linux-kernel - Re: [kernel-hardening] [RFC 2/7] init: add set_ro_mostly_after_init

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20170220102226.GB9003@leverpostej>
Date:   Mon, 20 Feb 2017 10:22:26 +0000
From:   Mark Rutland <mark.rutland@....com>
To:     Hoeun Ryu <hoeun.ryu@...il.com>
Cc:     kernel-hardening@...ts.openwall.com, linux-kernel@...r.kernel.org,
        Kees Cook <keescook@...omium.org>,
        Jessica Yu <jeyu@...hat.com>, Ingo Molnar <mingo@...nel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Emese Revfy <re.emese@...il.com>,
        AKASHI Takahiro <takahiro.akashi@...aro.org>,
        Fabian Frederick <fabf@...net.be>,
        Helge Deller <deller@....de>,
        Laura Abbott <labbott@...hat.com>,
        Nicholas Piggin <npiggin@...il.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Petr Mladek <pmladek@...e.com>, Yang Shi <yang.shi@...aro.org>,
        Rasmus Villemoes <linux@...musvillemoes.dk>,
        Tejun Heo <tj@...nel.org>, Prarit Bhargava <prarit@...hat.com>,
        Lokesh Vutla <lokeshvutla@...com>
Subject: Re: [kernel-hardening] [RFC 2/7] init: add
 set_ro_mostly_after_init_rw/ro function

On Sun, Feb 19, 2017 at 07:04:05PM +0900, Hoeun Ryu wrote:
>  Add set_ro_mostly_after_init_rw/ro pair to modify memory attributes for
> memory marked as `ro_mostly_after_init`.
> 
>  I am doubtful that this is the right place where these functions reside and
> these functions are suitable for all architectures for memory attributes
> modification. Please comment.

These won't work for arm64, since set_memory_* only work on
page-granular mappings in the vmalloc area.

The "real" kernel mappings can use larger block mappings, and would need
to be split (which cannot be done at runtime) before permissions could
be changed at page granularity.

Thanks,
Mark.

> Signed-off-by: Hoeun Ryu <hoeun.ryu@...il.com>
> ---
>  include/linux/init.h |  6 ++++++
>  init/main.c          | 24 ++++++++++++++++++++++++
>  2 files changed, 30 insertions(+)
> 
> diff --git a/include/linux/init.h b/include/linux/init.h
> index 79af096..d68e4f7 100644
> --- a/include/linux/init.h
> +++ b/include/linux/init.h
> @@ -131,6 +131,12 @@ extern bool rodata_enabled;
>  #endif
>  #ifdef CONFIG_STRICT_KERNEL_RWX
>  void mark_rodata_ro(void);
> +
> +void set_ro_mostly_after_init_rw(void);
> +void set_ro_mostly_after_init_ro(void);
> +#else
> +static inline void set_ro_mostly_after_init_rw(void) { }
> +static inline void set_ro_mostly_after_init_ro(void) { }
>  #endif
>  
>  extern void (*late_time_init)(void);
> diff --git a/init/main.c b/init/main.c
> index 4719abf..a5d4873 100644
> --- a/init/main.c
> +++ b/init/main.c
> @@ -941,6 +941,30 @@ static void mark_readonly(void)
>  	} else
>  		pr_info("Kernel memory protection disabled.\n");
>  }
> +
> +void set_ro_mostly_after_init_rw(void)
> +{
> +	unsigned long start = PFN_ALIGN(__start_data_ro_mostly_after_init);
> +	unsigned long end = PFN_ALIGN(&__end_data_ro_mostly_after_init);
> +	unsigned long nr_pages = (end - start) >> PAGE_SHIFT;
> +
> +	if (!rodata_enabled)
> +		return;
> +
> +	set_memory_rw(start, nr_pages);
> +}
> +
> +void set_ro_mostly_after_init_ro(void)
> +{
> +	unsigned long start = PFN_ALIGN(__start_data_ro_mostly_after_init);
> +	unsigned long end = PFN_ALIGN(&__end_data_ro_mostly_after_init);
> +	unsigned long nr_pages = (end - start) >> PAGE_SHIFT;
> +
> +	if (!rodata_enabled)
> +		return;
> +
> +	set_memory_ro(start, nr_pages);
> +}
>  #else
>  static inline void mark_readonly(void)
>  {
> -- 
> 2.7.4
>