ipq5018: use correct board for GL-iNet B3000 IPQ5018 radio

[psyke83]

This is created using the real device board file (bdwlan.b23) instead of the generic board file (bdwlan.bin) present in vendor firmware. The new file solves signal and throughput issues I was having only with this radio.

$ cat vendor/vendor_dmesg.txt  | grep bdwlan
[   34.131533] cnss[2]: INFO: Downloading BDF: IPQ5018/bdwlan.b23, size: 131072
[   34.721213] cnss[42]: INFO: Downloading BDF: qcn6122/bdwlan.b60, size: 131072

Further confirmation:

# For IPQ5018 - use bdwlan.bin (the generic vendor board that is not used at runtime):
$ rm -f 'bus=ahb,qmi-chip-id=0,qmi-board-id=255,variant=GL-iNet-GL-B3000.bin'
$ cp ~/work/openwrt/vendor-firmware/IPQ5018/WIFI_FW/bdwlan.bin 'bus=ahb,qmi-chip-id=0,qmi-board-id=255,variant=GL-iNet-GL-B3000.bin'
./ath11k-bdencoder -c board-2.json -o board-glinet_gl-b3000.ipq5018.OLD
board binary file 'board-glinet_gl-b3000.ipq5018.OLD' is created
# Compare with current in openwrt tree:
$ sha256sum board-glinet_gl-b3000.ipq5018.OLD ~/work/openwrt/firmware_qca-wireless/board-glinet_gl-b3000.ipq5018
776410c3c8a7ae657c5b9cbafd726fe16add54a573d35526b8fca343e21c0fc2  board-glinet_gl-b3000.ipq5018.OLD
776410c3c8a7ae657c5b9cbafd726fe16add54a573d35526b8fca343e21c0fc2  /home/conn/work/openwrt/firmware_qca-wireless/board-glinet_gl-b3000.ipq5018

# For IPQ5018 - use bdwlan.b23 (the correct one vendor uses):
$ rm -f 'bus=ahb,qmi-chip-id=0,qmi-board-id=255,variant=GL-iNet-GL-B3000.bin'
$ cp ~/work/openwrt/vendor-firmware/IPQ5018/WIFI_FW/bdwlan.b23 'bus=ahb,qmi-chip-id=0,qmi-board-id=255,variant=GL-iNet-GL-B3000.bin'
./ath11k-bdencoder -c board-2.json -o board-glinet_gl-b3000.ipq5018.NEW
board binary file 'board-glinet_gl-b3000.ipq5018.NEW' is created
# Compare with current in openwrt tree:
$ sha256sum board-glinet_gl-b3000.ipq5018.NEW ~/work/openwrt/firmware_qca-wireless/board-glinet_gl-b3000.ipq5018
685184ace8ab091228c9546ea86e5b8613169c147c2c795bc7a4d7506921ae1c  board-glinet_gl-b3000.ipq5018.NEW
776410c3c8a7ae657c5b9cbafd726fe16add54a573d35526b8fca343e21c0fc2  /home/conn/work/openwrt/firmware_qca-wireless/board-glinet_gl-b3000.ipq5018

# For QCN6122 - use bdwlan.b60 (the correct one):
$ rm 'bus=ahb,qmi-chip-id=0,qmi-board-id=255,variant=GL-iNet-GL-B3000.bin'
$ cp ~/work/openwrt/vendor-firmware/IPQ5018/WIFI_FW/qcn6122/bdwlan.b60 'bus=ahb,qmi-chip-id=0,qmi-board-id=255,variant=GL-iNet-GL-B3000.bin'
./ath11k-bdencoder -c board-2.json -o board-glinet_gl-b3000.qcn6122.NEW
board binary file 'board-glinet_gl-b3000.qcn6122.NEW' is created
# Compare with current in openwrt tree:
$ sha256sum board-glinet_gl-b3000.qcn6122.NEW ~/work/openwrt/firmware_qca-wireless/board-glinet_gl-b3000.qcn6122
8d8192d06a06c29c2fc4e7de228040490e7a18edba23c7e27c1090ff1cb91e14  board-glinet_gl-b3000.qcn6122.NEW
8d8192d06a06c29c2fc4e7de228040490e7a18edba23c7e27c1090ff1cb91e14  /home/conn/work/openwrt/firmware_qca-wireless/board-glinet_gl-b3000.qcn6122

[TheRootED24]

Those are the same board files used already ? perhaps yours are from a newer firmware ?? The bdfs 23 & 60 I used where from oem firmware openwrt-b3000-4.5.18-0731-1722397535.img.

EDIT

sha256sum board-glinet_gl-b3000.ipq5018
776410c3c8a7ae657c5b9cbafd726fe16add54a573d35526b8fca343e21c0fc2 board-glinet_gl-b3000.ipq5018

After going through them and comparing ...seems you are correct, the 5018 bdf is currently incorrect. Not sure how it got mixed, good catch in any case.

@robimarko should the pr be sent to firmware_qca-wireless instead ?

[robimarko]

Yes, this should be corrected directly in firmware_qca-wireless.

Once CI passes I will merge this

[psyke83]

To clarify, the files were sourced from: https://fw.gl-inet.com/firmware/b3000/testing/openwrt-b3000-4.7.15-0801-1754021959.img

I didn't check the stable release, but the generic ipq5018 and real qca6122 board files match what you submitted.

I could tell immediately this was right as the average signal of my farthest client improved from -61 to -49 dBm.

[TheRootED24]

yes, somehow i overwrote contents of my bdwlan.b23 with bdwlan.bin. I think it was when I was trying to get bootargs going .. regardless, I was under the impression i had submitted both .b23 and .b60 . I likely never would have notice had you not investigated and pointed out the sha256sum mismatch 👍

I could tell immediately this was right as the average signal of my farthest client improved from -61 to -49 dBm.

I have been saying all along that the noise level did not seem correct, in my setup, my test devices are all within excellent range, so i never really notice the tx power loss.

-63/-107 dBm   <-- current bdf (bdwlan.bin) 

-45/-94  dBm   <-- with proper bdf (bdwlan.b23) 

everything looks correct now

[robimarko]

Thanks! Rebased on top of main and merged!