Re: [PATCH 3/4] io_uring: clean __io_import_iovec()

[Dan Carpenter <dan.carpenter@oracle.com>]

On Tue, Nov 23, 2021 at 12:07:48AM +0000, Pavel Begunkov wrote:
> Apparently, implicit 0 to NULL conversion with ERR_PTR is not
> recommended and makes some tooling like Smatch to complain. Handle it
> explicitly, compilers are perfectly capable to optimise it out.
> 
> Link: https://lore.kernel.org/all/20211108134937.GA2863@kili/ 
> Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
> Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>

I like that this this code is now really explicit that the NULL returns
are intentional.  I had figured that out from the git log already.

Passing zero to ERR_PTR() is valid.  Linus complained about this Smatch
warning.  But I'm not going to delete the check because probably around
70% (complete guess) of the cases in new code are bugs.  In older
kernels the Smatch warnings tend to be 100% false positives because we
fix the real bugs.  Also the kbuild-bot hits a bunch of error pointer
false positives for cross compile builds but I don't have a cross
compile system set up so I haven't debugged that.  :/  It has something
to do with pointers being treated as signed on those arches.

But what I really like to see is documentation explaining when a
function is going to return NULL vs an error pointer.

regards,
dan carpenter