linux-kernel - Re: [PATCH V4 2/5] mailbox: Add support for QTI CPUCP mailbox controller

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] [thread-next>] [day] [month] [year] [list]

Date: Tue, 23 Apr 2024 22:40:22 +0530
From: Sibi Sankar <quic_sibis@...cinc.com>
To: Konrad Dybcio <konrad.dybcio@...aro.org>, <sudeep.holla@....com>,
        <cristian.marussi@....com>, <andersson@...nel.org>,
        <jassisinghbrar@...il.com>, <robh+dt@...nel.org>,
        <krzysztof.kozlowski+dt@...aro.org>, <dmitry.baryshkov@...aro.org>
CC: <linux-kernel@...r.kernel.org>, <linux-arm-msm@...r.kernel.org>,
        <devicetree@...r.kernel.org>, <quic_rgottimu@...cinc.com>,
        <quic_kshivnan@...cinc.com>, <conor+dt@...nel.org>,
        <quic_gkohli@...cinc.com>, <quic_nkela@...cinc.com>,
        <quic_psodagud@...cinc.com>, <abel.vesa@...aro.org>
Subject: Re: [PATCH V4 2/5] mailbox: Add support for QTI CPUCP mailbox
 controller



On 4/23/24 04:47, Konrad Dybcio wrote:
> 
> 
> On 4/22/24 18:40, Sibi Sankar wrote:
>> Add support for CPUSS Control Processor (CPUCP) mailbox controller,
>> this driver enables communication between AP and CPUCP by acting as
>> a doorbell between them.
>>
>> Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@...aro.org>
>> Signed-off-by: Sibi Sankar <quic_sibis@...cinc.com>
>> ---
> 
> [...]
> 
>> +
>> +static int qcom_cpucp_mbox_send_data(struct mbox_chan *chan, void *data)
>> +{
>> +    struct qcom_cpucp_mbox *cpucp = container_of(chan->mbox, struct 
>> qcom_cpucp_mbox, mbox);
>> +    unsigned long chan_id = channel_number(chan);
>> +    u32 *val = data;
>> +
>> +    writel(*val, cpucp->tx_base + APSS_CPUCP_TX_MBOX_CMD(chan_id) + 
>> APSS_CPUCP_MBOX_CMD_OFF);
> 

Hey Konrad,

Thanks for taking time to review the series.

> Just checking in, is *this access only* supposed to be 32b instead of 64 
> like others?

yeah, the readl and writely in the driver were used intentionally.

> 
> [...]
> 
>> +
>> +    writeq(0, cpucp->rx_base + APSS_CPUCP_RX_MBOX_EN);
>> +    writeq(0, cpucp->rx_base + APSS_CPUCP_RX_MBOX_CLEAR);
>> +    writeq(0, cpucp->rx_base + APSS_CPUCP_RX_MBOX_MAP);
> 
> If these writes are here to prevent a possible interrupt storm type 
> tragedy,
> you need to read back these registers to ensure the writes have left the 
> CPU
> complex and reached the observer at the other end of the bus (not to be
> confused with barriers which only ensure that such accesses are ordered
> *when still possibly within the CPU complex*).

I couldn't find anything alluding to ^^. This sequence was just
meant to reset the mailbox. Looks like we do need to preserve the
ordering so relaxed read/writes aren't an option.

-Sibi

> 
> Moreover, if the order of them arriving (en/clear/mask) doesn't matter, you
> can add _relaxed for a possible nanosecond-order perf gain
> 
>> +
>> +    irq = platform_get_irq(pdev, 0);
>> +    if (irq < 0)
>> +        return irq;
>> +
>> +    ret = devm_request_irq(dev, irq, qcom_cpucp_mbox_irq_fn,
>> +                   IRQF_TRIGGER_HIGH, "apss_cpucp_mbox", cpucp);
>> +    if (ret < 0)
>> +        return dev_err_probe(dev, ret, "Failed to register irq: 
>> %d\n", irq);
>> +
>> +    writeq(APSS_CPUCP_RX_MBOX_CMD_MASK, cpucp->rx_base + 
>> APSS_CPUCP_RX_MBOX_MAP);
> 
> Similarly here, unless read back, we may potentially miss some 
> interrupts if
> e.g. a channel is opened and that write "is decided" (by the silicon) to 
> leave
> the internal buffer first

At this point in time we don't expect any interrupts. They are expected
only after channel activation. Also there were no recommendations for
reading it back here as well.

-Sibi

> 
> 
>> +
>> +    mbox = &cpucp->mbox;
>> +    mbox->dev = dev;
>> +    mbox->num_chans = APSS_CPUCP_IPC_CHAN_SUPPORTED;
>> +    mbox->chans = cpucp->chans;
>> +    mbox->ops = &qcom_cpucp_mbox_chan_ops;
>> +    mbox->txdone_irq = false;
>> +    mbox->txdone_poll = false;
> 
> "false" == 0 is the default value (as you're using k*z*alloc)
> 
> 
>> +
>> +    ret = devm_mbox_controller_register(dev, mbox);
>> +    if (ret)
>> +        return dev_err_probe(dev, ret, "Failed to create mailbox\n");
>> +
>> +    return 0;
>> +}
>> +
>> +static const struct of_device_id qcom_cpucp_mbox_of_match[] = {
>> +    { .compatible = "qcom,x1e80100-cpucp-mbox" },
>> +    {}
>> +};
>> +MODULE_DEVICE_TABLE(of, qcom_cpucp_mbox_of_match);
>> +
>> +static struct platform_driver qcom_cpucp_mbox_driver = {
>> +    .probe = qcom_cpucp_mbox_probe,
>> +    .driver = {
>> +        .name = "qcom_cpucp_mbox",
>> +        .of_match_table = qcom_cpucp_mbox_of_match,
>> +    },
>> +};
>> +module_platform_driver(qcom_cpucp_mbox_driver);
> 
> That's turbo late. Go core_initcall.
> 
> Konrad