erofs: scan devices from device table
authorJeffle Xu <jefflexu@linux.alibaba.com>
Thu, 12 May 2022 05:56:01 +0000 (13:56 +0800)
committerGao Xiang <hsiangkao@linux.alibaba.com>
Tue, 17 May 2022 16:11:21 +0000 (00:11 +0800)
When "-o device" mount option is not specified, scan the device table
and instantiate the devices if there's any in the device table. In this
case, the tag field of each device slot uniquely specifies a device.

Signed-off-by: Jeffle Xu <jefflexu@linux.alibaba.com>
Reviewed-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Link: https://lore.kernel.org/r/20220512055601.106109-1-jefflexu@linux.alibaba.com
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
fs/erofs/erofs_fs.h
fs/erofs/super.c

index 20fc43c..2b48373 100644 (file)
 #define EROFS_SB_EXTSLOT_SIZE  16
 
 struct erofs_deviceslot {
-       union {
-               u8 uuid[16];            /* used for device manager later */
-               u8 userdata[64];        /* digest(sha256), etc. */
-       } u;
-       __le32 blocks;                  /* total fs blocks of this device */
-       __le32 mapped_blkaddr;          /* map starting at mapped_blkaddr */
+       u8 tag[64];             /* digest(sha256), etc. */
+       __le32 blocks;          /* total fs blocks of this device */
+       __le32 mapped_blkaddr;  /* map starting at mapped_blkaddr */
        u8 reserved[56];
 };
 #define EROFS_DEVT_SLOT_SIZE   sizeof(struct erofs_deviceslot)
index d0d33fd..c6f5fa4 100644 (file)
@@ -220,7 +220,52 @@ static int erofs_load_compr_cfgs(struct super_block *sb,
 }
 #endif
 
-static int erofs_init_devices(struct super_block *sb,
+static int erofs_init_device(struct erofs_buf *buf, struct super_block *sb,
+                            struct erofs_device_info *dif, erofs_off_t *pos)
+{
+       struct erofs_sb_info *sbi = EROFS_SB(sb);
+       struct erofs_deviceslot *dis;
+       struct block_device *bdev;
+       void *ptr;
+       int ret;
+
+       ptr = erofs_read_metabuf(buf, sb, erofs_blknr(*pos), EROFS_KMAP);
+       if (IS_ERR(ptr))
+               return PTR_ERR(ptr);
+       dis = ptr + erofs_blkoff(*pos);
+
+       if (!dif->path) {
+               if (!dis->tag[0]) {
+                       erofs_err(sb, "empty device tag @ pos %llu", *pos);
+                       return -EINVAL;
+               }
+               dif->path = kmemdup_nul(dis->tag, sizeof(dis->tag), GFP_KERNEL);
+               if (!dif->path)
+                       return -ENOMEM;
+       }
+
+       if (erofs_is_fscache_mode(sb)) {
+               ret = erofs_fscache_register_cookie(sb, &dif->fscache,
+                               dif->path, false);
+               if (ret)
+                       return ret;
+       } else {
+               bdev = blkdev_get_by_path(dif->path, FMODE_READ | FMODE_EXCL,
+                                         sb->s_type);
+               if (IS_ERR(bdev))
+                       return PTR_ERR(bdev);
+               dif->bdev = bdev;
+               dif->dax_dev = fs_dax_get_by_bdev(bdev, &dif->dax_part_off);
+       }
+
+       dif->blocks = le32_to_cpu(dis->blocks);
+       dif->mapped_blkaddr = le32_to_cpu(dis->mapped_blkaddr);
+       sbi->total_blocks += dif->blocks;
+       *pos += EROFS_DEVT_SLOT_SIZE;
+       return 0;
+}
+
+static int erofs_scan_devices(struct super_block *sb,
                              struct erofs_super_block *dsb)
 {
        struct erofs_sb_info *sbi = EROFS_SB(sb);
@@ -228,8 +273,6 @@ static int erofs_init_devices(struct super_block *sb,
        erofs_off_t pos;
        struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
        struct erofs_device_info *dif;
-       struct erofs_deviceslot *dis;
-       void *ptr;
        int id, err = 0;
 
        sbi->total_blocks = sbi->primarydevice_blocks;
@@ -238,7 +281,8 @@ static int erofs_init_devices(struct super_block *sb,
        else
                ondisk_extradevs = le16_to_cpu(dsb->extra_devices);
 
