Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH v4] tpm: fix reference counting for struct tpm_chip
To:     Jason Gunthorpe <jgg@ziepe.ca>
Cc:     peterhuewe@gmx.de, jarkko@kernel.org, stefanb@linux.vnet.ibm.com,
        James.Bottomley@hansenpartnership.com,
        linux-integrity@vger.kernel.org, linux-kernel@vger.kernel.org,
        Lino Sanfilippo <l.sanfilippo@kunbus.com>,
        stable@vger.kernel.org
References: <1613435460-4377-1-git-send-email-LinoSanfilippo@gmx.de>
 <1613435460-4377-2-git-send-email-LinoSanfilippo@gmx.de>
 <20210216125342.GU4718@ziepe.ca>
From:   Lino Sanfilippo <LinoSanfilippo@gmx.de>
Message-ID: <792e6e77-ef0a-d678-88fe-71efb7dcd52e@gmx.de>
Date:   Tue, 16 Feb 2021 20:04:23 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.10.0
MIME-Version: 1.0
In-Reply-To: <20210216125342.GU4718@ziepe.ca>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hi,

On 16.02.21 at 13:53, Jason Gunthorpe wrote:
> On Tue, Feb 16, 2021 at 01:31:00AM +0100, Lino Sanfilippo wrote:
>>
>> +static int tpm_add_tpm2_char_device(struct tpm_chip *chip)
>> +{
>> +	int rc;
>> +
>> +	device_initialize(&chip->devs);
>> +	chip->devs.parent =3D chip->dev.parent;
>> +	chip->devs.class =3D tpmrm_class;
>> +
>> +	rc =3D dev_set_name(&chip->devs, "tpmrm%d", chip->dev_num);
>> +	if (rc)
>> +		goto out_put_devs;
>> +	/*
>> +	 * get extra reference on main device to hold on behalf of devs.
>> +	 * This holds the chip structure while cdevs is in use. The
>> +	 * corresponding put is in the tpm_devs_release.
>> +	 */
>> +	get_device(&chip->dev);
>> +	chip->devs.release =3D tpm_devs_release;
>> +	chip->devs.devt =3D
>> +		MKDEV(MAJOR(tpm_devt), chip->dev_num + TPM_NUM_DEVICES);
>> +	cdev_init(&chip->cdevs, &tpmrm_fops);
>> +	chip->cdevs.owner =3D THIS_MODULE;
>> +
>> +	rc =3D cdev_device_add(&chip->cdevs, &chip->devs);
>> +	if (rc) {
>> +		dev_err(&chip->devs,
>> +			"unable to cdev_device_add() %s, major %d, minor %d, err=3D%d\n",
>> +			dev_name(&chip->devs), MAJOR(chip->devs.devt),
>> +			MINOR(chip->devs.devt), rc);
>> +		goto out_put_devs;
>> +	}
>> +
>> +	return 0;
>> +
>> +out_put_devs:
>> +	put_device(&chip->devs);
>
> I'd rather you organize this so chip->devs.release and the get_device
> is always sent instead of having the possiblity for a put_device that
> doesn't call release
>

Agreed, I will change it. It should not make a difference in terms of corr=
ectness
but I see that it is less confusing if both error cases are handled simila=
rly (plus its
only a minimal change).


Best regards,
Lino