linux-kernel - Re: [PATCH] arm64: Fix 'lock held when returning to user space' lockdep warning

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: <ZHXF86CioMoyWD1d@FVFF77S0Q05N>
Date:   Tue, 30 May 2023 10:46:27 +0100
From:   Mark Rutland <mark.rutland@....com>
To:     Eric Chan <ericchancf@...gle.com>
Cc:     Catalin Marinas <catalin.marinas@....com>,
        Will Deacon <will@...nel.org>,
        "Paul E . McKenney" <paulmck@...nel.org>,
        Mark Brown <broonie@...nel.org>,
        Frederic Weisbecker <frederic@...nel.org>,
        Mukesh Ojha <quic_mojha@...cinc.com>,
        Josh Poimboeuf <jpoimboe@...nel.org>,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] arm64: Fix 'lock held when returning to user space'
 lockdep warning

Hi Eric,

The subject line for this patch is misleading, as it makes it sound like we're
currently hitting a warning of the form:

  lock held when returning to user space

... when in actuality this patch is *adding* support for that warning.

This is a missing feature, not a fix (and e.g. doesn't require backporting to
stable).

On Fri, May 26, 2023 at 06:48:23PM +0000, Eric Chan wrote:
> The arm64 architecture lacks support for CONFIG_GENERIC_ENTRY, resulting
> in the failure to report the lockdep warning
> "lock held when returning to user space" when lockdep is enabled.
> 
> Rename the function to align with exit_to_user_mode_prepare in
> kernel/entry/common.c to improve readability.

Considering the point about the commit title, could you please reword this to
be clear that we're adding support for a feature, rather than fixing something
that's broken, e.g.

| arm64: lockdep: enable checks for held locks when returning to userspace
|
| Currently arm64 doesn't use CONFIG_GENERIC_ENTRY and doesn't call
| lockdep_sys_exit() when returning to userspace. This means that lockdep won't
| check for held locks when returning to userspace, which would be useful to
| detect kernel bugs.
|
| Call lockdep_sys_exit() when returning to userspace, enabling checking for
| held locks.
|
| At the same time, rename arm64's prepare_exit_to_user_mode() to
| exit_to_user_mode_prepare() to more clearly align with the naming in the
| generic entry code.

With that wording:

Acked-by: Mark Rutland <mark.rutland@....com>

Thanks,
Mark.

> Signed-off-by: Eric Chan <ericchancf@...gle.com>
> ---
>  arch/arm64/kernel/entry-common.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c
> index 3af3c01c93a6..9d6827201b6c 100644
> --- a/arch/arm64/kernel/entry-common.c
> +++ b/arch/arm64/kernel/entry-common.c
> @@ -126,7 +126,7 @@ static __always_inline void __exit_to_user_mode(void)
>  	lockdep_hardirqs_on(CALLER_ADDR0);
>  }
>  
> -static __always_inline void prepare_exit_to_user_mode(struct pt_regs *regs)
> +static __always_inline void exit_to_user_mode_prepare(struct pt_regs *regs)
>  {
>  	unsigned long flags;
>  
> @@ -135,11 +135,13 @@ static __always_inline void prepare_exit_to_user_mode(struct pt_regs *regs)
>  	flags = read_thread_flags();
>  	if (unlikely(flags & _TIF_WORK_MASK))
>  		do_notify_resume(regs, flags);
> +
> +	lockdep_sys_exit();
>  }
>  
>  static __always_inline void exit_to_user_mode(struct pt_regs *regs)
>  {
> -	prepare_exit_to_user_mode(regs);
> +	exit_to_user_mode_prepare(regs);
>  	mte_check_tfsr_exit();
>  	__exit_to_user_mode();
>  }
> -- 
> 2.41.0.rc0.172.g3f132b7071-goog
>