iommu/arm-smmu: Enable split pagetable support
Subsystems (8)
v9 cover iommu/arm-smmu: Enable split pagetable support
Another iteration of the split-pagetable support for arm-smmu and the Adreno GPU SMMU. After email discussions [1] we opted to make a arm-smmu implementation for specifically for the Adreno GPU and use that to enable split pagetable support and later other implementation specific bits that we need.
On the hardware side this is very close to the same code from before [2] only the TTBR1 quirk is turned on by the implementation and not a domain attribute. In drm/msm we use the returned size of the aperture as a clue to let us know which virtual address space we should use for global memory objects.
There are two open items that you should be aware of. First, in the implementation specific code we have to check the compatible string of the device so that we only enable TTBR1 for the GPU (SID 0) and not the GMU (SID 4). I went back and forth trying to decide if I wanted to use the compatible string or the SID as the filter and settled on the compatible string but I could be talked out of it.
The other open item is that in drm/msm the hardware only uses 49 bits of the address space but arm-smmu expects the address to be sign extended all the way to 64 bits. This isn't a problem normally unless you look at the hardware registers that contain a IOVA and then the upper bits will be zero. I opted to restrict the internal drm/msm IOVA range to only 49 bits and then sign extend right before calling iommu_map / iommu_unmap. This is a bit wonky but I thought that matching the hardware would be less confusing when debugging a hang.
v9: Fix bot-detected merge conflict
v7: Add attached device to smmu_domain to pass to implementation specific functions
[1] https://lists.linuxfoundation.org/pipermail/iommu/2020-May/044537.html
[2] https://patchwork.kernel.org/patch/11482591/
Jordan Crouse (7):
iommu/arm-smmu: Pass io-pgtable config to implementation specific
function
iommu/arm-smmu: Add support for split pagetables
dt-bindings: arm-smmu: Add compatible string for Adreno GPU SMMU iommu/arm-smmu: Add a pointer to the attached device to smmu_domain iommu/arm-smmu: Add implementation for the adreno GPU SMMU drm/msm: Set the global virtual address range from the IOMMU domain
arm: dts: qcom: sm845: Set the compatible string for the GPU SMMU
.../devicetree/bindings/iommu/arm,smmu.yaml | 4 ++ arch/arm64/boot/dts/qcom/sdm845.dtsi | 2 +- drivers/gpu/drm/msm/adreno/adreno_gpu.c | 13 +++++- drivers/gpu/drm/msm/msm_iommu.c | 7 +++ drivers/iommu/arm-smmu-impl.c | 6 ++- drivers/iommu/arm-smmu-qcom.c | 45 ++++++++++++++++++- drivers/iommu/arm-smmu.c | 38 +++++++++++----- drivers/iommu/arm-smmu.h | 30 ++++++++++--- 8 files changed, 120 insertions(+), 25 deletions(-)
Signature · 1 line
2.17.1
PATCH v9 1/7 iommu/arm-smmu: Pass io-pgtable config to implementation specific function
Construct the io-pgtable config before calling the implementation specific init_context function and pass it so the implementation specific function can get a chance to change it before the io-pgtable is created.
---
drivers/iommu/arm-smmu-impl.c | 3 ++- drivers/iommu/arm-smmu.c | 11 ++++++----- drivers/iommu/arm-smmu.h | 3 ++- 3 files changed, 10 insertions(+), 7 deletions(-)
drivers/iommu/arm-smmu-impl.c +2 -1
@@ -68,7 +68,8 @@ static int cavium_cfg_probe(struct arm_smmu_device *smmu)return 0 ; } -static int cavium_init_context (struct arm_smmu_domain *smmu_domain) +static int cavium_init_context (struct arm_smmu_domain *smmu_domain, +struct io_pgtable_cfg *pgtbl_cfg) {struct cavium_smmu *cs =container_of (smmu_domain->smmu, struct cavium_smmu, smmu);
drivers/iommu/arm-smmu.c +6 -5
@@ -797,11 +797,6 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain, cfg->asid = cfg->cbndx; smmu_domain->smmu = smmu; -if (smmu->impl && smmu->impl->init_context) { - ret = smmu->impl->init_context (smmu_domain); -if (ret) - goto out_unlock; - } pgtbl_cfg = (struct io_pgtable_cfg ) { .pgsize_bitmap = smmu->pgsize_bitmap,
@@ -812,6 +807,12 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain, .iommu_dev = smmu->dev, }; +if (smmu->impl && smmu->impl->init_context) { + ret = smmu->impl->init_context (smmu_domain, &pgtbl_cfg); +if (ret) + goto out_unlock; + } +if (smmu_domain->non_strict) pgtbl_cfg.quirks |=IO_PGTABLE_QUIRK_NON_STRICT ;
drivers/iommu/arm-smmu.h +2 -1
@@ -383,7 +383,8 @@ struct arm_smmu_impl { u64 val);int (*cfg_probe)(struct arm_smmu_device *smmu);int (*reset)(struct arm_smmu_device *smmu); -int (*init_context)(struct arm_smmu_domain *smmu_domain); +int (*init_context)(struct arm_smmu_domain *smmu_domain, +struct io_pgtable_cfg *cfg);void (*tlb_sync)(struct arm_smmu_device *smmu,int page,int sync,int status);int (*def_domain_type)(struct device *dev);
Signature · 6 lines
2.17.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
PATCH v9 2/7 iommu/arm-smmu: Add support for split pagetables
In reply to Jordan CrouseEnable TTBR1 for a context bank if IO_PGTABLE_QUIRK_ARM_TTBR1 is selected by the io-pgtable configuration.
