bq25895 reports 'not charging' although it is charging in /sys/class/power_supply/bq25890-charger/status

the status register 0b is not read correctly on status changes yielding 0 although it's 6b (fast charging).

have you compared to a 5.8 based kernel? at least there are quite some changes to the driver:

git log --oneline v5.7..v5.8-rc5|grep -i bq25
478efc79ee32 power: bq25890: implement INPUT_CURRENT_LIMIT property
c942656ddb3a power: bq25890: implement PRECHARGE_CURRENT property
b302a0ae7205 power: bq25890: implement CHARGE_TYPE property
3b4df57bef69 power: bq25890: update state on property read
21d90eda433f power: bq25890: fix ADC mode configuration
1e4724d0b7d1 power: bq25890: use proper CURRENT_NOW property for I_BAT
cf5701bf7496 power: bq25890: unlock on error paths in bq25890_resume()
72d9cd9cdc18 power: bq25890: protect view of the chip's state
a9c2419406b8 power: bq25890: remove redundant I2C bus check
a6a48fac96ae power: bq25890: make property table const
5956fca78f5c power: bq25890: simplify chip name property getter

could "update state on property read" fix that?

3b4df57b seems like a thing (i was looking yesterday why we use interrupts but it seems upstream does not bother) and b302a0ae can obsolete https://source.puri.sm/Librem5/linux-next/-/merge_requests/44

for me, 5.8 improves this in this regard and

upower -d|grep -i state
    state:               fully-charged
    state:               charging
    state:               charging

while on 5.7 the 2nd entry (bq) would not show "charging" (when actually charging).

yes, same here' - its 3b4df57b that might even be worth backporting for 5.7.

mentioned in commit 247b6244

mentioned in issue #390

Seems not relevant anymore - closing.

closed