Skip to contentSkip to conversation

qcow2: Drop synchronous qcow_write()

PATCH Superseded Kevin Wolf 2 participants 4 messages May 13, 2009 – May 18, 2009

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Kevin Wolf

author
#0 lore raw

PATCH 1/1 qcow2: Drop synchronous qcow_write()

There is only one (internal) user left and it can be switched to the normal emulation provided in block.c. Leaving it as it is would mean that creating compressed images uses a different codepath than everything else, so there would be a risk of a breakage that wouldn't be covered by much testing.

---
 block-qcow2.c |   42 +-----------------------------------------
 1 files changed, 1 insertions(+), 41 deletions(-)
block-qcow2.c +1 -41
index 9a49777..d674501 100644
@@ -1226,46 +1226,6 @@ static int qcow_read(BlockDriverState *bs, int64_t sector_num,
     return 0;
 }
 
-static int qcow_write(BlockDriverState *bs, int64_t sector_num,
-                     const uint8_t *buf, int nb_sectors)
-{
-    BDRVQcowState *s = bs->opaque;
-    int ret, index_in_cluster, n;
-    uint64_t cluster_offset;
-    int n_end;
-    QCowL2Meta l2meta;
-
-    while (nb_sectors > 0) {
-        index_in_cluster = sector_num & (s->cluster_sectors - 1);
-        n_end = index_in_cluster + nb_sectors;
-        if (s->crypt_method &&
-            n_end > QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors)
-            n_end = QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors;
-        cluster_offset = alloc_cluster_offset(bs, sector_num << 9,
-                                              index_in_cluster,
-                                              n_end, &n, &l2meta);
-        if (!cluster_offset)
-            return -1;
-        if (s->crypt_method) {
-            encrypt_sectors(s, sector_num, s->cluster_data, buf, n, 1,
-                            &s->aes_encrypt_key);
-            ret = bdrv_pwrite(s->hd, cluster_offset + index_in_cluster * 512,
-                              s->cluster_data, n * 512);
-        } else {
-            ret = bdrv_pwrite(s->hd, cluster_offset + index_in_cluster * 512, buf, n * 512);
-        }
-        if (ret != n * 512 || alloc_cluster_link_l2(bs, cluster_offset, &l2meta) < 0) {
-            free_any_clusters(bs, cluster_offset, l2meta.nb_clusters);
-            return -1;
-        }
-        nb_sectors -= n;
-        sector_num += n;
-        buf += n * 512;
-    }
-    s->cluster_cache_offset = -1; /* disable compressed cache */
-    return 0;
-}
-
 typedef struct QCowAIOCB {
     BlockDriverAIOCB common;
     int64_t sector_num;
@@ -1774,7 +1734,7 @@ static int qcow_write_compressed(BlockDriverState *bs, int64_t sector_num,
 
     if (ret != Z_STREAM_END || out_len >= s->cluster_size) {
         /* could not compress: write normal cluster */
-        qcow_write(bs, sector_num, buf, s->cluster_sectors);
+        bdrv_write(bs, sector_num, buf, s->cluster_sectors);
     } else {
         cluster_offset = alloc_compressed_cluster_offset(bs, sector_num << 9,
                                               out_len);
Signature · 1 line
1.6.0.6
#1 lore raw
Quoted 4 lines · Kevin Wolf

There is only one (internal) user left and it can be switched to the normal emulation provided in block.c. Leaving it as it is would mean that creating compressed images uses a different codepath than everything else, so there would be a risk of a breakage that wouldn't be covered by much testing.

Looks good to me, that strary qcow2_write irked me, too. Do you plan to kill qcow_read, too?

Also the same patch should into qcow, too.

author
#2 lore raw
Quoted 2 lines · Christoph Hellwig

Looks good to me, that strary qcow2_write irked me, too. Do you plan to kill qcow_read, too?

Hm, haven't planned that because qcow_read() is used in a much more common code path, but in theory the very same approach should work there (and someone seems to have done this already for qcow1?).

Don't know how the emulation behaves performance wise though, qcow_read() is used for each cluster allocation. I guess the difference is negligible?

Quoted 1 line

Also the same patch should into qcow, too.

Right, this looks the same.

Kevin

#3 lore raw
Quoted 11 lines · Kevin Wolf
Quoted 2 lines · Christoph Hellwig

Looks good to me, that strary qcow2_write irked me, too. Do you plan to kill qcow_read, too?

Hm, haven't planned that because qcow_read() is used in a much more common code path, but in theory the very same approach should work there (and someone seems to have done this already for qcow1?).

Don't know how the emulation behaves performance wise though, qcow_read() is used for each cluster allocation. I guess the difference is negligible?

Good question. Probably needs some benchmarking first.