---
drivers/iommu/arm-smmu.c | 21 ++++++++++++++++----- drivers/iommu/arm-smmu.h | 25 +++++++++++++++++++------ 2 files changed, 35 insertions(+), 11 deletions(-)
drivers/iommu/arm-smmu.c +16 -5
@@ -555,11 +555,15 @@ static void arm_smmu_init_context_bank(struct arm_smmu_domain *smmu_domain, cb->ttbr[0 ] = pgtbl_cfg->arm_v7s_cfg.ttbr; cb->ttbr[1 ] =0 ; }else { - cb->ttbr[0 ] = pgtbl_cfg->arm_lpae_s1_cfg.ttbr; - cb->ttbr[0 ] |=FIELD_PREP (ARM_SMMU_TTBRn_ASID, - cfg->asid); + cb->ttbr[0 ] =FIELD_PREP (ARM_SMMU_TTBRn_ASID, + cfg->asid); cb->ttbr[1 ] =FIELD_PREP (ARM_SMMU_TTBRn_ASID, - cfg->asid); + cfg->asid); + +if (pgtbl_cfg->quirks &IO_PGTABLE_QUIRK_ARM_TTBR1 ) + cb->ttbr[1 ] |= pgtbl_cfg->arm_lpae_s1_cfg.ttbr; +else + cb->ttbr[0 ] |= pgtbl_cfg->arm_lpae_s1_cfg.ttbr; } }else { cb->ttbr[0 ] = pgtbl_cfg->arm_lpae_s2_cfg.vttbr;
@@ -824,7 +828,14 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,/* Update the domain's page sizes to reflect the page table format */ domain->pgsize_bitmap = pgtbl_cfg.pgsize_bitmap; - domain->geometry.aperture_end = (1UL << ias) -1 ; + +if (pgtbl_cfg.quirks &IO_PGTABLE_QUIRK_ARM_TTBR1 ) { + domain->geometry.aperture_start = ~0UL << ias; + domain->geometry.aperture_end = ~0UL ; + }else { + domain->geometry.aperture_end = (1UL << ias) -1 ; + } + domain->geometry.force_aperture = true;/* Initialise the context bank with our page table cfg */
drivers/iommu/arm-smmu.h +19 -6
@@ -168,10 +168,12 @@ enum arm_smmu_cbar_type {#define ARM_SMMU_CB_TCR 0x30#define ARM_SMMU_TCR_EAE BIT(31)#define ARM_SMMU_TCR_EPD1 BIT(23) +#define ARM_SMMU_TCR_A1 BIT(22)#define ARM_SMMU_TCR_TG0 GENMASK(15, 14)#define ARM_SMMU_TCR_SH0 GENMASK(13, 12)#define ARM_SMMU_TCR_ORGN0 GENMASK(11, 10)#define ARM_SMMU_TCR_IRGN0 GENMASK(9, 8) +#define ARM_SMMU_TCR_EPD0 BIT(7)#define ARM_SMMU_TCR_T0SZ GENMASK(5, 0)#define ARM_SMMU_VTCR_RES1 BIT(31)
@@ -347,12 +349,23 @@ struct arm_smmu_domain {static inline u32 arm_smmu_lpae_tcr (struct io_pgtable_cfg *cfg) { -return ARM_SMMU_TCR_EPD1 | -FIELD_PREP (ARM_SMMU_TCR_TG0 , cfg->arm_lpae_s1_cfg.tcr.tg) | -FIELD_PREP (ARM_SMMU_TCR_SH0 , cfg->arm_lpae_s1_cfg.tcr.sh) | -FIELD_PREP (ARM_SMMU_TCR_ORGN0 , cfg->arm_lpae_s1_cfg.tcr.orgn) | -FIELD_PREP (ARM_SMMU_TCR_IRGN0 , cfg->arm_lpae_s1_cfg.tcr.irgn) | -FIELD_PREP (ARM_SMMU_TCR_T0SZ , cfg->arm_lpae_s1_cfg.tcr.tsz); +u32 tcr =FIELD_PREP (ARM_SMMU_TCR_TG0 , cfg->arm_lpae_s1_cfg.tcr.tg) | +FIELD_PREP (ARM_SMMU_TCR_SH0 , cfg->arm_lpae_s1_cfg.tcr.sh) | +FIELD_PREP (ARM_SMMU_TCR_ORGN0 , cfg->arm_lpae_s1_cfg.tcr.orgn) | +FIELD_PREP (ARM_SMMU_TCR_IRGN0 , cfg->arm_lpae_s1_cfg.tcr.irgn) | +FIELD_PREP (ARM_SMMU_TCR_T0SZ , cfg->arm_lpae_s1_cfg.tcr.tsz); + +/* +* When TTBR1 is selected shift the TCR fields by 16 bits and disable +* translation in TTBR0 +*/ +if (cfg->quirks &IO_PGTABLE_QUIRK_ARM_TTBR1 ) { + tcr = (tcr <<16 ) & ~ARM_SMMU_TCR_A1 ; + tcr |=ARM_SMMU_TCR_EPD0 ; + }else + tcr |=ARM_SMMU_TCR_EPD1 ; + +return tcr; }static inline u32 arm_smmu_lpae_tcr2 (struct io_pgtable_cfg *cfg)
Signature · 1 line
2.17.1
PATCH v9 7/7 arm: dts: qcom: sm845: Set the compatible string for the GPU SMMU
In reply to Jordan CrouseSet the qcom,adreno-smmu compatible string for the GPU SMMU to enable split pagetables.
---
arch/arm64/boot/dts/qcom/sdm845.dtsi | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm64/boot/dts/qcom/sdm845.dtsi +1 -1
@@ -3556,7 +3556,7 @@ }; adreno_smmu: iommu@5040000 { - compatible = "qcom,sdm845-smmu-v2", "qcom,smmu-v2"; + compatible = "qcom,adreno-smmu", "qcom,smmu-v2"; reg = <0 0x5040000 0 0x10000>; #iommu-cells = <1>; #global-interrupts = <2>;
Signature · 1 line
2.17.1
PATCH v9 6/7 drm/msm: Set the global virtual address range from the IOMMU domain
In reply to Jordan CrouseUse the aperture settings from the IOMMU domain to set up the virtual address range for the GPU. This allows us to transparently deal with IOMMU side features (like split pagetables).
