Re: [PATCHv2 2/2] iommu/arm-smmu-qcom: Move the adreno smmu specific impl earlier

5 messages, 3 authors, 2021-02-26 · open the first message on its own page

Re: [PATCHv2 2/2] iommu/arm-smmu-qcom: Move the adreno smmu specific impl earlier

From: Bjorn Andersson <hidden>
Date: 2021-02-26 17:26:00

On Fri 26 Feb 03:55 CST 2021, Sai Prakash Ranjan wrote:
Adreno(GPU) SMMU and APSS(Application Processor SubSystem) SMMU
both implement "arm,mmu-500" in some QTI SoCs and to run through
adreno smmu specific implementation such as enabling split pagetables
support, we need to match the "qcom,adreno-smmu" compatible first
before apss smmu or else we will be running apps smmu implementation
for adreno smmu and the additional features for adreno smmu is never
set. For ex: we have "qcom,sc7280-smmu-500" compatible for both apps
and adreno smmu implementing "arm,mmu-500", so the adreno smmu
implementation is never reached because the current sequence checks
for apps smmu compatible(qcom,sc7280-smmu-500) first and runs that
specific impl and we never reach adreno smmu specific implementation.
So you're saying that you have a single SMMU instance that's compatible
with both an entry in qcom_smmu_impl_of_match[] and "qcom,adreno-smmu"?

Per your proposed change we will pick the adreno ops _only_ for this
component, essentially disabling the non-Adreno quirks selected by the
qcom impl. As such keeping the non-adreno compatible in the
qcom_smmu_impl_init[] seems to only serve to obfuscate the situation.

Don't we somehow need the combined set of quirks? (At least if we're
running this with a standard UEFI based boot flow?)

Regards,
Bjorn
quoted hunk
Suggested-by: Akhil P Oommen <redacted>
Signed-off-by: Sai Prakash Ranjan <redacted>
---
 drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
index bea3ee0dabc2..03f048aebb80 100644
--- a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
+++ b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
@@ -345,11 +345,17 @@ struct arm_smmu_device *qcom_smmu_impl_init(struct arm_smmu_device *smmu)
 {
 	const struct device_node *np = smmu->dev->of_node;
 
-	if (of_match_node(qcom_smmu_impl_of_match, np))
-		return qcom_smmu_create(smmu, &qcom_smmu_impl);
-
+	/*
+	 * Do not change this order of implementation, i.e., first adreno
+	 * smmu impl and then apss smmu since we can have both implementing
+	 * arm,mmu-500 in which case we will miss setting adreno smmu specific
+	 * features if the order is changed.
+	 */
 	if (of_device_is_compatible(np, "qcom,adreno-smmu"))
 		return qcom_smmu_create(smmu, &qcom_adreno_smmu_impl);
 
+	if (of_match_node(qcom_smmu_impl_of_match, np))
+		return qcom_smmu_create(smmu, &qcom_smmu_impl);
+
 	return smmu;
 }
-- 
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
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCHv2 2/2] iommu/arm-smmu-qcom: Move the adreno smmu specific impl earlier

From: Rob Clark <hidden>
Date: 2021-02-26 18:24:41

On Fri, Feb 26, 2021 at 9:24 AM Bjorn Andersson
[off-list ref] wrote:
On Fri 26 Feb 03:55 CST 2021, Sai Prakash Ranjan wrote:
quoted
Adreno(GPU) SMMU and APSS(Application Processor SubSystem) SMMU
both implement "arm,mmu-500" in some QTI SoCs and to run through
adreno smmu specific implementation such as enabling split pagetables
support, we need to match the "qcom,adreno-smmu" compatible first
before apss smmu or else we will be running apps smmu implementation
for adreno smmu and the additional features for adreno smmu is never
set. For ex: we have "qcom,sc7280-smmu-500" compatible for both apps
and adreno smmu implementing "arm,mmu-500", so the adreno smmu
implementation is never reached because the current sequence checks
for apps smmu compatible(qcom,sc7280-smmu-500) first and runs that
specific impl and we never reach adreno smmu specific implementation.
So you're saying that you have a single SMMU instance that's compatible
with both an entry in qcom_smmu_impl_of_match[] and "qcom,adreno-smmu"?

Per your proposed change we will pick the adreno ops _only_ for this
component, essentially disabling the non-Adreno quirks selected by the
qcom impl. As such keeping the non-adreno compatible in the
qcom_smmu_impl_init[] seems to only serve to obfuscate the situation.

Don't we somehow need the combined set of quirks? (At least if we're
running this with a standard UEFI based boot flow?)
are you thinking of the apps-smmu handover of display context bank?
That shouldn't change, the only thing that changes is that gpu-smmu
becomes an mmu-500, whereas previously only apps-smmu was..