-       if (ondisk_extradevs != sbi->devs->extra_devices) {
+       if (sbi->devs->extra_devices &&
+           ondisk_extradevs != sbi->devs->extra_devices) {
                erofs_err(sb, "extra devices don't match (ondisk %u, given %u)",
                          ondisk_extradevs, sbi->devs->extra_devices);
                return -EINVAL;
@@ -249,39 +293,31 @@ static int erofs_init_devices(struct super_block *sb,
        sbi->device_id_mask = roundup_pow_of_two(ondisk_extradevs + 1) - 1;
        pos = le16_to_cpu(dsb->devt_slotoff) * EROFS_DEVT_SLOT_SIZE;
        down_read(&sbi->devs->rwsem);
-       idr_for_each_entry(&sbi->devs->tree, dif, id) {
-               struct block_device *bdev;
-
-               ptr = erofs_read_metabuf(&buf, sb, erofs_blknr(pos),
-                                        EROFS_KMAP);
-               if (IS_ERR(ptr)) {
-                       err = PTR_ERR(ptr);
-                       break;
-               }
-               dis = ptr + erofs_blkoff(pos);
-
-               if (erofs_is_fscache_mode(sb)) {
-                       err = erofs_fscache_register_cookie(sb, &dif->fscache,
-                                                           dif->path, false);
+       if (sbi->devs->extra_devices) {
+               idr_for_each_entry(&sbi->devs->tree, dif, id) {
+                       err = erofs_init_device(&buf, sb, dif, &pos);
                        if (err)
                                break;
-               } else {
-                       bdev = blkdev_get_by_path(dif->path,
-                                                 FMODE_READ | FMODE_EXCL,
-                                                 sb->s_type);
-                       if (IS_ERR(bdev)) {
-                               err = PTR_ERR(bdev);
+               }
+       } else {
+               for (id = 0; id < ondisk_extradevs; id++) {
+                       dif = kzalloc(sizeof(*dif), GFP_KERNEL);
+                       if (!dif) {
+                               err = -ENOMEM;
                                break;
                        }
-                       dif->bdev = bdev;
-                       dif->dax_dev = fs_dax_get_by_bdev(bdev,
-                                                         &dif->dax_part_off);
-               }
 
-               dif->blocks = le32_to_cpu(dis->blocks);
-               dif->mapped_blkaddr = le32_to_cpu(dis->mapped_blkaddr);
-               sbi->total_blocks += dif->blocks;
-               pos += EROFS_DEVT_SLOT_SIZE;
+                       err = idr_alloc(&sbi->devs->tree, dif, 0, 0, GFP_KERNEL);
+                       if (err < 0) {
+                               kfree(dif);
+                               break;
+                       }
+                       ++sbi->devs->extra_devices;
+
+                       err = erofs_init_device(&buf, sb, dif, &pos);
+                       if (err)
+                               break;
+               }
        }
        up_read(&sbi->devs->rwsem);
        erofs_put_metabuf(&buf);
@@ -368,7 +404,7 @@ static int erofs_read_superblock(struct super_block *sb)
                goto out;
 
        /* handle multiple devices */
-       ret = erofs_init_devices(sb, dsb);
+       ret = erofs_scan_devices(sb, dsb);
 
        if (erofs_sb_has_ztailpacking(sbi))
                erofs_info(sb, "EXPERIMENTAL compressed inline data feature in use. Use at your own risk!");