---
drivers/gpu/drm/msm/adreno/adreno_gpu.c | 13 +++++++++++-- drivers/gpu/drm/msm/msm_iommu.c | 7 +++++++ 2 files changed, 18 insertions(+), 2 deletions(-)
drivers/gpu/drm/msm/adreno/adreno_gpu.c +11 -2
@@ -192,9 +192,18 @@ adreno_iommu_create_address_space(struct msm_gpu *gpu,struct iommu_domain *iommu =iommu_domain_alloc (&platform_bus_type);struct msm_mmu *mmu =msm_iommu_new (&pdev->dev, iommu);struct msm_gem_address_space *aspace; +u64 start, size; - aspace =msm_gem_address_space_create (mmu,"gpu" ,SZ_16M , -0xffffffff -SZ_16M ); +/* +* Use the aperture start or SZ_16M, whichever is greater. This will +* ensure that we align with the allocated pagetable range while still +* allowing room in the lower 32 bits for GMEM and whatnot +*/ + start =max_t (u64,SZ_16M , iommu->geometry.aperture_start); + size = iommu->geometry.aperture_end - start +1 ; + + aspace =msm_gem_address_space_create (mmu,"gpu" , + start &GENMASK (48 ,0 ), size);if (IS_ERR (aspace) && !IS_ERR (mmu)) mmu->funcs->destroy (mmu);
drivers/gpu/drm/msm/msm_iommu.c +7 -0
@@ -36,6 +36,10 @@ static int msm_iommu_map(struct msm_mmu *mmu, uint64_t iova,struct msm_iommu *iommu =to_msm_iommu (mmu);size_t ret; +/* The arm-smmu driver expects the addresses to be sign extended */ +if (iova &BIT_ULL (48 )) + iova |=GENMASK_ULL (63 ,49 ); + ret =iommu_map_sg (iommu->domain, iova, sgt->sgl, sgt->nents, prot);WARN_ON (!ret);
@@ -46,6 +50,9 @@ static int msm_iommu_unmap(struct msm_mmu *mmu, uint64_t iova, size_t len) {struct msm_iommu *iommu =to_msm_iommu (mmu); +if (iova &BIT_ULL (48 )) + iova |=GENMASK_ULL (63 ,49 ); +iommu_unmap (iommu->domain, iova, len);return 0 ;
Signature · 1 line
2.17.1
PATCH v9 4/7 iommu/arm-smmu: Add a pointer to the attached device to smmu_domain
In reply to Jordan CrouseAdd a link to the pointer to the struct device that is attached to a domain. This makes it easy to get the pointer if it is needed in the implementation specific code.
---
drivers/iommu/arm-smmu.c | 6 ++++-- drivers/iommu/arm-smmu.h | 1 + 2 files changed, 5 insertions(+), 2 deletions(-)
drivers/iommu/arm-smmu.c +4 -2
@@ -668,7 +668,8 @@ static void arm_smmu_write_context_bank(struct arm_smmu_device *smmu, int idx) }static int arm_smmu_init_domain_context (struct iommu_domain *domain, -struct arm_smmu_device *smmu) +struct arm_smmu_device *smmu, +struct device *dev) {int irq, start, ret =0 ;unsigned long ias, oas;
@@ -801,6 +802,7 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain, cfg->asid = cfg->cbndx; smmu_domain->smmu = smmu; + smmu_domain->dev = dev; pgtbl_cfg = (struct io_pgtable_cfg ) { .pgsize_bitmap = smmu->pgsize_bitmap,
@@ -1190,7 +1192,7 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev)return ret;/* Ensure that the domain is finalised */ - ret =arm_smmu_init_domain_context (domain, smmu); + ret =arm_smmu_init_domain_context (domain, smmu, dev);if (ret <0 ) goto rpm_put;
drivers/iommu/arm-smmu.h +1 -0
@@ -345,6 +345,7 @@ struct arm_smmu_domain {struct mutex init_mutex;/* Protects smmu pointer */ spinlock_t cb_lock;/* Serialises ATS1* ops and TLB syncs */ struct iommu_domain domain; +struct device *dev;/* Device attached to this domain */ };static inline u32 arm_smmu_lpae_tcr (struct io_pgtable_cfg *cfg)
Signature · 1 line
2.17.1
PATCH v9 3/7 dt-bindings: arm-smmu: Add compatible string for Adreno GPU SMMU
In reply to Jordan CrouseEvery Qcom Adreno GPU has an embedded SMMU for its own use. These devices depend on unique features such as split pagetables, different stall/halt requirements and other settings. Identify them with a compatible string so that they can be identified in the arm-smmu implementation specific code.
---
Documentation/devicetree/bindings/iommu/arm,smmu.yaml | 4 ++++ 1 file changed, 4 insertions(+)
Documentation/devicetree/bindings/iommu/arm,smmu.yaml +4 -0
@@ -38,6 +38,10 @@ properties: - qcom,sc7180-smmu-500 - qcom,sdm845-smmu-500 - const: arm,mmu-500 + - description: Qcom Adreno GPUs implementing "arm,smmu-v2" + items: + - const: qcom,adreno-smmu + - const: qcom,smmu-v2 - items: - const: arm,mmu-500 - const: arm,smmu-v2
Signature · 1 line
2.17.1
PATCH v9 5/7 iommu/arm-smmu: Add implementation for the adreno GPU SMMU
In reply to Jordan CrouseAdd a special implementation for the SMMU attached to most Adreno GPU target triggered from the qcom,adreno-gpu-smmu compatible string. When selected the driver will attempt to enable split pagetables.
