linux-kernel - Re: [PATCH v3] locking: Document that some lock types must stay alive during unlock

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] [day] [month] [year] [list]

Message-ID: <ZW5lDougvHxM_F64@archie.me>
Date:   Tue, 5 Dec 2023 06:47:26 +0700
From:   Bagas Sanjaya <bagasdotme@...il.com>
To:     Jann Horn <jannh@...gle.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>
Cc:     Will Deacon <will@...nel.org>, Waiman Long <longman@...hat.com>,
        Jonathan Corbet <corbet@....net>,
        David Laight <David.Laight@...lab.com>,
        attreyee-muk <tintinm2017@...il.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Linux Documentation <linux-doc@...r.kernel.org>
Subject: Re: [PATCH v3] locking: Document that some lock types must stay
 alive during unlock

On Mon, Dec 04, 2023 at 02:22:59PM +0100, Jann Horn wrote:
> diff --git a/Documentation/locking/locktypes.rst b/Documentation/locking/locktypes.rst
> index 80c914f6eae7..c9a4bcc967ea 100644
> --- a/Documentation/locking/locktypes.rst
> +++ b/Documentation/locking/locktypes.rst
> @@ -95,6 +95,29 @@ rw_semaphores have a special interface which allows non-owner release for
>  readers.
>  
>  
> +Releasing and freeing
> +=====================
> +For some lock types, such as spinlocks, the lock release operation is designed
> +to allow another concurrent task to free the lock as soon as the lock has been
> +released - in other words, similarly to refcounts, the unlock operation will not
> +access the lock object anymore after marking it as unlocked.
> +
> +This behavior is guaranteed for:
> +
> + - spinlock_t (including in PREEMPT_RT kernels, where spinlock_t is
> +   implemented as an rtmutex)
> +
> +There are other lock types where the lock release operation makes no such
> +guarantee and the caller must ensure that the lock is not destroyed before the
> +unlock operation has returned.
> +Most sleeping locks are in this category.
> +
> +This is the case in particular for (not an exhaustive list):
> +
> + - mutex
> + - rw_semaphore
> +
> +
>  rtmutex
>  =======
>  
> diff --git a/Documentation/locking/mutex-design.rst b/Documentation/locking/mutex-design.rst
> index 78540cd7f54b..bbb4c4d56ed0 100644
> --- a/Documentation/locking/mutex-design.rst
> +++ b/Documentation/locking/mutex-design.rst
> @@ -101,6 +101,8 @@ features that make lock debugging easier and faster:
>      - Detects multi-task circular deadlocks and prints out all affected
>        locks and tasks (and only those tasks).
>  
> +The mutex user must ensure that the mutex is not destroyed while a unlock
> +operation is still in progress, see Documentation/locking/locktypes.rst.
>  
>  Interfaces
>  ----------
> diff --git a/kernel/locking/mutex.c b/kernel/locking/mutex.c
> index 2deeeca3e71b..fa4834dba407 100644
> --- a/kernel/locking/mutex.c
> +++ b/kernel/locking/mutex.c
> @@ -532,6 +532,11 @@ static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigne
>   * This function must not be used in interrupt context. Unlocking
>   * of a not locked mutex is not allowed.
>   *
> + * The caller must ensure that the mutex stays alive until this function has
> + * returned - mutex_unlock() can NOT directly be used to release an object such
> + * that another concurrent task can free it.
> + * See Documentation/locking/locktypes.rst.
> + *
>   * This function is similar to (but not equivalent to) up().
>   */
>  void __sched mutex_unlock(struct mutex *lock)
> diff --git a/kernel/locking/rwsem.c b/kernel/locking/rwsem.c
> index 2340b6d90ec6..cbc00a269deb 100644
> --- a/kernel/locking/rwsem.c
> +++ b/kernel/locking/rwsem.c
> @@ -1615,6 +1615,11 @@ EXPORT_SYMBOL(down_write_trylock);
>  
>  /*
>   * release a read lock
> + *
> + * The caller must ensure that the rw_semaphore stays alive until this function
> + * has returned - up_read() can NOT directly be used to release an object such
> + * that another concurrent task can free it.
> + * See Documentation/locking/locktypes.rst.
>   */
>  void up_read(struct rw_semaphore *sem)
>  {
> @@ -1625,6 +1630,11 @@ EXPORT_SYMBOL(up_read);
>  
>  /*
>   * release a write lock
> + *
> + * The caller must ensure that the rw_semaphore stays alive until this function
> + * has returned - up_write() can NOT directly be used to release an object such
> + * that another concurrent task can free it.
> + * See Documentation/locking/locktypes.rst.
>   */
>  void up_write(struct rw_semaphore *sem)
>  {
> 
> base-commit: 3b47bc037bd44f142ac09848e8d3ecccc726be99

LGTM, thanks!

Reviewed-by: Bagas Sanjaya <bagasdotme@...il.com>

-- 
An old man doll... just what I always wanted! - Clara

Download attachment "signature.asc" of type "application/pgp-signature" (229 bytes)