forge/linux-pinenote
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

[tps65185] when rebasing from v5.17-rc6-11023-gdadcf190596b to 6.2 the

Browse source 
vdrive regulator, from time to time, failed to resume within the allowed
time (ca. 83 ms), leading to a complete crash of the graphical stack and
most of the kernel.

My understanding of the problem is that sometimes the vdrive activation
takes a bit longer, probably due to the active discharge feature of the
tps65185 chip (the REG_ENABLE register shows that the VEE_EN holds at 1 over
the whole waiting time and never going to 0, indicating that the chip is
waiting for the discharge before proceeding with the full activation of
all rails).

We here simply add 1 second to the timeout for vdrive, which should give
the driver enough freedom to handle long discharges.

I the underlying problem is indeed the active discharge, then another
option would be to add an off_on_delay to the vdrive description, which
would make sure the regulator is kept turned off for a given minimal
time before attempting to turn it on again:

	.off_on_delay = 1000000,

Note that this behavior was not observed on older (5.17) kernels due to
a bug in the regulator code that effectively prevented the
activation-timeout from being triggered. See bug fix commit
8d8e165920
This commit is contained in:
Maximilian Weigand 2023-03-04 13:41:12 +01:00 • committed by Antoine Martin
parent fd42ed15cd
commit a7dc651a5e
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
drivers/regulator/tps65185-regulator.c
View file

@ -187,7 +187,7 @@ static int tps65185_vdrive_enable_time(struct regulator_dev *rdev)
{
struct tps65185 *tps = rdev_get_drvdata(rdev);
return tps->total_down_delay_us + tps->total_up_delay_us + 50000;
return tps->total_down_delay_us + tps->total_up_delay_us + 50000 + 1000000;
}
static int tps65185_vdrive_get_status(struct regulator_dev *rdev)