---
drivers/iommu/arm-smmu-impl.c | 3 +++ drivers/iommu/arm-smmu-qcom.c | 45 +++++++++++++++++++++++++++++++++-- drivers/iommu/arm-smmu.h | 1 + 3 files changed, 47 insertions(+), 2 deletions(-)
drivers/iommu/arm-smmu-impl.c +3 -0
@@ -176,5 +176,8 @@ struct arm_smmu_device *arm_smmu_impl_init(struct arm_smmu_device *smmu) of_device_is_compatible(np, "qcom,sc7180-smmu-500"))return qcom_smmu_impl_init (smmu); +if (of_device_is_compatible (smmu->dev->of_node,"qcom,adreno-smmu" )) +return qcom_adreno_smmu_impl_init (smmu); +return smmu; }
drivers/iommu/arm-smmu-qcom.c +43 -2
64 lines not on this page · raw message
drivers/iommu/arm-smmu.h +1 -0
@@ -466,6 +466,7 @@ static inline void arm_smmu_writeq(struct arm_smmu_device *smmu, int page,struct arm_smmu_device *arm_smmu_impl_init (struct arm_smmu_device *smmu);struct arm_smmu_device *qcom_smmu_impl_init (struct arm_smmu_device *smmu); +struct arm_smmu_device *qcom_adreno_smmu_impl_init (struct arm_smmu_device *smmu);int arm_mmu500_reset (struct arm_smmu_device *smmu);
Signature · 1 line
2.17.1
Re: [Freedreno] [PATCH v9 6/7] drm/msm: Set the global virtual address range from the IOMMU domain
In reply to Jordan CrouseQuoted 24 lines · Jordan Crouse
Use the aperture settings from the IOMMU domain to set up the virtual address range for the GPU. This allows us to transparently deal with IOMMU side features (like split pagetables).
Signed-off-by Jordan Crouse <[email protected]>---
drivers/gpu/drm/msm/adreno/adreno_gpu.c | 13 +++++++++++-- drivers/gpu/drm/msm/msm_iommu.c | 7 +++++++ 2 files changed, 18 insertions(+), 2 deletions(-)drivers/gpu/drm/msm/adreno/adreno_gpu.c +1
-2@@ -192,9 +192,18 @@ adreno_iommu_create_address_space(struct msm_gpu *gpu,struct iommu_domain *iommu =iommu_domain_alloc (&platform_bus_type);struct msm_mmu *mmu =msm_iommu_new (&pdev->dev, iommu);struct msm_gem_address_space *aspace; +u64 start, size; - aspace =msm_gem_address_space_create (mmu,"gpu" ,SZ_16M , -0xffffffff -SZ_16M );
drivers/gpu/drm/msm/adreno/adreno_gpu.c adreno_iommu_create_address_space(struct msm_gpu *gpu, · line 206
Quoted 10 lines
drivers/gpu/drm/msm/adreno/adreno_gpu.c +10
-0+/* +* Use the aperture start or SZ_16M, whichever is greater. This will +* ensure that we align with the allocated pagetable range while still +* allowing room in the lower 32 bits for GMEM and whatnot +*/ + start =max_t (u64,SZ_16M , iommu->geometry.aperture_start); + size = iommu->geometry.aperture_end - start +1 ; + + aspace =msm_gem_address_space_create (mmu,"gpu" , + start &GENMASK (48 ,0 ), size);
hmm, I kinda think this isn't going to play well for the 32b gpus (pre-a5xx).. possibly we should add address space size to 'struct adreno_info'?
Or I guess it is always going to be the same for all devices within a generation? So it could just be passed in to adreno_gpu_init()
Hopefully that makes things smoother if we someday had more than 48bits..
BR,
-R
Quoted 31 lines
drivers/gpu/drm/msm/adreno/adreno_gpu.c +0
-0if (IS_ERR (aspace) && !IS_ERR (mmu)) mmu->funcs->destroy (mmu);drivers/gpu/drm/msm/msm_iommu.c +7
-1@@ -36,6 +36,10 @@ static int msm_iommu_map(struct msm_mmu *mmu, uint64_t iova,struct msm_iommu *iommu =to_msm_iommu (mmu);size_t ret; +/* The arm-smmu driver expects the addresses to be sign extended */ +if (iova &BIT_ULL (48 )) + iova |=GENMASK_ULL (63 ,49 ); + ret =iommu_map_sg (iommu->domain, iova, sgt->sgl, sgt->nents, prot);WARN_ON (!ret);@@ -46,6 +50,9 @@ static int msm_iommu_unmap(struct msm_mmu *mmu, uint64_t iova, size_t len) {struct msm_iommu *iommu =to_msm_iommu (mmu); +if (iova &BIT_ULL (48 )) + iova |=GENMASK_ULL (63 ,49 ); +iommu_unmap (iommu->domain, iova, len);return 0 ; --2.17.1
Quoted 4 lines
_______________________________________________
Freedreno mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/freedreno
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Re: [Freedreno] [PATCH v9 6/7] drm/msm: Set the global virtual address range from the IOMMU domain
Quoted 23 lines · Rob Clark
Quoted 22 lines · Jordan Crouse
Use the aperture settings from the IOMMU domain to set up the virtual address range for the GPU. This allows us to transparently deal with IOMMU side features (like split pagetables).
