[v2,0/3] HID: i2c-hid: Probe and wake device with HID descriptor fetch

This revises my previous patch[0] to add the sleep STM chips seem to
require as per discussion on the original patch from Lukasz and
Radoslaw[1]. I had initially tried without as it had not previously been
needed in the similar logic in our resume path, but it would appear that
this was simply luck as the affected device was woken up in that case by
"noise" from other sources.

To reiterate, the idea is to add the retry that Lukasz and Radoslaw
discovered was necessary, but do away with the dummy smbus probe and
instead just let HID descriptor fetch retry as needed, aligning more
with the existing retry logic used after resume while saving some noise
on the bus and speeding up initialization a tiny bit.

I added Co-developed-by tags, I hope that's appropriate. We should await
an ACK from Lukasz on it fixing their hardware quirk.

[0]: https://lore.kernel.org/all/20240415170517.18780-1-kl@kl.wtf/
[1]: https://lore.kernel.org/all/CAE5UKNqPA4SnnXyaB7Hwk0kcKMMQ_DUuxogDphnnvSGP8g1nAQ@mail.gmail.com/

On Tue, Apr 23, 2024 at 2:26 PM Kenny Levinsen <kl@kl.wtf> wrote:
>
> This revises my previous patch[0] to add the sleep STM chips seem to
> require as per discussion on the original patch from Lukasz and
> Radoslaw[1]. I had initially tried without as it had not previously been
> needed in the similar logic in our resume path, but it would appear that
> this was simply luck as the affected device was woken up in that case by
> "noise" from other sources.
>
> To reiterate, the idea is to add the retry that Lukasz and Radoslaw
> discovered was necessary, but do away with the dummy smbus probe and
> instead just let HID descriptor fetch retry as needed, aligning more
> with the existing retry logic used after resume while saving some noise
> on the bus and speeding up initialization a tiny bit.
>
> I added Co-developed-by tags, I hope that's appropriate. We should await
> an ACK from Lukasz on it fixing their hardware quirk.
>
> [0]: https://lore.kernel.org/all/20240415170517.18780-1-kl@kl.wtf/
> [1]: https://lore.kernel.org/all/CAE5UKNqPA4SnnXyaB7Hwk0kcKMMQ_DUuxogDphnnvSGP8g1nAQ@mail.gmail.com/
>
Hi Kenny,

Your solution works as it should - I have tested it on my Eve with
enabled debugs and
the retries works as expected with power-on, reboot and suspend/resume paths.

I have also disabled cros_ec_i2c driver to be 100% sure it doesn't do
any i2c transactions on the bus
and again the touchpad initialized successfully (with a retry) on all paths.

So you can add:
Tested-by: Lukasz Majczak <lma@chromium.org>
Reviewed-by: Lukasz Majczak <lma@chromium.org>

Thank you Kenny for your work :)

Best regards,
Lukasz

Hi,

On Tue, Apr 23, 2024 at 5:26 AM Kenny Levinsen <kl@kl.wtf> wrote:
>
> To avoid error messages when a device is not present, b3a81b6c4fc6 added
> an initial bus probe using a dummy i2c_smbus_read_byte() call.
>
> Without this probe, i2c_hid_fetch_hid_descriptor() will fail with
> EREMOTEIO. Propagate the error up so the caller can handle EREMOTEIO
> gracefully, and remove the probe as it is no longer necessary.
>
> Signed-off-by: Kenny Levinsen <kl@kl.wtf>
> ---
>  drivers/hid/i2c-hid/i2c-hid-core.c | 20 ++++++--------------
>  1 file changed, 6 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/hid/i2c-hid/i2c-hid-core.c b/drivers/hid/i2c-hid/i2c-hid-core.c
> index 2df1ab3c31cc..515a80dbf6c7 100644
> --- a/drivers/hid/i2c-hid/i2c-hid-core.c
> +++ b/drivers/hid/i2c-hid/i2c-hid-core.c
> @@ -894,12 +894,8 @@ static int i2c_hid_fetch_hid_descriptor(struct i2c_hid *ihid)
>                                               ihid->wHIDDescRegister,
>                                               &ihid->hdesc,
>                                               sizeof(ihid->hdesc));
> -               if (error) {
> -                       dev_err(&ihid->client->dev,
> -                               "failed to fetch HID descriptor: %d\n",
> -                               error);
> -                       return -ENODEV;
> -               }
> +               if (error)
> +                       return error;
>         }
>
>         /* Validate the length of HID descriptor, the 4 first bytes:
> @@ -1014,17 +1010,13 @@ static int __i2c_hid_core_probe(struct i2c_hid *ihid)
>         struct hid_device *hid = ihid->hid;
>         int ret;
>
> -       /* Make sure there is something at this address */
> -       ret = i2c_smbus_read_byte(client);
> -       if (ret < 0) {
> +       ret = i2c_hid_fetch_hid_descriptor(ihid);
> +       if (ret == -EREMOTEIO) {

I worry a little bit about keying just off of -EREMOTEIO. If I'm
skimming the code properly it's up to the different i2c bus controller
to decide which error code to return here. Looking at, for instance,
"i2c-qcom-geni.c", I see:

[NACK] = {-ENXIO, "NACK: slv unresponsive, check its power/reset-ln"},

Maybe we should just use dev_dbg() in all cases here when we fail to
fetch the descriptor? ...otherwise I think some boards will start
getting a noisy error message.

...and confirmed that my skim seemed to be accurate. i put your
patches on my system and then changed the system to think my
hid-over-i2c device was at 0x11 instead of the normal 0x10. Now, I
get:

[    5.973417] i2c_hid_of 4-0011: failed to fetch HID descriptor: -6
[    5.979701] i2c_hid_of 4-0011: Power on failed: -6


-Doug

On 4/24/24 7:00 PM, Doug Anderson wrote:
> I worry a little bit about keying just off of -EREMOTEIO. If I'm
> skimming the code properly it's up to the different i2c bus controller
> to decide which error code to return here. Looking at, for instance,
> "i2c-qcom-geni.c", I see:
>
> [NACK] = {-ENXIO, "NACK: slv unresponsive, check its power/reset-ln"},


Hmm, good point. I decided to go through the drivers and check their 
behavior on NACK, and based on my quick glance I found (insert accuracy 
disclaimer):

- 52 drivers emitting ENXIO
- 14 drivers emitting EREMOTEIO
- 11 driver emitting EIO
- 5 drivers emitting ETIMEDOUT
- 1 driver emitting EAGAIN
- 1 driver emitting I2C_ERR_BERR (???)

So just EREMOTEIO is definitely not good enough. Looking at the drivers, 
it seems like the majority of drivers emitting generic error codes could 
just as well emit ENXIO on NACK. Room for improvement.

> Maybe we should just use dev_dbg() in all cases here when we fail to
> fetch the descriptor? ...otherwise I think some boards will start
> getting a noisy error message.

I'm okay with that. I don't like hiding a useful error message, but the 
smbus probe would also have hidden bus errors.

I'll send a v3 with just dev_dbg, then if I (or someone else) end up 
aligning more i2c drivers on their NACK error we can go to the stricter 
check and incentivize the drivers to give meaningful error values...