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 v3 2/3] iommu/tegra-smmu: Rework .probe_device and
 .attach_dev
To:     Nicolin Chen <nicoleotsuka@gmail.com>, thierry.reding@gmail.com,
        joro@8bytes.org, krzk@kernel.org
Cc:     vdumpa@nvidia.com, jonathanh@nvidia.com,
        linux-tegra@vger.kernel.org, iommu@lists.linux-foundation.org,
        linux-kernel@vger.kernel.org
References: <20200930084258.25493-1-nicoleotsuka@gmail.com>
 <20200930084258.25493-3-nicoleotsuka@gmail.com>
From:   Dmitry Osipenko <digetx@gmail.com>
Message-ID: <01728f61-df0c-291d-5589-4468be25f00c@gmail.com>
Date:   Wed, 30 Sep 2020 17:41:42 +0300
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: <20200930084258.25493-3-nicoleotsuka@gmail.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

...
> +	struct tegra_mc *mc = devm_tegra_get_memory_controller(dev);
> +	struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev);
>  
> -		of_node_put(args.np);
> -		index++;
> -	}
> +	/* An invalid mc pointer means mc and smmu drivers are not ready */
> +	if (IS_ERR(mc))
> +		return ERR_PTR(-EPROBE_DEFER);
>  
> -	if (!smmu)
> +	/*
> +	 * IOMMU core allows -ENODEV return to carry on. So bypass any call
> +	 * from bus_set_iommu() during tegra_smmu_probe(), as a device will
> +	 * call in again via of_iommu_configure when fwspec is prepared.
> +	 */
> +	if (!mc->smmu || !fwspec || fwspec->ops != &tegra_smmu_ops)
>  		return ERR_PTR(-ENODEV);
>  
> -	return &smmu->iommu;
> +	dev_iommu_priv_set(dev, mc->smmu);
> +
> +	return &mc->smmu->iommu;
>  }

Is it really okay to use devm_tegra_get_memory_controller() here?

I assume it should be more preferred to do it only for devices that have
fwspec->ops == &tegra_smmu_ops.

Secondly, it also looks to me that a non-devm variant should be more
appropriate here because tegra_smmu_probe_device() isn't invoked by the
devices themselves.