[Qemu-devel] [PATCH 0/2] RAM cleanup

[Qemu-devel] [PATCH 0/2] RAM cleanup

[Alex Williamson]

These are the first two patches from the RFC sent out a couple days
ago.  There seemed to be concensus that these were headed in the
right direction, so I'll post them for commit while we decided how
to name RAMBlocks.

Changes - rename 'ram' to 'ram_list' to avoid polluting the global
namespace quite so much, restore the error message and abort for
bad address in qemu_get_ram_ptr().

Alex

---

Alex Williamson (2):
      ram_blocks: Convert to a QLIST
      qemu_ram_alloc: Remove duplicate code


 arch_init.c |   14 ++++---
 cpu-all.h   |   28 +++++++++++----
 exec.c      |  112 ++++++++++++++++++-----------------------------------------
 3 files changed, 62 insertions(+), 92 deletions(-)

[Qemu-devel] [PATCH 1/2] qemu_ram_alloc: Remove duplicate code

[Alex Williamson]

No reason not to call qemu_ram_map() once we have the allocation
and remove duplicate code.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Chris Wright <chrisw@redhat.com>
---

 exec.c |   37 ++++++++++---------------------------
 1 files changed, 10 insertions(+), 27 deletions(-)

diff --git a/exec.c b/exec.c
index 7b0e1c5..c60f9e7 100644
--- a/exec.c
+++ b/exec.c
@@ -2816,18 +2816,17 @@ ram_addr_t qemu_ram_map(ram_addr_t size, void *host)
 
 ram_addr_t qemu_ram_alloc(ram_addr_t size)
 {
-    RAMBlock *new_block;
+    void *host;
 
     size = TARGET_PAGE_ALIGN(size);
-    new_block = qemu_malloc(sizeof(*new_block));
 
     if (mem_path) {
 #if defined (__linux__) && !defined(TARGET_S390X)
-        new_block->host = file_ram_alloc(size, mem_path);
-        if (!new_block->host) {
-            new_block->host = qemu_vmalloc(size);
+        host = file_ram_alloc(size, mem_path);
+        if (!host) {
+            host = qemu_vmalloc(size);
 #ifdef MADV_MERGEABLE
-            madvise(new_block->host, size, MADV_MERGEABLE);
+            madvise(host, size, MADV_MERGEABLE);
 #endif
         }
 #else
@@ -2837,33 +2836,17 @@ ram_addr_t qemu_ram_alloc(ram_addr_t size)
     } else {
 #if defined(TARGET_S390X) && defined(CONFIG_KVM)
         /* XXX S390 KVM requires the topmost vma of the RAM to be < 256GB */
-        new_block->host = mmap((void*)0x1000000, size,
-                                PROT_EXEC|PROT_READ|PROT_WRITE,
-                                MAP_SHARED | MAP_ANONYMOUS, -1, 0);
+        host = mmap((void*)0x1000000, size, PROT_EXEC|PROT_READ|PROT_WRITE,
+                    MAP_SHARED | MAP_ANONYMOUS, -1, 0);
 #else
-        new_block->host = qemu_vmalloc(size);
+        host = qemu_vmalloc(size);
 #endif
 #ifdef MADV_MERGEABLE
-        madvise(new_block->host, size, MADV_MERGEABLE);
+        madvise(host, size, MADV_MERGEABLE);
 #endif
     }
-    new_block->offset = last_ram_offset;
-    new_block->length = size;
-
-    new_block->next = ram_blocks;
-    ram_blocks = new_block;
-
-    phys_ram_dirty = qemu_realloc(phys_ram_dirty,
-        (last_ram_offset + size) >> TARGET_PAGE_BITS);
-    memset(phys_ram_dirty + (last_ram_offset >> TARGET_PAGE_BITS),
-           0xff, size >> TARGET_PAGE_BITS);
-
-    last_ram_offset += size;
 
-    if (kvm_enabled())
-        kvm_setup_guest_memory(new_block->host, size);
-
-    return new_block->offset;
+    return qemu_ram_map(size, host);
 }
 
 void qemu_ram_free(ram_addr_t addr)

