Skip to contentSkip to conversation

PCIe Host request to reserve IOVA

PATCH Applied Srinath Mannam 3 participants 9 messages May 3, 2019 – May 7, 2019

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

v6 cover PCIe Host request to reserve IOVA

This patch set will reserve IOVA addresses for DMA memory holes.

The IPROC host controller allows only a few ranges of physical address as inbound PCI addresses which are listed through dma-ranges DT property. Added dma_ranges list field of PCI host bridge structure to hold these allowed inbound address ranges in sorted order.

Process this list and reserve IOVA addresses that are not present in its resource entries (ie DMA memory holes) to prevent allocating IOVA addresses that cannot be allocated as inbound addresses.

This patch set is based on Linux-5.1-rc3.

Changes from v5:
- Addressed Robin Murphy, Lorenzo review comments.
- Error handling in dma ranges list processing.
- Used commit messages given by Lorenzo to all patches.

Changes from v4:
- Addressed Bjorn, Robin Murphy and Auger Eric review comments.
- Commit message modification.
- Change DMA_BIT_MASK to "~(dma_addr_t)0".

Changes from v3:
- Addressed Robin Murphy review comments.
- pcie-iproc: parse dma-ranges and make sorted resource list.
- dma-iommu: process list and reserve gaps between entries

Changes from v2:
- Patch set rebased to Linux-5.0-rc2

Changes from v1:
- Addressed Oza review comments.

Srinath Mannam (3):
PCI: Add dma_ranges window list
iommu/dma: Reserve IOVA for PCIe inaccessible DMA address
PCI: iproc: Add sorted dma ranges resource entries to host bridge

 drivers/iommu/dma-iommu.c           | 35 ++++++++++++++++++++++++++---
 drivers/pci/controller/pcie-iproc.c | 44 ++++++++++++++++++++++++++++++++++++-
 drivers/pci/probe.c                 |  3 +++
 include/linux/pci.h                 |  1 +
 4 files changed, 79 insertions(+), 4 deletions(-)
Signature · 1 line
2.7.4
#1 lore raw

PATCH v6 1/3 PCI: Add dma_ranges window list

Add a dma_ranges field in PCI host bridge structure to hold resource entries list of memory regions in sorted order representing memory ranges that can be accessed through DMA transactions.

---
 drivers/pci/probe.c | 3 +++
 include/linux/pci.h | 1 +
 2 files changed, 4 insertions(+)
drivers/pci/probe.c +3 -0
index 7e12d01..72563c1 100644
@@ -595,6 +595,7 @@ struct pci_host_bridge *pci_alloc_host_bridge(size_t priv)
 		return NULL;
 
 	INIT_LIST_HEAD(&bridge->windows);
+	INIT_LIST_HEAD(&bridge->dma_ranges);
 	bridge->dev.release = pci_release_host_bridge_dev;
 
 	/*
@@ -623,6 +624,7 @@ struct pci_host_bridge *devm_pci_alloc_host_bridge(struct device *dev,
 		return NULL;
 
 	INIT_LIST_HEAD(&bridge->windows);
+	INIT_LIST_HEAD(&bridge->dma_ranges);
 	bridge->dev.release = devm_pci_release_host_bridge_dev;
 
 	return bridge;
@@ -632,6 +634,7 @@ EXPORT_SYMBOL(devm_pci_alloc_host_bridge);
 void pci_free_host_bridge(struct pci_host_bridge *bridge)
 {
 	pci_free_resource_list(&bridge->windows);
+	pci_free_resource_list(&bridge->dma_ranges);
 
 	kfree(bridge);
 }
include/linux/pci.h +1 -0
index 7744821..bba0a29 100644
@@ -490,6 +490,7 @@ struct pci_host_bridge {
 	void		*sysdata;
 	int		busnr;
 	struct list_head windows;	/* resource_entry */
+	struct list_head dma_ranges;	/* dma ranges resource list */
 	u8 (*swizzle_irq)(struct pci_dev *, u8 *); /* Platform IRQ swizzler */
 	int (*map_irq)(const struct pci_dev *, u8, u8);
 	void (*release_fn)(struct pci_host_bridge *);
Signature · 1 line
2.7.4
#2 lore raw

PATCH v6 2/3 iommu/dma: Reserve IOVA for PCIe inaccessible DMA address

In reply to Srinath Mannam

The dma_ranges list field of PCI host bridge structure has resource entries in sorted order representing address ranges allowed for DMA transfers.

