From 962acd498b11ae5ccc040d76ec89990add119dec Mon Sep 17 00:00:00 2001
From: libai <libai12@huawei.com>
Date: Mon, 4 Dec 2023 15:09:26 +0800
Subject: [PATCH] vhost: introduce bytemap for vhost backend logging

As vhost backend may use bytemap for logging, when get log_size
of vhost device, check whether vhost device support VHOST_BACKEND_F_BYTEMAPLOG.
If vhost device support, use bytemap for logging.

By the way, add log_resize func pointer check and vhost_log_sync return
value check.

Signed-off-by: libai <libai12@huawei.com>
---
 hw/virtio/vhost.c         | 89 ++++++++++++++++++++++++++++++++++++---
 include/exec/memory.h     |  9 ++++
 include/exec/ram_addr.h   | 44 +++++++++++++++++++
 include/hw/virtio/vhost.h |  1 +
 system/physmem.c          | 11 +++++
 5 files changed, 148 insertions(+), 6 deletions(-)

diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index 038ac37dd0..438182d850 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -29,6 +29,7 @@
 #include "migration/migration.h"
 #include "sysemu/dma.h"
 #include "trace.h"
+#include "qapi/qapi-commands-migration.h"
 
 /* enabled until disconnected backend stabilizes */
 #define _VHOST_DEBUG 1
@@ -44,6 +45,11 @@
     do { } while (0)
 #endif
 
+static inline bool vhost_bytemap_log_support(struct vhost_dev *dev)
+{
+    return (dev->backend_cap & BIT_ULL(VHOST_BACKEND_F_BYTEMAPLOG));
+}
+
 static struct vhost_log *vhost_log;
 static struct vhost_log *vhost_log_shm;
 
@@ -232,12 +238,40 @@ static int vhost_sync_dirty_bitmap(struct vhost_dev *dev,
     return 0;
 }
 
+static int vhost_sync_dirty_bytemap(struct vhost_dev *dev,
+                                    MemoryRegionSection *section)
+{
+    unsigned long *bytemap = dev->log->log;
+    return memory_section_set_dirty_bytemap(section, bytemap);
+}
+
 static void vhost_log_sync(MemoryListener *listener,
                           MemoryRegionSection *section)
 {
     struct vhost_dev *dev = container_of(listener, struct vhost_dev,
                                          memory_listener);
-    vhost_sync_dirty_bitmap(dev, section, 0x0, ~0x0ULL);
+    MigrationState *ms = migrate_get_current();
+
+    if (!dev->log_enabled || !dev->started) {
+        return;
+    }
+
+    if (dev->vhost_ops->vhost_log_sync) {
+        int r = dev->vhost_ops->vhost_log_sync(dev);
+        if (r < 0) {
+            error_report("Failed to sync dirty log: 0x%x\n", r);
+            if (migration_is_running(ms->state)) {
+                qmp_migrate_cancel(NULL);
+            }
+            return;
+        }
+    }
+
+    if (vhost_bytemap_log_support(dev)) {
+        vhost_sync_dirty_bytemap(dev, section);
+    } else {
+        vhost_sync_dirty_bitmap(dev, section, 0x0, ~0x0ULL);
+    }
 }
 
 static void vhost_log_sync_range(struct vhost_dev *dev,
@@ -247,7 +281,11 @@ static void vhost_log_sync_range(struct vhost_dev *dev,
     /* FIXME: this is N^2 in number of sections */
     for (i = 0; i < dev->n_mem_sections; ++i) {
         MemoryRegionSection *section = &dev->mem_sections[i];
-        vhost_sync_dirty_bitmap(dev, section, first, last);
+        if (vhost_bytemap_log_support(dev)) {
+            vhost_sync_dirty_bytemap(dev, section);
+        } else {
+            vhost_sync_dirty_bitmap(dev, section, first, last);
+        }
     }
 }
 
