On 20/02/2024 08:29, Christoph Hellwig wrote:
+#define BDEV_STATX_SUPPORTED_MASK (STATX_DIOALIGN | STATX_WRITE_ATOMIC)
+ if (!(request_mask & BDEV_STATX_SUPPORTED_MASK))
+ return;
BDEV_STATX_SUPPORTED_MASK is misleading here. bdevs support a lot more
fields, these are just the ones needing special attention. I'd do away
with the extra define and just open code it.
ok, fine
+ /* If this is a block device inode, override the filesystem
+ * attributes with the block device specific parameters
+ * that need to be obtained from the bdev backing inode
+ */
This is not the normal kernel multi-line comment format.
will fix
+ if (S_ISBLK(d_backing_inode(path.dentry)->i_mode))
+ bdev_statx(path.dentry, stat, request_mask);
I know I touched this last, but does anyone remember why we have
various random fixups in vfs_statx and not in vfs_getattr_nosec, where
they we have more of them and also the inode at hand?