[PATCH 23/28] kvm tools: Endian-sanitise pci.h and PCI device setup

[Matt Evans <matt@ozlabs.org>]

vesa, pci-shmem and virtio-pci devices need to set up config space with
little-endian conversions (as config space is LE).  The pci_config_address
bitfield also needs to be reversed when building on BE systems.

Signed-off-by: Matt Evans <matt@ozlabs.org>
---
 tools/kvm/hw/pci-shmem.c       |   23 +++++++++++----------
 tools/kvm/hw/vesa.c            |   15 +++++++------
 tools/kvm/include/kvm/ioport.h |   11 +++++----
 tools/kvm/include/kvm/pci.h    |   24 +++++++++++++++++-----
 tools/kvm/pci.c                |    4 +-
 tools/kvm/virtio/pci.c         |   41 +++++++++++++++++++++------------------
 6 files changed, 68 insertions(+), 50 deletions(-)

diff --git a/tools/kvm/hw/pci-shmem.c b/tools/kvm/hw/pci-shmem.c
index 780a377..fd954c5 100644
--- a/tools/kvm/hw/pci-shmem.c
+++ b/tools/kvm/hw/pci-shmem.c
@@ -8,21 +8,22 @@
 #include "kvm/ioeventfd.h"
 
 #include <linux/kvm.h>
+#include <linux/byteorder.h>
 #include <sys/ioctl.h>
 #include <fcntl.h>
 #include <sys/mman.h>
 
 static struct pci_device_header pci_shmem_pci_device = {
-	.vendor_id	= PCI_VENDOR_ID_REDHAT_QUMRANET,
-	.device_id	= 0x1110,
+	.vendor_id	= cpu_to_le16(PCI_VENDOR_ID_REDHAT_QUMRANET),
+	.device_id	= cpu_to_le16(0x1110),
 	.header_type	= PCI_HEADER_TYPE_NORMAL,
-	.class		= 0xFF0000,	/* misc pci device */
-	.status		= PCI_STATUS_CAP_LIST,
+	.class[2]	= 0xFF,	/* misc pci device */
+	.status		= cpu_to_le16(PCI_STATUS_CAP_LIST),
 	.capabilities	= (void *)&pci_shmem_pci_device.msix - (void *)&pci_shmem_pci_device,
 	.msix.cap	= PCI_CAP_ID_MSIX,
-	.msix.ctrl	= 1,
-	.msix.table_offset = 1,		/* Use BAR 1 */
-	.msix.pba_offset = 0x1001,	/* Use BAR 1 */
+	.msix.ctrl	= cpu_to_le16(1),
+	.msix.table_offset = cpu_to_le32(1),		/* Use BAR 1 */
+	.msix.pba_offset = cpu_to_le32(0x1001),		/* Use BAR 1 */
 };
 
 /* registers for the Inter-VM shared memory device */
@@ -123,7 +124,7 @@ int pci_shmem__get_local_irqfd(struct kvm *kvm)
 		if (fd < 0)
 			return fd;
 
-		if (pci_shmem_pci_device.msix.ctrl & PCI_MSIX_FLAGS_ENABLE) {
+		if (pci_shmem_pci_device.msix.ctrl & cpu_to_le16(PCI_MSIX_FLAGS_ENABLE)) {
 			gsi = irq__add_msix_route(kvm, &msix_table[0].msg);
 		} else {
 			gsi = pci_shmem_pci_device.irq_line;
@@ -241,11 +242,11 @@ int pci_shmem__init(struct kvm *kvm)
 	 * 1 - MSI-X MMIO space
 	 * 2 - Shared memory block
 	 */
-	pci_shmem_pci_device.bar[0] = ivshmem_registers | PCI_BASE_ADDRESS_SPACE_IO;
+	pci_shmem_pci_device.bar[0] = cpu_to_le32(ivshmem_registers | PCI_BASE_ADDRESS_SPACE_IO);
 	pci_shmem_pci_device.bar_size[0] = shmem_region->size;
-	pci_shmem_pci_device.bar[1] = msix_block | PCI_BASE_ADDRESS_SPACE_MEMORY;
+	pci_shmem_pci_device.bar[1] = cpu_to_le32(msix_block | PCI_BASE_ADDRESS_SPACE_MEMORY);
 	pci_shmem_pci_device.bar_size[1] = 0x1010;
-	pci_shmem_pci_device.bar[2] = shmem_region->phys_addr | PCI_BASE_ADDRESS_SPACE_MEMORY;
+	pci_shmem_pci_device.bar[2] = cpu_to_le32(shmem_region->phys_addr | PCI_BASE_ADDRESS_SPACE_MEMORY);
 	pci_shmem_pci_device.bar_size[2] = shmem_region->size;
 
