[PATCH] mfd / platform: cros_ec: Query EC protocol version if EC transitions between RO/RW

[PATCH] mfd / platform: cros_ec: Query EC protocol version if EC transitions between RO/RW

[Yicheng Li]

RO and RW of EC may have different EC protocol version. If EC transitions
between RO and RW, but AP does not reboot (this is true for fingerprint
microcontroller / cros_fp, but not true for main ec / cros_ec), the AP
still uses the protocol version queried before transition, which can
cause problems. In the case of fingerprint microcontroller, this causes
AP to send the wrong version of EC_CMD_GET_NEXT_EVENT to RO in the
interrupt handler, which in turn prevents RO to clear the interrupt
line to AP, in an infinite loop.

Once an EC_HOST_EVENT_INTERFACE_READY is received, we know that there
might have been a transition between RO and RW, so re-query the protocol.

Signed-off-by: Yicheng Li <yichengli@chromium.org>

Change-Id: Ib58032ff4a8e113bdbd07212e8aff42807afff38
Series-to: LKML <linux-kernel@vger.kernel.org>
Series-cc: Benson Leung <bleung@chromium.org>, Enric Balletbo i Serra <enric.balletbo@collabora.com>, Gwendal Grignou <gwendal@chromium.org>
---
 drivers/platform/chrome/cros_ec.c           | 24 +++++++++++++++++++++
 include/linux/platform_data/cros_ec_proto.h |  1 +
 2 files changed, 25 insertions(+)

diff --git a/drivers/platform/chrome/cros_ec.c b/drivers/platform/chrome/cros_ec.c
index 9b2d07422e17..0c910846d99d 100644
--- a/drivers/platform/chrome/cros_ec.c
+++ b/drivers/platform/chrome/cros_ec.c
@@ -104,6 +104,23 @@ static int cros_ec_sleep_event(struct cros_ec_device *ec_dev, u8 sleep_event)
 	return ret;
 }
 
+static int cros_ec_ready_event(struct notifier_block *nb,
+	unsigned long queued_during_suspend, void *_notify)
+{
+	struct cros_ec_device *ec_dev = container_of(nb, struct cros_ec_device,
+						     notifier_ready);
+	u32 host_event = cros_ec_get_host_event(ec_dev);
+
+	if (host_event & EC_HOST_EVENT_MASK(EC_HOST_EVENT_INTERFACE_READY)) {
+		mutex_lock(&ec_dev->lock);
+		cros_ec_query_all(ec_dev);
+		mutex_unlock(&ec_dev->lock);
+		return NOTIFY_OK;
+	} else {
+		return NOTIFY_DONE;
+	}
+}
+
 /**
  * cros_ec_register() - Register a new ChromeOS EC, using the provided info.
  * @ec_dev: Device to register.
@@ -201,6 +218,13 @@ int cros_ec_register(struct cros_ec_device *ec_dev)
 		dev_dbg(ec_dev->dev, "Error %d clearing sleep event to ec",
 			err);
 
+	/* Register the notifier for EC_HOST_EVENT_INTERFACE_READY event. */
+	ec_dev->notifier_ready.notifier_call = cros_ec_ready_event;
+	err = blocking_notifier_chain_register(&ec_dev->event_notifier,
+					       &ec_dev->notifier_ready);
+	if (err < 0)
+		dev_warn(ec_dev->dev, "Failed to register notifier\n");
+
 	dev_info(dev, "Chrome EC device registered\n");
 
 	return 0;
diff --git a/include/linux/platform_data/cros_ec_proto.h b/include/linux/platform_data/cros_ec_proto.h
index 0d4e4aaed37a..9840408c0b01 100644
--- a/include/linux/platform_data/cros_ec_proto.h
+++ b/include/linux/platform_data/cros_ec_proto.h
@@ -161,6 +161,7 @@ struct cros_ec_device {
 	int event_size;
 	u32 host_event_wake_mask;
 	u32 last_resume_result;
+	struct notifier_block notifier_ready;
 
 	/* The platform devices used by the mfd driver */
 	struct platform_device *ec;
-- 
2.21.0

Re: [PATCH] mfd / platform: cros_ec: Query EC protocol version if EC transitions between RO/RW

[Enric Balletbo i Serra]

Hi Yicheng,

Many thanks for sending the patch upstream.

On 18/11/19 21:00, Yicheng Li wrote:
> RO and RW of EC may have different EC protocol version. If EC transitions
> between RO and RW, but AP does not reboot (this is true for fingerprint
> microcontroller / cros_fp, but not true for main ec / cros_ec), the AP
> still uses the protocol version queried before transition, which can
> cause problems. In the case of fingerprint microcontroller, this causes
> AP to send the wrong version of EC_CMD_GET_NEXT_EVENT to RO in the
> interrupt handler, which in turn prevents RO to clear the interrupt
> line to AP, in an infinite loop.
> 

cc'ing: Fabien.