@@ -255,11 +293,19 @@ static uint64_t vhost_get_log_size(struct vhost_dev *dev)
 {
     uint64_t log_size = 0;
     int i;
+    uint64_t vhost_log_chunk_size;
+
+    if (vhost_bytemap_log_support(dev)) {
+        vhost_log_chunk_size = VHOST_LOG_CHUNK_BYTES;
+    } else {
+        vhost_log_chunk_size = VHOST_LOG_CHUNK;
+    }
+
     for (i = 0; i < dev->mem->nregions; ++i) {
         struct vhost_memory_region *reg = dev->mem->regions + i;
         uint64_t last = range_get_last(reg->guest_phys_addr,
                                        reg->memory_size);
-        log_size = MAX(log_size, last / VHOST_LOG_CHUNK + 1);
+        log_size = MAX(log_size, last / vhost_log_chunk_size + 1);
     }
     return log_size;
 }
@@ -377,12 +423,21 @@ static bool vhost_dev_log_is_shared(struct vhost_dev *dev)
            dev->vhost_ops->vhost_requires_shm_log(dev);
 }
 
-static inline void vhost_dev_log_resize(struct vhost_dev *dev, uint64_t size)
+static inline int vhost_dev_log_resize(struct vhost_dev *dev, uint64_t size)
 {
     struct vhost_log *log = vhost_log_get(size, vhost_dev_log_is_shared(dev));
-    uint64_t log_base = (uintptr_t)log->log;
+    uint64_t log_base;
+    int log_fd;
     int r;
 
+    if (!log) {
+        r = -ENOMEM;
+        goto out;
+    }
+
+    log_base = (uint64_t)log->log;
+    log_fd = log_fd;
+
     /* inform backend of log switching, this must be done before
        releasing the current log, to ensure no logging is lost */
     r = dev->vhost_ops->vhost_set_log_base(dev, log_base, log);
@@ -390,9 +445,19 @@ static inline void vhost_dev_log_resize(struct vhost_dev *dev, uint64_t size)
         VHOST_OPS_DEBUG(r, "vhost_set_log_base failed");
     }
 
+    if (dev->vhost_ops->vhost_set_log_size) {
+        r = dev->vhost_ops->vhost_set_log_size(dev, size, dev->log);
+        if (r < 0) {
+            VHOST_OPS_DEBUG(r, "vhost_set_log_size failed");
+        }
+    }
+
     vhost_log_put(dev, true);
     dev->log = log;
     dev->log_size = size;
+
+out:
+    return r;
 }
 
 static void *vhost_memory_map(struct vhost_dev *dev, hwaddr addr,
@@ -1018,7 +1083,11 @@ static int vhost_migration_log(MemoryListener *listener, bool enable)
         }
         vhost_log_put(dev, false);
     } else {
-        vhost_dev_log_resize(dev, vhost_get_log_size(dev));
+        r = vhost_dev_log_resize(dev, vhost_get_log_size(dev));
+        if ( r < 0 ) {
+            return r;
+        }
+
         r = vhost_dev_set_log(dev, true);
         if (r < 0) {
             goto check_dev_state;
@@ -2057,6 +2126,14 @@ int vhost_dev_start(struct vhost_dev *hdev, VirtIODevice *vdev, bool vrings)
             VHOST_OPS_DEBUG(r, "vhost_set_log_base failed");
             goto fail_log;
         }
+
+        if (hdev->vhost_ops->vhost_set_log_size) {
+            r = hdev->vhost_ops->vhost_set_log_size(hdev, hdev->log_size, hdev->log);
+            if (r < 0) {
+                VHOST_OPS_DEBUG(r, "vhost_set_log_size failed");
+                goto fail_log;
+            }
+        }
     }
     if (vrings) {
         r = vhost_dev_set_vring_enable(hdev, true);
diff --git a/include/exec/memory.h b/include/exec/memory.h
index 831f7c996d..e131c2682c 100644
--- a/include/exec/memory.h
+++ b/include/exec/memory.h
@@ -2594,6 +2594,15 @@ MemTxResult memory_region_dispatch_write(MemoryRegion *mr,
                                          MemOp op,
                                          MemTxAttrs attrs);
 
