test: restore MCU1 cache fix for V3 firmware

This commit is contained in:
Shen Mintao
2026-07-23 09:10:43 +08:00
parent 7b2541e5c0
commit 027a7a8662
2 changed files with 42 additions and 17 deletions
@@ -234,6 +234,7 @@ int system_config_8800d80(struct aic_usb_dev *usb_dev){
int syscfg_num;
int ret, cnt;
const u32 mem_addr = 0x40500000;
const u32 cache_mem_addr = 0x40100020;
struct dbg_mem_read_cfm rd_mem_addr_cfm;
ret = rwnx_send_dbg_mem_read_req(usb_dev, mem_addr, &rd_mem_addr_cfm);
if (ret) {
@@ -245,7 +246,20 @@ int system_config_8800d80(struct aic_usb_dev *usb_dev){
}
chip_id = (u8)(rd_mem_addr_cfm.memdata >> 16);
printk("chip_id=%x, chip_mcu_id = %d\n", chip_id, chip_mcu_id);
printk("issue58: using Radxa SDK V3 D80 loader profile\n");
printk("issue58: using Radxa SDK V3 D80 loader profile with MCU1 cache fix\n");
if (chip_mcu_id == 1) {
ret = rwnx_send_dbg_mem_read_req(usb_dev, cache_mem_addr, &rd_mem_addr_cfm);
if (ret) {
printk("%x rd fail: %d\n", cache_mem_addr, ret);
return ret;
}
rd_mem_addr_cfm.memdata |= 0x01;
ret = rwnx_send_dbg_mem_write_req(usb_dev, cache_mem_addr, rd_mem_addr_cfm.memdata);
if (ret) {
printk("%x write fail: %d\n", cache_mem_addr, ret);
return ret;
}
}
#if 1
syscfg_num = sizeof(syscfg_tbl_8800d80) / sizeof(u32) / 2;
for (cnt = 0; cnt < syscfg_num; cnt++) {
+27 -16
View File
@@ -1,4 +1,4 @@
# Issue #58: legacy firmware test for D80 MCU revision 1
# Issue #58: V3 firmware and MCU1 Bluetooth cache test
This branch is an experimental build for
[issue #58](https://github.com/shenmintao/aic8800d80/issues/58). It is intended
@@ -10,10 +10,11 @@ chip_id=7, chip_mcu_id=1
Do not merge this branch as a general firmware downgrade. It replaces the
complete `fw/aic8800D80` firmware set and matches the D80 loader's FMAC patch
table, patch-buffer layout, and MCU cache setup to that firmware generation.
The USB transport and modern-kernel compatibility code remain current.
Firmware and loader paths for D80N, D80X2, DC, and other variants are not
changed.
table and patch-buffer layout to that firmware generation. It additionally
sets bit 0 of register `0x40100020` only when `chip_mcu_id=1`, as tested in
[PR #35](https://github.com/shenmintao/aic8800d80/pull/35). The USB transport
and modern-kernel compatibility code remain current. Firmware and loader paths
for D80N, D80X2, DC, and other variants are not changed.
## Hypothesis
@@ -32,11 +33,17 @@ fixed at upstream commit
The normal FMAC file has SHA-256
`1ec680c2b63dcaa0e5d33c5fb6d1857d030f8145c05c385e243760388a61a0da`.
The first hardware test proved that this image uploads completely without the
`0x170400` timeout, but the device then failed to re-enumerate while the newer
loader was still applying SDK V5-era FMAC patches and MCU cache setup. This
follow-up matches those loader operations to V3 so that their effect can be
tested separately, while retaining unrelated fixes in the current driver.
Hardware testing of commit `7b2541e` proved that the matched V3 firmware and
loader upload completely without the `0x170400` timeout, re-enumerate as
`a69c:8d81`, and provide working 2.4 GHz and 5 GHz Wi-Fi. Its Bluetooth
interfaces bind to the kernel's standard `btusb` driver, but HCI initialization
fails with `Opcode 0x0c03 failed: -110`.
This follow-up changes one loader behavior only: for MCU revision 1 it reads
register `0x40100020`, sets bit 0, and writes the value back before firmware
upload. PR #35 independently found that this is required for correct Bluetooth
firmware block writes on MCU1 and verified it with the system `btusb` driver.
This branch does not contain or install `aic_btusb`.
## Install the test branch
@@ -44,7 +51,7 @@ From an existing clone:
```bash
git fetch origin
git switch test/issue-58-mcu1-legacy-fw
git switch test/issue-58-mcu1-v3-fw-cache
git pull --ff-only
sudo ./install.sh
```
@@ -61,8 +68,8 @@ re-enumerates after `a69c:8d80`:
```bash
sudo dmesg -C
# Disconnect and reconnect the device, then wait for initialization.
sudo dmesg | tee issue58-legacy-fw-dmesg.txt
sudo dmesg | grep -iE 'aic|issue58|chip_id|chip_mcu_id|fmacfw|bin upload|cmd timed-out|error -110'
sudo dmesg | tee issue58-v3-cache-dmesg.txt
sudo dmesg | grep -iE 'aic|issue58|chip_id|chip_mcu_id|fmacfw|bin upload|cmd timed-out|Bluetooth|btusb|0x0c03|error -110'
lsusb
lsusb -t
```
@@ -70,7 +77,7 @@ lsusb -t
The log must contain:
```text
issue58: using Radxa SDK V3 D80 loader profile
issue58: using Radxa SDK V3 D80 loader profile with MCU1 cache fix
```
Please report all of the following, even if an earlier item fails:
@@ -83,8 +90,10 @@ Please report all of the following, even if an earlier item fails:
5. Whether the interface receives an address by DHCP.
6. Whether the gateway and an Internet address can be pinged, and whether real
traffic works.
7. Whether Bluetooth still enumerates and works through the kernel's standard
`btusb` driver. This test does not install or use `aic_btusb`.
7. Whether both Bluetooth interfaces are bound to the kernel's standard
`btusb` driver and the earlier HCI Reset timeout is gone.
8. Whether Bluetooth can scan, pair, and establish a real connection. This
test does not install or use `aic_btusb`.
Useful commands:
@@ -96,6 +105,8 @@ ip address
ip route
lsusb -t
bluetoothctl list
bluetoothctl show
bluetoothctl scan on
```
Identify the new AIC interface with `iw dev`, then replace `wlan0` below with