[PATCH v5 06/17] soc: qcom: Add Qualcomm APSS minidump kernel driver

Konrad Dybcio konrad.dybcio at linaro.org
Mon Sep 11 12:33:43 PDT 2023


On 11.09.2023 21:09, Jeff Johnson wrote:
> On 9/11/2023 4:07 AM, Krzysztof Kozlowski wrote:
>> On 09/09/2023 22:16, Mukesh Ojha wrote:
>>> +/**
>>> + * qcom_minidump_region_register() - Register region in APSS Minidump table.
>>> + * @region: minidump region.
>>> + *
>>> + * Return: On success, it returns 0 and negative error value on failure.
>>> + */
>>> +int qcom_minidump_region_register(const struct qcom_minidump_region *region)
>>> +{
>>> +    struct minidump *md;
>>> +    int ret;
>>> +
>>> +    md = qcom_smem_minidump_ready();
>>> +    if (!md)
>>> +        return -EPROBE_DEFER;
>>> +
>>> +    if (!qcom_minidump_valid_region(region))
>>> +        return -EINVAL;
>>> +
>>> +    mutex_lock(&md->md_lock);
>>> +    ret = qcom_md_region_register(md, region);
>>> +    if (ret)
>>> +        goto unlock;
>>> +
>>> +    qcom_md_update_elfheader(md, region);
>>> +unlock:
>>> +    mutex_unlock(&md->md_lock);
>>> +    return ret;
>>> +}
>>> +EXPORT_SYMBOL_GPL(qcom_minidump_region_register);
>>
>> NAK, there is no user for this.
>>
>> Drop all exports from minidump drivers. Your upstream driver *must not*
>> expose stuff to your vendor drivers.
> 
> Do we not expect that upstream drivers would want to register?
> Mind you, in the downstream code the following was a bad limitation:
> #define MAX_NUM_OF_SS           10
No, Krzysztof meant that you are not allowed to export symbols
without immediately providing a user for them - meaning if the
functions are not going to be used upstream, this change will
not be accepted.

Konrad



More information about the linux-arm-kernel mailing list