From: Vinod Koul <vkoul@kernel.org>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: dmaengine@vger.kernel.org, Michal Simek <michal.simek@xilinx.com>,
Hyun Kwon <hyun.kwon@xilinx.com>,
Tejas Upadhyay <tejasu@xilinx.com>,
Satish Kumar Nagireddy <SATISHNA@xilinx.com>
Subject: Re: [PATCH v2 2/4] dma: xilinx: dpdma: Add the Xilinx DisplayPort DMA engine driver
Date: Thu, 5 Dec 2019 22:09:09 +0530 [thread overview]
Message-ID: <20191205163909.GH82508@vkoul-mobl> (raw)
In-Reply-To: <20191205150407.GL4734@pendragon.ideasonboard.com>
Hi Laurent,
On 05-12-19, 17:04, Laurent Pinchart wrote:
> > > +/*
> > > + * DPDMA descriptor placement
> > > + * --------------------------
> > > + * DPDMA descritpor life time is described with following placements:
> > > + *
> > > + * allocated_desc -> submitted_desc -> pending_desc -> active_desc -> done_list
> > > + *
> > > + * Transition is triggered as following:
> > > + *
> > > + * -> allocated_desc : a descriptor allocation
> > > + * allocated_desc -> submitted_desc: a descriptor submission
> > > + * submitted_desc -> pending_desc: request to issue pending a descriptor
> > > + * pending_desc -> active_desc: VSYNC intr when a desc is scheduled to DPDMA
> > > + * active_desc -> done_list: VSYNC intr when DPDMA switches to a new desc
> >
> > Well this tells me driver is not using vchan infrastructure, the
> > drivers/dma/virt-dma.c is common infra which does pretty decent list
> > management and drivers do not need to open code this.
> >
> > Please convert the driver to use virt-dma
>
> As noted in the cover letter,
>
> "There is one review comment that is still pending: switching to
> virt-dma. I started investigating this, and it quickly appeared that
> this would result in an almost complete rewrite of the driver's logic.
> While the end result may be better, I wonder if this is worth it, given
> that the DPDMA is tied to the DisplayPort subsystem and can't be used
> with other DMA slaves. The DPDMA is thus used with very specific usage
> patterns, which don't need the genericity of descriptor handling
> provided by virt-dma. Vinod, what's your opinion on this ? Is virt-dma
> usage a blocker to merge this driver, could we switch to it later, or is
> it just overkill in this case ?"
>
> I'd like to ask an additional question : is the dmaengine API the best
> solution for this ? The DPDMA is a separate IP core, but it is tied with
> the DP subsystem. I'm tempted to just fold it in the display driver. The
> only reason why I'm hesitant on this is that the DPDMA also handles
> audio channels, that are also part of the DP subsystem, but that could
> be handled by a separate ALSA driver. Still, handling display, audio and
> DMA in drivers that we pretend are independent and generic would be a
> bit of a lie.
Yeah if it is _only_ going to be used in display and no other client
using it, then I really do not see any advantage of this being a
dmaengine driver. That is pretty much we have been telling folks over
the years.
Btw since this is xilinx and I guess everything is an IP how difficult
would it be to put this on a non display core :)
If you decide to use dmaengine I would prefer it use virt-dma that mean
rewrite yes but helps you term
--
~Vinod
next prev parent reply other threads:[~2019-12-05 16:39 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-11-07 2:13 [PATCH v2 0/4] dma: Add Xilinx ZynqMP DPDMA driver Laurent Pinchart
2019-11-07 2:13 ` [PATCH v2 1/4] dt: bindings: dma: xilinx: dpdma: DT bindings for Xilinx DPDMA Laurent Pinchart
2019-11-13 13:24 ` Rob Herring
2019-11-07 2:13 ` [PATCH v2 2/4] dma: xilinx: dpdma: Add the Xilinx DisplayPort DMA engine driver Laurent Pinchart
2019-11-09 17:59 ` Vinod Koul
2019-12-05 15:04 ` Laurent Pinchart
2019-12-05 16:39 ` Vinod Koul [this message]
2019-12-05 20:27 ` Hyun Kwon
2019-12-08 16:03 ` Laurent Pinchart
2019-12-20 5:13 ` Laurent Pinchart
2019-12-20 8:01 ` Vinod Koul
2019-12-20 12:35 ` Laurent Pinchart
2019-12-20 15:40 ` Vinod Koul
2019-12-20 16:02 ` Laurent Pinchart
2020-01-03 0:59 ` Laurent Pinchart
2020-01-09 15:57 ` Laurent Pinchart
2020-01-10 7:41 ` Vinod Koul
2020-01-22 16:52 ` Laurent Pinchart
2019-11-07 2:13 ` [PATCH v2 3/4] dma: xilinx: dpdma: Add debugfs support Laurent Pinchart
2019-11-07 2:14 ` [PATCH v2 4/4] arm64: dts: zynqmp: Add DPDMA node Laurent Pinchart
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=20191205163909.GH82508@vkoul-mobl \
--to=vkoul@kernel.org \
--cc=SATISHNA@xilinx.com \
--cc=dmaengine@vger.kernel.org \
--cc=hyun.kwon@xilinx.com \
--cc=laurent.pinchart@ideasonboard.com \
--cc=michal.simek@xilinx.com \
--cc=tejasu@xilinx.com \
/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;
as well as URLs for NNTP newsgroup(s).