Signed-off-by Jordan Crouse <[email protected]>---
drivers/gpu/drm/msm/adreno/adreno_gpu.c | 13 +++++++++++-- drivers/gpu/drm/msm/msm_iommu.c | 7 +++++++ 2 files changed, 18 insertions(+), 2 deletions(-)drivers/gpu/drm/msm/adreno/adreno_gpu.c +1
-0@@ -192,9 +192,18 @@ adreno_iommu_create_address_space(struct msm_gpu *gpu,struct iommu_domain *iommu =iommu_domain_alloc (&platform_bus_type);struct msm_mmu *mmu =msm_iommu_new (&pdev->dev, iommu);struct msm_gem_address_space *aspace; +u64 start, size;
Quoted 16 lines
drivers/gpu/drm/msm/adreno/adreno_gpu.c adreno_iommu_create_address_space(struct msm_gpu *gpu, · line 206
Quoted 12 lines
drivers/gpu/drm/msm/adreno/adreno_gpu.c +10
-2- aspace =msm_gem_address_space_create (mmu,"gpu" ,SZ_16M , -0xffffffff -SZ_16M ); +/* +* Use the aperture start or SZ_16M, whichever is greater. This will +* ensure that we align with the allocated pagetable range while still +* allowing room in the lower 32 bits for GMEM and whatnot +*/ + start =max_t (u64,SZ_16M , iommu->geometry.aperture_start); + size = iommu->geometry.aperture_end - start +1 ; + + aspace =msm_gem_address_space_create (mmu,"gpu" , + start &GENMASK (48 ,0 ), size);hmm, I kinda think this isn't going to play well for the 32b gpus (pre-a5xx).. possibly we should add address space size to 'struct
adreno_info'?
I checked and qcom-iommu sets the aperture correctly so this should be okay for everybody. To be honest, I'm nots sure if we even need to mask the start to 49 bits. It seems that all of the iommu implementations do the right thing. Of course it would be worth a check if you have a 4xx handy.
Quoted 2 lines
Or I guess it is always going to be the same for all devices within a generation? So it could just be passed in to
adreno_gpu_init()
We can do that easily if we are worried about it (see also: a2xx). I just figured this might save us a bit of code.
Quoted 1 line
Hopefully that makes things smoother if we someday had more than 48bits..
We'll be at 49 bits for as far ahead as I can see. 49 bits has a special meaning in the SMMU so it is a natural fit for the GPU hardware. If we change in N generations we can just shift to a family specific function at that point.
Jordan
Quoted 38 lines
BR,
-RQuoted 35 lines
drivers/gpu/drm/msm/adreno/adreno_gpu.c +0
-0if (IS_ERR (aspace) && !IS_ERR (mmu)) mmu->funcs->destroy (mmu);drivers/gpu/drm/msm/msm_iommu.c +7
-1@@ -36,6 +36,10 @@ static int msm_iommu_map(struct msm_mmu *mmu, uint64_t iova,struct msm_iommu *iommu =to_msm_iommu (mmu);size_t ret; +/* The arm-smmu driver expects the addresses to be sign extended */ +if (iova &BIT_ULL (48 )) + iova |=GENMASK_ULL (63 ,49 ); + ret =iommu_map_sg (iommu->domain, iova, sgt->sgl, sgt->nents, prot);WARN_ON (!ret);@@ -46,6 +50,9 @@ static int msm_iommu_unmap(struct msm_mmu *mmu, uint64_t iova, size_t len) {struct msm_iommu *iommu =to_msm_iommu (mmu); +if (iova &BIT_ULL (48 )) + iova |=GENMASK_ULL (63 ,49 ); +iommu_unmap (iommu->domain, iova, len);return 0 ; --2.17.1
_______________________________________________
Freedreno mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/freedreno
Signature · 6 lines
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project _______________________________________________ dri-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/dri-devel
Hi Will, Robin,
Quoted 43 lines · Jordan Crouse
Another iteration of the split-pagetable support for arm-smmu and the Adreno GPU SMMU. After email discussions [1] we opted to make a arm-smmu implementation for specifically for the Adreno GPU and use that to enable split pagetable support and later other implementation specific bits that we need.
On the hardware side this is very close to the same code from before [2] only the TTBR1 quirk is turned on by the implementation and not a domain attribute. In drm/msm we use the returned size of the aperture as a clue to let us know which virtual address space we should use for global memory objects.
There are two open items that you should be aware of. First, in the implementation specific code we have to check the compatible string of the device so that we only enable TTBR1 for the GPU (SID 0) and not the GMU (SID 4). I went back and forth trying to decide if I wanted to use the compatible string or the SID as the filter and settled on the compatible string but I could be talked out of it.
The other open item is that in drm/msm the hardware only uses 49 bits of the address space but arm-smmu expects the address to be sign extended all the way to 64 bits. This isn't a problem normally unless you look at the hardware registers that contain a IOVA and then the upper bits will be zero. I opted to restrict the internal drm/msm IOVA range to only 49 bits and then sign extend right before calling
iommu_map/iommu_unmap. This is a bit wonky but I thought that matching the hardware would be less confusing when debugging a hang.v9: Fix bot-detected merge conflict
v7: Add attached device tosmmu_domainto pass to implementation specific functions[1] https://lists.linuxfoundation.org/pipermail/iommu/2020-May/044537.html
[2] https://patchwork.kernel.org/patch/11482591/Jordan Crouse (7):
iommu/arm-smmu: Pass io-pgtable config to implementation specific
function
iommu/arm-smmu: Add support for split pagetables
dt-bindings: arm-smmu: Add compatible string for Adreno GPU SMMU iommu/arm-smmu: Add a pointer to the attached device tosmmu_domainiommu/arm-smmu: Add implementation for the adreno GPU SMMU drm/msm: Set the global virtual address range from the IOMMU domain
arm: dts: qcom: sm845: Set the compatible string for the GPU SMMU
Quoted 9 lines
.../devicetree/bindings/iommu/arm,smmu.yaml | 4 ++ arch/arm64/boot/dts/qcom/sdm845.dtsi | 2 +- drivers/gpu/drm/msm/adreno/adreno_gpu.c | 13 +++++- drivers/gpu/drm/msm/msm_iommu.c | 7 +++ drivers/iommu/arm-smmu-impl.c | 6 ++- drivers/iommu/arm-smmu-qcom.c | 45 ++++++++++++++++++- drivers/iommu/arm-smmu.c | 38 +++++++++++----- drivers/iommu/arm-smmu.h | 30 ++++++++++--- 8 files changed, 120 insertions(+), 25 deletions(-)
Any chance reviewing this?
