diff options
author | David 'Digit' Turner <digit@android.com> | 2010-12-23 02:54:08 +0100 |
---|---|---|
committer | David 'Digit' Turner <digit@android.com> | 2011-01-02 22:31:27 +0100 |
commit | cb42a1b1461e02efb034582ac5d8f71534723b92 (patch) | |
tree | 5dbd619671d4591d33834488f1c15690dabf50ba /block/qcow2-refcount.c | |
parent | ca6a2e034bce665a08d9d748ac11d6a7cfcd7c48 (diff) | |
download | external_qemu-cb42a1b1461e02efb034582ac5d8f71534723b92.zip external_qemu-cb42a1b1461e02efb034582ac5d8f71534723b92.tar.gz external_qemu-cb42a1b1461e02efb034582ac5d8f71534723b92.tar.bz2 |
upstream: integrate block changes
This large patch upgrades the block support code to the upstream
version available in ba5e7f82169f32ab8163c707d97c799ca09f8924
dated 2010-08-08
Change-Id: I8b24df0c287e72f6620650a4d6a62e1bb315453e
Diffstat (limited to 'block/qcow2-refcount.c')
-rw-r--r-- | block/qcow2-refcount.c | 750 |
1 files changed, 549 insertions, 201 deletions
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c index dd6e293..4c19e7e 100644 --- a/block/qcow2-refcount.c +++ b/block/qcow2-refcount.c @@ -27,10 +27,32 @@ #include "block/qcow2.h" static int64_t alloc_clusters_noref(BlockDriverState *bs, int64_t size); -static int update_refcount(BlockDriverState *bs, +static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs, int64_t offset, int64_t length, int addend); + +static int cache_refcount_updates = 0; + +static int write_refcount_block(BlockDriverState *bs) +{ + BDRVQcowState *s = bs->opaque; + size_t size = s->cluster_size; + + if (s->refcount_block_cache_offset == 0) { + return 0; + } + + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_UPDATE); + if (bdrv_pwrite_sync(bs->file, s->refcount_block_cache_offset, + s->refcount_block_cache, size) < 0) + { + return -EIO; + } + + return 0; +} + /*********************************************************/ /* refcount handling */ @@ -43,7 +65,8 @@ int qcow2_refcount_init(BlockDriverState *bs) refcount_table_size2 = s->refcount_table_size * sizeof(uint64_t); s->refcount_table = qemu_malloc(refcount_table_size2); if (s->refcount_table_size > 0) { - ret = bdrv_pread(s->hd, s->refcount_table_offset, + BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_LOAD); + ret = bdrv_pread(bs->file, s->refcount_table_offset, s->refcount_table, refcount_table_size2); if (ret != refcount_table_size2) goto fail; @@ -68,19 +91,36 @@ static int load_refcount_block(BlockDriverState *bs, { BDRVQcowState *s = bs->opaque; int ret; - ret = bdrv_pread(s->hd, refcount_block_offset, s->refcount_block_cache, + + if (cache_refcount_updates) { + ret = write_refcount_block(bs); + if (ret < 0) { + return ret; + } + } + + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_LOAD); + ret = bdrv_pread(bs->file, refcount_block_offset, s->refcount_block_cache, s->cluster_size); - if (ret != s->cluster_size) - return -EIO; + if (ret < 0) { + return ret; + } + s->refcount_block_cache_offset = refcount_block_offset; return 0; } +/* + * Returns the refcount of the cluster given by its index. Any non-negative + * return value is the refcount of the cluster, negative values are -errno + * and indicate an error. + */ static int get_refcount(BlockDriverState *bs, int64_t cluster_index) { BDRVQcowState *s = bs->opaque; int refcount_table_index, block_index; int64_t refcount_block_offset; + int ret; refcount_table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT); if (refcount_table_index >= s->refcount_table_size) @@ -90,163 +130,351 @@ static int get_refcount(BlockDriverState *bs, int64_t cluster_index) return 0; if (refcount_block_offset != s->refcount_block_cache_offset) { /* better than nothing: return allocated if read error */ - if (load_refcount_block(bs, refcount_block_offset) < 0) - return 1; + ret = load_refcount_block(bs, refcount_block_offset); + if (ret < 0) { + return ret; + } } block_index = cluster_index & ((1 << (s->cluster_bits - REFCOUNT_SHIFT)) - 1); return be16_to_cpu(s->refcount_block_cache[block_index]); } -static int grow_refcount_table(BlockDriverState *bs, int min_size) +/* + * Rounds the refcount table size up to avoid growing the table for each single + * refcount block that is allocated. + */ +static unsigned int next_refcount_table_size(BDRVQcowState *s, + unsigned int min_size) { - BDRVQcowState *s = bs->opaque; - int new_table_size, new_table_size2, refcount_table_clusters, i, ret; - uint64_t *new_table; - int64_t table_offset; - uint8_t data[12]; - int old_table_size; - int64_t old_table_offset; + unsigned int min_clusters = (min_size >> (s->cluster_bits - 3)) + 1; + unsigned int refcount_table_clusters = + MAX(1, s->refcount_table_size >> (s->cluster_bits - 3)); - if (min_size <= s->refcount_table_size) - return 0; - /* compute new table size */ - refcount_table_clusters = s->refcount_table_size >> (s->cluster_bits - 3); - for(;;) { - if (refcount_table_clusters == 0) { - refcount_table_clusters = 1; - } else { - refcount_table_clusters = (refcount_table_clusters * 3 + 1) / 2; - } - new_table_size = refcount_table_clusters << (s->cluster_bits - 3); - if (min_size <= new_table_size) - break; + while (min_clusters > refcount_table_clusters) { + refcount_table_clusters = (refcount_table_clusters * 3 + 1) / 2; } -#ifdef DEBUG_ALLOC2 - printf("grow_refcount_table from %d to %d\n", - s->refcount_table_size, - new_table_size); -#endif - new_table_size2 = new_table_size * sizeof(uint64_t); - new_table = qemu_mallocz(new_table_size2); - memcpy(new_table, s->refcount_table, - s->refcount_table_size * sizeof(uint64_t)); - for(i = 0; i < s->refcount_table_size; i++) - cpu_to_be64s(&new_table[i]); - /* Note: we cannot update the refcount now to avoid recursion */ - table_offset = alloc_clusters_noref(bs, new_table_size2); - ret = bdrv_pwrite(s->hd, table_offset, new_table, new_table_size2); - if (ret != new_table_size2) - goto fail; - for(i = 0; i < s->refcount_table_size; i++) - be64_to_cpus(&new_table[i]); - - cpu_to_be64w((uint64_t*)data, table_offset); - cpu_to_be32w((uint32_t*)(data + 8), refcount_table_clusters); - if (bdrv_pwrite(s->hd, offsetof(QCowHeader, refcount_table_offset), - data, sizeof(data)) != sizeof(data)) - goto fail; - qemu_free(s->refcount_table); - old_table_offset = s->refcount_table_offset; - old_table_size = s->refcount_table_size; - s->refcount_table = new_table; - s->refcount_table_size = new_table_size; - s->refcount_table_offset = table_offset; - update_refcount(bs, table_offset, new_table_size2, 1); - qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t)); - return 0; - fail: - qcow2_free_clusters(bs, table_offset, new_table_size2); - qemu_free(new_table); - return -EIO; + return refcount_table_clusters << (s->cluster_bits - 3); } +/* Checks if two offsets are described by the same refcount block */ +static int in_same_refcount_block(BDRVQcowState *s, uint64_t offset_a, + uint64_t offset_b) +{ + uint64_t block_a = offset_a >> (2 * s->cluster_bits - REFCOUNT_SHIFT); + uint64_t block_b = offset_b >> (2 * s->cluster_bits - REFCOUNT_SHIFT); + + return (block_a == block_b); +} + +/* + * Loads a refcount block. If it doesn't exist yet, it is allocated first + * (including growing the refcount table if needed). + * + * Returns the offset of the refcount block on success or -errno in error case + */ static int64_t alloc_refcount_block(BlockDriverState *bs, int64_t cluster_index) { BDRVQcowState *s = bs->opaque; - int64_t offset, refcount_block_offset; - int ret, refcount_table_index; - uint64_t data64; + unsigned int refcount_table_index; + int ret; - /* Find L1 index and grow refcount table if needed */ + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC); + + /* Find the refcount block for the given cluster */ refcount_table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT); - if (refcount_table_index >= s->refcount_table_size) { - ret = grow_refcount_table(bs, refcount_table_index + 1); - if (ret < 0) + + if (refcount_table_index < s->refcount_table_size) { + + uint64_t refcount_block_offset = + s->refcount_table[refcount_table_index]; + + /* If it's already there, we're done */ + if (refcount_block_offset) { + if (refcount_block_offset != s->refcount_block_cache_offset) { + ret = load_refcount_block(bs, refcount_block_offset); + if (ret < 0) { + return ret; + } + } + return refcount_block_offset; + } + } + + /* + * If we came here, we need to allocate something. Something is at least + * a cluster for the new refcount block. It may also include a new refcount + * table if the old refcount table is too small. + * + * Note that allocating clusters here needs some special care: + * + * - We can't use the normal qcow2_alloc_clusters(), it would try to + * increase the refcount and very likely we would end up with an endless + * recursion. Instead we must place the refcount blocks in a way that + * they can describe them themselves. + * + * - We need to consider that at this point we are inside update_refcounts + * and doing the initial refcount increase. This means that some clusters + * have already been allocated by the caller, but their refcount isn't + * accurate yet. free_cluster_index tells us where this allocation ends + * as long as we don't overwrite it by freeing