RMII 8b tx by ed-xmos · Pull Request #85 · xmos/lib_ethernet

ed-xmos · 2025-01-09T12:19:09Z

See https://xmosjira.atlassian.net/wiki/spaces/Ethernet/pages/4478271544/8b+port+RMII+Tx+Feasibility

This has been tested in a local 8b_tx -> 1b_rx loopback app only but results are positive so far and it passes at 70MHz min thread speed so just meets our goal of 75MHz. The wrapping logic took this from 60 to 70MHz.

Captured in #88

It has been verified using the tests/bringup app (ping) also in hw.

This still needs the sim/hw tests extended and IFG calculations done as well as extending the docs.

Conflicts: lib_ethernet/api/ethernet.h lib_ethernet/src/rmii_ethernet_rt_mac.xc lib_ethernet/src/rmii_master.h lib_ethernet/src/rmii_master.xc

Conflicts: Jenkinsfile examples/app_rmii_100Mbit_icmp/src/icmp.h examples/app_rmii_100Mbit_icmp/src/icmp.xc examples/app_rmii_100Mbit_icmp/src/main.xc examples/app_rmii_100Mbit_icmp/src/xk-eth-xu316-dual-100m.xn examples/deps.cmake lib_ethernet/api/ethernet.h lib_ethernet/api/smi.h lib_ethernet/src/mii_ethernet_rt_mac.xc lib_ethernet/src/rgmii_buffering.xc lib_ethernet/src/rmii_ethernet_rt_mac.xc lib_ethernet/src/rmii_master.h lib_ethernet/src/rmii_master.xc lib_ethernet/src/smi.xc requirements.txt tests/CMakeLists.txt tests/bringup_xk_eth_xu316_dual_100m/CMakeLists.txt tests/bringup_xk_eth_xu316_dual_100m/src/icmp.h tests/bringup_xk_eth_xu316_dual_100m/src/icmp.xc tests/bringup_xk_eth_xu316_dual_100m/src/main.xc tests/bringup_xk_eth_xu316_dual_100m/src/xk-eth-xu316-dual-100m.xn tests/test_smi/src/main.xc

Conflicts: examples/app_rmii_100Mbit_icmp/src/main.xc lib_ethernet/api/ethernet.h

# Conflicts: # examples/app_ethernet_diagnostics/src/main.xc # lib_ethernet/api/ethernet.h # lib_ethernet/api/smi.h

# Conflicts: # tests/bringup_xk_eth_xu316_dual_100m/CMakeLists.txt

- Extend test_tx, test_time_tx and test_rmii_timing to test 8b TX configs - Updated the TX IFG calculation for 8b TX, although, the turnaround time between consecutive TX is greater than the required IFG wait so the IFG wait is never really exercised to 8b TX

humphrey-xmos

Good stuff

humphrey-xmos · 2025-09-17T13:54:06Z

lib_ethernet/src/rmii_master.xc

+    unsigned time;
+    const unsigned poly = 0xEDB88320;
+    unsigned * unsafe dptr = &buf->data[0];
+    unsigned * unsafe wrap_ptr = mii_get_wrap_ptr(tx_mem);;


Double semi-colon

Copilot

Pull Request Overview

This PR adds support for 8-bit RMII TX functionality to the Ethernet library. The implementation extends the existing RMII TX capabilities to support 8-bit port configurations in addition to the existing 4-bit and 1-bit options.

Key changes:

Adds new 8-bit TX assembly implementation for improved performance
Updates test framework to support 8-bit TX test configurations
Refactors pin assignment handling to support flexible bit positioning for 8-bit ports

Reviewed Changes

Copilot reviewed 26 out of 28 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
lib_ethernet/src/rmii_master_tx_pins_8b.S	New assembly implementation for 8-bit RMII TX with optimized performance
lib_ethernet/src/rmii_master.xc	Core logic updates to support 8-bit TX including lookup tables and packet transmission
lib_ethernet/api/ethernet.h	API updates to support 8-bit port configuration and pin assignment macros
tests/rmii_phy.py	Test infrastructure updates to handle 8-bit port pin assignments
tests/helpers.py	Helper functions for 8-bit RMII PHY configuration
tests/helpers.cmake	CMake macros to parse 8-bit TX width configurations
Various test files	Updated test parameters and configurations to include 8-bit TX test cases

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

Copilot · 2025-09-17T14:39:18Z

tests/test_tx/src/main.xc

+        on tile[0]: {
+          rmii_ethernet_rt_mac( i_cfg, NUM_CFG_IF,


[nitpick] The addition of braces around the rmii_ethernet_rt_mac call creates inconsistent formatting with the MII case above. Consider applying the same formatting to both cases for consistency.

