linux-kernel - Re: [PATCH] locking/ww_mutex: add ww_mutex_is_owned_by function v4

Open Source and information security mailing list archives

Message-ID: <20180220163016.GZ22199@phenom.ffwll.local>
Date:   Tue, 20 Feb 2018 17:30:16 +0100
From:   Daniel Vetter <daniel@...ll.ch>
To:     Christian König 
        <ckoenig.leichtzumerken@...il.com>
Cc:     peterz@...radead.org, amd-gfx@...ts.freedesktop.org,
        dri-devel@...ts.freedesktop.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] locking/ww_mutex: add ww_mutex_is_owned_by function v4

On Tue, Feb 20, 2018 at 03:49:46PM +0100, Christian König wrote:
> amdgpu needs to verify if userspace sends us valid addresses and the simplest
> way of doing this is to check if the buffer object is locked with the ticket
> of the current submission.
> 
> Clean up the access to the ww_mutex internals by providing a function
> for this and extend the check to the thread owning the underlying mutex.
> 
> v2: split amdgpu changes into separate patch as suggested by Alex
> v3: change logic as suggested by Daniel
> v4: fix test in case ctx is NULL
> 
> Signed-off-by: Christian König <christian.koenig@....com>
> ---
>  include/linux/ww_mutex.h | 19 +++++++++++++++++++
>  1 file changed, 19 insertions(+)
> 
> diff --git a/include/linux/ww_mutex.h b/include/linux/ww_mutex.h
> index 39fda195bf78..fea4acc0bcbc 100644
> --- a/include/linux/ww_mutex.h
> +++ b/include/linux/ww_mutex.h
> @@ -358,4 +358,23 @@ static inline bool ww_mutex_is_locked(struct ww_mutex *lock)
>  	return mutex_is_locked(&lock->base);
>  }
>  
> +/**
> + * ww_mutex_is_owned_by - is the w/w mutex locked by this task in that context
> + * @lock: the mutex to be queried
> + * @ctx: the w/w acquire context to test
> + *
> + * If @ctx is not NULL test if the mutex is owned by this context.
> + * If @ctx is NULL test if the mutex is owned by the current thread and not
> + * locked in any context.

Ok I think with the increased guarantees this needs a better kerneldoc, I
think we need to at least add ".... and not locked in any context, i.e.
acquired using either ww_mutex_trylock() or ww_mutex_lock() without
supplying a &ww_acquire_ctx."

I think with that we have a nice 1:1 mapping between the context you
supply to the ww_mutex_lock function and to this one. That means generic
code which takes a ctx (but can handle NULL) should dtrt no matter what.

With the kerneldoc augmented:

Reviewed-by: Daniel Vetter <daniel.vetter@...ll.ch>

I'll probably regret this r-b too, but hey to err is human :-)

Cheers, Daniel

> + */
> +static inline bool ww_mutex_is_owned_by(struct ww_mutex *lock,
> +					struct ww_acquire_ctx *ctx)
> +{
> +	if (ctx)
> +		return likely(READ_ONCE(lock->ctx) == ctx);
> +
> +	return likely(__mutex_owner(&lock->base) == current) &&
> +		likely(READ_ONCE(lock->ctx) == NULL;
> +}
> +
>  #endif
> -- 
> 2.14.1
> 
> _______________________________________________
> dri-devel mailing list
> dri-devel@...ts.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/dri-devel

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

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