Commit 9990f2f6264c ("usb: typec: tipd: Enable event interrupts by default")
writes a fixed set of interrupts to TPS_REG_INT_MASK1. In case interrupts
had been enabled by the firmware by default, these get disabled now
which can break use cases. Only append to what is already enabled instead.
Fixes: 9990f2f6264c ("usb: typec: tipd: Enable event interrupts by default")
Signed-off-by: Martin Kepplinger <[email protected]>
---
drivers/usb/typec/tipd/core.c | 10 +++++++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/drivers/usb/typec/tipd/core.c b/drivers/usb/typec/tipd/core.c
index 6d27a5b5e3ca..365ee5494c15 100644
--- a/drivers/usb/typec/tipd/core.c
+++ b/drivers/usb/typec/tipd/core.c
@@ -744,10 +744,14 @@ static int tps6598x_probe(struct i2c_client *client)
irq_handler = cd321x_interrupt;
} else {
+ ret = tps6598x_read64(tps, TPS_REG_INT_MASK1, &mask1);
+ if (ret)
+ return ret;
+
/* Enable power status, data status and plug event interrupts */
- mask1 = TPS_REG_INT_POWER_STATUS_UPDATE |
- TPS_REG_INT_DATA_STATUS_UPDATE |
- TPS_REG_INT_PLUG_EVENT;
+ mask1 |= TPS_REG_INT_POWER_STATUS_UPDATE |
+ TPS_REG_INT_DATA_STATUS_UPDATE |
+ TPS_REG_INT_PLUG_EVENT;
}
/* Make sure the controller has application firmware running */
--
2.30.2
Hi Martin,
On Mon, Jan 10, 2022 at 05:35:59PM +0100, Martin Kepplinger wrote:
> Commit 9990f2f6264c ("usb: typec: tipd: Enable event interrupts by default")
> writes a fixed set of interrupts to TPS_REG_INT_MASK1. In case interrupts
> had been enabled by the firmware by default, these get disabled now
> which can break use cases. Only append to what is already enabled instead.
No, we have to know what are the interrupts that are actually needed.
Commmit 9990f2f6264c fixed an interrupt flood issue that was seen on
some platforms where the firmware I guess enabled absolutely
everything. This would just revert that.
So only enable the interrupts that you know you need.
> Fixes: 9990f2f6264c ("usb: typec: tipd: Enable event interrupts by default")
What is this fixing? What are the use cases you are talking about?
> Signed-off-by: Martin Kepplinger <[email protected]>
> ---
> drivers/usb/typec/tipd/core.c | 10 +++++++---
> 1 file changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/usb/typec/tipd/core.c b/drivers/usb/typec/tipd/core.c
> index 6d27a5b5e3ca..365ee5494c15 100644
> --- a/drivers/usb/typec/tipd/core.c
> +++ b/drivers/usb/typec/tipd/core.c
> @@ -744,10 +744,14 @@ static int tps6598x_probe(struct i2c_client *client)
>
> irq_handler = cd321x_interrupt;
> } else {
> + ret = tps6598x_read64(tps, TPS_REG_INT_MASK1, &mask1);
> + if (ret)
> + return ret;
> +
> /* Enable power status, data status and plug event interrupts */
> - mask1 = TPS_REG_INT_POWER_STATUS_UPDATE |
> - TPS_REG_INT_DATA_STATUS_UPDATE |
> - TPS_REG_INT_PLUG_EVENT;
> + mask1 |= TPS_REG_INT_POWER_STATUS_UPDATE |
> + TPS_REG_INT_DATA_STATUS_UPDATE |
> + TPS_REG_INT_PLUG_EVENT;
> }
>
> /* Make sure the controller has application firmware running */
thanks,
--
heikki
On 2022/01/11 1:35, Martin Kepplinger wrote:
> Commit 9990f2f6264c ("usb: typec: tipd: Enable event interrupts by default")
> writes a fixed set of interrupts to TPS_REG_INT_MASK1. In case interrupts
> had been enabled by the firmware by default, these get disabled now
> which can break use cases. Only append to what is already enabled instead.
>
I'm confused. The kernel drives the hardware, it needs to enable only
the interrupts it can handle. Do you have some kind of firmware trying
to share access to the same I2C port that needs other interrupts? That
sounds like a recipe for trouble... or am I misunderstanding things?
If the *kernel* needs other interrupts enabled to make something work,
then they should also be enabled unconditionally, and you'd have to
check the IRQ handler to make sure it actually handles it.
--
Hector Martin ([email protected])
Public Key: https://mrcn.st/pub
Am Dienstag, dem 11.01.2022 um 20:10 +0900 schrieb Hector Martin:
> On 2022/01/11 1:35, Martin Kepplinger wrote:
> > Commit 9990f2f6264c ("usb: typec: tipd: Enable event interrupts by
> > default")
> > writes a fixed set of interrupts to TPS_REG_INT_MASK1. In case
> > interrupts
> > had been enabled by the firmware by default, these get disabled now
> > which can break use cases. Only append to what is already enabled
> > instead.
> >
>
> I'm confused. The kernel drives the hardware, it needs to enable only
> the interrupts it can handle. Do you have some kind of firmware
> trying
> to share access to the same I2C port that needs other interrupts?
> That
> sounds like a recipe for trouble... or am I misunderstanding things?
>
> If the *kernel* needs other interrupts enabled to make something
> work,
> then they should also be enabled unconditionally, and you'd have to
> check the IRQ handler to make sure it actually handles it.
>
true. sorry for the confusion. we need to submit the patches for the
interrupt handler to handle what we need and then we'll extend the mask
accordingly. please ignore this patch.
thank you,
martin