2020-02-12 20:23:06

by Rob Herring (Arm)

[permalink] [raw]
Subject: [PATCH v2] drm/panfrost: Don't try to map on error faults

From: Tomeu Vizoso <[email protected]>

If the exception type isn't a translation fault, don't try to map and
instead go straight to a terminal fault.

Otherwise, we can get flooded by kernel warnings and further faults.

Signed-off-by: Tomeu Vizoso <[email protected]>
Signed-off-by: Rob Herring <[email protected]>
---
I rewrote this some simplifying the code and somewhat following Steven's
suggested. Still not using defines though. No defines here was good
enough before IMO.

Only compile tested.

drivers/gpu/drm/panfrost/panfrost_mmu.c | 44 +++++++++++--------------
1 file changed, 19 insertions(+), 25 deletions(-)

diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c
index 763cfca886a7..4f2836bd9215 100644
--- a/drivers/gpu/drm/panfrost/panfrost_mmu.c
+++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c
@@ -596,33 +596,27 @@ static irqreturn_t panfrost_mmu_irq_handler_thread(int irq, void *data)
source_id = (fault_status >> 16);

/* Page fault only */
- if ((status & mask) == BIT(i)) {
- WARN_ON(exception_type < 0xC1 || exception_type > 0xC4);
-
+ ret = -1;
+ if ((status & mask) == BIT(i) && (exception_type & 0xF8) == 0xC0)
ret = panfrost_mmu_map_fault_addr(pfdev, i, addr);
- if (!ret) {
- mmu_write(pfdev, MMU_INT_CLEAR, BIT(i));
- status &= ~mask;
- continue;
- }
- }

- /* terminal fault, print info about the fault */
- dev_err(pfdev->dev,
- "Unhandled Page fault in AS%d at VA 0x%016llX\n"
- "Reason: %s\n"
- "raw fault status: 0x%X\n"
- "decoded fault status: %s\n"
- "exception type 0x%X: %s\n"
- "access type 0x%X: %s\n"
- "source id 0x%X\n",
- i, addr,
- "TODO",
- fault_status,
- (fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
- exception_type, panfrost_exception_name(pfdev, exception_type),
- access_type, access_type_name(pfdev, fault_status),
- source_id);
+ if (ret)
+ /* terminal fault, print info about the fault */
+ dev_err(pfdev->dev,
+ "Unhandled Page fault in AS%d at VA 0x%016llX\n"
+ "Reason: %s\n"
+ "raw fault status: 0x%X\n"
+ "decoded fault status: %s\n"
+ "exception type 0x%X: %s\n"
+ "access type 0x%X: %s\n"
+ "source id 0x%X\n",
+ i, addr,
+ "TODO",
+ fault_status,
+ (fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
+ exception_type, panfrost_exception_name(pfdev, exception_type),
+ access_type, access_type_name(pfdev, fault_status),
+ source_id);

mmu_write(pfdev, MMU_INT_CLEAR, mask);

--
2.20.1


2020-02-13 11:14:58

by Steven Price

[permalink] [raw]
Subject: Re: [PATCH v2] drm/panfrost: Don't try to map on error faults

On 12/02/2020 20:22, Rob Herring wrote:
> From: Tomeu Vizoso <[email protected]>
>
> If the exception type isn't a translation fault, don't try to map and
> instead go straight to a terminal fault.
>
> Otherwise, we can get flooded by kernel warnings and further faults.
>
> Signed-off-by: Tomeu Vizoso <[email protected]>
> Signed-off-by: Rob Herring <[email protected]>
> ---
> I rewrote this some simplifying the code and somewhat following Steven's
> suggested. Still not using defines though. No defines here was good
> enough before IMO.

Heresy! It's a good thing you're not writing kbase code - there you
(seemingly) need to pick a #define which is as long as possible, but
then still wrap the code to avoid the 80 character limit... For
example[1]. Although shifting the code right might get you extra bonus
points, deep nesting seems to be encouraged! ;)

[1]
https://gitlab.freedesktop.org/panfrost/mali_kbase/blob/master/driver/product/kernel/drivers/gpu/arm/midgard/backend/gpu/mali_kbase_js_backend.c#L156

>
> Only compile tested.

I've done some quick testing on a Firefly RK3288. But I don't have any
(handy) tests to trigger non-TRANSLATION_FAULT MMU faults.

