[4/6] dma: tegra: add accurate reporting of dma state

[Dmitry Osipenko <digetx@gmail.com>]

On 05.11.2018 12:03, Ben Dooks wrote:
> 
> 
> On 2018-11-03 12:24, Dmitry Osipenko wrote:
>> On 31.10.2018 19:03, Ben Dooks wrote:
>>> The tx_status callback does not report the state of the transfer
>>> beyond complete segments. This causes problems with users such as
>>> ALSA when applications want to know accurately how much data has
>>> been moved.
>>>
>>> This patch addes a function tegra_dma_update_residual() to query
>>> the hardware and modify the residual information accordinly. It
>>> takes into account any hardware issues when trying to read the
>>> state, such as delays between finishing a buffer and signalling
>>> the interrupt.
>>>
>>> Signed-off-by: Ben Dooks <ben.dooks@codethink.co.uk>
>>
>> Hello Ben,
>>
>> Thank you very much for the patch! It fixes "farting sound" for
>> chromium-browser and applications that use chromium-engine (tested on
>> Tegra20) because apparently it tries to use low latency for everything
>> and audio buffer is constantly underflowing without more detailed
>> DMA-progress reporting. See couple more comments below.
>>
>>> ---
>>>  drivers/dma/tegra20-apb-dma.c | 94 ++++++++++++++++++++++++++++++++---
>>>  1 file changed, 87 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/drivers/dma/tegra20-apb-dma.c b/drivers/dma/tegra20-apb-dma.c
>>> index 4f7d1e576d03..3fa3a1ac4f57 100644
>>> --- a/drivers/dma/tegra20-apb-dma.c
>>> +++ b/drivers/dma/tegra20-apb-dma.c
>>> @@ -802,12 +802,96 @@ static int tegra_dma_terminate_all(struct dma_chan *dc)
>>>      return 0;
>>>  }
>>>
>>> +static unsigned int tegra_dma_update_residual(struct tegra_dma_channel *tdc,
>>> +                          struct tegra_dma_sg_req *sg_req,
>>> +                          struct tegra_dma_desc *dma_desc,
>>> +                          unsigned int residual)
>>> +{
>>> +    unsigned long status = 0x0;
>>
>> There is no need to initialize "status" variable.
> 
> ok, will check this.
> 
>>> +    unsigned long wcount;
>>> +    unsigned long ahbptr;
>>> +    unsigned long tmp = 0x0;
>>> +    unsigned int result;
>>> +    int retries = TEGRA_APBDMA_BURST_COMPLETE_TIME * 10;
>>> +    int done;
>>> +
>>> +    /* if we're not the current request, then don't alter the residual */
>>> +    if (sg_req != list_first_entry(&tdc->pending_sg_req,
>>> +                       struct tegra_dma_sg_req, node)) {
>>> +        result = residual;
>>> +        ahbptr = 0xffffffff;
>>> +        goto done;
>>> +    }
>>> +
>>> +    /* loop until we have a reliable result for residual */
>>> +    do {
>>> +        ahbptr = tdc_read(tdc, TEGRA_APBDMA_CHAN_AHBPTR);
>>> +        status = tdc_read(tdc, TEGRA_APBDMA_CHAN_STATUS);
>>
>>> +        tmp =  tdc_read(tdc, 0x08);    /* total count for debug */
>>
>> Register 0x08 (DMA_BYTE_STA) doesn't present on Tegra20 and "tmp"
>> isn't used anywhere in the code. Please remove it entirely.
> 
> ok, fixed
> 
>>> +
>>> +        /* check status, if channel isn't busy then skip */
>>> +        if (!(status & TEGRA_APBDMA_STATUS_BUSY)) {
>>> +            result = residual;
>>> +            break;
>>> +        }
>>
>> If "BUSY" is unset, doesn't this mean that transaction could be
>> completed already? I don't quite understand why you want to skip here.
> 
> I can't remember what the reasoning behind this was, this work was
> originally done over a year ago and I am not sure if I can find any
> of the notes from this.
> 

Looks like it should be safe to remove this hunk. Please consider the removal or provide reasoning in the comment to the code if you'll recall it.