I am wondering if we should be able to reproduce this from userspace, i.e by
setting firmware to RO and issuing a EC_CMD_REBOOT.

What's the actual issue, if the firmware hangs switches to RO and reboots? Is
this happening in products or is something that you detected while playing with
the firmware?

> Once an EC_HOST_EVENT_INTERFACE_READY is received, we know that there
> might have been a transition between RO and RW, so re-query the protocol.
> 
> Signed-off-by: Yicheng Li <yichengli@chromium.org>
> 
> Change-Id: Ib58032ff4a8e113bdbd07212e8aff42807afff38
> Series-to: LKML <linux-kernel@vger.kernel.org>
> Series-cc: Benson Leung <bleung@chromium.org>, Enric Balletbo i Serra <enric.balletbo@collabora.com>, Gwendal Grignou <gwendal@chromium.org>

You should get rid of the three above lines, did you use patman correctly to
send this patch?

> ---
>  drivers/platform/chrome/cros_ec.c           | 24 +++++++++++++++++++++
>  include/linux/platform_data/cros_ec_proto.h |  1 +
>  2 files changed, 25 insertions(+)
> 
> diff --git a/drivers/platform/chrome/cros_ec.c b/drivers/platform/chrome/cros_ec.c
> index 9b2d07422e17..0c910846d99d 100644
> --- a/drivers/platform/chrome/cros_ec.c
> +++ b/drivers/platform/chrome/cros_ec.c
> @@ -104,6 +104,23 @@ static int cros_ec_sleep_event(struct cros_ec_device *ec_dev, u8 sleep_event)
>  	return ret;
>  }
>  
> +static int cros_ec_ready_event(struct notifier_block *nb,
> +	unsigned long queued_during_suspend, void *_notify)
> +{
> +	struct cros_ec_device *ec_dev = container_of(nb, struct cros_ec_device,
> +						     notifier_ready);
> +	u32 host_event = cros_ec_get_host_event(ec_dev);
> +
> +	if (host_event & EC_HOST_EVENT_MASK(EC_HOST_EVENT_INTERFACE_READY)) {
> +		mutex_lock(&ec_dev->lock);
> +		cros_ec_query_all(ec_dev);
> +		mutex_unlock(&ec_dev->lock);
> +		return NOTIFY_OK;
> +	} else {
> +		return NOTIFY_DONE;
> +	}

no else, just

        return NOTIFY_DONE;

> +}
> +
>  /**
>   * cros_ec_register() - Register a new ChromeOS EC, using the provided info.
>   * @ec_dev: Device to register.
> @@ -201,6 +218,13 @@ int cros_ec_register(struct cros_ec_device *ec_dev)
>  		dev_dbg(ec_dev->dev, "Error %d clearing sleep event to ec",
>  			err);
>  
> +	/* Register the notifier for EC_HOST_EVENT_INTERFACE_READY event. */
> +	ec_dev->notifier_ready.notifier_call = cros_ec_ready_event;
> +	err = blocking_notifier_chain_register(&ec_dev->event_notifier,
> +					       &ec_dev->notifier_ready);
> +	if (err < 0)
> +		dev_warn(ec_dev->dev, "Failed to register notifier\n");
> +

Is that only a warning and should be ignored if it fails?

BTW, blocking_notfier_chain_register _always_ returns 0. So I'd say that better
do in case at some point starts to returning something different.

if (err)
   return err;


>  	dev_info(dev, "Chrome EC device registered\n");
>  
>  	return 0;
> diff --git a/include/linux/platform_data/cros_ec_proto.h b/include/linux/platform_data/cros_ec_proto.h
> index 0d4e4aaed37a..9840408c0b01 100644
> --- a/include/linux/platform_data/cros_ec_proto.h
> +++ b/include/linux/platform_data/cros_ec_proto.h
> @@ -161,6 +161,7 @@ struct cros_ec_device {
>  	int event_size;
>  	u32 host_event_wake_mask;
>  	u32 last_resume_result;
> +	struct notifier_block notifier_ready;

You should add documentations for this new entry.

>  
>  	/* The platform devices used by the mfd driver */
>  	struct platform_device *ec;
> 

Thanks,
 Enric

Re: [PATCH] mfd / platform: cros_ec: Query EC protocol version if EC transitions between RO/RW

[Fabien Lahoudere]

Hi,

I tried to test that patch but I cannot switch to RO firmware.
I do the following steps:

root@debian:/sys/class/chromeos/cros_fp# cat version 
RO version:    nocturne_fp_v2.2.64-58cf5974e
RW version:    nocturne_fp_v2.2.110-b936c0a3c
Firmware copy: RW
Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02 14:16:46
@swarm-cros-461
Chip vendor:   stm
Chip name:     stm32h7x3
Chip revision: 
Board version: EC error 1
root@debian:/sys/class/chromeos/cros_fp# echo ro > reboot
root@debian:/sys/class/chromeos/cros_fp# cat version 
RO version:    nocturne_fp_v2.2.64-58cf5974e
RW version:    nocturne_fp_v2.2.110-b936c0a3c
Firmware copy: RW
Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02 14:16:46
@swarm-cros-461
Chip vendor:   stm
Chip name:     stm32h7x3
Chip revision: 
Board version: EC error 1
root@debian:/sys/class/chromeos/cros_fp#

We see here that cros_fp is still RW.

