NACK: [SRU][J][PATCH 2/3] cgroup: Make cgroup_get_from_id() prettier

[Koichiro Den]

On Tue, Feb 11, 2025 at 09:58:40AM GMT, Philip Cox wrote:
> From: Tejun Heo <tj at kernel.org>
> 
> BugLink: https://bugs.launchpad.net/bugs/2089318
> 
> After merging 836ac87d ("cgroup: fix cgroup_get_from_id") into for-6.1, its
> combination with two commits in for-6.1 - 4534dee9 ("cgroup: cgroup: Honor
> caller's cgroup NS when resolving cgroup id") and fa7e439c ("cgroup:
> Homogenize cgroup_get_from_id() return value") - makes the gotos in the
> error handling path too ugly while not adding anything of value.
> 
> All that the gotos are saving is one extra kernfs_put() call. Let's remove
> the gotos and perform error returns directly.
> 
> Signed-off-by: Tejun Heo <tj at kernel.org>
> Cc: Ming Lei <ming.lei at redhat.com>
> Cc: Michal Koutný <mkoutny at suse.com>
> (backported from commit 7e1eb5437d3c3fdb61d45378579aab383cafc694)
> [context changes in cgroup_get_from_id()]
> Signed-off-by: Philip Cox <philip.cox at canonical.com>
> ---
>  kernel/cgroup/cgroup.c | 17 +++++++++--------
>  1 file changed, 9 insertions(+), 8 deletions(-)
> 
> diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
> index 27d503fd17fe5..bc528b8635a5f 100644
> --- a/kernel/cgroup/cgroup.c
> +++ b/kernel/cgroup/cgroup.c
> @@ -6038,14 +6038,16 @@ void cgroup_path_from_kernfs_id(u64 id, char *buf, size_t buflen)
>  struct cgroup *cgroup_get_from_id(u64 id)
>  {
>  	struct kernfs_node *kn;
> -	struct cgroup *cgrp = NULL, *root_cgrp;
> +	struct cgroup *cgrp, *root_cgrp;
>  
>  	kn = kernfs_find_and_get_node_by_id(cgrp_dfl_root.kf_root, id);
>  	if (!kn)
> -		goto out;
> +		return ERR_PTR(-ENOENT);
>  
> -	if (kernfs_type(kn) != KERNFS_DIR)
> -		goto put;
> +	if (kernfs_type(kn) != KERNFS_DIR) {
> +		kernfs_put(kn);
> +		return ERR_PTR(-ENOENT);
> +	}
>  
>  	rcu_read_lock();
>  
> @@ -6054,20 +6056,19 @@ struct cgroup *cgroup_get_from_id(u64 id)
>  		cgrp = NULL;
>  
>  	rcu_read_unlock();
> -put:
>  	kernfs_put(kn);
>  
>  	if (!cgrp)
> -		goto out;
> +		return ERR_PTR(-ENOENT);
>  
>  	spin_lock_irq(&css_set_lock);
>  	root_cgrp = current_cgns_cgroup_from_root(&cgrp_dfl_root);
>  	spin_unlock_irq(&css_set_lock);
>  	if (!cgroup_is_descendant(cgrp, root_cgrp)) {
>  		cgroup_put(cgrp);
> -		cgrp = NULL;
> +		return ERR_PTR(-ENOENT);
>  	}
> -out:
> +
>  	return cgrp;
>  }
>  EXPORT_SYMBOL_GPL(cgroup_get_from_id);
> -- 
> 2.43.0
> 

This breaks blkcg_set_fc_appid(). With this backport, the return value of
cgroup_get_from_id() can be e.g. ERR_PTR(-ENOENT). Commit fa7e439cf90b
("cgroup: Homogenize cgroup_get_from_id() return value") is missing, so
we need to adjust (or backport it as prereq).

Also, please follow the long-existing rule below:

 (https://canonical-kernel-docs.readthedocs-hosted.com/en/latest/reference/stable-patch-format/#comment-body)
 > [...]
 >
 >  There must be a brief explanation immediately after the "(backported
 >  from ...)" block, between square brackets, with the name of the person
 >  who introduced the change.                          ^^^^^^^^^^^^^^^^^^