linux-kernel - Re: [PATCH v3] ASoC: Intel: kbl_rt5663_rt5514_max98927: Split be_hw_params

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: <CAFJ_xbq3DU8NMmymYHLjy0z+QbGRAOmB4NaiXPphAr2Zu0MPAg@mail.gmail.com>
Date:   Mon, 29 Jun 2020 20:19:38 +0200
From:   Łukasz Majczak <lma@...ihalf.com>
To:     "N, Harshapriya" <harshapriya.n@...el.com>
Cc:     "alsa-devel@...a-project.org" <alsa-devel@...a-project.org>,
        Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>,
        Jie Yang <yang.jie@...ux.intel.com>,
        Radoslaw Biernacki <rad@...ihalf.com>,
        Ross Zwisler <zwisler@...gle.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Liam Girdwood <liam.r.girdwood@...ux.intel.com>,
        Bob Brandt <brndt@...gle.com>, Marcin Wojtas <mw@...ihalf.com>,
        Alex Levin <levinale@...omium.org>,
        "M R, Sathya Prakash" <sathya.prakash.m.r@...el.com>
Subject: Re: [PATCH v3] ASoC: Intel: kbl_rt5663_rt5514_max98927: Split
 be_hw_params_fixup function

Hi Harsha,

I've put the reason for this change in the commit message.
I had to split be_hw_params_fixup function for different codecs
because current approach (made for kernel 4.4) used in kernel 5.4,
leads to crash while trying to get snd_soc_dpcm with container_of()
macro in kabylake_ssp_fixup().
The crash call path looks as below:
soc_pcm_hw_params()
snd_soc_dai_hw_params(codec_dai, substream, &codec_params);
rtd->dai_link->be_hw_params_fixup(rtd, params)
kabylake_ssp_fixup()
In this case, codec_params is just a copy of an internal structure and is
not embedded into struct snd_soc_dpcm thus we cannot use
container_of() on it.

Best regards,
Lukasz

pon., 29 cze 2020 o 18:51 N, Harshapriya <harshapriya.n@...el.com> napisał(a):
>
> > -----Original Message-----
> > From: Łukasz Majczak <lma@...ihalf.com>
> > Sent: Monday, June 29, 2020 4:11 AM
> > To: N, Harshapriya <harshapriya.n@...el.com>
> > Cc: alsa-devel@...a-project.org; Pierre-Louis Bossart <pierre-
> > louis.bossart@...ux.intel.com>; Jie Yang <yang.jie@...ux.intel.com>; Radoslaw
> > Biernacki <rad@...ihalf.com>; Ross Zwisler <zwisler@...gle.com>; linux-
> > kernel@...r.kernel.org; Liam Girdwood <liam.r.girdwood@...ux.intel.com>;
> > Bob Brandt <brndt@...gle.com>; Marcin Wojtas <mw@...ihalf.com>; Alex
> > Levin <levinale@...omium.org>
> > Subject: Re: [PATCH v3] ASoC: Intel: kbl_rt5663_rt5514_max98927: Split
> > be_hw_params_fixup function
> >
> > Hi Harsha,
> >
> > We would like to continue the work on this, could you please suggest the
> > correct approach.
> >
> > Best regards,
> > Lukasz
> >
> > czw., 21 maj 2020 o 20:10 Pierre-Louis Bossart <pierre-
> > louis.bossart@...ux.intel.com> napisał(a):
> > >
> > >
> > >
> > > On 5/21/20 12:30 PM, Łukasz Majczak wrote:
> > > > Hi Pierre
> > > >
> > > > If you will take a look at the original kabylake_ssp_fixup() you
> > > > will see that it is checking whether the related FE is "Kbl Audio
> > > > Port", "Kbl Audio Headset Playback", "Kbl Audio Capture Port" or
> > > > "Kbl Audio DMIC cap" - then for the first 3 cases it sets min/max
> > > > channels to 2 while for the "Kbl DMIC cap" it can be 2 or 4, that's
> > > > is why I'm trying to split this, but maybe I'm missing here something.
> > >
> > > I don't understand this code either.
> > >
> > > I believe the intent is that for all SSP1-RT5663 usages, we should use
> > >
> > >                 rate->min = rate->max = 48000;
> > >                 chan->min = chan->max = 2;
> > >                 snd_mask_none(fmt);
> > >                 snd_mask_set_format(fmt, SNDRV_PCM_FORMAT_S24_LE);
> > >
> > > That is pretty easy to move to a dedicated ssp1 fixup.
> > >
> > > for SSP0, we have RT5514 for capture and max98927 for playback, but
> > > the existing code does not explicitly deal with rate/channels/format
> > > for all cases, so it's not clear what should happen.
> > >
> > > Harsha, can you help here?
> Apologies for missing the email I had to respond to.
>
> SSP0 - has the speakers
> SSP1 - has headset and dmic
> For speakers and headsets its 48Khz, 2 ch and 24 bit and this setting is done based on the front-end dai
> For speakers, however support only 16 bit, so we set it back to 16 bit
> If the front end dai is dmic, then the channel is set to either 2 or 4 dmic_constraints. No other formats need to be set.
>
> All the SSP1 usages do not have the same parameters (as dmic is on SSP1 and its different as given above)
> Most parameters are same for speakers and headset which are on different SSP. This is the reason we had a single fixup function.
>
> Is there a reason why the fixup function needs to be split?
>
> > >
> > > >
> > > > Best regards,
> > > > Lukasz
> > > >
> > > > czw., 21 maj 2020 o 19:17 Pierre-Louis Bossart
> > > > <pierre-louis.bossart@...ux.intel.com> napisał(a):
> > > >>
> > > >>
> > > >>
> > > >> On 5/21/20 12:08 PM, Łukasz Majczak wrote:
> > > >>>>
> > > >>>> don't add a new dailink, this is not right.
> > > >>>>
> > > >>> Can you advise a better solution how to assign different fixup
> > > >>> functions to mic and to speakers? I was looking at "dmic01"
> > > >>> dailink in skl_nau88l25_max98357a.c as an example.
> > > >>
> > > >> I am not sure I follow. the DMICs are handled on a shared SSP, so
> > > >> how would one set a different fixup? The word length have to be the same.