Reviewed-by: Steven Price <[email protected]>

Thanks,

Steve

>
> drivers/gpu/drm/panfrost/panfrost_mmu.c | 44 +++++++++++--------------
> 1 file changed, 19 insertions(+), 25 deletions(-)
>
> diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c
> index 763cfca886a7..4f2836bd9215 100644
> --- a/drivers/gpu/drm/panfrost/panfrost_mmu.c
> +++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c
> @@ -596,33 +596,27 @@ static irqreturn_t panfrost_mmu_irq_handler_thread(int irq, void *data)
> source_id = (fault_status >> 16);
>
> /* Page fault only */
> - if ((status & mask) == BIT(i)) {
> - WARN_ON(exception_type < 0xC1 || exception_type > 0xC4);
> -
> + ret = -1;
> + if ((status & mask) == BIT(i) && (exception_type & 0xF8) == 0xC0)
> ret = panfrost_mmu_map_fault_addr(pfdev, i, addr);
> - if (!ret) {
> - mmu_write(pfdev, MMU_INT_CLEAR, BIT(i));
> - status &= ~mask;
> - continue;
> - }
> - }
>
> - /* terminal fault, print info about the fault */
> - dev_err(pfdev->dev,
> - "Unhandled Page fault in AS%d at VA 0x%016llX\n"
> - "Reason: %s\n"
> - "raw fault status: 0x%X\n"
> - "decoded fault status: %s\n"
> - "exception type 0x%X: %s\n"
> - "access type 0x%X: %s\n"
> - "source id 0x%X\n",
> - i, addr,
> - "TODO",
> - fault_status,
> - (fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
> - exception_type, panfrost_exception_name(pfdev, exception_type),
> - access_type, access_type_name(pfdev, fault_status),
> - source_id);
> + if (ret)
> + /* terminal fault, print info about the fault */
> + dev_err(pfdev->dev,
> + "Unhandled Page fault in AS%d at VA 0x%016llX\n"
> + "Reason: %s\n"
> + "raw fault status: 0x%X\n"
> + "decoded fault status: %s\n"
> + "exception type 0x%X: %s\n"
> + "access type 0x%X: %s\n"
> + "source id 0x%X\n",
> + i, addr,
> + "TODO",
> + fault_status,
> + (fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
> + exception_type, panfrost_exception_name(pfdev, exception_type),
> + access_type, access_type_name(pfdev, fault_status),
> + source_id);
>
> mmu_write(pfdev, MMU_INT_CLEAR, mask);
>
>

2020-02-24 07:24:04

by Tomeu Vizoso

[permalink] [raw]
Subject: Re: [PATCH v2] drm/panfrost: Don't try to map on error faults

On Wed, 12 Feb 2020 at 21:22, Rob Herring <[email protected]> wrote:
>
> From: Tomeu Vizoso <[email protected]>
>
> If the exception type isn't a translation fault, don't try to map and
> instead go straight to a terminal fault.
>
> Otherwise, we can get flooded by kernel warnings and further faults.
>
> Signed-off-by: Tomeu Vizoso <[email protected]>
> Signed-off-by: Rob Herring <[email protected]>
> ---
> I rewrote this some simplifying the code and somewhat following Steven's
> suggested. Still not using defines though. No defines here was good
> enough before IMO.
>
> Only compile tested.

Looks good, I also tested it on RK3399.

Reviewed-by: Tomeu Vizoso <[email protected]>

Thanks,

Tomeu

> drivers/gpu/drm/panfrost/panfrost_mmu.c | 44 +++++++++++--------------
> 1 file changed, 19 insertions(+), 25 deletions(-)
>
> diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c
> index 763cfca886a7..4f2836bd9215 100644
> --- a/drivers/gpu/drm/panfrost/panfrost_mmu.c
> +++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c
> @@ -596,33 +596,27 @@ static irqreturn_t panfrost_mmu_irq_handler_thread(int irq, void *data)
> source_id = (fault_status >> 16);
>
> /* Page fault only */
> - if ((status & mask) == BIT(i)) {
> - WARN_ON(exception_type < 0xC1 || exception_type > 0xC4);
> -
> + ret = -1;
> + if ((status & mask) == BIT(i) && (exception_type & 0xF8) == 0xC0)
> ret = panfrost_mmu_map_fault_addr(pfdev, i, addr);
> - if (!ret) {
> - mmu_write(pfdev, MMU_INT_CLEAR, BIT(i));
> - status &= ~mask;
> - continue;
> - }
> - }
>
> - /* terminal fault, print info about the fault */
> - dev_err(pfdev->dev,
> - "Unhandled Page fault in AS%d at VA 0x%016llX\n"
> - "Reason: %s\n"
> - "raw fault status: 0x%X\n"
> - "decoded fault status: %s\n"
> - "exception type 0x%X: %s\n"
> - "access type 0x%X: %s\n"
> - "source id 0x%X\n",
> - i, addr,
> - "TODO",
> - fault_status,
> - (fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
> - exception_type, panfrost_exception_name(pfdev, exception_type),
> - access_type, access_type_name(pfdev, fault_status),
> - source_id);
> + if (ret)
> + /* terminal fault, print info about the fault */
> + dev_err(pfdev->dev,
> + "Unhandled Page fault in AS%d at VA 0x%016llX\n"
> + "Reason: %s\n"
> + "raw fault status: 0x%X\n"
> + "decoded fault status: %s\n"
> + "exception type 0x%X: %s\n"
> + "access type 0x%X: %s\n"
> + "source id 0x%X\n",
> + i, addr,
> + "TODO",
> + fault_status,
> + (fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
> + exception_type, panfrost_exception_name(pfdev, exception_type),
> + access_type, access_type_name(pfdev, fault_status),
> + source_id);
>
> mmu_write(pfdev, MMU_INT_CLEAR, mask);
>
> --
> 2.20.1
>
> _______________________________________________
> dri-devel mailing list
> [email protected]
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