Process the list and reserve IOVA addresses that are not present in its resource entries (ie DMA memory holes) to prevent allocating IOVA addresses that cannot be accessed by PCI devices.

---
 drivers/iommu/dma-iommu.c | 35 ++++++++++++++++++++++++++++++++---
 1 file changed, 32 insertions(+), 3 deletions(-)
drivers/iommu/dma-iommu.c +32 -3
index 77aabe6..954ae11 100644
@@ -206,12 +206,13 @@ static int cookie_init_hw_msi_region(struct iommu_dma_cookie *cookie,
 	return 0;
 }
 
-static void iova_reserve_pci_windows(struct pci_dev *dev,
+static int iova_reserve_pci_windows(struct pci_dev *dev,
 		struct iova_domain *iovad)
 {
 	struct pci_host_bridge *bridge = pci_find_host_bridge(dev->bus);
 	struct resource_entry *window;
 	unsigned long lo, hi;
+	phys_addr_t start = 0, end;
 
 	resource_list_for_each_entry(window, &bridge->windows) {
 		if (resource_type(window->res) != IORESOURCE_MEM)
@@ -221,6 +222,31 @@ static void iova_reserve_pci_windows(struct pci_dev *dev,
 		hi = iova_pfn(iovad, window->res->end - window->offset);
 		reserve_iova(iovad, lo, hi);
 	}
+
+	/* Get reserved DMA windows from host bridge */
+	resource_list_for_each_entry(window, &bridge->dma_ranges) {
+		end = window->res->start - window->offset;
+resv_iova:
+		if (end > start) {
+			lo = iova_pfn(iovad, start);
+			hi = iova_pfn(iovad, end);
+			reserve_iova(iovad, lo, hi);
+		} else {
+			/* dma_ranges list should be sorted */
+			dev_err(&dev->dev, "Failed to reserve IOVA\n");
+			return -EINVAL;
+		}
+
+		start = window->res->end - window->offset + 1;
+		/* If window is last entry */
+		if (window->node.next == &bridge->dma_ranges &&
+		    end != ~(dma_addr_t)0) {
+			end = ~(dma_addr_t)0;
+			goto resv_iova;
+		}
+	}
+
+	return 0;
 }
 
 static int iova_reserve_iommu_regions(struct device *dev,
@@ -232,8 +258,11 @@ static int iova_reserve_iommu_regions(struct device *dev,
 	LIST_HEAD(resv_regions);
 	int ret = 0;
 
-	if (dev_is_pci(dev))
-		iova_reserve_pci_windows(to_pci_dev(dev), iovad);
+	if (dev_is_pci(dev)) {
+		ret = iova_reserve_pci_windows(to_pci_dev(dev), iovad);
+		if (ret)
+			return ret;
+	}
 
 	iommu_get_resv_regions(dev, &resv_regions);
 	list_for_each_entry(region, &resv_regions, list) {
Signature · 1 line
2.7.4
#3 lore raw

PATCH v6 3/3 PCI: iproc: Add sorted dma ranges resource entries to host bridge

In reply to Srinath Mannam

The IPROC host controller allows only a subset of physical address space as target of inbound PCI memory transactions addresses.

PCIe devices memory transactions targeting memory regions that are not allowed for inbound transactions in the host controller are rejected by the host controller and cannot reach the upstream buses.