Thanks,
Sai
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Re: [PATCH v9 2/7] iommu/arm-smmu: Add support for split pagetables
In reply to Jordan CrouseQuoted 17 lines · Jordan Crouse
Enable TTBR1 for a context bank if
IO_PGTABLE_QUIRK_ARM_TTBR1is selected by the io-pgtable configuration.Signed-off-by Jordan Crouse <[email protected]>---
drivers/iommu/arm-smmu.c | 21 ++++++++++++++++----- drivers/iommu/arm-smmu.h | 25 +++++++++++++++++++------ 2 files changed, 35 insertions(+), 11 deletions(-)drivers/iommu/arm-smmu.c +0
-0@@ -555,11 +555,15 @@ static void arm_smmu_init_context_bank(struct arm_smmu_domain *smmu_domain, cb->ttbr[0 ] = pgtbl_cfg->arm_v7s_cfg.ttbr;
drivers/iommu/arm-smmu.c static void arm_smmu_init_context_bank(struct arm_smmu_domain *smmu_domain, · line 561
Quoted 10 lines
drivers/iommu/arm-smmu.c +3
-4cb->ttbr[1 ] =0 ; }else { - cb->ttbr[0 ] = pgtbl_cfg->arm_lpae_s1_cfg.ttbr; - cb->ttbr[0 ] |=FIELD_PREP (ARM_SMMU_TTBRn_ASID, - cfg->asid); + cb->ttbr[0 ] =FIELD_PREP (ARM_SMMU_TTBRn_ASID, + cfg->asid); cb->ttbr[1 ] =FIELD_PREP (ARM_SMMU_TTBRn_ASID, - cfg->asid); + cfg->asid);
above looks like stray whitespace changes?
Quoted 56 lines
drivers/iommu/arm-smmu.c +13
-1+ +if (pgtbl_cfg->quirks &IO_PGTABLE_QUIRK_ARM_TTBR1 ) + cb->ttbr[1 ] |= pgtbl_cfg->arm_lpae_s1_cfg.ttbr; +else + cb->ttbr[0 ] |= pgtbl_cfg->arm_lpae_s1_cfg.ttbr; } }else { cb->ttbr[0 ] = pgtbl_cfg->arm_lpae_s2_cfg.vttbr;@@ -824,7 +828,14 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,/* Update the domain's page sizes to reflect the page table format */ domain->pgsize_bitmap = pgtbl_cfg.pgsize_bitmap; - domain->geometry.aperture_end = (1UL << ias) -1 ; + +if (pgtbl_cfg.quirks &IO_PGTABLE_QUIRK_ARM_TTBR1 ) { + domain->geometry.aperture_start = ~0UL << ias; + domain->geometry.aperture_end = ~0UL ; + }else { + domain->geometry.aperture_end = (1UL << ias) -1 ; + } + domain->geometry.force_aperture = true;/* Initialise the context bank with our page table cfg */ drivers/iommu/arm-smmu.h +7
-6@@ -168,10 +168,12 @@ enum arm_smmu_cbar_type {#define ARM_SMMU_CB_TCR 0x30#define ARM_SMMU_TCR_EAE BIT(31)#define ARM_SMMU_TCR_EPD1 BIT(23) +#define ARM_SMMU_TCR_A1 BIT(22)#define ARM_SMMU_TCR_TG0 GENMASK(15, 14)#define ARM_SMMU_TCR_SH0 GENMASK(13, 12)#define ARM_SMMU_TCR_ORGN0 GENMASK(11, 10)#define ARM_SMMU_TCR_IRGN0 GENMASK(9, 8) +#define ARM_SMMU_TCR_EPD0 BIT(7)#define ARM_SMMU_TCR_T0SZ GENMASK(5, 0)#define ARM_SMMU_VTCR_RES1 BIT(31)@@ -347,12 +349,23 @@ struct arm_smmu_domain {static inline u32 arm_smmu_lpae_tcr (struct io_pgtable_cfg *cfg) { -return ARM_SMMU_TCR_EPD1 | -FIELD_PREP (ARM_SMMU_TCR_TG0 , cfg->arm_lpae_s1_cfg.tcr.tg) | -FIELD_PREP (ARM_SMMU_TCR_SH0 , cfg->arm_lpae_s1_cfg.tcr.sh) | -FIELD_PREP (ARM_SMMU_TCR_ORGN0 , cfg->arm_lpae_s1_cfg.tcr.orgn) | -FIELD_PREP (ARM_SMMU_TCR_IRGN0 , cfg->arm_lpae_s1_cfg.tcr.irgn) | -FIELD_PREP (ARM_SMMU_TCR_T0SZ , cfg->arm_lpae_s1_cfg.tcr.tsz); +u32 tcr =FIELD_PREP (ARM_SMMU_TCR_TG0 , cfg->arm_lpae_s1_cfg.tcr.tg) | +FIELD_PREP (ARM_SMMU_TCR_SH0 , cfg->arm_lpae_s1_cfg.tcr.sh) | +FIELD_PREP (ARM_SMMU_TCR_ORGN0 , cfg->arm_lpae_s1_cfg.tcr.orgn) | +FIELD_PREP (ARM_SMMU_TCR_IRGN0 , cfg->arm_lpae_s1_cfg.tcr.irgn) | +FIELD_PREP (ARM_SMMU_TCR_T0SZ , cfg->arm_lpae_s1_cfg.tcr.tsz);
drivers/iommu/arm-smmu.h struct arm_smmu_domain { · line 366
Quoted 10 lines
drivers/iommu/arm-smmu.h +10
-0+ +/* +* When TTBR1 is selected shift the TCR fields by 16 bits and disable +* translation in TTBR0 +*/ +if (cfg->quirks &IO_PGTABLE_QUIRK_ARM_TTBR1 ) { + tcr = (tcr <<16 ) & ~ARM_SMMU_TCR_A1 ; + tcr |=ARM_SMMU_TCR_EPD0 ; + }else + tcr |=ARM_SMMU_TCR_EPD1 ;
I'm not personally a fan of if/else ladders that mix {}'s, but Will/Robin may have a different opinion
BR,
-R
Quoted 12 lines
drivers/iommu/arm-smmu.h +2
-1+ +return tcr; }static inline u32 arm_smmu_lpae_tcr2 (struct io_pgtable_cfg *cfg) --2.17.1
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
Will Deacon
maintainerRe: [PATCH v9 4/7] iommu/arm-smmu: Add a pointer to the attached device to smmu_domain
In reply to Jordan CrouseQuoted 47 lines · Jordan Crouse
Add a link to the pointer to the struct device that is attached to a domain. This makes it easy to get the pointer if it is needed in the implementation specific code.
