On 2/26/24 2:03 PM, Sumit Garg wrote:
On Mon, 26 Feb 2024 at 16:26, Marek Vasut marex@denx.de wrote:
On 2/26/24 11:43 AM, Sumit Garg wrote:
On Mon, 26 Feb 2024 at 14:24, Marek Vasut marex@denx.de wrote:
On 2/26/24 9:04 AM, Sumit Garg wrote:
[...]
+static int imx7_reset_assert_imx8mp(struct reset_ctl *rst)
Linux calls those imx8mp_reset_set() can co. which is less confusing than imx7...imx8mp() , use it.
IMO, it would be more confusing if we choose two different naming patterns for the same driver, see existing function names imx7_reset_{deassert/assert}_imx* pattern.
Maybe create a patch which updates the naming scheme ?
I can do that as a preparatory patch.
Using two SoC names in one function name is confusing I think .
In fact, why not copy the code from Linux outright ?
I suppose the question here is to not convert the driver to use regmap, right? If it is then that should be done as a separate cleanup patch.
If it isn't too hard, please do the clean up first and then add the MX8MP bits on top.
The coding part is easy but it is rather the testing part which I am worried about as I don't have access to all the platforms which rely on this driver.
Don't worry about that part, I'm reasonably sure if something breaks, it will be spotted sooner rather than later.