BR,
-R

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCHv2 2/2] iommu/arm-smmu-qcom: Move the adreno smmu specific impl earlier

From: Jordan Crouse <hidden>
Date: 2021-02-26 18:49:08

On Fri, Feb 26, 2021 at 11:24:52AM -0600, Bjorn Andersson wrote:
On Fri 26 Feb 03:55 CST 2021, Sai Prakash Ranjan wrote:
quoted
Adreno(GPU) SMMU and APSS(Application Processor SubSystem) SMMU
both implement "arm,mmu-500" in some QTI SoCs and to run through
adreno smmu specific implementation such as enabling split pagetables
support, we need to match the "qcom,adreno-smmu" compatible first
before apss smmu or else we will be running apps smmu implementation
for adreno smmu and the additional features for adreno smmu is never
set. For ex: we have "qcom,sc7280-smmu-500" compatible for both apps
and adreno smmu implementing "arm,mmu-500", so the adreno smmu
implementation is never reached because the current sequence checks
for apps smmu compatible(qcom,sc7280-smmu-500) first and runs that
specific impl and we never reach adreno smmu specific implementation.
So you're saying that you have a single SMMU instance that's compatible
with both an entry in qcom_smmu_impl_of_match[] and "qcom,adreno-smmu"?

Per your proposed change we will pick the adreno ops _only_ for this
component, essentially disabling the non-Adreno quirks selected by the
qcom impl. As such keeping the non-adreno compatible in the
qcom_smmu_impl_init[] seems to only serve to obfuscate the situation.

Don't we somehow need the combined set of quirks? (At least if we're
running this with a standard UEFI based boot flow?)
We *do* need the combined set of quirks, so there has to be an adreno-smmu
impelmentation that matches the "generic" implementation with a few extra
function hooks added on. I'm not sure if there is a clever way to figure out how
to meld the implementation hooks at runtime but the alternative is to just make
sure that the adreno-smmu static struct calls the same quirks as its generic
partner.

Jordan
quoted
Suggested-by: Akhil P Oommen <redacted>
Signed-off-by: Sai Prakash Ranjan <redacted>
---
 drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
index bea3ee0dabc2..03f048aebb80 100644
--- a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
+++ b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
@@ -345,11 +345,17 @@ struct arm_smmu_device *qcom_smmu_impl_init(struct arm_smmu_device *smmu)
 {
 	const struct device_node *np = smmu->dev->of_node;
 
-	if (of_match_node(qcom_smmu_impl_of_match, np))
-		return qcom_smmu_create(smmu, &qcom_smmu_impl);
-
+	/*
+	 * Do not change this order of implementation, i.e., first adreno
+	 * smmu impl and then apss smmu since we can have both implementing
+	 * arm,mmu-500 in which case we will miss setting adreno smmu specific
+	 * features if the order is changed.
+	 */
 	if (of_device_is_compatible(np, "qcom,adreno-smmu"))
 		return qcom_smmu_create(smmu, &qcom_adreno_smmu_impl);
 
+	if (of_match_node(qcom_smmu_impl_of_match, np))
+		return qcom_smmu_create(smmu, &qcom_smmu_impl);
+
 	return smmu;
 }
-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCHv2 2/2] iommu/arm-smmu-qcom: Move the adreno smmu specific impl earlier

From: Bjorn Andersson <hidden>
Date: 2021-02-26 19:15:54

On Fri 26 Feb 12:23 CST 2021, Rob Clark wrote:
On Fri, Feb 26, 2021 at 9:24 AM Bjorn Andersson
[off-list ref] wrote:
quoted
On Fri 26 Feb 03:55 CST 2021, Sai Prakash Ranjan wrote:
quoted
Adreno(GPU) SMMU and APSS(Application Processor SubSystem) SMMU
both implement "arm,mmu-500" in some QTI SoCs and to run through
adreno smmu specific implementation such as enabling split pagetables
support, we need to match the "qcom,adreno-smmu" compatible first
before apss smmu or else we will be running apps smmu implementation
for adreno smmu and the additional features for adreno smmu is never
set. For ex: we have "qcom,sc7280-smmu-500" compatible for both apps
and adreno smmu implementing "arm,mmu-500", so the adreno smmu
implementation is never reached because the current sequence checks
for apps smmu compatible(qcom,sc7280-smmu-500) first and runs that
specific impl and we never reach adreno smmu specific implementation.
So you're saying that you have a single SMMU instance that's compatible
with both an entry in qcom_smmu_impl_of_match[] and "qcom,adreno-smmu"?

Per your proposed change we will pick the adreno ops _only_ for this
component, essentially disabling the non-Adreno quirks selected by the
qcom impl. As such keeping the non-adreno compatible in the
qcom_smmu_impl_init[] seems to only serve to obfuscate the situation.