[Qemu-devel] [PATCH 2/2] ram_blocks: Convert to a QLIST

[Alex Williamson]

This makes the RAM block list easier to manipulate.  Also incorporate
relevant variables into the RAMList struct.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Chris Wright <chrisw@redhat.com>
---

 arch_init.c |   14 ++++++-----
 cpu-all.h   |   28 ++++++++++++++++------
 exec.c      |   75 ++++++++++++++++++++---------------------------------------
 3 files changed, 52 insertions(+), 65 deletions(-)

diff --git a/arch_init.c b/arch_init.c
index 8e849a8..eb5b67c 100644
--- a/arch_init.c
+++ b/arch_init.c
@@ -110,7 +110,7 @@ static int ram_save_block(QEMUFile *f)
     ram_addr_t addr = 0;
     int bytes_sent = 0;
 
-    while (addr < last_ram_offset) {
+    while (addr < ram_list.last_offset) {
         if (cpu_physical_memory_get_dirty(current_addr, MIGRATION_DIRTY_FLAG)) {
             uint8_t *p;
 
@@ -133,7 +133,7 @@ static int ram_save_block(QEMUFile *f)
             break;
         }
         addr += TARGET_PAGE_SIZE;
-        current_addr = (saved_addr + addr) % last_ram_offset;
+        current_addr = (saved_addr + addr) % ram_list.last_offset;
     }
 
     return bytes_sent;
@@ -146,7 +146,7 @@ static ram_addr_t ram_save_remaining(void)
     ram_addr_t addr;
     ram_addr_t count = 0;
 
-    for (addr = 0; addr < last_ram_offset; addr += TARGET_PAGE_SIZE) {
+    for (addr = 0; addr < ram_list.last_offset; addr += TARGET_PAGE_SIZE) {
         if (cpu_physical_memory_get_dirty(addr, MIGRATION_DIRTY_FLAG)) {
             count++;
         }
@@ -167,7 +167,7 @@ uint64_t ram_bytes_transferred(void)
 
 uint64_t ram_bytes_total(void)
 {
-    return last_ram_offset;
+    return ram_list.last_offset;
 }
 
 int ram_save_live(Monitor *mon, QEMUFile *f, int stage, void *opaque)
@@ -191,7 +191,7 @@ int ram_save_live(Monitor *mon, QEMUFile *f, int stage, void *opaque)
         bytes_transferred = 0;
 
         /* Make sure all dirty bits are set */
-        for (addr = 0; addr < last_ram_offset; addr += TARGET_PAGE_SIZE) {
+        for (addr = 0; addr < ram_list.last_offset; addr += TARGET_PAGE_SIZE) {
             if (!cpu_physical_memory_get_dirty(addr, MIGRATION_DIRTY_FLAG)) {
                 cpu_physical_memory_set_dirty(addr);
             }
@@ -200,7 +200,7 @@ int ram_save_live(Monitor *mon, QEMUFile *f, int stage, void *opaque)
         /* Enable dirty memory tracking */
         cpu_physical_memory_set_dirty_tracking(1);
 
-        qemu_put_be64(f, last_ram_offset | RAM_SAVE_FLAG_MEM_SIZE);
+        qemu_put_be64(f, ram_list.last_offset | RAM_SAVE_FLAG_MEM_SIZE);
     }
 
     bytes_transferred_last = bytes_transferred;
@@ -259,7 +259,7 @@ int ram_load(QEMUFile *f, void *opaque, int version_id)
         addr &= TARGET_PAGE_MASK;
 
         if (flags & RAM_SAVE_FLAG_MEM_SIZE) {
-            if (addr != last_ram_offset) {
+            if (addr != ram_list.last_offset) {
                 return -EINVAL;
             }
         }
diff --git a/cpu-all.h b/cpu-all.h
index 77eaf85..e31c2de 100644
--- a/cpu-all.h
+++ b/cpu-all.h
@@ -859,9 +859,21 @@ target_phys_addr_t cpu_get_phys_page_debug(CPUState *env, target_ulong addr);
 /* memory API */
 
