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] dmaengine: pl330: introduce debugfs interface
To:     Vinod Koul <vkoul@kernel.org>
Cc:     dmaengine@vger.kernel.org, linux-kernel@vger.kernel.org
References: <20190314184953.8129-1-katsuhiro@katsuster.net>
 <20190315170059.GF5348@vkoul-mobl>
From:   Katsuhiro Suzuki <katsuhiro@katsuster.net>
Message-ID: <38787dd0-0999-adab-6a7d-88792fb7ad61@katsuster.net>
Date:   Sun, 17 Mar 2019 18:58:54 +0900
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.5.3
MIME-Version: 1.0
In-Reply-To: <20190315170059.GF5348@vkoul-mobl>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hello Vinod,

Thank you for your comment.
I fix it all and re-post v2 patch.


On 2019/03/16 2:00, Vinod Koul wrote:
> On 15-03-19, 03:49, Katsuhiro Suzuki wrote:
>> This patch adds debugfs interface to show the relationship between
>> DMA threads (hardware resource for transferring data) and DMA
>> channel ID of DMA slave.
>>
>> Typically, PL330 has many slaves than number of DMA threads.
>> So sometimes PL330 cannot allocate DMA threads for all slaves even
>> if an user specify DMA channel ID in devicetree. This interface will
> 
> a user :)
> 
>> be useful for checking that DMA threads are allocated or not.
>>
>> Below is an output sample:
>>
>> $ sudo cat /sys/kernel/debug/ff1f0000.dmac
>> PL330 physical channels:
>> THREAD:         CHANNEL:
>> --------        -----
>> 0               8
>> 1               9
>> 2               11
>> 3               12
>> 4               14
>> 5               15
>> 6               10
>> 7               --
>>
>> Signed-off-by: Katsuhiro Suzuki <katsuhiro@katsuster.net>
>> ---
>>   drivers/dma/pl330.c | 51 +++++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 51 insertions(+)
>>
>> diff --git a/drivers/dma/pl330.c b/drivers/dma/pl330.c
>> index eec79fdf27a5..aab71863757c 100644
>> --- a/drivers/dma/pl330.c
>> +++ b/drivers/dma/pl330.c
>> @@ -11,6 +11,7 @@
>>    * (at your option) any later version.
>>    */
>>   
>> +#include <linux/debugfs.h>
>>   #include <linux/kernel.h>
>>   #include <linux/io.h>
>>   #include <linux/init.h>
>> @@ -2896,6 +2897,55 @@ static irqreturn_t pl330_irq_handler(int irq, void *data)
>>   	BIT(DMA_SLAVE_BUSWIDTH_4_BYTES) | \
>>   	BIT(DMA_SLAVE_BUSWIDTH_8_BYTES)
>>   
>> +#ifdef CONFIG_DEBUG_FS
>> +static int pl330_debugfs_show(struct seq_file *s, void *data)
>> +{
>> +	struct pl330_dmac *pl330 = s->private;
>> +	int chans, pchs, ch, pr, found;
>> +
>> +	chans = pl330->pcfg.num_chan;
>> +	pchs = pl330->num_peripherals;
>> +
>> +	seq_puts(s, "PL330 physical channels:\n");
>> +	seq_puts(s, "THREAD:\t\tCHANNEL:\n");
>> +	seq_puts(s, "--------\t-----\n");
>> +	for (ch = 0; ch < chans; ch++) {
>> +		struct pl330_thread *thrd = &pl330->channels[ch];
>> +
>> +		found = -1;
> 
> found can be uninitialized if we skip this for, which should not be an
> issue as chans> 0!
> 

Exactly. Thanks, 'found' should be changed to for-loop local variable.


>> +		for (pr = 0; pr < pchs; pr++) {
>> +			struct dma_pl330_chan *pch = &pl330->peripherals[pr];
>> +
>> +			if (!pch->thread || thrd->id != pch->thread->id)
>> +				continue;
>> +
>> +			found = pr;
>> +		}
>> +
>> +		seq_printf(s, "%d\t\t", thrd->id);
>> +		if (found == -1)
>> +			seq_puts(s, "--\n");
>> +		else
>> +			seq_printf(s, "%d\n", found);
>> +	}
>> +
>> +	return 0;
>> +}
>> +
>> +DEFINE_SHOW_ATTRIBUTE(pl330_debugfs);
>> +
>> +static inline void init_pl330_debugfs(struct pl330_dmac *pl330)
>> +{
>> +	debugfs_create_file(dev_name(pl330->ddma.dev),
>> +			    S_IFREG | S_IRUGO, NULL, pl330,
> 
> are we not supposed to use octal permissions?
> 
>> +			    &pl330_debugfs_fops);
>> +}
>> +#else
>> +static inline void init_pl330_debugfs(struct pl330_dmac *pl330)
>> +{
>> +}
>> +#endif
>> +
>>   /*
>>    * Runtime PM callbacks are provided by amba/bus.c driver.
>>    *
>> @@ -3082,6 +3132,7 @@ pl330_probe(struct amba_device *adev, const struct amba_id *id)
>>   		dev_err(&adev->dev, "unable to set the seg size\n");
>>   
>>   
>> +	init_pl330_debugfs(pl330);
>>   	dev_info(&adev->dev,
>>   		"Loaded driver for PL330 DMAC-%x\n", adev->periphid);
>>   	dev_info(&adev->dev,
>> -- 
>> 2.20.1
> 

Best Regards,
Katsuhiro Suzuki