On Wed, 17 Sep 2014 08:09:36 -0400, nick said:
On 14-09-17 08:05 AM, Greg Freemyer wrote:
I don't know that chunk of code, but error messages that go to the kernel log exist for a specific reason. Taking them out requires a specific reason.
Ie. This would make a good commit message "At this point the condition is well understood and the code that handles it is well tested and has been stable for 3 years, thus removing the error message."
This is what I meant with my patch, why have a unneeded error message if the code is already tested and only uses the return value in that function.
Sorry Nick, but that's *not* what Greg meant. What he *meant* was that removal of an error message should be its *own* commit, explaining *why* it was being removed and why it was OK to do so. He did *not* say that this particular removal was in fact correct. He *did* say that such a hypothetical removal *should not* be in a patch calling itself a checkpatch cleanup. He *did* say that the patch will require proof that you've examined the code, understood it, and can explain *why* the patch is OK.