Quality of wireless drivers (rtl8821ae, ath10k, wl18xx)
Hi!
On cip-members there's discussion about quality of selected wireless drivers.
AFAICT we are reviewing changes to those drivers on -stable series (good). We are not testing the drivers in any meaningful way, and it may be good to mark them built-in (=y, not modular) in some configs so at least probing is tested. Proper testing would be quite hard, and ammount of changes to them in stable probably does not require it.
To me it looks like wl18xx is getting very little changes; rtl8821ae gets a bit more, but they are simple cleanups; only ath10k seems to be under any kind of active development.
I have no experience with any of those drivers. I don't expect much work would need to be done on any of these. If there's work I'd expect ath10k to be slightly easier to work with.
Any comments?
Best regards,
Pavel
* rtl8821ae (wifi + BT)
CONFIG_RTL8821AE./drivers/net/wireless/realtek/rtlwifi/rtl8821ae./drivers/bluetooth/btrtl.c
./4.19.y-cip/x86/plathome_obsvx2_defconfig:CONFIG_RTL8821AE=m ./5.10.y-cip-rt/x86/siemens_i386-rt.config:CONFIG_BT_RTL=m
=> We should be reviewing it. We probably are not testing it at all. Single change in 5.10-stable. Some developement from 4.19, but looks mostly like staging-type cleanups.
* ath10k
CONFIG_ATH10K./drivers/net/wireless/ath/ath10k/
./5.10.y-cip/arm64/qemu_arm64_defconfig:CONFIG_ATH10K=m ./5.10.y-cip/arm64/qemu_arm64_defconfig:CONFIG_ATH10K_PCI=m ./5.10.y-cip/x86/plathome_obsvx2_defconfig:CONFIG_ATH10K=m ./5.10.y-cip/x86/plathome_obsvx2_defconfig:CONFIG_ATH10K_PCI=m ./5.10.y-cip/arm/moxa_mxc_defconfig:CONFIG_ATH10K=m ./5.10.y-cip/arm/moxa_mxc_defconfig:CONFIG_ATH10K_DFS_CERTIFIED=y
=> We should be reviewing it. We probably are not testing it at all. 19 patches in 5.10-stable. Under active development, changes from 4.19.
* wl18xx
(wl1837 hardware is handled by wl18xx driver.)
CONFIG_WL18XXCONFIG_BT_HCIUART_LL./drivers/net/wireless/ti/wl18xx./drivers/bluetooth/hci_ll.c
./5.10.y-cip/arm64/qemu_arm64_defconfig:CONFIG_WL18XX=m ./5.10.y-cip/arm64/ctj_zynqmp_defconfig:CONFIG_WL18XX=y ./5.10.y-cip/arm/moxa_mxc_defconfig:CONFIG_WL18XX=m
./5.10.y-cip/arm64/qemu_arm64_defconfig:CONFIG_BT_HCIUART_LL=y ./5.10.y-cip/arm64/ctj_zynqmp_defconfig:CONFIG_BT_HCIUART_LL=y
=> We should be reviewing it. We do some minimal testing by compiling module in ZYNQMP configuration. No changes in 5.10-stable. Very few changes from 4.19, mostly treewide cleanups.
Signature · 2 lines
DENX Software Engineering GmbH, Managing Director: Wolfgang Denk HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Hi!
Quoted 5 lines · Pavel Machek
AFAICT we are reviewing changes to those drivers on -stable series (good). We are not testing the drivers in any meaningful way, and it may be good to mark them built-in (=y, not modular) in some configs so at least probing is tested. Proper testing would be quite hard, and ammount of changes to them in stable probably does not require it.
I think functional testing for wireless drivers in CI system is hard, so building and
probing seem reasonable tests currently.
Quoted 13 lines
* wl18xx
(wl1837 hardware is handled by wl18xx driver.)
CONFIG_WL18XXCONFIG_BT_HCIUART_LL./drivers/net/wireless/ti/wl18xx./drivers/bluetooth/hci_ll.c./5.10.y-cip/arm64/qemu_arm64_defconfig:CONFIG_WL18XX=m ./5.10.y-cip/arm64/ctj_zynqmp_defconfig:CONFIG_WL18XX=y ./5.10.y-cip/arm/moxa_mxc_defconfig:CONFIG_WL18XX=m
Quoted 6 lines
./5.10.y-cip/arm64/qemu_arm64_defconfig:CONFIG_BT_HCIUART_LL=y ./5.10.y-cip/arm64/ctj_zynqmp_defconfig:CONFIG_BT_HCIUART_LL=y
=> We should be reviewing it. We do some minimal testing by compiling module in ZYNQMP configuration. No changes in 5.10-stable. Very few changes from 4.19, mostly treewide cleanups.
zynqmp-zcu102 which is connected to CIP LAVA doesn't have the wl18xx device actually.
On the other hand, RZ/G2 boards (r8a774a1-hihope-rzg2m-ex, r8a774b1-hihope-rzg2n-ex,
r8a774c0-ek874) have the device. So these boards look suitable if probing is tested, although
wl18xx is currently not enabled in the corresponding config files.
Best regards,
Hiraku Toyooka
Cybertrust Japan Co., Ltd.