Firmware device tree description defines the DMA ranges that are addressable by devices DMA transactions; parse the device tree dma-ranges property and add its ranges to the PCI host bridge dma_ranges list; the iova_reserve_pci_windows() call in the driver will reserve the IOVA address ranges that are not addressable (ie memory holes in the dma-ranges set) so that they are not allocated to PCI devices for DMA transfers.

All allowed address ranges are listed in dma-ranges DT parameter.

Example:

dma-ranges = < \
0x43000000 0x00 0x80000000 0x00 0x80000000 0x00 0x80000000 \
0x43000000 0x08 0x00000000 0x08 0x00000000 0x08 0x00000000 \
0x43000000 0x80 0x00000000 0x80 0x00000000 0x40 0x00000000>

In the above example of dma-ranges, memory address from

0x0 - 0x80000000,
0x100000000 - 0x800000000,
0x1000000000 - 0x8000000000 and
0x10000000000 - 0xffffffffffffffff.

are not allowed to be used as inbound addresses.

---
 drivers/pci/controller/pcie-iproc.c | 44 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)
drivers/pci/controller/pcie-iproc.c +43 -1
index c20fd6b..94ba5c0 100644
@@ -1146,11 +1146,43 @@ static int iproc_pcie_setup_ib(struct iproc_pcie *pcie,
 	return ret;
 }
 
+static int
+iproc_pcie_add_dma_range(struct device *dev, struct list_head *resources,
+			 struct of_pci_range *range)
+{
+	struct resource *res;
+	struct resource_entry *entry, *tmp;
+	struct list_head *head = resources;
+
+	res = devm_kzalloc(dev, sizeof(struct resource), GFP_KERNEL);
+	if (!res)
+		return -ENOMEM;
+
+	resource_list_for_each_entry(tmp, resources) {
+		if (tmp->res->start < range->cpu_addr)
+			head = &tmp->node;
+	}
+
+	res->start = range->cpu_addr;
+	res->end = res->start + range->size - 1;
+
+	entry = resource_list_create_entry(res, 0);
+	if (!entry)
+		return -ENOMEM;
+
+	entry->offset = res->start - range->cpu_addr;
+	resource_list_add(entry, head);
+
+	return 0;
+}
+
 static int iproc_pcie_map_dma_ranges(struct iproc_pcie *pcie)
 {
+	struct pci_host_bridge *host = pci_host_bridge_from_priv(pcie);
 	struct of_pci_range range;
 	struct of_pci_range_parser parser;
 	int ret;
+	LIST_HEAD(resources);
 
 	/* Get the dma-ranges from DT */
 	ret = of_pci_dma_range_parser_init(&parser, pcie->dev->of_node);
@@ -1158,13 +1190,23 @@ static int iproc_pcie_map_dma_ranges(struct iproc_pcie *pcie)
 		return ret;
 
 	for_each_of_pci_range(&parser, &range) {
+		ret = iproc_pcie_add_dma_range(pcie->dev,
+					       &resources,
+					       &range);
+		if (ret)
+			goto out;
 		/* Each range entry corresponds to an inbound mapping region */
 		ret = iproc_pcie_setup_ib(pcie, &range, IPROC_PCIE_IB_MAP_MEM);
 		if (ret)
-			return ret;
+			goto out;
 	}
 
+	list_splice_init(&resources, &host->dma_ranges);
+
 	return 0;
+out:
+	pci_free_resource_list(&resources);
+	return ret;
 }
 
 static int iproce_pcie_get_msi(struct iproc_pcie *pcie,
Signature · 1 line
2.7.4
#4 lore raw

Re: [PATCH v6 3/3] PCI: iproc: Add sorted dma ranges resource entries to host bridge

Quoted 49 lines · Srinath Mannam

The IPROC host controller allows only a subset of physical address space as target of inbound PCI memory transactions addresses.

PCIe devices memory transactions targeting memory regions that are not allowed for inbound transactions in the host controller are rejected by the host controller and cannot reach the upstream buses.

