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  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:   Wed, 5 Jun 2019 10:13:34 -0500
From:   Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
To:     Amadeusz Sławiński 
        <amadeuszx.slawinski@...ux.intel.com>, alsa-devel@...a-project.org
Cc:     Cezary Rojewski <cezary.rojewski@...el.com>,
        linux-kernel@...r.kernel.org, Takashi Iwai <tiwai@...e.com>,
        Jie Yang <yang.jie@...ux.intel.com>,
        Liam Girdwood <lgirdwood@...il.com>,
        Mark Brown <broonie@...nel.org>,
        Amadeusz Sławiński 
        <amadeuszx.slawinski@...el.com>
Subject: Re: [alsa-devel] [PATCH 03/14] ALSA: hdac: Fix codec name after
 machine driver is unloaded and reloaded

On 6/5/19 8:45 AM, Amadeusz Sławiński wrote:
> From: Amadeusz Sławiński <amadeuszx.slawinski@...el.com>
> 
> This resets internal index used for enumarating codecs. This will only
> work on assumption that platform has one codec. Anyway if there is more,
> it won't work with current machine drivers, because we can't guarantee
> order in which they are enumerated. This workarounds the fact that most
> intel machine drivers have the following defined:
> .codec_name = "ehdaudio0D2",
> However when we unload and reload machine driver idx gets incremented,
> so .codec_name would've needed to be set to ehdaudio1D2 on first reload
> and so on.
> 
> Signed-off-by: Amadeusz Sławiński <amadeuszx.slawinski@...el.com>
> ---
>   sound/hda/ext/hdac_ext_bus.c | 11 ++++++++++-
>   1 file changed, 10 insertions(+), 1 deletion(-)
> 
> diff --git a/sound/hda/ext/hdac_ext_bus.c b/sound/hda/ext/hdac_ext_bus.c
> index f33ba58b753c..c84d69c2eba4 100644
> --- a/sound/hda/ext/hdac_ext_bus.c
> +++ b/sound/hda/ext/hdac_ext_bus.c
> @@ -77,6 +77,8 @@ static const struct hdac_io_ops hdac_ext_default_io = {
>   	.dma_free_pages = hdac_ext_dma_free_pages,
>   };
>   
> +static int idx;
> +
>   /**
>    * snd_hdac_ext_bus_init - initialize a HD-audio extended bus
>    * @ebus: the pointer to extended bus object
> @@ -93,7 +95,6 @@ int snd_hdac_ext_bus_init(struct hdac_bus *bus, struct device *dev,
>   			const struct hdac_ext_bus_ops *ext_ops)
>   {
>   	int ret;
> -	static int idx;
>   
>   	/* check if io ops are provided, if not load the defaults */
>   	if (io_ops == NULL)
> @@ -118,6 +119,14 @@ EXPORT_SYMBOL_GPL(snd_hdac_ext_bus_init);
>   void snd_hdac_ext_bus_exit(struct hdac_bus *bus)
>   {
>   	snd_hdac_bus_exit(bus);
> +	/* FIXME: this is workaround
> +	 * reset index used for bus->idx, because machine drivers expect
> +	 * the codec name to be ehdaudio0D2, where 0 is bus->idx
> +	 * we only perform reset if there is one used device, if there is more
> +	 * all bets are off
> +	 */
> +	if (idx == 1)
> +		idx = 0;

The real fix would be to stop incrementing idx in snd_hdac_ext_bus_init, 
which would make sense only if we had multiple controllers. SOF pegged 
bus->idx to zero.


>   	WARN_ON(!list_empty(&bus->hlink_list));
>   }
>   EXPORT_SYMBOL_GPL(snd_hdac_ext_bus_exit);
> 

Powered by blists - more mailing lists