On Wed, 26 Aug 2020 12:13:01 +0200, Mark Brown wrote:
On Wed, Aug 26, 2020 at 12:09:28PM +0200, Takashi Iwai wrote:
Mark Brown wrote:
checkpatch is broken.
Heh, I'm not objecting it :)
OTOH, it's also true that ENOTSUPP is no good error code if returned to user-space. Unfortunately many codes (including what I wrote) use this code mistakenly, and they can't be changed any longer...
It's also used internally in various places without being returned to userspace, that's what's going on here - the regmap core has some specific checks for -ENOTSUPP.
Sure, for such an internal usage any code can be used. The question is a case like this -- where the return code might be carried to outside. Though, looking through the grep output, all callers simply return -EINVAL for any errors, so it doesn't matter much for now.
BTW, there are a few callers of devm_regmap_init_sdw() checking the return value with NULL. This will crash as the function returns the error pointer, and they must be checked with IS_ERR() instead.
Takashi