I also tried with:

debian@debian:/src/ec$ sudo build/bds/util/ectool --name=cros_fp
reboot_ec RO
debian@debian:/src/ec$ sudo build/bds/util/ectool --name=cros_fp
version     
RO version:    nocturne_fp_v2.2.64-58cf5974e
RW version:    nocturne_fp_v2.2.110-b936c0a3c
Firmware copy: RW
Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02 14:16:46
@swarm-cros-461
Tool version:  v2.0.3074-a5052d4e7 2020-01-16 10:23:05 debian@debian
debian@debian:/src/ec$

with the same result.

Can you decribe us steps you follow to test that patch?

Thanks

Fabien

Le lundi 18 novembre 2019 à 12:00 -0800, Yicheng Li a écrit :
> RO and RW of EC may have different EC protocol version. If EC
> transitions
> between RO and RW, but AP does not reboot (this is true for
> fingerprint
> microcontroller / cros_fp, but not true for main ec / cros_ec), the
> AP
> still uses the protocol version queried before transition, which can
> cause problems. In the case of fingerprint microcontroller, this
> causes
> AP to send the wrong version of EC_CMD_GET_NEXT_EVENT to RO in the
> interrupt handler, which in turn prevents RO to clear the interrupt
> line to AP, in an infinite loop.
> 
> Once an EC_HOST_EVENT_INTERFACE_READY is received, we know that there
> might have been a transition between RO and RW, so re-query the
> protocol.
> 
> Signed-off-by: Yicheng Li <yichengli@chromium.org>
> 
> Change-Id: Ib58032ff4a8e113bdbd07212e8aff42807afff38
> Series-to: LKML <linux-kernel@vger.kernel.org>
> Series-cc: Benson Leung <bleung@chromium.org>, Enric Balletbo i Serra
> <enric.balletbo@collabora.com>, Gwendal Grignou <gwendal@chromium.org
> >
> ---
>  drivers/platform/chrome/cros_ec.c           | 24
> +++++++++++++++++++++
>  include/linux/platform_data/cros_ec_proto.h |  1 +
>  2 files changed, 25 insertions(+)
> 
> diff --git a/drivers/platform/chrome/cros_ec.c
> b/drivers/platform/chrome/cros_ec.c
> index 9b2d07422e17..0c910846d99d 100644
> --- a/drivers/platform/chrome/cros_ec.c
> +++ b/drivers/platform/chrome/cros_ec.c
> @@ -104,6 +104,23 @@ static int cros_ec_sleep_event(struct
> cros_ec_device *ec_dev, u8 sleep_event)
>  	return ret;
>  }
>  
> +static int cros_ec_ready_event(struct notifier_block *nb,
> +	unsigned long queued_during_suspend, void *_notify)
> +{
> +	struct cros_ec_device *ec_dev = container_of(nb, struct
> cros_ec_device,
> +						     notifier_ready);
> +	u32 host_event = cros_ec_get_host_event(ec_dev);
> +
> +	if (host_event &
> EC_HOST_EVENT_MASK(EC_HOST_EVENT_INTERFACE_READY)) {
> +		mutex_lock(&ec_dev->lock);
> +		cros_ec_query_all(ec_dev);
> +		mutex_unlock(&ec_dev->lock);
> +		return NOTIFY_OK;
> +	} else {
> +		return NOTIFY_DONE;
> +	}
> +}
> +
>  /**
>   * cros_ec_register() - Register a new ChromeOS EC, using the
> provided info.
>   * @ec_dev: Device to register.
> @@ -201,6 +218,13 @@ int cros_ec_register(struct cros_ec_device
> *ec_dev)
>  		dev_dbg(ec_dev->dev, "Error %d clearing sleep event to
> ec",
>  			err);
>  
> +	/* Register the notifier for EC_HOST_EVENT_INTERFACE_READY
> event. */
> +	ec_dev->notifier_ready.notifier_call = cros_ec_ready_event;
> +	err = blocking_notifier_chain_register(&ec_dev->event_notifier,
> +					       &ec_dev-
> >notifier_ready);
> +	if (err < 0)
> +		dev_warn(ec_dev->dev, "Failed to register notifier\n");
> +
>  	dev_info(dev, "Chrome EC device registered\n");
>  
>  	return 0;
> diff --git a/include/linux/platform_data/cros_ec_proto.h
> b/include/linux/platform_data/cros_ec_proto.h
> index 0d4e4aaed37a..9840408c0b01 100644
> --- a/include/linux/platform_data/cros_ec_proto.h
> +++ b/include/linux/platform_data/cros_ec_proto.h
> @@ -161,6 +161,7 @@ struct cros_ec_device {
>  	int event_size;
>  	u32 host_event_wake_mask;
>  	u32 last_resume_result;
> +	struct notifier_block notifier_ready;
>  
>  	/* The platform devices used by the mfd driver */
>  	struct platform_device *ec;

Re: [PATCH] mfd / platform: cros_ec: Query EC protocol version if EC transitions between RO/RW

[Fabien Lahoudere]

Hi Yicheng,

Thanks for your quick answer.

