Jetpack/kernel/nvidia/Documentation/devicetree/bindings/platform/tegra/tegra-eqos.txt

177 lines
8.0 KiB
Plaintext
Raw Permalink Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

Device tree binding for NVIDIA Tegra186 Ethernet QOS Controller
=================================================================
The Tegra186 EQOS (Ethernet QOS) device is an Ethernet controller which
supports Quality of Service features.
Supported Hardware Features are
=====================================================/
10/100 Mbps Support : YES
1000 Mbps Support : YES
Half-duplex Support : YES
PCS Registers(TBI/SGMII/RTBI PHY interface) : NO
VLAN Hash Filter Selected : NO
SMA (MDIO) Interface : YES
PMT Remote Wake-up Packet Enable : YES
PMT Magic Packet Enable : YES
RMON/MMC Module Enable : YES
ARP Offload Enabled : YES
IEEE 1588-2008 Timestamp Enabled : YES
Energy Efficient Ethernet Enabled : YES
Transmit Checksum Offload Enabled : YES
Receive Checksum Offload Enabled : YES
MAC Addresses 1631 Selected : YES
MAC Addresses 3263 Selected : YES
MAC Addresses 64127 Selected : YES
Timestamp System Time Source : INTERNAL
Source Address or VLAN Insertion Enable : YES
Active PHY Selected : RGMII
MTL Receive FIFO Size : 16 KBytes
MTL Transmit FIFO Size : 16 KBytes
IEEE 1588 High Word Register Enable : YES
DCB Feature Enable : NO
Split Header Feature Enable : YES
TCP Segmentation Offload Enable : YES
DMA Debug Registers Enabled : YES
AV Feature Enabled : YES
Low Power Mode Enabled : YES
Hash Table Size : No hash table selected
Total number of L3 or L4 Filters : 8 L3/L4 Filter
Number of MTL Receive Queues : 4
Number of MTL Transmit Queues : 4
Number of DMA Receive Channels : 4
Number of DMA Transmit Channels : 4
Number of PPS Outputs : No PPS output
Number of Auxiliary Snapshot Inputs : 1 auxiliary input
=====================================================/
Required properties:
- compatible: should be "nvidia,eqos"
- reg: Physical base address and length of registers.
- interrupts: An array of tuples specifying IRQ number for
each tx/rx queue in case of multi-queue, and
for the only tx/rx queue,in case of single queue.
please check example at end for reference.
- clocks: Specifies the needed clocks.
- clock-names: Specifies the names of the needed clocks.
- resets: Specifies the reset line.
- reset-names: Specifies the name of the reset line.
- iommu-group-id: Specifies IOMMU group ID.
- nvidia,csr_clock_speed: This is a free-running clock input provided
by the application. The MAC Control Interface,
CSR, and Station Management Agent (SMA) of the
MAC run on this clock. The valid frequency range
of this clock is 20300 MHz (minimum frequency
is 25 MHz when the MMC module is active in
1000 Mbps mode). Frequencies outside this range
may result in incorrect operation
- nvidia,use_multi_queues: EQOS HW supports upto 4 Tx/Rx Queues/Channels.
This property enables us to make use of 4Queues,
each queue size is configured to 4KB. if this
property is not defined then EQOS HW is
configured to make use of a single queue with a
total of 16KB size.To support Jumbo frame
feature we makes use of a single queue so that
9k packet size can be fit into a single queue.
- nvidia,ptp_ref_clock_speed: This is reference for PTP clock frequency and
the value set for tegra is 125MHz.
The frequency of this clock is programmable and
can be configured based on the requirements.
The maximum frequency it can take is 125MHz and
minimum frequency is 25MHz
- nvidia,queue_prio: An array where each entry specifies priority of
rx queue. Based on the VLAN tag PCP priority
value rx packets will be routed the corresponding
channel.
For example - If the values are "0 7 2 3", it
means that packets with priority 0 will be
mapped to DMA channel 0, similarly 7 will be
mapped to 1 and so on.
"nvidia,queue_prio = <0 7 2 3> " means
priority-0 = DMA-0, priority-7 = DMA-1,
priority-2 = DMA-2, priority-3 = DMA-3
- nvidia,phy-reset-gpio: If this gpio is provided, then the PHY is reset
by toggling this gpio LOW during driver probe.
