Front-end driver for PCIe Pass-through on Hyper-V

index c85235e..d0e8832 100644
@@ -72,10 +72,7 @@ static struct notifier_block hyperv_panic_block = {
 	.notifier_call = hyperv_panic_event,
 };
 
-struct resource hyperv_mmio = {
-	.name  = "hyperv mmio",
-	.flags = IORESOURCE_MEM,
-};
+struct resource *hyperv_mmio;
 EXPORT_SYMBOL_GPL(hyperv_mmio);
 
 static int vmbus_exists(void)

@@ -982,30 +979,105 @@ void vmbus_device_unregister(struct hv_device *device_obj)
 
 
 /*
- * VMBUS is an acpi enumerated device. Get the the information we
+ * VMBUS is an acpi enumerated device. Get the information we
  * need from DSDT.
  */
-
+#define VTPM_BASE_ADDRESS 0xfed40000
 static acpi_status vmbus_walk_resources(struct acpi_resource *res, void *ctx)
 {
+	resource_size_t start = 0;
+	resource_size_t end = 0;
+	struct resource *new_res;
+	struct resource **old_res = &hyperv_mmio;
+	struct resource **prev_res = NULL;
+
 	switch (res->type) {
 	case ACPI_RESOURCE_TYPE_IRQ:
 		irq = res->data.irq.interrupts[0];
+		return AE_OK;
+
+	/*
+	 * "Address" descriptors are for bus windows. Ignore
+	 * "memory" descriptors, which are for registers on
+	 * devices.
+	 */
+	case ACPI_RESOURCE_TYPE_ADDRESS32:
+		start = res->data.address32.address.minimum;
+		end = res->data.address32.address.maximum;
 		break;
 
 	case ACPI_RESOURCE_TYPE_ADDRESS64:
-		hyperv_mmio.start = res->data.address64.address.minimum;
-		hyperv_mmio.end = res->data.address64.address.maximum;
+		start = res->data.address64.address.minimum;
+		end = res->data.address64.address.maximum;
 		break;
+
+	default:
+		/* Unused resource type */
+		return AE_OK;
+
 	}
+	/*
+	 * Ignore ranges that are below 1MB, as they're not
+	 * necessary or useful here.
+	 */
+	if (end < 0x100000)
+		return AE_OK;
+
+	new_res = kzalloc(sizeof(*new_res), GFP_ATOMIC);
+	if (!new_res)
+		return AE_NO_MEMORY;
+
+	/* If this range overlaps the virtual TPM, truncate it. */
+	if (end > VTPM_BASE_ADDRESS && start < VTPM_BASE_ADDRESS)
+		end = VTPM_BASE_ADDRESS;
+
+	new_res->name = "hyperv mmio";
+	new_res->flags = IORESOURCE_MEM;
+	new_res->start = start;
+	new_res->end = end;
+
+	do {
+		if (!*old_res) {
+			*old_res = new_res;
+			break;
+		}
+
+		if ((*old_res)->end < new_res->start) {
+			new_res->sibling = *old_res;
+			if (prev_res)
+				(*prev_res)->sibling = new_res;
+			*old_res = new_res;
+			break;
+		}
+
+		prev_res = old_res;
+		old_res = &(*old_res)->sibling;
+
+	} while (1);
 
 	return AE_OK;
 }
 
+static int vmbus_acpi_remove(struct acpi_device *device)
+{
+	struct resource *cur_res;
+	struct resource *next_res;
+
+	if (hyperv_mmio) {
+		for (cur_res = hyperv_mmio; cur_res; cur_res = next_res) {
+			next_res = cur_res->sibling;
+			kfree(cur_res);
+		}
+	}
+
+	return 0;
+}
+
 static int vmbus_acpi_add(struct acpi_device *device)
 {
 	acpi_status result;
 	int ret_val = -ENODEV;
+	struct acpi_device *ancestor;
 
 	hv_acpi_dev = device;
 

