linux-kernel - Re: [PATCH v2] regmap: debugfs: Fix name collision without atomic operations

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <2025030437-copious-irrigate-1b63@gregkh>
Date: Tue, 4 Mar 2025 17:23:02 +0100
From: Greg KH <gregkh@...uxfoundation.org>
To: Zxyan Zhu <zxyan20@....com>
Cc: broonie@...nel.org, rafael@...nel.org, dakr@...nel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] regmap: debugfs: Fix name collision without atomic
 operations

On Tue, Mar 04, 2025 at 10:24:52PM +0800, Zxyan Zhu wrote:
> The `dummy_index` global variable caused debugfs file name conflicts
> during re-entry, leading to creation failures. Use atomic operations
> to ensure safe and unique debugfs `dummy%d` naming.
> 
> Changes since v1:
> - Replaced atomic_read + atomic_inc with atomic_inc_return.
> - Added atomic_dec in the error path to maintain index consistency.
> - Updated the commit message to clarify the fix.
> 
> Signed-off-by: Zxyan Zhu <zxyan20@....com>
> ---
>  drivers/base/regmap/regmap-debugfs.c | 12 +++++++-----
>  1 file changed, 7 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/base/regmap/regmap-debugfs.c b/drivers/base/regmap/regmap-debugfs.c
> index fb84cda92a75..60c8d9a673b8 100644
> --- a/drivers/base/regmap/regmap-debugfs.c
> +++ b/drivers/base/regmap/regmap-debugfs.c
> @@ -20,7 +20,7 @@ struct regmap_debugfs_node {
>  	struct list_head link;
>  };
>  
> -static unsigned int dummy_index;
> +static atomic_t dummy_index = ATOMIC_INIT(0);
>  static struct dentry *regmap_debugfs_root;
>  static LIST_HEAD(regmap_debugfs_early_list);
>  static DEFINE_MUTEX(regmap_debugfs_early_lock);
> @@ -549,6 +549,7 @@ void regmap_debugfs_init(struct regmap *map)
>  	struct regmap_range_node *range_node;
>  	const char *devname = "dummy";
>  	const char *name = map->name;
> +	int index;
>  
>  	/*
>  	 * Userspace can initiate reads from the hardware over debugfs.
> @@ -595,12 +596,13 @@ void regmap_debugfs_init(struct regmap *map)
>  
>  	if (!strcmp(name, "dummy")) {
>  		kfree(map->debugfs_name);
> -		map->debugfs_name = kasprintf(GFP_KERNEL, "dummy%d",
> -						dummy_index);
> -		if (!map->debugfs_name)
> +		index = atomic_inc_return(&dummy_index);
> +		map->debugfs_name = kasprintf(GFP_KERNEL, "dummy%d", index);
> +		if (!map->debugfs_name) {
> +			atomic_dec(&dummy_index);
>  			return;
> +		}
>  		name = map->debugfs_name;
> -		dummy_index++;

Shouldn't you just use an idr here if there is a race condition?
There's a lock built into it that should solve all of these issues.

thanks,

greg k-h