
 	/*
 	 * We will inevitably have to combine model-specific implementation
 	 * quirks with platform-specific integration quirks, but everything
@@ -166,11 +168,11 @@ struct arm_smmu_device *arm_smmu_impl_init(struct arm_smmu_device *smmu)
 		break;
 	}
 
-	if (of_property_read_bool(smmu->dev->of_node,
-				  "calxeda,smmu-secure-config-access"))
+	if (of_property_read_bool(np, "calxeda,smmu-secure-config-access"))
 		smmu->impl = &calxeda_impl;
 
-	if (of_device_is_compatible(smmu->dev->of_node, "qcom,sdm845-smmu-500"))
+	if (of_device_is_compatible(np, "qcom,sdm845-smmu-500") ||
+	    of_device_is_compatible(np, "qcom,sc7180-smmu-500"))
 		return qcom_smmu_impl_init(smmu);
 
 	return smmu;
drivers/iommu/arm-smmu-qcom.c +13 -3
index 24c071c1d8b0..64a4ab270ab7 100644
@@ -15,8 +15,6 @@ static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
 {
 	int ret;
 
-	arm_mmu500_reset(smmu);
-
 	/*
 	 * To address performance degradation in non-real time clients,
 	 * such as USB and UFS, turn off wait-for-safe on sdm845 based boards,
@@ -30,8 +28,20 @@ static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
 	return ret;
 }
 
+static int qcom_smmu500_reset(struct arm_smmu_device *smmu)
+{
+	const struct device_node *np = smmu->dev->of_node;
+
+	arm_mmu500_reset(smmu);
+
+	if (of_device_is_compatible(np, "qcom,sdm845-smmu-500"))
+		return qcom_sdm845_smmu500_reset(smmu);
+
+	return 0;
+}
+
 static const struct arm_smmu_impl qcom_smmu_impl = {
-	.reset = qcom_sdm845_smmu500_reset,
+	.reset = qcom_smmu500_reset,
 };
 
 struct arm_smmu_device *qcom_smmu_impl_init(struct arm_smmu_device *smmu)
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
#5 lore raw

PATCH 2/2 iommu/arm-smmu: Allow client devices to select direct mapping

In reply to Sai Prakash Ranjan

From: Jordan Crouse <[email protected]>

Some client devices want to directly map the IOMMU themselves instead of using the DMA domain. Allow those devices to opt in to direct mapping by way of a list of compatible strings.

---
 drivers/iommu/arm-smmu-qcom.c | 39 +++++++++++++++++++++++++++++++++++
 drivers/iommu/arm-smmu.c      |  3 +++
 drivers/iommu/arm-smmu.h      |  5 +++++
 3 files changed, 47 insertions(+)
drivers/iommu/arm-smmu-qcom.c +39 -0
index 64a4ab270ab7..ff746acd1c81 100644
@@ -3,6 +3,7 @@
  * Copyright (c) 2019, The Linux Foundation. All rights reserved.
  */
 
+#include <linux/of_device.h>
 #include <linux/qcom_scm.h>
 
 #include "arm-smmu.h"
@@ -11,6 +12,43 @@ struct qcom_smmu {
 	struct arm_smmu_device smmu;
 };
 
+static const struct arm_smmu_client_match_data qcom_adreno = {
+	.direct_mapping = true,
+};
+
+static const struct arm_smmu_client_match_data qcom_mdss = {
+	.direct_mapping = true,
+};
+
+static const struct of_device_id qcom_smmu_client_of_match[] = {
+	{ .compatible = "qcom,adreno", .data = &qcom_adreno },
+	{ .compatible = "qcom,mdp4", .data = &qcom_mdss },
+	{ .compatible = "qcom,mdss", .data = &qcom_mdss },
+	{ .compatible = "qcom,sc7180-mdss", .data = &qcom_mdss },
+	{ .compatible = "qcom,sdm845-mdss", .data = &qcom_mdss },
+	{},
+};
+
+static const struct arm_smmu_client_match_data *
+qcom_smmu_client_data(struct device *dev)
+{
+	const struct of_device_id *match =
+		of_match_device(qcom_smmu_client_of_match, dev);
+
+	return match ? match->data : NULL;
+}
+
+static int qcom_smmu_request_domain(struct device *dev)
+{
+	const struct arm_smmu_client_match_data *client;
+
+	client = qcom_smmu_client_data(dev);
+	if (client)
+		iommu_request_dm_for_dev(dev);
+
+	return 0;
+}
+
 static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
 {
 	int ret;
@@ -41,6 +79,7 @@ static int qcom_smmu500_reset(struct arm_smmu_device *smmu)
 }
 
 static const struct arm_smmu_impl qcom_smmu_impl = {
+	.req_domain = qcom_smmu_request_domain,
 	.reset = qcom_smmu500_reset,
 };
 
drivers/iommu/arm-smmu.c +3 -0
index 16c4b87af42b..67dd9326247a 100644
@@ -1448,6 +1448,9 @@ static int arm_smmu_add_device(struct device *dev)
 	device_link_add(dev, smmu->dev,
 			DL_FLAG_PM_RUNTIME | DL_FLAG_AUTOREMOVE_SUPPLIER);
 
