linux-kernel - Re: [PATCH 3/3] [media] em28xx: fix Terratec Grabby AC97 codec detection

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]

Date:	Thu, 3 Mar 2016 14:26:21 -0300
From:	Mauro Carvalho Chehab <mchehab@....samsung.com>
To:	Matthieu Rogez <matthieu.rogez@...il.com>
Cc:	linux-media@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3/3] [media] em28xx: fix Terratec Grabby AC97 codec
 detection

Em Sun, 28 Feb 2016 12:26:23 +0100
Matthieu Rogez <matthieu.rogez@...il.com> escreveu:

> EMP202 chip inside Terratec Grabby (hw rev 2) seems to require some time before
> accessing reliably its registers. Otherwise it returns some values previously
> put on the I2C bus.
> 
> To account for that period, we delay card setup until we have a proof that
> accessing AC97 registers is reliable. We get this proof by polling AC97_RESET
> until the expected value is read.
> 
> Signed-off-by: Matthieu Rogez <matthieu.rogez@...il.com>
> ---
>  drivers/media/usb/em28xx/em28xx-cards.c | 26 ++++++++++++++++++++++++++
>  1 file changed, 26 insertions(+)
> 
> diff --git a/drivers/media/usb/em28xx/em28xx-cards.c b/drivers/media/usb/em28xx/em28xx-cards.c
> index 5e127e4..2e04902 100644
> --- a/drivers/media/usb/em28xx/em28xx-cards.c
> +++ b/drivers/media/usb/em28xx/em28xx-cards.c
> @@ -37,6 +37,7 @@
>  #include <media/i2c-addr.h>
>  #include <media/tveeprom.h>
>  #include <media/v4l2-common.h>
> +#include <sound/ac97_codec.h>
>  
>  #include "em28xx.h"
>  
> @@ -2563,6 +2564,24 @@ static inline void em28xx_set_model(struct em28xx *dev)
>  	dev->def_i2c_bus = dev->board.def_i2c_bus;
>  }
>  
> +/* Wait until AC97_RESET reports the expected value (or reading error)
> + * before proceeding.
> + * This can help ensuring AC97 register accesses are reliable.
> + */
> +static int em28xx_wait_until_ac97_features_equals(struct em28xx *dev,
> +						  int expected_feat)
> +{
> +	int feat;
> +
> +	do {
> +		feat = em28xx_read_ac97(dev, AC97_RESET);
> +		if (feat < 0)
> +			return feat;
> +	} while (feat != expected_feat);

Please add some sort of timeout here... We don't want the Kernel to
be in a dead lock here...

> +
> +	return 0;
> +}
> +
>  /* Since em28xx_pre_card_setup() requires a proper dev->model,
>   * this won't work for boards with generic PCI IDs
>   */
> @@ -2668,6 +2687,13 @@ static void em28xx_pre_card_setup(struct em28xx *dev)
>  		em28xx_write_reg(dev, EM2820_R08_GPIO_CTRL, 0xfd);
>  		msleep(70);
>  		break;
> +
> +	case EM2860_BOARD_TERRATEC_GRABBY:
> +		/* HACK?: Ensure AC97 register reading is reliable before
> +		 * proceeding. In practice, this will wait about 1.6 seconds.
> +		 */
> +		em28xx_wait_until_ac97_features_equals(dev, 0x6a90);
> +		break;
>  	}
>  
>  	em28xx_gpio_set(dev, dev->board.tuner_gpio);


-- 
Thanks,
Mauro