linux-kernel - Re: [PATCH 2/2] serial: earlycon: Make early_con as _

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]

Date:   Fri, 7 Jul 2017 13:47:12 +0900
From:   Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>
To:     Matt Redfearn <matt.redfearn@...tec.com>
Cc:     Petr Mladek <pmladek@...e.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Jiri Slaby <jslaby@...e.com>, linux-serial@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] serial: earlycon: Make early_con as __initdata

On (07/06/17 11:38), Matt Redfearn wrote:
> All early console drivers that may be registered as the earlycon are
> marked __init to be placed in the init section. The drivers' code and
> data are freed during free_initmem_default() but the early console is
> not unregistered in printk_late_init() as the init_section_intersects()
> test fails. This leads to the earlycon still being active, potentially
> with dangling pointers into the freed init section, which may have been
> poisoned by the slab debugger. Attempting to use the boot console after
> this will likely lead to weird behaviour and kernel crashes.

people want to use early-con as a panic() console fallback.

	-ss

> Fix this by marking the early_con struct __initdata so that the
> init_section_intersects() test succeeds and the console is unregistered
> in printk_late_init() before the init section is freed.
> 
> The 8250 earlycon, on which the generic earlycon was based, had these
> attributes on it's console struct. The switch to the generic
> implementation in commit d2fd6810a823 ("tty/serial: convert 8250 to
> generic earlycon") appears to have broken unregistraton of the boot
> console when its code and data are in __init.
> 
> Fixes: d2fd6810a823 ("tty/serial: convert 8250 to generic earlycon")
> Signed-off-by: Matt Redfearn <matt.redfearn@...tec.com>
> 
> ---
> 
>  drivers/tty/serial/earlycon.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/tty/serial/earlycon.c b/drivers/tty/serial/earlycon.c
> index c3651540e1ba..388aaafcca2b 100644
> --- a/drivers/tty/serial/earlycon.c
> +++ b/drivers/tty/serial/earlycon.c
> @@ -29,13 +29,13 @@
>  
>  #include <asm/serial.h>
>  
> -static struct console early_con = {
> +static struct console early_con __initdata = {
>  	.name =		"uart",		/* fixed up at earlycon registration */
>  	.flags =	CON_PRINTBUFFER | CON_BOOT,
>  	.index =	0,
>  };
>  
> -static struct earlycon_device early_console_dev = {
> +static struct earlycon_device early_console_dev __initdata = {
>  	.con = &early_con,
>  };
>  
> -- 
> 2.7.4
>