linux-kernel - Re: [PATCH v2] rbd: replace strcpy() with strscpy()

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 for Android: free password hash cracker in your pocket

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

Message-ID: <f92ddea4-edf1-42f9-a738-51233ce3d45e@ieee.org>
Date: Tue, 20 May 2025 11:44:15 -0500
From: Alex Elder <elder@...e.org>
To: Siddarth Gundu <siddarthsgml@...il.com>, idryomov@...il.com,
 dongsheng.yang@...ystack.cn, axboe@...nel.dk, ceph-devel@...r.kernel.org,
 linux-block@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] rbd: replace strcpy() with strscpy()

On 5/19/25 1:38 AM, Siddarth Gundu wrote:
> strcpy() is deprecated; use strscpy() instead.
> 
> Both the destination and source buffer are of fixed length
> so strscpy with 2-arguments is used.
> 
> Introduce a typedef for cookie array to improve code clarity.
> 
> Link: https://github.com/KSPP/linux/issues/88
> Signed-off-by: Siddarth Gundu <siddarthsgml@...il.com>
> ---
> changes since v1
> - added a typedef for cookie arrays
> 
> About the typedef: I was a bit hesitant to add it since the kernel
> style guide is against adding new typedef but I wanted to follow
> the review feedback for this.

I personally think the typedef here is the appropriate.  But
it's really up to Ilya whether he likes this approach.  Get
his input before you do more.

There's a basic question about whether this is a useful
abstraction.  It's used for "lock cookies" but do they
serve a broader purpose?

The other part of my suggestion was to define functions that
provide an API.  For example:

static inline rbd_cookie_t rbd_cookie_set(rbd_cookie_t cookie, u64 id);
static inline u64 rbd_cookie_get(rbd_cookie_t cookie);

Anyway, before I say any more let's see if Ilya even wants
to go in this direction.  Your original proposal was OK, I
just thought specifying the length might be safer.

					-Alex

>   drivers/block/rbd.c | 13 ++++++++-----
>   1 file changed, 8 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
> index faafd7ff43d6..863d9c591aa5 100644
> --- a/drivers/block/rbd.c
> +++ b/drivers/block/rbd.c
> @@ -46,11 +46,14 @@
>   #include <linux/slab.h>
>   #include <linux/idr.h>
>   #include <linux/workqueue.h>
> +#include <linux/string.h>
>   
>   #include "rbd_types.h"
>   
>   #define RBD_DEBUG	/* Activate rbd_assert() calls */
>   
> +typedef char rbd_cookie_t[32];
> +
>   /*
>    * Increment the given counter and return its updated value.
>    * If the counter is already 0 it will not be incremented.
> @@ -411,7 +414,7 @@ struct rbd_device {
>   
>   	struct rw_semaphore	lock_rwsem;
>   	enum rbd_lock_state	lock_state;
> -	char			lock_cookie[32];
> +	rbd_cookie_t		lock_cookie;
>   	struct rbd_client_id	owner_cid;
>   	struct work_struct	acquired_lock_work;
>   	struct work_struct	released_lock_work;
> @@ -3649,12 +3652,12 @@ static void format_lock_cookie(struct rbd_device *rbd_dev, char *buf)
>   	mutex_unlock(&rbd_dev->watch_mutex);
>   }
>   
> -static void __rbd_lock(struct rbd_device *rbd_dev, const char *cookie)
> +static void __rbd_lock(struct rbd_device *rbd_dev, const rbd_cookie_t cookie)
>   {
>   	struct rbd_client_id cid = rbd_get_cid(rbd_dev);
>   
>   	rbd_dev->lock_state = RBD_LOCK_STATE_LOCKED;
> -	strcpy(rbd_dev->lock_cookie, cookie);
> +	strscpy(rbd_dev->lock_cookie, cookie);
>   	rbd_set_owner_cid(rbd_dev, &cid);
>   	queue_work(rbd_dev->task_wq, &rbd_dev->acquired_lock_work);
>   }
> @@ -3665,7 +3668,7 @@ static void __rbd_lock(struct rbd_device *rbd_dev, const char *cookie)
>   static int rbd_lock(struct rbd_device *rbd_dev)
>   {
>   	struct ceph_osd_client *osdc = &rbd_dev->rbd_client->client->osdc;
> -	char cookie[32];
> +	rbd_cookie_t cookie;
>   	int ret;
>   
>   	WARN_ON(__rbd_is_lock_owner(rbd_dev) ||
> @@ -4581,7 +4584,7 @@ static void rbd_unregister_watch(struct rbd_device *rbd_dev)
>   static void rbd_reacquire_lock(struct rbd_device *rbd_dev)
>   {
>   	struct ceph_osd_client *osdc = &rbd_dev->rbd_client->client->osdc;
> -	char cookie[32];
> +	rbd_cookie_t cookie;
>   	int ret;
>   
>   	if (!rbd_quiesce_lock(rbd_dev))