+	if (smmu->impl && smmu->impl->req_domain)
+		return smmu->impl->req_domain(dev);
+
 	return 0;
 
 out_cfg_free:
drivers/iommu/arm-smmu.h +5 -0
index 8d1cd54d82a6..059dc9c39f64 100644
@@ -244,6 +244,10 @@ enum arm_smmu_arch_version {
 	ARM_SMMU_V2,
 };
 
+struct arm_smmu_client_match_data {
+	bool direct_mapping;
+};
+
 enum arm_smmu_implementation {
 	GENERIC_SMMU,
 	ARM_MMU500,
@@ -386,6 +390,7 @@ struct arm_smmu_impl {
 	int (*init_context)(struct arm_smmu_domain *smmu_domain);
 	void (*tlb_sync)(struct arm_smmu_device *smmu, int page, int sync,
 			 int status);
+	int (*req_domain)(struct device *dev);
 };
 
 static inline void __iomem *arm_smmu_page(struct arm_smmu_device *smmu, int n)
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
#12 lore raw
In reply to Sai Prakash Ranjan

Hello Robin, Will

Quoted 9 lines · Sai Prakash Ranjan

This series allows drm devices to set a default identity mapping using iommu_request_dm_for_dev(). First patch is a cleanup to support other SoCs to call into QCOM specific implementation and preparation for second patch. Second patch sets the default identity domain for drm devices.

Jordan Crouse (1):
iommu/arm-smmu: Allow client devices to select direct mapping

Quoted 8 lines

Sai Prakash Ranjan (1):
iommu: arm-smmu-impl: Convert to a generic reset implementation

 drivers/iommu/arm-smmu-impl.c |  8 +++--
 drivers/iommu/arm-smmu-qcom.c | 55 +++++++++++++++++++++++++++++++++--
 drivers/iommu/arm-smmu.c      |  3 ++
 drivers/iommu/arm-smmu.h      |  5 ++++
 4 files changed, 65 insertions(+), 6 deletions(-)

Any review comments?

Thanks,
Sai

Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
Reviewed-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] iommu: arm-smmu-impl: Convert to a generic reset implementation

In reply to Sai Prakash Ranjan
Quoted 7 lines · Sai Prakash Ranjan

Currently the QCOM specific smmu reset implementation is very specific to SDM845 SoC and has a wait-for-safe logic which may not be required for other SoCs. So move the SDM845 specific logic to its specific reset function. Also add SC7180 SMMU compatible for calling into QCOM specific implementation.

Regards,
Bjorn

Quoted 71 lines
---
 drivers/iommu/arm-smmu-impl.c |  8 +++++---
 drivers/iommu/arm-smmu-qcom.c | 16 +++++++++++++---
 2 files changed, 18 insertions(+), 6 deletions(-)
drivers/iommu/arm-smmu-impl.c +5 -3
index 74d97a886e93..c75b9d957b70 100644
@@ -150,6 +150,8 @@ static const struct arm_smmu_impl arm_mmu500_impl = {
 
 struct arm_smmu_device *arm_smmu_impl_init(struct arm_smmu_device *smmu)
 {
+	const struct device_node *np = smmu->dev->of_node;
+
 	/*
 	 * We will inevitably have to combine model-specific implementation
 	 * quirks with platform-specific integration quirks, but everything
@@ -166,11 +168,11 @@ struct arm_smmu_device *arm_smmu_impl_init(struct arm_smmu_device *smmu)
 		break;
 	}
 
-	if (of_property_read_bool(smmu->dev->of_node,
-				  "calxeda,smmu-secure-config-access"))
+	if (of_property_read_bool(np, "calxeda,smmu-secure-config-access"))
 		smmu->impl = &calxeda_impl;
 
-	if (of_device_is_compatible(smmu->dev->of_node, "qcom,sdm845-smmu-500"))
+	if (of_device_is_compatible(np, "qcom,sdm845-smmu-500") ||
+	    of_device_is_compatible(np, "qcom,sc7180-smmu-500"))
 		return qcom_smmu_impl_init(smmu);
 
 	return smmu;
drivers/iommu/arm-smmu-qcom.c +13 -3
index 24c071c1d8b0..64a4ab270ab7 100644
@@ -15,8 +15,6 @@ static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
 {
 	int ret;
 
-	arm_mmu500_reset(smmu);
-
 	/*
 	 * To address performance degradation in non-real time clients,
 	 * such as USB and UFS, turn off wait-for-safe on sdm845 based boards,
@@ -30,8 +28,20 @@ static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
 	return ret;
 }
 
