Re: [PATCH] soc: qcom: pmic_glink_altmode: use a freezable workqueue

[Val Packett]

On 7/27/26 10:25 PM, Daniel J Blueman wrote:
> On suspend wake, the DisplayPort altmode setup often executes
> before devices have compeleted resuming; this triggers a legitimate
> warning:
>    i2c i2c-2: Transfer while suspended
>    WARNING: drivers/i2c/i2c-core.h:57 at __i2c_transfer+0xc0/0x9a8
>    Workqueue: events pmic_glink_altmode_worker
>    Call trace:
>     __i2c_transfer
>     i2c_transfer
>     regmap_i2c_read
>     ...
>     ps883x_sw_set
>     typec_switch_set
>     pmic_glink_altmode_worker
>
> The setup doesn't succeed, leaving the DP-alt monitor non-working.
>
> Fix this by queuing the work on the freezable workqueue instead,
> deferring the setup until devices have resumed.
>
> Observed and validated on a Lenovo Yoga Slim 7x X1E80100 when waking
> from suspend with a USB-C DisplayPort monitor.
>
> Fixes: 080b4e24852b ("soc: qcom: pmic_glink: Introduce altmode support")
> Cc: stable@xxxxxxxxxxxxxxx
> Signed-off-by: Daniel J Blueman <daniel@xxxxxxxxx>
> ---
>   drivers/soc/qcom/pmic_glink_altmode.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/soc/qcom/pmic_glink_altmode.c b/drivers/soc/qcom/pmic_glink_altmode.c
> index 619bad2c27ee..939a444ebe6e 100644
> --- a/drivers/soc/qcom/pmic_glink_altmode.c
> +++ b/drivers/soc/qcom/pmic_glink_altmode.c
> @@ -450,7 +450,7 @@ static void pmic_glink_altmode_sc8180xp_notify(struct pmic_glink_altmode *altmod
>   	alt_port->mode = mode;
>   	alt_port->hpd_state = hpd_state;
>   	alt_port->hpd_irq = hpd_irq;
> -	schedule_work(&alt_port->work);
> +	queue_work(system_freezable_wq, &alt_port->work);
>   }
>   
>   #define SC8280XP_DPAM_MASK	0x3f
> @@ -501,7 +501,7 @@ static void pmic_glink_altmode_sc8280xp_notify(struct pmic_glink_altmode *altmod
>   		alt_port->tbt_data = *tbt;
>   	}
>   
> -	schedule_work(&alt_port->work);
> +	queue_work(system_freezable_wq, &alt_port->work);
>   }
>   
>   static void pmic_glink_altmode_callback(const void *data, size_t len, void *priv)


BTW exactly this has been proposed previously:

https://lore.kernel.org/all/20250110-soc-qcom-pmic-glink-fix-device-access-on-worker-while-suspended-v1-1-e32fd6bf322e@xxxxxxxxxx/

It was concluded that this is a stop-gap solution until the 
notifications are fixed to not be sent at all while suspended… and of 
course neither the "real fix" nor this has landed :(

~val