mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Srinivas Kandagatla <srinivas.kandagatla@oss.qualcomm.com>
To: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Cc: broonie@kernel.org, jens.glathe@oldschoolsolutions.biz,
	linux-sound@vger.kernel.org, lgirdwood@gmail.com, perex@perex.cz,
	tiwai@suse.com, johan@kernel.org, konrad.dybcio@oss.qualcomm.com,
	linux-arm-msm@vger.kernel.org, linux-kernel@vger.kernel.org,
	val@packett.cool, mailingradian@gmail.com,
	Stable@vger.kernel.org
Subject: Re: [PATCH] ASoC: qcom: q6apm-dai: Allocate an extra page for PCM buffers
Date: Thu, 14 May 2026 10:55:15 +0000	[thread overview]
Message-ID: <c6be57b9-ce26-4e96-b029-b18798904e2f@oss.qualcomm.com> (raw)
In-Reply-To: <2pe7rk7jhc36osc6i4rxeyw342mvza2m7i4ztsmm6pjgwtlemc@k4gkw5b4jg7g>

On 5/14/26 9:22 AM, Dmitry Baryshkov wrote:
> On Thu, May 14, 2026 at 09:06:07AM +0000, Srinivas Kandagatla wrote:
>> Some Old DSP firmware versions use 32-bit address arithmetic and size for
>> validating the PCM buffer address range. If a buffer is allocated near
>> the top of the 32-bit address space, arithmetic calculations involving
>> the end address can overflow and fail checks.
> 
> Should we limit the workaround to those platforms only?
I would love to do that, but I have no idea which platforms or firmware
versions have this bug.


--srini
> 
>>
>> Work around this by increasing the preallocated PCM buffer size by one
>> page. The DSP is still passed the usable buffer size, excluding the extra
>> page, which prevents the firmware from seeing an end address that crosses
>> the 32-bit boundary.
>>
>> This was not hit before because PCM buffer allocation and DSP-side
>> mapping happened at different points, and the size mapped on the DSP was
>> usually nperiods * period_size. Therefore the mapped size was unlikely to
>> match the full preallocated buffer size exactly, although the issue was
>> still possible. With early buffer mapping on the DSP, the full
>> preallocated buffer is mapped during PCM creation, making the failure
>> reproducible at boot.
>>
>> Fixes: 8ea6e25c8536 ("ASoC: qcom: q6apm: Add support for early buffer mapping on DSP")
>> Cc: Stable@vger.kernel.org
>> Reported-by: Jens Glathe <jens.glathe@oldschoolsolutions.biz>
>> Closes: https://lore.kernel.org/all/7f10abbd-fb78-4c3a-ab90-7ca78239891a@oldschoolsolutions.biz/
>> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@oss.qualcomm.com>
>> ---
>>  sound/soc/qcom/qdsp6/q6apm-dai.c | 7 ++++++-
>>  1 file changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/sound/soc/qcom/qdsp6/q6apm-dai.c b/sound/soc/qcom/qdsp6/q6apm-dai.c
>> index ede19fdea6e9..3a1be41df096 100644
>> --- a/sound/soc/qcom/qdsp6/q6apm-dai.c
>> +++ b/sound/soc/qcom/qdsp6/q6apm-dai.c
>> @@ -497,7 +497,12 @@ static int q6apm_dai_pcm_new(struct snd_soc_component *component, struct snd_soc
>>  {
>>  	struct snd_soc_dai *cpu_dai = snd_soc_rtd_to_cpu(rtd, 0);
>>  	struct snd_pcm *pcm = rtd->pcm;
>> -	int size = BUFFER_BYTES_MAX;
>> +	/*
>> +	 * Allocate one extra page as a workaround for a DSP bug where 32-bit
>> +	 * address arithmetic can overflow when the buffer is placed near the
>> +	 * end of the addressable range.
>> +	 */
>> +	int size = BUFFER_BYTES_MAX + PAGE_SIZE;
>>  	int graph_id, ret;
>>  	struct snd_pcm_substream *substream;
>>  
>> -- 
>> 2.47.3
>>
> 


  reply	other threads:[~2026-05-14 10:55 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-14  9:06 Srinivas Kandagatla
2026-05-14  9:22 ` Dmitry Baryshkov
2026-05-14 10:55   ` Srinivas Kandagatla [this message]
2026-05-15 18:34 ` Jens Glathe
2026-05-16  2:14 ` Mark Brown

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=c6be57b9-ce26-4e96-b029-b18798904e2f@oss.qualcomm.com \
    --to=srinivas.kandagatla@oss.qualcomm.com \
    --cc=Stable@vger.kernel.org \
    --cc=broonie@kernel.org \
    --cc=dmitry.baryshkov@oss.qualcomm.com \
    --cc=jens.glathe@oldschoolsolutions.biz \
    --cc=johan@kernel.org \
    --cc=konrad.dybcio@oss.qualcomm.com \
    --cc=lgirdwood@gmail.com \
    --cc=linux-arm-msm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-sound@vger.kernel.org \
    --cc=mailingradian@gmail.com \
    --cc=perex@perex.cz \
    --cc=tiwai@suse.com \
    --cc=val@packett.cool \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®