linux-kernel - Re: [PATCH v2] staging: rtl8723bs: os_dep: simplify the return statement.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 11 Oct 2021 14:33:21 +0200
From:   "Fabio M. De Francesco" <fmdefrancesco@...il.com>
To:     gregkh@...uxfoundation.org, fabioaiuto83@...il.com,
        ross.schm.dev@...il.com, marcocesati@...il.com,
        saurav.girepunje@...il.com, insafonov@...il.com,
        linux-staging@...ts.linux.dev, linux-kernel@...r.kernel.org,
        Saurav Girepunje <saurav.girepunje@...il.com>
Cc:     saurav.girepunje@...mail.com
Subject: Re: [PATCH v2] staging: rtl8723bs: os_dep: simplify the return statement.

On Monday, October 11, 2021 10:54:11 AM CEST Fabio M. De Francesco wrote:
> On Sunday, October 10, 2021 7:06:05 AM CEST Saurav Girepunje wrote:
> > Remove the unneeded and redundant check of variable on goto out.
> > Simplify the return using multiple goto label to avoid
> > unneeded check.
> > 
> > Signed-off-by: Saurav Girepunje <saurav.girepunje@...il.com>
> > ---
> > 
> > ChangeLog V2:
> > 	-Add goto out after the memcpy for no error case return with
> > 	 ret only. Free is not required on no error case.
> 
> Please write versions logs that reflect clearly and unequivocally what you 
> changed between revisions and why. Subjects, Commit messages (Changelogs), 
> and Versions logs are the "specifics" of your work. There must be no 
> inconsistencies between these and the code or the history of the changes of 
> the code.
> 
> You may think that I'm pedantic, but since I acked your patch, I don't want 
> to be misunderstood to be a promoter of approximate or clearly incorrect 
> messages.
> 
> "Free is not required on no error case" conveys the message that you have 
> changed something that is not required but that is still potentially 
allowed.
> 
> This is not the case because the problem that you fix with v2 is _not_ 
> something that is merely not required and unnecessary. You have fixed a bug 
> that is introduced in v1. Introducing bugs is not allowed. If you do 
> something that is not allowed you cannot simply say that it is not 
required. 
> 
> > 
> > ChangeLog V1:
> > 	-Remove the unneeded and redundant check of variable on
> > 	 goto out.
> > 	-Simplify the return using multiple goto label to avoid
> > 	 unneeded check.
> > 
> >  .../staging/rtl8723bs/os_dep/ioctl_cfg80211.c | 22 +++++++++----------
> >  1 file changed, 10 insertions(+), 12 deletions(-)
> > 
> > diff --git a/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c b/drivers/
> staging/rtl8723bs/os_dep/ioctl_cfg80211.c
> > index 0868f56e2979..ae9579dc0848 100644
> > --- a/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c
> > +++ b/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c
> > @@ -2312,7 +2312,7 @@ static int rtw_cfg80211_add_monitor_if(struct 
adapter 
> *padapter, char *name, str
> >  	mon_wdev = rtw_zmalloc(sizeof(struct wireless_dev));
> >  	if (!mon_wdev) {
> >  		ret = -ENOMEM;
> > -		goto out;
> > +		goto err_zmalloc;
> >  	}
> > 
> >  	mon_wdev->wiphy = padapter->rtw_wdev->wiphy;
> > @@ -2322,23 +2322,21 @@ static int rtw_cfg80211_add_monitor_if(struct 
> adapter *padapter, char *name, str
> > 
> >  	ret = cfg80211_register_netdevice(mon_ndev);
> >  	if (ret) {
> > -		goto out;
> > +		goto err_register;
> >  	}
> > 
> >  	*ndev = pwdev_priv->pmon_ndev = mon_ndev;
> >  	memcpy(pwdev_priv->ifname_mon, name, IFNAMSIZ+1);
> > +	goto out;

As Pavel noticed, probably you'd better return 'ret' here. The logic does not 
change, but I guess that this is what Linux developers usually do.

>From a review by Dan Carpenter of one of your other patches: "[] Do nothing
labels only hurt readability and introduce "forgot to set the error code
bugs.".

Thanks,

Fabio

> > 
> > -out:
> > -	if (ret && mon_wdev) {
> > -		kfree(mon_wdev);
> > -		mon_wdev = NULL;
> > -	}
> > -
> > -	if (ret && mon_ndev) {
> > -		free_netdev(mon_ndev);
> > -		*ndev = mon_ndev = NULL;
> > -	}
> > +err_register:
> > +	kfree(mon_wdev);
> > +	mon_wdev = NULL;
> 
> Probably you have already read a message by Greg Kroah-Hartman that 
confirms 
> what I wrote in another message: "There is no need to set a local variable 
> like this to NULL.".
> 
> So please submit a v3. With the two changes requested above, my "acked-by" 
> tag is confirmed again.
> 
> Thanks,
> 
> Fabio
> 
> > 
> > +err_zmalloc:
> > +	free_netdev(mon_ndev);
> > +	*ndev = mon_ndev = NULL;
> > +out:
> >  	return ret;
> >  }
> > 
> > --
> > 2.32.0
> > 
> > 
> > 
> 
> 
> 
> 
>