From: Miquel Raynal <miquel.raynal@bootlin.com>
To: Lizhi Hou <lizhi.hou@amd.com>
Cc: Vinod Koul <vkoul@kernel.org>, Brian Xu <brian.xu@amd.com>,
Raj Kumar Rampelli <raj.kumar.rampelli@amd.com>,
<dmaengine@vger.kernel.org>, Michal Simek <monstr@monstr.eu>,
Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Subject: Re: [PATCH 1/3] dmaengine: xilinx: xdma: Fix the count of elapsed periods in cyclic mode
Date: Wed, 29 Nov 2023 01:13:44 +0100 [thread overview]
Message-ID: <20231129011344.35450ee3@xps-13> (raw)
In-Reply-To: <3403ac76-db69-d2c2-0bd1-03e8210d309e@amd.com>
Hi Lizhi,
lizhi.hou@amd.com wrote on Tue, 28 Nov 2023 14:44:35 -0800:
> On 11/24/23 07:09, Miquel Raynal wrote:
> > Xilinx DMA engine is capable of keeping track of the number of elapsed
> > periods and this is an increasing 32-bit counter which is only reset
> > when turning off the engine. No need to add this value to our local
> > counter.
> >
> > Fixes: cd8c732ce1a5 ("dmaengine: xilinx: xdma: Support cyclic transfers")
> > Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
> > ---
> >
> > Hello, so far all my testing was performed by looping the playback
> > output to the recording input and comparing the files using
> > FFTs. Unfortunately, when the DMA engine suffers from the same issue on
> > both sides, some issues may appear un-noticed, which is likely what
> > happened here as the tooling did not report any issue while analyzing
> > the output until I actually listened to real audio now that I have in my
> > hands the relevant hardware/connectors to do so.
> > ---
> > drivers/dma/xilinx/xdma.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/dma/xilinx/xdma.c b/drivers/dma/xilinx/xdma.c
> > index 84a88029226f..75533e787414 100644
> > --- a/drivers/dma/xilinx/xdma.c
> > +++ b/drivers/dma/xilinx/xdma.c
> > @@ -754,7 +754,7 @@ static irqreturn_t xdma_channel_isr(int irq, void *dev_id)
> > if (ret)
> > goto out;
> > > - desc->completed_desc_num += complete_desc_num;
> > + desc->completed_desc_num = complete_desc_num;
>
> Based on PG195, completed descriptor count will be reset to 0 on rising edge of Control register Run bit. That means it resets to zero for each transaction.
>
> This change breaks our long sg list test.
Ok, so we need act differently depending on the type of transfer (sg or
cyclic). Thanks for the feedback!
>
> > > if (desc->cyclic) {
> > ret = regmap_read(xdev->rmap, xchan->base + XDMA_CHAN_STATUS,
Miquèl
next prev parent reply other threads:[~2023-11-29 0:13 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-11-24 15:09 [PATCH 0/3] dmaengine: xilinx: Better cyclic transfers Miquel Raynal
2023-11-24 15:09 ` [PATCH 1/3] dmaengine: xilinx: xdma: Fix the count of elapsed periods in cyclic mode Miquel Raynal
2023-11-28 22:44 ` Lizhi Hou
2023-11-29 0:13 ` Miquel Raynal [this message]
2023-11-24 15:09 ` [PATCH 2/3] dmaengine: xilinx: xdma: Better handling of the busy variable Miquel Raynal
2023-11-24 15:09 ` [PATCH 3/3] dmaengine: xilinx: xdma: Add terminate_all/synchronize callbacks Miquel Raynal
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20231129011344.35450ee3@xps-13 \
--to=miquel.raynal@bootlin.com \
--cc=brian.xu@amd.com \
--cc=dmaengine@vger.kernel.org \
--cc=lizhi.hou@amd.com \
--cc=monstr@monstr.eu \
--cc=raj.kumar.rampelli@amd.com \
--cc=thomas.petazzoni@bootlin.com \
--cc=vkoul@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox