[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20250515150629.GD12165@willie-the-truck>
Date: Thu, 15 May 2025 16:06:32 +0100
From: Will Deacon <will@...nel.org>
To: Nicolin Chen <nicolinc@...dia.com>
Cc: robin.murphy@....com, jgg@...dia.com, joro@...tes.org,
kevin.tian@...el.com, praan@...gle.com, nathan@...nel.org,
yi.l.liu@...el.com, peterz@...radead.org, mshavit@...gle.com,
jsnitsel@...hat.com, smostafa@...gle.com,
jeff.johnson@....qualcomm.com, zhangzekun11@...wei.com,
linux-arm-kernel@...ts.infradead.org, iommu@...ts.linux.dev,
linux-kernel@...r.kernel.org, shameerali.kolothum.thodi@...wei.com
Subject: Re: [PATCH v2 02/11] iommu/arm-smmu-v3: Pass in smmu/iommu_domain to
__arm_smmu_tlb_inv_range()
On Mon, Apr 14, 2025 at 09:57:37PM -0700, Nicolin Chen wrote:
> What __arm_smmu_tlb_inv_range() really needs is the smmu and iommu_domain
> pointers from the smmu_domain.
>
> For a nest_parent smmu_domain, it will no longer store an smmu pointer as
> it can be shared across vSMMU instances. A vSMMU structure sharing the S2
> smmu_domain instead would hold the smmu pointer.
>
> Pass them in explicitly to fit both !nest_parent and nest_parent cases.
>
> While changing it, share it in the header with arm-smmu-v3-iommmufd that
> will call it too.
>
> Signed-off-by: Nicolin Chen <nicolinc@...dia.com>
> ---
> drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 4 ++++
> drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 17 +++++++++--------
> 2 files changed, 13 insertions(+), 8 deletions(-)
>
> 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 d4837a33fb81..5dbdc61558a9 100644
> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> @@ -955,6 +955,10 @@ void arm_smmu_tlb_inv_range_asid(unsigned long iova, size_t size, int asid,
> struct arm_smmu_domain *smmu_domain);
> int arm_smmu_atc_inv_domain(struct arm_smmu_domain *smmu_domain,
> unsigned long iova, size_t size);
> +void __arm_smmu_tlb_inv_range(struct arm_smmu_device *smmu,
> + struct arm_smmu_cmdq_ent *cmd, unsigned long iova,
> + size_t size, size_t granule,
> + struct iommu_domain *domain);
I don't think this function makes a particularly good "public" API --
the caller even sets the cmd opcode!
Can we expose some TLB invalidation helpers instead rather than the
low-level helpers?
Will
Powered by blists - more mailing lists