linux-kernel - Re: [PATCH] hw_random: atmel-rng: fix race condition leading to repeated bits

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: <4FBF5916.4070705@atmel.com>
Date:	Fri, 25 May 2012 12:04:06 +0200
From:	Nicolas Ferre <nicolas.ferre@...el.com>
To:	Peter Korsgaard <jacmet@...site.dk>, <herbert@...dor.apana.org.au>
CC:	<jamie@...ieiles.com>, <linux-kernel@...r.kernel.org>,
	<GPontis@...com>, <stable@...r.kernel.org>
Subject: Re: [PATCH] hw_random: atmel-rng: fix race condition leading to repeated
 bits

On 05/25/2012 11:12 AM, Peter Korsgaard :
> Data valid gets cleared by reading the ISR (status register) and NOT from
> reading ODATA (data register). A new data word can become available between
> checking ISR and reading ODATA, causing us to reuse the same data word next
> time atmel_trng_read() gets called, if that happens before the following
> data word is ready.

Hi Peter, thanks for finding this...

> With this fixed, rngtest no longer complains of 'Continous run' errors.
> Before:
> 
> rngtest -c 1000 < /dev/hwrng
> rngtest 3
> Copyright (c) 2004 by Henrique de Moraes Holschuh
> This is free software; see the source for copying conditions.  There is NO warr.
> 
> rngtest: starting FIPS tests...
> rngtest: bits received from input: 20000032
> rngtest: FIPS 140-2 successes: 923
> rngtest: FIPS 140-2 failures: 77
> rngtest: FIPS 140-2(2001-10-10) Monobit: 0
> rngtest: FIPS 140-2(2001-10-10) Poker: 0
> rngtest: FIPS 140-2(2001-10-10) Runs: 1
> rngtest: FIPS 140-2(2001-10-10) Long run: 0
> rngtest: FIPS 140-2(2001-10-10) Continuous run: 76
> rngtest: input channel speed: (min=721.402; avg=46003.510; max=49321.338)Kibitss
> rngtest: FIPS tests speed: (min=11.442; avg=12.714; max=12.801)Mibits/s
> rngtest: Program run time: 1931860 microseconds
> 
> After:
> 
> rngtest -c 1000 < /dev/hwrng
> rngtest 3
> Copyright (c) 2004 by Henrique de Moraes Holschuh
> This is free software; see the source for copying conditions.  There is NO warr.
> 
> rngtest: starting FIPS tests...
> rngtest: bits received from input: 20000032
> rngtest: FIPS 140-2 successes: 1000
> rngtest: FIPS 140-2 failures: 0
> rngtest: FIPS 140-2(2001-10-10) Monobit: 0
> rngtest: FIPS 140-2(2001-10-10) Poker: 0
> rngtest: FIPS 140-2(2001-10-10) Runs: 0
> rngtest: FIPS 140-2(2001-10-10) Long run: 0
> rngtest: FIPS 140-2(2001-10-10) Continuous run: 0
> rngtest: input channel speed: (min=777.518; avg=36988.482; max=43115.342)Kibitss
> rngtest: FIPS tests speed: (min=11.951; avg=12.715; max=12.887)Mibits/s
> rngtest: Program run time: 2035543 microseconds
> 
> Cc: stable@...r.kernel.org
> Signed-off-by: Peter Korsgaard <jacmet@...site.dk>
> Reported-by: George Pontis <GPontis@...com>
> ---
>  drivers/char/hw_random/atmel-rng.c |    7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/drivers/char/hw_random/atmel-rng.c b/drivers/char/hw_random/atmel-rng.c
> index d7ab920..731c904 100644
> --- a/drivers/char/hw_random/atmel-rng.c
> +++ b/drivers/char/hw_random/atmel-rng.c
> @@ -36,6 +36,13 @@ static int atmel_trng_read(struct hwrng *rng, void *buf, size_t max,
>  	/* data ready? */
>  	if (readl(trng->base + TRNG_ISR) & 1) {
>  		*data = readl(trng->base + TRNG_ODATA);
> +		/*
> +		  ensure data ready is only set again AFTER the next data
> +		  word is ready in case it got set between checking ISR
> +		  and reading ODATA, so we don't risk re-reading the
> +		  same word
> +		*/
> +		readl(trng->base + TRNG_ISR);
>  		return 4;
>  	} else
>  		return 0;

What about a single read to ISR like this:

tmp = readl(trng->base + TRNG_ODATA);
if (readl(trng->base + TRNG_ISR) & 1) {
	*data = tmp;
	return 4;
} else {
  	return 0;
}

But it is true that there is always 2 reads in case of data not
available, instead of just one: I cannot figure out which solution is
the fastest: your thoughts?

Bye,
-- 
Nicolas Ferre
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/