[RFC PATCH v4 4/7] PCI: Modify resource_alignment to support multiple devices

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



When vfio passthrough a PCI device of which MMIO BARs
are smaller than PAGE_SIZE, guest will not handle the
mmio accesses to the BARs which leads to mmio emulations
in host.

This is because vfio will not allow to passthrough one
BAR's mmio page which may be shared with other BARs.

To solve this performance issue, this patch modifies
resource_alignment to support syntax where multiple
devices get the same alignment. So we can use something
like "pci=resource_alignment=*:*:*.*:noresize" to
enforce the alignment of all MMIO BARs to be at least
PAGE_SIZE so that one BAR's mmio page would not be
shared with other BARs.

Signed-off-by: Yongji Xie <xyjxie@xxxxxxxxxxxxxxxxxx>
---
 Documentation/kernel-parameters.txt |    2 +
 drivers/pci/pci.c                   |   90 ++++++++++++++++++++++++++++++-----
 include/linux/pci.h                 |    4 ++
 3 files changed, 85 insertions(+), 11 deletions(-)

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 8028631..74b38ab 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -2918,6 +2918,8 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
 				aligned memory resources.
 				If <order of align> is not specified,
 				PAGE_SIZE is used as alignment.
+				<domain>, <bus>, <slot> and <func> can be set to
+				"*" which means match all values.
 				PCI-PCI bridge can be specified, if resource
 				windows need to be expanded.
 				noresize: Don't change the resources' sizes when
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 760cce5..44ab59f 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -102,6 +102,8 @@ unsigned int pcibios_max_latency = 255;
 /* If set, the PCIe ARI capability will not be used. */
 static bool pcie_ari_disabled;
 
+bool pci_resources_page_aligned;
+
 /**
  * pci_bus_max_busnr - returns maximum PCI bus number of given bus' children
  * @bus: pointer to PCI bus structure to search
@@ -4604,6 +4606,7 @@ static resource_size_t pci_specified_resource_alignment(struct pci_dev *dev,
 	int seg, bus, slot, func, align_order, count;
 	resource_size_t align = 0;
 	char *p;
+	bool invalid = false;
 
 	spin_lock(&resource_alignment_lock);
 	p = resource_alignment_param;
@@ -4615,16 +4618,49 @@ static resource_size_t pci_specified_resource_alignment(struct pci_dev *dev,
 		} else {
 			align_order = -1;
 		}
-		if (sscanf(p, "%x:%x:%x.%x%n",
-			&seg, &bus, &slot, &func, &count) != 4) {
+		if (p[0] == '*' && p[1] == ':') {
+			seg = -1;
+			count = 1;
+		} else if (sscanf(p, "%x%n", &seg, &count) != 1 ||
+				p[count] != ':') {
+			invalid = true;
+			break;
+		}
+		p += count + 1;
+		if (*p == '*') {
+			bus = -1;
+			count = 1;
+		} else if (sscanf(p, "%x%n", &bus, &count) != 1) {
+			invalid = true;
+			break;
+		}
+		p += count;
+		if (*p == '.') {
+			slot = bus;
+			bus = seg;
 			seg = 0;
-			if (sscanf(p, "%x:%x.%x%n",
-					&bus, &slot, &func, &count) != 3) {
-				/* Invalid format */
-				printk(KERN_ERR "PCI: Can't parse resource_alignment parameter: %s\n",
-					p);
+			p++;
+		} else if (*p == ':') {
+			p++;
+			if (p[0] == '*' && p[1] == '.') {
+				slot = -1;
+				count = 1;
+			} else if (sscanf(p, "%x%n", &slot, &count) != 1 ||
+					p[count] != '.') {
+				invalid = true;
 				break;
 			}
+			p += count + 1;
+		} else {
+			invalid = true;
+			break;
+		}
+		if (*p == '*') {
+			func = -1;
+			count = 1;
+		} else if (sscanf(p, "%x%n", &func, &count) != 1) {
+			invalid = true;
+			break;
 		}
 		p += count;
 		if (!strncmp(p, ":noresize", 9)) {
@@ -4632,23 +4668,34 @@ static resource_size_t pci_specified_resource_alignment(struct pci_dev *dev,
 			p += 9;
 		} else
 			*resize = true;
-		if (seg == pci_domain_nr(dev->bus) &&
-			bus == dev->bus->number &&
-			slot == PCI_SLOT(dev->devfn) &&
-			func == PCI_FUNC(dev->devfn)) {
+		if ((seg == pci_domain_nr(dev->bus) || seg == -1) &&
+			(bus == dev->bus->number || bus == -1) &&
+			(slot == PCI_SLOT(dev->devfn) || slot == -1) &&
+			(func == PCI_FUNC(dev->devfn) || func == -1)) {
 			if (align_order == -1)
 				align = PAGE_SIZE;
 			else
 				align = 1 << align_order;
+			if (!pci_resources_page_aligned &&
+				(align >= PAGE_SIZE &&
+				seg == -1 && bus == -1 &&
+				slot == -1 && func == -1))
+				pci_resources_page_aligned = true;
 			/* Found */
 			break;
 		}
 		if (*p != ';' && *p != ',') {
 			/* End of param or invalid format */
+			invalid = true;
 			break;
 		}
 		p++;
 	}
+	if (invalid == true) {
+		/* Invalid format */
+		printk(KERN_ERR "PCI: Can't parse resource_alignment parameter:%s\n",
+				p);
+	}
 	spin_unlock(&resource_alignment_lock);
 	return align;
 }
@@ -4769,6 +4816,27 @@ static int __init pci_resource_alignment_sysfs_init(void)
 }
 late_initcall(pci_resource_alignment_sysfs_init);
 
+/*
+ * This function checks whether PCI BARs' mmio page will be shared
+ * with other BARs.
+ */
+bool pci_resources_share_page(struct pci_dev *dev, int resno)
+{
+	struct resource *res = dev->resource + resno;
+
+	if (resource_size(res) >= PAGE_SIZE)
+		return false;
+	if (pci_resources_page_aligned && !(res->start & ~PAGE_MASK) &&
+		res->flags & IORESOURCE_MEM) {
+		if (res->sibling)
+			return (res->sibling->start & ~PAGE_MASK);
+		else
+			return false;
+	}
+	return true;
+}
+EXPORT_SYMBOL_GPL(pci_resources_share_page);
+
 static void pci_no_domains(void)
 {
 #ifdef CONFIG_PCI_DOMAINS
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 2771625..064a1b6 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -1511,6 +1511,10 @@ static inline int pci_get_new_domain_nr(void) { return -ENOSYS; }
 	 (pci_resource_end((dev), (bar)) -		\
 	  pci_resource_start((dev), (bar)) + 1))
 
+extern bool pci_resources_page_aligned;
+
+bool pci_resources_share_page(struct pci_dev *dev, int resno);
+
 /* Similar to the helpers above, these manipulate per-pci_dev
  * driver-specific data.  They are really just a wrapper around
  * the generic device structure functions of these calls.
-- 
1.7.9.5

--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html



[Index of Archives]     [DMA Engine]     [Linux Coverity]     [Linux USB]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Greybus]

  Powered by Linux