[PATCH v6 0/2] iommu/arm-smmu: hard iova_to_phys

[PATCH v6 0/2] iommu/arm-smmu: hard iova_to_phys

[Mitchel Humpherys]

This series introduces support for performing iova-to-phys translations via
the ARM SMMU hardware on supported implementations. We also make use of
some new generic macros for polling hardware registers.

v5..v6:

  - iopoll: use shift instead of divide
  - arm-smmu: no changes, resending series due to iopoll change.

v4..v5:

  - iopoll: Added support for other accessor functions
  - iopoll: Unified atomic and non-atomic interfaces
  - iopoll: Fixed erroneous `might_sleep'
  - arm-smmu: Lowered timeout and moved to new iopoll atomic interface

v3..v4:

  - Updated the iopoll commit message to reflect the patch better
  - Added locking around address translation op
  - Return 0 on iova_to_phys failure

v2..v3:

  - Removed unnecessary `dev_name's

v1..v2:

  - Renamed one of the iopoll macros to use the more standard `_atomic'
    suffix
  - Removed some convenience iopoll wrappers to encourage explicitness


Matt Wagantall (1):
  iopoll: Introduce memory-mapped IO polling macros

Mitchel Humpherys (1):
  iommu/arm-smmu: add support for iova_to_phys through ATS1PR

 drivers/iommu/arm-smmu.c |  79 +++++++++++++++++-
 include/linux/iopoll.h   | 213 +++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 291 insertions(+), 1 deletion(-)
 create mode 100644 include/linux/iopoll.h

-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH v6 1/2] iopoll: Introduce memory-mapped IO polling macros

[Mitchel Humpherys]

From: Matt Wagantall <mattw-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>

It is sometimes necessary to poll a memory-mapped register until its value
satisfies some condition. Introduce a family of convenience macros that do
this. Tight-looping, sleeping, and timing out can all be accomplished using
these macros.

Cc: Thierry Reding <thierry.reding-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Cc: Will Deacon <will.deacon-5wv7dgnIgG8@public.gmane.org>
Signed-off-by: Matt Wagantall <mattw-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
Signed-off-by: Mitchel Humpherys <mitchelh-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
---
Changes since v5:
  - Use a shift instead of a divide in the poll loop.
---
 include/linux/iopoll.h | 213 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 213 insertions(+)
 create mode 100644 include/linux/iopoll.h

