linux-kernel - Re: [PATCH V9 3/5] i2c: tegra: Add DMA support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <3490a279-3584-161b-cbd4-c7ef2b5288e1@gmail.com>
Date:   Sat, 2 Feb 2019 00:45:49 +0300
From:   Dmitry Osipenko <digetx@...il.com>
To:     Sowjanya Komatineni <skomatineni@...dia.com>,
        "thierry.reding@...il.com" <thierry.reding@...il.com>,
        Jonathan Hunter <jonathanh@...dia.com>,
        Mantravadi Karthik <mkarthik@...dia.com>,
        Shardar Mohammed <smohammed@...dia.com>,
        Timo Alho <talho@...dia.com>
Cc:     "linux-tegra@...r.kernel.org" <linux-tegra@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-i2c@...r.kernel.org" <linux-i2c@...r.kernel.org>
Subject: Re: [PATCH V9 3/5] i2c: tegra: Add DMA support

02.02.2019 0:36, Sowjanya Komatineni пишет:
> 
>>>
>>>> 	rx_chan = dma_request_slave_channel_reason(i2c_dev->dev, "rx");
>>>> 	if (IS_ERR(rx_chan))
>>>> 		return PTR_ERR(rx_chan); 
>>>>
>>>>
>>>>> +
>>>>> +	dma_chan = dma_request_slave_channel_reason(i2c_dev->dev, "tx");
>>>>> +	if (IS_ERR(dma_chan)) {
>>>>> +		err = PTR_ERR(dma_chan);
>>>>> +		goto error;
>>>>
>>>> It's a good practice to release resources in opposite order to the allocation. Hence better to write this as:
>>>>
>>>> 		goto err_release_rx;
>>>>
>>>>> +	}
>>>>> +
>>>>> +	i2c_dev->tx_dma_chan = dma_chan;
>>>>> +
>>>>> +	dma_buf = dma_alloc_coherent(i2c_dev->dev,
>>>>> +				     i2c_dev->dma_buf_size, &dma_phys,
>>>>> +				     GFP_KERNEL | __GFP_NOWARN);
>>>>> +
>>>>> +	if (!dma_buf) {
>>>>> +		dev_err(i2c_dev->dev, "failed to allocate the DMA buffer\n");
>>>>> +		err = -ENOMEM;
>>>>> +		goto error;
>>>>
>>>> 		goto err_release_tx;
>>>>
>>>>> +	}
>>>>> +
>>>>> +	i2c_dev->dma_buf = dma_buf;
>>>>> +	i2c_dev->dma_phys = dma_phys;	
>>
>> 	i2c_dev->rx_dma_chan = rx_chan;
>>>> 	i2c_dev->tx_dma_chan = tx_chan;
>>>>
>>>>> +	return 0;
>>>>> +
>>>>> +error:
>>>>> +	if (i2c_dev->tx_dma_chan)
>>>>> +		dma_release_channel(i2c_dev->tx_dma_chan);
>>>>> +
>>>>> +	if (i2c_dev->rx_dma_chan)
>>>>> +		dma_release_channel(i2c_dev->rx_dma_chan); 
>>>>
>>>> error_release_tx:
>>>> 	dma_release_channel(tx_chan);
>>>> error_release_rx:
>>>> 	dma_release_channel(rx_chan);
>>>>
>>>>> +
>>>>> +	return err;
>>>
>>> I am releasing resources in reverse order to allocation.
>>> Trying for rx allocation followed by tx allocation During release 
>>> releasing tx and then rx.
>>> In case if tx allocation fails, doesn’t go thru release. If rx or buf 
>>> allocation fails, releases tx first and then rx
>>>
>>
>>
>> Okay. Anyway it's a good-n-common practice to write it in the way I'm suggesting.
>>
>> And please set rx_chan and tx_chan after dma_buf allocation as I'm suggesting because you current variant will crash kernel since if dma_buf allocation fails, both rx and tx channels will be released and you're not setting them to NULL in that case.
> 
> OK, my wrong assumption. Thought dma_release_channel will NULL chan after its freed
> Will update and send V10
> 

There is no need to rush, you may wait at least for the testing feedback for now. Also maybe Thierry or somebody else will have something to say as well.