linux-kernel - RE: [PATCH V3] HID: HID-rmi - ignore to rmi_hid_read

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <PH0PR03MB58487C980CC7B97FD2A78C7CA3649@PH0PR03MB5848.namprd03.prod.outlook.com>
Date:   Thu, 11 Aug 2022 07:26:13 +0000
From:   Marge Yang <Marge.Yang@...synaptics.com>
To:     Dmitry Torokhov <dmitry.torokhov@...il.com>,
        margeyang <marge.yang@...aptics.corp-partner.google.com>,
        "dtor@...gle.com" <dtor@...gle.com>,
        Marge Yang <Marge.Yang@...synaptics.com>
CC:     "linux-input@...r.kernel.org" <linux-input@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "hdegoede@...hat.com" <hdegoede@...hat.com>,
        "benjamin.tissoires@...hat.com" <benjamin.tissoires@...hat.com>,
        Derek Cheng <derek.cheng@...synaptics.com>,
        Vincent Huang <Vincent.huang@...synaptics.com>
Subject: RE: [PATCH V3] HID: HID-rmi - ignore to rmi_hid_read_block after
 system resumes.

Hi Dmitry,
	Thanks for your questions.
I have updated the following questions

-----Original Message-----
From: Dmitry Torokhov <dmitry.torokhov@...il.com> 
Sent: Thursday, August 11, 2022 12:21 PM
To: margeyang <marge.yang@...aptics.corp-partner.google.com>
Cc: linux-input@...r.kernel.org; linux-kernel@...r.kernel.org; hdegoede@...hat.com; benjamin.tissoires@...hat.com; Marge Yang <Marge.Yang@...synaptics.com>; Derek Cheng <derek.cheng@...synaptics.com>; Vincent Huang <Vincent.huang@...synaptics.com>
Subject: Re: [PATCH V3] HID: HID-rmi - ignore to rmi_hid_read_block after system resumes.

CAUTION: Email originated externally, do not click links or open attachments unless you recognize the sender and know the content is safe.


Hi Marge,

On Thu, Aug 11, 2022 at 10:58:32AM +0800, margeyang wrote:
> From: Marge Yang <marge.yang@...aptics.corp-partner.google.com>

I recommend using your official @synaptics.com email address for upstream submissions.

[Marge 08/11] Because Synaptics IT's policy is changed, my Linux device is not  added into white list .
Hence, I can't use marge.yang@...synaptics.com and use "git send-email" command to send Linux patch file.
After Synaptics IT changes this setting, I will use marge.yang@...synaptics.com email address for upstream submissions
>
> The interrupt GPIO will be pulled down once after RMI driver reads 
> this command(Report ID:0x0A).
> It will cause "Dark resume test fail" for chromebook device.

What exactly is the failure mode? Please describe what exactly happens.
Maybe you can also add dtor@...gle.com to the relevant internal bugs.

[Marge 08/11] Add dtor@...gle.com into the list of email.
For Dark Resume (https://chromium.googlesource.com/chromiumos/platform2/+/master/power_manager/docs/dark_resume.md) is enabled on this device/device family (this should be enabled per board / unibuild family)
It is a Chromebook auto test tool.
Check if the device driver is incrementing the wakeup counts correctly. 
    Note the wakeup count of the device before suspend.
    Suspend the device.
    Trigger the wake using the wake source.
    Resume the device.
    Check if the wakeup count has incremented.

The interrupt GPIO will be pulled down when RMI driver reads this command(Report ID: RMI_READ_ADDR_REPORT_ID).
It causes that the wakeup count has incremented and Dark Resume fails.
 

> Hence, TP driver will ignore rmi_hid_read_block function once after 
> system resumes.

Do you really want to avoid reads, or we happen to signal wakeup from attention handler somewhere? Is it still valid to skip read if the touchpad or touchscreen is truly the wakeup source?
[Marge 08/11]
Yes, we want to avoid reads (Report ID: RMI_READ_ADDR_REPORT_ID).
For Synaptics FW design, the interrupt GPIO will be pulled down after RMI driver reads this command(Report ID: RMI_READ_ADDR_REPORT_ID).
"Dark resume" test case on Chromebook device will detect one interrupt (not finger data) during the process of resume function.
"Dark resume" test case will fail because wakeup count has incremented between "Suspend" and "Resume".
Hence, this issue should happen on RMI mode for all Synaptics devices.
This change is used for Synaptics touchpad/touchscreen devices.

Yes, it is valid.
When the touchpad or touchscreen is truly the wakeup source, the interrupt GPIO will be pulled.
"Dark resume" test case on Chromebook device will detect one interrupt (finger data) during the process of resume function.
"Dark resume" test case will show "input" event.

>
> Signed-off-by: Marge 
> Yang<marge.yang@...aptics.corp-partner.google.com>
> ---
>  drivers/hid/hid-rmi.c | 14 ++++++++++++--
>  include/linux/rmi.h   |  2 ++
>  2 files changed, 14 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/hid/hid-rmi.c b/drivers/hid/hid-rmi.c index 
> 311eee599ce9..fdaf2c4ce181 100644
> --- a/drivers/hid/hid-rmi.c
> +++ b/drivers/hid/hid-rmi.c
> @@ -203,7 +203,13 @@ static int rmi_hid_read_block(struct rmi_transport_dev *xport, u16 addr,
>               if (ret < 0)
>                       goto exit;
>       }
> -
> +     if (xport->ignoreonce == 1) {
> +             dev_err(&hdev->dev,
> +                     "ignoreonce (%d)\n",
> +                     xport->ignoreonce);
> +             xport->ignoreonce = 0;
> +             goto exit;
> +     }

Do we need to have it after updating the page register?

[Marge 8/11] Yes, we need it to avoid reads (Report ID: RMI_READ_ADDR_REPORT_ID).

>       for (retries = 5; retries > 0; retries--) {
>               data->writeReport[0] = RMI_READ_ADDR_REPORT_ID;
>               data->writeReport[1] = 0; /* old 1 byte read count */ @@ 
> -468,8 +474,12 @@ static int rmi_post_resume(struct hid_device *hdev)
>       ret = hid_hw_open(hdev);
>       if (ret)
>               return ret;
> -
> +     // Avoid to read rmi_hid_read_block once after system resumes.
> +     // The interrupt will be pulled down
> +     // after RMI Read command(Report ID:0x0A).
> +     data->xport.ignoreonce = 1;
>       ret = rmi_reset_attn_mode(hdev)
> +     data->xport.ignoreonce = 0;
>       if (ret)
>               goto out;
>
> diff --git a/include/linux/rmi.h b/include/linux/rmi.h index 
> ab7eea01ab42..24f63ad00970 100644
> --- a/include/linux/rmi.h
> +++ b/include/linux/rmi.h
> @@ -270,6 +270,8 @@ struct rmi_transport_dev {
>       struct rmi_device_platform_data pdata;
>
>       struct input_dev *input;
> +
> +     int ignoreonce;
>  };
>
>  /**
> --
> 2.22.0.windows.1
>

Thanks.

--
Dmitry