 extern int phys_ram_fd;
-extern uint8_t *phys_ram_dirty;
 extern ram_addr_t ram_size;
-extern ram_addr_t last_ram_offset;
+
+typedef struct RAMBlock {
+    uint8_t *host;
+    ram_addr_t offset;
+    ram_addr_t length;
+    QLIST_ENTRY(RAMBlock) next;
+} RAMBlock;
+
+typedef struct RAMList {
+    uint8_t *phys_dirty;
+    ram_addr_t last_offset;
+    QLIST_HEAD(ram, RAMBlock) blocks;
+} RAMList;
+extern RAMList ram_list;
 
 extern const char *mem_path;
 extern int mem_prealloc;
@@ -891,29 +903,29 @@ extern int mem_prealloc;
 /* read dirty bit (return 0 or 1) */
 static inline int cpu_physical_memory_is_dirty(ram_addr_t addr)
 {
-    return phys_ram_dirty[addr >> TARGET_PAGE_BITS] == 0xff;
+    return ram_list.phys_dirty[addr >> TARGET_PAGE_BITS] == 0xff;
 }
 
 static inline int cpu_physical_memory_get_dirty_flags(ram_addr_t addr)
 {
-    return phys_ram_dirty[addr >> TARGET_PAGE_BITS];
+    return ram_list.phys_dirty[addr >> TARGET_PAGE_BITS];
 }
 
 static inline int cpu_physical_memory_get_dirty(ram_addr_t addr,
                                                 int dirty_flags)
 {
-    return phys_ram_dirty[addr >> TARGET_PAGE_BITS] & dirty_flags;
+    return ram_list.phys_dirty[addr >> TARGET_PAGE_BITS] & dirty_flags;
 }
 
 static inline void cpu_physical_memory_set_dirty(ram_addr_t addr)
 {
-    phys_ram_dirty[addr >> TARGET_PAGE_BITS] = 0xff;
+    ram_list.phys_dirty[addr >> TARGET_PAGE_BITS] = 0xff;
 }
 
 static inline int cpu_physical_memory_set_dirty_flags(ram_addr_t addr,
                                                       int dirty_flags)
 {
-    return phys_ram_dirty[addr >> TARGET_PAGE_BITS] |= dirty_flags;
+    return ram_list.phys_dirty[addr >> TARGET_PAGE_BITS] |= dirty_flags;
 }
 
 static inline void cpu_physical_memory_mask_dirty_range(ram_addr_t start,
@@ -925,7 +937,7 @@ static inline void cpu_physical_memory_mask_dirty_range(ram_addr_t start,
 
     len = length >> TARGET_PAGE_BITS;
     mask = ~dirty_flags;
-    p = phys_ram_dirty + (start >> TARGET_PAGE_BITS);
+    p = ram_list.phys_dirty + (start >> TARGET_PAGE_BITS);
     for (i = 0; i < len; i++) {
         p[i] &= mask;
     }
diff --git a/exec.c b/exec.c
index c60f9e7..96aa989 100644
--- a/exec.c
+++ b/exec.c
@@ -116,21 +116,9 @@ uint8_t *code_gen_ptr;
 
 #if !defined(CONFIG_USER_ONLY)
 int phys_ram_fd;
-uint8_t *phys_ram_dirty;
 static int in_migration;
 
-typedef struct RAMBlock {
-    uint8_t *host;
-    ram_addr_t offset;
-    ram_addr_t length;
-    struct RAMBlock *next;
-} RAMBlock;
-
-static RAMBlock *ram_blocks;
-/* TODO: When we implement (and use) ram deallocation (e.g. for hotplug)
-   then we can no longer assume contiguous ram offsets, and external uses
-   of this variable will break.  */
-ram_addr_t last_ram_offset;
+RAMList ram_list = { .blocks = QLIST_HEAD_INITIALIZER(ram_list) };
 #endif
 
 CPUState *first_cpu;
@@ -2795,18 +2783,17 @@ ram_addr_t qemu_ram_map(ram_addr_t size, void *host)
 
     new_block->host = host;
 
-    new_block->offset = last_ram_offset;
+    new_block->offset = ram_list.last_offset;
     new_block->length = size;
 
