From: Eugen Hristev <eugen.hristev@collabora.com>
To: Trevor Wu <trevor.wu@mediatek.com>,
pierre-louis.bossart@linux.intel.com,
peter.ujfalusi@linux.intel.com, yung-chuan.liao@linux.intel.com,
ranjani.sridharan@linux.intel.com, kai.vehmanen@linux.intel.com,
daniel.baluta@nxp.com, broonie@kernel.org, lgirdwood@gmail.com,
tiwai@suse.com, perex@perex.cz, matthias.bgg@gmail.com,
angelogioacchino.delregno@collabora.com
Cc: yc.hung@mediatek.com, allen-kh.cheng@mediatek.com,
sound-open-firmware@alsa-project.org,
alsa-devel@alsa-project.org, linux-mediatek@lists.infradead.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH] ASoC: SOF: mediatek: mt8186 modify dram type as non-cache
Date: Thu, 3 Aug 2023 11:04:03 +0300 [thread overview]
Message-ID: <e88a1074-ca86-709f-8c26-f875320e728a@collabora.com> (raw)
In-Reply-To: <20230803075028.32170-1-trevor.wu@mediatek.com>
Hi Trevor,
On 8/3/23 10:50, Trevor Wu wrote:
> To prevent incorrect access between the host and DSP sides, we need to
> modify DRAM as a non-cache memory type. Additionally, we can retrieve
> the size of shared DMA from the device tree.
>
> Signed-off-by: Trevor Wu <trevor.wu@mediatek.com>
> Reviewed-by: Yaochun Hung <yc.hung@mediatek.com>
> Reviewed-by: Kuan-Hsun Cheng <Allen-KH.Cheng@mediatek.com>
> ---
> sound/soc/sof/mediatek/mt8186/mt8186.c | 40 +++++++++++++++-----------
> 1 file changed, 23 insertions(+), 17 deletions(-)
>
> diff --git a/sound/soc/sof/mediatek/mt8186/mt8186.c b/sound/soc/sof/mediatek/mt8186/mt8186.c
> index 3e0ea0e109e2..f587edf9e0a7 100644
> --- a/sound/soc/sof/mediatek/mt8186/mt8186.c
> +++ b/sound/soc/sof/mediatek/mt8186/mt8186.c
> @@ -111,6 +111,14 @@ static int platform_parse_resource(struct platform_device *pdev, void *data)
>
> dev_dbg(dev, "DMA %pR\n", &res);
>
> + adsp->pa_shared_dram = (phys_addr_t)res.start;
> + adsp->shared_size = resource_size(&res);
> + if (adsp->pa_shared_dram & DRAM_REMAP_MASK) {
> + dev_err(dev, "adsp shared dma memory(%#x) is not 4K-aligned\n",
> + (u32)adsp->pa_shared_dram);
> + return -EINVAL;
> + }
> +
Would it be better to just realign to the next 4k boundary ?
Or, isn't it more usual to use dma_coerce_mask_and_coherent ?
> ret = of_reserved_mem_device_init(dev);
> if (ret) {
> dev_err(dev, "of_reserved_mem_device_init failed\n");
> @@ -244,23 +252,18 @@ static int adsp_shared_base_ioremap(struct platform_device *pdev, void *data)
> {
> struct device *dev = &pdev->dev;
> struct mtk_adsp_chip_info *adsp = data;
> - u32 shared_size;
>
> /* remap shared-dram base to be non-cachable */
> - shared_size = TOTAL_SIZE_SHARED_DRAM_FROM_TAIL;
> - adsp->pa_shared_dram = adsp->pa_dram + adsp->dramsize - shared_size;
> - if (adsp->va_dram) {
> - adsp->shared_dram = adsp->va_dram + DSP_DRAM_SIZE - shared_size;
> - } else {
> - adsp->shared_dram = devm_ioremap(dev, adsp->pa_shared_dram,
> - shared_size);
> - if (!adsp->shared_dram) {
> - dev_err(dev, "ioremap failed for shared DRAM\n");
> - return -ENOMEM;
> - }
> + adsp->shared_dram = devm_ioremap(dev, adsp->pa_shared_dram,
> + adsp->shared_size);
You cannot use dma_alloc_coherent ? This should take care of all the
cache maintainance for you.
> + if (!adsp->shared_dram) {
> + dev_err(dev, "failed to ioremap base %pa size %#x\n",
> + adsp->shared_dram, adsp->shared_size);
> + return -ENOMEM;
> }
> - dev_dbg(dev, "shared-dram vbase=%p, phy addr :%pa, size=%#x\n",
> - adsp->shared_dram, &adsp->pa_shared_dram, shared_size);
> +
> + dev_dbg(dev, "shared-dram vbase=%p, phy addr :%pa, size=%#x\n",
> + adsp->shared_dram, &adsp->pa_shared_dram, adsp->shared_size);
>
> return 0;
> }
> @@ -307,9 +310,12 @@ static int mt8186_dsp_probe(struct snd_sof_dev *sdev)
> return -ENOMEM;
> }
>
> - sdev->bar[SOF_FW_BLK_TYPE_SRAM] = devm_ioremap_wc(sdev->dev,
> - priv->adsp->pa_dram,
> - priv->adsp->dramsize);
> + priv->adsp->va_sram = sdev->bar[SOF_FW_BLK_TYPE_IRAM];
> +
> + sdev->bar[SOF_FW_BLK_TYPE_SRAM] = devm_ioremap(sdev->dev,
> + priv->adsp->pa_dram,
> + priv->adsp->dramsize);
> +
Same here
> if (!sdev->bar[SOF_FW_BLK_TYPE_SRAM]) {
> dev_err(sdev->dev, "failed to ioremap base %pa size %#x\n",
> &priv->adsp->pa_dram, priv->adsp->dramsize);
Regards,
Eugen
next prev parent reply other threads:[~2023-08-03 8:13 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-08-03 7:50 Trevor Wu
2023-08-03 8:04 ` Eugen Hristev [this message]
2023-08-04 10:31 ` Trevor Wu (吳文良)
2023-08-04 16:57 ` 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=e88a1074-ca86-709f-8c26-f875320e728a@collabora.com \
--to=eugen.hristev@collabora.com \
--cc=allen-kh.cheng@mediatek.com \
--cc=alsa-devel@alsa-project.org \
--cc=angelogioacchino.delregno@collabora.com \
--cc=broonie@kernel.org \
--cc=daniel.baluta@nxp.com \
--cc=kai.vehmanen@linux.intel.com \
--cc=lgirdwood@gmail.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mediatek@lists.infradead.org \
--cc=matthias.bgg@gmail.com \
--cc=perex@perex.cz \
--cc=peter.ujfalusi@linux.intel.com \
--cc=pierre-louis.bossart@linux.intel.com \
--cc=ranjani.sridharan@linux.intel.com \
--cc=sound-open-firmware@alsa-project.org \
--cc=tiwai@suse.com \
--cc=trevor.wu@mediatek.com \
--cc=yc.hung@mediatek.com \
--cc=yung-chuan.liao@linux.intel.com \
/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®