Before the device_id conversion this error named its target:
cannot allocate data in the metazone when unavailable for %s
It is now a single device message, and device_id < 0 means the metadata
zone rather than a device, so that case prints the sentinel:
cannot allocate data on unavailable device -1 for ...
Reachable with mkfs.erofs --incremental --MZ over an image built with
--MZ. Name the metadata zone again there.
Signed-off-by: Zhan Xusheng <[email protected]>
---
lib/inode.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/lib/inode.c b/lib/inode.c
index d71a9d1..2cd7bed 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -213,8 +213,12 @@ int erofs_allocate_inode_bh_data(struct erofs_inode
*inode, erofs_blk_t nblocks,
bmgr = sbi->devs[device_id - 1].bmgr;
if (!bmgr) {
- erofs_err("cannot allocate data on unavailable device %d for
%s",
- device_id, inode->i_srcpath);
+ if (device_id < 0)
+ erofs_err("cannot allocate data in the unavailable
metadata zone for %s",
+ inode->i_srcpath);
+ else
+ erofs_err("cannot allocate data on unavailable device
%d for %s",
+ device_id, inode->i_srcpath);
return -EINVAL;
}
--
2.43.0
