Add system mmu support for Armada-806
Subsystems (5)
Gregory CLEMENT
maintainerv2 cover Add system mmu support for Armada-806
Hello,
last year a first version of this series was submitted to add support for IOMMU for AP806, including workaround for accessing ARM SMMU 64bit registers[1].
For the record, AP-806 can't access SMMU registers with 64bit width, this patches split the readq/writeq for 32bit access, due to erratanum #582743.
Based on the feedback from Robin Murphy, I also add code ensuring that we won't try to use AArch64 format with 32 bits acces.
It was also discussed to not use compatible but propertu to support this workaround. I agree to make this change if needed, but for now I would like to have a feedback on the current code to know if it is acceptable if there is still potential issue.
The series was tested on a vanilla v5.1 kernel, and without the series, an USB stick was not detected under QEMU whereas with this series it worked as expected.
Greogry
[1]: https://lkml.org/lkml/2018/10/15/373
Gregory CLEMENT (1):
arm64: dts: marvell: armada-ap806: add smmu support
Hanna Hawa (3):
iommu/arm-smmu: Introduce wrapper for writeq/readq iommu/arm-smmu: Workaround for Marvell Armada-AP806 SoC erratum
#582743
dt-bindings: iommu/arm,smmu: add compatible string for Marvell
Documentation/arm64/silicon-errata.txt | 2 + .../devicetree/bindings/iommu/arm,smmu.txt | 1 + arch/arm64/boot/dts/marvell/armada-ap806.dtsi | 17 +++++ drivers/iommu/arm-smmu.c | 74 ++++++++++++++++--- 4 files changed, 83 insertions(+), 11 deletions(-)
Signature · 1 line
2.20.1
Gregory CLEMENT
maintainerPATCH v2 1/4 iommu/arm-smmu: Introduce wrapper for writeq/readq
From: Hanna Hawa <[email protected]>
This patch introduces the smmu_writeq_relaxed/smmu_readq_relaxed helpers, as preparation to add specific Marvell work-around for accessing 64 bits width registers of ARM SMMU.
--- drivers/iommu/arm-smmu.c | 36 +++++++++++++++++++++++++++--------- 1 file changed, 27 insertions(+), 9 deletions(-)
drivers/iommu/arm-smmu.c +27 -9
@@ -91,9 +91,11 @@* therefore this actually makes more sense than it might first appear. */ #ifdef CONFIG_64BIT -#define smmu_write_atomic_lq writeq_relaxed +#define smmu_write_atomic_lq (smmu, val, reg) \ + smmu_writeq_relaxed(smmu, val, reg)#else -#define smmu_write_atomic_lq writel_relaxed +#define smmu_write_atomic_lq (smmu, val, reg) \ + writel_relaxed(val, reg)#endif /* Translation context bank */
@@ -295,6 +297,19 @@ static struct arm_smmu_domain *to_smmu_domain(struct iommu_domain *dom) return container_of(dom, struct arm_smmu_domain, domain); } +static inline void smmu_writeq_relaxed (struct arm_smmu_device *smmu, +u64 val, +void __iomem *addr) +{ +writeq_relaxed (val, addr); +} + +static inline u64 smmu_readq_relaxed (struct arm_smmu_device *smmu, +void __iomem *addr) +{ +return readq_relaxed (addr); +} +static void parse_driver_options (struct arm_smmu_device *smmu) {int i =0 ;
@@ -495,6 +510,7 @@ static void arm_smmu_tlb_inv_range_nosync(unsigned long iova, size_t size, size_t granule, bool leaf, void *cookie) { struct arm_smmu_domain *smmu_domain = cookie; +struct arm_smmu_device *smmu = smmu_domain->smmu;struct arm_smmu_cfg *cfg = &smmu_domain->cfg;bool stage1 = cfg->cbar !=CBAR_TYPE_S2_TRANS ;void __iomem *reg =ARM_SMMU_CB (smmu_domain->smmu, cfg->cbndx);
@@ -516,7 +532,7 @@ static void arm_smmu_tlb_inv_range_nosync(unsigned long iova, size_t size, iova >>=12 ; iova |= (u64 )cfg->asid <<48 ;do { -writeq_relaxed (iova, reg); +smmu_writeq_relaxed (smmu, iova, reg); iova += granule >>12 ; }while (size -= granule); }
@@ -525,7 +541,7 @@ static void arm_smmu_tlb_inv_range_nosync(unsigned long iova, size_t size,ARM_SMMU_CB_S2_TLBIIPAS2 ; iova >>=12 ;do { -smmu_write_atomic_lq (iova, reg); +smmu_write_atomic_lq (smmu, iova, reg); iova += granule >>12 ; }while (size -= granule); }
@@ -584,7 +600,7 @@ static irqreturn_t arm_smmu_context_fault(int irq, void *dev)return IRQ_NONE ; fsynr =readl_relaxed (cb_base +ARM_SMMU_CB_FSYNR0 ); - iova =readq_relaxed (cb_base +ARM_SMMU_CB_FAR ); + iova =smmu_readq_relaxed (smmu, cb_base +ARM_SMMU_CB_FAR ); dev_err_ratelimited(smmu->dev, "Unhandled context fault: fsr=0x%x, iova=0x%08lx, fsynr=0x%x, cb=%d\n",
