On 15.02.22 08:08, Ramon Fried wrote:
On Mon, Feb 14, 2022 at 12:33 PM Daniel Klauer daniel.klauer@gin.de wrote:
On 12.02.22 12:50, Ramon Fried wrote:
On Wed, Feb 9, 2022 at 5:41 PM Daniel Klauer daniel.klauer@gin.de wrote:
The miiphy_read/miiphy_write functions return 1 on error, not -errno.
Why don't you just fix the miiphy_read/miiphy_write functions ?
Other functions from that module do the same, so I assumed it's intentional. It could be fixed too though, with a corresponding fix up of the few callers that expect the positive return value on error.
Sometimes a real fix needs more work. What caused you to change it on your side to -EIO, is there someone checking for EIO in the framework code ?
It's not that anything specifically expects -EIO, I just picked it because I had to pick something and it seemed like a good solution. Of course that only makes sense with the current miiphy_read/write functions which don't propagate the real errno. If that were to be changed, this patch for mv88e6352.c would not be useful anymore.