qcow2: Discard freed clusters

index 5975dde..459a87e 100644
@@ -1737,120 +1737,10 @@ QemuOptsList qemu_drive_opts = {
     .name = "drive",
     .head = QTAILQ_HEAD_INITIALIZER(qemu_drive_opts.head),
     .desc = {
-        {
-            .name = "bus",
-            .type = QEMU_OPT_NUMBER,
-            .help = "bus number",
-        },{
-            .name = "unit",
-            .type = QEMU_OPT_NUMBER,
-            .help = "unit number (i.e. lun for scsi)",
-        },{
-            .name = "if",
-            .type = QEMU_OPT_STRING,
-            .help = "interface (ide, scsi, sd, mtd, floppy, pflash, virtio)",
-        },{
-            .name = "index",
-            .type = QEMU_OPT_NUMBER,
-            .help = "index number",
-        },{
-            .name = "cyls",
-            .type = QEMU_OPT_NUMBER,
-            .help = "number of cylinders (ide disk geometry)",
-        },{
-            .name = "heads",
-            .type = QEMU_OPT_NUMBER,
-            .help = "number of heads (ide disk geometry)",
-        },{
-            .name = "secs",
-            .type = QEMU_OPT_NUMBER,
-            .help = "number of sectors (ide disk geometry)",
-        },{
-            .name = "trans",
-            .type = QEMU_OPT_STRING,
-            .help = "chs translation (auto, lba. none)",
-        },{
-            .name = "media",
-            .type = QEMU_OPT_STRING,
-            .help = "media type (disk, cdrom)",
-        },{
-            .name = "snapshot",
-            .type = QEMU_OPT_BOOL,
-            .help = "enable/disable snapshot mode",
-        },{
-            .name = "file",
-            .type = QEMU_OPT_STRING,
-            .help = "disk image",
-        },{
-            .name = "discard",
-            .type = QEMU_OPT_STRING,
-            .help = "discard operation (ignore/off, unmap/on)",
-        },{
-            .name = "cache",
-            .type = QEMU_OPT_STRING,
-            .help = "host cache usage (none, writeback, writethrough, "
-                    "directsync, unsafe)",
-        },{
-            .name = "aio",
-            .type = QEMU_OPT_STRING,
-            .help = "host AIO implementation (threads, native)",
-        },{
-            .name = "format",
-            .type = QEMU_OPT_STRING,
-            .help = "disk format (raw, qcow2, ...)",
-        },{
-            .name = "serial",
-            .type = QEMU_OPT_STRING,
-            .help = "disk serial number",
-        },{
-            .name = "rerror",
-            .type = QEMU_OPT_STRING,
-            .help = "read error action",
-        },{
-            .name = "werror",
-            .type = QEMU_OPT_STRING,
-            .help = "write error action",
-        },{
-            .name = "addr",
-            .type = QEMU_OPT_STRING,
-            .help = "pci address (virtio only)",
-        },{
-            .name = "readonly",
-            .type = QEMU_OPT_BOOL,
-            .help = "open drive file as read-only",
-        },{
-            .name = "iops",
-            .type = QEMU_OPT_NUMBER,
-            .help = "limit total I/O operations per second",
-        },{
-            .name = "iops_rd",
-            .type = QEMU_OPT_NUMBER,
-            .help = "limit read operations per second",
-        },{
-            .name = "iops_wr",
-            .type = QEMU_OPT_NUMBER,
-            .help = "limit write operations per second",
-        },{
-            .name = "bps",
-            .type = QEMU_OPT_NUMBER,
-            .help = "limit total bytes per second",
-        },{
-            .name = "bps_rd",
-            .type = QEMU_OPT_NUMBER,
-            .help = "limit read bytes per second",
-        },{
-            .name = "bps_wr",
-            .type = QEMU_OPT_NUMBER,
-            .help = "limit write bytes per second",
-        },{
-            .name = "copy-on-read",
-            .type = QEMU_OPT_BOOL,
-            .help = "copy read data from backing file into image file",
-        },{
-            .name = "boot",
-            .type = QEMU_OPT_BOOL,
-            .help = "(deprecated, ignored)",
-        },
+        /*
+         * no elements => accept any params
+         * validation will happen later
+         */
         { /* end of list */ }
     },
 };