Copilot · 2025-09-17T14:39:18Z

tests/bringup_xk_eth_xu316_dual_100m/src/icmp.h

@@ -4,7 +4,6 @@
 #define __icmp_h__
 #include <ethernet.h>



[nitpick] The removal of blank lines creates inconsistent spacing. The blank line should be preserved to maintain consistent formatting with the rest of the codebase.

Suggested change

Copilot · 2025-09-17T14:39:19Z

lib_ethernet/src/rmii_master.xc

+            case 8: {
+                    time = rmii_transmit_packet_8b(tx_mem, buf, *p_mii_txd_0, lookup_8b_tx, ifg_tmr, ifg_time, eof_time);
+                    eof_time = ifg_time;
+                    ifg_time += RMII_ETHERNET_IFG_AS_REF_CLOCK_COUNT_8b; // TODO WORK ME OUT


The TODO comment indicates incomplete IFG calculations for 8-bit mode. This should be resolved before merging to production as it affects timing correctness.

Copilot · 2025-09-17T14:39:19Z

lib_ethernet/src/rmii_master.xc

+static void init_8b_tx_lookup(unsigned lookup[], int bitpos_0, int bitpos_1){
+    for(int i = 0; i < 256; i++){
+        lookup[i] = 0; // All unused pins driven low
+        lookup[i] |= (i & 0x1) << (bitpos_0 + 0);
+        lookup[i] |= (i & 0x2) << (bitpos_1 - 1);
+        lookup[i] |= (i & 0x4) << (bitpos_0 + 6);
+        lookup[i] |= (i & 0x8) << (bitpos_1 + 5);
+        lookup[i] |= (i & 0x10) << (bitpos_0 + 12);
+        lookup[i] |= (i & 0x20) << (bitpos_1 + 11);
+        lookup[i] |= (i & 0x40) << (bitpos_0 + 18);


The bit manipulation logic uses magic numbers without explanation. Consider adding comments to explain the bit positioning logic, especially the relationship between the input byte bits and output port positions.

Suggested change

static void init_8b_tx_lookup(unsigned lookup[], int bitpos_0, int bitpos_1){

for(int i = 0; i < 256; i++){

lookup[i] = 0; // All unused pins driven low

lookup[i] |= (i & 0x1) << (bitpos_0 + 0);

lookup[i] |= (i & 0x2) << (bitpos_1 - 1);

lookup[i] |= (i & 0x4) << (bitpos_0 + 6);

lookup[i] |= (i & 0x8) << (bitpos_1 + 5);

lookup[i] |= (i & 0x10) << (bitpos_0 + 12);

lookup[i] |= (i & 0x20) << (bitpos_1 + 11);

lookup[i] |= (i & 0x40) << (bitpos_0 + 18);

static void init_8b_tx_lookup(unsigned lookup[], int bitpos_0, int bitpos_1){

/*

* This function creates a lookup table mapping each possible 8-bit value (0-255)

* to a port output value, where each bit in the input byte is mapped to a specific

* output port position. The mapping uses two base bit positions: bitpos_0 and bitpos_1,

* which correspond to the physical pin positions for the two RMII TXD ports.

*

* The mapping is as follows:

* - Input bit 0 (0x01): output at bitpos_0 + 0

* - Input bit 1 (0x02): output at bitpos_1 - 1

* - Input bit 2 (0x04): output at bitpos_0 + 6

* - Input bit 3 (0x08): output at bitpos_1 + 5

* - Input bit 4 (0x10): output at bitpos_0 + 12

* - Input bit 5 (0x20): output at bitpos_1 + 11

* - Input bit 6 (0x40): output at bitpos_0 + 18

* - Input bit 7 (0x80): output at bitpos_1 + 17

*

* This arrangement is determined by the hardware wiring and protocol requirements.