+static int qcom_smmu500_reset(struct arm_smmu_device *smmu)
+{
+	const struct device_node *np = smmu->dev->of_node;
+
+	arm_mmu500_reset(smmu);
+
+	if (of_device_is_compatible(np, "qcom,sdm845-smmu-500"))
+		return qcom_sdm845_smmu500_reset(smmu);
+
+	return 0;
+}
+
 static const struct arm_smmu_impl qcom_smmu_impl = {
-	.reset = qcom_sdm845_smmu500_reset,
+	.reset = qcom_smmu500_reset,
 };
 
 struct arm_smmu_device *qcom_smmu_impl_init(struct arm_smmu_device *smmu)
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
Reviewed-by on v1 1/2
#3 lore raw

Re: [PATCH 1/2] iommu: arm-smmu-impl: Convert to a generic reset implementation

In reply to Sai Prakash Ranjan

Quoting Sai Prakash Ranjan (2020-01-22 03:48:01)

Quoted 8 lines

Currently the QCOM specific smmu reset implementation is very specific to SDM845 SoC and has a wait-for-safe logic which may not be required for other SoCs. So move the SDM845 specific logic to its specific reset function. Also add SC7180 SMMU compatible for calling into QCOM specific implementation.

---

#13 lore raw
In reply to Sai Prakash Ranjan
Quoted 15 lines · Sai Prakash Ranjan

Hello Robin, Will

Quoted 12 lines · Sai Prakash Ranjan

This series allows drm devices to set a default identity mapping using iommu_request_dm_for_dev(). First patch is a cleanup to support other SoCs to call into QCOM specific implementation and preparation for second patch. Second patch sets the default identity domain for drm devices.

Jordan Crouse (1):
iommu/arm-smmu: Allow client devices to select direct mapping

Sai Prakash Ranjan (1):
iommu: arm-smmu-impl: Convert to a generic reset implementation

Quoted 7 lines
Quoted 5 lines
 drivers/iommu/arm-smmu-impl.c |  8 +++--
 drivers/iommu/arm-smmu-qcom.c | 55 +++++++++++++++++++++++++++++++++--
 drivers/iommu/arm-smmu.c      |  3 ++
 drivers/iommu/arm-smmu.h      |  5 ++++
 4 files changed, 65 insertions(+), 6 deletions(-)

Any review comments?

Ping

What is the status of this series, is it ready to land or are any changes needed?

Thanks

Matthias

#14 lore raw

Hi Matthias,

Quoted 22 lines · Matthias Kaehlcke
Quoted 21 lines · Sai Prakash Ranjan

Hello Robin, Will

Quoted 17 lines · Sai Prakash Ranjan

This series allows drm devices to set a default identity mapping using iommu_request_dm_for_dev(). First patch is a cleanup to support other SoCs to call into QCOM specific implementation and preparation for second patch. Second patch sets the default identity domain for drm devices.

Jordan Crouse (1):
iommu/arm-smmu: Allow client devices to select direct mapping

Sai Prakash Ranjan (1):
iommu: arm-smmu-impl: Convert to a generic reset implementation

 drivers/iommu/arm-smmu-impl.c |  8 +++--
 drivers/iommu/arm-smmu-qcom.c | 55 +++++++++++++++++++++++++++++++++--
 drivers/iommu/arm-smmu.c      |  3 ++
 drivers/iommu/arm-smmu.h      |  5 ++++
 4 files changed, 65 insertions(+), 6 deletions(-)
Quoted 7 lines
Quoted 1 line

Any review comments?

Ping

What is the status of this series, is it ready to land or are any changes needed?

I am wondering the same :)

Thanks,
Sai

Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
#15 lore raw
In reply to Matthias Kaehlcke
Quoted 24 lines · Matthias Kaehlcke
Quoted 15 lines · Sai Prakash Ranjan

Hello Robin, Will

Quoted 12 lines · Sai Prakash Ranjan

This series allows drm devices to set a default identity mapping using iommu_request_dm_for_dev(). First patch is a cleanup to support other SoCs to call into QCOM specific implementation and preparation for second patch. Second patch sets the default identity domain for drm devices.

Jordan Crouse (1):
iommu/arm-smmu: Allow client devices to select direct mapping

Sai Prakash Ranjan (1):
iommu: arm-smmu-impl: Convert to a generic reset implementation

Quoted 7 lines
Quoted 5 lines
 drivers/iommu/arm-smmu-impl.c |  8 +++--
 drivers/iommu/arm-smmu-qcom.c | 55 +++++++++++++++++++++++++++++++++--
 drivers/iommu/arm-smmu.c      |  3 ++
 drivers/iommu/arm-smmu.h      |  5 ++++
 4 files changed, 65 insertions(+), 6 deletions(-)

Any review comments?

Quoted 8 lines

Ping

What is the status of this series, is it ready to land or are any changes needed?

Thanks

Matthias

I think this is up in the air following the changes that Joerg suggested:
https://lists.linuxfoundation.org/pipermail/iommu/2020-April/043017.html

Jordan

Signature · 6 lines
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
#6 lore raw
In reply to Sai Prakash Ranjan
Quoted 29 lines · Sai Prakash Ranjan

From: Jordan Crouse <[email protected]>

Some client devices want to directly map the IOMMU themselves instead of using the DMA domain. Allow those devices to opt in to direct mapping by way of a list of compatible strings.