diff --git a/include/linux/iopoll.h b/include/linux/iopoll.h
new file mode 100644
index 0000000000..21dd41942b
--- /dev/null
+++ b/include/linux/iopoll.h
@@ -0,0 +1,213 @@
+/*
+ * Copyright (c) 2012-2014 The Linux Foundation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ */
+
+#ifndef _LINUX_IOPOLL_H
+#define _LINUX_IOPOLL_H
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/hrtimer.h>
+#include <linux/delay.h>
+#include <linux/errno.h>
+#include <linux/io.h>
+
+/**
+ * readx_poll_timeout - Periodically poll an address until a condition is met or a timeout occurs
+ * @op: accessor function (takes @addr as its only argument)
+ * @addr: Address to poll
+ * @val: Variable to read the value into
+ * @cond: Break condition (usually involving @val)
+ * @sleep_us: Maximum time to sleep between reads in us (0 tight-loops)
+ * @timeout_us: Timeout in us, 0 means never timeout
+ *
+ * Returns 0 on success and -ETIMEDOUT upon a timeout. In either
+ * case, the last read value at @addr is stored in @val. Must not
+ * be called from atomic context if sleep_us or timeout_us are used.
+ *
+ * Generally you'll want to use one of the specialized macros defined below
+ * rather than this macro directly.
+ */
+#define readx_poll_timeout(op, addr, val, cond, sleep_us, timeout_us)	\
+({ \
+	ktime_t timeout = ktime_add_us(ktime_get(), timeout_us); \
+	might_sleep_if(sleep_us); \
+	for (;;) { \
+		(val) = op(addr); \
+		if (cond) \
+			break; \
+		if (timeout_us && ktime_compare(ktime_get(), timeout) > 0) { \
+			(val) = op(addr); \
+			break; \
+		} \
+		if (sleep_us) \
+			usleep_range((sleep_us >> 2) + 1, sleep_us); \
+	} \
+	(cond) ? 0 : -ETIMEDOUT; \
+})
+
+/**
+ * readx_poll_timeout_atomic - Periodically poll an address until a condition is met or a timeout occurs
+ * @op: accessor function (takes @addr as its only argument)
+ * @addr: Address to poll
+ * @val: Variable to read the value into
+ * @cond: Break condition (usually involving @val)
+ * @delay_us: Time to udelay between reads in us (0 tight-loops)
+ * @timeout_us: Timeout in us, 0 means never timeout
+ *
+ * Returns 0 on success and -ETIMEDOUT upon a timeout. In either
+ * case, the last read value at @addr is stored in @val.
+ *
+ * Generally you'll want to use one of the specialized macros defined below
+ * rather than this macro directly.
+ */
+#define readx_poll_timeout_atomic(op, addr, val, cond, delay_us, timeout_us) \
+({ \
+	ktime_t timeout = ktime_add_us(ktime_get(), timeout_us); \
+	for (;;) { \
+		(val) = op(addr); \
+		if (cond) \
+			break; \
+		if (timeout_us && ktime_compare(ktime_get(), timeout) > 0) { \
+			(val) = op(addr); \
+			break; \
+		} \
+		if (delay_us) \
+			udelay(delay_us);	\
+	} \
+	(cond) ? 0 : -ETIMEDOUT; \
+})
+
+
+#define readl_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readl, addr, val, cond, delay_us, timeout_us)
+
+#define readl_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readl, addr, val, cond, delay_us, timeout_us)
+
+#define readb_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readb, addr, val, cond, delay_us, timeout_us)
+
+#define readb_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readb, addr, val, cond, delay_us, timeout_us)
+
+#define readw_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readw, addr, val, cond, delay_us, timeout_us)
+
+#define readw_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readw, addr, val, cond, delay_us, timeout_us)
+
+#define readq_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readq, addr, val, cond, delay_us, timeout_us)
+
+#define readq_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readq, addr, val, cond, delay_us, timeout_us)
+
+#define readb_relaxed_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readb_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readb_relaxed_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readb_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readw_relaxed_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readw_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readw_relaxed_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readw_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readl_relaxed_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readl_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readl_relaxed_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readl_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readq_relaxed_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readq_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readq_relaxed_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readq_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define ioread8_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(ioread8, addr, val, cond, delay_us, timeout_us)
+
+#define ioread8_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(ioread8, addr, val, cond, delay_us, timeout_us)
+
+#define ioread16_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(ioread16, addr, val, cond, delay_us, timeout_us)
+
+#define ioread16_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(ioread16, addr, val, cond, delay_us, timeout_us)
+
+#define ioread16be_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(ioread16be, addr, val, cond, delay_us, timeout_us)
+
+#define ioread16be_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(ioread16be, addr, val, cond, delay_us, timeout_us)
+
+#define ioread32_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(ioread32, addr, val, cond, delay_us, timeout_us)
+
+#define ioread32_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(ioread32, addr, val, cond, delay_us, timeout_us)
+
+#define ioread32b3_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(ioread32b3, addr, val, cond, delay_us, timeout_us)
+
+#define ioread32b3_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(ioread32b3, addr, val, cond, delay_us, timeout_us)
+
+#define inb_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(inb, addr, val, cond, delay_us, timeout_us)
+
+#define inb_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(inb, addr, val, cond, delay_us, timeout_us)
+
+#define inb_p_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(inb_p, addr, val, cond, delay_us, timeout_us)
+
+#define inb_p_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(inb_p, addr, val, cond, delay_us, timeout_us)
+
+#define inw_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(inw, addr, val, cond, delay_us, timeout_us)
+
+#define inw_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(inw, addr, val, cond, delay_us, timeout_us)
+
+#define inw_p_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(inw_p, addr, val, cond, delay_us, timeout_us)
+
+#define inw_p_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(inw_p, addr, val, cond, delay_us, timeout_us)
+
+#define inw_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(inw, addr, val, cond, delay_us, timeout_us)
+
+#define inw_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(inw, addr, val, cond, delay_us, timeout_us)
+
+#define inl_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(inl, addr, val, cond, delay_us, timeout_us)
+
+#define inl_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(inl, addr, val, cond, delay_us, timeout_us)
+
+#define inl_p_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(inl_p, addr, val, cond, delay_us, timeout_us)
+
+#define inl_p_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(inl_p, addr, val, cond, delay_us, timeout_us)
+
+
+#endif /* _LINUX_IOPOLL_H */
-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