I updated the firmware after my first email on friday.
Now the version is correct:

   # ectool version --name=cros_fp
   RO version:    nocturne_fp_v2.2.64-58cf5974e
   RW version:    nocturne_fp_v2.2.191-1d529566e
   Firmware copy: RW
   Build info:    nocturne_fp_v2.2.191-1d529566e 2019-11-01 20:30:58
   @chromeos-ci-legacy-us-central1-b-x32-72-dltd
   Tool version:  v2.0.3031-9e0f24fad 2020-01-13 06:11:02 @chromeos-ci-
   legacy-us-east1-d-x32-89-i09h

The result does not change. This is the console output after I sent the
reboot command.

   $ ectool name=cros_fp output
   ...
   [250664.176204 HC 0xa9]
   [250664.176287 HC 0xa9 err 1]
   [250664.177413 HC 0x67]
   [250664.177489 HC 0x67 err 9]
   [250664.178434 HC 0x02]
   [250664.180308 HC Suppressed: 0x97=127 0x98=138 0x115=0]
   [250664.477987 HC 0x01]
   [250892.898421 HC 0x02]
   [250892.900447 HC 0x07]
   [250892.901286 HC 0x0b]
   [250902.139053 HC 0x02]
   [250902.141059 HC 0x07]
   [250902.141847 HC 0x0b]
   [250902.142988 HC 0xd2]
   [250902.143072 Executing host reboot command 1]
   [250902.143176 HC 0xd2 err 4]
   [250908.838254 HC 0x02]
   [250908.840474 HC 0x07]
   [250908.841391 HC 0x0b]

it seems we have an error when executing the 0zd2 command.

Any idea what is error 4?

