kvm: release: merge from trunkkvm-8

........
  r4116 | avi | 2006-12-17 10:52:25 +0200 (Sun, 17 Dec 2006) | 11 lines

  kvm: do not export unsupported msrs to userspace

  From: Michael Riepe <michael@mr511.de>

  Some msrs, such as MSR_STAR, are not available on all processors.  Exporting
  them causes qemu to try to fetch them, which will fail.

  So, check all msrs for validity at module load time.

  Signed-off-by: Michael Riepe <michael@mr511.de>
........
  r4117 | avi | 2006-12-17 11:10:47 +0200 (Sun, 17 Dec 2006) | 4 lines

  kvm: web: fix -noacpi option in faq

  noticed by Ingemar Nilsson <init@kth.se>
........
  r4119 | avi | 2006-12-17 14:39:24 +0200 (Sun, 17 Dec 2006) | 8 lines

  kvm: force real-mode cs limit to 64K

  From: Michael Riepe <michael@mr511.de>

  this allows opensolaris to boot on kvm/intel.

  Signed-off-by: Michael Riepe <michael@mr511.de>
........
  r4120 | avi | 2006-12-17 14:42:18 +0200 (Sun, 17 Dec 2006) | 8 lines

  kvm: handle p5 mce msrs

  From: Michael Riepe <michael@mr511.de>

  this allows plan9 to get a little further booting.

  Signed-off-by: Michael Riepe <michael@mr511.de>
........
  r4126 | uri | 2006-12-19 11:48:31 +0200 (Tue, 19 Dec 2006) | 4 lines

  qemu migration interface: add "listening" status #143

  Simon's request is my (migration status) command (to be changed)
........
  r4127 | uri | 2006-12-19 13:13:08 +0200 (Tue, 19 Dec 2006) | 5 lines

  qemu migration interface: set FAIL status for every failure (#143)

  Including failure to open/bind/listen-to a socket.
  Also migration_cleanup() now always sets the status (even if fd is FD_UNUSED)
........
  r4128 | avi | 2006-12-19 17:04:30 +0200 (Tue, 19 Dec 2006) | 2 lines

  kvm: web: clarify /proc/cpuinfo availability wrt kernel versions
........
  r4132 | avi | 2006-12-20 14:11:44 +0200 (Wed, 20 Dec 2006) | 2 lines

  kvm: web: mention kvm-commits mailing list
........
  r4143 | avi | 2006-12-21 11:36:10 +0200 (Thu, 21 Dec 2006) | 2 lines

  kvm: add api versioning
........
  r4144 | avi | 2006-12-21 11:36:43 +0200 (Thu, 21 Dec 2006) | 2 lines

  kvm: libkvm: add kernel api version check
........

6 files changed, 84 insertions, 18 deletions

diff --git a/kvm/kernel/include/linux/kvm.h b/kvm/kernel/include/linux/kvm.h
index 5bb2c3c58..28fdce1ac 100644
--- a/kvm/kernel/include/linux/kvm.h
+++ b/kvm/kernel/include/linux/kvm.h
@@ -11,6 +11,8 @@
 #include <asm/types.h>
 #include <linux/ioctl.h>
 
+#define KVM_API_VERSION 1
+
 /*
  * Architectural interrupt line count, and the size of the bitmap needed
  * to hold them.
@@ -209,6 +211,7 @@ struct kvm_dirty_log {
 
 #define KVMIO 0xAE
 
+#define KVM_GET_API_VERSION       _IO(KVMIO, 1)
 #define KVM_RUN                   _IOWR(KVMIO, 2, struct kvm_run)
 #define KVM_GET_REGS              _IOWR(KVMIO, 3, struct kvm_regs)
 #define KVM_SET_REGS              _IOW(KVMIO, 4, struct kvm_regs)
diff --git a/kvm/kernel/kvm_main.c b/kvm/kernel/kvm_main.c
index 4621b4ee6..3df93eff8 100644
--- a/kvm/kernel/kvm_main.c
+++ b/kvm/kernel/kvm_main.c
@@ -1417,6 +1417,9 @@ static int kvm_dev_ioctl_set_sregs(struct kvm *kvm, struct kvm_sregs *sregs)
 /*
  * List of msr numbers which we expose to userspace through KVM_GET_MSRS
  * and KVM_SET_MSRS, and KVM_GET_MSR_INDEX_LIST.
+ *
+ * This list is modified at module load time to reflect the
+ * capabilities of the host cpu.
  */
 static u32 msrs_to_save[] = {
 	MSR_IA32_SYSENTER_CS, MSR_IA32_SYSENTER_ESP, MSR_IA32_SYSENTER_EIP,
@@ -1427,6 +1430,22 @@ static u32 msrs_to_save[] = {
 	MSR_IA32_TIME_STAMP_COUNTER,
 };
 
+static unsigned num_msrs_to_save = 0;
+
+static __init void kvm_init_msr_list(void)
+{
+	u32 dummy[2];
+	unsigned i, j;
+
+	for (i = j = 0; i < ARRAY_SIZE(msrs_to_save); i++) {
+		if (rdmsr_safe(msrs_to_save[i], &dummy[0], &dummy[1]) < 0)
+			continue;
+		if (j < i)
+			msrs_to_save[j] = msrs_to_save[i];
+		j++;
+	}
+	num_msrs_to_save = j;
+}
 
 /*
  * Adapt set_msr() to msr_io()'s calling convention
@@ -1584,6 +1603,9 @@ static long kvm_dev_ioctl(struct file *filp,
 	int r = -EINVAL;
 
 	switch (ioctl) {
+	case KVM_GET_API_VERSION:
+		r = KVM_API_VERSION;
+		break;
 	case KVM_CREATE_VCPU: {
 		r = kvm_dev_ioctl_create_vcpu(kvm, arg);
 		if (r)
@@ -1735,15 +1757,15 @@ static long kvm_dev_ioctl(struct file *filp,
 		if (copy_from_user(&msr_list, user_msr_list, sizeof msr_list))
 			goto out;
 		n = msr_list.nmsrs;
-		msr_list.nmsrs = ARRAY_SIZE(msrs_to_save);
+		msr_list.nmsrs = num_msrs_to_save;
 		if (copy_to_user(user_msr_list, &msr_list, sizeof msr_list))
 			goto out;
 		r = -E2BIG;
-		if (n < ARRAY_SIZE(msrs_to_save))
+		if (n < num_msrs_to_save)
 			goto out;
 		r = -EFAULT;
 		if (copy_to_user(user_msr_list->indices, &msrs_to_save,
-				 sizeof msrs_to_save))
+				 num_msrs_to_save * sizeof(u32)))
 			goto out;
 		r = 0;
 	}
@@ -1894,6 +1916,8 @@ static __init int kvm_init(void)
 
 	kvm_init_debug();
 
+	kvm_init_msr_list();
+
 	if ((bad_page = alloc_page(GFP_KERNEL)) == NULL) {
 		r = -ENOMEM;
 		goto out;
diff --git a/kvm/kernel/svm.c b/kvm/kernel/svm.c
index 59c930f27..9ec87636d 100644
--- a/kvm/kernel/svm.c
+++ b/kvm/kernel/svm.c
@@ -1073,6 +1073,8 @@ static int emulate_on_interception(struct kvm_vcpu *vcpu, struct kvm_run *kvm_ru
 static int svm_get_msr(struct kvm_vcpu *vcpu, unsigned ecx, u64 *data)
 {
 	switch (ecx) {
+	case MSR_IA32_P5_MC_ADDR:
+	case MSR_IA32_P5_MC_TYPE:
 	case MSR_IA32_MC0_CTL:
 	case MSR_IA32_MCG_STATUS:
 	case MSR_IA32_MCG_CAP:
diff --git a/kvm/kernel/vmx.c b/kvm/kernel/vmx.c
index f0f0b1a78..983a15b19 100644
--- a/kvm/kernel/vmx.c
+++ b/kvm/kernel/vmx.c
@@ -359,6 +359,8 @@ static int vmx_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata)
 	case MSR_IA32_SYSENTER_ESP:
 		data = vmcs_read32(GUEST_SYSENTER_ESP);
 		break;
+	case MSR_IA32_P5_MC_ADDR:
+	case MSR_IA32_P5_MC_TYPE:
 	case MSR_IA32_MC0_CTL:
 	case MSR_IA32_MCG_STATUS:
 	case MSR_IA32_MCG_CAP:
@@ -726,6 +728,7 @@ static void enter_rmode(struct kvm_vcpu *vcpu)
 	vmcs_write32(GUEST_SS_AR_BYTES, 0xf3);
 
 	vmcs_write32(GUEST_CS_AR_BYTES, 0xf3);
+	vmcs_write32(GUEST_CS_LIMIT, 0xffff);
 	vmcs_write16(GUEST_CS_SELECTOR, vmcs_readl(GUEST_CS_BASE) >> 4);
 
 	fix_rmode_seg(VCPU_SREG_ES, &vcpu->rmode.es);
diff --git a/kvm/user/kvmctl.c b/kvm/user/kvmctl.c
index 844fc57b7..7f712e71a 100644
--- a/kvm/user/kvmctl.c
+++ b/kvm/user/kvmctl.c
@@ -23,6 +23,12 @@
 #include <errno.h>
 #include "kvmctl.h"
 