index 76f30e5..3191d6b 100644
@@ -98,14 +98,16 @@ int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
         goto fail;
     }
     g_free(s->l1_table);
-    qcow2_free_clusters(bs, s->l1_table_offset, s->l1_size * sizeof(uint64_t));
+    qcow2_free_clusters(bs, s->l1_table_offset, s->l1_size * sizeof(uint64_t),
+                        QCOW2_DISCARD_OTHER);
     s->l1_table_offset = new_l1_table_offset;
     s->l1_table = new_l1_table;
     s->l1_size = new_l1_size;
     return 0;
  fail:
     g_free(new_l1_table);
-    qcow2_free_clusters(bs, new_l1_table_offset, new_l1_size2);
+    qcow2_free_clusters(bs, new_l1_table_offset, new_l1_size2,
+                        QCOW2_DISCARD_OTHER);
     return ret;
 }
 

@@ -548,7 +550,8 @@ static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
 
         /* Then decrease the refcount of the old table */
         if (l2_offset) {
-            qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t));
+            qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
+                                QCOW2_DISCARD_OTHER);
         }
     }
 

@@ -715,10 +718,14 @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
     /*
      * If this was a COW, we need to decrease the refcount of the old cluster.
      * Also flush bs->file to get the right order for L2 and refcount update.
+     *
+     * Don't discard clusters that reach a refcount of 0 (e.g. compressed
+     * clusters), the next write will reuse them anyway.
      */
     if (j != 0) {
         for (i = 0; i < j; i++) {
-            qcow2_free_any_clusters(bs, be64_to_cpu(old_cluster[i]), 1);
+            qcow2_free_any_clusters(bs, be64_to_cpu(old_cluster[i]), 1,
+                                    QCOW2_DISCARD_NEVER);
         }
     }
 

@@ -1339,7 +1346,7 @@ static int discard_single_l2(BlockDriverState *bs, uint64_t offset,
         l2_table[l2_index + i] = cpu_to_be64(0);
 
         /* Then decrease the refcount */
-        qcow2_free_any_clusters(bs, old_offset, 1);
+        qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
     }
 
     ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);

@@ -1415,7 +1422,7 @@ static int zero_single_l2(BlockDriverState *bs, uint64_t offset,
         qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
         if (old_offset & QCOW_OFLAG_COMPRESSED) {
             l2_table[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
-            qcow2_free_any_clusters(bs, old_offset, 1);
+            qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
         } else {
             l2_table[l2_index + i] |= cpu_to_be64(QCOW_OFLAG_ZERO);
         }

169 lines not on this page · raw message

index 992a5c8..0caac90 100644
@@ -262,7 +262,8 @@ static int qcow2_write_snapshots(BlockDriverState *bs)
     }
 
     /* free the old snapshot table */
-    qcow2_free_clusters(bs, s->snapshots_offset, s->snapshots_size);
+    qcow2_free_clusters(bs, s->snapshots_offset, s->snapshots_size,
+                        QCOW2_DISCARD_SNAPSHOT);
     s->snapshots_offset = snapshots_offset;
     s->snapshots_size = snapshots_size;
     return 0;

@@ -569,7 +570,8 @@ int qcow2_snapshot_delete(BlockDriverState *bs, const char *snapshot_id)
     if (ret < 0) {
         return ret;
     }
-    qcow2_free_clusters(bs, sn.l1_table_offset, sn.l1_size * sizeof(uint64_t));
+    qcow2_free_clusters(bs, sn.l1_table_offset, sn.l1_size * sizeof(uint64_t),
+                        QCOW2_DISCARD_SNAPSHOT);
 