* The use of bitpos_0 and bitpos_1 allows for flexible assignment of output pins.

*/

for(int i = 0; i < 256; i++){

lookup[i] = 0; // All unused pins driven low

// Map input bit 0 (LSB) to output at bitpos_0 + 0

lookup[i] |= (i & 0x1) << (bitpos_0 + 0);

// Map input bit 1 to output at bitpos_1 - 1

lookup[i] |= (i & 0x2) << (bitpos_1 - 1);

// Map input bit 2 to output at bitpos_0 + 6

lookup[i] |= (i & 0x4) << (bitpos_0 + 6);

// Map input bit 3 to output at bitpos_1 + 5

lookup[i] |= (i & 0x8) << (bitpos_1 + 5);

// Map input bit 4 to output at bitpos_0 + 12

lookup[i] |= (i & 0x10) << (bitpos_0 + 12);

// Map input bit 5 to output at bitpos_1 + 11

lookup[i] |= (i & 0x20) << (bitpos_1 + 11);

// Map input bit 6 to output at bitpos_0 + 18

lookup[i] |= (i & 0x40) << (bitpos_0 + 18);

// Map input bit 7 (MSB) to output at bitpos_1 + 17

humphrey-xmos

Sorry, should have approved the other day.
Other than a few formatting changes and maybe clearing up some 'TODO's its good

Initial support for 8b tx in API

2c83d8d

ed-xmos self-assigned this Jan 9, 2025

Successful non-wrapped 8b Tx at 60MHz thread speed

596cef5

ed-xmos added the type:enhancement New feature or request label Jan 10, 2025

ed-xmos added 3 commits January 10, 2025 10:54

Refresh API to avoid extra braces on initialiser

6c0af60

Implement wrap checking for tx8

e8c858c

Comments and minor DI opt

1d266e4

shuchitak mentioned this pull request Jan 16, 2025

RMII TX on a 8b port #88

Open

7 tasks

ed-xmos added 4 commits February 28, 2025 14:47

Merge branch 'develop'

d86ce13

Conflicts: lib_ethernet/api/ethernet.h lib_ethernet/src/rmii_ethernet_rt_mac.xc lib_ethernet/src/rmii_master.h lib_ethernet/src/rmii_master.xc

Tidy merge

93248b9

Tidy bringup app

8f64867

ed-xmos marked this pull request as ready for review February 28, 2025 15:42

ed-xmos added 9 commits February 28, 2025 15:46

Update deps.cmake

e2d1c9c

Tidy up items - merge cleanup and use 8b port initialiser

415904a

Fix cmake in tests

8e73bfd

Copyright

d7d33c4

Merge branch 'develop'

c235a3d

Merge branch 'develop'

0c6bdf8

Merge branch 'develop'

049b9b5

Conflicts: examples/app_rmii_100Mbit_icmp/src/main.xc lib_ethernet/api/ethernet.h

Fix duplicate IP

8ccb40f

Merge branch 'develop'

e9f3d2b

humphrey-xmos mentioned this pull request Aug 12, 2025

Adding dual ethernet 316 dev-kit xmos/lib_board_support#66

Merged

humphrey-xmos added 7 commits August 28, 2025 10:56

Merge branch 'develop' into feature/rmii_8b_tx

781a527

# Conflicts: # examples/app_ethernet_diagnostics/src/main.xc # lib_ethernet/api/ethernet.h # lib_ethernet/api/smi.h

Updating bringup test to latest lib

a42308c

Merge branch 'develop' into feature/rmii_8b_tx

b157f42

Documentation update

5236397

icmp client refactored, brought in from app-note

9302ed0

Fixing bringup test 8b select

0e4e7cd

Merge branch 'develop' into feature/rmii_8b_tx

e81f6db

# Conflicts: # tests/bringup_xk_eth_xu316_dual_100m/CMakeLists.txt

shuchitak requested review from humphrey-xmos and shuchitak September 17, 2025 10:16

humphrey-xmos reviewed Sep 17, 2025

View reviewed changes

humphrey-xmos requested a review from Copilot September 17, 2025 14:37

Copilot AI reviewed Sep 17, 2025

View reviewed changes

humphrey-xmos approved these changes Sep 24, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

RMII 8b tx#85

RMII 8b tx#85
ed-xmos wants to merge 26 commits intoxmos:developfrom
ed-xmos:feature/rmii_8b_tx

ed-xmos commented Jan 9, 2025 •

edited

Loading

Uh oh!

humphrey-xmos left a comment

Uh oh!

humphrey-xmos Sep 17, 2025

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Sep 17, 2025

Uh oh!

