| 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
| ||
|
Message-ID: <CAAObsKDc3pS32QWB47AhawmE7i2MiMXR_xzE7i+jjaQNBWP+BA@mail.gmail.com>
Date: Mon, 11 Apr 2016 13:45:47 +0200
From: Tomeu Vizoso <tomeu.vizoso@...labora.com>
To: Lee Jones <lee.jones@...aro.org>
Cc: "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
Sameer Nanda <snanda@...omium.org>,
Javier Martinez Canillas <javier@....samsung.com>,
Benson Leung <bleung@...omium.org>,
Enric Balletbò <enric.balletbo@...labora.co.uk>,
Vic Yang <victoryang@...omium.org>,
Stephen Boyd <sboyd@...eaurora.org>,
Vincent Palatin <vpalatin@...omium.org>,
Randall Spangler <rspangler@...omium.org>,
Todd Broch <tbroch@...omium.org>,
Gwendal Grignou <gwendal@...omium.org>,
Vic Yang <victoryang@...gle.com>,
Olof Johansson <olof@...om.net>, linux-input@...r.kernel.org,
Dmitry Torokhov <dmitry.torokhov@...il.com>
Subject: Re: [RESEND PATCH v7 1/6] mfd: cros_ec: Add MKBP event support
On 7 April 2016 at 17:29, Lee Jones <lee.jones@...aro.org> wrote:
> On Tue, 05 Apr 2016, Tomeu Vizoso wrote:
>
>> From: Vic Yang <victoryang@...gle.com>
>>
>> Newer revisions of the ChromeOS EC add more events besides the keyboard
>> ones. So handle interrupts in the MFD driver and let consumers register
>> for notifications for the events they might care.
>>
>> To keep backward compatibility, if the EC doesn't support MKBP event, we
>> fall back to the old MKBP key matrix host command.
>>
>> Signed-off-by: Vic Yang <victoryang@...omium.org>
>> [bleung: fixup some context changes going from v3.14 to v3.18]
>> Signed-off-by: Benson Leung <bleung@...omium.org>
>> [tomeu: adapted to changes in mainline (in power-supply and platform/chrome)]
>> Signed-off-by: Tomeu Vizoso <tomeu.vizoso@...labora.com>
>
> I'm not interested in your BSP submission path. As far as I'm
> concerned *this* is the first submission. If these guys are happy
> with the patch, they can either choose to Ack or Review it. Drop the
> blurb in the middle.
Ok.
>> Cc: Randall Spangler <rspangler@...omium.org>
>> Cc: Vincent Palatin <vpalatin@...omium.org>
>> ---
>>
>> Changes in v7: None
>> Changes in v6: None
>> Changes in v5: None
>> Changes in v4:
>> - Calculate correctly the size of the payloads in
>> cros_ec_get_host_command_version_mask.
>>
>> Changes in v3:
>> - Remove duplicated prototype of cros_ec_get_host_event.
>>
>> Changes in v2:
>> - Allocate enough for the structs in cros_ec_get_host_command_version_mask,
>> not their pointers.
>> - Allocate msg in the stack in get_next_event and
>> get_keyboard_state_event, as suggested by Gwendal.
>>
>> drivers/input/keyboard/cros_ec_keyb.c | 135 ++++++++------------------------
>> drivers/mfd/cros_ec.c | 57 +++++++++++++-
>> drivers/platform/chrome/cros_ec_proto.c | 92 ++++++++++++++++++++++
>> include/linux/mfd/cros_ec.h | 34 ++++++++
>> include/linux/mfd/cros_ec_commands.h | 34 ++++++++
>
> What are the *build time* dependencies that warrant all of these
> changes happening in one patch?
We can split further the changes without breaking the builds, but
cros-ec functionality will be broken in between if both the mfd driver
and the input driver handle the same interrupt, which will hurt
bisectability.
>> 5 files changed, 245 insertions(+), 107 deletions(-)
>
> [...]
