[PATCH v4 4/5] drm/amdgpu: address remove from fault filter

[Christian König]

Am 26.04.21 um 23:26 schrieb Philip Yang:
> Add interface to remove address from fault filter ring by resetting
> fault ring entry key, then future vm fault on the address will be
> processed to recover.
>
> Define fault key as atomic64_t type to use atomic read/set/cmpxchg key
> to protect fault ring access by interrupt handler and interrupt deferred
> work for vg20. Change fault->timestamp to 56-bit to share same uint64_t
> with 8-bit fault->next, it is big enough for 48bit IH timestamp.
>
> Signed-off-by: Philip Yang <Philip.Yang at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c | 54 +++++++++++++++++++++++--
>   drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h |  6 ++-
>   2 files changed, 55 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
> index c39ed9eb0987..888b749bd75e 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
> @@ -332,6 +332,17 @@ void amdgpu_gmc_agp_location(struct amdgpu_device *adev, struct amdgpu_gmc *mc)
>   			mc->agp_size >> 20, mc->agp_start, mc->agp_end);
>   }
>   
> +/**
> + * amdgpu_gmc_fault_key - get hask key from vm fault address and pasid
> + *
> + * @addr: 48bit physical address
> + * @pasid: 4 bit
> + */
> +static inline uint64_t amdgpu_gmc_fault_key(uint64_t addr, uint16_t pasid)
> +{
> +	return addr << 4 | pasid;
> +}
> +
>   /**
>    * amdgpu_gmc_filter_faults - filter VM faults
>    *
> @@ -349,13 +360,14 @@ bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev, uint64_t addr,
>   {
>   	struct amdgpu_gmc *gmc = &adev->gmc;
>   
> -	uint64_t stamp, key = addr << 4 | pasid;
> +	uint64_t stamp, key = amdgpu_gmc_fault_key(addr, pasid);
>   	struct amdgpu_gmc_fault *fault;
>   	uint32_t hash;
>   
>   	/* If we don't have space left in the ring buffer return immediately */
>   	stamp = max(timestamp, AMDGPU_GMC_FAULT_TIMEOUT + 1) -
>   		AMDGPU_GMC_FAULT_TIMEOUT;
> +
>   	if (gmc->fault_ring[gmc->last_fault].timestamp >= stamp)
>   		return true;
>   
> @@ -365,7 +377,7 @@ bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev, uint64_t addr,
>   	while (fault->timestamp >= stamp) {
>   		uint64_t tmp;
>   
> -		if (fault->key == key)
> +		if (atomic64_read(&fault->key) == key)
>   			return true;
>   
>   		tmp = fault->timestamp;
> @@ -378,7 +390,7 @@ bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev, uint64_t addr,
>   
>   	/* Add the fault to the ring */
>   	fault = &gmc->fault_ring[gmc->last_fault];
> -	fault->key = key;
> +	atomic64_set(&fault->key, key);
>   	fault->timestamp = timestamp;
>   
>   	/* And update the hash */
> @@ -387,6 +399,42 @@ bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev, uint64_t addr,
>   	return false;
>   }
>   
> +/**
> + * amdgpu_gmc_filter_faults_remove - remove address from VM faults filter
> + *
> + * @adev: amdgpu device structure
> + * @addr: address of the VM fault
> + * @pasid: PASID of the process causing the fault
> + *
> + * Remove the address from fault filter, then future vm fault on this address
> + * will pass to retry fault handler to recover.
> + */
> +void amdgpu_gmc_filter_faults_remove(struct amdgpu_device *adev, uint64_t addr,
> +				     uint16_t pasid)
> +{
> +	struct amdgpu_gmc *gmc = &adev->gmc;
> +
> +	uint64_t key = amdgpu_gmc_fault_key(addr, pasid);
> +	struct amdgpu_gmc_fault *fault;
> +	uint32_t hash;
> +
> +	hash = hash_64(key, AMDGPU_GMC_FAULT_HASH_ORDER);
> +	fault = &gmc->fault_ring[gmc->fault_hash[hash].idx];
> +	while (true) {
> +		uint64_t tmp;
> +
> +		if (atomic64_cmpxchg(&fault->key, key, 0) == key)
> +			break;
> +
> +		tmp = fault->timestamp;
> +		fault = &gmc->fault_ring[fault->next];
> +
> +		/* Check if the entry was reused */
> +		if (fault->timestamp >= tmp)
> +			break;
> +	}

Maybe rewrite this as "do { ... } while (fault->timestamp < tmp)".

> +}
> +
>   int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev)
>   {
>   	int r;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
> index 9d11c02a3938..95e18ef83aec 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
> @@ -66,9 +66,9 @@ struct firmware;
>    * GMC page fault information
>    */
>   struct amdgpu_gmc_fault {
> -	uint64_t	timestamp;
> +	uint64_t	timestamp:56;

I think 48 bits should be enough for the timestamp for current hardware.

Apart from that looks good to me now,
Christian.

>   	uint64_t	next:AMDGPU_GMC_FAULT_RING_ORDER;
> -	uint64_t	key:52;
> +	atomic64_t	key;
>   };
>   
>   /*
> @@ -318,6 +318,8 @@ void amdgpu_gmc_agp_location(struct amdgpu_device *adev,
>   			     struct amdgpu_gmc *mc);
>   bool amdgpu_gmc_filter_faults(struct amdgpu_device *adev, uint64_t addr,
>   			      uint16_t pasid, uint64_t timestamp);
> +void amdgpu_gmc_filter_faults_remove(struct amdgpu_device *adev, uint64_t addr,
> +				     uint16_t pasid);
>   int amdgpu_gmc_ras_late_init(struct amdgpu_device *adev);
>   void amdgpu_gmc_ras_fini(struct amdgpu_device *adev);
>   int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev);