X-Git-Url: http://git.maemo.org/git/?a=blobdiff_plain;f=kvm-all.c;h=73e814a31d81194ddff0f38636e87f504515e7dd;hb=d3f243676addaef6c8d818934565292c698f91cc;hp=3e4e4213891d34dedf08b31b74c4c93a1b7b21e7;hpb=d3f8d37fe2d0c24ec8bac9c94d5b0e2dc09c0d2a;p=qemu

diff --git a/kvm-all.c b/kvm-all.c
index 3e4e421..73e814a 100644
--- a/kvm-all.c
+++ b/kvm-all.c
@@ -98,19 +98,31 @@ static KVMSlot *kvm_lookup_matching_slot(KVMState *s,
     return NULL;
 }
 
-static KVMSlot *kvm_lookup_slot(KVMState *s, target_phys_addr_t start_addr)
+/*
+ * Find overlapping slot with lowest start address
+ */
+static KVMSlot *kvm_lookup_overlapping_slot(KVMState *s,
+                                            target_phys_addr_t start_addr,
+                                            target_phys_addr_t end_addr)
 {
+    KVMSlot *found = NULL;
     int i;
 
     for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
         KVMSlot *mem = &s->slots[i];
 
-        if (start_addr >= mem->start_addr &&
-            start_addr < (mem->start_addr + mem->memory_size))
-            return mem;
+        if (mem->memory_size == 0 ||
+            (found && found->start_addr < mem->start_addr)) {
+            continue;
+        }
+
+        if (end_addr > mem->start_addr &&
+            start_addr < mem->start_addr + mem->memory_size) {
+            found = mem;
+        }
     }
 
-    return NULL;
+    return found;
 }
 
 static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
@@ -308,14 +320,28 @@ int kvm_uncoalesce_mmio_region(target_phys_addr_t start, ram_addr_t size)
     return ret;
 }
 
+int kvm_check_extension(KVMState *s, unsigned int extension)
+{
+    int ret;
+
+    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, extension);
+    if (ret < 0) {
+        ret = 0;
+    }
+
+    return ret;
+}
+
 int kvm_init(int smp_cpus)
 {
     KVMState *s;
     int ret;
     int i;
 
-    if (smp_cpus > 1)
+    if (smp_cpus > 1) {
+        fprintf(stderr, "No SMP KVM support, use '-smp 1'\n");
         return -EINVAL;
+    }
 
     s = qemu_mallocz(sizeof(KVMState));
 
@@ -356,10 +382,8 @@ int kvm_init(int smp_cpus)
      * just use a user allocated buffer so we can use regular pages
      * unmodified.  Make sure we have a sufficiently modern version of KVM.
      */
-    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, KVM_CAP_USER_MEMORY);
-    if (ret <= 0) {
-        if (ret == 0)
-            ret = -EINVAL;
+    if (!kvm_check_extension(s, KVM_CAP_USER_MEMORY)) {
+        ret = -EINVAL;
         fprintf(stderr, "kvm does not support KVM_CAP_USER_MEMORY\n");
         goto err;
     }
@@ -367,11 +391,8 @@ int kvm_init(int smp_cpus)
     /* There was a nasty bug in < kvm-80 that prevents memory slots from being
      * destroyed properly.  Since we rely on this capability, refuse to work
      * with any kernel without this capability. */
-    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION,
-                    KVM_CAP_DESTROY_MEMORY_REGION_WORKS);
-    if (ret <= 0) {
-        if (ret == 0)
-            ret = -EINVAL;
+    if (!kvm_check_extension(s, KVM_CAP_DESTROY_MEMORY_REGION_WORKS)) {
+        ret = -EINVAL;
 
         fprintf(stderr,
                 "KVM kernel module broken (DESTROY_MEMORY_REGION)\n"
@@ -379,11 +400,10 @@ int kvm_init(int smp_cpus)
         goto err;
     }
 