>
>> diff --git a/drivers/mfd/cros_ec.c b/drivers/mfd/cros_ec.c
>> index 0eee63542038..fbe78b819fdd 100644
>> --- a/drivers/mfd/cros_ec.c
>> +++ b/drivers/mfd/cros_ec.c
>> @@ -23,6 +23,7 @@
>> #include <linux/module.h>
>> #include <linux/mfd/core.h>
>> #include <linux/mfd/cros_ec.h>
>> +#include <asm/unaligned.h>
>>
>> #define CROS_EC_DEV_EC_INDEX 0
>> #define CROS_EC_DEV_PD_INDEX 1
>> @@ -49,11 +50,28 @@ static const struct mfd_cell ec_pd_cell = {
>> .pdata_size = sizeof(pd_p),
>> };
>>
>> +static irqreturn_t ec_irq_thread(int irq, void *data)
>> +{
>> + struct cros_ec_device *ec_dev = data;
>> + int ret;
>> +
>> + if (device_may_wakeup(ec_dev->dev))
>> + pm_wakeup_event(ec_dev->dev, 0);
>> +
>> + ret = cros_ec_get_next_event(ec_dev);
>> + if (ret > 0)
>> + blocking_notifier_call_chain(&ec_dev->event_notifier,
>> + 0, ec_dev);
>> + return IRQ_HANDLED;
>> +}
>> +
>> int cros_ec_register(struct cros_ec_device *ec_dev)
>> {
>> struct device *dev = ec_dev->dev;
>> int err = 0;
>>
>> + BLOCKING_INIT_NOTIFIER_HEAD(&ec_dev->event_notifier);
>> +
>> ec_dev->max_request = sizeof(struct ec_params_hello);
>> ec_dev->max_response = sizeof(struct ec_response_get_protocol_info);
>> ec_dev->max_passthru = 0;
>> @@ -70,13 +88,24 @@ int cros_ec_register(struct cros_ec_device *ec_dev)
>>
>> cros_ec_query_all(ec_dev);
>>
>> + if (ec_dev->irq) {
>> + err = request_threaded_irq(ec_dev->irq, NULL, ec_irq_thread,
>> + IRQF_TRIGGER_LOW | IRQF_ONESHOT,
>> + "chromeos-ec", ec_dev);
>> + if (err) {
>> + dev_err(dev, "request irq %d: error %d\n",
>
> This is an ugly error message. Write them like you (as I user) would
> like to see. I suggest using proper English and grammar.
>
> "Failed to request IRQ %d: %d", irq, err ?
Sounds good.
>> + ec_dev->irq, err);
>> + return err;
>> + }
>> + }
>> +
>> err = mfd_add_devices(ec_dev->dev, PLATFORM_DEVID_AUTO, &ec_cell, 1,
>> NULL, ec_dev->irq, NULL);
>> if (err) {
>> dev_err(dev,
>> "Failed to register Embedded Controller subdevice %d\n",
>> err);
>> - return err;
>> + goto fail_mfd;
>> }
>>
>> if (ec_dev->max_passthru) {
>> @@ -94,7 +123,7 @@ int cros_ec_register(struct cros_ec_device *ec_dev)
>> dev_err(dev,
>> "Failed to register Power Delivery subdevice %d\n",
>> err);
>> - return err;
>> + goto fail_mfd;
>> }
>> }
>>
>> @@ -103,13 +132,18 @@ int cros_ec_register(struct cros_ec_device *ec_dev)
>> if (err) {
>> mfd_remove_devices(dev);
>> dev_err(dev, "Failed to register sub-devices\n");
>> - return err;
>> + goto fail_mfd;
>> }
>> }
>>
>> dev_info(dev, "Chrome EC device registered\n");
>>
>> return 0;
>> +
>> +fail_mfd:
>> + if (ec_dev->irq)
>> + free_irq(ec_dev->irq, ec_dev);
>> + return err;
>> }
>> EXPORT_SYMBOL(cros_ec_register);
>>
>> @@ -136,13 +170,30 @@ int cros_ec_suspend(struct cros_ec_device *ec_dev)
>> }
>> EXPORT_SYMBOL(cros_ec_suspend);
>>
>> +static void cros_ec_drain_events(struct cros_ec_device *ec_dev)
>> +{
>> + while (cros_ec_get_next_event(ec_dev) > 0)
>> + blocking_notifier_call_chain(&ec_dev->event_notifier,
>> + 1, ec_dev);
>> +}
>> +
>> int cros_ec_resume(struct cros_ec_device *ec_dev)
>> {
>> enable_irq(ec_dev->irq);
>>
>> + /*
>> + * In some case, we need to distinguish events that occur during
>
> s/case/cases/
>
> s/distinguish/distinguish between/
Cool.
