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] [day] [month] [year] [list]
Date:   Thu, 16 Jun 2022 10:30:45 +0000
From:   Charles Keepax <ckeepax@...nsource.cirrus.com>
To:     Sameer Pujar <spujar@...dia.com>
CC:     <broonie@...nel.org>, <lgirdwood@...il.com>, <perex@...ex.cz>,
        <tiwai@...e.com>, <thierry.reding@...il.com>,
        <jonathanh@...dia.com>, <alsa-devel@...a-project.org>,
        <linux-tegra@...r.kernel.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] ASoC: tegra: Fix clock DAI format on Tegra210

On Wed, Jun 15, 2022 at 01:34:37PM +0530, Sameer Pujar wrote:
> I2S reset failures are seen on Tegra210 and later platforms. This indicates
> absence of I2S bit clock, which is required to perform the reset operation.
> Following failures are seen with I2S based tests on Tegra210 and later:
> 
>   tegra210-i2s 2901100.i2s: timeout: failed to reset I2S for playback
>   tegra210-i2s 2901100.i2s: ASoC: PRE_PMU: I2S2 RX event failed: -110
>   tegra210-i2s 2901100.i2s: timeout: failed to reset I2S for capture
>   tegra210-i2s 2901100.i2s: ASoC: PRE_PMU: I2S2 TX event failed: -110
> 
> The commit d92ad6633fa7 ("ASoC: tegra: Update to use set_fmt_new callback")
> regressed I2S functionality on Tegra platforms. Basically it flipped clock
> provider and consumer DAI formats. This configures Tegra I2S in consumer
> mode by default now and there is none to provide bit clock during loopback
> tests. The external codec based tests also fail because both Tegra I2S and
> codec I2S get configured in consumer mode.
> 
> ASoC core flips the DAI format before calling set_fmt() for CPU DAIs. This
> is negated in above commit. Fix this by swapping SND_SOC_DAIFMT_BC_FC and
> SND_SOC_DAIFMT_BP_FP switch cases.
> 
> Fixes: d92ad6633fa7 ("ASoC: tegra: Update to use set_fmt_new callback")
> Signed-off-by: Sameer Pujar <spujar@...dia.com>
> Cc: Charles Keepax <ckeepax@...nsource.cirrus.com>
> ---

Ah.... I see what has happened there. As this device is one of
the few CPU side components defining non_legacy_dai_naming whilst
processing the DAI links the core would have been flipping the
format for this driver as it would have mistaken it for a CODEC
to CODEC link, so unlike the other CPU side components the flags
shouldn't have been inverted when I converted this driver.

Apologies for missing this whilst doing the updates, and thank
you for fixing. I will check if there are any other cases of this
that need to be hit.

A bit late but for what it is worth:

Reviewed-by: Charles Keepax <ckeepax@...nsource.cirrus.com>

Thanks,
Charles

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