-    new_block->next = ram_blocks;
-    ram_blocks = new_block;
+    QLIST_INSERT_HEAD(&ram_list.blocks, new_block, next);
 
-    phys_ram_dirty = qemu_realloc(phys_ram_dirty,
-        (last_ram_offset + size) >> TARGET_PAGE_BITS);
-    memset(phys_ram_dirty + (last_ram_offset >> TARGET_PAGE_BITS),
+    ram_list.phys_dirty = qemu_realloc(ram_list.phys_dirty,
+        (ram_list.last_offset + size) >> TARGET_PAGE_BITS);
+    memset(ram_list.phys_dirty + (ram_list.last_offset >> TARGET_PAGE_BITS),
            0xff, size >> TARGET_PAGE_BITS);
 
-    last_ram_offset += size;
+    ram_list.last_offset += size;
 
     if (kvm_enabled())
         kvm_setup_guest_memory(new_block->host, size);
@@ -2864,31 +2851,20 @@ void qemu_ram_free(ram_addr_t addr)
  */
 void *qemu_get_ram_ptr(ram_addr_t addr)
 {
-    RAMBlock *prev;
-    RAMBlock **prevp;
     RAMBlock *block;
 
-    prev = NULL;
-    prevp = &ram_blocks;
-    block = ram_blocks;
-    while (block && (block->offset > addr
-                     || block->offset + block->length <= addr)) {
-        if (prev)
-          prevp = &prev->next;
-        prev = block;
-        block = block->next;
-    }
-    if (!block) {
-        fprintf(stderr, "Bad ram offset %" PRIx64 "\n", (uint64_t)addr);
-        abort();
-    }
-    /* Move this entry to to start of the list.  */
-    if (prev) {
-        prev->next = block->next;
-        block->next = *prevp;
-        *prevp = block;
+    QLIST_FOREACH(block, &ram_list.blocks, next) {
+        if (addr - block->offset < block->length) {
+            QLIST_REMOVE(block, next);
+            QLIST_INSERT_HEAD(&ram_list.blocks, block, next);
+            return block->host + (addr - block->offset);
+        }
     }
-    return block->host + (addr - block->offset);
+
+    fprintf(stderr, "Bad ram offset %" PRIx64 "\n", (uint64_t)addr);
+    abort();
+
+    return NULL;
 }
 
 int do_qemu_ram_addr_from_host(void *ptr, ram_addr_t *ram_addr)
@@ -2896,15 +2872,14 @@ int do_qemu_ram_addr_from_host(void *ptr, ram_addr_t *ram_addr)
     RAMBlock *block;
     uint8_t *host = ptr;
 
-    block = ram_blocks;
-    while (block && (block->host > host
-                     || block->host + block->length <= host)) {
-        block = block->next;
+    QLIST_FOREACH(block, &ram_list.blocks, next) {
+        if (host - block->host < block->length) {
+            *ram_addr = block->offset + (host - block->host);
+            return 0;
+        }
     }
-    if (!block)
-        return -1;
-    *ram_addr = block->offset + (host - block->host);
-    return 0;
+
+    return -1;
 }
 
 /* Some of the softmmu routines need to translate from a host pointer

[Qemu-devel] Re: [PATCH 0/2] RAM cleanup

[Alex Williamson]

On Fri, 2010-06-11 at 08:15 -0600, Alex Williamson wrote:
> These are the first two patches from the RFC sent out a couple days
> ago.  There seemed to be concensus that these were headed in the
> right direction, so I'll post them for commit while we decided how
> to name RAMBlocks.
> 
> Changes - rename 'ram' to 'ram_list' to avoid polluting the global
> namespace quite so much, restore the error message and abort for
> bad address in qemu_get_ram_ptr().
> 
> Alex
> 
> ---
> 
> Alex Williamson (2):
>       ram_blocks: Convert to a QLIST
>       qemu_ram_alloc: Remove duplicate code

Marcelo pointed out that I forgot to rebase against qemu, where
qemu_ram_map doesn't exist yet.  So I'll drop the first patch, we can do
that in kvm later, and I'll repost the qemu version of the second.

Alex