linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Srinivasa Rao Mandadapu <quic_srivasam@quicinc.com>
To: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>,
	<linux-remoteproc@vger.kernel.org>, <agross@kernel.org>,
	<bjorn.andersson@linaro.org>, <lgirdwood@gmail.com>,
	<broonie@kernel.org>, <robh+dt@kernel.org>,
	<quic_plai@quicinc.com>, <bgoswami@quicinc.com>, <perex@perex.cz>,
	<tiwai@suse.com>, <srinivas.kandagatla@linaro.org>,
	<quic_rohkumar@quicinc.com>, <linux-arm-msm@vger.kernel.org>,
	<linux-kernel@vger.kernel.org>, <swboyd@chromium.org>,
	<judyhsiao@chromium.org>, <devicetree@vger.kernel.org>
Subject: Re: [PATCH 7/8] remoteproc: qcom: Add support for memory sandbox
Date: Tue, 9 Aug 2022 13:52:26 +0530	[thread overview]
Message-ID: <1f340f3d-83f3-6455-7671-34ef40abe6c4@quicinc.com> (raw)
In-Reply-To: <9d78a571-8d02-2967-1f29-21ca737a582f@linaro.org>


On 8/7/2022 2:04 AM, Dmitry Baryshkov wrote:
Thanks for your time and Valuable inputs Dmitry!!!
> On 03/08/2022 17:21, Srinivasa Rao Mandadapu wrote:
>> Add memory sandbox support for ADSP based platforms secure booting.
>
> This repeats commit subject. Please replace it with proper commit 
> message text describing what is done and why.
Okay. Will update it.
>
>>
>> Signed-off-by: Srinivasa Rao Mandadapu <quic_srivasam@quicinc.com>
>> ---
>>   drivers/remoteproc/qcom_q6v5_adsp.c | 101 
>> +++++++++++++++++++++++++++++++++++-
>>   1 file changed, 99 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/remoteproc/qcom_q6v5_adsp.c 
>> b/drivers/remoteproc/qcom_q6v5_adsp.c
>> index 3dbd035..f81da47 100644
>> --- a/drivers/remoteproc/qcom_q6v5_adsp.c
>> +++ b/drivers/remoteproc/qcom_q6v5_adsp.c
>> @@ -9,6 +9,7 @@
>>   #include <linux/firmware.h>
>>   #include <linux/interrupt.h>
>>   #include <linux/io.h>
>> +#include <linux/iommu.h>
>>   #include <linux/iopoll.h>
>>   #include <linux/kernel.h>
>>   #include <linux/mfd/syscon.h>
>> @@ -48,6 +49,8 @@
>>   #define LPASS_PWR_ON_REG        0x10
>>   #define LPASS_HALTREQ_REG        0x0
>>   +#define SID_MASK_DEFAULT        0xF
>> +
>>   #define QDSP6SS_XO_CBCR        0x38
>>   #define QDSP6SS_CORE_CBCR    0x20
>>   #define QDSP6SS_SLEEP_CBCR    0x3c
>> @@ -77,7 +80,7 @@ struct adsp_pil_data {
>>   struct qcom_adsp {
>>       struct device *dev;
>>       struct rproc *rproc;
>> -
>> +    struct iommu_domain *iommu_dom;
>>       struct qcom_q6v5 q6v5;
>>         struct clk *xo;
>> @@ -332,6 +335,91 @@ static int adsp_load(struct rproc *rproc, const 
>> struct firmware *fw)
>>       return 0;
>>   }
>>   +static int adsp_map_smmu(struct qcom_adsp *adsp, struct rproc *rproc)
>> +{
>> +    struct of_phandle_args args;
>> +    int ret, rc, i;
>> +    long long sid;
>> +
>> +    unsigned long mem_phys;
>> +    unsigned long iova;
>> +    const __be32 *prop;
>> +    int access_level;
>> +    uint32_t len, flag, mem_size;
>> +    int offset;
>> +    struct fw_rsc_hdr *hdr;
>> +    struct fw_rsc_devmem *rsc_fw;
>> +
>> +    rc = of_parse_phandle_with_fixed_args(adsp->dev->of_node, 
>> "iommus", 1, 0, &args);
>
> Please do not add implicit dependency on #iommu-cells value.
Okay. Will change it to "of_parse_phandle_with_args()"
>
>> +    if (rc < 0)
>> +        sid = -1;
>> +    else
>> +        sid = args.args[0] & SID_MASK_DEFAULT;
>> +
>> +    adsp->iommu_dom = iommu_domain_alloc(&platform_bus_type);