     /* must update the copied flag on the current cluster offsets */
     ret = qcow2_update_snapshot_refcount(bs, s->l1_table_offset, s->l1_size, 0);

index 0fa5cb2..e28ea47 100644
@@ -1196,7 +1196,8 @@ static int preallocate(BlockDriverState *bs)
 
         ret = qcow2_alloc_cluster_link_l2(bs, meta);
         if (ret < 0) {
-            qcow2_free_any_clusters(bs, meta->alloc_offset, meta->nb_clusters);
+            qcow2_free_any_clusters(bs, meta->alloc_offset, meta->nb_clusters,
+                                    QCOW2_DISCARD_NEVER);
             return ret;
         }
 

index 6959c6a..64a6479 100644
@@ -129,6 +129,15 @@ enum {
     QCOW2_COMPAT_FEAT_MASK            = QCOW2_COMPAT_LAZY_REFCOUNTS,
 };
 
+enum qcow2_discard_type {
+    QCOW2_DISCARD_NEVER = 0,
+    QCOW2_DISCARD_ALWAYS,
+    QCOW2_DISCARD_REQUEST,
+    QCOW2_DISCARD_SNAPSHOT,
+    QCOW2_DISCARD_OTHER,
+    QCOW2_DISCARD_MAX
+};
+
 typedef struct Qcow2Feature {
     uint8_t type;
     uint8_t bit;

@@ -349,9 +358,10 @@ int qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset,
     int nb_clusters);
 int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size);
 void qcow2_free_clusters(BlockDriverState *bs,
-    int64_t offset, int64_t size);
-void qcow2_free_any_clusters(BlockDriverState *bs,
-    uint64_t cluster_offset, int nb_clusters);
+                          int64_t offset, int64_t size,
+                          enum qcow2_discard_type type);
+void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry,
+                             int nb_clusters, enum qcow2_discard_type type);
 
 int qcow2_update_snapshot_refcount(BlockDriverState *bs,
     int64_t l1_table_offset, int l1_size, int addend);

index 6d35e49..7488988 100644
@@ -488,6 +488,11 @@ static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
             s->free_cluster_index = cluster_index;
         }
         refcount_block[block_index] = cpu_to_be16(refcount);
+        if (refcount == 0 && s->discard_passthrough[type]) {
+            /* Try discarding, ignore errors */
+            /* FIXME Doing this cluster by cluster will be painfully slow */
+            bdrv_discard(bs->file, cluster_offset, 1);
+        }
     }
 
     ret = 0;

index e28ea47..ef8a2ca 100644
@@ -295,6 +295,22 @@ static QemuOptsList qcow2_runtime_opts = {
             .type = QEMU_OPT_BOOL,
             .help = "Postpone refcount updates",
         },
+        {
+            .name = QCOW2_OPT_DISCARD_REQUEST,
+            .type = QEMU_OPT_BOOL,
+            .help = "Pass guest discard requests to the layer below",
+        },
+        {
+            .name = QCOW2_OPT_DISCARD_SNAPSHOT,
+            .type = QEMU_OPT_BOOL,
+            .help = "Generate discard requests when snapshot related space "
+                    "is freed",
+        },
+        {
+            .name = QCOW2_OPT_DISCARD_OTHER,
+            .type = QEMU_OPT_BOOL,
+            .help = "Generate discard requests when other clusters are freed",
+        },
         { /* end of list */ }
     },
 };

@@ -532,6 +548,16 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags)
     s->use_lazy_refcounts = qemu_opt_get_bool(opts, QCOW2_OPT_LAZY_REFCOUNTS,
         (s->compatible_features & QCOW2_COMPAT_LAZY_REFCOUNTS));
 
+    s->discard_passthrough[QCOW2_DISCARD_NEVER] = false;
+    s->discard_passthrough[QCOW2_DISCARD_ALWAYS] = true;
+    s->discard_passthrough[QCOW2_DISCARD_REQUEST] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_REQUEST,
+                          flags & BDRV_O_UNMAP);
+    s->discard_passthrough[QCOW2_DISCARD_SNAPSHOT] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_SNAPSHOT, true);
+    s->discard_passthrough[QCOW2_DISCARD_OTHER] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_OTHER, false);
+
     qemu_opts_del(opts);
 
     if (s->use_lazy_refcounts && s->qcow_version < 3) {

index 64a6479..6f91b9a 100644
@@ -60,6 +60,9 @@
 
 
 #define QCOW2_OPT_LAZY_REFCOUNTS "lazy_refcounts"
+#define QCOW2_OPT_DISCARD_REQUEST "pass_discard_request"
+#define QCOW2_OPT_DISCARD_SNAPSHOT "pass_discard_snapshot"
+#define QCOW2_OPT_DISCARD_OTHER "pass_discard_other"
 
 typedef struct QCowHeader {
     uint32_t magic;