clusters. + * + * - alloc_clusters_noref and qcow2_free_clusters may load a different + * refcount block into the cache + */ + + if (cache_refcount_updates) { + ret = write_refcount_block(bs); + if (ret < 0) { return ret; + } } - /* Load or allocate the refcount block */ - refcount_block_offset = s->refcount_table[refcount_table_index]; - if (!refcount_block_offset) { - /* create a new refcount block */ - /* Note: we cannot update the refcount now to avoid recursion */ - offset = alloc_clusters_noref(bs, s->cluster_size); + /* Allocate the refcount block itself and mark it as used */ + int64_t new_block = alloc_clusters_noref(bs, s->cluster_size); + if (new_block < 0) { + return new_block; + } + +#ifdef DEBUG_ALLOC2 + fprintf(stderr, "qcow2: Allocate refcount block %d for %" PRIx64 + " at %" PRIx64 "\n", + refcount_table_index, cluster_index << s->cluster_bits, new_block); +#endif + + if (in_same_refcount_block(s, new_block, cluster_index << s->cluster_bits)) { + /* Zero the new refcount block before updating it */ memset(s->refcount_block_cache, 0, s->cluster_size); - ret = bdrv_pwrite(s->hd, offset, s->refcount_block_cache, s->cluster_size); - if (ret != s->cluster_size) - return -EINVAL; - s->refcount_table[refcount_table_index] = offset; - data64 = cpu_to_be64(offset); - ret = bdrv_pwrite(s->hd, s->refcount_table_offset + - refcount_table_index * sizeof(uint64_t), - &data64, sizeof(data64)); - if (ret != sizeof(data64)) - return -EINVAL; - - refcount_block_offset = offset; - s->refcount_block_cache_offset = offset; - update_refcount(bs, offset, s->cluster_size, 1); + s->refcount_block_cache_offset = new_block; + + /* The block describes itself, need to update the cache */ + int block_index = (new_block >> s->cluster_bits) & + ((1 << (s->cluster_bits - REFCOUNT_SHIFT)) - 1); + s->refcount_block_cache[block_index] = cpu_to_be16(1); } else { - if (refcount_block_offset != s->refcount_block_cache_offset) { - if (load_refcount_block(bs, refcount_block_offset) < 0) - return -EIO; + /* Described somewhere else. This can recurse at most twice before we + * arrive at a block that describes itself. */ + ret = update_refcount(bs, new_block, s->cluster_size, 1); + if (ret < 0) { + goto fail_block; + } + + /* Initialize the new refcount block only after updating its refcount, + * update_refcount uses the refcount cache itself */ + memset(s->refcount_block_cache, 0, s->cluster_size); + s->refcount_block_cache_offset = new_block; + } + + /* Now the new refcount block needs to be written to disk */ + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE); + ret = bdrv_pwrite_sync(bs->file, new_block, s->refcount_block_cache, + s->cluster_size); + if (ret < 0) { + goto fail_block; + } + + /* If the refcount table is big enough, just hook the block up there */ + if (refcount_table_index < s->refcount_table_size) { + uint64_t data64 = cpu_to_be64(new_block); + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_HOOKUP); + ret = bdrv_pwrite_sync(bs->file, + s->refcount_table_offset + refcount_table_index * sizeof(uint64_t), + &data64, sizeof(data64)); + if (ret < 0) { + goto fail_block; } + + s->refcount_table[refcount_table_index] = new_block; + return new_block; + } + + /* + * If we come here, we need to grow the refcount table. Again, a new + * refcount table needs some space and we can't simply allocate to avoid + * endless recursion. + * + * Therefore let's grab new refcount blocks at the end of the image, which + * will describe themselves and the new refcount table. This way we can + * reference them only in the new table and do the switch to the new + * refcount table at once without producing an inconsistent state in + * between. + */ + BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_GROW); + + /* Calculate the number of refcount blocks needed so far */ + uint64_t refcount_block_clusters = 1 << (s->cluster_bits - REFCOUNT_SHIFT); + uint64_t blocks_used = (s->free_cluster_index + + refcount_block_clusters - 1) / refcount_block_clusters; + + /* And now we need at least one block more for the new metadata */ + uint64_t table_size = next_refcount_table_size(s, blocks_used + 1); + uint64_t last_table_size; + uint64_t blocks_clusters; + do { + uint64_t table_clusters = size_to_clusters(s, table_size); + blocks_clusters = 1 + + ((table_clusters + refcount_block_clusters - 1) + / refcount_block_clusters); + uint64_t meta_clusters = table_clusters + blocks_clusters; + + last_table_size = table_size; + table_size = next_refcount_table_size(s, blocks_used + + ((meta_clusters + refcount_block_clusters - 1) + / refcount_block_clusters)); + + } while (last_table_size != table_size); + +#ifdef DEBUG_ALLOC2 + fprintf(stderr, "qcow2: Grow refcount table %" PRId32 " => %" PRId64 "\n", + s->refcount_table_size, table_size); +#endif + + /* Create the new refcount table and blocks */ + uint64_t meta_offset = (blocks_used * refcount_block_clusters) * + s->cluster_size; + uint64_t table_offset = meta_offset + blocks_clusters * s->cluster_size; + uint16_t *new_blocks = qemu_mallocz(blocks_clusters * s->cluster_size); + uint64_t *new_table = qemu_mallocz(table_size * sizeof(uint64_t)); + + assert(meta_offset >= (s->free_cluster_index * s->cluster_size)); + + /* Fill the new refcount table */ + memcpy(new_table, s->refcount_table, + s->refcount_table_size * sizeof(uint64_t)); + new_table[refcount_table_index] = new_block; + + int i; + for (i = 0; i < blocks_clusters; i++) { + new_table[blocks_used + i] = meta_offset + (i * s->cluster_size); + } + + /* Fill the refcount blocks */ + uint64_t table_clusters = size_to_clusters(s, table_size * sizeof(uint64_t)); + int block = 0; + for (i = 0; i < table_clusters + blocks_clusters; i++) { + new_blocks[block++] = cpu_to_be16(1); + } + + /* Write refcount blocks to disk */ + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_BLOCKS); + ret = bdrv_pwrite_sync(bs->file, meta_offset, new_blocks, + blocks_clusters * s->cluster_size); + qemu_free(new_blocks); + if (ret < 0) { + goto fail_table; + } + + /* Write refcount table to disk */ + for(i = 0; i < table_size; i++) { + cpu_to_be64s(&new_table[i]); + } + + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_TABLE); + ret = bdrv_pwrite_sync(bs->file, table_offset, new_table, + table_size * sizeof(uint64_t)); + if (ret < 0) { + goto fail_table; + } + + for(i = 0; i < table_size; i++) { + cpu_to_be64s(&new_table[i]); + } + + /* Hook up the new refcount table in the qcow2 header */ + uint8_t data[12]; + cpu_to_be64w((uint64_t*)data, table_offset); + cpu_to_be32w((uint32_t*)(data + 8), table_clusters); + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_SWITCH_TABLE); + ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, refcount_table_offset), + data, sizeof(data)); + if (ret < 0) { + goto fail_table; } - return refcount_block_offset; + /* And switch it in memory */ + uint64_t old_table_offset = s->refcount_table_offset; + uint64_t old_table_size = s->refcount_table_size; + + qemu_free(s->refcount_table); + s->refcount_table = new_table; + s->refcount_table_size = table_size; + s->refcount_table_offset = table_offset; + + /* Free old table. Remember, we must not change free_cluster_index */ + uint64_t old_free_cluster_index = s->free_cluster_index; + qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t)); + s->free_cluster_index = old_free_cluster_index; + + ret = load_refcount_block(bs, new_block); + if (ret < 0) { + goto fail_block; + } + + return new_block; + +fail_table: + qemu_free(new_table); +fail_block: + s->refcount_block_cache_offset = 0; + return ret; } #define REFCOUNTS_PER_SECTOR (512 >> REFCOUNT_SHIFT) -static int write_refcount_block_entries(BDRVQcowState *s, +static int write_refcount_block_entries(BlockDriverState *bs, int64_t refcount_block_offset, int first_index, int last_index) { + BDRVQcowState *s = bs->opaque; size_t size; + int ret; + + if (cache_refcount_updates) { + return 0; + } + + if (first_index < 0) { + return 0; + } first_index &= ~(REFCOUNTS_PER_SECTOR - 1); last_index = (last_index + REFCOUNTS_PER_SECTOR) & ~(REFCOUNTS_PER_SECTOR - 1); size = (last_index - first_index) << REFCOUNT_SHIFT; - if (bdrv_pwrite(s->hd, + + BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_UPDATE_PART); + ret = bdrv_pwrite_sync(bs->file, refcount_block_offset + (first_index << REFCOUNT_SHIFT), - &s->refcount_block_cache[first_index], size) != size) - { - return -EIO; + &s->refcount_block_cache[first_index], size); + if (ret < 0) { + return ret; } return 0; } /* XXX: cache several refcount block clusters ? */ -static int update_refcount(BlockDriverState *bs, - int64_t offset, int64_t length, - int addend) +static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs, + int64_t offset, int64_t length, int addend) { BDRVQcowState *s = bs->opaque; int64_t start, last, cluster_offset; int64_t refcount_block_offset = 0; int64_t table_index = -1, old_table_index; int first_index = -1, last_index = -1; + int ret; #ifdef DEBUG_ALLOC2 - printf("update_refcount: offset=%lld size=%lld addend=%d\n", + printf("update_refcount: offset=%" PRId64 " size=%" PRId64 " addend=%d\n", offset, length, addend); #endif - if (length <= 0) + if (length < 0) { return -EINVAL; + } else if (length == 0) { + return 0; + } + start = offset & ~(s->cluster_size - 1); last = (offset + length - 1) & ~(s->cluster_size - 1); for(cluster_offset = start; cluster_offset <= last; @@ -254,16 +482,17 @@ static int update_refcount(BlockDriverState *bs, { int block_index, refcount; int64_t cluster_index = cluster_offset >> s->cluster_bits; + int64_t new_block; /* Only write refcount block to disk when we are done with it */ old_table_index = table_index; table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT); if ((old_table_index >= 0) && (table_index != old_table_index)) { - if (write_refcount_block_entries(s, refcount_block_offset, - first_index, last_index) < 0) - { - return -EIO; + ret = write_refcount_block_entries(bs, refcount_block_offset, + first_index, last_index); + if (ret < 0) { + return ret; } first_index = -1; @@ -271,10 +500,12 @@ static int update_refcount(BlockDriverState *bs, } /* Load the refcount block and allocate it if needed */ - refcount_block_offset = alloc_refcount_block(bs, cluster_index); - if (refcount_block_offset < 0) { - return refcount_block_offset; + new_block = alloc_refcount_block(bs, cluster_index); + if (new_block < 0) { + ret = new_block; + goto fail; } + refcount_block_offset = new_block; /* we can update the count and save it */ block_index = cluster_index & @@ -288,27 +519,48 @@ static int update_refcount(BlockDriverState *bs, refcount = be16_to_cpu(s->refcount_block_cache[block_index]); refcount += addend; - if (refcount < 0 || refcount > 0xffff) - return -EINVAL; + if (refcount < 0 || refcount > 0xffff) { + ret = -EINVAL; + goto fail; + } if (refcount == 0 && cluster_index < s->free_cluster_index) { s->free_cluster_index = cluster_index; } s->refcount_block_cache[block_index] = cpu_to_be16(refcount); } + ret = 0; +fail: + /* Write last changed block to disk */ if (refcount_block_offset != 0) { - if (write_refcount_block_entries(s, refcount_block_offset, - first_index, last_index) < 0) - { - return -EIO; + int wret; + wret = write_refcount_block_entries(bs, refcount_block_offset, + first_index, last_index); + if (wret < 0) { + return ret < 0 ? ret : wret; } } - return 0; + /* + * Try do undo any updates if an error is returned (This may succeed in + * some cases like ENOSPC for allocating a new refcount block) + */ + if (ret < 0) { + int dummy; + dummy = update_refcount(bs, offset, cluster_offset - offset, -addend); + } + + return ret; } -/* addend must be 1 or -1 */ +/* + * Increases or decreases the refcount of a given cluster by one. + * addend must be 1 or -1. + * + * If the return value is non-negative, it is the new refcount of the cluster. + * If it is negative, it is -errno and indicates an error. + */ static int update_cluster_refcount(BlockDriverState *bs, int64_t cluster_index, int addend) @@ -335,17 +587,22 @@ static int update_cluster_refcount(BlockDriverState *bs, static int64_t alloc_clusters_noref(BlockDriverState *bs, int64_t size) { BDRVQcowState *s = bs->opaque; - int i, nb_clusters; + int i, nb_clusters, refcount; nb_clusters = size_to_clusters(s, size); retry: for(i = 0; i < nb_clusters; i++) { - int64_t i = s->free_cluster_index++; - if (get_refcount(bs, i) != 0) + int64_t next_cluster_index = s->free_cluster_index++; + refcount = get_refcount(bs, next_cluster_index); + + if (refcount < 0) { + return refcount; + } else if (refcount != 0) { goto retry; + } } #ifdef DEBUG_ALLOC2 - printf("alloc_clusters: size=%lld -> %lld\n", + printf("alloc_clusters: size=%" PRId64 " -> %" PRId64 "\n", size, (s->free_cluster_index - nb_clusters) << s->cluster_bits); #endif @@ -355,9 +612,18 @@ retry: int64_t qcow2_alloc_clusters(BlockDriverState *bs, int64_t size) { int64_t offset; + int ret; + BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC); offset = alloc_clusters_noref(bs, size); - update_refcount(bs, offset, size, 1); + if (offset < 0) { + return offset; + } + + ret = update_refcount(bs, offset, size, 1); + if (ret < 0) { + return ret; + } return offset; } @@ -369,9 +635,13 @@ int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size) int64_t offset, cluster_offset; int free_in_cluster; + BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC_BYTES); assert(size > 0 && size <= s->cluster_size); if (s->free_byte_offset == 0) { s->free_byte_offset = qcow2_alloc_clusters(bs, s->cluster_size); + if (s->free_byte_offset < 0) { + return s->free_byte_offset; + } } redo: free_in_cluster = s->cluster_size - @@ -387,6 +657,9 @@ int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size) update_cluster_refcount(bs, offset >> s->cluster_bits, 1); } else { offset = qcow2_alloc_clusters(bs, s->cluster_size); + if (offset < 0) { + return offset; + } cluster_offset = s->free_byte_offset & ~(s->cluster_size - 1); if ((cluster_offset + s->cluster_size) == offset) { /* we are lucky: contiguous data */ @@ -404,7 +677,14 @@ int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size) void qcow2_free_clusters(BlockDriverState *bs, int64_t offset, int64_t size) { - update_refcount(bs, offset, size, -1); + int ret; + + BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_FREE); + ret = update_refcount(bs, offset, size, -1); + if (ret < 0) { + fprintf(stderr, "qcow2_free_clusters failed: %s\n", strerror(-ret)); + /* TODO Remember the clusters to free them later and avoid leaking */ + } } /* @@ -471,15 +751,19 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, int l2_size, i, j, l1_modified, l2_modified, nb_csectors, refcount; qcow2_l2_cache_reset(bs); + cache_refcount_updates = 1; l2_table = NULL; l1_table = NULL; l1_size2 = l1_size * sizeof(uint64_t); - l1_allocated = 0; if (l1_table_offset != s->l1_table_offset) { - l1_table = qemu_malloc(l1_size2); + if (l1_size2 != 0) { + l1_table = qemu_mallocz(align_offset(l1_size2, 512)); + } else { + l1_table = NULL; + } l1_allocated = 1; - if (bdrv_pread(s->hd, l1_table_offset, + if (bdrv_pread(bs->file, l1_table_offset, l1_table, l1_size2) != l1_size2) goto fail; for(i = 0;i < l1_size; i++) @@ -499,7 +783,7 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, old_l2_offset = l2_offset; l2_offset &= ~QCOW_OFLAG_COPIED; l2_modified = 0; - if (bdrv_pread(s->hd, l2_offset, l2_table, l2_size) != l2_size) + if (bdrv_pread(bs->file, l2_offset, l2_table, l2_size) != l2_size) goto fail; for(j = 0; j < s->l2_size; j++) { offset = be64_to_cpu(l2_table[j]); @@ -509,9 +793,15 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, if (offset & QCOW_OFLAG_COMPRESSED) { nb_csectors = ((offset >> s->csize_shift) & s->csize_mask) + 1; - if (addend != 0) - update_refcount(bs, (offset & s->cluster_offset_mask) & ~511, - nb_csectors * 512, addend); + if (addend != 0) { + int ret; + ret = update_refcount(bs, + (offset & s->cluster_offset_mask) & ~511, + nb_csectors * 512, addend); + if (ret < 0) { + goto fail; + } + } /* compressed clusters are never modified */ refcount = 2; } else { @@ -520,6 +810,10 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, } else { refcount = get_refcount(bs, offset >> s->cluster_bits); } + + if (refcount < 0) { + goto fail; + } } if (refcount == 1) { @@ -532,8 +826,8 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, } } if (l2_modified) { - if (bdrv_pwrite(s->hd, - l2_offset, l2_table, l2_size) != l2_size) + if (bdrv_pwrite_sync(bs->file, + l2_offset, l2_table, l2_size) < 0) goto fail; } @@ -542,7 +836,9 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, } else { refcount = get_refcount(bs, l2_offset >> s->cluster_bits); } - if (refcount == 1) { + if (refcount < 0) { + goto fail; + } else if (refcount == 1) { l2_offset |= QCOW_OFLAG_COPIED; } if (l2_offset != old_l2_offset) { @@ -554,8 +850,8 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, if (l1_modified) { for(i = 0; i < l1_size; i++) cpu_to_be64s(&l1_table[i]); - if (bdrv_pwrite(s->hd, l1_table_offset, l1_table, - l1_size2) != l1_size2) + if (bdrv_pwrite_sync(bs->file, l1_table_offset, l1_table, + l1_size2) < 0) goto fail; for(i = 0; i < l1_size; i++) be64_to_cpus(&l1_table[i]); @@ -563,11 +859,15 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, if (l1_allocated) qemu_free(l1_table); qemu_free(l2_table); + cache_refcount_updates = 0; + write_refcount_block(bs); return 0; fail: if (l1_allocated) qemu_free(l1_table); qemu_free(l2_table); + cache_refcount_updates = 0; + write_refcount_block(bs); return -EIO; } @@ -584,9 +884,10 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs, * This is used to construct a temporary refcount table out of L1 and L2 tables * which can be compared the the refcount table saved in the image. * - * Returns the number of errors in the image that were found + * Modifies the number of errors in res. */ -static int inc_refcounts(BlockDriverState *bs, +static void inc_refcounts(BlockDriverState *bs, + BdrvCheckResult *res, uint16_t *refcount_table, int refcount_table_size, int64_t offset, int64_t size) @@ -594,30 +895,32 @@ static int inc_refcounts(BlockDriverState *bs, BDRVQcowState *s = bs->opaque; int64_t start, last, cluster_offset; int k; - int errors = 0; if (size <= 0) - return 0; + return; start = offset & ~(s->cluster_size - 1); last = (offset + size - 1) & ~(s->cluster_size - 1); for(cluster_offset = start; cluster_offset <= last; cluster_offset += s->cluster_size) { k = cluster_offset >> s->cluster_bits; - if (k < 0 || k >= refcount_table_size) { + if (k < 0) { fprintf(stderr, "ERROR: invalid cluster offset=0x%" PRIx64 "\n", cluster_offset); - errors++; + res->corruptions++; + } else if (k >= refcount_table_size) { + fprintf(stderr, "Warning: cluster offset=0x%" PRIx64 " is after " + "the end of the image file, can't properly check refcounts.