This is a cache of https://discuss.96boards.org/t/some-wireless-issue-with-current-linux-4-4-kernel/5535. It is a snapshot of the page at 2024-10-31T09:10:35.789+0000.
Some wireless issue with current Linux 4.4 kernel - Rock960 - 96Boards Forum

Some wireless issue with current Linux 4.4 kernel

Products Support Rock960
#1

There seems to be a small issue with wireless. It works but dmesg is flooded with the following messages:

[142240.887170] dhd_tcpdata_info_get 1092: No more free tdata_psh_info!!
[142240.902190] _tdata_psh_info_pool_deq 200: Out of tdata_disc_grp
[142240.902216] dhd_tcpdata_info_get 1092: No more free tdata_psh_info!!
[142240.917179] _tdata_psh_info_pool_deq 200: Out of tdata_disc_grp
[142240.917206] dhd_tcpdata_info_get 1092: No more free tdata_psh_info!!
[142240.935494] _tdata_psh_info_pool_deq 200: Out of tdata_disc_grp
[142240.935520] dhd_tcpdata_info_get 1092: No more free tdata_psh_info!!
[142241.139611] _tdata_psh_info_pool_deq 200: Out of tdata_disc_grp
[142241.139638] dhd_tcpdata_info_get 1092: No more free tdata_psh_info!!

#2

hmm…

which image are you running? i didn’t notice the print before…

#3

I am using the ubuntu 16.04 image. Wireless is started by

/sbin/wpa_supplicant -c/etc/wpa_supplicant/wpa_supplicant-wlan0.conf -Dnl80211,wext -iwlan0

wpa_supplicant is controlled by systemd-networkd (I borrowed this configuration from ubuntu 18.04).

Maybe there is a problem with the parameters e.g. drivers I am using here? Should it work (in another way atm) without this issue?

#4

hmm…

The following command works fine here, can you try to remove wext?

wpa_supplicant -B -Dnl80211 -i wlan0 -c /etc/wpa_supplicant/wpa.conf

or

nmcli dev wifi connect "wifi-hotspot-name" password "xxxx"

nmcli works better since it will automatically connect after first time input the wifi hotspot name and password.

#5

I tried your first command manually but this did not improve anything. So I thought there must be still something active that interferes with the otherwise working configuration and I removed “wext” from the configuration permanently and rebooted the system. When I did this in the beginning the errors still showed up and then the following (for me) quite strange things happened:

[ 50.260689] _tdata_psh_info_pool_deq 200: Out of tdata_disc_grp
[ 50.260710] dhd_tcpdata_info_get 1092: No more free tdata_psh_info!!
[ 50.275884] _tdata_psh_info_pool_deq 200: Out of tdata_disc_grp
[ 50.275904] dhd_tcpdata_info_get 1092: No more free tdata_psh_info!!
[ 421.074917] dwmmc_rockchip fe310000.dwmmc: All phases bad!
[ 421.074946] mmc2: tuning execution failed
[ 421.075447] bcmsdh_sdmmc: Failed to Read byte F1:@0x1001f=ff, Err: -5
[ 421.078613] dwmmc_rockchip fe310000.dwmmc: All phases bad!
[ 421.078632] mmc2: tuning execution failed
[ 421.079148] bcmsdh_sdmmc: Failed to Read byte F1:@0x1001f=ff, Err: -5
[ 421.081317] dwmmc_rockchip fe310000.dwmmc: All phases bad!
[ 421.081343] mmc2: tuning execution failed
[ 421.081722] bcmsdh_sdmmc: Failed to Write byte F1:@0x1001f=00, Err: -5
[ 421.084934] dwmmc_rockchip fe310000.dwmmc: All phases bad!
[ 421.084951] mmc2: tuning execution failed
[ 421.085402] bcmsdh_sdmmc: Failed to Write byte F1:@0x1001f=00, Err: -5
[ 421.088630] dwmmc_rockchip fe310000.dwmmc: All phases bad!
[ 421.088647] mmc2: tuning execution failed
[ 421.089188] bcmsdh_sdmmc: Failed to Write byte F1:@0x1001f=00, Err: -5
[ 421.091313] dwmmc_rockchip fe310000.dwmmc: All phases bad!
[ 421.091330] mmc2: tuning execution failed
[ 421.091710] bcmsdh_sdmmc: Failed to Read byte F1:@0x1001f=ff, Err: -5
[ 421.094636] dwmmc_rockchip fe310000.dwmmc: All phases bad!
[ 421.094655] mmc2: tuning execution failed
[ 421.095089] bcmsdh_sdmmc: Failed to Read byte F1:@0x1001f=ff, Err: -5
[ 421.098034] dwmmc_rockchip fe310000.dwmmc: All phases bad!
[ 421.098048] mmc2: tuning execution failed
[ 421.098416] bcmsdh_sdmmc: Failed to Read byte F1:@0x1001f=ff, Err: -5
[ 421.279552] dwmmc_rockchip fe310000.dwmmc: Successfully tuned phase to 235

After that I rebooted the system a second time and it looked really promising that these errors were gone. But after a few minutes all content of dmesg is now filled with these messages again :frowning:

#6

Still get errors from time to time like the one in my post above and I think this is a kernel/driver or hardware issue.

Is there nobody else who gets this kind of warnings and error messages?

#7

does the issue exist in the newer debian image?

New Debian LXDE Desktop image release - #5 by hipboi

#8

It took some time to test and yes this issue also exists in the new Debian image.

  1. I tested the new Debian image with two rock960 boxes and I am pretty sure now that this is no hardware issue because the issue showed up on both and

  2. it can be reproduced by copying some data over w-lan, I did “scp -r rock960:/ /tmp/tmpdir” from a remote host and after a few seconds the first messages appeared in dmesg.

Additional information about the current images:

On one of the rock960 boxes I was not able to start Debian until a used maskrom to disable onboard storage. The difference seems to be that one box has stock Android installed onboard (it works here) and the other has ubuntu image installed onboard and here boot stops before the kernel is started at the following position:

LoadTrustBL
No find bl30.bin
No find bl32.bin
RunBL31 0x10000
NOTICE: BL31: v1.3(debug):65aa5ce
NOTICE: BL31: Built : 10:47:37, Jun 19 2018
NOTICE: BL31: Rockchip release version: v1.1
INFO: GICv3 with legacy support detected. ARM GICV3 driver initialized in EL3
INFO: Using opteed sec cpu_context!
INFO: boot cpu mask: 0
INFO: plat_rockchip_pmu_init(1151): pd status 3e
INFO: BL31: Initializing runtime services
INFO: BL31: Initializing BL32
<— no additional output appears here

And in case of maskrom boot the onboard root filesystem of ubuntu was then mounted under /media. This is really weird so something must be completely wrong here like it is also described in another thread in this forum