- nvidia,phy-intr-gpio: Specifies gpio pin for PHY interrupts.
- nvidia,rxq_enable_ctrl: Mode of the RxQ to enabled either AV mode or legacy mode.
example: rxq_enable_ctrl = <rx0 rx1 rx2 rx3>
0x0 = Not enabled, 0x1 = Enabled for AV
0x2 = Enabled for Legacy, 0x3 = Reserved
- nvidia,phy-max-frame-size: Specifies maximum frame size supported by PHY
in kilobytes. Default is 10kb. It Will determine
jumbo frame size.
- phy-mode: should be "rgmii".
- pinctrl-names If we define these property, then we can control
EQOS pin-mux states dynamically so that we can
save power based on the Ethernet link.
- pinctrl-0 Valid only when "pinctrl-names" is defined, this
property has the configuration that need to be
set to EQOS signals when Ethernet link is down.
- pinctrl-1 Valid only when "pinctrl-names" is defined, this
property has the configuration that need to be
set when Ethernet link is up.
Optional properties:
- nvidia,use_tagged_ptp: if this property is set we will make use of the
channel/Queue which is defined in "ptp_dma_ch"
for routing the rx ptp packets.
- nvidia,ptp_dma_ch: Specifies channel/Queue number where the rx PTP
packets need to be be routed.
Channel 0 is default.
- nvidia,chan_napi_quota: Specifies number of tx/rx completions to process
when napi handler is called. Default is "64".
- nvidia,pause_frames: Used to disable/enable PAUSE frames support.
This is a flag. If this property is present
then PAUSE frames will be enabled.
- nvidia,brcm_phy_apd_mode: Flag specifies the Auto Power Down mode for
BRCM89610 PHY.
- nvidia,iso_bw: ISO BW, indicates sum of read and write bandwidth.
- nvidia,eth_iso_enable: Specifies whether ISO enabled for EQOS or not.
- nvidia,slot_intvl_val: Specifies slot interval for fetching the data from DMA.
In T186 EQOS - 125usec is default and fixed value.
In T194 EQOS this value can be changed based on the usecase.
- nvidia,rx_riwt: Specifies the RX wathcdog interrupt timeout in usec
for an Rx descriptor for which IOC bit is not set.
The RIWT field is programmed as (<val> * 256) / 62.5
Example:
ether_qos@2490000 {
compatible = "nvidia,eqos";
reg = <0x0 0x02490000 0x0 0x10000>; /* EQOS Base Register */
reg-names = "eqos_base";
interrupts = <0 194 0x4>, /* common */
<0 195 0x4>, /* power */
<0 190 0x4>, /* rx0 */
<0 186 0x4>, /* tx0 */
<0 191 0x4>, /* rx1 */
<0 187 0x4>, /* tx1 */
<0 192 0x4>, /* rx2 */
<0 188 0x4>, /* tx2 */
<0 193 0x4>, /* rx3 */
<0 189 0x4>; /* tx3 */
nvidia,csr_clock_speed = <0x19>; /* CSR clock speed 25MHz */
clocks = <&tegra_car TEGRA186_CLK_EQOS_AXI>,
<&tegra_car TEGRA186_CLK_EQOS_RX>,
<&tegra_car TEGRA186_CLK_EQOS_PTP_REF>,
<&tegra_car TEGRA186_CLK_EQOS_TX>,
<&tegra_car TEGRA186_CLK_AXI_CBB>;
clock-names = "eqos_axi", "eqos_rx", "eqos_ptp_ref", "eqos_tx", "axi_cbb";
resets = <&tegra_car TEGRA186_RESET_EQOS>;
reset-names = "eqos_rst";
phy-mode = "rgmii";
status = "disabled";
pinctrl-names = "idle", "default";
pinctrl-0 = <&eqos_tx_tri_state_idle>;
pinctrl-1 = <&eqos_tx_tri_state_default>;
nvidia,queue_prio = <0 1 2 3>;
nvidia,use_tagged_ptp;
nvidia,chan_napi_quota = <64 64 64 64>;
nvidia,ptp_dma_ch = <3>;
nvidia,phy-max-frame-size = <10>;
nvidia,ptp_ref_clock_speed = <125>;
nvidia,use_multi_queues
nvidia,brcm_phy_apd_mode;
}