Re: [PATCH v2 007/110] ext4: use PRIino format for i_ino
From: Jeff Layton <jlayton@kernel.org>
Date: 2026-03-03 11:42:09
Also in:
amd-gfx, autofs, bpf, ceph-devel, dri-devel, linux-bluetooth, linux-can, linux-cifs, linux-ext4, linux-f2fs-devel, linux-fscrypt, linux-fsdevel, linux-hams, linux-integrity, linux-media, linux-mm, linux-nfs, linux-perf-users, linux-sctp, linux-trace-kernel, linux-unionfs, linux-xfs, lkml, netdev, netfs, ntfs3, nvdimm, ocfs2-devel, selinux, v9fs
On Tue, 2026-03-03 at 12:20 +0100, Jan Kara wrote:
On Mon 02-03-26 15:23:51, Jeff Layton wrote:
quoted
Convert ext4 i_ino format strings to use the PRIino format
macro in preparation for the widening of i_ino via kino_t.
In trace events, change __field(ino_t, ...) to __field(u64, ...)
and update TP_printk format strings to %llu/%llx to match the
widened field type.
Update local variables and function parameters that hold i_ino
values from unsigned long to kino_t.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Two small comments. Otherwise feel free to add:
Reviewed-by: Jan Kara <jack@suse.cz>
quoted
diff --git a/fs/ext4/migrate.c b/fs/ext4/migrate.c
index 96ab95167bd6e10ba86e61a60cb0be9fbafe157f..43103816b80ef4901858bcd789acb0ffb2612317 100644
--- a/fs/ext4/migrate.c
+++ b/fs/ext4/migrate.c
@@ -455,7 +455,7 @@ int ext4_ext_migrate(struct inode *inode)
* log, so disable fast commits for this transaction.
*/
ext4_fc_mark_ineligible(inode->i_sb, EXT4_FC_REASON_MIGRATE, handle);
- goal = (((inode->i_ino - 1) / EXT4_INODES_PER_GROUP(inode->i_sb)) *
+ goal = (div_u64(inode->i_ino - 1, EXT4_INODES_PER_GROUP(inode->i_sb)) *
Ext4 doesn't support more than 2^32 inodes (due to on-disk format). Thus
i_ino is always guaranteed to be a number that fits in 32-bits. Thus I'd
here just type i_ino to (unsigned int) and be done with it like you've done
it at other places.
...
Thanks. Fixed both places. I ended up casting the above to a u32 since
this patchset has given me a stronger affinity for explicit-width
types.
quoted
@@ -1823,7 +1823,7 @@ TRACE_EVENT(ext4_journal_start_inode,
TP_ARGS(inode, blocks, rsv_blocks, revoke_creds, type, IP),
TP_STRUCT__entry(
- __field( unsigned long, ino )
+ __field( u64, ino )
__field( dev_t, dev )
__field( unsigned long, ip )
__field( int, blocks )
@@ -1843,9 +1843,10 @@ TRACE_EVENT(ext4_journal_start_inode,
),
TP_printk("dev %d,%d blocks %d, rsv_blocks %d, revoke_creds %d,"
- " type %d, ino %lu, caller %pS", MAJOR(__entry->dev),
+ " type %d, ino %llu, caller %pS", MAJOR(__entry->dev),
MINOR(__entry->dev), __entry->blocks, __entry->rsv_blocks,
- __entry->revoke_creds, __entry->type, __entry->ino,
+ __entry->revoke_creds, __entry->type,
+ (unsigned long long) __entry->ino,
Not point in the type cast?
Honza
--
Jeff Layton [off-list ref]