[PATCH] drm/msm: Fix display fault handling

[PATCH] drm/msm: Fix display fault handling

[Rob Clark]

From: Rob Clark <robdclark@chromium.org>

It turns out that when the display is enabled by the bootloader, we can
get some transient iommu faults from the display.  Which doesn't go over
too well when we install a fault handler that is gpu specific.  To avoid
this, defer installing the fault handler until we get around to setting
up per-process pgtables (which is adreno_smmu specific).  The arm-smmu
fallback error reporting is sufficient for reporting display related
faults (and in fact was all we had prior to f8f934c180f629bb927a04fd90d)

Reported-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reported-by: Yassine Oudjana <y.oudjana@protonmail.com>
Fixes: 2a574cc05d38 ("drm/msm: Improve the a6xx page fault handler")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Tested-by: John Stultz <john.stultz@linaro.org>
---
 drivers/gpu/drm/msm/msm_iommu.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/msm/msm_iommu.c b/drivers/gpu/drm/msm/msm_iommu.c
index eed2a762e9dd..bcaddbba564d 100644
--- a/drivers/gpu/drm/msm/msm_iommu.c
+++ b/drivers/gpu/drm/msm/msm_iommu.c
@@ -142,6 +142,9 @@ static const struct iommu_flush_ops null_tlb_ops = {
 	.tlb_add_page = msm_iommu_tlb_add_page,
 };
 
+static int msm_fault_handler(struct iommu_domain *domain, struct device *dev,
+		unsigned long iova, int flags, void *arg);
+
 struct msm_mmu *msm_iommu_pagetable_create(struct msm_mmu *parent)
 {
 	struct adreno_smmu_priv *adreno_smmu = dev_get_drvdata(parent->dev);
@@ -157,6 +160,13 @@ struct msm_mmu *msm_iommu_pagetable_create(struct msm_mmu *parent)
 	if (!ttbr1_cfg)
 		return ERR_PTR(-ENODEV);
 
+	/*
+	 * Defer setting the fault handler until we have a valid adreno_smmu
+	 * to avoid accidentially installing a GPU specific fault handler for
+	 * the display's iommu
+	 */
+	iommu_set_fault_handler(iommu->domain, msm_fault_handler, iommu);
+
 	pagetable = kzalloc(sizeof(*pagetable), GFP_KERNEL);
 	if (!pagetable)
 		return ERR_PTR(-ENOMEM);
@@ -300,7 +310,6 @@ struct msm_mmu *msm_iommu_new(struct device *dev, struct iommu_domain *domain)
 
 	iommu->domain = domain;
 	msm_mmu_init(&iommu->base, dev, &funcs, MSM_MMU_IOMMU);
-	iommu_set_fault_handler(domain, msm_fault_handler, iommu);
 
 	atomic_set(&iommu->pagetables, 0);
 
-- 
2.31.1

Re: [PATCH] drm/msm: Fix display fault handling

[Yassine Oudjana]

On Wed, Jul 7 2021 at 21:57:05 +0400, Rob Clark <robdclark@gmail.com> 
wrote:
> From: Rob Clark <robdclark@chromium.org>
> 
> It turns out that when the display is enabled by the bootloader, we 
> can
> get some transient iommu faults from the display.  Which doesn't go 
> over
> too well when we install a fault handler that is gpu specific.  To 
> avoid
> this, defer installing the fault handler until we get around to 
> setting
> up per-process pgtables (which is adreno_smmu specific).  The arm-smmu
> fallback error reporting is sufficient for reporting display related
> faults (and in fact was all we had prior to 
> f8f934c180f629bb927a04fd90d)
> 
> Reported-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
> Reported-by: Yassine Oudjana <y.oudjana@protonmail.com>
> Fixes: 2a574cc05d38 ("drm/msm: Improve the a6xx page fault handler")
> Signed-off-by: Rob Clark <robdclark@chromium.org>
> Tested-by: John Stultz <john.stultz@linaro.org>
> ---
>  drivers/gpu/drm/msm/msm_iommu.c | 11 ++++++++++-
>  1 file changed, 10 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/msm/msm_iommu.c 
> b/drivers/gpu/drm/msm/msm_iommu.c
> index eed2a762e9dd..bcaddbba564d 100644
> --- a/drivers/gpu/drm/msm/msm_iommu.c
> +++ b/drivers/gpu/drm/msm/msm_iommu.c
> @@ -142,6 +142,9 @@ static const struct iommu_flush_ops null_tlb_ops 
> = {
>  	.tlb_add_page = msm_iommu_tlb_add_page,
>  };
> 
> +static int msm_fault_handler(struct iommu_domain *domain, struct 
> device *dev,
> +		unsigned long iova, int flags, void *arg);
> +
>  struct msm_mmu *msm_iommu_pagetable_create(struct msm_mmu *parent)
>  {
>  	struct adreno_smmu_priv *adreno_smmu = dev_get_drvdata(parent->dev);
> @@ -157,6 +160,13 @@ struct msm_mmu 
> *msm_iommu_pagetable_create(struct msm_mmu *parent)
>  	if (!ttbr1_cfg)
>  		return ERR_PTR(-ENODEV);
> 
> +	/*
> +	 * Defer setting the fault handler until we have a valid adreno_smmu
> +	 * to avoid accidentially installing a GPU specific fault handler 
> for
> +	 * the display's iommu
> +	 */
> +	iommu_set_fault_handler(iommu->domain, msm_fault_handler, iommu);
> +
>  	pagetable = kzalloc(sizeof(*pagetable), GFP_KERNEL);
>  	if (!pagetable)
>  		return ERR_PTR(-ENOMEM);
> @@ -300,7 +310,6 @@ struct msm_mmu *msm_iommu_new(struct device *dev, 
> struct iommu_domain *domain)
> 
>  	iommu->domain = domain;
>  	msm_mmu_init(&iommu->base, dev, &funcs, MSM_MMU_IOMMU);
> -	iommu_set_fault_handler(domain, msm_fault_handler, iommu);
> 
>  	atomic_set(&iommu->pagetables, 0);
> 
> --
> 2.31.1
> 