@@ -734,9 +750,11 @@ static void arm_smmu_write_context_bank(struct arm_smmu_device *smmu, int idx)writel_relaxed (cb->ttbr[0 ], cb_base +ARM_SMMU_CB_TTBR0 );writel_relaxed (cb->ttbr[1 ], cb_base +ARM_SMMU_CB_TTBR1 ); }else { -writeq_relaxed (cb->ttbr[0 ], cb_base +ARM_SMMU_CB_TTBR0 ); +smmu_writeq_relaxed (smmu, cb->ttbr[0 ], + cb_base +ARM_SMMU_CB_TTBR0 );if (stage1) -writeq_relaxed (cb->ttbr[1 ], cb_base +ARM_SMMU_CB_TTBR1 ); +smmu_writeq_relaxed (smmu, cb->ttbr[1 ], + cb_base +ARM_SMMU_CB_TTBR1 ); }/* MAIRs (stage-1 only) */
@@ -1367,7 +1385,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain,/* ATS1 registers can only be written atomically */ va = iova & ~0xfffUL ;if (smmu->version ==ARM_SMMU_V2 ) -smmu_write_atomic_lq (va, cb_base +ARM_SMMU_CB_ATS1PR ); +smmu_write_atomic_lq (smmu, va, cb_base +ARM_SMMU_CB_ATS1PR );else /* Register is only 32-bit in v1 */ writel_relaxed (va, cb_base +ARM_SMMU_CB_ATS1PR );
@@ -1380,7 +1398,7 @@ static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain,return ops->iova_to_phys (ops, iova); } - phys =readq_relaxed (cb_base +ARM_SMMU_CB_PAR ); + phys =smmu_readq_relaxed (smmu, cb_base +ARM_SMMU_CB_PAR );spin_unlock_irqrestore (&smmu_domain->cb_lock, flags);if (phys &CB_PAR_F ) {dev_err (dev,"translation fault!\n" );
Signature · 1 line
2.20.1
Will Deacon
maintainerRe: [PATCH v2 1/4] iommu/arm-smmu: Introduce wrapper for writeq/readq
Hi Gregory, Hanna,
Quoted 11 lines · Gregory CLEMENT
From: Hanna Hawa <[email protected]>
This patch introduces the smmu_writeq_relaxed/smmu_readq_relaxed helpers, as preparation to add specific Marvell work-around for accessing 64 bits width registers of ARM SMMU.
Signed-off-by Hanna Hawa <[email protected]>Signed-off-by Gregory CLEMENT <[email protected]>--- drivers/iommu/arm-smmu.c | 36 +++++++++++++++++++++++++++--------- 1 file changed, 27 insertions(+), 9 deletions(-)
Sorry for the delay in replying to this -- Robin's been reworking the driver so that implementation quirks can be specified more cleanly. Please can you take a look at:
https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-joerg/arm-smmu/refactoring
and try to respin your patches on top of that?
Thanks,
Will
Robin Murphy
reviewerRe: [PATCH v2 1/4] iommu/arm-smmu: Introduce wrapper for writeq/readq
Quoted 15 lines · Will Deacon
Hi Gregory, Hanna,
Quoted 11 lines · Gregory CLEMENT
From: Hanna Hawa <[email protected]>
This patch introduces the smmu_writeq_relaxed/smmu_readq_relaxed helpers, as preparation to add specific Marvell work-around for accessing 64 bits width registers of ARM SMMU.
Signed-off-by Hanna Hawa <[email protected]>Signed-off-by Gregory CLEMENT <[email protected]>--- drivers/iommu/arm-smmu.c | 36 +++++++++++++++++++++++++++--------- 1 file changed, 27 insertions(+), 9 deletions(-)
Quoted 7 lines
Sorry for the delay in replying to this -- Robin's been reworking the driver so that implementation quirks can be specified more cleanly. Please can you take a look at:
https://git.kernel.org/pub/scm/linux/kernel/git/will/linux.git/log/?h=for-joerg/arm-smmu/refactoring
and try to respin your patches on top of that?
Right, the arm_smmu_impl design was specifically anticipating this quirk as well - it should just be a case of a cfg_probe hook to hide the features which can't work, plus {read,write}_reg64 hooks to override any remaining 64-bit accesses with the explicit hi_lo_* variants, munged together (either statically or dynamically) with the standard MMU-500 hooks.
Robin.
Gregory CLEMENT
maintainerPATCH v2 2/4 iommu/arm-smmu: Workaround for Marvell Armada-AP806 SoC erratum #582743
In reply to Gregory CLEMENTFrom: Hanna Hawa <[email protected]>
Due to erratum #582743, the Marvell Armada-AP806 can't access 64bit to ARM SMMUv2 registers.
This patch split the writeq/readq to two accesses of writel/readl.
We also mask the MMU_IDR2.PTFSv8 fields to not use AArch64 format but only AARCH32_L. Indeed with AArch64 format 32 bits acces is not supported.
