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: <87ed5q2v9m.wl-tiwai@suse.de>
Date: Wed, 11 Sep 2024 15:08:05 +0200
From: Takashi Iwai <tiwai@...e.de>
To: Jerome Brunet <jbrunet@...libre.com>
Cc: Jaroslav Kysela <perex@...ex.cz>,	Péter Ujfalusi
 <peter.ujfalusi@...ux.intel.com>,	Pierre-Louis Bossart
 <pierre-louis.bossart@...ux.intel.com>,	Takashi Iwai <tiwai@...e.com>,
	David Rhodes <david.rhodes@...rus.com>,	Richard Fitzgerald
 <rf@...nsource.cirrus.com>,	Liam Girdwood <lgirdwood@...il.com>,	Mark Brown
 <broonie@...nel.org>,	Cezary Rojewski <cezary.rojewski@...el.com>,	Liam
 Girdwood <liam.r.girdwood@...ux.intel.com>,	Bard Liao
 <yung-chuan.liao@...ux.intel.com>,	Ranjani Sridharan
 <ranjani.sridharan@...ux.intel.com>,	Kai Vehmanen
 <kai.vehmanen@...ux.intel.com>,	Srinivas Kandagatla
 <srinivas.kandagatla@...aro.org>,	Chen-Yu Tsai <wens@...e.org>,	Jernej
 Skrabec <jernej.skrabec@...il.com>,	Samuel Holland <samuel@...lland.org>,
	linux-sound@...r.kernel.org,	linux-kernel@...r.kernel.org,
	patches@...nsource.cirrus.com,	alsa-devel@...a-project.org,
	linux-arm-msm@...r.kernel.org,	linux-arm-kernel@...ts.infradead.org,
	linux-sunxi@...ts.linux.dev
Subject: Re: [PATCH 01/13] ALSA: pcm: add more sample rate definitions

