From: Wolfram Sang <wsa@the-dreams.de>
To: Bastian Hecht <hechtb@gmail.com>
Cc: linux-i2c@vger.kernel.org, Linux-SH <linux-sh@vger.kernel.org>,
Tomoya MORINAGA <tomoya.rohm@gmail.com>,
Naveen Krishna Chatradhi <ch.naveen@samsung.com>,
Taekgyun Ko <taeggyun.ko@samsung.com>,
Ben Dooks <ben@simtec.co.uk>, Magnus Damm <magnus.damm@gmail.com>
Subject: Re: [PATCH 1/5] i2c: Don't start transfers when suspended
Date: Thu, 17 Jul 2014 15:00:48 +0200 [thread overview]
Message-ID: <20140717130047.GF2740@katana> (raw)
In-Reply-To: <1405165771-8732-1-git-send-email-hechtb@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 3996 bytes --]
Hi Bastian,
On Sat, Jul 12, 2014 at 01:49:27PM +0200, Bastian Hecht wrote:
> i2c transfer requests come in very uncontrolled, like from interrupt routines.
> We might be suspended when this happens. To avoid i2c timeouts caused by
> powered down busses we check for suspension.
>
> Signed-off-by: Bastian Hecht <hechtb@gmail.com>
Thanks for doing this! I add linux-pm to CC because of two questions.
> ---
> I ended up getting i2c timeouts when suspending and provoking touchscreen
> IRQs on an Armadillo board with an sh_mobile i2c bus. Instead of copying
> again the protection mechanism from other bus drivers, we can clean things
> up and move this into the core.
This could be part of the commit message IMO.
>
> I assume that all bus drivers use kzalloc or equivalent for the
> struct i2c_adapter. So adap->suspended can't end up uninitialized,
> right?
Better safe than sorry, I'd say. Please add the initialization.
> We could consider using this scheme for freeze/restore too.
linux-pm: opinions?
>
>
> drivers/i2c/i2c-core.c | 24 +++++++++++++++++++++++-
> include/linux/i2c.h | 1 +
> 2 files changed, 24 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/i2c/i2c-core.c b/drivers/i2c/i2c-core.c
> index 7c7f4b8..9fe9581 100644
> --- a/drivers/i2c/i2c-core.c
> +++ b/drivers/i2c/i2c-core.c
> @@ -343,6 +343,13 @@ static int i2c_legacy_resume(struct device *dev)
> static int i2c_device_pm_suspend(struct device *dev)
> {
> const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
> + struct i2c_adapter *adap = i2c_verify_adapter(dev);
> +
> + if (adap) {
> + i2c_lock_adapter(adap);
> + adap->suspended = true;
> + i2c_unlock_adapter(adap);
> + }
>
> if (pm)
> return pm_generic_suspend(dev);
> @@ -353,6 +360,13 @@ static int i2c_device_pm_suspend(struct device *dev)
> static int i2c_device_pm_resume(struct device *dev)
> {
> const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
> + struct i2c_adapter *adap = i2c_verify_adapter(dev);
> +
> + if (adap) {
> + i2c_lock_adapter(adap);
> + adap->suspended = false;
> + i2c_unlock_adapter(adap);
> + }
>
> if (pm)
> return pm_generic_resume(dev);
> @@ -1819,7 +1833,10 @@ int i2c_transfer(struct i2c_adapter *adap, struct i2c_msg *msgs, int num)
> i2c_lock_adapter(adap);
> }
>
> - ret = __i2c_transfer(adap, msgs, num);
> + if (adap->suspended)
> + ret = -EIO;
> + else
> + ret = __i2c_transfer(adap, msgs, num);
a) Minor: I'd swap the branches and invert the condition, so that the
more expected code path comes first
b) linux-pm: I wonder about the return code. Currently, some drivers
return -EIO, some -EBUSY. Do you care which? Since we try to have
designated fault codes in the i2c subsystem, I wonder about -ESHUTDOWN?
Or is this too much of hijacking fault codes?
> i2c_unlock_adapter(adap);
>
> return ret;
> @@ -2577,6 +2594,10 @@ s32 i2c_smbus_xfer(struct i2c_adapter *adapter, u16 addr, unsigned short flags,
>
> if (adapter->algo->smbus_xfer) {
> i2c_lock_adapter(adapter);
> + if (adapter->suspended) {
> + res = -EIO;
> + goto unlock;
> + }
>
> /* Retry automatically on arbitration loss */
> orig_jiffies = jiffies;
> @@ -2590,6 +2611,7 @@ s32 i2c_smbus_xfer(struct i2c_adapter *adapter, u16 addr, unsigned short flags,
> orig_jiffies + adapter->timeout))
> break;
> }
> +unlock:
> i2c_unlock_adapter(adapter);
>
> if (res != -EOPNOTSUPP || !adapter->algo->master_xfer)
> diff --git a/include/linux/i2c.h b/include/linux/i2c.h
> index b556e0a..af08c75 100644
> --- a/include/linux/i2c.h
> +++ b/include/linux/i2c.h
> @@ -434,6 +434,7 @@ struct i2c_adapter {
> int timeout; /* in jiffies */
> int retries;
> struct device dev; /* the adapter device */
> + unsigned int suspended:1;
>
> int nr;
> char name[48];
> --
> 1.9.1
>
[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]
next prev parent reply other threads:[~2014-07-17 13:00 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-07-12 11:49 [PATCH 1/5] i2c: Don't start transfers when suspended Bastian Hecht
2014-07-12 11:49 ` [PATCH 4/5] i2c: sc3c2410: Remove suspension check Bastian Hecht
[not found] ` <1405165771-8732-1-git-send-email-hechtb-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
2014-07-12 11:49 ` [PATCH 2/5] i2c: eg20t: " Bastian Hecht
2014-07-12 11:49 ` [PATCH 3/5] i2c: exynos5: " Bastian Hecht
2014-07-12 11:49 ` [PATCH 5/5] i2c: tegra: " Bastian Hecht
2014-07-17 13:00 ` Wolfram Sang [this message]
2014-07-17 13:04 ` [PATCH 1/5] i2c: Don't start transfers when suspended Wolfram Sang
2014-07-17 14:34 ` Bastian Hecht
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20140717130047.GF2740@katana \
--to=wsa@the-dreams.de \
--cc=ben@simtec.co.uk \
--cc=ch.naveen@samsung.com \
--cc=hechtb@gmail.com \
--cc=linux-i2c@vger.kernel.org \
--cc=linux-sh@vger.kernel.org \
--cc=magnus.damm@gmail.com \
--cc=taeggyun.ko@samsung.com \
--cc=tomoya.rohm@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).