>
> please use adsp->dev->bus instead of platform_bus_type here.
Okay. will update it.
>
>> +    if (!adsp->iommu_dom) {
>> +        dev_err(adsp->dev, "failed to allocate iommu domain\n");
>> +        return -ENOMEM;
>> +    }
>> +
>> +    ret = iommu_attach_device(adsp->iommu_dom, adsp->dev);
>> +    if (ret) {
>> +        dev_err(adsp->dev, "could not attach device ret = %d\n", ret);
>> +        return -EBUSY;
>> +    }
>> +
>> +    /* Add SID configuration for ADSP Firmware to SMMU */
>> +    adsp->mem_phys =  adsp->mem_phys | (sid << 32);
>> +
>> +    ret = iommu_map(adsp->iommu_dom, adsp->mem_phys, adsp->mem_phys,
>> +            adsp->mem_size,    IOMMU_READ | IOMMU_WRITE);
>> +    if (ret) {
>> +        dev_err(adsp->dev, "Unable to map ADSP Physical Memory\n");
>> +        return ret;
>> +    }
>> +
>> +    prop = of_get_property(adsp->dev->of_node, "qcom,adsp-memory", 
>> &len);
>
> Non-documented property. So, this chunk is not acceptable.
Okay. Will add it in dt-bindings too.
>
>> +    if (prop) {
>> +        len /= sizeof(__be32);
>> +        for (i = 0; i < len; i++) {
>> +            iova = be32_to_cpu(prop[i++]);
>> +            mem_phys = be32_to_cpu(prop[i++]);
>> +            mem_size = be32_to_cpu(prop[i++]);
>> +            access_level = be32_to_cpu(prop[i]);
>> +
>> +            if (access_level)
>> +                flag = IOMMU_READ | IOMMU_WRITE;
>> +            else
>> +                flag = IOMMU_READ;
>> +
>> +            ret = iommu_map(adsp->iommu_dom, iova, mem_phys, 
>> mem_size, flag);
>> +            if (ret) {
>> +                dev_err(adsp->dev, "failed to map addr = %p mem_size 
>> = %x\n",
>> +                        &(mem_phys), mem_size);
>> +                return ret;
>> +            }
>> +        }
>> +    } else {
>> +        if (!rproc->table_ptr)
>> +            return 0;
>> +
>> +        for (i = 0; i < rproc->table_ptr->num; i++) {
>> +            offset = rproc->table_ptr->offset[i];
>> +            hdr = (void *)rproc->table_ptr + offset;
>> +            rsc_fw = (struct fw_rsc_devmem *)hdr + sizeof(*hdr);
>> +
>> +            ret = iommu_map(rproc->domain, rsc_fw->da, rsc_fw->pa,
>> +                        rsc_fw->len, rsc_fw->flags);
>
> What about filling an sgtable instead and using it?

Here we are just doing IO mapping and allowing ADSP to access the 
specified memory.

I am not sure,  sg_table applicable here or not as it's not any DMA 
activity.

Please correct me if my understanding is not enough and It would help me 
a lot, if any good example shared.

>
>> +            if (ret) {
>> +                pr_err("%s; unable to map adsp memory address\n", 
>> __func__);
>> +                return ret;
>> +            }
>> +        }
>> +    }
>> +    return 0;
>> +}
>> +
>> +
>>   static int adsp_start(struct rproc *rproc)
>>   {
>>       struct qcom_adsp *adsp = (struct qcom_adsp *)rproc->priv;
>> @@ -341,7 +429,13 @@ static int adsp_start(struct rproc *rproc)
>>       ret = qcom_q6v5_prepare(&adsp->q6v5);
>>       if (ret)
>>           return ret;
>> -
>> +    if (!adsp->is_wpss) {
>> +        ret = adsp_map_smmu(adsp, rproc);
>
> Is this also applicable to cDSP? To sdm845 adsp?

It's applicable to all ADSP SoC variants. I think it's better to add 
adsp flag("is_adsp") for

distinguishing adsp use cases. Please suggest here.