On Wed, 11 Sep 2024 14:59:39 +0200,
Jerome Brunet wrote:
> 
> On Wed 11 Sep 2024 at 14:42, Takashi Iwai <tiwai@...e.de> wrote:
> 
> > On Wed, 11 Sep 2024 12:58:53 +0200,
> > Jaroslav Kysela wrote:
> >> 
> >> On 11. 09. 24 12:51, Takashi Iwai wrote:
> >> > On Wed, 11 Sep 2024 12:33:01 +0200,
> >> > Péter Ujfalusi wrote:
> >> >> 
> >> >> On 11/09/2024 12:21, Takashi Iwai wrote:
> >> >>>> Wondering if this is backwards compatible with the alsa-lib definitions,
> >> >>>> specifically the topology parts which did unfortunately have a list of
> >> >>>> rates that will map to a different index now:
> >> >>>> 
> >> >>>> 
> >> >>>> typedef enum _snd_pcm_rates {
> >> >>>> 	SND_PCM_RATE_UNKNOWN = -1,
> >> >>>> 	SND_PCM_RATE_5512 = 0,
> >> >>>> 	SND_PCM_RATE_8000,
> >> >>>> 	SND_PCM_RATE_11025,
> >> >>>> 	SND_PCM_RATE_16000,
> >> >>>> 	SND_PCM_RATE_22050,
> >> >>>> 	SND_PCM_RATE_32000,
> >> >>>> 	SND_PCM_RATE_44100,
> >> >>>> 	SND_PCM_RATE_48000,
> >> >>>> 	SND_PCM_RATE_64000,
> >> >>>> 	SND_PCM_RATE_88200,
> >> >>>> 	SND_PCM_RATE_96000,
> >> >>>> 	SND_PCM_RATE_176400,
> >> >>>> 	SND_PCM_RATE_192000,
> >> >>>> 	SND_PCM_RATE_CONTINUOUS = 30,
> >> >>>> 	SND_PCM_RATE_KNOT = 31,
> >> >>>> 	SND_PCM_RATE_LAST = SND_PCM_RATE_KNOT,
> >> >>>> } snd_pcm_rates_t;
> >> >>> 
> >> >>> As far as I understand correctly, those rate bits used for topology
> >> >>> are independent from the bits used for PCM core, although it used to
> >> >>> be the same.  Maybe better to rename (such as SND_TPLG_RATE_*) so that
> >> >>> it's clearer only for topology stuff.
> >> >> 
> >> >> Even if we rename these in alsa-lib we will need translation from
> >> >> SND_TPLG_RATE_ to SND_PCM_RATE_ in kernel likely?
> >> >> 
> >> >> The topology files are out there and this is an ABI...
> >> >> 
> >> >>> But it'd be better if anyone can double-check.
> >> >> 
> >> >> Since the kernel just copies the rates bitfield, any rate above 11025
> >> >> will be misaligned and result broken setup.
> >> > 
> >> > Yep, I noticed it now, too.
> >> > 
> >> > Below is the fix patch, totally untested.
> >> > It'd be appreciated if anyone can test it quickly.
> >> > 
> >> > 
> >> > thanks,
> >> > 
> >> > Takashi
> >> > 
> >> > -- 8< --
> >> > From: Takashi Iwai <tiwai@...e.de>
> >> > Subject: [PATCH] ALSA: pcm: Fix breakage of PCM rates used for topology
> >> > 
> >> > It turned out that the topology ABI takes the standard PCM rate bits
> >> > as is, and it means that the recent change of the PCM rate bits would
> >> > lead to the inconsistent rate values used for topology.
> >> > 
> >> > This patch reverts the original PCM rate bit definitions while adding
> >> > the new rates to the extended bits instead.  This needed the change of
> >> > snd_pcm_known_rates, too.  And this also required to fix the handling
> >> > in snd_pcm_hw_limit_rates() that blindly assumed that the list is
> >> > sorted while it became unsorted now.
> >> > 
> >> > Fixes: 090624b7dc83 ("ALSA: pcm: add more sample rate definitions")
> >> > Signed-off-by: Takashi Iwai <tiwai@...e.de>
> >> 
> >> This looks fine. But the topology rate bits should not depend on those
> >> bits. It's a bit pitty that the standard PCM ABI does not use those
> >> bits for user space and we are doing this change just for topology
> >> ABI.
> >
> > Yeah, and theoretically it's possible to fix in topology side, but
> > it'll be more cumbersome.
> >
> > Although it's not really a part of PCM ABI, I believe we should move
> > the PCM rate bit definitions to uapi, for showing that it's set in
> > stone.  Something like below.
> >
> >
> > Takashi
> >
> > -- 8< --
> > From: Takashi Iwai <tiwai@...e.de>
> > Subject: [PATCH] ALSA: pcm: Move standard rate bit definitions into uapi
> >
> > Since the standard PCM rate bits are used for the topology ABI, it's a
> > part of public ABI that must not be changed.  Move the definitions
> > into uapi to indicate it more clearly.
> >
> > Signed-off-by: Takashi Iwai <tiwai@...e.de>
> > ---
> >  include/sound/pcm.h         | 26 --------------------------
> >  include/uapi/sound/asound.h | 26 ++++++++++++++++++++++++++
> >  2 files changed, 26 insertions(+), 26 deletions(-)
> >
> > diff --git a/include/sound/pcm.h b/include/sound/pcm.h
> > index 824216799070..f28f6d6ac996 100644
> > --- a/include/sound/pcm.h
> > +++ b/include/sound/pcm.h
> > @@ -105,32 +105,6 @@ struct snd_pcm_ops {
> >  
> >  #define SNDRV_PCM_POS_XRUN		((snd_pcm_uframes_t)-1)
> >  
> > -/* If you change this don't forget to change rates[] table in pcm_native.c */
> > -#define SNDRV_PCM_RATE_5512		(1U<<0)		/* 5512Hz */
> > -#define SNDRV_PCM_RATE_8000		(1U<<1)		/* 8000Hz */
> > -#define SNDRV_PCM_RATE_11025		(1U<<2)		/* 11025Hz */
> > -#define SNDRV_PCM_RATE_16000		(1U<<3)		/* 16000Hz */
> > -#define SNDRV_PCM_RATE_22050		(1U<<4)		/* 22050Hz */
> > -#define SNDRV_PCM_RATE_32000		(1U<<5)		/* 32000Hz */
> > -#define SNDRV_PCM_RATE_44100		(1U<<6)		/* 44100Hz */
> > -#define SNDRV_PCM_RATE_48000		(1U<<7)		/* 48000Hz */
> > -#define SNDRV_PCM_RATE_64000		(1U<<8)		/* 64000Hz */
> > -#define SNDRV_PCM_RATE_88200		(1U<<9)		/* 88200Hz */
> > -#define SNDRV_PCM_RATE_96000		(1U<<10)	/* 96000Hz */
> > -#define SNDRV_PCM_RATE_176400		(1U<<11)	/* 176400Hz */
> > -#define SNDRV_PCM_RATE_192000		(1U<<12)	/* 192000Hz */
> > -#define SNDRV_PCM_RATE_352800		(1U<<13)	/* 352800Hz */
> > -#define SNDRV_PCM_RATE_384000		(1U<<14)	/* 384000Hz */
> > -#define SNDRV_PCM_RATE_705600		(1U<<15)	/* 705600Hz */
> > -#define SNDRV_PCM_RATE_768000		(1U<<16)	/* 768000Hz */
> > -/* extended rates */
> 
> It is cosmetic but I wonder, does the extended really start here ?

Maybe a bad choice of the words.  This was rather meant as the
extension since 6.12.  So I'll replace it with "extended rates since
6.12", to be clearer.

> From the table Pierre-Louis sent, I suppose that all the recently added rates could
> been seen as extended too (352.8 to 768). Those did not mess with the
> order though 

AFAIU, the topology stuff seems supporting only up to 192kHz for now,
but it's a matter of topology-only; the limitation could be commented
in somewhere in topology's headers, but it's basically independent
from the definitions in pcm.h.


thanks,

Takashi

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