@@ -748,7 +748,7 @@ int acquire_bus(struct wilc *wilc, enum bus_acquire acquire)
int ret = 0;
mutex_lock(&wilc->hif_cs);
- if (acquire == WILC_BUS_ACQUIRE_AND_WAKEUP && wilc->power_save_mode) {
+ if (acquire == WILC_BUS_ACQUIRE_AND_WAKEUP) {
ret = chip_wakeup(wilc);
if (ret)
mutex_unlock(&wilc->hif_cs);
@@ -761,7 +761,7 @@ int release_bus(struct wilc *wilc, enum bus_release release)
{
int ret = 0;
- if (release == WILC_BUS_RELEASE_ALLOW_SLEEP && wilc->power_save_mode)
+ if (release == WILC_BUS_RELEASE_ALLOW_SLEEP)
ret = chip_allow_sleep(wilc);
mutex_unlock(&wilc->hif_cs);
The wilc chips family has at least two mechanisms dedicated to power save: - a firmware configuration message, currently wired to the set_power_mgmt_ops of the cfg80211_ops structure (and so, configured through iw) - dedicated raw registers which allow to wake up the chip each time we need to communicate with it and to allow it again to sleep once done: those registers are currently manipulated almost any time we are about to use the SDIO/SPI bus to drive the chip. Those mechanisms are currently coupled together, and so the second mechanism is driven only if the first one is enabled, but it is wrong: even without the power save feature being configured in the wlan firmware, there are cases where the wake up and clock registers need to be written correctly, otherwise the chip stays unresponsive. The downstream driver seems to acknowledge this issue, since the flag matching the first feature has been completely removed. Decouple those two features by removing the condition on the powersave flag in acquire_bus/release_bus to make sure that the wakeup/allow_sleep registers are actually written when the passed enums (WILC_BUS_ACQUIRE_AND_WAKEUP, WILC_BUS_RELEASE_ALLOW_SLEEP) explicitly require those writes. Signed-off-by: Alexis Lothoré <alexis.lothore@bootlin.com> --- drivers/net/wireless/microchip/wilc1000/wlan.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)