\n", + cluster_offset); + res->check_errors++; } else { if (++refcount_table[k] == 0) { fprintf(stderr, "ERROR: overflow cluster offset=0x%" PRIx64 "\n", cluster_offset); - errors++; + res->corruptions++; } } } - - return errors; } /* @@ -628,20 +931,19 @@ static int inc_refcounts(BlockDriverState *bs, * Returns the number of errors found by the checks or -errno if an internal * error occurred. */ -static int check_refcounts_l2(BlockDriverState *bs, +static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res, uint16_t *refcount_table, int refcount_table_size, int64_t l2_offset, int check_copied) { BDRVQcowState *s = bs->opaque; uint64_t *l2_table, offset; int i, l2_size, nb_csectors, refcount; - int errors = 0; /* Read L2 table from disk */ l2_size = s->l2_size * sizeof(uint64_t); l2_table = qemu_malloc(l2_size); - if (bdrv_pread(s->hd, l2_offset, l2_table, l2_size) != l2_size) + if (bdrv_pread(bs->file, l2_offset, l2_table, l2_size) != l2_size) goto fail; /* Do the actual checks */ @@ -655,50 +957,53 @@ static int check_refcounts_l2(BlockDriverState *bs, "copied flag must never be set for compressed " "clusters\n", offset >> s->cluster_bits); offset &= ~QCOW_OFLAG_COPIED; - errors++; + res->corruptions++; } /* Mark cluster as used */ nb_csectors = ((offset >> s->csize_shift) & s->csize_mask) + 1; offset &= s->cluster_offset_mask; - errors += inc_refcounts(bs, refcount_table, - refcount_table_size, - offset & ~511, nb_csectors * 512); + inc_refcounts(bs, res, refcount_table, refcount_table_size, + offset & ~511, nb_csectors * 512); } else { /* QCOW_OFLAG_COPIED must be set iff refcount == 1 */ if (check_copied) { uint64_t entry = offset; offset &= ~QCOW_OFLAG_COPIED; refcount = get_refcount(bs, offset >> s->cluster_bits); + if (refcount < 0) { + fprintf(stderr, "Can't get refcount for offset %" + PRIx64 ": %s\n", entry, strerror(-refcount)); + goto fail; + } if ((refcount == 1) != ((entry & QCOW_OFLAG_COPIED) != 0)) { fprintf(stderr, "ERROR OFLAG_COPIED: offset=%" PRIx64 " refcount=%d\n", entry, refcount); - errors++; + res->corruptions++; } } /* Mark cluster as used */ offset &= ~QCOW_OFLAG_COPIED; - errors += inc_refcounts(bs, refcount_table, - refcount_table_size, - offset, s->cluster_size); + inc_refcounts(bs, res, refcount_table,refcount_table_size, + offset, s->cluster_size); /* Correct offsets are cluster aligned */ if (offset & (s->cluster_size - 1)) { fprintf(stderr, "ERROR offset=%" PRIx64 ": Cluster is not " "properly aligned; L2 entry corrupted.\n", offset); - errors++; + res->corruptions++; } } } } qemu_free(l2_table); - return errors; + return 0; fail: - fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n"); + fprintf(stderr, "ERROR: I/O error in check_refcounts_l2\n"); qemu_free(l2_table); return -EIO; } @@ -712,6 +1017,7 @@ fail: * error occurred. */ static int check_refcounts_l1(BlockDriverState *bs, + BdrvCheckResult *res, uint16_t *refcount_table, int refcount_table_size, int64_t l1_table_offset, int l1_size, @@ -720,21 +1026,24 @@ static int check_refcounts_l1(BlockDriverState *bs, BDRVQcowState *s = bs->opaque; uint64_t *l1_table, l2_offset, l1_size2; int i, refcount, ret; - int errors = 0; l1_size2 = l1_size * sizeof(uint64_t); /* Mark L1 table as used */ - errors += inc_refcounts(bs, refcount_table, refcount_table_size, - l1_table_offset, l1_size2); + inc_refcounts(bs, res, refcount_table, refcount_table_size, + l1_table_offset, l1_size2); /* Read L1 table entries from disk */ - l1_table = qemu_malloc(l1_size2); - if (bdrv_pread(s->hd, l1_table_offset, - l1_table, l1_size2) != l1_size2) - goto fail; - for(i = 0;i < l1_size; i++) - be64_to_cpus(&l1_table[i]); + if (l1_size2 == 0) { + l1_table = NULL; + } else { + l1_table = qemu_malloc(l1_size2); + if (bdrv_pread(bs->file, l1_table_offset, + l1_table, l1_size2) != l1_size2) + goto fail; + for(i = 0;i < l1_size; i++) + be64_to_cpus(&l1_table[i]); + } /* Do the actual checks */ for(i = 0; i < l1_size; i++) { @@ -744,41 +1053,44 @@ static int check_refcounts_l1(BlockDriverState *bs, if (check_copied) { refcount = get_refcount(bs, (l2_offset & ~QCOW_OFLAG_COPIED) >> s->cluster_bits); + if (refcount < 0) { + fprintf(stderr, "Can't