@@ -1015,23 +1087,25 @@ static int vmbus_acpi_add(struct acpi_device *device)
 	if (ACPI_FAILURE(result))
 		goto acpi_walk_err;
 	/*
-	 * The parent of the vmbus acpi device (Gen2 firmware) is the VMOD that
-	 * has the mmio ranges. Get that.
+	 * Some ancestor of the vmbus acpi device (Gen1 or Gen2
+	 * firmware) is the VMOD that has the mmio ranges. Get that.
 	 */
-	if (device->parent) {
-		result = acpi_walk_resources(device->parent->handle,
+	for (ancestor = device->parent; ancestor; ancestor = ancestor->parent) {
+		result = acpi_walk_resources(ancestor->handle,
 					METHOD_NAME__CRS,
 					vmbus_walk_resources, NULL);
 
 		if (ACPI_FAILURE(result))
-			goto acpi_walk_err;
-		if (hyperv_mmio.start && hyperv_mmio.end)
-			request_resource(&iomem_resource, &hyperv_mmio);
+			continue;
+		if (hyperv_mmio)
+			break;
 	}
 	ret_val = 0;
 
 acpi_walk_err:
 	complete(&probe_event);
+	if (ret_val)
+		vmbus_acpi_remove(device);
 	return ret_val;
 }
 

@@ -1047,6 +1121,7 @@ static struct acpi_driver vmbus_acpi_driver = {
 	.ids = vmbus_acpi_device_ids,
 	.ops = {
 		.add = vmbus_acpi_add,
+		.remove = vmbus_acpi_remove,
 	},
 };
 

index 807ee22..b54ee1c 100644
@@ -688,7 +688,7 @@ static int hvfb_getmem(struct fb_info *info)
 	par->mem.name = KBUILD_MODNAME;
 	par->mem.flags = IORESOURCE_MEM | IORESOURCE_BUSY;
 	if (gen2vm) {
-		ret = allocate_resource(&hyperv_mmio, &par->mem,
+		ret = allocate_resource(hyperv_mmio, &par->mem,
 					screen_fb_size,
 					0, -1,
 					screen_fb_size,

index 902c37a..21fa867 100644
@@ -1245,7 +1245,7 @@ void hv_vss_deinit(void);
 void hv_vss_onchannelcallback(void *);
 void hv_process_channel_removal(struct vmbus_channel *channel, u32 relid);
 
-extern struct resource hyperv_mmio;
+extern struct resource *hyperv_mmio;
 
 /*
  * Negotiated version with the Host.

index d0e8832..067a469 100644
@@ -39,6 +39,7 @@
 #include <asm/mshyperv.h>
 #include <linux/notifier.h>
 #include <linux/ptrace.h>
+#include <linux/screen_info.h>
 #include "hyperv_vmbus.h"
 
 static struct acpi_device  *hv_acpi_dev;

@@ -73,7 +74,6 @@ static struct notifier_block hyperv_panic_block = {
 };
 
 struct resource *hyperv_mmio;
-EXPORT_SYMBOL_GPL(hyperv_mmio);
 
 static int vmbus_exists(void)
 {

@@ -1073,6 +1073,91 @@ static int vmbus_acpi_remove(struct acpi_device *device)
 	return 0;
 }
 