2020-02-24 12:28:34

by Alyssa Rosenzweig

[permalink] [raw]
Subject: Re: [PATCH v2] drm/panfrost: Don't try to map on error faults

Acked-by: Alyssa Rosenzweig <[email protected]>

On Wed, Feb 12, 2020 at 02:22:36PM -0600, Rob Herring wrote:
> From: Tomeu Vizoso <[email protected]>
>
> If the exception type isn't a translation fault, don't try to map and
> instead go straight to a terminal fault.
>
> Otherwise, we can get flooded by kernel warnings and further faults.
>
> Signed-off-by: Tomeu Vizoso <[email protected]>
> Signed-off-by: Rob Herring <[email protected]>
> ---
> I rewrote this some simplifying the code and somewhat following Steven's
> suggested. Still not using defines though. No defines here was good
> enough before IMO.
>
> Only compile tested.
>
> drivers/gpu/drm/panfrost/panfrost_mmu.c | 44 +++++++++++--------------
> 1 file changed, 19 insertions(+), 25 deletions(-)
>
> diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c
> index 763cfca886a7..4f2836bd9215 100644
> --- a/drivers/gpu/drm/panfrost/panfrost_mmu.c
> +++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c
> @@ -596,33 +596,27 @@ static irqreturn_t panfrost_mmu_irq_handler_thread(int irq, void *data)
> source_id = (fault_status >> 16);
>
> /* Page fault only */
> - if ((status & mask) == BIT(i)) {
> - WARN_ON(exception_type < 0xC1 || exception_type > 0xC4);
> -
> + ret = -1;
> + if ((status & mask) == BIT(i) && (exception_type & 0xF8) == 0xC0)
> ret = panfrost_mmu_map_fault_addr(pfdev, i, addr);
> - if (!ret) {
> - mmu_write(pfdev, MMU_INT_CLEAR, BIT(i));
> - status &= ~mask;
> - continue;
> - }
> - }
>
> - /* terminal fault, print info about the fault */
> - dev_err(pfdev->dev,
> - "Unhandled Page fault in AS%d at VA 0x%016llX\n"
> - "Reason: %s\n"
> - "raw fault status: 0x%X\n"
> - "decoded fault status: %s\n"
> - "exception type 0x%X: %s\n"
> - "access type 0x%X: %s\n"
> - "source id 0x%X\n",
> - i, addr,
> - "TODO",
> - fault_status,
> - (fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
> - exception_type, panfrost_exception_name(pfdev, exception_type),
> - access_type, access_type_name(pfdev, fault_status),
> - source_id);
> + if (ret)
> + /* terminal fault, print info about the fault */
> + dev_err(pfdev->dev,
> + "Unhandled Page fault in AS%d at VA 0x%016llX\n"
> + "Reason: %s\n"
> + "raw fault status: 0x%X\n"
> + "decoded fault status: %s\n"
> + "exception type 0x%X: %s\n"
> + "access type 0x%X: %s\n"
> + "source id 0x%X\n",
> + i, addr,
> + "TODO",
> + fault_status,
> + (fault_status & (1 << 10) ? "DECODER FAULT" : "SLAVE FAULT"),
> + exception_type, panfrost_exception_name(pfdev, exception_type),
> + access_type, access_type_name(pfdev, fault_status),
> + source_id);
>
> mmu_write(pfdev, MMU_INT_CLEAR, mask);
>
> --
> 2.20.1
>


Attachments:
(No filename) (2.85 kB)
signature.asc (849.00 B)
Download all attachments