Copilot AI Sep 17, 2025

Uh oh!

Copilot AI Sep 17, 2025

Uh oh!

Copilot AI Sep 17, 2025

Uh oh!

humphrey-xmos left a comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-static void init_8b_tx_lookup(unsigned lookup[], int bitpos_0, int bitpos_1){
-    for(int i = 0; i < 256; i++){
-        lookup[i] = 0; // All unused pins driven low
-        lookup[i] |= (i & 0x1) << (bitpos_0 + 0);
-        lookup[i] |= (i & 0x2) << (bitpos_1 - 1);
-        lookup[i] |= (i & 0x4) << (bitpos_0 + 6);
-        lookup[i] |= (i & 0x8) << (bitpos_1 + 5);
-        lookup[i] |= (i & 0x10) << (bitpos_0 + 12);
-        lookup[i] |= (i & 0x20) << (bitpos_1 + 11);
-        lookup[i] |= (i & 0x40) << (bitpos_0 + 18);
+static void init_8b_tx_lookup(unsigned lookup[], int bitpos_0, int bitpos_1){
+    /*
+     * This function creates a lookup table mapping each possible 8-bit value (0-255)
+     * to a port output value, where each bit in the input byte is mapped to a specific
+     * output port position. The mapping uses two base bit positions: bitpos_0 and bitpos_1,
+     * which correspond to the physical pin positions for the two RMII TXD ports.
+     *
+     * The mapping is as follows:
+     *   - Input bit 0 (0x01): output at bitpos_0 + 0
+     *   - Input bit 1 (0x02): output at bitpos_1 - 1
+     *   - Input bit 2 (0x04): output at bitpos_0 + 6
+     *   - Input bit 3 (0x08): output at bitpos_1 + 5
+     *   - Input bit 4 (0x10): output at bitpos_0 + 12
+     *   - Input bit 5 (0x20): output at bitpos_1 + 11
+     *   - Input bit 6 (0x40): output at bitpos_0 + 18
+     *   - Input bit 7 (0x80): output at bitpos_1 + 17
+     *
+     * This arrangement is determined by the hardware wiring and protocol requirements.
+     * The use of bitpos_0 and bitpos_1 allows for flexible assignment of output pins.
+     */
+    for(int i = 0; i < 256; i++){
+        lookup[i] = 0; // All unused pins driven low
+        // Map input bit 0 (LSB) to output at bitpos_0 + 0
+        lookup[i] |= (i & 0x1) << (bitpos_0 + 0);
+        // Map input bit 1 to output at bitpos_1 - 1
+        lookup[i] |= (i & 0x2) << (bitpos_1 - 1);
+        // Map input bit 2 to output at bitpos_0 + 6
+        lookup[i] |= (i & 0x4) << (bitpos_0 + 6);
+        // Map input bit 3 to output at bitpos_1 + 5
+        lookup[i] |= (i & 0x8) << (bitpos_1 + 5);
+        // Map input bit 4 to output at bitpos_0 + 12
+        lookup[i] |= (i & 0x10) << (bitpos_0 + 12);
+        // Map input bit 5 to output at bitpos_1 + 11
+        lookup[i] |= (i & 0x20) << (bitpos_1 + 11);
+        // Map input bit 6 to output at bitpos_0 + 18
+        lookup[i] |= (i & 0x40) << (bitpos_0 + 18);
+        // Map input bit 7 (MSB) to output at bitpos_1 + 17

Conversation

ed-xmos commented Jan 9, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

humphrey-xmos left a comment

Choose a reason for hiding this comment

Uh oh!

humphrey-xmos Sep 17, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Sep 17, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Sep 17, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Sep 17, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Sep 17, 2025

Choose a reason for hiding this comment

Uh oh!

humphrey-xmos left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

ed-xmos commented Jan 9, 2025 •

edited

Loading