On 02.08.2018 04:00, Andy Duan wrote:
> From: Stefan Agner <ste...@agner.ch> Sent: 2018年8月1日 19:45
>> Check DMA addressing limitations as suggested by the DMA API how-to.
>> This does not fix a particular issue seen but is considered good style.
>>
>> Signed-off-by: Stefan Agner <ste...@agner.ch>
>> ---
>>  drivers/net/ethernet/freescale/fec_main.c | 6 ++++++
>>  1 file changed, 6 insertions(+)
>>
>> diff --git a/drivers/net/ethernet/freescale/fec_main.c
>> b/drivers/net/ethernet/freescale/fec_main.c
>> index c729665107f5..af0fb200e936 100644
>> --- a/drivers/net/ethernet/freescale/fec_main.c
>> +++ b/drivers/net/ethernet/freescale/fec_main.c
>> @@ -3146,6 +3146,12 @@ static int fec_enet_init(struct net_device
>> *ndev)
>>      fep->tx_align = 0x3;
>>  #endif
>>
>> +    /* Check mask of the streaming and coherent API */
>> +    if (dma_set_mask_and_coherent(&fep->pdev->dev,
>> DMA_BIT_MASK(32))) {
>> +            dev_warn(&fep->pdev->dev, "No suitable DMA available\n");
>> +            return -ENODEV;
> It is better:
> 
> ret = dma_set_mask_and_coherent(&fep->pdev->dev, DMA_BIT_MASK(32));
> if (ret < 0) {
>       dev_warn(&fep->pdev->dev, "No suitable DMA available\n");
>       return ret;
> }

The code comes from the example in Documentation/DMA-API-HOWTO.txt.

I can rearrange if you prefer.

> 
> 
> If the patch aim to "OF: Don't set default coherent DMA mask", I think
> not only this driver need to add the DMA mask limitations,  many other
> drivers also need.

It doesn't exactly address the issue since with that patch DMA mask
wasn't set at all. This code would just make the driver fail with
-ENODEV instead of -ENOMEM.

To force a DMA mask I would have to set the DMA mask using
dma_coerce_mask_and_coherent(). But as discussed with Robin it is safe
to assume that an initial mask is set by the bus code:
http://lkml.kernel.org/r/892f9d14-e6fd-7b1b-d07b-af0be6e62...@arm.com

Setting the DMA mask on DT buses has been addressed by Robin with
"of/platform: Initialise default DMA masks". So everything should be
fine again.

I agree, other drivers need fixing too.

--
Stefan

> 
> 
>> +    }
>> +
>>      fec_enet_alloc_queue(ndev);
>>
>>      bd_size = (fep->total_tx_ring_size + fep->total_rx_ring_size) * dsize;
>> --
>> 2.18.0

Reply via email to