 	pci__register(&pci_shmem_pci_device, dev);
diff --git a/tools/kvm/hw/vesa.c b/tools/kvm/hw/vesa.c
index 22b1652..63f1082 100644
--- a/tools/kvm/hw/vesa.c
+++ b/tools/kvm/hw/vesa.c
@@ -8,6 +8,7 @@
 #include "kvm/irq.h"
 #include "kvm/kvm.h"
 #include "kvm/pci.h"
+#include <linux/byteorder.h>
 #include <sys/mman.h>
 
 #include <sys/types.h>
@@ -31,14 +32,14 @@ static struct ioport_operations vesa_io_ops = {
 };
 
 static struct pci_device_header vesa_pci_device = {
-	.vendor_id		= PCI_VENDOR_ID_REDHAT_QUMRANET,
-	.device_id		= PCI_DEVICE_ID_VESA,
+	.vendor_id		= cpu_to_le16(PCI_VENDOR_ID_REDHAT_QUMRANET),
+	.device_id		= cpu_to_le16(PCI_DEVICE_ID_VESA),
 	.header_type		= PCI_HEADER_TYPE_NORMAL,
 	.revision_id		= 0,
-	.class			= 0x030000,
-	.subsys_vendor_id	= PCI_SUBSYSTEM_VENDOR_ID_REDHAT_QUMRANET,
-	.subsys_id		= PCI_SUBSYSTEM_ID_VESA,
-	.bar[1]			= VESA_MEM_ADDR | PCI_BASE_ADDRESS_SPACE_MEMORY,
+	.class[2]		= 0x03,
+	.subsys_vendor_id	= cpu_to_le16(PCI_SUBSYSTEM_VENDOR_ID_REDHAT_QUMRANET),
+	.subsys_id		= cpu_to_le16(PCI_SUBSYSTEM_ID_VESA),
+	.bar[1]			= cpu_to_le32(VESA_MEM_ADDR | PCI_BASE_ADDRESS_SPACE_MEMORY),
 	.bar_size[1]		= VESA_MEM_SIZE,
 };
 
@@ -56,7 +57,7 @@ struct framebuffer *vesa__init(struct kvm *kvm)
 	vesa_pci_device.irq_pin		= pin;
 	vesa_pci_device.irq_line	= line;
 	vesa_base_addr			= ioport__register(IOPORT_EMPTY, &vesa_io_ops, IOPORT_SIZE, NULL);
-	vesa_pci_device.bar[0]		= vesa_base_addr | PCI_BASE_ADDRESS_SPACE_IO;
+	vesa_pci_device.bar[0]		= cpu_to_le32(vesa_base_addr | PCI_BASE_ADDRESS_SPACE_IO);
 	pci__register(&vesa_pci_device, dev);
 
 	mem = mmap(NULL, VESA_MEM_SIZE, PROT_RW, MAP_ANON_NORESERVE, -1, 0);
diff --git a/tools/kvm/include/kvm/ioport.h b/tools/kvm/include/kvm/ioport.h
index 61a70ec..09bf876 100644
--- a/tools/kvm/include/kvm/ioport.h
+++ b/tools/kvm/include/kvm/ioport.h
@@ -7,6 +7,7 @@
 #include <limits.h>
 #include <asm/types.h>
 #include <linux/types.h>
+#include <linux/byteorder.h>
 
