From mboxrd@z Thu Jan  1 00:00:00 1970
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Sat, 17 May 2014 21:56:15 +0000
Subject: Re: [PATCH 2/4] cpupower: Remove redundant error check
Message-Id: <20140517215615.GI16255@mwanda>
List-Id: <kernel-janitors.vger.kernel.org>
References: <1400350980-30455-2-git-send-email-peter.senna@gmail.com>
 <20140517202255.GF15585@mwanda>
 <CA+MoWDqLtr-O=VQNC7qqNUJvWkEPtVr8cw9t5sB=adQWkywLZw@mail.gmail.com>
In-Reply-To: <CA+MoWDqLtr-O=VQNC7qqNUJvWkEPtVr8cw9t5sB=adQWkywLZw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: Peter Senna Tschudin <peter.senna@gmail.com>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>, Thomas Renninger <trenn@suse.de>, "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>, "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>, Alan Cox <alan@linux.intel.com>, kernel-janitors@vger.kernel.org

On Sat, May 17, 2014 at 11:34:46PM +0200, Peter Senna Tschudin wrote:
> On Sat, May 17, 2014 at 10:22 PM, Dan Carpenter
> <dan.carpenter@oracle.com> wrote:
> > On Sat, May 17, 2014 at 08:22:58PM +0200, Peter Senna Tschudin wrote:
> >> diff --git a/tools/power/cpupower/utils/cpufreq-set.c b/tools/power/cpupower/utils/cpufreq-set.c
> >> index a416de8..4e2f35a 100644
> >> --- a/tools/power/cpupower/utils/cpufreq-set.c
> >> +++ b/tools/power/cpupower/utils/cpufreq-set.c
> >> @@ -320,12 +320,11 @@ int cmd_freq_set(int argc, char **argv)
> >>
> >>               printf(_("Setting cpu: %d\n"), cpu);
> >>               ret = do_one_cpu(cpu, &new_pol, freq, policychange);
> >> -             if (ret)
> >> +             if (ret) {
> >> +                     print_error();
> >>                       break;
> >
> > Just return directly instead of break return;
> >
> >> +             }
> >>       }
> >>
> >> -     if (ret)
> >> -             print_error();
> >> -
> >>       return ret;
> >
> > Are you sure this patch is correct?  Theoretically, it's possible to
> > reach the end of this function without going hitting the
> > "ret = do_one_cpu(...);" assignment.
> >
> > Don't be fooled by the "int ret = 0;" initialization, that is a trick
> > initialization to mislead the unwary.  By the end of the do while loop
> > then "ret" is always -1.
> I have missed that, thank you for pointing this out. This patch is
> wrong and should not be applied, please ignore it.
> 
> Dan, should I just leave this file as it is?

I think in reality we should always hit the "ret = do_one_cpu()"
assignment.  But your static analysis tool should say that we don't know
that, so that's why I brought it up.

My guess is that the original code is bad and we should say:

		ret = do_one_cpu(cpu, &new_pol, freq, policychange);
		if (ret) {
			print_error();
			return ret;
		}
	}

	return 0;

I am currently involved in a number of threads, not just yours, where I
am encouraging people to replace ambiguous returns with "return 0;".
This is my life now.

regards,
dan carpenter