@@ -187,6 +190,8 @@ typedef struct BDRVQcowState {
     int qcow_version;
     bool use_lazy_refcounts;
 
+    bool discard_passthrough[QCOW2_DISCARD_MAX];
+
     uint64_t incompatible_features;
     uint64_t compatible_features;
     uint64_t autoclear_features;

index 3191d6b..cca76d4 100644
@@ -1377,18 +1377,25 @@ int qcow2_discard_clusters(BlockDriverState *bs, uint64_t offset,
 
     nb_clusters = size_to_clusters(s, end_offset - offset);
 
+    s->cache_discards = true;
+
     /* Each L2 table is handled by its own loop iteration */
     while (nb_clusters > 0) {
         ret = discard_single_l2(bs, offset, nb_clusters);
         if (ret < 0) {
-            return ret;
+            goto fail;
         }
 
         nb_clusters -= ret;
         offset += (ret * s->cluster_size);
     }
 
-    return 0;
+    ret = 0;
+fail:
+    s->cache_discards = false;
+    qcow2_process_discards(bs, ret);
+
+    return ret;
 }
 
 /*

@@ -1450,15 +1457,22 @@ int qcow2_zero_clusters(BlockDriverState *bs, uint64_t offset, int nb_sectors)
     /* Each L2 table is handled by its own loop iteration */
     nb_clusters = size_to_clusters(s, nb_sectors << BDRV_SECTOR_BITS);
 
+    s->cache_discards = true;
+
     while (nb_clusters > 0) {
         ret = zero_single_l2(bs, offset, nb_clusters);
         if (ret < 0) {
-            return ret;
+            goto fail;
         }
 
         nb_clusters -= ret;
         offset += (ret * s->cluster_size);
     }
 
-    return 0;
+    ret = 0;
+fail:
+    s->cache_discards = false;
+    qcow2_process_discards(bs, ret);
+
+    return ret;
 }

112 lines not on this page · raw message

index ef8a2ca..9383990 100644
@@ -486,6 +486,7 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags)
     }
 
     QLIST_INIT(&s->cluster_allocs);
+    QTAILQ_INIT(&s->discards);
 
     /* read qcow2 extensions */
     if (qcow2_read_extensions(bs, header.header_length, ext_end, NULL)) {

index 6f91b9a..3b2d5cd 100644
@@ -147,6 +147,13 @@ typedef struct Qcow2Feature {
     char    name[46];
 } QEMU_PACKED Qcow2Feature;
 
+typedef struct Qcow2DiscardRegion {
+    BlockDriverState *bs;
+    uint64_t offset;
+    uint64_t bytes;
+    QTAILQ_ENTRY(Qcow2DiscardRegion) next;
+} Qcow2DiscardRegion;
+
 typedef struct BDRVQcowState {
     int cluster_bits;
     int cluster_size;

@@ -199,6 +206,8 @@ typedef struct BDRVQcowState {
     size_t unknown_header_fields_size;
     void* unknown_header_fields;
     QLIST_HEAD(, Qcow2UnknownHeaderExtension) unknown_header_ext;
+    QTAILQ_HEAD (, Qcow2DiscardRegion) discards;
+    bool cache_discards;
 } BDRVQcowState;
 
 /* XXX: use std qcow open function ? */

@@ -374,6 +383,8 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs,
 int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
                           BdrvCheckMode fix);
 
+void qcow2_process_discards(BlockDriverState *bs, int ret);
+
 /* qcow2-cluster.c functions */
 int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
                         bool exact_size);

index b88ad2f..8e77d46 100644
@@ -1045,7 +1045,7 @@ int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
     extract_subqdict(options, &file_options, "file.");
 
     ret = bdrv_file_open(&file, filename, file_options,
-                         bdrv_open_flags(bs, flags));
+                         bdrv_open_flags(bs, flags | BDRV_O_UNMAP));
     if (ret < 0) {
         goto fail;
     }