Thanks

Fabien

Le jeudi 16 janvier 2020 à 05:04 -0800, Yicheng Li a écrit :
> Hi Fabien,
> 
> First, you need the new firmware which has RW 2.2.191 instead of
> 2.2.110. It's available in release M80. The old firmware you were
> using does not have the RO/RW protocol incompatibility problem
> because RW 2.2.110 still only support versions 0 and 1 of
> EC_CMD_GET_NEXT_EVENT, same as RO 2.2.64. RW 2.2.191 has version 2 of
> EC_CMD_GET_NEXT_ EVENT, hence the issue described in the commit
> message.
> 
> With RW 2.2.191, if the kernal does not have this patch, rebooting
> cros_fp will leave you in an infinite loop in RO, as described in the
> commit message. This patch solves the problem and you should be in RW
> after the reboot. You can see more details on the commands sent to
> cros_fp during RO if you have cros_fp.log 
> 
> Let me know if you have other questions. Thanks!
> 
> Best,
> Yicheng
> 
> 
> On Thu, Jan 16, 2020, 4:27 AM Fabien Lahoudere <
> fabien.lahoudere@collabora.com> wrote:
> > Hi,
> > 
> > I tried to test that patch but I cannot switch to RO firmware.
> > I do the following steps:
> > 
> > root@debian:/sys/class/chromeos/cros_fp# cat version 
> > RO version:    nocturne_fp_v2.2.64-58cf5974e
> > RW version:    nocturne_fp_v2.2.110-b936c0a3c
> > Firmware copy: RW
> > Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02 14:16:46
> > @swarm-cros-461
> > Chip vendor:   stm
> > Chip name:     stm32h7x3
> > Chip revision: 
> > Board version: EC error 1
> > root@debian:/sys/class/chromeos/cros_fp# echo ro > reboot
> > root@debian:/sys/class/chromeos/cros_fp# cat version 
> > RO version:    nocturne_fp_v2.2.64-58cf5974e
> > RW version:    nocturne_fp_v2.2.110-b936c0a3c
> > Firmware copy: RW
> > Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02 14:16:46
> > @swarm-cros-461
> > Chip vendor:   stm
> > Chip name:     stm32h7x3
> > Chip revision: 
> > Board version: EC error 1
> > root@debian:/sys/class/chromeos/cros_fp#
> > 
> > We see here that cros_fp is still RW.
> > 
> > I also tried with:
> > 
> > debian@debian:/src/ec$ sudo build/bds/util/ectool --name=cros_fp
> > reboot_ec RO
> > debian@debian:/src/ec$ sudo build/bds/util/ectool --name=cros_fp
> > version     
> > RO version:    nocturne_fp_v2.2.64-58cf5974e
> > RW version:    nocturne_fp_v2.2.110-b936c0a3c
> > Firmware copy: RW
> > Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02 14:16:46
> > @swarm-cros-461
> > Tool version:  v2.0.3074-a5052d4e7 2020-01-16 10:23:05 
> > debian@debian
> > debian@debian:/src/ec$
> > 
> > with the same result.
> > 
> > Can you decribe us steps you follow to test that patch?
> > 
> > Thanks
> > 
> > Fabien
> > 
> > Le lundi 18 novembre 2019 à 12:00 -0800, Yicheng Li a écrit :
> > > RO and RW of EC may have different EC protocol version. If EC
> > > transitions
> > > between RO and RW, but AP does not reboot (this is true for
> > > fingerprint
> > > microcontroller / cros_fp, but not true for main ec / cros_ec),
> > the
> > > AP
> > > still uses the protocol version queried before transition, which
> > can
> > > cause problems. In the case of fingerprint microcontroller, this
> > > causes
> > > AP to send the wrong version of EC_CMD_GET_NEXT_EVENT to RO in
> > the
> > > interrupt handler, which in turn prevents RO to clear the
> > interrupt
> > > line to AP, in an infinite loop.
> > > 
> > > Once an EC_HOST_EVENT_INTERFACE_READY is received, we know that
> > there
> > > might have been a transition between RO and RW, so re-query the
> > > protocol.
> > > 
> > > Signed-off-by: Yicheng Li <yichengli@chromium.org>
> > > 
> > > Change-Id: Ib58032ff4a8e113bdbd07212e8aff42807afff38
> > > Series-to: LKML <linux-kernel@vger.kernel.org>
> > > Series-cc: Benson Leung <bleung@chromium.org>, Enric Balletbo i
> > Serra
> > > <enric.balletbo@collabora.com>, Gwendal Grignou <
> > gwendal@chromium.org
> > > >
> > > ---
> > >  drivers/platform/chrome/cros_ec.c           | 24
> > > +++++++++++++++++++++
> > >  include/linux/platform_data/cros_ec_proto.h |  1 +
> > >  2 files changed, 25 insertions(+)
> > > 
> > > diff --git a/drivers/platform/chrome/cros_ec.c
> > > b/drivers/platform/chrome/cros_ec.c
> > > index 9b2d07422e17..0c910846d99d 100644
> > > --- a/drivers/platform/chrome/cros_ec.c
> > > +++ b/drivers/platform/chrome/cros_ec.c
> > > @@ -104,6 +104,23 @@ static int cros_ec_sleep_event(struct
> > > cros_ec_device *ec_dev, u8 sleep_event)
> > >       return ret;
> > >  }
> > >  
> > > +static int cros_ec_ready_event(struct notifier_block *nb,
> > > +     unsigned long queued_during_suspend, void *_notify)
> > > +{
> > > +     struct cros_ec_device *ec_dev = container_of(nb, struct
> > > cros_ec_device,
> > > +                                                 
> > notifier_ready);
> > > +     u32 host_event = cros_ec_get_host_event(ec_dev);
> > > +
> > > +     if (host_event &
> > > EC_HOST_EVENT_MASK(EC_HOST_EVENT_INTERFACE_READY)) {
> > > +             mutex_lock(&ec_dev->lock);
> > > +             cros_ec_query_all(ec_dev);
> > > +             mutex_unlock(&ec_dev->lock);
> > > +             return NOTIFY_OK;
> > > +     } else {
> > > +             return NOTIFY_DONE;
> > > +     }
> > > +}
> > > +
> > >  /**
> > >   * cros_ec_register() - Register a new ChromeOS EC, using the
> > > provided info.
> > >   * @ec_dev: Device to register.
> > > @@ -201,6 +218,13 @@ int cros_ec_register(struct cros_ec_device
> > > *ec_dev)
> > >               dev_dbg(ec_dev->dev, "Error %d clearing sleep event
> > to
> > > ec",
> > >                       err);
> > >  
> > > +     /* Register the notifier for EC_HOST_EVENT_INTERFACE_READY
> > > event. */
> > > +     ec_dev->notifier_ready.notifier_call = cros_ec_ready_event;
> > > +     err = blocking_notifier_chain_register(&ec_dev-
> > >event_notifier,
> > > +                                            &ec_dev-
> > > >notifier_ready);
> > > +     if (err < 0)
> > > +             dev_warn(ec_dev->dev, "Failed to register
> > notifier\n");
> > > +
> > >       dev_info(dev, "Chrome EC device registered\n");
> > >  
> > >       return 0;
> > > diff --git a/include/linux/platform_data/cros_ec_proto.h
> > > b/include/linux/platform_data/cros_ec_proto.h
> > > index 0d4e4aaed37a..9840408c0b01 100644
> > > --- a/include/linux/platform_data/cros_ec_proto.h
> > > +++ b/include/linux/platform_data/cros_ec_proto.h
> > > @@ -161,6 +161,7 @@ struct cros_ec_device {
> > >       int event_size;
> > >       u32 host_event_wake_mask;
> > >       u32 last_resume_result;
> > > +     struct notifier_block notifier_ready;
> > >  
> > >       /* The platform devices used by the mfd driver */
> > >       struct platform_device *ec;
> > 

