I said, QCN BDF and FW!
If you are in the testing mood, then try testing the new 2.7 FW for the QCN card
For the PCI WiFi you need to add this firmware:
and the following BDF:
https://source.codeaurora.org/quic/qsdk/oss/ath11k-bdf/tree/QCN9074/hw1.0/WLAN.HK.2.6.0.1/WLAN.HK.2.6.0.1-00828-QCAHKSWPL_SILICONZ-1?h=NHSS.QSDK.11.5.0.5.r7
Not sure if calibration data related shell scripts handles this device though
1 Like
You cant use that BDF directly, comment the ath11k-variant in the DTS PCI node first.
No point in trying 2.6, try 2.7 beta they released today.
2 Likes
So updated the Makefile to load the test QCN9074 and removed ath11k-variant - wlan0 start but not the other radios
boot messages
[ 11.974021] remoteproc remoteproc0: remote processor cd00000.q6v5_wcss is now up
[ 11.975264] ath11k c000000.wifi: qmi ignore invalid mem req type 3
[ 11.980836] ath11k_pci 0000:01:00.0: BAR 0: assigned [mem 0x20400000-0x205fffff 64bit]
[ 11.980996] ath11k c000000.wifi: chip_id 0x0 chip_family 0x0 board_id 0x294 soc_id 0xffffffff
[ 11.986546] ath11k_pci 0000:01:00.0: enabling device (0000 -> 0002)
[ 11.994389] ath11k c000000.wifi: fw_version 0x260484a5 fw_build_timestamp 2022-03-09 12:43 fw_build_id QC_IMAGE_VERSION_STRING=WLAN.HK.2.6.0.1-00861-QCAHKSWPL_SILICONZ-1
[ 12.003349] ath11k_pci 0000:01:00.0: MSI vectors: 16
[ 12.024374] ath11k_pci 0000:01:00.0: qcn9074 hw1.0
[ 12.187421] mhi mhi0: Requested to power ON
[ 12.187644] mhi mhi0: Power on setup success
[ 12.289222] mhi mhi0: Wait for device to enter SBL or Mission mode
[ 12.335759] ath11k c000000.wifi: host capability request failed: 1 90
[ 12.335799] ath11k c000000.wifi: failed to send qmi host cap: -22
network start - remoteproc crash
502.536259] qcom-q6v5-wcss-pil cd00000.q6v5_wcss: fatal error received:
[ 502.536259] QC Image Version: QC_IMAGE_VERSION_STRING=WLAN.HK.2.6.0.1-00861-QCAHKSWPL_SILICONZ-1
[ 502.536259] Image Variant : IMAGE_VARIANT_STRING=8074.wlanfw.eval_v2Q
[ 502.536259]
[ 502.536259] :Excep :0 Exception detectedparam0 :zero, param1 :zero, param2 :zero.
[ 502.536259] Thread ID : 0x00000069 Thread name : WLAN RT0 Process ID : 0
[ 502.536259] Register:
[ 502.536259] SP : 0x4bfa84a8
[ 502.536259] FP : 0x4bfa84c0
[ 502.536259] PC : 0x4b18adf8
[ 502.536259] SSR : 0x00000001
[ 502.536259] BADVA : 0x0064644a
[ 502.536259] LR : 0x4b18ad78
[ 502.536259]
[ 502.536259] Stack Dump
[ 502.536259] from : 0x4bfa84a8
[ 502.536259] to : 0x4bfa8ae0
[ 502.536259]
[ 502.582774] remoteproc remoteproc0: crash detected in cd00000.q6v5_wcss: type fatal error
[ 502.605028] remoteproc remoteproc0: handling crash #1 in cd00000.q6v5_wcss
[ 502.613168] remoteproc remoteproc0: recovering cd00000.q6v5_wcss
[ 502.645781] remoteproc remoteproc0: stopped remote processor cd00000.q6v5_wcss
wlan0 works fine as it is attached to ath10k ... wlan1 & wlan2 are attached to ath11k
Could you share the full logs, i.e. using gist or https://pastebin.com please?
