Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH v3 09/13] dmaengine: tegra-apb: Remove runtime PM usage
To:     Jon Hunter <jonathanh@nvidia.com>,
        Laxman Dewangan <ldewangan@nvidia.com>,
        Vinod Koul <vkoul@kernel.org>,
        Dan Williams <dan.j.williams@intel.com>,
        Thierry Reding <thierry.reding@gmail.com>,
        =?UTF-8?B?TWljaGHFgiBNaXJvc8WCYXc=?= <mirq-linux@rere.qmqm.pl>
Cc:     dmaengine@vger.kernel.org, linux-tegra@vger.kernel.org,
        linux-kernel@vger.kernel.org
References: <20200106011708.7463-1-digetx@gmail.com>
 <20200106011708.7463-10-digetx@gmail.com>
 <f63a68cf-bb9d-0e79-23f3-233fc97ca6f9@nvidia.com>
From:   Dmitry Osipenko <digetx@gmail.com>
Message-ID: <fd6215ac-a646-4e13-ee22-e815a69cd099@gmail.com>
Date:   Tue, 7 Jan 2020 20:12:45 +0300
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.3.0
MIME-Version: 1.0
In-Reply-To: <f63a68cf-bb9d-0e79-23f3-233fc97ca6f9@nvidia.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hello Jon,

07.01.2020 18:13, Jon Hunter пишет:
> 
> On 06/01/2020 01:17, Dmitry Osipenko wrote:
>> There is no benefit from runtime PM usage for the APB DMA driver because
>> it enables clock at the time of channel's allocation and thus clock stays
>> enabled all the time in practice, secondly there is benefit from manually
>> disabled clock because hardware auto-gates it during idle by itself.
> 
> This assumes that the channel is allocated during a driver
> initialisation. That may not always be the case. I believe audio is one
> case where channels are requested at the start of audio playback.

At least serial, I2C, SPI and T20 FUSE are permanently keeping channels
allocated, thus audio is an exception here. I don't think that it's
practical to assume that there is a real-world use-case where audio
driver is the only active DMA client.

The benefits of gating the DMA clock are also dim, do you have any
power-consumption numbers that show that it's really worth to care about
the clock-gating?