+#define EXPECTED_KVM_API_VERSION 1
+
+#if EXPECTED_KVM_API_VERSION != KVM_API_VERSION
+#error libkvm: userspace and kernel version mismatch
+#endif
+
 #define PAGE_SIZE 4096ul
 
 struct kvm_context {
@@ -74,17 +80,34 @@ kvm_context_t kvm_init(struct kvm_callbacks *callbacks,
 {
 	int fd;
 	kvm_context_t kvm;
+	int r;
 
 	fd = open("/dev/kvm", O_RDWR);
 	if (fd == -1) {
-		perror("open");
+		perror("open /dev/kvm");
 		return NULL;
 	}
+	r = ioctl(fd, KVM_GET_API_VERSION, 0);
+	if (r == -1) {
+	    fprintf(stderr, "kvm kernel version too old\n");
+	    goto out_close;
+	}
+	if (r < EXPECTED_KVM_API_VERSION) {
+	    fprintf(stderr, "kvm kernel version too old\n");
+	    goto out_close;
+	}
+	if (r > EXPECTED_KVM_API_VERSION) {
+	    fprintf(stderr, "kvm userspace version too old\n");
+	    goto out_close;
+	}
 	kvm = malloc(sizeof(*kvm));
 	kvm->fd = fd;
 	kvm->callbacks = callbacks;
 	kvm->opaque = opaque;
 	return kvm;
+ out_close:
+	close(fd);
+	return NULL;
 }
 
 void kvm_finalize(kvm_context_t kvm)
diff --git a/migration.c b/migration.c
index 80ec292d5..1bf515efd 100644
--- a/migration.c
+++ b/migration.c
@@ -28,11 +28,12 @@ typedef enum {
 
 typedef enum {
     MIG_STAT_NONE   = 0, /* disconnected */
-    MIG_STAT_CONN   = 1, /* connection established */
-    MIG_STAT_START  = 2, /* migration started */
-    MIG_STAT_SUCC   = 3, /* migration completed successfully */
-    MIG_STAT_FAIL   = 4, /* migration failed */
-    MIG_STAT_CANCEL = 5  /* migration canceled */
+    MIG_STAT_LISTEN = 1, /* listening, waiting for the other to connect */
+    MIG_STAT_CONN   = 2, /* connection established */
+    MIG_STAT_START  = 3, /* migration started */
+    MIG_STAT_SUCC   = 4, /* migration completed successfully */
+    MIG_STAT_FAIL   = 5, /* migration failed */
+    MIG_STAT_CANCEL = 6  /* migration canceled */
 } migration_status_t;
 
 typedef struct migration_bandwith_params {
@@ -68,6 +69,7 @@ static const char *writer_default_addr="localhost:4456";
 
 /* forward declarations */
 static void migration_start_dst(int online);
+static void migration_cleanup(migration_state_t *pms, migration_status_t stat);
 
 
 static const char *mig_stat_str(migration_status_t mig_stat)
@@ -76,9 +78,10 @@ static const char *mig_stat_str(migration_status_t mig_stat)
         migration_status_t stat;
         const char *str;
     } stat_strs[] = {    
-        {MIG_STAT_NONE,  "disconnected"},
-        {MIG_STAT_CONN,  "connected"},
-        {MIG_STAT_START, "migration stared"},
+        {MIG_STAT_NONE,   "disconnected"},
+        {MIG_STAT_LISTEN, "listening"},
+        {MIG_STAT_CONN,   "connected"},
+        {MIG_STAT_START,  "migration stared"},
         {MIG_STAT_SUCC,   "migration completed successfully"},
         {MIG_STAT_FAIL,   "migration failed"},
         {MIG_STAT_CANCEL, "migration canceled"}
@@ -141,7 +144,8 @@ static int parse_host_port_and_message(struct sockaddr_in *saddr,
     if (!arg)
         arg = default_addr;
     if (parse_host_port(saddr, arg) < 0) {
-        term_printf("%s: invalid argument '%s'", name, arg);
+        term_printf("%s: invalid argument '%s'\n", name, arg);
+        migration_cleanup(&ms, MIG_STAT_FAIL);
         return -1;
     }
     return 0;
@@ -155,8 +159,8 @@ static void migration_cleanup(migration_state_t *pms, migration_status_t stat)
 #endif
         close(pms->fd);
         pms->fd = FD_UNUSED;
-        pms->status = stat;
     }
+    pms->status = stat;
 }
 