>> diff --git a/include/linux/mfd/cros_ec.h b/include/linux/mfd/cros_ec.h
>> index a677c2bd485c..ddc935ef1911 100644
>> --- a/include/linux/mfd/cros_ec.h
>> +++ b/include/linux/mfd/cros_ec.h
>> @@ -72,6 +72,24 @@ struct cros_ec_command {
>> uint8_t data[0];
>> };
>>
>> +/*
>> + * event_data is used by keyboard or event notifier:
>> + * event_data format:
>> + * If MKBP protocol is supported:
>> + * 0 1
>> + * +-----------+--------------------------------
>> + * | type | payload
>> + * +-----------+--------------------------------
>> + * |HOST_EVENT | EVENT (32 bit)
>> + * |KEY_MATRIX | Keyboard keys pressed.
>> + * |SENSOR_FIFO| Sensors FIFO information.
>> + *
>> + * Otherwise:
>> + * 0 1
>> + * +-----------+--------------------------------
>> + * |Unused | Keyboard keys pressed.
>> + */
>
> Personally, I don't think this documentation is required. But if you
> insist on supplying it, I think it'll be better placed near the
> 'struct ec_response_get_next_event' definition.
I'm personally fine with leaving it out.
>> /**
>> * struct cros_ec_device - Information about a ChromeOS EC device
>> *
>> @@ -107,6 +125,9 @@ struct cros_ec_command {
>> * should check msg.result for the EC's result code.
>> * @pkt_xfer: send packet to EC and get response
>> * @lock: one transaction at a time
>> + * @event_notifier: interrupt event notifier for transport devices.
>> + * @event_data: raw payload transferred with the MKBP event.
>> + * @event_size: size in bytes of the event data.
>> */
>> struct cros_ec_device {
>>
>> @@ -135,6 +156,11 @@ struct cros_ec_device {
>> int (*pkt_xfer)(struct cros_ec_device *ec,
>> struct cros_ec_command *msg);
>> struct mutex lock;
>> + bool mkbp_event_supported;
>
> Did you document this?
Oops.
>> + struct blocking_notifier_head event_notifier;
>> +
>> + struct ec_response_get_next_event event_data;
>> + int event_size;
>> };
>>
>> /* struct cros_ec_platform - ChromeOS EC platform information
>> @@ -252,6 +278,14 @@ int cros_ec_register(struct cros_ec_device *ec_dev);
>> */
>> int cros_ec_query_all(struct cros_ec_device *ec_dev);
>>
>> +/**
>> + * cros_ec_get_next_event - Fetch next event from the ChromeOS EC
>> + *
>> + * @ec_dev: Device to fetch event from
>> + * @return 0 if ok, -ve on error
>
> I'd prefer easy to read/descriptive over trying to be smart.
>
> The 'return' value doesn' require a @. Instead, the return section
> should look like "Return: <blah>".
>
> I suggest: "Return: 0 on success, Linux error number on failure"
Sounds good to me.
>> + */
>> +int cros_ec_get_next_event(struct cros_ec_device *ec_dev);
>> +
>> /* sysfs stuff */
>> extern struct attribute_group cros_ec_attr_group;
>> extern struct attribute_group cros_ec_lightbar_attr_group;
>> diff --git a/include/linux/mfd/cros_ec_commands.h b/include/linux/mfd/cros_ec_commands.h
>> index 13b630c10d4c..d86526f0bd8e 100644
>> --- a/include/linux/mfd/cros_ec_commands.h
>> +++ b/include/linux/mfd/cros_ec_commands.h
>> @@ -1762,6 +1762,40 @@ struct ec_result_keyscan_seq_ctrl {
>> };
>> } __packed;
>>
>> +/*
>> + * Get the next pending MKBP event.
>> + *
>> + * Returns EC_RES_UNAVAILABLE if there is no event pending.
>> + */
>
> You're documenting this command as if it's a function. This command
> does nothing by it's self, rather it is supplied to a function call,
> which does the work. Similarly this command returns nothing, the
> device will provide the UNAVAILABLE return value. Please update the
> comment.
I see. This was following the style of the existing command docs, so
it may end up being a bit surprising.
Thanks a lot,
Tomeu
Powered by blists - more mailing lists