Re: [PATCH v6 1/2] iopoll: Introduce memory-mapped IO polling macros

[Mitchel Humpherys]

On Tue, Oct 14 2014 at 02:53:29 PM, Mitchel Humpherys <mitchelh-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org> wrote:
> From: Matt Wagantall <mattw-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
>
> It is sometimes necessary to poll a memory-mapped register until its value
> satisfies some condition. Introduce a family of convenience macros that do
> this. Tight-looping, sleeping, and timing out can all be accomplished using
> these macros.
>
> Cc: Thierry Reding <thierry.reding-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Cc: Will Deacon <will.deacon-5wv7dgnIgG8@public.gmane.org>
> Signed-off-by: Matt Wagantall <mattw-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
> Signed-off-by: Mitchel Humpherys <mitchelh-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
> ---
> Changes since v5:
>   - Use a shift instead of a divide in the poll loop.
> ---
>  include/linux/iopoll.h | 213 +++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 213 insertions(+)
>  create mode 100644 include/linux/iopoll.h

I realize I sent this at a bad time (ELCE) but were there any more
comments on this patch?


-Mitch

-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

[PATCH v6 2/2] iommu/arm-smmu: add support for iova_to_phys through ATS1PR

[Mitchel Humpherys]

Currently, we provide the iommu_ops.iova_to_phys service by doing a
table walk in software to translate IO virtual addresses to physical
addresses. On SMMUs that support it, it can be useful to ask the SMMU
itself to do the translation. This can be used to warm the TLBs for an
SMMU. It can also be useful for testing and hardware validation.

Since the address translation registers are optional on SMMUv2, only
enable hardware translations when using SMMUv1 or when SMMU_IDR0.S1TS=1
and SMMU_IDR0.ATOSNS=0, as described in the ARM SMMU v1-v2 spec.

Signed-off-by: Mitchel Humpherys <mitchelh-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>
---
Changes since v5:
  - None. Re-sending series due to change in patch 1/2 in series
---
 drivers/iommu/arm-smmu.c | 79 +++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 78 insertions(+), 1 deletion(-)

diff --git a/drivers/iommu/arm-smmu.c b/drivers/iommu/arm-smmu.c
index 37dc3dd0df..ef57043994 100644
--- a/drivers/iommu/arm-smmu.c
+++ b/drivers/iommu/arm-smmu.c
@@ -36,6 +36,7 @@
 #include <linux/interrupt.h>
 #include <linux/io.h>
 #include <linux/iommu.h>
+#include <linux/iopoll.h>
 #include <linux/mm.h>
 #include <linux/module.h>
 #include <linux/of.h>
@@ -140,6 +141,7 @@
 #define ID0_S2TS			(1 << 29)
 #define ID0_NTS				(1 << 28)
 #define ID0_SMS				(1 << 27)
+#define ID0_ATOSNS			(1 << 26)
 #define ID0_PTFS_SHIFT			24
 #define ID0_PTFS_MASK			0x2
 #define ID0_PTFS_V8_ONLY		0x2
@@ -233,11 +235,16 @@
 #define ARM_SMMU_CB_TTBR0_HI		0x24
 #define ARM_SMMU_CB_TTBCR		0x30
 #define ARM_SMMU_CB_S1_MAIR0		0x38
+#define ARM_SMMU_CB_PAR_LO		0x50
+#define ARM_SMMU_CB_PAR_HI		0x54
 #define ARM_SMMU_CB_FSR			0x58
 #define ARM_SMMU_CB_FAR_LO		0x60
 #define ARM_SMMU_CB_FAR_HI		0x64
 #define ARM_SMMU_CB_FSYNR0		0x68
 #define ARM_SMMU_CB_S1_TLBIASID		0x610
+#define ARM_SMMU_CB_ATS1PR_LO		0x800
+#define ARM_SMMU_CB_ATS1PR_HI		0x804
+#define ARM_SMMU_CB_ATSR		0x8f0
 