 /* some ports we reserve for own use */
 #define IOPORT_DBG			0xe0
@@ -36,15 +37,15 @@ static inline u8 ioport__read8(u8 *data)
 {
 	return *data;
 }
-
+/* On BE platforms, PCI I/O is byteswapped, i.e. LE, so swap back. */
 static inline u16 ioport__read16(u16 *data)
 {
-	return *data;
+	return le16_to_cpu(*data);
 }
 
 static inline u32 ioport__read32(u32 *data)
 {
-	return *data;
+	return le32_to_cpu(*data);
 }
 
 static inline void ioport__write8(u8 *data, u8 value)
@@ -54,12 +55,12 @@ static inline void ioport__write8(u8 *data, u8 value)
 
 static inline void ioport__write16(u16 *data, u16 value)
 {
-	*data		 = value;
+	*data		 = cpu_to_le16(value);
 }
 
 static inline void ioport__write32(u32 *data, u32 value)
 {
-	*data		 = value;
+	*data		 = cpu_to_le32(value);
 }
 
 #endif /* KVM__IOPORT_H */
diff --git a/tools/kvm/include/kvm/pci.h b/tools/kvm/include/kvm/pci.h
index b578ad7..88e92dc 100644
--- a/tools/kvm/include/kvm/pci.h
+++ b/tools/kvm/include/kvm/pci.h
@@ -5,6 +5,7 @@
 #include <linux/kvm.h>
 #include <linux/pci_regs.h>
 #include <linux/msi.h>
+#include <endian.h>
 
 #define PCI_MAX_DEVICES			256
 /*
@@ -17,7 +18,8 @@
 #define PCI_CONFIG_BUS_FORWARD	0xcfa
 #define PCI_IO_SIZE		0x100
 
-struct pci_config_address {
+union pci_config_address {
+#if __BYTE_ORDER == __LITTLE_ENDIAN
 	unsigned	zeros		: 2;		/* 1  .. 0  */
 	unsigned	register_number	: 6;		/* 7  .. 2  */
 	unsigned	function_number	: 3;		/* 10 .. 8  */
@@ -25,6 +27,16 @@ struct pci_config_address {
 	unsigned	bus_number	: 8;		/* 23 .. 16 */
 	unsigned	reserved	: 7;		/* 30 .. 24 */
 	unsigned	enable_bit	: 1;		/* 31       */
+#else
+	unsigned	enable_bit	: 1;		/* 31       */
+	unsigned	reserved	: 7;		/* 30 .. 24 */
+	unsigned	bus_number	: 8;		/* 23 .. 16 */
+	unsigned	device_number	: 5;		/* 15 .. 11 */
+	unsigned	function_number	: 3;		/* 10 .. 8  */
+	unsigned	register_number	: 6;		/* 7  .. 2  */
+	unsigned	zeros		: 2;		/* 1  .. 0  */
+#endif
+	u32 w;
 };
 
 struct msix_table {
@@ -45,8 +57,8 @@ struct pci_device_header {
 	u16		device_id;
 	u16		command;
 	u16		status;
-	u16		revision_id		:  8;
-	u32		class			: 24;
+	u8		revision_id;
+	u8		class[3];
 	u8		cacheline_size;
 	u8		latency_timer;
 	u8		header_type;
@@ -56,8 +68,8 @@ struct pci_device_header {
 	u16		subsys_vendor_id;
 	u16		subsys_id;
 	u32		exp_rom_bar;
-	u32		capabilities		:  8;
-	u32		reserved1		: 24;
+	u8		capabilities;
+	u8		reserved1[3];
 	u32		reserved2;
 	u8		irq_line;
 	u8		irq_pin;
@@ -66,7 +78,7 @@ struct pci_device_header {
 	struct msix_cap msix;
 	u8		empty[136]; /* Rest of PCI config space */
 	u32		bar_size[6];
-};
+} __attribute__((packed));
 
 void pci__init(void);
 void pci__register(struct pci_device_header *dev, u8 dev_num);
diff --git a/tools/kvm/pci.c b/tools/kvm/pci.c
index 920e13e..5bbcbc7 100644
--- a/tools/kvm/pci.c
+++ b/tools/kvm/pci.c
@@ -9,7 +9,7 @@
 
 static struct pci_device_header		*pci_devices[PCI_MAX_DEVICES];
 
