linux-kernel - Re: [PATCH 1/5] locking: Add rwsem_is_write

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]

Date:   Sun, 10 Sep 2023 22:15:59 -0400
From:   Waiman Long <longman@...hat.com>
To:     Dave Chinner <david@...morbit.com>,
        Matthew Wilcox <willy@...radead.org>
Cc:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>, Will Deacon <will@...nel.org>,
        linux-kernel@...r.kernel.org, linux-mm@...ck.org,
        Chandan Babu R <chandan.babu@...cle.com>,
        "Darrick J . Wong" <djwong@...nel.org>, linux-xfs@...r.kernel.org
Subject: Re: [PATCH 1/5] locking: Add rwsem_is_write_locked()


On 9/10/23 20:55, Dave Chinner wrote:
> On Mon, Sep 11, 2023 at 12:17:18AM +0100, Matthew Wilcox wrote:
>> On Mon, Sep 11, 2023 at 08:56:45AM +1000, Dave Chinner wrote:
>>> On Fri, Sep 08, 2023 at 12:44:34PM +0200, Peter Zijlstra wrote:
>>>> Agreed, and this is fine. However there's been some very creative
>>>> 'use' of the _is_locked() class of functions in the past that did not
>>>> follow 'common' sense.
>>>>
>>>> If all usage was: I should be holding this, lets check. I probably
>>>> wouldn't have this bad feeling about things.
>>> So your argument against such an interface is essentially "we can't
>>> have nice things because someone might abuse them"?
>> Some people are very creative ...
> Sure, but that's no reason to stop anyone else from making progress.
>
>> I was thinking about how to handle this better.  We could have
>>
>> static inline void rwsem_assert_locked(const struct rw_semaphore *sem)
>> {
>> 	BUG_ON(atomic_long_read(&sem->count) == 0);
>> }
>>
>> static inline void rwsem_assert_write_locked(const struct rw_semaphore *sem)
>> {
>> 	BUG_ON((atomic_long_read(&sem->count) & 1) != 1);
>> }
> We already have CONFIG_DEBUG_RWSEMS, so we can put these
> introspection interfaces inside debug code, and make any attempt to
> use them outside of debug builds break the build. e.g:
>
> #if DEBUG_RWSEMS
> /*
>   * rwsem locked checks can only be used by conditionally compiled
>   * subsystem debug code. It is not valid to use them in normal
>   * production code.
>   */
> static inline bool rwsem_is_write_locked()
> {
> 	....
> }
>
> static inline bool rwsem_is_locked()
> {
> 	....
> }
> #else /* !DEBUG_RWSEMS */
> #define rwsem_is_write_locked()		BUILD_BUG()
> #define rwsem_is_locked()		BUILD_BUG()
> #endif /* DEBUG_RWSEMS */
>
> And now we simply add a single line to subsystem Kconfig debug
> options to turn on rwsem introspection for their debug checks like
> so:
>
>   config XFS_DEBUG
>   	bool "XFS Debugging support"
>   	depends on XFS_FS
> +	select RWSEM_DEBUG
>   	help
>   	  Say Y here to get an XFS build with many debugging features,
>   	  including ASSERT checks, function wrappers around macros,

That may be a possible compromise. Actually, I am not against having 
them defined even outside the DEBUG_RWSEMS. We already have 
mutex_is_locked() defined and used in a lot of places. So this is just 
providing the rwsem equivalents.

I also agreed that these APIs can be misused by other users. I think we 
should clearly document the caveats of using these. So unless there are 
other means of maintaining the stability of the lock state, the test 
result may no longer be true right after the test. It is simply just the 
lock state at a certain moment in time. Callers are using them at their 
own risk.

Cheers,
Longman