---
 drivers/iommu/arm-smmu-qcom.c | 39 +++++++++++++++++++++++++++++++++++
 drivers/iommu/arm-smmu.c      |  3 +++
 drivers/iommu/arm-smmu.h      |  5 +++++
 3 files changed, 47 insertions(+)
drivers/iommu/arm-smmu-qcom.c +1 -0
index 64a4ab270ab7..ff746acd1c81 100644
@@ -3,6 +3,7 @@
  * Copyright (c) 2019, The Linux Foundation. All rights reserved.
  */
 
+#include <linux/of_device.h>
 #include <linux/qcom_scm.h>
 
 #include "arm-smmu.h"

drivers/iommu/arm-smmu-qcom.c struct qcom_smmu { · line 20

Quoted 9 lines
drivers/iommu/arm-smmu-qcom.c +6 -0
@@ -11,6 +12,43 @@ struct qcom_smmu {
        struct arm_smmu_device smmu;
 };
 
+static const struct arm_smmu_client_match_data qcom_adreno = {
+       .direct_mapping = true,
+};
+
+static const struct arm_smmu_client_match_data qcom_mdss = {
+       .direct_mapping = true,

I don't actually see direct_mapping being used. Shouldn't this member be checked somewhere?

-Evan

#7 lore raw

Hi Evan,

Quoted 35 lines · Evan Green
Quoted 33 lines · Sai Prakash Ranjan

From: Jordan Crouse <[email protected]>

Some client devices want to directly map the IOMMU themselves instead of using the DMA domain. Allow those devices to opt in to direct mapping by way of a list of compatible strings.

---
 drivers/iommu/arm-smmu-qcom.c | 39 
+++++++++++++++++++++++++++++++++++
 drivers/iommu/arm-smmu.c      |  3 +++
 drivers/iommu/arm-smmu.h      |  5 +++++
 3 files changed, 47 insertions(+)
drivers/iommu/arm-smmu-qcom.c +1 -0
index 64a4ab270ab7..ff746acd1c81 100644
@@ -3,6 +3,7 @@
  * Copyright (c) 2019, The Linux Foundation. All rights reserved.
  */
 
+#include <linux/of_device.h>
 #include <linux/qcom_scm.h>
 
 #include "arm-smmu.h"
@@ -11,6 +12,43 @@ struct qcom_smmu {
        struct arm_smmu_device smmu;
 };
 
Quoted 10 lines

drivers/iommu/arm-smmu-qcom.c struct qcom_smmu { · line 20

Quoted 6 lines
drivers/iommu/arm-smmu-qcom.c +6 -0
+static const struct arm_smmu_client_match_data qcom_adreno = {
+       .direct_mapping = true,
+};
+
+static const struct arm_smmu_client_match_data qcom_mdss = {
+       .direct_mapping = true,

I don't actually see direct_mapping being used. Shouldn't this member be checked somewhere?

Thanks for spotting this, my bad. It should be checked in qcom_smmu_request_domain().

drivers/iommu/arm-smmu-qcom.c +1 -1
index ff746acd1c81..3ff62ca13ad5 100644
@@ -43,7 +43,7 @@ static int qcom_smmu_request_domain(struct device *dev)
        const struct arm_smmu_client_match_data *client;
 
        client = qcom_smmu_client_data(dev);
-       if (client)
+       if (client && client->direct_mapping)
                iommu_request_dm_for_dev(dev);
 
        return 0;

-Sai

Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
reviewer Reviewed-by on v1 1/2
#4 lore raw

Re: [PATCH 1/2] iommu: arm-smmu-impl: Convert to a generic reset implementation

In reply to Sai Prakash Ranjan
Quoted 63 lines · Sai Prakash Ranjan

Currently the QCOM specific smmu reset implementation is very specific to SDM845 SoC and has a wait-for-safe logic which may not be required for other SoCs. So move the SDM845 specific logic to its specific reset function. Also add SC7180 SMMU compatible for calling into QCOM specific implementation.

---
  drivers/iommu/arm-smmu-impl.c |  8 +++++---
  drivers/iommu/arm-smmu-qcom.c | 16 +++++++++++++---
  2 files changed, 18 insertions(+), 6 deletions(-)
drivers/iommu/arm-smmu-impl.c +5 -3
index 74d97a886e93..c75b9d957b70 100644
@@ -150,6 +150,8 @@ static const struct arm_smmu_impl arm_mmu500_impl = {
 
  struct arm_smmu_device *arm_smmu_impl_init(struct arm_smmu_device *smmu)
  {
+	const struct device_node *np = smmu->dev->of_node;
+
  	/*
  	 * We will inevitably have to combine model-specific implementation
  	 * quirks with platform-specific integration quirks, but everything
@@ -166,11 +168,11 @@ struct arm_smmu_device *arm_smmu_impl_init(struct arm_smmu_device *smmu)
  		break;
  	}
 