get refcount for l2_offset %" + PRIx64 ": %s\n", l2_offset, strerror(-refcount)); + goto fail; + } if ((refcount == 1) != ((l2_offset & QCOW_OFLAG_COPIED) != 0)) { fprintf(stderr, "ERROR OFLAG_COPIED: l2_offset=%" PRIx64 " refcount=%d\n", l2_offset, refcount); - errors++; + res->corruptions++; } } /* Mark L2 table as used */ l2_offset &= ~QCOW_OFLAG_COPIED; - errors += inc_refcounts(bs, refcount_table, - refcount_table_size, - l2_offset, - s->cluster_size); + inc_refcounts(bs, res, refcount_table, refcount_table_size, + l2_offset, s->cluster_size); /* L2 tables are cluster aligned */ if (l2_offset & (s->cluster_size - 1)) { fprintf(stderr, "ERROR l2_offset=%" PRIx64 ": Table is not " "cluster aligned; L1 entry corrupted\n", l2_offset); - errors++; + res->corruptions++; } /* Process and check L2 entries */ - ret = check_refcounts_l2(bs, refcount_table, refcount_table_size, - l2_offset, check_copied); + ret = check_refcounts_l2(bs, res, refcount_table, + refcount_table_size, l2_offset, check_copied); if (ret < 0) { goto fail; } - errors += ret; } } qemu_free(l1_table); - return errors; + return 0; fail: fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n"); + res->check_errors++; qemu_free(l1_table); return -EIO; } @@ -789,66 +1101,102 @@ fail: * Returns 0 if no errors are found, the number of errors in case the image is * detected as corrupted, and -errno when an internal error occured. */ -int qcow2_check_refcounts(BlockDriverState *bs) +int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res) { BDRVQcowState *s = bs->opaque; int64_t size; int nb_clusters, refcount1, refcount2, i; QCowSnapshot *sn; uint16_t *refcount_table; - int ret, errors = 0; + int ret; - size = bdrv_getlength(s->hd); + size = bdrv_getlength(bs->file); nb_clusters = size_to_clusters(s, size); refcount_table = qemu_mallocz(nb_clusters * sizeof(uint16_t)); /* header */ - errors += inc_refcounts(bs, refcount_table, nb_clusters, - 0, s->cluster_size); + inc_refcounts(bs, res, refcount_table, nb_clusters, + 0, s->cluster_size); /* current L1 table */ - ret = check_refcounts_l1(bs, refcount_table, nb_clusters, + ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters, s->l1_table_offset, s->l1_size, 1); if (ret < 0) { return ret; } - errors += ret; /* snapshots */ for(i = 0; i < s->nb_snapshots; i++) { sn = s->snapshots + i; - check_refcounts_l1(bs, refcount_table, nb_clusters, - sn->l1_table_offset, sn->l1_size, 0); + ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters, + sn->l1_table_offset, sn->l1_size, 0); + if (ret < 0) { + return ret; + } } - errors += inc_refcounts(bs, refcount_table, nb_clusters, - s->snapshots_offset, s->snapshots_size); + inc_refcounts(bs, res, refcount_table, nb_clusters, + s->snapshots_offset, s->snapshots_size); /* refcount data */ - errors += inc_refcounts(bs, refcount_table, nb_clusters, - s->refcount_table_offset, - s->refcount_table_size * sizeof(uint64_t)); + inc_refcounts(bs, res, refcount_table, nb_clusters, + s->refcount_table_offset, + s->refcount_table_size * sizeof(uint64_t)); + for(i = 0; i < s->refcount_table_size; i++) { - int64_t offset; + uint64_t offset, cluster; offset = s->refcount_table[i]; + cluster = offset >> s->cluster_bits; + + /* Refcount blocks are cluster aligned */ + if (offset & (s->cluster_size - 1)) { + fprintf(stderr, "ERROR refcount block %d is not " + "cluster aligned; refcount table entry corrupted\n", i); + res->corruptions++; + continue; + } + + if (cluster >= nb_clusters) { + fprintf(stderr, "ERROR refcount block %d is outside image\n", i); + res->corruptions++; + continue; + } + if (offset != 0) { - errors += inc_refcounts(bs, refcount_table, nb_clusters, - offset, s->cluster_size); + inc_refcounts(bs, res, refcount_table, nb_clusters, + offset, s->cluster_size); + if (refcount_table[cluster] != 1) { + fprintf(stderr, "ERROR refcount block %d refcount=%d\n", + i, refcount_table[cluster]); + res->corruptions++; + } } } /* compare ref counts */ for(i = 0; i < nb_clusters; i++) { refcount1 = get_refcount(bs, i); + if (refcount1 < 0) { + fprintf(stderr, "Can't get refcount for cluster %d: %s\n", + i, strerror(-refcount1)); + res->check_errors++; + continue; + } + refcount2 = refcount_table[i]; if (refcount1 != refcount2) { - fprintf(stderr, "ERROR cluster %d refcount=%d reference=%d\n", + fprintf(stderr, "%s cluster %d refcount=%d reference=%d\n", + refcount1 < refcount2 ? "ERROR" : "Leaked", i, refcount1, refcount2); - errors++; + if (refcount1 < refcount2) { + res->corruptions++; + } else { + res->leaks++; + } } } qemu_free(refcount_table); - return errors; + return 0; } |