linux-kernel - Re: [PATCH V10 02/28] powerpc, process: Add the function flush_tmregs_to

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: <20160302111500.7b50fac9@camb691>
Date:	Wed, 2 Mar 2016 11:15:00 +1100
From:	Cyril Bur <cyrilbur@...il.com>
To:	Anshuman Khandual <khandual@...ux.vnet.ibm.com>
Cc:	linux-kernel@...r.kernel.org, linuxppc-dev@...abs.org
Subject: Re: [PATCH V10 02/28] powerpc, process: Add the function
 flush_tmregs_to_thread

On Tue, 16 Feb 2016 14:29:32 +0530
Anshuman Khandual <khandual@...ux.vnet.ibm.com> wrote:

> This patch creates a function flush_tmregs_to_thread which
> will then be used by subsequent patches in this series. The
> function checks for self tracing ptrace interface attempts
> while in the TM context and logs appropriate warning message.
> 

Hi Anshuman,

You'll have to bare with me, my ptrace knowledge is non existent so you might
have to walk me though some aspects.

I have been playing with FPU/VMX and VSX saving so I thought I'd take a look.

> Signed-off-by: Anshuman Khandual <khandual@...ux.vnet.ibm.com>
> ---
>  arch/powerpc/include/asm/switch_to.h |  8 ++++++++
>  arch/powerpc/kernel/process.c        | 20 ++++++++++++++++++++
>  2 files changed, 28 insertions(+)
> 
> diff --git a/arch/powerpc/include/asm/switch_to.h b/arch/powerpc/include/asm/switch_to.h
> index 5b268b6..7b297bf 100644
> --- a/arch/powerpc/include/asm/switch_to.h
> +++ b/arch/powerpc/include/asm/switch_to.h
> @@ -70,6 +70,14 @@ static inline void disable_kernel_spe(void)
>  }
>  #endif
>  
> +#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
> +extern void flush_tmregs_to_thread(struct task_struct *);
> +#else
> +static inline void flush_tmregs_to_thread(struct task_struct *t)
> +{
> +}
> +#endif
> +
>  static inline void clear_task_ebb(struct task_struct *t)
>  {
>  #ifdef CONFIG_PPC_BOOK3S_64
> diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
> index dccc87e..2c4fa7f 100644
> --- a/arch/powerpc/kernel/process.c
> +++ b/arch/powerpc/kernel/process.c
> @@ -918,6 +918,26 @@ static inline void restore_sprs(struct thread_struct *old_thread,
>  #endif
>  }
>  



> +#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
> +void flush_tmregs_to_thread(struct task_struct *tsk)
> +{
> +	/*
> +	 * Process self tracing is not yet supported through
> +	 * ptrace interface. Ptrace generic code should have
> +	 * prevented this from happening in the first place.
> +	 * Warn once here with the message, if some how it
> +	 * is attempted.
> +	 */
> +	WARN_ONCE(tsk == current,
> +		"Not expecting ptrace on self: TM regs may be incorrect\n");
> +
> +	/*
> +	 * If task is not current, it should have been flushed
> +	 * already to it's thread_struct during __switch_to().
> +	 */

I totally agree except this highlights something that I notice in subsequent
patches, and existing code. All the *_{get,set}() functions call
flush_*_to_thread() when, as per your comment (and my understanding of task
switching) there really shouldn't be a need to do that. My only thought is that
this could be a relic of uniprocessor days when it would have been necessary but
Anton recently stripped that out. Are you able to shed some light here?

The reason I ask is that if the flush_*_to_thread() calls ARE actually
important then I worry that this function is inadequate...

> +}
> +#endif
> +
>  struct task_struct *__switch_to(struct task_struct *prev,
>  	struct task_struct *new)
>  {