From: "Émeric Vigier" <emeric.vigier-4ysUXcep3aM1wj+D4I0NRVaTQe2KTcn/@public.gmane.org>
To: linux-i2c-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Subject: how to gracefully unload an i2c driver if chip not detected?
Date: Mon, 18 Mar 2013 10:47:26 -0400 (EDT) [thread overview]
Message-ID: <35709576.270615.1363618046395.JavaMail.root@mail> (raw)
In-Reply-To: <931575758.126890.1363615027376.JavaMail.root@mail>
Hi,
I work on an omap platform running linux 3.0.21. The kernel image embeds two built-in i2c drivers for two different proximity sensors.
The hardware platform embeds only one sensor, so there is a detection mechanism of which chip is present in each driver (code below).
I recently changed the sensor on my board. Leading to kernel crash when entering suspend.
Thanks to "no_console_suspend" cmdline argument, I found out that the suspend function of the "absent" chip gets called. It tries to take a mutex which has been freed in probe's "device not found" fallback code. Leading to kernel panic.
I naively tried to add "i2c_del_driver(&vcnl4010_driver);" in the driver's probe function.
But it triggers a null ptr deref in i2c_do_del_adapter().
How can I gracefully unload this i2c driver if chip is not detected, while keeping this driver built-in in the kernel?
What is the recommended method to do just that?
Thanks,
Emeric
---------
static struct i2c_driver vcnl4010_driver;
static int __devinit vcnl4010_driver_probe(struct i2c_client *client, const struct i2c_device_id *id)
{
[...]
if(vcnl4010_read_transfer(data, VCNL4010_PROD_ID_VER, ®_val, 1) !=0 )
{
pr_err("vcnl4010: Device not found!");
goto nochip;
}
if(reg_val != 0x21)
{
pr_err("vcnl4010: Found device isn't a vcnl4010, is a vcnl4000 installed?");
goto badchip;
}
[...]
badchip:
nochip:
input_unregister_device(data->input_dev);
dev_register_error:
input_free_device(data->input_dev);
dev_allocate_error:
mutex_destroy(&data->lock);
kfree(data);
error:
/* i2c_del_driver(&vcnl4010_driver); triggers NULL ptr deref in i2c_do_del_adapter() */
return ret;
}
static int vcnl4010_driver_suspend(struct device *dev)
{
struct platform_device *pdev = to_platform_device(dev);
struct vcnl4010_data *data = platform_get_drvdata(pdev);
mutex_lock(&data->lock); /* panic kernel if chip is not present */
vcnl4010_write(data, VCNL4010_CMD_REG, 0x00);
vcnl4010_write(data, VCNL4010_IR_LED_CURR, 0);
mutex_unlock(&data->lock);
return 0;
}
static struct i2c_driver vcnl4010_driver = {
.probe = vcnl4010_driver_probe,
.remove = vcnl4010_driver_remove,
.id_table = vcnl4010_idtable,
.driver = {
.name = DRIVER_NAME,
#ifdef CONFIG_PM
.pm = &vcnl4010_pm_ops,
#endif
},
};
next parent reply other threads:[~2013-03-18 14:47 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <931575758.126890.1363615027376.JavaMail.root@mail>
2013-03-18 14:47 ` Émeric Vigier [this message]
2013-04-09 9:26 ` how to gracefully unload an i2c driver if chip not detected? Wolfram Sang
[not found] ` <20130409092625.GD3624-z923LK4zBo2bacvFa/9K2g@public.gmane.org>
2013-04-21 3:38 ` Émeric Vigier
2013-04-21 15:48 ` Wolfram Sang
[not found] ` <20130421154832.GA9593-z923LK4zBo2bacvFa/9K2g@public.gmane.org>
2013-04-22 11:18 ` Émeric Vigier
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=35709576.270615.1363618046395.JavaMail.root@mail \
--to=emeric.vigier-4ysuxcep3am1wj+d4i0nrvatqe2ktcn/@public.gmane.org \
--cc=linux-i2c-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
/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).