>
>> +        if (ret) {
>> +            dev_err(adsp->dev, "ADSP smmu mapping failed\n");
>> +            goto adsp_smmu_unmap;
>> +        }
>> +    }
>>       ret = clk_prepare_enable(adsp->xo);
>>       if (ret)
>>           goto disable_irqs;
>> @@ -402,6 +496,9 @@ static int adsp_start(struct rproc *rproc)
>>       clk_disable_unprepare(adsp->xo);
>>   disable_irqs:
>>       qcom_q6v5_unprepare(&adsp->q6v5);
>> +adsp_smmu_unmap:
>> +    iommu_unmap(adsp->iommu_dom, adsp->mem_phys, adsp->mem_size);
>> +    iommu_domain_free(adsp->iommu_dom);
>>         return ret;
>>   }
>
>

  reply	other threads:[~2022-08-09  8:22 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-08-03 14:21 [PATCH 0/8] Update ADSP pil loader for SC7280 platform Srinivasa Rao Mandadapu
2022-08-03 14:21 ` [PATCH 1/8] dt-bindings: remoteproc: qcom: adsp: Make ADSP pil loader as generic Srinivasa Rao Mandadapu
2022-08-04 19:47   ` Caleb Connolly
2022-08-09  9:51     ` Srinivasa Rao Mandadapu
2022-08-03 14:21 ` [PATCH 2/8] dt-bindings: remoteproc: qcom: adsp: Add compatible name for SC7280 Srinivasa Rao Mandadapu
2022-08-03 20:34   ` Rob Herring
2022-08-03 20:43   ` Rob Herring
2022-08-04 10:14     ` Srinivasa Rao Mandadapu
2022-08-03 14:21 ` [PATCH 3/8] remoteproc: qcom: Add compatible name for SC7280 ADSP Srinivasa Rao Mandadapu
2022-08-06 20:57   ` Dmitry Baryshkov
2022-08-09  9:49     ` Srinivasa Rao Mandadapu
2022-08-03 14:21 ` [PATCH 4/8] remoteproc: qcom: Update hard coded values with macros Srinivasa Rao Mandadapu
2022-08-06 20:24   ` Dmitry Baryshkov
2022-08-03 14:21 ` [PATCH 5/8] remoteproc: qcom: Add efuse evb selection control Srinivasa Rao Mandadapu
2022-08-06 20:26   ` Dmitry Baryshkov
2022-08-09  6:31     ` Srinivasa Rao Mandadapu
2022-08-03 14:21 ` [PATCH 6/8] remoteproc: qcom: Add flag in adsp private data structure Srinivasa Rao Mandadapu
2022-08-03 14:21 ` [PATCH 7/8] remoteproc: qcom: Add support for memory sandbox Srinivasa Rao Mandadapu
2022-08-06 20:34   ` Dmitry Baryshkov
2022-08-09  8:22     ` Srinivasa Rao Mandadapu [this message]
2022-08-09 11:04       ` Srinivasa Rao Mandadapu
2022-08-06 20:39   ` Christophe JAILLET
2022-08-09  6:34     ` Srinivasa Rao Mandadapu
2022-08-03 14:21 ` [PATCH 8/8] remoteproc: qcom: Update QDSP6 out-of-reset timeout value Srinivasa Rao Mandadapu

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=1f340f3d-83f3-6455-7671-34ef40abe6c4@quicinc.com \
    --to=quic_srivasam@quicinc.com \
    --cc=agross@kernel.org \
    --cc=bgoswami@quicinc.com \
    --cc=bjorn.andersson@linaro.org \
    --cc=broonie@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=dmitry.baryshkov@linaro.org \
    --cc=judyhsiao@chromium.org \
    --cc=lgirdwood@gmail.com \
    --cc=linux-arm-msm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-remoteproc@vger.kernel.org \
    --cc=perex@perex.cz \
    --cc=quic_plai@quicinc.com \
    --cc=quic_rohkumar@quicinc.com \
    --cc=robh+dt@kernel.org \
    --cc=srinivas.kandagatla@linaro.org \
    --cc=swboyd@chromium.org \
    --cc=tiwai@suse.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;
as well as URLs for NNTP newsgroup(s).