+/**
+ * This function walks the resources granted to VMBus by the
+ * _CRS object in the ACPI namespace underneath the parent
+ * "bridge" whether that's a root PCI bus in the Generation 1
+ * case or a Module Device in the Generation 2 case.  It then
+ * attempts to allocate from the global MMIO pool in a way that
+ * matches the constraints supplied in these parameters and by
+ * that _CRS.
+ *
+ * @param new - If successful, supplied a pointer to the
+ *            allocated MMIO space.
+ * @param device_obj - Identifies the caller
+ * @param min - Minimum guest physical address of the allocation
+ * @param max - Maximum guest physical address
+ * @param size - Size of the range to be allocated
+ * @param align - Alignment of the range to be allocated
+ * @param fb_overlap_ok - Whether this allocation can be allowed
+ *			to overlap the video frame buffer.
+ *
+ * @return int
+ */
+int vmbus_allocate_mmio(struct resource **new, struct hv_device *device_obj,
+			resource_size_t min, resource_size_t max,
+			resource_size_t size, resource_size_t align,
+			bool fb_overlap_ok)
+{
+	struct resource *iter;
+	resource_size_t range_min, range_max, start, local_min, local_max;
+	const char *dev_n = dev_name(&device_obj->device);
+	u32 fb_end = screen_info.lfb_base + (screen_info.lfb_size << 1);
+	int i;
+
+	pr_devel("allocating %llxx bytes alignment %llx in %llx->%llx for %s\n",
+		 size, align, min, max, dev_n);
+
+	for (iter = hyperv_mmio; iter; iter = iter->sibling) {
+
+		if ((iter->start >= max) || (iter->end <= min))
+			continue;
+
+		range_min = iter->start;
+		range_max = iter->end;
+
+		/* If this range overlaps the frame buffer, split it into
+		   two tries. */
+		for (i = 0; i < 2; i++) {
+			local_min = range_min;
+			local_max = range_max;
+			if (fb_overlap_ok || ((range_min >= fb_end) ||
+			      (range_max <= screen_info.lfb_base))) {
+				i++;
+			} else {
+				if ((range_min <= screen_info.lfb_base) &&
+				    (range_max >= screen_info.lfb_base)) {
+
+					/*
+					 * The frame buffer is in this window,
+					 * so trim this into the part that
+					 * preceeds the frame buffer.
+					 */
+					local_max = screen_info.lfb_base-1;
+					range_min = fb_end;
+				} else {
+					range_min = fb_end;
+					continue;
+				}
+			}
+
+			pr_devel("looking between %llx:%llx\n", range_min,
+				 range_max);
+			start = (local_min + align - 1) & ~(align - 1);
+			for (; start + size - 1 <= local_max; start += align) {
+				*new = request_mem_region_exclusive(start, size,
+								    dev_n);
+				if (*new)
+					return 0;
+			}
+		}
+	}
+
+	pr_warn("vmbus_allocate_mmio failed\n");
+	return -ENXIO;
+}
+EXPORT_SYMBOL_GPL(vmbus_allocate_mmio);
+
 static int vmbus_acpi_add(struct acpi_device *device)
 {
 	acpi_status result;

98 lines not on this page · raw message

index 21fa867..ffd98d7 100644
@@ -982,6 +982,11 @@ int __must_check __vmbus_driver_register(struct hv_driver *hv_driver,
 					 const char *mod_name);
 void vmbus_driver_unregister(struct hv_driver *hv_driver);
 
+int vmbus_allocate_mmio(struct resource **new, struct hv_device *device_obj,
+			resource_size_t min, resource_size_t max,
+			resource_size_t size, resource_size_t align,
+			bool fb_overlap_ok);
+
 /**
  * VMBUS_DEVICE - macro used to describe a specific hyperv vmbus device
  *

@@ -1245,8 +1250,6 @@ void hv_vss_deinit(void);
 void hv_vss_onchannelcallback(void *);
 void hv_process_channel_removal(struct vmbus_channel *channel, u32 relid);
 
-extern struct resource *hyperv_mmio;
-
 /*
  * Negotiated version with the Host.
  */

index c163215..bf7789e 100644
@@ -19,5 +19,7 @@ void hyperv_callback_vector(void);
 void hyperv_vector_handler(struct pt_regs *regs);
 void hv_setup_vmbus_irq(void (*handler)(void));
 void hv_remove_vmbus_irq(void);
+void hyperv_install_interrupt_translation(struct x86_msi_ops *new_ops);
+void hyperv_uninstall_interrupt_translation(void);
 
 #endif

index 939155f..816e329 100644
@@ -69,6 +69,36 @@ void hv_remove_vmbus_irq(void)
 }
 EXPORT_SYMBOL_GPL(hv_setup_vmbus_irq);
 EXPORT_SYMBOL_GPL(hv_remove_vmbus_irq);
+
+struct x86_msi_ops old_msi_ops;
+
+void hyperv_install_interrupt_translation(struct x86_msi_ops *new_ops)
+{
+	old_msi_ops.setup_msi_irqs = xchg(&(x86_msi.setup_msi_irqs),
+					  new_ops->setup_msi_irqs);
+	old_msi_ops.compose_msi_msg = xchg(&(x86_msi.compose_msi_msg),
+					   new_ops->compose_msi_msg);
+	old_msi_ops.teardown_msi_irqs = xchg(&(x86_msi.teardown_msi_irqs),
+					     new_ops->teardown_msi_irqs);
+	old_msi_ops.restore_msi_irqs = xchg(&(x86_msi.restore_msi_irqs),
+					    new_ops->restore_msi_irqs);
+
+	new_ops->setup_msi_irqs = old_msi_ops.setup_msi_irqs;
+	new_ops->compose_msi_msg = old_msi_ops.compose_msi_msg;
+	new_ops->teardown_msi_irqs = old_msi_ops.teardown_msi_irqs;
+	new_ops->restore_msi_irqs = old_msi_ops.restore_msi_irqs;
+}
+EXPORT_SYMBOL_GPL(hyperv_install_interrupt_translation);
+
+void hyperv_uninstall_interrupt_translation(void)
+{
+	xchg(&(x86_msi.setup_msi_irqs), old_msi_ops.setup_msi_irqs);
+	xchg(&(x86_msi.compose_msi_msg), old_msi_ops.compose_msi_msg);
+	xchg(&(x86_msi.teardown_msi_irqs), old_msi_ops.teardown_msi_irqs);
+	xchg(&(x86_msi.restore_msi_irqs), old_msi_ops.restore_msi_irqs);
+}
+EXPORT_SYMBOL_GPL(hyperv_uninstall_interrupt_translation);
+
 #endif
 