 #define SCTLR_S1_ASIDPNE		(1 << 12)
 #define SCTLR_CFCFG			(1 << 7)
@@ -249,6 +256,10 @@
 #define SCTLR_M				(1 << 0)
 #define SCTLR_EAE_SBOP			(SCTLR_AFE | SCTLR_TRE)
 
+#define CB_PAR_F			(1 << 0)
+
+#define ATSR_ACTIVE			(1 << 0)
+
 #define RESUME_RETRY			(0 << 0)
 #define RESUME_TERMINATE		(1 << 0)
 
@@ -366,6 +377,7 @@ struct arm_smmu_device {
 #define ARM_SMMU_FEAT_TRANS_S1		(1 << 2)
 #define ARM_SMMU_FEAT_TRANS_S2		(1 << 3)
 #define ARM_SMMU_FEAT_TRANS_NESTED	(1 << 4)
+#define ARM_SMMU_FEAT_TRANS_OPS		(1 << 5)
 	u32				features;
 
 #define ARM_SMMU_OPT_SECURE_CFG_ACCESS (1 << 0)
@@ -1524,7 +1536,7 @@ static size_t arm_smmu_unmap(struct iommu_domain *domain, unsigned long iova,
 	return ret ? 0 : size;
 }
 
-static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain,
+static phys_addr_t arm_smmu_iova_to_phys_soft(struct iommu_domain *domain,
 					 dma_addr_t iova)
 {
 	pgd_t *pgdp, pgd;
@@ -1557,6 +1569,66 @@ static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain,
 	return __pfn_to_phys(pte_pfn(pte)) | (iova & ~PAGE_MASK);
 }
 
+static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain,
+					dma_addr_t iova)
+{
+	struct arm_smmu_domain *smmu_domain = domain->priv;
+	struct arm_smmu_device *smmu = smmu_domain->smmu;
+	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
+	struct device *dev = smmu->dev;
+	void __iomem *cb_base;
+	u32 tmp;
+	u64 phys;
+	unsigned long flags;
+
+	cb_base = ARM_SMMU_CB_BASE(smmu) + ARM_SMMU_CB(smmu, cfg->cbndx);
+
+	spin_lock_irqsave(&smmu_domain->lock, flags);
+
+	if (smmu->version == 1) {
+		u32 reg = iova & ~0xfff;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_LO);
+	} else {
+		u32 reg = iova & ~0xfff;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_LO);
+		reg = (iova & ~0xfff) >> 32;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_HI);
+	}
+
+	if (readl_poll_timeout_atomic(cb_base + ARM_SMMU_CB_ATSR, tmp,
+				!(tmp & ATSR_ACTIVE), 5, 50)) {
+		dev_err(dev,
+			"iova to phys timed out on 0x%pa. Falling back to software table walk.\n",
+			&iova);
+		return arm_smmu_iova_to_phys_soft(domain, iova);
+	}
+
+	phys = readl_relaxed(cb_base + ARM_SMMU_CB_PAR_LO);
+	phys |= ((u64) readl_relaxed(cb_base + ARM_SMMU_CB_PAR_HI)) << 32;
+
+	spin_unlock_irqrestore(&smmu_domain->lock, flags);
+
+	if (phys & CB_PAR_F) {
+		dev_err(dev, "translation fault!\n");
+		dev_err(dev, "PAR = 0x%llx\n", phys);
+		phys = 0;
+	} else {
+		phys = (phys & 0xfffffff000ULL) | (iova & 0x00000fff);
+	}
+
+	return phys;
+}
+
+static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain,
+					dma_addr_t iova)
+{
+	struct arm_smmu_domain *smmu_domain = domain->priv;
+
+	if (smmu_domain->smmu->features & ARM_SMMU_FEAT_TRANS_OPS)
+		return arm_smmu_iova_to_phys_hard(domain, iova);
+	return arm_smmu_iova_to_phys_soft(domain, iova);
+}
+
 static int arm_smmu_domain_has_cap(struct iommu_domain *domain,
 				   unsigned long cap)
 {
@@ -1777,6 +1849,11 @@ static int arm_smmu_device_cfg_probe(struct arm_smmu_device *smmu)
 		return -ENODEV;
 	}
 