Signed-off-by Jordan Crouse <[email protected]>---
drivers/iommu/arm-smmu.c | 6 ++++-- drivers/iommu/arm-smmu.h | 1 + 2 files changed, 5 insertions(+), 2 deletions(-)drivers/iommu/arm-smmu.c +4
-2@@ -668,7 +668,8 @@ static void arm_smmu_write_context_bank(struct arm_smmu_device *smmu, int idx) }static int arm_smmu_init_domain_context (struct iommu_domain *domain, -struct arm_smmu_device *smmu) +struct arm_smmu_device *smmu, +struct device *dev) {int irq, start, ret =0 ;unsigned long ias, oas;@@ -801,6 +802,7 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain, cfg->asid = cfg->cbndx; smmu_domain->smmu = smmu; + smmu_domain->dev = dev; pgtbl_cfg = (struct io_pgtable_cfg ) { .pgsize_bitmap = smmu->pgsize_bitmap,@@ -1190,7 +1192,7 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev)return ret;/* Ensure that the domain is finalised */ - ret =arm_smmu_init_domain_context (domain, smmu); + ret =arm_smmu_init_domain_context (domain, smmu, dev);if (ret <0 ) goto rpm_put;drivers/iommu/arm-smmu.h +0
-0
drivers/iommu/arm-smmu.h struct arm_smmu_domain { · line 348
Quoted 4 lines
drivers/iommu/arm-smmu.h +1
-0@@ -345,6 +345,7 @@ struct arm_smmu_domain {struct mutex init_mutex;/* Protects smmu pointer */ spinlock_t cb_lock;/* Serialises ATS1* ops and TLB syncs */ struct iommu_domain domain; +struct device *dev;/* Device attached to this domain */
This really doesn't feel right to me -- you can generally have multiple devices attached to a domain and they can come and go without the domain being destroyed. Perhaps you could instead identify the GPU during cfg_probe() and squirrel that information away somewhere?
The rest of the series looks ok to me.
Will
Re: [Freedreno] [PATCH v9 4/7] iommu/arm-smmu: Add a pointer to the attached device to smmu_domain
Quoted 43 lines · Will Deacon
Quoted 42 lines · Jordan Crouse
Add a link to the pointer to the struct device that is attached to a domain. This makes it easy to get the pointer if it is needed in the implementation specific code.
Signed-off-by Jordan Crouse <[email protected]>---
drivers/iommu/arm-smmu.c | 6 ++++-- drivers/iommu/arm-smmu.h | 1 + 2 files changed, 5 insertions(+), 2 deletions(-)drivers/iommu/arm-smmu.c +4
-2@@ -668,7 +668,8 @@ static void arm_smmu_write_context_bank(struct arm_smmu_device *smmu, int idx) }static int arm_smmu_init_domain_context (struct iommu_domain *domain, -struct arm_smmu_device *smmu) +struct arm_smmu_device *smmu, +struct device *dev) {int irq, start, ret =0 ;unsigned long ias, oas;@@ -801,6 +802,7 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain, cfg->asid = cfg->cbndx; smmu_domain->smmu = smmu; + smmu_domain->dev = dev; pgtbl_cfg = (struct io_pgtable_cfg ) { .pgsize_bitmap = smmu->pgsize_bitmap,@@ -1190,7 +1192,7 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev)return ret;/* Ensure that the domain is finalised */ - ret =arm_smmu_init_domain_context (domain, smmu); + ret =arm_smmu_init_domain_context (domain, smmu, dev);if (ret <0 ) goto rpm_put;
Quoted 14 lines
drivers/iommu/arm-smmu.h struct arm_smmu_domain { · line 348
Quoted 9 lines
drivers/iommu/arm-smmu.h +1
-0@@ -345,6 +345,7 @@ struct arm_smmu_domain {struct mutex init_mutex;/* Protects smmu pointer */ spinlock_t cb_lock;/* Serialises ATS1* ops and TLB syncs */ struct iommu_domain domain; +struct device *dev;/* Device attached to this domain */ This really doesn't feel right to me -- you can generally have multiple devices attached to a domain and they can come and go without the domain being destroyed. Perhaps you could instead identify the GPU during
cfg_probe()and squirrel that information away somewhere?
I need some help here. The SMMU device (qcom,adreno-smmu) will have at least two stream ids from two different platform devices (GPU and GMU) and I need to configure split-pagetable and stall/terminate differently on the two domains.