-	if (of_property_read_bool(smmu->dev->of_node,
-				  "calxeda,smmu-secure-config-access"))
+	if (of_property_read_bool(np, "calxeda,smmu-secure-config-access"))
  		smmu->impl = &calxeda_impl;
 
-	if (of_device_is_compatible(smmu->dev->of_node, "qcom,sdm845-smmu-500"))
+	if (of_device_is_compatible(np, "qcom,sdm845-smmu-500") ||
+	    of_device_is_compatible(np, "qcom,sc7180-smmu-500"))
  		return qcom_smmu_impl_init(smmu);
 
  	return smmu;
drivers/iommu/arm-smmu-qcom.c +6 -2
index 24c071c1d8b0..64a4ab270ab7 100644
@@ -15,8 +15,6 @@ static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
  {
  	int ret;
 
-	arm_mmu500_reset(smmu);
-
  	/*
  	 * To address performance degradation in non-real time clients,
  	 * such as USB and UFS, turn off wait-for-safe on sdm845 based boards,
@@ -30,8 +28,20 @@ static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
  	return ret;
  }
 
+static int qcom_smmu500_reset(struct arm_smmu_device *smmu)
+{
+	const struct device_node *np = smmu->dev->of_node;
+
+	arm_mmu500_reset(smmu);
+

drivers/iommu/arm-smmu-qcom.c static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu) · line 45

Quoted 10 lines
drivers/iommu/arm-smmu-qcom.c +7 -1
+	if (of_device_is_compatible(np, "qcom,sdm845-smmu-500"))
+		return qcom_sdm845_smmu500_reset(smmu);
+
+	return 0;
+}
+
  static const struct arm_smmu_impl qcom_smmu_impl = {
-	.reset = qcom_sdm845_smmu500_reset,
+	.reset = qcom_smmu500_reset,
  };

It might be logical to have a separate SDM845 impl rather than indirecting within the callback itself, but I'm not too concerned either way. For the arm-smmu-impl.c changes,

Thanks,
Robin.

drivers/iommu/arm-smmu-qcom.c static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu) · line 47

Quoted 3 lines
drivers/iommu/arm-smmu-qcom.c +0 -0
  struct arm_smmu_device *qcom_smmu_impl_init(struct arm_smmu_device *smmu)
 

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

reviewer
#8 lore raw
In reply to Sai Prakash Ranjan
Quoted 28 lines · Sai Prakash Ranjan

From: Jordan Crouse <[email protected]>

Some client devices want to directly map the IOMMU themselves instead of using the DMA domain. Allow those devices to opt in to direct mapping by way of a list of compatible strings.

---
  drivers/iommu/arm-smmu-qcom.c | 39 +++++++++++++++++++++++++++++++++++
  drivers/iommu/arm-smmu.c      |  3 +++
  drivers/iommu/arm-smmu.h      |  5 +++++
  3 files changed, 47 insertions(+)
drivers/iommu/arm-smmu-qcom.c +1 -0
index 64a4ab270ab7..ff746acd1c81 100644
@@ -3,6 +3,7 @@
   * Copyright (c) 2019, The Linux Foundation. All rights reserved.
   */
 
+#include <linux/of_device.h>
  #include <linux/qcom_scm.h>
 
  #include "arm-smmu.h"

drivers/iommu/arm-smmu-qcom.c struct qcom_smmu { · line 21

Quoted 10 lines
drivers/iommu/arm-smmu-qcom.c +7 -0
@@ -11,6 +12,43 @@ struct qcom_smmu {
  	struct arm_smmu_device smmu;
  };
 
+static const struct arm_smmu_client_match_data qcom_adreno = {
+	.direct_mapping = true,
+};
+
+static const struct arm_smmu_client_match_data qcom_mdss = {
+	.direct_mapping = true,
+};

Might it make sense to group these by the desired SMMU behaviour rather than (apparently) what kind of device the client happens to be, which seems like a completely arbitrary distinction from the SMMU driver's PoV?

Quoted 7 lines
drivers/iommu/arm-smmu-qcom.c +7 -0
+
+static const struct of_device_id qcom_smmu_client_of_match[] = {
+	{ .compatible = "qcom,adreno", .data = &qcom_adreno },
+	{ .compatible = "qcom,mdp4", .data = &qcom_mdss },
+	{ .compatible = "qcom,mdss", .data = &qcom_mdss },
+	{ .compatible = "qcom,sc7180-mdss", .data = &qcom_mdss },
+	{ .compatible = "qcom,sdm845-mdss", .data = &qcom_mdss },

drivers/iommu/arm-smmu-qcom.c struct qcom_smmu { · line 38

Quoted 10 lines
drivers/iommu/arm-smmu-qcom.c +10 -0
+	{},
+};
+
+static const struct arm_smmu_client_match_data *
+qcom_smmu_client_data(struct device *dev)
+{
+	const struct of_device_id *match =
+		of_match_device(qcom_smmu_client_of_match, dev);
+
+	return match ? match->data : NULL;

of_device_get_match_data() is your friend.