Re: [PATCH] mfd / platform: cros_ec: Query EC protocol version if EC transitions between RO/RW

[Fabien Lahoudere]

Hi,

With the patch applied,

   root@debian:/src/ec# ./build/bds/util/ectool --name=cros_fp
   reboot_ec                         
   ioctl -1, errno 110 (Connection timed out), EC result 255
   (<unknown>)
   root@debian:/src/ec# ./build/bds/util/ectool --name=cros_fp version
   RO version:    nocturne_fp_v2.2.64-58cf5974e
   RW version:    nocturne_fp_v2.2.191-1d529566e
   Firmware copy: RW
   Build info:    nocturne_fp_v2.2.191-1d529566e 2019-11-01 20:30:58
   @chromeos-ci-legacy-us-central1-b-x32-72-dltd
   Tool version:  v2.0.3074-a5052d4e7 2020-01-16 10:23:05 debian@debian

in that case we see that the firmware copy is RW after the reboot.

Without the patch :

   root@debian:/src/ec# ./build/bds/util/ectool --name=cros_fp
   reboot_ec
   ioctl -1, errno 110 (Connection timed out), EC result 255
   (<unknown>)
   root@debian:/src/ec# ./build/bds/util/ectool --name=cros_fp version
   RO version:    nocturne_fp_v2.2.64-58cf5974e
   RW version:    nocturne_fp_v2.2.191-1d529566e
   Firmware copy: RO
   Build info:    nocturne_fp_v2.2.64-58cf5974e 2018-10-01 19:11:56
   @cros-beefy532-c2
   Tool version:  v2.0.3074-a5052d4e7 2020-01-16 10:23:05 debian@debian
root@debian:/src/ec#