 static uint32_t  __init ms_hyperv_platform(void)

index 067a469..65e4eb6 100644
@@ -1158,6 +1158,22 @@ int vmbus_allocate_mmio(struct resource **new, struct hv_device *device_obj,
 }
 EXPORT_SYMBOL_GPL(vmbus_allocate_mmio);
 
+/**
+ * This function returns the mapping between the Linux processor
+ * number and
+ * the hypervisor's virtual processor number, useful in making
+ * hypercalls and such that talk about specific processors.
+ *
+ * @param procnum - in Linux terms
+ *
+ * @return int - in Hyper-V terms
+ */
+int vmbus_procnum_to_vpnum(int procnum)
+{
+	return hv_context.vp_index[procnum];
+}
+EXPORT_SYMBOL_GPL(vmbus_procnum_to_vpnum);
+
 static int vmbus_acpi_add(struct acpi_device *device)
 {
 	acpi_status result;

index ffd98d7..6fec42d 100644
@@ -987,6 +987,8 @@ int vmbus_allocate_mmio(struct resource **new, struct hv_device *device_obj,
 			resource_size_t size, resource_size_t align,
 			bool fb_overlap_ok);
 
+int vmbus_procnum_to_vpnum(int procnum);
+
 /**
  * VMBUS_DEVICE - macro used to describe a specific hyperv vmbus device
  *

index 6fec42d..1903238 100644
@@ -1144,6 +1144,17 @@ int vmbus_procnum_to_vpnum(int procnum);
 		}
 
 /*
+ * PCI Express Pass Through
+ * {44C4F61D-4444-4400-9D52-802E27EDE19F}
+ */
+
+#define HV_PCIE_GUID \
+	.guid = { \
+			0x1D, 0xF6, 0xC4, 0x44, 0x44, 0x44, 0x00, 0x44, \
+			0x9D, 0x52, 0x80, 0x2E, 0x27, 0xED, 0xE1, 0x9F \
+		}
+
+/*
  * Common header for Hyper-V ICs
  */
 

2,177 lines not on this page · raw message

index 1903238..0acf9ff 100644
@@ -1269,4 +1269,22 @@ void hv_process_channel_removal(struct vmbus_channel *channel, u32 relid);
 
 extern __u32 vmbus_proto_version;
 
+/*
+ * Functions for passing data between SR-IOV PF and VF drivers.  The VF driver
+ * sends requests to read and write blocks.  Each block must be 128 bytes or
+ * smaller so that the host can guarantee a successful response.  (All
+ * intermediate buffers are preallocated.)  Optionally, the VF driver can
+ * register a callback function which will be invoked when the host says that
+ * one or more of the first 64 block IDs is "invalid" which means that the
+ * VF driver should reread them.
+ */
+#define CONFIG_BLOCK_SIZE_MAX 128
+int hv_read_config_block(struct pci_dev *dev, void *buf, int buf_len,
+			 int block_id, int *bytes_returned);
+int hv_write_config_block(struct pci_dev *dev, void *buf, int len,
+			  int block_id);
+int hv_register_block_invalidate(struct pci_dev *dev, void *context,
+				 void (*block_invalidate)(void *context,
+							  u64 block_mask));
+
 #endif /* _HYPERV_H */