+/**
+ * memory_section_set_dirty_bytemap: Mark a range of bytes as dirty for a memory section
+ * using a bytemap
+ *
+ * @section: the memory section being dirtied.
+ * @bytemap: bytemap that stores dirty page range information.
+ */
+int64_t memory_section_set_dirty_bytemap(MemoryRegionSection *section, unsigned long *bytemap);
+
 /**
  * address_space_init: initializes an address space
  *
diff --git a/include/exec/ram_addr.h b/include/exec/ram_addr.h
index 90676093f5..ef6988b445 100644
--- a/include/exec/ram_addr.h
+++ b/include/exec/ram_addr.h
@@ -535,5 +535,49 @@ uint64_t cpu_physical_memory_sync_dirty_bitmap(RAMBlock *rb,
 
     return num_dirty;
 }
+
+#define BYTES_PER_LONG           (sizeof(unsigned long))
+#define BYTE_WORD(nr)            ((nr) / BYTES_PER_LONG)
+#define BYTES_TO_LONGS(nr)       DIV_ROUND_UP(nr, BYTES_PER_LONG)
+
+static inline int64_t _set_dirty_bytemap_atomic(unsigned long *bytemap, unsigned long cur_pfn)
+{
+    char *byte_of_long = (char *)bytemap;
+    int i;
+    int64_t dirty_num = 0;
+
+    for (i = 0; i < BYTES_PER_LONG; i++) {
+        if (byte_of_long[i]) {
+            cpu_physical_memory_set_dirty_range((cur_pfn + i) << TARGET_PAGE_BITS,
+                                                TARGET_PAGE_SIZE,
+                                                1 << DIRTY_MEMORY_MIGRATION);
+            /* Per byte ops, no need to atomic_xchg */
+            byte_of_long[i] = 0;
+            dirty_num++;
+        }
+    }
+
+    return dirty_num;
+}
+
+static inline int64_t cpu_physical_memory_set_dirty_bytemap(unsigned long *bytemap,
+                                      ram_addr_t start,
+                                      ram_addr_t pages)
+{
+    unsigned long i;
+    unsigned long len = BYTES_TO_LONGS(pages);
+    unsigned long pfn = (start >> TARGET_PAGE_BITS) /
+                         BYTES_PER_LONG * BYTES_PER_LONG;
+    int64_t dirty_mig_bits = 0;
+
+    for (i = 0; i < len; i++) {
+        if (bytemap[i]) {
+            dirty_mig_bits += _set_dirty_bytemap_atomic(&bytemap[i],
+                                                        pfn + BYTES_PER_LONG * i);
+        }
+    }
+
+    return dirty_mig_bits;
+}
 #endif
 #endif
diff --git a/include/hw/virtio/vhost.h b/include/hw/virtio/vhost.h
index 444ca0ad42..6ae86833e3 100644
--- a/include/hw/virtio/vhost.h
+++ b/include/hw/virtio/vhost.h
@@ -43,6 +43,7 @@ typedef unsigned long vhost_log_chunk_t;
 #define VHOST_LOG_PAGE 0x1000
 #define VHOST_LOG_BITS (8 * sizeof(vhost_log_chunk_t))
 #define VHOST_LOG_CHUNK (VHOST_LOG_PAGE * VHOST_LOG_BITS)
+#define VHOST_LOG_CHUNK_BYTES (VHOST_LOG_PAGE * sizeof(vhost_log_chunk_t))
 #define VHOST_INVALID_FEATURE_BIT   (0xff)
 #define VHOST_QUEUE_NUM_CONFIG_INR 0
 
diff --git a/system/physmem.c b/system/physmem.c
index f14d64819b..247c252e53 100644
--- a/system/physmem.c
+++ b/system/physmem.c
@@ -2602,6 +2602,17 @@ static void invalidate_and_set_dirty(MemoryRegion *mr, hwaddr addr,
     cpu_physical_memory_set_dirty_range(addr, length, dirty_log_mask);
 }
 
+int64_t memory_section_set_dirty_bytemap(MemoryRegionSection *section, unsigned long *bytemap)
+{
+    ram_addr_t start = section->offset_within_region +
+                       memory_region_get_ram_addr(section->mr);
+    ram_addr_t pages = int128_get64(section->size) >> TARGET_PAGE_BITS;
+
+    hwaddr idx = BYTE_WORD(
+        section->offset_within_address_space >> TARGET_PAGE_BITS);
+    return cpu_physical_memory_set_dirty_bytemap(bytemap + idx, start, pages);
+}
+
 void memory_region_flush_rom_device(MemoryRegion *mr, hwaddr addr, hwaddr size)
 {
     /*
-- 
2.27.0