Quoted 29 lines
drivers/iommu/arm-smmu-qcom.c +14 -0
+}
+
+static int qcom_smmu_request_domain(struct device *dev)
+{
+	const struct arm_smmu_client_match_data *client;
+
+	client = qcom_smmu_client_data(dev);
+	if (client)
+		iommu_request_dm_for_dev(dev);
+
+	return 0;
+}
+
  static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
  {
  	int ret;
@@ -41,6 +79,7 @@ static int qcom_smmu500_reset(struct arm_smmu_device *smmu)
  }
 
  static const struct arm_smmu_impl qcom_smmu_impl = {
+	.req_domain = qcom_smmu_request_domain,
  	.reset = qcom_smmu500_reset,
  };
 
drivers/iommu/arm-smmu.c +0 -0
index 16c4b87af42b..67dd9326247a 100644

drivers/iommu/arm-smmu.c static int arm_smmu_add_device(struct device *dev) · line 1453

Quoted 6 lines
drivers/iommu/arm-smmu.c +3 -0
@@ -1448,6 +1448,9 @@ static int arm_smmu_add_device(struct device *dev)
  	device_link_add(dev, smmu->dev,
  			DL_FLAG_PM_RUNTIME | DL_FLAG_AUTOREMOVE_SUPPLIER);
 
+	if (smmu->impl && smmu->impl->req_domain)
+		return smmu->impl->req_domain(dev);
+

There are about 5 different patchsets flying around at the moment that all touch default domain allocation, so this is a fast-moving target, but I think where the dust should settle is with arm_smmu_ops forwarding .def_domain_type (or whatever it ends up as) calls to arm_smmu_impl as appropriate.

Quoted 8 lines
drivers/iommu/arm-smmu.c +0 -0
  	return 0;
 
  out_cfg_free:
drivers/iommu/arm-smmu.h +0 -0
index 8d1cd54d82a6..059dc9c39f64 100644

drivers/iommu/arm-smmu.h enum arm_smmu_arch_version { · line 249

Quoted 6 lines
drivers/iommu/arm-smmu.h +3 -0
@@ -244,6 +244,10 @@ enum arm_smmu_arch_version {
  	ARM_SMMU_V2,
  };
 
+struct arm_smmu_client_match_data {
+	bool direct_mapping;
+};

Does this need to be public? I don't see the other users...

Robin.

drivers/iommu/arm-smmu.h struct arm_smmu_impl { · line 396

Quoted 13 lines
drivers/iommu/arm-smmu.h +2 -0
+
  enum arm_smmu_implementation {
  	GENERIC_SMMU,
  	ARM_MMU500,
@@ -386,6 +390,7 @@ struct arm_smmu_impl {
  	int (*init_context)(struct arm_smmu_domain *smmu_domain);
  	void (*tlb_sync)(struct arm_smmu_device *smmu, int page, int sync,
  			 int status);
+	int (*req_domain)(struct device *dev);
  };
 
  static inline void __iomem *arm_smmu_page(struct arm_smmu_device *smmu, int n)

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#9 lore raw

Hi Robin,

Quoted 34 lines · Robin Murphy
Quoted 33 lines · Sai Prakash Ranjan

From: Jordan Crouse <[email protected]>

Some client devices want to directly map the IOMMU themselves instead of using the DMA domain. Allow those devices to opt in to direct mapping by way of a list of compatible strings.

---
  drivers/iommu/arm-smmu-qcom.c | 39 
+++++++++++++++++++++++++++++++++++
  drivers/iommu/arm-smmu.c      |  3 +++
  drivers/iommu/arm-smmu.h      |  5 +++++
  3 files changed, 47 insertions(+)
drivers/iommu/arm-smmu-qcom.c +3 -0
index 64a4ab270ab7..ff746acd1c81 100644
@@ -3,6 +3,7 @@
   * Copyright (c) 2019, The Linux Foundation. All rights reserved.
   */
  +#include <linux/of_device.h>
  #include <linux/qcom_scm.h>
    #include "arm-smmu.h"
@@ -11,6 +12,43 @@ struct qcom_smmu {
  	struct arm_smmu_device smmu;
  };
  +static const struct arm_smmu_client_match_data qcom_adreno = {
+	.direct_mapping = true,
+};
+
Quoted 9 lines

drivers/iommu/arm-smmu-qcom.c struct qcom_smmu { · line 21

Quoted 3 lines
drivers/iommu/arm-smmu-qcom.c +3 -0
+static const struct arm_smmu_client_match_data qcom_mdss = {
+	.direct_mapping = true,
+};

Might it make sense to group these by the desired SMMU behaviour rather than (apparently) what kind of device the client happens to be, which seems like a completely arbitrary distinction from the SMMU driver's PoV?

Sorry, I did not get the "grouping by the desired SMMU behaviour" thing. Could you please give some more details?