Firmware device tree description defines the DMA ranges that are addressable by devices DMA transactions; parse the device tree dma-ranges property and add its ranges to the PCI host bridge dma_ranges list; the iova_reserve_pci_windows() call in the driver will reserve the IOVA address ranges that are not addressable (ie memory holes in the dma-ranges set) so that they are not allocated to PCI devices for DMA transfers.

All allowed address ranges are listed in dma-ranges DT parameter.

Example:

dma-ranges = < \
0x43000000 0x00 0x80000000 0x00 0x80000000 0x00 0x80000000 \
0x43000000 0x08 0x00000000 0x08 0x00000000 0x08 0x00000000 \
0x43000000 0x80 0x00000000 0x80 0x00000000 0x40 0x00000000>

In the above example of dma-ranges, memory address from

0x0 - 0x80000000,
0x100000000 - 0x800000000,
0x1000000000 - 0x8000000000 and
0x10000000000 - 0xffffffffffffffff.

are not allowed to be used as inbound addresses.

---
 drivers/pci/controller/pcie-iproc.c | 44 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)
drivers/pci/controller/pcie-iproc.c +0 -0
index c20fd6b..94ba5c0 100644

drivers/pci/controller/pcie-iproc.c static int iproc_pcie_setup_ib(struct iproc_pcie *pcie, · line 1151

Quoted 6 lines
drivers/pci/controller/pcie-iproc.c +3 -0
@@ -1146,11 +1146,43 @@ static int iproc_pcie_setup_ib(struct iproc_pcie *pcie,
 	return ret;
 }
 
+static int
+iproc_pcie_add_dma_range(struct device *dev, struct list_head *resources,
+			 struct of_pci_range *range)

Just FYI, I cherry-picked these commits from Lorenzo's branch to fix the formatting of this prototype to match the rest of the file, e.g.:

drivers/pci/controller/pcie-iproc.c static int iproc_pcie_setup_ib(struct iproc_pcie *pcie,

Quoted 3 lines
drivers/pci/controller/pcie-iproc.c +0 -0
 static int iproc_pcie_map_dma_ranges(struct iproc_pcie *pcie)
...
 static int iproce_pcie_get_msi(struct iproc_pcie *pcie,

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu

#5 lore raw

Re: [PATCH v6 3/3] PCI: iproc: Add sorted dma ranges resource entries to host bridge

Quoted 53 lines · Bjorn Helgaas
Quoted 52 lines · Srinath Mannam

The IPROC host controller allows only a subset of physical address space as target of inbound PCI memory transactions addresses.

PCIe devices memory transactions targeting memory regions that are not allowed for inbound transactions in the host controller are rejected by the host controller and cannot reach the upstream buses.

Firmware device tree description defines the DMA ranges that are addressable by devices DMA transactions; parse the device tree dma-ranges property and add its ranges to the PCI host bridge dma_ranges list; the iova_reserve_pci_windows() call in the driver will reserve the IOVA address ranges that are not addressable (ie memory holes in the dma-ranges set) so that they are not allocated to PCI devices for DMA transfers.

All allowed address ranges are listed in dma-ranges DT parameter.

Example:

dma-ranges = < \
0x43000000 0x00 0x80000000 0x00 0x80000000 0x00 0x80000000 \
0x43000000 0x08 0x00000000 0x08 0x00000000 0x08 0x00000000 \
0x43000000 0x80 0x00000000 0x80 0x00000000 0x40 0x00000000>

In the above example of dma-ranges, memory address from

0x0 - 0x80000000,
0x100000000 - 0x800000000,
0x1000000000 - 0x8000000000 and
0x10000000000 - 0xffffffffffffffff.

are not allowed to be used as inbound addresses.

