Hi,
On Sat, Nov 9, 2019 at 2:31 AM Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:
>
> Changes in v2:
> - Add adaptations to the mwifiex driver.
> - Keep existing syncronous reset behaviour if the SDIO card has a single
> func driver.
>
> It has turned out that it's not a good idea to try to power cycle and to
> re-initialize the SDIO card, as currently done through mmc_hw_reset(). This
> because there may be multiple SDIO funcs attached to the same SDIO card and
> some of the others that didn't execute the call to mmc_hw_reset(), may then
> simply experience an undefined behaviour.
>
> The following patches in this series attempts to address this problem, by
> reworking the mmc_hw_reset() behaviour for SDIO and by adopting the Marvel
> mwifiex driver to these changes.
>
> Note that, I don't have the HW at hand so the the code has only compile tested.
> Test on HW is greatly appreciated!
>
> Ulf Hansson (3):
> mwifiex: Re-work support for SDIO HW reset
> mmc: core: Drop check for mmc_card_is_removable() in mmc_rescan()
> mmc: core: Re-work HW reset for SDIO cards
>
> drivers/mmc/core/core.c | 12 +++-----
> drivers/mmc/core/core.h | 2 ++
> drivers/mmc/core/sdio.c | 28 ++++++++++++++++-
> drivers/mmc/core/sdio_bus.c | 9 +++++-
> drivers/net/wireless/marvell/mwifiex/main.c | 6 +++-
> drivers/net/wireless/marvell/mwifiex/main.h | 1 +
> drivers/net/wireless/marvell/mwifiex/sdio.c | 33 ++++++++++++++-------
> include/linux/mmc/card.h | 1 +
> 8 files changed, 70 insertions(+), 22 deletions(-)
I put this on rk3288-veyron-jerry atop v5.4-rc7 and I could run my
test case for a while, AKA I got over 50 cycles of:
---
for i in $(seq 1000); do
echo "LOOP $i --------"
echo 1 > /sys/kernel/debug/mwifiex/mlan0/reset
while true; do
if ! ping6 -w15 -c1 "${GW}" >/dev/null 2>&1; then
fail=$(( fail + 1 ))
echo "Fail WiFi ${fail}"
if [[ ${fail} == 3 ]]; then
exit 1
fi
else
fail=0
break
fi
done
hciconfig hci0 down
sleep 1
if ! hciconfig hci0 up; then
echo "Fail BT"
exit 1
fi
done
---
NOTE: with no patches I couldn't even get my test case to pass w/out
the BT bits and I swear that used to work before. ...but I didn't
debug since the end result (with full card hotplug) is happy-working
for me. I'll still use it as further argument that (IMO) full unplug
/ plug of the card is better it uses more standard code paths and is
less likely to break. ;-)
Tested-by: Douglas Anderson <dianders@xxxxxxxxxxxx>