Quoted 11 lines
Quoted 11 lines
drivers/iommu/arm-smmu-qcom.c +11 -0
+
+static const struct of_device_id qcom_smmu_client_of_match[] = {
+	{ .compatible = "qcom,adreno", .data = &qcom_adreno },
+	{ .compatible = "qcom,mdp4", .data = &qcom_mdss },
+	{ .compatible = "qcom,mdss", .data = &qcom_mdss },
+	{ .compatible = "qcom,sc7180-mdss", .data = &qcom_mdss },
+	{ .compatible = "qcom,sdm845-mdss", .data = &qcom_mdss },
+	{},
+};
+
+static const struct arm_smmu_client_match_data *
Quoted 9 lines

drivers/iommu/arm-smmu-qcom.c struct qcom_smmu { · line 38

Quoted 6 lines
drivers/iommu/arm-smmu-qcom.c +6 -0
+qcom_smmu_client_data(struct device *dev)
+{
+	const struct of_device_id *match =
+		of_match_device(qcom_smmu_client_of_match, dev);
+
+	return match ? match->data : NULL;

of_device_get_match_data() is your friend.

Ok will use it.

Quoted 32 lines
Quoted 32 lines
drivers/iommu/arm-smmu-qcom.c +14 -0
+}
+
+static int qcom_smmu_request_domain(struct device *dev)
+{
+	const struct arm_smmu_client_match_data *client;
+
+	client = qcom_smmu_client_data(dev);
+	if (client)
+		iommu_request_dm_for_dev(dev);
+
+	return 0;
+}
+
  static int qcom_sdm845_smmu500_reset(struct arm_smmu_device *smmu)
  {
  	int ret;
@@ -41,6 +79,7 @@ static int qcom_smmu500_reset(struct arm_smmu_device 
 *smmu)
  }
    static const struct arm_smmu_impl qcom_smmu_impl = {
+	.req_domain = qcom_smmu_request_domain,
  	.reset = qcom_smmu500_reset,
  };
  diff --git a/drivers/iommu/arm-smmu.c b/drivers/iommu/arm-smmu.c

index 16c4b87af42b..67dd9326247a 100644

drivers/iommu/arm-smmu.c +0 -0
@@ -1448,6 +1448,9 @@ static int arm_smmu_add_device(struct device 
 *dev)
  	device_link_add(dev, smmu->dev,
  			DL_FLAG_PM_RUNTIME | DL_FLAG_AUTOREMOVE_SUPPLIER);
  +	if (smmu->impl && smmu->impl->req_domain)
Quoted 9 lines

drivers/iommu/arm-smmu.c static int arm_smmu_add_device(struct device *dev) · line 1453

Quoted 2 lines
drivers/iommu/arm-smmu.c +2 -0
+		return smmu->impl->req_domain(dev);
+

There are about 5 different patchsets flying around at the moment that all touch default domain allocation, so this is a fast-moving target, but I think where the dust should settle is with arm_smmu_ops forwarding .def_domain_type (or whatever it ends up as) calls to arm_smmu_impl as appropriate.

I'll wait till the dust settles down and then post the next version.

Quoted 6 lines
Quoted 6 lines
drivers/iommu/arm-smmu.c +0 -0
  	return 0;
    out_cfg_free:
drivers/iommu/arm-smmu.h +0 -0
index 8d1cd54d82a6..059dc9c39f64 100644
Quoted 9 lines

drivers/iommu/arm-smmu.h enum arm_smmu_arch_version { · line 248

Quoted 6 lines
drivers/iommu/arm-smmu.h +2 -0
@@ -244,6 +244,10 @@ enum arm_smmu_arch_version {
  	ARM_SMMU_V2,
  };
  +struct arm_smmu_client_match_data {
+	bool direct_mapping;
+};

Does this need to be public? I don't see the other users...

Will move this out.

Thanks,
Sai

Signature · 7 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
reviewer
#10 lore raw
Quoted 46 lines · Sai Prakash Ranjan

Hi Robin,

Quoted 33 lines · Robin Murphy
Quoted 32 lines · Sai Prakash Ranjan

From: Jordan Crouse <[email protected]>

Some client devices want to directly map the IOMMU themselves instead of using the DMA domain. Allow those devices to opt in to direct mapping by way of a list of compatible strings.

---
  drivers/iommu/arm-smmu-qcom.c | 39 +++++++++++++++++++++++++++++++++++ drivers/iommu/arm-smmu.c      |  3 +++ drivers/iommu/arm-smmu.h      |  5 +++++

  3 files changed, 47 insertions(+)
drivers/iommu/arm-smmu-qcom.c +0 -0
index 64a4ab270ab7..ff746acd1c81 100644
@@ -3,6 +3,7 @@
    * Copyright (c) 2019, The Linux Foundation. All rights reserved.
    */

  +#include <linux/of_device.h>
  #include <linux/qcom_scm.h>
    #include "arm-smmu.h"

drivers/iommu/arm-smmu-qcom.c +3 -0
@@ -11,6 +12,43 @@ struct qcom_smmu {
       struct arm_smmu_device smmu;
   };
   +static const struct arm_smmu_client_match_data qcom_adreno = {
+    .direct_mapping = true,
+};
+
Quoted 9 lines

drivers/iommu/arm-smmu-qcom.c struct qcom_smmu { · line 21

Quoted 3 lines
drivers/iommu/arm-smmu-qcom.c +3 -0
+static const struct arm_smmu_client_match_data qcom_mdss = {
+    .direct_mapping = true,
+};

Might it make sense to group these by the desired SMMU behaviour rather than (apparently) what kind of device the client happens to be, which seems like a completely arbitrary distinction from the SMMU driver's PoV?