---
 drivers/pci/controller/pcie-iproc.c | 44 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)
drivers/pci/controller/pcie-iproc.c +0 -0
index c20fd6b..94ba5c0 100644
@@ -1146,11 +1146,43 @@ static int iproc_pcie_setup_ib(struct iproc_pcie *pcie,
 	return ret;
 }
 
Quoted 6 lines

drivers/pci/controller/pcie-iproc.c static int iproc_pcie_setup_ib(struct iproc_pcie *pcie, · line 1151

Quoted 3 lines
drivers/pci/controller/pcie-iproc.c +3 -0
+static int
+iproc_pcie_add_dma_range(struct device *dev, struct list_head *resources,
+			 struct of_pci_range *range)

Just FYI, I cherry-picked these commits from Lorenzo's branch to fix the formatting of this prototype to match the rest of the file, e.g.:

Thank you, I noticed too but I forgot to update it before merging v6 from the list.

Thanks,
Lorenzo

Quoted 3 lines
Quoted 3 lines
drivers/pci/controller/pcie-iproc.c +0 -0
 static int iproc_pcie_map_dma_ranges(struct iproc_pcie *pcie)
...
 static int iproce_pcie_get_msi(struct iproc_pcie *pcie,
#6 lore raw

Re: [PATCH v6 3/3] PCI: iproc: Add sorted dma ranges resource entries to host bridge

Hi Bjorn,

Thank you.

Regards,
Srinath.

Quoted 71 lines · Lorenzo Pieralisi
Quoted 53 lines · Bjorn Helgaas
Quoted 52 lines · Srinath Mannam

The IPROC host controller allows only a subset of physical address space as target of inbound PCI memory transactions addresses.

PCIe devices memory transactions targeting memory regions that are not allowed for inbound transactions in the host controller are rejected by the host controller and cannot reach the upstream buses.

Firmware device tree description defines the DMA ranges that are addressable by devices DMA transactions; parse the device tree dma-ranges property and add its ranges to the PCI host bridge dma_ranges list; the iova_reserve_pci_windows() call in the driver will reserve the IOVA address ranges that are not addressable (ie memory holes in the dma-ranges set) so that they are not allocated to PCI devices for DMA transfers.

All allowed address ranges are listed in dma-ranges DT parameter.

Example:

dma-ranges = < \
0x43000000 0x00 0x80000000 0x00 0x80000000 0x00 0x80000000 \
0x43000000 0x08 0x00000000 0x08 0x00000000 0x08 0x00000000 \
0x43000000 0x80 0x00000000 0x80 0x00000000 0x40 0x00000000>

In the above example of dma-ranges, memory address from

0x0 - 0x80000000,
0x100000000 - 0x800000000,
0x1000000000 - 0x8000000000 and
0x10000000000 - 0xffffffffffffffff.

are not allowed to be used as inbound addresses.

---
 drivers/pci/controller/pcie-iproc.c | 44 ++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)
drivers/pci/controller/pcie-iproc.c +0 -0
index c20fd6b..94ba5c0 100644
@@ -1146,11 +1146,43 @@ static int iproc_pcie_setup_ib(struct iproc_pcie *pcie,
    return ret;
 }
 
Quoted 6 lines

drivers/pci/controller/pcie-iproc.c static int iproc_pcie_setup_ib(struct iproc_pcie *pcie, · line 1151

Quoted 3 lines
drivers/pci/controller/pcie-iproc.c +3 -0
+static int
+iproc_pcie_add_dma_range(struct device *dev, struct list_head *resources,
+                    struct of_pci_range *range)

Just FYI, I cherry-picked these commits from Lorenzo's branch to fix the formatting of this prototype to match the rest of the file, e.g.:

Thank you, I noticed too but I forgot to update it before merging v6 from the list.

Thanks,
Lorenzo

Quoted 3 lines
Quoted 3 lines
drivers/pci/controller/pcie-iproc.c +0 -0
 static int iproc_pcie_map_dma_ranges(struct iproc_pcie *pcie)
...
 static int iproce_pcie_get_msi(struct iproc_pcie *pcie,
#7 lore raw
In reply to Srinath Mannam
Quoted 39 lines · Srinath Mannam

This patch set will reserve IOVA addresses for DMA memory holes.