Tested-by: Yassine Oudjana <y.oudjana@protonmail.com>

Re: [PATCH] drm/msm: Fix display fault handling

[John Stultz]

On Wed, Jul 7, 2021 at 10:57 AM Rob Clark <robdclark@gmail.com> wrote:
>
> From: Rob Clark <robdclark@chromium.org>
>
> It turns out that when the display is enabled by the bootloader, we can
> get some transient iommu faults from the display.  Which doesn't go over
> too well when we install a fault handler that is gpu specific.  To avoid
> this, defer installing the fault handler until we get around to setting
> up per-process pgtables (which is adreno_smmu specific).  The arm-smmu
> fallback error reporting is sufficient for reporting display related
> faults (and in fact was all we had prior to f8f934c180f629bb927a04fd90d)
>
> Reported-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
> Reported-by: Yassine Oudjana <y.oudjana@protonmail.com>
> Fixes: 2a574cc05d38 ("drm/msm: Improve the a6xx page fault handler")
> Signed-off-by: Rob Clark <robdclark@chromium.org>
> Tested-by: John Stultz <john.stultz@linaro.org>
> ---

Hey folks!
  Just wanted to follow up on this, as it's still missing from
5.14-rc3 and is critical for resolving a boot regression on db845c. Is
there anything keeping this from heading upstream?

thanks
-john

Re: [PATCH] drm/msm: Fix display fault handling

[Bjorn Andersson]

On Wed 07 Jul 11:01 PDT 2021, Rob Clark wrote:

> From: Rob Clark <robdclark@chromium.org>
> 
> It turns out that when the display is enabled by the bootloader, we can
> get some transient iommu faults from the display.  Which doesn't go over
> too well when we install a fault handler that is gpu specific.  To avoid
> this, defer installing the fault handler until we get around to setting
> up per-process pgtables (which is adreno_smmu specific).  The arm-smmu
> fallback error reporting is sufficient for reporting display related
> faults (and in fact was all we had prior to f8f934c180f629bb927a04fd90d)
> 
> Reported-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
> Reported-by: Yassine Oudjana <y.oudjana@protonmail.com>
> Fixes: 2a574cc05d38 ("drm/msm: Improve the a6xx page fault handler")
> Signed-off-by: Rob Clark <robdclark@chromium.org>
> Tested-by: John Stultz <john.stultz@linaro.org>

Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>

Regards,
Bjorn

> ---
>  drivers/gpu/drm/msm/msm_iommu.c | 11 ++++++++++-
>  1 file changed, 10 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/msm/msm_iommu.c b/drivers/gpu/drm/msm/msm_iommu.c
> index eed2a762e9dd..bcaddbba564d 100644
> --- a/drivers/gpu/drm/msm/msm_iommu.c
> +++ b/drivers/gpu/drm/msm/msm_iommu.c
> @@ -142,6 +142,9 @@ static const struct iommu_flush_ops null_tlb_ops = {
>  	.tlb_add_page = msm_iommu_tlb_add_page,
>  };
>  
> +static int msm_fault_handler(struct iommu_domain *domain, struct device *dev,
> +		unsigned long iova, int flags, void *arg);
> +
>  struct msm_mmu *msm_iommu_pagetable_create(struct msm_mmu *parent)
>  {
>  	struct adreno_smmu_priv *adreno_smmu = dev_get_drvdata(parent->dev);
> @@ -157,6 +160,13 @@ struct msm_mmu *msm_iommu_pagetable_create(struct msm_mmu *parent)
>  	if (!ttbr1_cfg)
>  		return ERR_PTR(-ENODEV);
>  
> +	/*
> +	 * Defer setting the fault handler until we have a valid adreno_smmu
> +	 * to avoid accidentially installing a GPU specific fault handler for
> +	 * the display's iommu
> +	 */
> +	iommu_set_fault_handler(iommu->domain, msm_fault_handler, iommu);
> +
>  	pagetable = kzalloc(sizeof(*pagetable), GFP_KERNEL);
>  	if (!pagetable)
>  		return ERR_PTR(-ENOMEM);
> @@ -300,7 +310,6 @@ struct msm_mmu *msm_iommu_new(struct device *dev, struct iommu_domain *domain)
>  
>  	iommu->domain = domain;
>  	msm_mmu_init(&iommu->base, dev, &funcs, MSM_MMU_IOMMU);
> -	iommu_set_fault_handler(domain, msm_fault_handler, iommu);
>  
>  	atomic_set(&iommu->pagetables, 0);
>  
> -- 
> 2.31.1
>