On 04/09/2023 14:30, Mark Brown wrote:
On Mon, Sep 04, 2023 at 02:17:43PM +0200, Krzysztof Kozlowski wrote:
On 04/09/2023 13:46, wangweidong.a@awinic.com wrote:
- ret = regmap_read(regmap, AW87390_ID_REG, &chip_id);
- if (ret) {
dev_err(&i2c->dev, "%s read chipid error. ret = %d\n", __func__, ret);return ret;- }
- if (chip_id != AW87390_CHIP_ID) {
dev_err(&i2c->dev, "unsupported device\n");Why? The compatible tells it cannot be anything else.
This is very common good practice, as well as validating communication
No, it is neither common nor good. The kernel's job is not to verify the supplied DTS. Rob also made here a point:
https://lore.kernel.org/all/CAL_Jsq+wcrOjh7+0c=mrg+Qz6dbhOUE-VEeQ4FoWC3Y7ENo...
with the device it verifies that the device descrbied in the DT is the one that is actually present in the system. This might create hassle down the line if there is a backwards compatible upgrade but that's much rarer for this class of hardware than cut'n'pasting of device trees.
Best regards, Krzysztof