No need, it fails the same as before
Did you test also the new beta 2.7.0.1.r2 from yesterday as suggested by robimarko ?
yes for QCN9074
I just looked at your boot logs and I couldn't find the string 2.7.0.1 there. I'm a noob in these matters.
it is build/compiled ... here's the log when it is doing checkout from the repo
HEAD is now at 9990f54 QCN9074 hw1.0: testing: 2.7.0.1.r2: add WLAN.HK.2.7.0.1.r2-00005-QCAHKSWPL_SILICONZ-1
Packing checkout...
interesting discussion on the ath11k mailing list on AHB & PCI patches from you @robimarko
https://lists.infradead.org/pipermail/ath11k/2022-April/002910.html
in separate (but related) I noticed that the ath11k kernel drivers come via the backports is this right? what drives the ath11k driver selections as in like one would do in kernel_menuconfig or via the openwrt menuconfig?
Yeah, backports are used to provide ath11k driver.
Backports have a KConfig limitation that unlike when you are building everything in-tree they cannot select symbols, but only depend on them.
Thus, ATH11K_PCI in backports depends on QRTR and QRTR_MHI instead of selecting them so we have to make the OpenWrt package depend on the kernel module so it gets selected in the kernel config.
You basically have to provide all of the dependencies when using backports manually.
Its been a long time since I noticed that QRTR_SMD is required for AHB support but completely forgot to push it upstream until today when I was working on Buildroot support for IPQ807x to make upstreaming easier cause I can run latest kernels.
I haven't been able to get SSDK to build due to its insane dependency on everything being like in OpenWrt
3 Likes
i was actually yesterday playing with kernel_menuconfig and forcing the ath11k but nothing was working so it feels I got my answer ... perhaps got telepathic on my desperation ! 
just one last thing
noticed Documentation/devicetree/bindings/net/wireless/qcom,ath11k.yaml... but you @robimarko noted on the ax9000 "/* ath11k has no DT compatible for PCI cards */"
has a pci definition ... is this worth patching ?
pci {
pcie0 {
#size-cells = <2>;
#address-cells = <3>;
wifi_0: wifi@0 {
reg = <0 0 0 0 0>;
memory-region = <&qcn9074_0>;
};
};
};
Yeah, it doesn't have generic compatible like ath10k has.
There is no point in adding that, all it does is use the memory region you gave it instead of just asking the kernel to allocate it, its meant for cases when you have 2 or more PCI cards.
I even think I tried using that a while back, might give it another go just in case since its already backported.
Need to backport recent ath11k commits after they merge one regression fix, and hostapd had v3 of the BSS coloring so I will update to that
1 Like
yes this mailing list is quite useful the March list seems full of stuff... they are obviously vry busy
There is new stuff constantly, only late december/early January was dead
@robimarko I applied a patch based on one that you posted on the ath11k forum
note that
@@ -22,7 +24,6 @@ config ATH11K_PCI
tristate "Atheros ath11k PCI support"
depends on ATH11K && PCI
select MHI_BUS
- select QRTR
select QRTR_MHI
help
This module adds support for PCIE bus
--
where QRTR was deselected from the PCI crashed the kernel so instead i applied against backports
--- a/drivers/net/wireless/ath/ath11k/Kconfig 2022-04-02 06:55:03.616289339 +0100
+++ b/drivers/net/wireless/ath/ath11k/Kconfig 2022-04-02 06:55:03.616289339 +0100
@@ -5,6 +5,7 @@
depends on MAC80211 && HAS_DMA
depends on CRYPTO_MICHAEL_MIC
select ATH_COMMON
+ select QRTR
depends on QCOM_QMI_HELPERS
help
This module adds support for Qualcomm Technologies 802.11ax family of
@@ -17,6 +18,7 @@
depends on m
depends on ATH11K
depends on REMOTEPROC
+ select QRTR_SMD
help
This module adds support for AHB bus
I got the three radios up this was something i didn't managed before (nor even on the ax3600)!
11.083617] ath11k c000000.wifi: qmi ignore invalid mem req type 3
[ 11.088663] ath11k c000000.wifi: chip_id 0x0 chip_family 0x0 board_id 0x294 soc_id 0xffffffff
[ 11.094123] ath11k c000000.wifi: fw_version 0x250a04a5 fw_build_timestamp 2021-12-20 07:09 fw_build_id QC_IMAGE_VERSION_STRING=WLAN.HK.2.5.0.1-01208-QCAHKSWPL_SILICONZ-1
But that patch doesnt do anything in OpenWrt as QRTR SMD is pulled in as dependency anyway.