The firmware copy stays RO

For me the patch works fine

Tested-by: Fabien Lahoudere <fabien.lahoudere@collabora.com>

Thanks

Le mardi 21 janvier 2020 à 10:13 -0800, Yicheng Li a écrit :
> Hi Fabien,
> 
> error 4 is EC_RES_ACCESS_DENIED. I think it's because you are trying
> to directly reboot to RO, which issues the command EC_CMD_REBOOT_EC
> (0xD2)
> https://chromium.git.corp.google.com/chromiumos/platform/ec/+/15edeaabe90c554d83faa4ff0c9143c66248e0e4/common/system.c#1495
> and is ultimately disallowed since system_is_locked() is true: 
> https://chromium.git.corp.google.com/chromiumos/platform/ec/+/15edeaabe90c554d83faa4ff0c9143c66248e0e4/common/system.c#551
> 
> The way I test it is to run "ectool --name=cros_fp reboot_ec" without
> the "RO" argument. This issues EC_CMD_REBOOT 
> https://chromium.git.corp.google.com/chromiumos/platform/ec/+/15edeaabe90c554d83faa4ff0c9143c66248e0e4/util/ectool.c#1055
>  which is 0xD1. (Sorry I didn't notice the RO argument in the first
> pass...)
> 
> With "ectool --name=cros_fp reboot_ec", we are telling cros_fp to do
> an immediate reboot (to RW), however because of the issue outlined in
> the commit description, it would hang in RO without this patch.
> 
> Thanks,
> Yicheng
> 
> 
> 
> On Mon, Jan 20, 2020 at 3:16 AM Fabien Lahoudere <
> fabien.lahoudere@collabora.com> wrote:
> > Hi Yicheng,
> > 
> > Thanks for your quick answer.
> > 
> > I updated the firmware after my first email on friday.
> > Now the version is correct:
> > 
> >    # ectool version --name=cros_fp
> >    RO version:    nocturne_fp_v2.2.64-58cf5974e
> >    RW version:    nocturne_fp_v2.2.191-1d529566e
> >    Firmware copy: RW
> >    Build info:    nocturne_fp_v2.2.191-1d529566e 2019-11-01
> > 20:30:58
> >    @chromeos-ci-legacy-us-central1-b-x32-72-dltd
> >    Tool version:  v2.0.3031-9e0f24fad 2020-01-13 06:11:02
> > @chromeos-ci-
> >    legacy-us-east1-d-x32-89-i09h
> > 
> > The result does not change. This is the console output after I sent
> > the
> > reboot command.
> > 
> >    $ ectool name=cros_fp output
> >    ...
> >    [250664.176204 HC 0xa9]
> >    [250664.176287 HC 0xa9 err 1]
> >    [250664.177413 HC 0x67]
> >    [250664.177489 HC 0x67 err 9]
> >    [250664.178434 HC 0x02]
> >    [250664.180308 HC Suppressed: 0x97=127 0x98=138 0x115=0]
> >    [250664.477987 HC 0x01]
> >    [250892.898421 HC 0x02]
> >    [250892.900447 HC 0x07]
> >    [250892.901286 HC 0x0b]
> >    [250902.139053 HC 0x02]
> >    [250902.141059 HC 0x07]
> >    [250902.141847 HC 0x0b]
> >    [250902.142988 HC 0xd2]
> >    [250902.143072 Executing host reboot command 1]
> >    [250902.143176 HC 0xd2 err 4]
> >    [250908.838254 HC 0x02]
> >    [250908.840474 HC 0x07]
> >    [250908.841391 HC 0x0b]
> > 
> > it seems we have an error when executing the 0zd2 command.
> > 
> > Any idea what is error 4?
> > 
> > Thanks
> > 
> > Fabien
> > 
> > Le jeudi 16 janvier 2020 à 05:04 -0800, Yicheng Li a écrit :
> > > Hi Fabien,
> > > 
> > > First, you need the new firmware which has RW 2.2.191 instead of
> > > 2.2.110. It's available in release M80. The old firmware you were
> > > using does not have the RO/RW protocol incompatibility problem
> > > because RW 2.2.110 still only support versions 0 and 1 of
> > > EC_CMD_GET_NEXT_EVENT, same as RO 2.2.64. RW 2.2.191 has version
> > 2 of
> > > EC_CMD_GET_NEXT_ EVENT, hence the issue described in the commit
> > > message.
> > > 
> > > With RW 2.2.191, if the kernal does not have this patch,
> > rebooting
> > > cros_fp will leave you in an infinite loop in RO, as described in
> > the
> > > commit message. This patch solves the problem and you should be
> > in RW
> > > after the reboot. You can see more details on the commands sent
> > to
> > > cros_fp during RO if you have cros_fp.log 
> > > 
> > > Let me know if you have other questions. Thanks!
> > > 
> > > Best,
> > > Yicheng
> > > 
> > > 
> > > On Thu, Jan 16, 2020, 4:27 AM Fabien Lahoudere <
> > > fabien.lahoudere@collabora.com> wrote:
> > > > Hi,
> > > > 
> > > > I tried to test that patch but I cannot switch to RO firmware.
> > > > I do the following steps:
> > > > 
> > > > root@debian:/sys/class/chromeos/cros_fp# cat version 
> > > > RO version:    nocturne_fp_v2.2.64-58cf5974e
> > > > RW version:    nocturne_fp_v2.2.110-b936c0a3c
> > > > Firmware copy: RW
> > > > Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02
> > 14:16:46
> > > > @swarm-cros-461
> > > > Chip vendor:   stm
> > > > Chip name:     stm32h7x3
> > > > Chip revision: 
> > > > Board version: EC error 1
> > > > root@debian:/sys/class/chromeos/cros_fp# echo ro > reboot
> > > > root@debian:/sys/class/chromeos/cros_fp# cat version 
> > > > RO version:    nocturne_fp_v2.2.64-58cf5974e
> > > > RW version:    nocturne_fp_v2.2.110-b936c0a3c
> > > > Firmware copy: RW
> > > > Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02
> > 14:16:46
> > > > @swarm-cros-461
> > > > Chip vendor:   stm
> > > > Chip name:     stm32h7x3
> > > > Chip revision: 
> > > > Board version: EC error 1
> > > > root@debian:/sys/class/chromeos/cros_fp#
> > > > 
> > > > We see here that cros_fp is still RW.
> > > > 
> > > > I also tried with:
> > > > 
