Received-SPF: pass (google.com: domain of linux-kernel+bounces-24636-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Message-ID: <98067d77-bfde-45a8-af6b-9eaf4bcc3839@tuxon.dev>
Date: Fri, 12 Jan 2024 14:05:17 +0200
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH] pinctrl: renesas: rzg2l: Fix double unlock in
 rzg2l_dt_subnode_to_map()
Content-Language: en-US
To: Dan Carpenter <dan.carpenter@linaro.org>
Cc: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com>,
 Geert Uytterhoeven <geert+renesas@glider.be>,
 Linus Walleij <linus.walleij@linaro.org>, linux-renesas-soc@vger.kernel.org,
 linux-gpio@vger.kernel.org, linux-kernel@vger.kernel.org,
 kernel-janitors@vger.kernel.org
References: <f8c3a3a0-7c48-4e40-8af0-ed4e9d9b049f@moroto.mountain>
 <185128ab-f229-4b40-91cd-ebdc138e11ac@tuxon.dev>
 <300d204b-8151-45f3-9977-7ceb3a5c5eb0@moroto.mountain>
From: claudiu beznea <claudiu.beznea@tuxon.dev>
In-Reply-To: <300d204b-8151-45f3-9977-7ceb3a5c5eb0@moroto.mountain>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit


On 12.01.2024 11:53, Dan Carpenter wrote:
> On Fri, Jan 12, 2024 at 10:55:40AM +0200, claudiu beznea wrote:
>> Hi, Dan,
>>
>> Thanks for your patch!
>>
>> On 10.01.2024 20:41, Dan Carpenter wrote:
>>> If rzg2l_map_add_config() fails then the error handling calls
>>> mutex_unlock(&pctrl->mutex) but we're not holding that mutex.  Move
>>> the unlocks to before the gotos to avoid this situation.
>>>
>>> Fixes: d3aaa7203a17 ("pinctrl: renesas: rzg2l: Add pin configuration support for pinmux groups")
>>> Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
>>> ---
>>> (Not tested).
>>
>> I've tested it on RZ/G3S SoC and all is good.
>>
>> However, I think, to keep the locking scheme unchanged and simpler (FMPOV),
>> commit d3aaa7203a17 ("pinctrl: renesas: rzg2l: Add pin configuration
>> support for pinmux groups") should have been call rzg2l_map_add_config()
>> just before the mutex is locked. That would be the following diff:
>>
>> --- a/drivers/pinctrl/renesas/pinctrl-rzg2l.c
>> +++ b/drivers/pinctrl/renesas/pinctrl-rzg2l.c
>> @@ -447,6 +447,16 @@ static int rzg2l_dt_subnode_to_map(struct pinctrl_dev
>> *pctldev,
>>  		name = np->name;
>>  	}
>>
>> +	if (num_configs) {
>> +		ret = rzg2l_map_add_config(&maps[idx], name,
>> +					   PIN_MAP_TYPE_CONFIGS_GROUP,
>> +					   configs, num_configs);
>> +		if (ret < 0)
>> +			goto done;
>> +
>> +		idx++;
>> +	}
>> +
>>  	mutex_lock(&pctrl->mutex);
>>
>>  	/* Register a single pin group listing all the pins we read from DT */
>> @@ -474,16 +484,6 @@ static int rzg2l_dt_subnode_to_map(struct pinctrl_dev
>> *pctldev,
>>  	maps[idx].data.mux.function = name;
>>  	idx++;
>         ^^^^^

This needs to be here for subsequent calls of rzg2l_dt_subnode_to_map() to
know which entry in maps[] to be populated next time.

> 
>>
>> -	if (num_configs) {
>> -		ret = rzg2l_map_add_config(&maps[idx], name,
>> -					   PIN_MAP_TYPE_CONFIGS_GROUP,
>> -					   configs, num_configs);
>> -		if (ret < 0)
>> -			goto remove_group;
>> -
>> -		idx++;
>> -	}
> 
> Does the ordering of the maps[] not matter?

It doesn't matter, AFAIKT. The core code checks for map type (e.g.
PIN_MAP_TYPE_CONFIGS_GROUP) when processes the data from maps[].

> 
>> -
>>  	dev_dbg(pctrl->dev, "Parsed %pOF with %d pins\n", np, num_pinmux);
>>  	ret = 0;
>>  	goto done;
>>
>> Would you mind doing it like this?
>>
>> Please, let me know if you want me to handle it.
> 
> Either way is fine.  Whatever is easiest.

Ok, I'll prepare a patch as I already tested it on my side on multiple
platforms.

Thank you,
Claudiu Beznea

> 
> regards,
> dan carpenter
>