> nvmet: refuse files on pseudo filesystems as namespace backing files
TL;DR: The series may not build: file_check_backing() in fs/read_write.c calls
d_real_data_inode(), which seems undefined anywhere in the tree, so a
prerequisite patch may be missing.
> diff --git a/drivers/nvme/target/io-cmd-file.c
> b/drivers/nvme/target/io-cmd-file.c
> --- a/drivers/nvme/target/io-cmd-file.c
> +++ b/drivers/nvme/target/io-cmd-file.c
> @@ -47,10 +47,9 @@ int nvmet_file_ns_enable(struct nvmet_ns *ns)
> return ret;
> }
>
> - if (!ns->file->f_op->read_iter || !ns->file->f_op->write_iter) {
> - ret = -EINVAL;
> + ret = file_check_backing(ns->file, FMODE_READ | FMODE_WRITE, S_IFBLK);
> + if (ret)
> goto err;
> - }
Can this be built? nvmet_file_ns_enable() now depends on
file_check_backing() from f66083ed0be9, which does:
fs/read_write.c:file_check_backing() {
...
/* Look through stacking filesystems at the inode with the data. */
real_inode = d_real_data_inode(file->f_path.dentry);
if (IS_ERR(real_inode))
...
}
d_real_data_inode() is not defined or declared anywhere in the tree, and
no commit up to the end of the series adds it. Only d_real_inode() exists
in include/linux/dcache.h, so fs/read_write.c looks like it fails with an
implicit function declaration.
If d_real_inode() was intended, it does d_inode(d_real(...)) without any
ERR_PTR handling, so the IS_ERR() check would never trigger the error path
the f66083ed0be9 commit message describes.
Is a prerequisite patch adding d_real_data_inode() missing from the series?
---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md
CI run summary: https://github.com/kernel-patches/bpf/actions/runs/43858134663