Re: [PATCH v2 1/3] mmc: sdhci: add support for using external DMA devices

[Adrian Hunter <adrian.hunter@intel.com>]

On 29/11/18 11:59 AM, Chunyan Zhang wrote:
> Hi Adrian,
> 
> On Thu, 29 Nov 2018 at 15:36, Adrian Hunter <adrian.hunter@intel.com> wrote:
>>
>> On 29/11/18 8:22 AM, Chunyan Zhang wrote:
>>> On Tue, 20 Nov 2018 at 21:41, Adrian Hunter <adrian.hunter@intel.com> wrote:
>>>>
>>>> On 12/11/18 9:26 AM, Chunyan Zhang wrote:
>>>>> Some standard SD host controllers can support both external dma
>>>>> controllers as well as ADMA/SDMA in which the SD host controller
>>>>> acts as DMA master. TI's omap controller is the case as an example.
>>>>>
>>>>> Currently the generic SDHCI code supports ADMA/SDMA integrated in
>>>>> the host controller but does not have any support for external DMA
>>>>> controllers implemented using dmaengine, meaning that custom code is
>>>>> needed for any systems that use an external DMA controller with SDHCI.
>>>>
>>>> I still think you probably need to reset the DMA if there are transfer
>>>> errors - perhaps you could comment on that.  Also there are some comments below.
>>>
>>> With regard to "transfer error", do you mean if
>>> sdhci_external_dma_setup() failed?
>>
>> No, I mean any error interrupt that can leave the DMA uncompleted.  For
>> SDHCI, resetting the data circuit cleans that up, but presumably something
>> is needed for external DMA?
> 
> Yes, it should need a dmaengine_terminate_all().
> 
> How about adding that at here (I will wrap it up of course):
> https://elixir.bootlin.com/linux/v4.19.5/source/drivers/mmc/host/sdhci.c#L2553

Yes except we really need to reverse
	if (host->flags & SDHCI_REQ_USE_DMA) {
	}
	if (sdhci_needs_reset(host, mrq)) {
	}
so that we do not unmap before killing the dma

Perhaps you could send that as a separate patch.

> Is there somewhere else I'm missing?

Testing ;-)