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-next>] [day] [month] [year] [list]
Message-Id: <1606907286-32104-1-git-send-email-mikhail_durnev@mentor.com>
Date:   Wed,  2 Dec 2020 21:08:06 +1000
From:   mdurnev@...il.com
To:     kuninori.morimoto.gx@...esas.com, alsa-devel@...a-project.org,
        linux-kernel@...r.kernel.org
Cc:     mikhail_durnev@...tor.com
Subject: [PATCH v0] ASoC: rsnd: core: Check convert rate in rsnd_hw_params

From: Mikhail Durnev <mikhail_durnev@...tor.com>

snd_pcm_hw_params_set_rate_near can return incorrect sample rate in
some cases, e.g. when the backend output rate is set to some value higher
than 48000 Hz and the input rate is 8000 Hz. So passing the value returned
by snd_pcm_hw_params_set_rate_near to snd_pcm_hw_params will result in
"FSO/FSI ratio error" and playing no audio at all while the userland
is not properly notified about the issue.

If SRC is unable to convert the requested sample rate to the sample rate
the backend is using, then the requested sample rate should be adjusted in
rsnd_hw_params. The userland will be notified about that change in the
returned hw_params structure.

Signed-off-by: Mikhail Durnev <mikhail_durnev@...tor.com>
---
 sound/soc/sh/rcar/core.c | 67 +++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 66 insertions(+), 1 deletion(-)

diff --git a/sound/soc/sh/rcar/core.c b/sound/soc/sh/rcar/core.c
index 6e670b3..8ca3fb7 100644
--- a/sound/soc/sh/rcar/core.c
+++ b/sound/soc/sh/rcar/core.c
@@ -1428,8 +1428,73 @@ static int rsnd_hw_params(struct snd_soc_component *component,
 		}
 		if (io->converted_chan)
 			dev_dbg(dev, "convert channels = %d\n", io->converted_chan);
-		if (io->converted_rate)
+		if (io->converted_rate) {
 			dev_dbg(dev, "convert rate     = %d\n", io->converted_rate);
+
+			/*
+			 * SRC supports convert rates from params_rate(hw_params)/k_down
+			 * to params_rate(hw_params)*k_up, where k_up is always 6, and
+			 * k_down depends on number of channels and SRC unit.
+			 * So all SRC units can upsample audio up to 6 times regardless
+			 * its number of channels. And all SRC units can downsample
+			 * 2 channel audio up to 6 times too.
+			 */
+			int k_up = 6;
+			int k_down = 6;
+			int channel;
+			struct rsnd_mod *src_mod = rsnd_io_to_mod_src(io);
+
+			channel = io->converted_chan ? io->converted_chan : fe_channel;
+
+			switch (rsnd_mod_id(src_mod)) {
+			/*
+			 * SRC0 can downsample 4, 6 and 8 channel audio up to 4 times.
+			 * SRC1, SRC3 and SRC4 can downsample 4 channel audio
+			 * up to 4 times.
+			 * SRC1, SRC3 and SRC4 can downsample 6 and 8 channel audio
+			 * no more than twice.
+			 */
+			case 1:
+			case 3:
+			case 4:
+				if (channel > 4) {
+					k_down = 2;
+					break;
+				}
+			case 0:
+				if (channel > 2)
+					k_down = 4;
+				break;
+
+			/* Other SRC units do not support more than 2 channels */
+			default:
+				if (channel > 2)
+					return -EINVAL;
+			}
+
+			if (params_rate(hw_params) > io->converted_rate * k_down) {
+				hw_param_interval(hw_params, SNDRV_PCM_HW_PARAM_RATE)->min =
+					io->converted_rate * k_down;
+				hw_param_interval(hw_params, SNDRV_PCM_HW_PARAM_RATE)->max =
+					io->converted_rate * k_down;
+				hw_params->cmask |= SNDRV_PCM_HW_PARAM_RATE;
+			} else if (params_rate(hw_params) * k_up < io->converted_rate) {
+				hw_param_interval(hw_params, SNDRV_PCM_HW_PARAM_RATE)->min =
+					(io->converted_rate + k_up - 1) / k_up;
+				hw_param_interval(hw_params, SNDRV_PCM_HW_PARAM_RATE)->max =
+					(io->converted_rate + k_up - 1) / k_up;
+				hw_params->cmask |= SNDRV_PCM_HW_PARAM_RATE;
+			}
+
+			/*
+			 * TBD: Max SRC input and output rates also depend on number
+			 * of channels and SRC unit:
+			 * SRC1, SRC3 and SRC4 do not support more than 128kHz
+			 * for 6 channel and 96kHz for 8 channel audio.
+			 * Perhaps this function should return EINVAL if the input or
+			 * the output rate exceeds the limitation.
+			 */
+		}
 	}
 
 	return rsnd_dai_call(hw_params, io, substream, hw_params);
-- 
2.7.4

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