Quoted 2 lines

Sorry, I did not get the "grouping by the desired SMMU behaviour" thing. Could you please give some more details?

I mean this pattern:

device_a_data {
	.thing = this;
}
device_b_data {
	.thing = this;
}
device_c_data {
	.thing = that;
}
match[] = {
	{ .compatible = "A", .data = &device_a_data },
	{ .compatible = "B", .data = &device_b_data },
	{ .compatible = "C", .data = &device_c_data },
};

...vs. this pattern:

do_this {
	.thing = this;
}
do_that {
	.thing = that;
}
match[] = {
	{ .compatible = "A", .data = &do_this },
	{ .compatible = "B", .data = &do_this },
	{ .compatible = "C", .data = &do_that },
};

From the perspective of the thing doing the thing, grouping the data by device is meaningless if all that matters is whether to do this or that. The second pattern expresses that more naturally.

Of course if every device turns out to need a unique combination of several behaviours, then there ends up being no practical difference except that it's probably easier to come up with nice names under the first pattern. I guess it's up to how you see this developing in future; whether you're likely to need fine-grained per-device control, or don't expect it to go much beyond domain type.

Robin.

#11 lore raw

Hi Robin,

Quoted 91 lines · Robin Murphy
Quoted 47 lines · Sai Prakash Ranjan

Hi Robin,

Quoted 34 lines · Robin Murphy
Quoted 33 lines · Sai Prakash Ranjan

From: Jordan Crouse <[email protected]>

Some client devices want to directly map the IOMMU themselves instead of using the DMA domain. Allow those devices to opt in to direct mapping by way of a list of compatible strings.

---
  drivers/iommu/arm-smmu-qcom.c | 39
+++++++++++++++++++++++++++++++++++
  drivers/iommu/arm-smmu.c      |  3 +++
  drivers/iommu/arm-smmu.h      |  5 +++++

  3 files changed, 47 insertions(+)
drivers/iommu/arm-smmu-qcom.c +0 -0
index 64a4ab270ab7..ff746acd1c81 100644
@@ -3,6 +3,7 @@
    * Copyright (c) 2019, The Linux Foundation. All rights reserved.
    */

  +#include <linux/of_device.h>
  #include <linux/qcom_scm.h>
    #include "arm-smmu.h"

drivers/iommu/arm-smmu-qcom.c +3 -0
@@ -11,6 +12,43 @@ struct qcom_smmu {
       struct arm_smmu_device smmu;
   };
   +static const struct arm_smmu_client_match_data qcom_adreno = {
+    .direct_mapping = true,
+};
+
Quoted 9 lines

drivers/iommu/arm-smmu-qcom.c struct qcom_smmu { · line 21

Quoted 3 lines
drivers/iommu/arm-smmu-qcom.c +3 -0
+static const struct arm_smmu_client_match_data qcom_mdss = {
+    .direct_mapping = true,
+};

Might it make sense to group these by the desired SMMU behaviour rather than (apparently) what kind of device the client happens to be, which seems like a completely arbitrary distinction from the SMMU driver's PoV?

Quoted 2 lines

Sorry, I did not get the "grouping by the desired SMMU behaviour" thing. Could you please give some more details?

I mean this pattern:

device_a_data {
	.thing = this;
}
device_b_data {
	.thing = this;
}
device_c_data {
	.thing = that;
}
match[] = {
	{ .compatible = "A", .data = &device_a_data },
	{ .compatible = "B", .data = &device_b_data },
	{ .compatible = "C", .data = &device_c_data },
};

...vs. this pattern:

do_this {
	.thing = this;
}
do_that {
	.thing = that;
}
match[] = {
	{ .compatible = "A", .data = &do_this },
	{ .compatible = "B", .data = &do_this },
	{ .compatible = "C", .data = &do_that },
};

From the perspective of the thing doing the thing, grouping the data by device is meaningless if all that matters is whether to do this or that. The second pattern expresses that more naturally.

Quoted 7 lines

Of course if every device turns out to need a unique combination of several behaviours, then there ends up being no practical difference except that it's probably easier to come up with nice names under the first pattern. I guess it's up to how you see this developing in future; whether you're likely to need fine-grained per-device control, or don't expect it to go much beyond domain type.

Thanks for explaining *this thing* :)
I will update the patch to follow the 2nd pattern as it makes more sense to do_this or do_that directly. I'm not expecting anything other than domain type atleast for now but hey we can always add the functionality if the need arises.

Thanks,
Sai

Signature · 7 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel