[PATCH 3/7] iommu/arm-smmu-v3: Shrink the strtab l1_desc array

[Mostafa Saleh]

Hi Jason,

On Mon, Jun 03, 2024 at 07:31:29PM -0300, Jason Gunthorpe wrote:
> The top of the 2 level stream table is (at most) 128k entries big, and two
> high order allocations are required. One of __le64 which is programmed
> into the HW (1M), and one of struct arm_smmu_strtab_l1_desc which holds
> the CPU pointer (3M).
> 
> There is no reason to store the l2ptr_dma as nothing reads it. devm stores
> a copy of it and the DMA memory will be freed via devm mechanisms. span is
> a constant of 8+1. Remove both.
> 
This caught my eye before, I imagine(although I was not there) there was some
thought about having different spans per SMMUs maybe, but that’s not the case.

> This removes 16 bytes from each arm_smmu_l1_ctx_desc and saves up to 2M of
> memory per iommu instance.
> 
> Signed-off-by: Jason Gunthorpe <jgg at nvidia.com>

Reviewed-by: Mostafa Saleh <smostafa at google.com>

> ---
>  drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 14 +++++++-------
>  drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h |  3 ---
>  2 files changed, 7 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> index d27dd0600bf1df..735dd9ff61890e 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> @@ -1448,12 +1448,12 @@ bool arm_smmu_free_asid(struct arm_smmu_ctx_desc *cd)
>  
>  /* Stream table manipulation functions */
>  static void
> -arm_smmu_write_strtab_l1_desc(__le64 *dst, struct arm_smmu_strtab_l1_desc *desc)
> +arm_smmu_write_strtab_l1_desc(__le64 *dst, dma_addr_t l2ptr_dma, u8 span)
>  {
>  	u64 val = 0;
>  
> -	val |= FIELD_PREP(STRTAB_L1_DESC_SPAN, desc->span);
> -	val |= desc->l2ptr_dma & STRTAB_L1_DESC_L2PTR_MASK;
> +	val |= FIELD_PREP(STRTAB_L1_DESC_SPAN, span);
> +	val |= l2ptr_dma & STRTAB_L1_DESC_L2PTR_MASK;
>  
>  	/* The HW has 64 bit atomicity with stores to the L2 STE table */
>  	WRITE_ONCE(*dst, cpu_to_le64(val));
> @@ -1655,6 +1655,7 @@ static int arm_smmu_init_l2_strtab(struct arm_smmu_device *smmu, u32 sid)
>  {
>  	size_t size;
>  	void *strtab;
> +	dma_addr_t l2ptr_dma;
>  	struct arm_smmu_strtab_cfg *cfg = &smmu->strtab_cfg;
>  	struct arm_smmu_strtab_l1_desc *desc = &cfg->l1_desc[sid >> STRTAB_SPLIT];
>  
> @@ -1664,9 +1665,8 @@ static int arm_smmu_init_l2_strtab(struct arm_smmu_device *smmu, u32 sid)
>  	size = (1 << STRTAB_SPLIT) * sizeof(struct arm_smmu_ste);
>  	strtab = &cfg->strtab.l1_desc[sid >> STRTAB_SPLIT];
>  
> -	desc->span = STRTAB_SPLIT + 1;
> -	desc->l2ptr = dmam_alloc_coherent(smmu->dev, size, &desc->l2ptr_dma,
> -					  GFP_KERNEL);
> +	desc->l2ptr =
> +		dmam_alloc_coherent(smmu->dev, size, &l2ptr_dma, GFP_KERNEL);
>  	if (!desc->l2ptr) {
>  		dev_err(smmu->dev,
>  			"failed to allocate l2 stream table for SID %u\n",
> @@ -1675,7 +1675,7 @@ static int arm_smmu_init_l2_strtab(struct arm_smmu_device *smmu, u32 sid)
>  	}
>  
>  	arm_smmu_init_initial_stes(desc->l2ptr, 1 << STRTAB_SPLIT);
> -	arm_smmu_write_strtab_l1_desc(strtab, desc);
> +	arm_smmu_write_strtab_l1_desc(strtab, l2ptr_dma, STRTAB_SPLIT + 1);
>  	return 0;
>  }
>  
> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> index 4769780259affc..280a04bfb7230c 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> @@ -577,10 +577,7 @@ struct arm_smmu_priq {
>  
>  /* High-level stream table and context descriptor structures */
>  struct arm_smmu_strtab_l1_desc {
> -	u8				span;
> -
>  	struct arm_smmu_ste		*l2ptr;
> -	dma_addr_t			l2ptr_dma;
>  };
>  
>  struct arm_smmu_ctx_desc {
> -- 
> 2.45.2
>