The IPROC host controller allows only a few ranges of physical address as inbound PCI addresses which are listed through dma-ranges DT property. Added dma_ranges list field of PCI host bridge structure to hold these allowed inbound address ranges in sorted order.

Process this list and reserve IOVA addresses that are not present in its resource entries (ie DMA memory holes) to prevent allocating IOVA addresses that cannot be allocated as inbound addresses.

This patch set is based on Linux-5.1-rc3.

Changes from v5:
- Addressed Robin Murphy, Lorenzo review comments.
- Error handling in dma ranges list processing.
- Used commit messages given by Lorenzo to all patches.

Changes from v4:
- Addressed Bjorn, Robin Murphy and Auger Eric review comments.
- Commit message modification.
- Change DMA_BIT_MASK to "~(dma_addr_t)0".

Changes from v3:
- Addressed Robin Murphy review comments.
- pcie-iproc: parse dma-ranges and make sorted resource list.
- dma-iommu: process list and reserve gaps between entries

Changes from v2:
- Patch set rebased to Linux-5.0-rc2

Changes from v1:
- Addressed Oza review comments.

Srinath Mannam (3):
PCI: Add dma_ranges window list
iommu/dma: Reserve IOVA for PCIe inaccessible DMA address
PCI: iproc: Add sorted dma ranges resource entries to host bridge

Quoted 5 lines
 drivers/iommu/dma-iommu.c           | 35 ++++++++++++++++++++++++++---
 drivers/pci/controller/pcie-iproc.c | 44 ++++++++++++++++++++++++++++++++++++-
 drivers/pci/probe.c                 |  3 +++
 include/linux/pci.h                 |  1 +
 4 files changed, 79 insertions(+), 4 deletions(-)

I have applied the series to pci/iova-dma-ranges, targeting v5.2, thanks.

Lorenzo

#8 lore raw

Hi Lorenzo,

Thanks a lot.

Regards,
Srinath.

Quoted 51 lines · Lorenzo Pieralisi
Quoted 39 lines · Srinath Mannam

This patch set will reserve IOVA addresses for DMA memory holes.

The IPROC host controller allows only a few ranges of physical address as inbound PCI addresses which are listed through dma-ranges DT property. Added dma_ranges list field of PCI host bridge structure to hold these allowed inbound address ranges in sorted order.

Process this list and reserve IOVA addresses that are not present in its resource entries (ie DMA memory holes) to prevent allocating IOVA addresses that cannot be allocated as inbound addresses.

This patch set is based on Linux-5.1-rc3.

Changes from v5:
- Addressed Robin Murphy, Lorenzo review comments.
- Error handling in dma ranges list processing.
- Used commit messages given by Lorenzo to all patches.

Changes from v4:
- Addressed Bjorn, Robin Murphy and Auger Eric review comments.
- Commit message modification.
- Change DMA_BIT_MASK to "~(dma_addr_t)0".

Changes from v3:
- Addressed Robin Murphy review comments.
- pcie-iproc: parse dma-ranges and make sorted resource list.
- dma-iommu: process list and reserve gaps between entries

Changes from v2:
- Patch set rebased to Linux-5.0-rc2

Changes from v1:
- Addressed Oza review comments.

Srinath Mannam (3):
PCI: Add dma_ranges window list
iommu/dma: Reserve IOVA for PCIe inaccessible DMA address
PCI: iproc: Add sorted dma ranges resource entries to host bridge

Quoted 5 lines
 drivers/iommu/dma-iommu.c           | 35 ++++++++++++++++++++++++++---
 drivers/pci/controller/pcie-iproc.c | 44 ++++++++++++++++++++++++++++++++++++-
 drivers/pci/probe.c                 |  3 +++
 include/linux/pci.h                 |  1 +
 4 files changed, 79 insertions(+), 4 deletions(-)

I have applied the series to pci/iova-dma-ranges, targeting v5.2, thanks.

Lorenzo