Note that separate writes/reads to 2 is not problem regards to atomicity, because the driver use the readq/writeq while initialize the SMMU, report for SMMU fault, and use spinlock in one case (iova_to_phys).
--- Documentation/arm64/silicon-errata.txt | 2 ++ drivers/iommu/arm-smmu.c | 42 +++++++++++++++++++++++--- 2 files changed, 40 insertions(+), 4 deletions(-)
Documentation/arm64/silicon-errata.txt +2 -0
@@ -72,6 +72,8 @@ stable kernels. | Cavium | ThunderX2 SMMUv3| #74 | N/A | | Cavium | ThunderX2 SMMUv3| #126 | N/A | | | | | | +| Marvell | ARM-MMU-500 | #582743 | N/A | +| | | | | | Freescale/NXP | LS2080A/LS1043A | A-008585 | FSL_ERRATUM_A008585 | | | | | | | Hisilicon | Hip0{5,6,7} | #161010101 | HISILICON_ERRATUM_161010101 |
drivers/iommu/arm-smmu.c +38 -4
89 lines not on this page · raw message
Signature · 1 line
2.20.1
Gregory CLEMENT
maintainerPATCH v2 3/4 dt-bindings: iommu/arm,smmu: add compatible string for Marvell
In reply to Gregory CLEMENTFrom: Hanna Hawa <[email protected]>
Add specific compatible string for Marvell usage due errata of accessing 64bits registers of ARM SMMU, in AP806.
AP806 SoC uses the generic ARM-MMU500, and there's no specific implementation of Marvell, this compatible is used for errata only.
--- Documentation/devicetree/bindings/iommu/arm,smmu.txt | 1 + 1 file changed, 1 insertion(+)
Documentation/devicetree/bindings/iommu/arm,smmu.txt +1 -0
@@ -16,6 +16,7 @@ conditions. "arm,mmu-400" "arm,mmu-401" "arm,mmu-500" + "marvell,mmu-500" "cavium,smmu-v2" "qcom,smmu-v2"
Signature · 1 line
2.20.1
Re: [PATCH v2 3/4] dt-bindings: iommu/arm,smmu: add compatible string for Marvell
Quoted 5 lines · Gregory CLEMENT
From: Hanna Hawa <[email protected]>
Add specific compatible string for Marvell usage due errata of accessing 64bits registers of ARM SMMU, in AP806.
Quoted 9 lines
AP806 SoC uses the generic ARM-MMU500, and there's no specific implementation of Marvell, this compatible is used for errata only.
Signed-off-by Hanna Hawa <[email protected]>Signed-off-by Gregory CLEMENT <[email protected]>--- Documentation/devicetree/bindings/iommu/arm,smmu.txt | 1 + 1 file changed, 1 insertion(+)
Robin Murphy
reviewerRe: [PATCH v2 3/4] dt-bindings: iommu/arm,smmu: add compatible string for Marvell
In reply to Gregory CLEMENTQuoted 7 lines · Gregory CLEMENT
From: Hanna Hawa <[email protected]>
Add specific compatible string for Marvell usage due errata of accessing 64bits registers of ARM SMMU, in AP806.
AP806 SoC uses the generic ARM-MMU500, and there's no specific implementation of Marvell, this compatible is used for errata only.
Forgive me for repeating myself[1], but:
"Given that, I think something more specific like:
"marvell,ap806-smmu", "arm,mmu-500";
would be most appropriate. Otherwise, if some future Marvell SoC were to ever come out with a *different* MMU-500 integration problem, you'd already have painted yourself into a corner."
Robin.
[1] https://lore.kernel.org/linux-arm-kernel/[email protected]/
Quoted 19 lines
Signed-off-by Hanna Hawa <[email protected]>Signed-off-by Gregory CLEMENT <[email protected]>--- Documentation/devicetree/bindings/iommu/arm,smmu.txt | 1 + 1 file changed, 1 insertion(+)Documentation/devicetree/bindings/iommu/arm,smmu.txt +1
-0@@ -16,6 +16,7 @@ conditions. "arm,mmu-400" "arm,mmu-401" "arm,mmu-500" + "marvell,mmu-500" "cavium,smmu-v2" "qcom,smmu-v2"
Gregory CLEMENT
maintainerPATCH v2 4/4 arm64: dts: marvell: armada-ap806: add smmu support
In reply to Gregory CLEMENTAdd IOMMU node for Marvell AP806 based SoCs.
--- arch/arm64/boot/dts/marvell/armada-ap806.dtsi | 17 +++++++++++++++++ 1 file changed, 17 insertions(+)
arch/arm64/boot/dts/marvell/armada-ap806.dtsi +17 -0
@@ -115,6 +115,23 @@ interrupts = <17>; }; + smmu: iommu@5000000 { + compatible = "marvell,mmu-500"; + reg = <0x100000 0x100000>; + dma-coherent; + #iommu-cells = <1>; + #global-interrupts = <1>; + interrupts = <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>, + <GIC_SPI 6 IRQ_TYPE_LEVEL_HIGH>; + }; + odmi: odmi@300000 { compatible = "marvell,odmi-controller"; interrupt-controller;
Signature · 1 line
2.20.1