>>> +
>>> +        /* if we've got an interrupt pending on the channel, don't
>>> +         * try and deal with the residue as the hardware has likely
>>> +         * moved on to the next buffer. return all data moved.
>>> +         */
>>> +        if (status & TEGRA_APBDMA_STATUS_ISE_EOC) {
>>> +            result = residual - sg_req->req_len;
>>> +            break;
>>> +        }
>>> +
>>> +        if (tdc->tdma->chip_data->support_separate_wcount_reg)
>>> +            wcount = tdc_read(tdc, TEGRA_APBDMA_CHAN_WORD_TRANSFER);
>>> +        else
>>> +            wcount = status;
>>> +
>>> +        /* If the request is at the full point, then there is a
>>> +         * chance that we have read the status register in the
>>> +         * middle of the hardware reloading the next buffer.
>>> +         *
>>> +         * The sequence seems to be at the end of the buffer, to
>>> +         * load the new word count before raising the EOC flag (or
>>> +         * changing the ping-pong flag which could have also been
>>> +         * used to determine a new buffer). This  means there is a
>>                                two whitespaces here-----|
>>
>>> +         * small window where we cannot determine zero-done for the
>>> +         * current buffer, or moved to next buffer.
>>> +         *
>>
>>> +         * If done shows 0, then retry the load, as it may hit the
>>> +         * above hardware race. We will either get a new value which
>>> +         * is from the first buffer, or we get an EOC (new buffer)
>>> +         * or both a new value and an EOC...
>>
>> I think we just need to wait 20usec after reading out "words count"
>> and then re-check interrupt status, so transfer is done if interrupt
>> is set and otherwise "words count" value is actual and reliable.
> 
> At the moment I have no way of going back and re-testing this code
> against the problem it originally fixed, so would rather not change
> the algorithm in this.

Please tell what was the problem.

>>
>>> +         */
>>> +        done = get_current_xferred_count(tdc, sg_req, wcount);
>>> +        if (done != 0) {
>>> +            result = residual - done;
>>> +            break;
>>> +        }
>>> +
>>> +        ndelay(100);
>>
>> There is no ndelay() on ARM, hence your 20usec timeout is 200usec.
>> Please use udelay().
> 
> I thought there was one based on the lpj calculations, I will check later.
> 
>>> +    } while (--retries > 0);
>>> +
>>> +    if (retries <= 0) {
>>> +        dev_err(tdc2dev(tdc), "timeout waiting for dma load\n");
>>> +        result = residual;
>>> +    }
>>> +
>>> +done:
>>
>> Please rename goto label as it duplicates local variable name.
> 
> ok, fixed.
> 
>>> +    dev_dbg(tdc2dev(tdc), "residual: req %08lx, ahb@%08lx, wcount %08lx, done %d\n",
>>> +         sg_req->ch_regs.ahb_ptr, ahbptr, wcount, done);
>>
>> Whitespace just after tab not needed.
>>
>>> +
>>> +    return result;
>>> +}
>>> +
>>>  static enum dma_status tegra_dma_tx_status(struct dma_chan *dc,
>>>      dma_cookie_t cookie, struct dma_tx_state *txstate)
>>>  {
>>>      struct tegra_dma_channel *tdc = to_tegra_dma_chan(dc);
>>>      struct tegra_dma_desc *dma_desc;
>>> -    struct tegra_dma_sg_req *sg_req;
>>> +    struct tegra_dma_sg_req *sg_req = NULL;
>>>      enum dma_status ret;
>>>      unsigned long flags;
>>>      unsigned int residual;
>>> @@ -843,6 +927,7 @@ static enum dma_status tegra_dma_tx_status(struct dma_chan *dc,
>>>          residual = dma_desc->bytes_requested -
>>>                 (dma_desc->bytes_transferred %
>>>                  dma_desc->bytes_requested);
>>> +        residual = tegra_dma_update_residual(tdc, sg_req, dma_desc, residual);

Line over 80 characters. Please make sure that "scripts/checkpatch.pl" doesn't have valid complains about the patches.

[snip]