+	if (smmu->version == 1 || (!(id & ID0_ATOSNS) && (id & ID0_S1TS))) {
+		smmu->features |= ARM_SMMU_FEAT_TRANS_OPS;
+		dev_notice(smmu->dev, "\taddress translation ops\n");
+	}
+
 	if (id & ID0_CTTW) {
 		smmu->features |= ARM_SMMU_FEAT_COHERENT_WALK;
 		dev_notice(smmu->dev, "\tcoherent table walk\n");
-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

Re: [PATCH v6 2/2] iommu/arm-smmu: add support for iova_to_phys through ATS1PR

[Will Deacon]

On Tue, Oct 14, 2014 at 10:53:30PM +0100, Mitchel Humpherys wrote:
> Currently, we provide the iommu_ops.iova_to_phys service by doing a
> table walk in software to translate IO virtual addresses to physical
> addresses. On SMMUs that support it, it can be useful to ask the SMMU
> itself to do the translation. This can be used to warm the TLBs for an
> SMMU. It can also be useful for testing and hardware validation.
> 
> Since the address translation registers are optional on SMMUv2, only
> enable hardware translations when using SMMUv1 or when SMMU_IDR0.S1TS=1
> and SMMU_IDR0.ATOSNS=0, as described in the ARM SMMU v1-v2 spec.

[...]

> -static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain,
> +static phys_addr_t arm_smmu_iova_to_phys_soft(struct iommu_domain *domain,
>  					 dma_addr_t iova)
>  {
>  	pgd_t *pgdp, pgd;
> @@ -1557,6 +1569,66 @@ static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain,
>  	return __pfn_to_phys(pte_pfn(pte)) | (iova & ~PAGE_MASK);
>  }
>  
> +static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain,
> +					dma_addr_t iova)
> +{
> +	struct arm_smmu_domain *smmu_domain = domain->priv;
> +	struct arm_smmu_device *smmu = smmu_domain->smmu;
> +	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
> +	struct device *dev = smmu->dev;
> +	void __iomem *cb_base;
> +	u32 tmp;
> +	u64 phys;
> +	unsigned long flags;
> +
> +	cb_base = ARM_SMMU_CB_BASE(smmu) + ARM_SMMU_CB(smmu, cfg->cbndx);
> +
> +	spin_lock_irqsave(&smmu_domain->lock, flags);
> +
> +	if (smmu->version == 1) {
> +		u32 reg = iova & ~0xfff;
> +		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_LO);
> +	} else {
> +		u32 reg = iova & ~0xfff;
> +		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_LO);
> +		reg = (iova & ~0xfff) >> 32;
> +		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_HI);
> +	}
> +
> +	if (readl_poll_timeout_atomic(cb_base + ARM_SMMU_CB_ATSR, tmp,
> +				!(tmp & ATSR_ACTIVE), 5, 50)) {
> +		dev_err(dev,
> +			"iova to phys timed out on 0x%pa. Falling back to software table walk.\n",
> +			&iova);
> +		return arm_smmu_iova_to_phys_soft(domain, iova);

Missing unlock here.

> +	}
> +
> +	phys = readl_relaxed(cb_base + ARM_SMMU_CB_PAR_LO);
> +	phys |= ((u64) readl_relaxed(cb_base + ARM_SMMU_CB_PAR_HI)) << 32;
> +
> +	spin_unlock_irqrestore(&smmu_domain->lock, flags);
> +
> +	if (phys & CB_PAR_F) {
> +		dev_err(dev, "translation fault!\n");
> +		dev_err(dev, "PAR = 0x%llx\n", phys);
> +		phys = 0;
> +	} else {
> +		phys = (phys & 0xfffffff000ULL) | (iova & 0x00000fff);
> +	}

I think your mask is too big here -- SMMUv2 puts the MAIR in the top byte.
It probably makes more sense to use PHYS_MASK & ~0xfffUL.

Will