netdev - Re: [PATCH net-next v2 1/2] dt-bindings: net: xilinx

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <c0df773ed9bc07172f38760fd1e4509b6164e790.camel@calian.com>
Date:   Fri, 12 Mar 2021 19:44:03 +0000
From:   Robert Hancock <robert.hancock@...ian.com>
To:     "radheys@...inx.com" <radheys@...inx.com>,
        "davem@...emloft.net" <davem@...emloft.net>,
        "kuba@...nel.org" <kuba@...nel.org>
CC:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "devicetree@...r.kernel.org" <devicetree@...r.kernel.org>
Subject: Re: [PATCH net-next v2 1/2] dt-bindings: net: xilinx_axienet:
 Document additional clocks

On Fri, 2021-03-12 at 18:39 +0000, Radhey Shyam Pandey wrote:
> > -----Original Message-----
> > From: Robert Hancock <robert.hancock@...ian.com>
> > Sent: Friday, March 12, 2021 11:13 PM
> > To: Radhey Shyam Pandey <radheys@...inx.com>; davem@...emloft.net;
> > kuba@...nel.org
> > Cc: netdev@...r.kernel.org; devicetree@...r.kernel.org; Robert Hancock
> > <robert.hancock@...ian.com>
> > Subject: [PATCH net-next v2 1/2] dt-bindings: net: xilinx_axienet: Document
> > additional clocks
> > 
> > Update DT bindings to describe all of the clocks that the axienet driver
> > will
> > now be able to make use of.
> > 
> > Signed-off-by: Robert Hancock <robert.hancock@...ian.com>
> > ---
> >  .../bindings/net/xilinx_axienet.txt           | 25 ++++++++++++++-----
> >  1 file changed, 19 insertions(+), 6 deletions(-)
> > 
> > diff --git a/Documentation/devicetree/bindings/net/xilinx_axienet.txt
> > b/Documentation/devicetree/bindings/net/xilinx_axienet.txt
> > index 2cd452419ed0..5df5ba449b8f 100644
> > --- a/Documentation/devicetree/bindings/net/xilinx_axienet.txt
> > +++ b/Documentation/devicetree/bindings/net/xilinx_axienet.txt
> > @@ -42,11 +42,23 @@ Optional properties:
> >  		  support both 1000BaseX and SGMII modes. If set, the phy-
> > mode
> >  		  should be set to match the mode selected on core reset
> > (i.e.
> >  		  by the basex_or_sgmii core input line).
> > -- clocks	: AXI bus clock for the device. Refer to common clock
> > bindings.
> > -		  Used to calculate MDIO clock divisor. If not specified, it is
> > -		  auto-detected from the CPU clock (but only on platforms
> > where
> > -		  this is possible). New device trees should specify this - the
> > -		  auto detection is only for backward compatibility.
> > +- clock-names: 	  Tuple listing input clock names. Possible clocks:
> > +		  s_axi_lite_clk: Clock for AXI register slave interface
> > +		  axis_clk: AXI stream clock for DMA block
> 
> Description for axis_clk should be-
> axis_clk: AXI4-Stream clock for TXD RXD TXC and RXS interfaces.

Missed this, will add in.

> In this patch I assume we are only adding additional clocks for 
> 1G ethernet subsystem. For dma clocking support we need to add 
> more clocks and better to add them in a separate patch. Please refer to
> xilinx tree.

Correct, here I am just focusing on clocks which are connected to the Ethernet
core itself. It seems the DMA core clocks would be more difficult since those
clocks are attached to the DMA node in the device tree (based on the Vitis-
generated PL device tree content) and not to the node belonging to the Ethernet
platform device. Looking at the Xilinx version of this driver, it seems to
implement retrieving these clocks, but it is using devm_clk_get on the same
platform device used for the Ethernet core, so I am confused how that is
working (or maybe it isn't?)

The DMA core clocks are less of a concern for us, since (at least in our
design) they are driven by the same clock source as one of the Ethernet core
clocks, so they should always be enabled when the Ethernet core clocks are
enabled.

> > +		  ref_clk: Ethernet reference clock, used by signal delay
> > +			   primitives and transceivers
> > +		  mgt_clk: MGT reference clock (used by optional internal
> > +			   PCS/PMA PHY)
> > +
> > +		  Note that if s_axi_lite_clk is not specified by name, the
> > +		  first clock of any name is used for this. If that is also not
> > +		  specified, the clock rate is auto-detected from the CPU
> > clock
> > +		  (but only on platforms where this is possible). New device
> > +		  trees should specify all applicable clocks by name - the
> > +		  fallbacks to an unnamed clock or to CPU clock are only for
> > +		  backward compatibility.
> > +- clocks: 	  Phandles to input clocks matching clock-names. Refer to
> > common
> > +		  clock bindings.
> >  - axistream-connected: Reference to another node which contains the
> > resources
> >  		       for the AXI DMA controller used by this device.
> >  		       If this is specified, the DMA-related resources from
> > that
> > @@ -62,7 +74,8 @@ Example:
> >  		device_type = "network";
> >  		interrupt-parent = <&microblaze_0_axi_intc>;
> >  		interrupts = <2 0 1>;
> > -		clocks = <&axi_clk>;
> > +		clock-names = "s_axi_lite_clk", "axis_clk", "ref_clk",
> > "mgt_clk";
> > +		clocks = <&axi_clk>, <&axi_clk>, <&pl_enet_ref_clk>,
> > <&mgt_clk>;
> >  		phy-mode = "mii";
> >  		reg = <0x40c00000 0x40000 0x50c00000 0x40000>;
> >  		xlnx,rxcsum = <0x2>;
> > --
> > 2.27.0