-    s->coalesced_mmio = 0;
 #ifdef KVM_CAP_COALESCED_MMIO
-    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, KVM_CAP_COALESCED_MMIO);
-    if (ret > 0)
-        s->coalesced_mmio = ret;
+    s->coalesced_mmio = kvm_check_extension(s, KVM_CAP_COALESCED_MMIO);
+#else
+    s->coalesced_mmio = 0;
 #endif
 
     ret = kvm_arch_init(s, smp_cpus);
@@ -567,65 +587,119 @@ void kvm_set_phys_mem(target_phys_addr_t start_addr,
 {
     KVMState *s = kvm_state;
     ram_addr_t flags = phys_offset & ~TARGET_PAGE_MASK;
-    KVMSlot *mem;
+    KVMSlot *mem, old;
+    int err;
 
     if (start_addr & ~TARGET_PAGE_MASK) {
-        fprintf(stderr, "Only page-aligned memory slots supported\n");
+        if (flags >= IO_MEM_UNASSIGNED) {
+            if (!kvm_lookup_overlapping_slot(s, start_addr,
+                                             start_addr + size)) {
+                return;
+            }
+            fprintf(stderr, "Unaligned split of a KVM memory slot\n");
+        } else {
+            fprintf(stderr, "Only page-aligned memory slots supported\n");
+        }
         abort();
     }
 
     /* KVM does not support read-only slots */
     phys_offset &= ~IO_MEM_ROM;
 
-    mem = kvm_lookup_slot(s, start_addr);
-    if (mem) {
-        if (flags >= IO_MEM_UNASSIGNED) {
-            mem->memory_size = 0;
-            mem->start_addr = start_addr;
-            mem->phys_offset = 0;
+    while (1) {
+        mem = kvm_lookup_overlapping_slot(s, start_addr, start_addr + size);
+        if (!mem) {
+            break;
+        }
+
+        if (flags < IO_MEM_UNASSIGNED && start_addr >= mem->start_addr &&
+            (start_addr + size <= mem->start_addr + mem->memory_size) &&
+            (phys_offset - start_addr == mem->phys_offset - mem->start_addr)) {
+            /* The new slot fits into the existing one and comes with
+             * identical parameters - nothing to be done. */
+            return;
+        }
+
+        old = *mem;
+
+        /* unregister the overlapping slot */
+        mem->memory_size = 0;
+        err = kvm_set_user_memory_region(s, mem);
+        if (err) {
+            fprintf(stderr, "%s: error unregistering overlapping slot: %s\n",
+                    __func__, strerror(-err));
+            abort();
+        }
+
+        /* Workaround for older KVM versions: we can't join slots, even not by
+         * unregistering the previous ones and then registering the larger
+         * slot. We have to maintain the existing fragmentation. Sigh.
+         *
+         * This workaround assumes that the new slot starts at the same
+         * address as the first existing one. If not or if some overlapping
+         * slot comes around later, we will fail (not seen in practice so far)
+         * - and actually require a recent KVM version. */
+        if (old.start_addr == start_addr && old.memory_size < size &&
+            flags < IO_MEM_UNASSIGNED) {
+            mem = kvm_alloc_slot(s);
+            mem->memory_size = old.memory_size;
+            mem->start_addr = old.start_addr;
+            mem->phys_offset = old.phys_offset;
             mem->flags = 0;
 
-            kvm_set_user_memory_region(s, mem);
-        } else if (start_addr >= mem->start_addr &&
-                   (start_addr + size) <= (mem->start_addr +
-                                           mem->memory_size)) {
-            KVMSlot slot;
-            target_phys_addr_t mem_start;
-            ram_addr_t mem_size, mem_offset;
-
-            /* Not splitting */
-            if ((phys_offset - (start_addr - mem->start_addr)) == 
-                mem->phys_offset)
-                return;
+            err = kvm_set_user_memory_region(s, mem);
+            if (err) {
+                fprintf(stderr, "%s: error updating slot: %s\n", __func__,
+                        strerror(-err));
+                abort();
+            }
 
-            /* unregister whole slot */
-            memcpy(&slot, mem, sizeof(slot));
-            mem->memory_size = 0;
-            kvm_set_user_memory_region(s, mem);
+            start_addr += old.memory_size;
+            phys_offset += old.memory_size;
+            size -= old.memory_size;
+            continue;
+        }
 
-            /* register prefix slot */
-            mem_start = slot.start_addr;
-            mem_size = start_addr - slot.start_addr;
-            mem_offset = slot.phys_offset;
-            if (mem_size)
-                kvm_set_phys_mem(mem_start, mem_size, mem_offset);
+        /* register prefix slot */
+        if (old.start_addr < start_addr) {
+            mem = kvm_alloc_slot(s);
+            mem->memory_size = start_addr - old.start_addr;
+            mem->start_addr = old.start_addr;
+            mem->phys_offset = old.phys_offset;
+            mem->flags = 0;
 
-            /* register new slot */
-            kvm_set_phys_mem(start_addr, size, phys_offset);
+            err = kvm_set_user_memory_region(s, mem);
+            if (err) {
+                fprintf(stderr, "%s: error registering prefix slot: %s\n",
+                        __func__, strerror(-err));
+                abort();
+            }
+        }
 
-            /* register suffix slot */
-            mem_start = start_addr + size;
-            mem_offset += mem_size + size;
-            mem_size = slot.memory_size - mem_size - size;
-            if (mem_size)
-                kvm_set_phys_mem(mem_start, mem_size, mem_offset);
+        /* register suffix slot */
+        if (old.start_addr + old.memory_size > start_addr + size) {
+            ram_addr_t size_delta;
 
-            return;
-        } else {
-            printf("Registering overlapping slot\n");
-            abort();
+            mem = kvm_alloc_slot(s);
+            mem->start_addr = start_addr + size;
+            size_delta = mem->start_addr - old.start_addr;
+            mem->memory_size = old.memory_size - size_delta;
+            mem->phys_offset = old.phys_offset + size_delta;
+            mem->flags = 0;
+
+            err = kvm_set_user_memory_region(s, mem);
+            if (err) {
+                fprintf(stderr, "%s: error registering suffix slot: %s\n",
+                        __func__, strerror(-err));
+                abort();
+            }
         }
     }
+
+    /* in case the KVM bug workaround already "consumed" the new slot */
+    if (!size)
+        return;
+
     /* KVM does not need to know about this memory */
     if (flags >= IO_MEM_UNASSIGNED)
         return;
@@ -636,8 +710,12 @@ void kvm_set_phys_mem(target_phys_addr_t start_addr,
     mem->phys_offset = phys_offset;
     mem->flags = 0;
 
-    kvm_set_user_memory_region(s, mem);
-    /* FIXME deal with errors */
+    err = kvm_set_user_memory_region(s, mem);
+    if (err) {
+        fprintf(stderr, "%s: error registering slot: %s\n", __func__,
+                strerror(-err));
+        abort();
+    }
 }
 
 int kvm_ioctl(KVMState *s, int type, ...)
@@ -696,11 +774,28 @@ int kvm_has_sync_mmu(void)
 #ifdef KVM_CAP_SYNC_MMU
     KVMState *s = kvm_state;
 
-    if (kvm_ioctl(s, KVM_CHECK_EXTENSION, KVM_CAP_SYNC_MMU) > 0)
-        return 1;
+    return kvm_check_extension(s, KVM_CAP_SYNC_MMU);
+#else
+    return 0;
 #endif
+}
 
-    return 0;
+void kvm_setup_guest_memory(void *start, size_t size)
+{
+    if (!kvm_has_sync_mmu()) {
+#ifdef MADV_DONTFORK
+        int ret = madvise(start, size, MADV_DONTFORK);
+
+        if (ret) {
+            perror("madvice");
+            exit(1);
+        }
+#else
+        fprintf(stderr,
+                "Need MADV_DONTFORK in absence of synchronous KVM MMU\n");
+        exit(1);
+#endif
+    }
 }
 
 #ifdef KVM_CAP_SET_GUEST_DEBUG