I couldn't figure out a way to identify the platform device before it attached itself with iommu_attach_device. I tried poking around in fwspec but got lost.
If there is a way we can uniquely identify the devices (by stream id maybe) then we could use that though I have reservations about hard coding stream IDs in the impl driver. That said, the stream IDs have never changed in the life of the GPU so maybe it's not a problem that needs solving.
Jordan
Quoted 7 lines
The rest of the series looks ok to me.
Will
_______________________________________________
Freedreno mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/freedreno
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
Will Deacon
maintainerRe: [Freedreno] [PATCH v9 4/7] iommu/arm-smmu: Add a pointer to the attached device to smmu_domain
Quoted 17 lines · Jordan Crouse
Quoted 15 lines · Will Deacon
drivers/iommu/arm-smmu.h struct arm_smmu_domain { · line 348
Quoted 9 lines · Jordan Crouse
drivers/iommu/arm-smmu.h +1
-0@@ -345,6 +345,7 @@ struct arm_smmu_domain {struct mutex init_mutex;/* Protects smmu pointer */ spinlock_t cb_lock;/* Serialises ATS1* ops and TLB syncs */ struct iommu_domain domain; +struct device *dev;/* Device attached to this domain */ This really doesn't feel right to me -- you can generally have multiple devices attached to a domain and they can come and go without the domain being destroyed. Perhaps you could instead identify the GPU during
cfg_probe()and squirrel that information away somewhere?
Quoted 3 lines
I need some help here. The SMMU device (qcom,adreno-smmu) will have at least two stream ids from two different platform devices (GPU and GMU) and I need to configure split-pagetable and stall/terminate differently on the two domains.
Hmm. How does the GPU driver know which context bank is assigned to the GPU and which one is assigned to the GMU? I assume it needs this information so that it can play its nasty tricks with the TTBR registers?
I ask because if we need to guarantee stability of the context-bank assignment, then you could match on that in the ->init_context() callback, but now I worry that it currently works by luck :/
Do we need to add an extra callback to allocate the context bank?
Will
Re: [Freedreno] [PATCH v9 4/7] iommu/arm-smmu: Add a pointer to the attached device to smmu_domain
Quoted 23 lines · Will Deacon
Quoted 17 lines · Jordan Crouse
Quoted 15 lines · Will Deacon
drivers/iommu/arm-smmu.h struct arm_smmu_domain { · line 348
Quoted 9 lines · Jordan Crouse
drivers/iommu/arm-smmu.h +1
-0@@ -345,6 +345,7 @@ struct arm_smmu_domain {struct mutex init_mutex;/* Protects smmu pointer */ spinlock_t cb_lock;/* Serialises ATS1* ops and TLB syncs */ struct iommu_domain domain; +struct device *dev;/* Device attached to this domain */ This really doesn't feel right to me -- you can generally have multiple devices attached to a domain and they can come and go without the domain being destroyed. Perhaps you could instead identify the GPU during
cfg_probe()and squirrel that information away somewhere?Quoted 3 lines
I need some help here. The SMMU device (qcom,adreno-smmu) will have at least two stream ids from two different platform devices (GPU and GMU) and I need to configure split-pagetable and stall/terminate differently on the two domains.
Quoted 7 lines
Hmm. How does the GPU driver know which context bank is assigned to the GPU and which one is assigned to the GMU? I assume it needs this information so that it can play its nasty tricks with the TTBR registers?
I ask because if we need to guarantee stability of the context-bank assignment, then you could match on that in the ->init_context() callback, but now I worry that it currently works by luck :/
Yeah, it is basically by luck right now.. some sort of impl callback which was passed the dev into impl->init_context() might do the trick (ie. then the impl could match on compat string)
BR,
-R
Quoted 7 lines
Do we need to add an extra callback to allocate the context bank?
Will
_______________________________________________
Freedreno mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/freedreno
Re: [Freedreno] [PATCH v9 4/7] iommu/arm-smmu: Add a pointer to the attached device to smmu_domain
In reply to Will DeaconQuoted 22 lines · Will Deacon
Quoted 17 lines · Jordan Crouse
Quoted 15 lines · Will Deacon
drivers/iommu/arm-smmu.h struct arm_smmu_domain { · line 348
Quoted 9 lines · Jordan Crouse
drivers/iommu/arm-smmu.h +1
-0@@ -345,6 +345,7 @@ struct arm_smmu_domain {struct mutex init_mutex;/* Protects smmu pointer */ spinlock_t cb_lock;/* Serialises ATS1* ops and TLB syncs */ struct iommu_domain domain; +struct device *dev;/* Device attached to this domain */ This really doesn't feel right to me -- you can generally have multiple devices attached to a domain and they can come and go without the domain being destroyed. Perhaps you could instead identify the GPU during
cfg_probe()and squirrel that information away somewhere?Quoted 3 lines
I need some help here. The SMMU device (qcom,adreno-smmu) will have at least two stream ids from two different platform devices (GPU and GMU) and I need to configure split-pagetable and stall/terminate differently on the two domains.
Quoted 7 lines
Hmm. How does the GPU driver know which context bank is assigned to the GPU and which one is assigned to the GMU? I assume it needs this information so that it can play its nasty tricks with the TTBR registers?
I ask because if we need to guarantee stability of the context-bank assignment, then you could match on that in the ->init_context() callback, but now I worry that it currently works by luck :/
Its more like "educated" luck. If we assign the domains in the right order we know that the GPU will be on context bank 0 but you are entirely right that if everything doesn't go exactly the way we need things go badly.
Some targets do have the ability to reprogram which context bank is used but that would require a domain attribute from the SMMU driver to communicate that information back to the GPU driver.
Quoted 1 line
Do we need to add an extra callback to allocate the context bank?
That seems like a reasonable option. That seems like it would work with legacy targets and rely less on luck.
Jordan
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project