Don't we somehow need the combined set of quirks? (At least if we're
running this with a standard UEFI based boot flow?)
are you thinking of the apps-smmu handover of display context bank?
That shouldn't change, the only thing that changes is that gpu-smmu
becomes an mmu-500, whereas previously only apps-smmu was..
The current logic picks one of:
1) is the compatible mentioned in qcom_smmu_impl_of_match[]
2) is the compatible an adreno
3) no quirks needed

The change flips the order of these, so the only way I can see this
change affecting things is if we expected a match on #2, but we got one
on #1.

Which implies that the instance that we want to act according to the
adreno impl was listed in qcom_smmu_impl_of_match[] - which either is
wrong, or there's a single instance that needs both behaviors.

(And I believe Jordan's answer confirms the latter - there's a single
SMMU instance that needs all them quirks at once)

Regards,
Bjorn

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCHv2 2/2] iommu/arm-smmu-qcom: Move the adreno smmu specific impl earlier

From: Jordan Crouse <hidden>
Date: 2021-02-26 19:55:18

On Fri, Feb 26, 2021 at 11:48:13AM -0700, Jordan Crouse wrote:
On Fri, Feb 26, 2021 at 11:24:52AM -0600, Bjorn Andersson wrote:
quoted
On Fri 26 Feb 03:55 CST 2021, Sai Prakash Ranjan wrote:
quoted
Adreno(GPU) SMMU and APSS(Application Processor SubSystem) SMMU
both implement "arm,mmu-500" in some QTI SoCs and to run through
adreno smmu specific implementation such as enabling split pagetables
support, we need to match the "qcom,adreno-smmu" compatible first
before apss smmu or else we will be running apps smmu implementation
for adreno smmu and the additional features for adreno smmu is never
set. For ex: we have "qcom,sc7280-smmu-500" compatible for both apps
and adreno smmu implementing "arm,mmu-500", so the adreno smmu
implementation is never reached because the current sequence checks
for apps smmu compatible(qcom,sc7280-smmu-500) first and runs that
specific impl and we never reach adreno smmu specific implementation.
So you're saying that you have a single SMMU instance that's compatible
with both an entry in qcom_smmu_impl_of_match[] and "qcom,adreno-smmu"?

Per your proposed change we will pick the adreno ops _only_ for this
component, essentially disabling the non-Adreno quirks selected by the
qcom impl. As such keeping the non-adreno compatible in the
qcom_smmu_impl_init[] seems to only serve to obfuscate the situation.

Don't we somehow need the combined set of quirks? (At least if we're
running this with a standard UEFI based boot flow?)
We *do* need the combined set of quirks, so there has to be an adreno-smmu
impelmentation that matches the "generic" implementation with a few extra
function hooks added on. I'm not sure if there is a clever way to figure out how
to meld the implementation hooks at runtime but the alternative is to just make
sure that the adreno-smmu static struct calls the same quirks as its generic
partner.
To clarify, the gpu-smmu doesn't strictly need the s2cr handoff or the cfg_probe
though it wouldn't hurt to have them since they would be essentially
passthroughs for the GPU. We do need to capture errata like the
sdm845_smmu500_reset which is already part of the upstream adreno
implementation.

I think the main takeaway is that if a new errata or quirk is added
for main mmu500 it needs to be considered for adreno-smmu too.

Jordan
quoted
quoted
Suggested-by: Akhil P Oommen <redacted>
Signed-off-by: Sai Prakash Ranjan <redacted>
---
 drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
index bea3ee0dabc2..03f048aebb80 100644
--- a/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
+++ b/drivers/iommu/arm/arm-smmu/arm-smmu-qcom.c
@@ -345,11 +345,17 @@ struct arm_smmu_device *qcom_smmu_impl_init(struct arm_smmu_device *smmu)
 {
 	const struct device_node *np = smmu->dev->of_node;
 
-	if (of_match_node(qcom_smmu_impl_of_match, np))
-		return qcom_smmu_create(smmu, &qcom_smmu_impl);
-
+	/*
+	 * Do not change this order of implementation, i.e., first adreno
+	 * smmu impl and then apss smmu since we can have both implementing
+	 * arm,mmu-500 in which case we will miss setting adreno smmu specific
+	 * features if the order is changed.
+	 */
 	if (of_device_is_compatible(np, "qcom,adreno-smmu"))
 		return qcom_smmu_create(smmu, &qcom_adreno_smmu_impl);
 
+	if (of_match_node(qcom_smmu_impl_of_match, np))
+		return qcom_smmu_create(smmu, &qcom_smmu_impl);
+
 	return smmu;
 }
-- 
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member
of Code Aurora Forum, hosted by The Linux Foundation
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help