-static struct pci_config_address	pci_config_address;
+static union pci_config_address		pci_config_address;
 
 /* This is within our PCI gap - in an unused area */
 static u32 io_space_blocks		= KVM_32BIT_GAP_START + 0x1000000;
@@ -105,7 +105,7 @@ static bool pci_config_data_out(struct ioport *ioport, struct kvm *kvm, u16 port
 			 * When the kernel got the size it would write the address
 			 * back.
 			 */
-			if (ioport__read32(p + offset)) {
+			if (*(u32 *)(p + offset)) {
 				/* See if kernel tries to mask one of the BARs */
 				if ((offset >= PCI_BAR_OFFSET(0)) &&
 				    (offset <= PCI_BAR_OFFSET(6)) &&
diff --git a/tools/kvm/virtio/pci.c b/tools/kvm/virtio/pci.c
index 0737ae7..0ae93fb 100644
--- a/tools/kvm/virtio/pci.c
+++ b/tools/kvm/virtio/pci.c
@@ -9,6 +9,7 @@
 #include "kvm/virtio-trans.h"
 
 #include <linux/virtio_pci.h>
+#include <linux/byteorder.h>
 #include <string.h>
 
 struct virtio_trans_ops *virtio_pci__get_trans_ops(void)
@@ -59,7 +60,7 @@ static int virtio_pci__init_ioeventfd(struct kvm *kvm, struct virtio_trans *vtra
 
 static inline bool virtio_pci__msix_enabled(struct virtio_pci *vpci)
 {
-	return vpci->pci_hdr.msix.ctrl & PCI_MSIX_FLAGS_ENABLE;
+	return vpci->pci_hdr.msix.ctrl & cpu_to_le16(PCI_MSIX_FLAGS_ENABLE);
 }
 
 static bool virtio_pci__specific_io_in(struct kvm *kvm, struct virtio_trans *vtrans, u16 port,
@@ -245,8 +246,8 @@ int virtio_pci__signal_vq(struct kvm *kvm, struct virtio_trans *vtrans, u32 vq)
 	int tbl = vpci->vq_vector[vq];
 
 	if (virtio_pci__msix_enabled(vpci)) {
-		if (vpci->pci_hdr.msix.ctrl & PCI_MSIX_FLAGS_MASKALL ||
-			vpci->msix_table[tbl].ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT) {
+		if (vpci->pci_hdr.msix.ctrl & cpu_to_le16(PCI_MSIX_FLAGS_MASKALL) ||
+		    vpci->msix_table[tbl].ctrl & cpu_to_le16(PCI_MSIX_ENTRY_CTRL_MASKBIT)) {
 
 			vpci->msix_pba |= 1 << tbl;
 			return 0;
@@ -266,8 +267,8 @@ int virtio_pci__signal_config(struct kvm *kvm, struct virtio_trans *vtrans)
 	int tbl = vpci->config_vector;
 
 	if (virtio_pci__msix_enabled(vpci)) {
-		if (vpci->pci_hdr.msix.ctrl & PCI_MSIX_FLAGS_MASKALL ||
-			vpci->msix_table[tbl].ctrl & PCI_MSIX_ENTRY_CTRL_MASKBIT) {
+		if (vpci->pci_hdr.msix.ctrl & cpu_to_le16(PCI_MSIX_FLAGS_MASKALL) ||
+		    vpci->msix_table[tbl].ctrl & cpu_to_le16(PCI_MSIX_ENTRY_CTRL_MASKBIT)) {
 