> > > > debian@debian:/src/ec$ sudo build/bds/util/ectool --
> > name=cros_fp
> > > > reboot_ec RO
> > > > debian@debian:/src/ec$ sudo build/bds/util/ectool --
> > name=cros_fp
> > > > version     
> > > > RO version:    nocturne_fp_v2.2.64-58cf5974e
> > > > RW version:    nocturne_fp_v2.2.110-b936c0a3c
> > > > Firmware copy: RW
> > > > Build info:    nocturne_fp_v2.2.110-b936c0a3c 2018-11-02
> > 14:16:46
> > > > @swarm-cros-461
> > > > Tool version:  v2.0.3074-a5052d4e7 2020-01-16 10:23:05 
> > > > debian@debian
> > > > debian@debian:/src/ec$
> > > > 
> > > > with the same result.
> > > > 
> > > > Can you decribe us steps you follow to test that patch?
> > > > 
> > > > Thanks
> > > > 
> > > > Fabien
> > > > 
> > > > Le lundi 18 novembre 2019 à 12:00 -0800, Yicheng Li a écrit :
> > > > > RO and RW of EC may have different EC protocol version. If EC
> > > > > transitions
> > > > > between RO and RW, but AP does not reboot (this is true for
> > > > > fingerprint
> > > > > microcontroller / cros_fp, but not true for main ec /
> > cros_ec),
> > > > the
> > > > > AP
> > > > > still uses the protocol version queried before transition,
> > which
> > > > can
> > > > > cause problems. In the case of fingerprint microcontroller,
> > this
> > > > > causes
> > > > > AP to send the wrong version of EC_CMD_GET_NEXT_EVENT to RO
> > in
> > > > the
> > > > > interrupt handler, which in turn prevents RO to clear the
> > > > interrupt
> > > > > line to AP, in an infinite loop.
> > > > > 
> > > > > Once an EC_HOST_EVENT_INTERFACE_READY is received, we know
> > that
> > > > there
> > > > > might have been a transition between RO and RW, so re-query
> > the
> > > > > protocol.
> > > > > 
> > > > > Signed-off-by: Yicheng Li <yichengli@chromium.org>
> > > > > 
> > > > > Change-Id: Ib58032ff4a8e113bdbd07212e8aff42807afff38
> > > > > Series-to: LKML <linux-kernel@vger.kernel.org>
> > > > > Series-cc: Benson Leung <bleung@chromium.org>, Enric Balletbo
> > i
> > > > Serra
> > > > > <enric.balletbo@collabora.com>, Gwendal Grignou <
> > > > gwendal@chromium.org
> > > > > >
> > > > > ---
> > > > >  drivers/platform/chrome/cros_ec.c           | 24
> > > > > +++++++++++++++++++++
> > > > >  include/linux/platform_data/cros_ec_proto.h |  1 +
> > > > >  2 files changed, 25 insertions(+)
> > > > > 
> > > > > diff --git a/drivers/platform/chrome/cros_ec.c
> > > > > b/drivers/platform/chrome/cros_ec.c
> > > > > index 9b2d07422e17..0c910846d99d 100644
> > > > > --- a/drivers/platform/chrome/cros_ec.c
> > > > > +++ b/drivers/platform/chrome/cros_ec.c
> > > > > @@ -104,6 +104,23 @@ static int cros_ec_sleep_event(struct
> > > > > cros_ec_device *ec_dev, u8 sleep_event)
> > > > >       return ret;
> > > > >  }
> > > > >  
> > > > > +static int cros_ec_ready_event(struct notifier_block *nb,
> > > > > +     unsigned long queued_during_suspend, void *_notify)
> > > > > +{
> > > > > +     struct cros_ec_device *ec_dev = container_of(nb, struct
> > > > > cros_ec_device,
> > > > > +                                                 
> > > > notifier_ready);
> > > > > +     u32 host_event = cros_ec_get_host_event(ec_dev);
> > > > > +
> > > > > +     if (host_event &
> > > > > EC_HOST_EVENT_MASK(EC_HOST_EVENT_INTERFACE_READY)) {
> > > > > +             mutex_lock(&ec_dev->lock);
> > > > > +             cros_ec_query_all(ec_dev);
> > > > > +             mutex_unlock(&ec_dev->lock);
> > > > > +             return NOTIFY_OK;
> > > > > +     } else {
> > > > > +             return NOTIFY_DONE;
> > > > > +     }
> > > > > +}
> > > > > +
> > > > >  /**
> > > > >   * cros_ec_register() - Register a new ChromeOS EC, using
> > the
> > > > > provided info.
> > > > >   * @ec_dev: Device to register.
> > > > > @@ -201,6 +218,13 @@ int cros_ec_register(struct
> > cros_ec_device
> > > > > *ec_dev)
> > > > >               dev_dbg(ec_dev->dev, "Error %d clearing sleep
> > event
> > > > to
> > > > > ec",
> > > > >                       err);
> > > > >  
> > > > > +     /* Register the notifier for
> > EC_HOST_EVENT_INTERFACE_READY
> > > > > event. */
> > > > > +     ec_dev->notifier_ready.notifier_call =
> > cros_ec_ready_event;
> > > > > +     err = blocking_notifier_chain_register(&ec_dev-
> > > > >event_notifier,
> > > > > +                                            &ec_dev-
> > > > > >notifier_ready);
> > > > > +     if (err < 0)
> > > > > +             dev_warn(ec_dev->dev, "Failed to register
> > > > notifier\n");
> > > > > +
> > > > >       dev_info(dev, "Chrome EC device registered\n");
> > > > >  
> > > > >       return 0;
> > > > > diff --git a/include/linux/platform_data/cros_ec_proto.h
> > > > > b/include/linux/platform_data/cros_ec_proto.h
> > > > > index 0d4e4aaed37a..9840408c0b01 100644
> > > > > --- a/include/linux/platform_data/cros_ec_proto.h
> > > > > +++ b/include/linux/platform_data/cros_ec_proto.h
> > > > > @@ -161,6 +161,7 @@ struct cros_ec_device {
> > > > >       int event_size;
> > > > >       u32 host_event_wake_mask;
> > > > >       u32 last_resume_result;
> > > > > +     struct notifier_block notifier_ready;
> > > > >  
> > > > >       /* The platform devices used by the mfd driver */
> > > > >       struct platform_device *ec;
> > > > 
> >