 static int migration_read_from_socket(void *opaque)
@@ -179,6 +183,7 @@ static int migration_read_from_socket(void *opaque)
             if (socket_error() == EINTR)
                 continue;
             term_printf("migration_read_from_socket: read failed (%s)\n", strerror(errno) );
+            migration_cleanup(pms, MIG_STAT_FAIL);
             return size;
         }
         if (size == 0) {
@@ -218,6 +223,7 @@ static int migration_write_into_socket(void *opaque, int len)
                 continue;
             term_printf("migration_write_into_socket: write failed (%s)\n", 
                         strerror(socket_error()) );
+            migration_cleanup(pms, MIG_STAT_FAIL);
             return size;
         }
         if (size == 0) {
@@ -252,6 +258,7 @@ static void migration_accept(void *opaque)
         if (new_fd < 0 && errno != EINTR) {
             term_printf("migration listen: accept failed (%s)\n",
                         strerror(errno));
+            migration_cleanup(pms, MIG_STAT_FAIL);
             return;
         } else if (new_fd >= 0) {
             break;
@@ -296,6 +303,7 @@ void do_migration_listen(char *arg1, char *arg2)
     if (ms.fd < 0) {
         term_printf("migration listen: socket() failed (%s)\n",
                     strerror(errno));
+        migration_cleanup(&ms, MIG_STAT_FAIL);
         return;
     }
 
@@ -304,17 +312,18 @@ void do_migration_listen(char *arg1, char *arg2)
     setsockopt(ms.fd, SOL_SOCKET, SO_REUSEADDR, (const char *)&val, sizeof(val));
     
     if (bind(ms.fd, &local, sizeof local) < 0 ) {
-        migration_cleanup(&ms, MIG_STAT_NONE);
+        migration_cleanup(&ms, MIG_STAT_FAIL);
         term_printf("migration listen: bind() failed (%s)\n", strerror(errno));
         return;
     }
     
     if (listen(ms.fd, 1) < 0) { /* allow only one connection */
-        migration_cleanup(&ms, MIG_STAT_NONE);
+        migration_cleanup(&ms, MIG_STAT_FAIL);
         term_printf("migration listen: listen() failed (%s)\n", strerror(errno));
         return;
     }
 
+    ms.status = MIG_STAT_LISTEN;
     term_printf("migration listen: listening on fd %d\n", ms.fd);
 
 #ifdef USE_NONBLOCKING_SOCKETS
@@ -451,7 +460,7 @@ static void migration_connect_check(void *opaque)
     rc = getsockopt(pms->fd, SOL_SOCKET, SO_ERROR, (void *)&err, &len);
     if (rc != 0) {
         term_printf("migration connect: getsockopt FAILED (%s)\n", strerror(errno));
-        migration_cleanup(pms, MIG_STAT_NONE);
+        migration_cleanup(pms, MIG_STAT_FAIL);
         return;
     }
     if (err == 0) {
@@ -460,6 +469,7 @@ static void migration_connect_check(void *opaque)
     }
     else {
         term_printf("migration connect: failed to conenct (%s)\n", strerror(err));
+        migration_cleanup(pms, MIG_STAT_FAIL);
         return;
     }
 
@@ -490,6 +500,7 @@ void do_migration_connect(char *arg1, char *arg2)
     if (ms.fd < 0) {
         term_printf("migration connect: socket() failed (%s)\n",
                     strerror(errno));
+        migration_cleanup(&ms, MIG_STAT_FAIL);
         return;
     }
 
@@ -504,7 +515,7 @@ void do_migration_connect(char *arg1, char *arg2)
         if (errno != EINPROGRESS) {
             term_printf("migration connect: connect() failed (%s)\n",
                         strerror(errno));
-            migration_cleanup(&ms, MIG_STAT_NONE);
+            migration_cleanup(&ms, MIG_STAT_FAIL);
         }
         return;
     }