Re: [PATCH net 2/5] net/smc: remove duplicate mutex_unlock

[Heiko Carstens <heiko.carstens@de.ibm.com>]

On Mon, Oct 01, 2018 at 05:22:22PM +0200, Ursula Braun wrote:
> 
> > 3b2dec260 Hans Wippel   2018-05-18  1282  	/* finish worker */
> > c69342ef9 Ursula Braun  2018-09-18  1283  	if (!ism_supported) {
> > c69342ef9 Ursula Braun  2018-09-18  1284  		if (smc_listen_rdma_finish(new_smc, &cclc, local_contact))
> > c69342ef9 Ursula Braun  2018-09-18  1285  			return;
> >                                                                 ^^^^^^
> > We need to mutex_unlock(&smc_create_lgr_pending); before the return.
> >
> 
> The smatch warning is not necessary, since the mutex_unlock(&smc_create_lgr_pending)
> for this case is done within smc_listen_rdma_finish().
> 
> > c69342ef9 Ursula Braun  2018-09-18  1286  	}
> > 3b2dec260 Hans Wippel   2018-05-18  1287  	smc_conn_save_peer_info(new_smc, &cclc);
> > 3b2dec260 Hans Wippel   2018-05-18  1288  	mutex_unlock(&smc_create_lgr_pending);
> > 3b2dec260 Hans Wippel   2018-05-18 @1289  	smc_listen_out_connected(new_smc);
> > a046d57da Ursula Braun  2017-01-09  1290  }
> > a046d57da Ursula Braun  2017-01-09  1291  
> > 
> > ---
> > 0-DAY kernel test infrastructure                Open Source Technology Center
> > https://lists.01.org/pipermail/kbuild-all                   Intel Corporation
> > 
> 
> But you could argue the code confuses smatch and other readers. Thus I could improve
> readability with a patch like this:

Yes, please.

> ---
>  net/smc/af_smc.c |    5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> --- a/net/smc/af_smc.c
> +++ b/net/smc/af_smc.c
> @@ -1184,7 +1184,6 @@ static int smc_listen_rdma_finish(struct
>  	return 0;
>  
>  decline:
> -	mutex_unlock(&smc_create_lgr_pending);
>  	smc_listen_decline(new_smc, reason_code, local_contact);
>  	return reason_code;
>  }

The lonely mutex_unlock() also _looks_ like a bug ;)