 			vpci->msix_pba |= 1 << tbl;
 			return 0;
@@ -297,19 +298,21 @@ int virtio_pci__init(struct kvm *kvm, struct virtio_trans *vtrans, void *dev,
 	kvm__register_mmio(kvm, vpci->msix_pba_block, 0x100, callback_mmio_pba, vpci);
 
 	vpci->pci_hdr = (struct pci_device_header) {
-		.vendor_id		= PCI_VENDOR_ID_REDHAT_QUMRANET,
-		.device_id		= device_id,
+		.vendor_id		= cpu_to_le16(PCI_VENDOR_ID_REDHAT_QUMRANET),
+		.device_id		= cpu_to_le16(device_id),
 		.header_type		= PCI_HEADER_TYPE_NORMAL,
 		.revision_id		= 0,
-		.class			= class,
-		.subsys_vendor_id	= PCI_SUBSYSTEM_VENDOR_ID_REDHAT_QUMRANET,
-		.subsys_id		= subsys_id,
-		.bar[0]			= vpci->base_addr | PCI_BASE_ADDRESS_SPACE_IO,
-		.bar[1]			= vpci->msix_io_block | PCI_BASE_ADDRESS_SPACE_MEMORY
-					| PCI_BASE_ADDRESS_MEM_TYPE_64,
-		.bar[3]			= vpci->msix_pba_block | PCI_BASE_ADDRESS_SPACE_MEMORY
-					| PCI_BASE_ADDRESS_MEM_TYPE_64,
-		.status			= PCI_STATUS_CAP_LIST,
+		.class[0]		= class & 0xff,
+		.class[1]		= (class >> 8) & 0xff,
+		.class[2]		= (class >> 16) & 0xff,
+		.subsys_vendor_id	= cpu_to_le16(PCI_SUBSYSTEM_VENDOR_ID_REDHAT_QUMRANET),
+		.subsys_id		= cpu_to_le16(subsys_id),
+		.bar[0]			= cpu_to_le32(vpci->base_addr | PCI_BASE_ADDRESS_SPACE_IO),
+		.bar[1]			= cpu_to_le32(vpci->msix_io_block | PCI_BASE_ADDRESS_SPACE_MEMORY
+						      | PCI_BASE_ADDRESS_MEM_TYPE_64),
+		.bar[3]			= cpu_to_le32(vpci->msix_pba_block | PCI_BASE_ADDRESS_SPACE_MEMORY
+						      | PCI_BASE_ADDRESS_MEM_TYPE_64),
+		.status			= cpu_to_le16(PCI_STATUS_CAP_LIST),
 		.capabilities		= (void *)&vpci->pci_hdr.msix - (void *)&vpci->pci_hdr,
 	};
 
@@ -326,14 +329,14 @@ int virtio_pci__init(struct kvm *kvm, struct virtio_trans *vtrans, void *dev,
 	 * For example, a returned value of "00000000011"
 	 * indicates a table size of 4.
 	 */
-	vpci->pci_hdr.msix.ctrl = (VIRTIO_PCI_MAX_VQ + VIRTIO_PCI_MAX_CONFIG - 1);
+	vpci->pci_hdr.msix.ctrl = cpu_to_le16(VIRTIO_PCI_MAX_VQ + VIRTIO_PCI_MAX_CONFIG - 1);
 
 	/*
 	 * Both table and PBA could be mapped on the same BAR, but for now
 	 * we're not in short of BARs
 	 */
-	vpci->pci_hdr.msix.table_offset = 1; /* Use BAR 1 */
-	vpci->pci_hdr.msix.pba_offset = 3; /* Use BAR 3 */
+	vpci->pci_hdr.msix.table_offset = cpu_to_le32(1); /* Use BAR 1 */
+	vpci->pci_hdr.msix.pba_offset = cpu_to_le32(3); /* Use BAR 3 */
 	vpci->config_vector = 0;
 
 	if (irq__register_device(subsys_id, &ndev, &pin, &line) < 0)