[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAHp75VdrS=aSZvg2LwxwV0sR03SQb6do=aG5KGmgs1WKyvDirg@mail.gmail.com>
Date: Sun, 18 Jun 2017 16:27:42 +0300
From: Andy Shevchenko <andy.shevchenko@...il.com>
To: Okash Khawaja <okash.khawaja@...il.com>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
Jiri Slaby <jslaby@...e.com>,
Samuel Thibault <samuel.thibault@...-lyon.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
William Hubbs <w.d.hubbs@...il.com>,
Chris Brannon <chris@...-brannons.com>,
Kirk Reiser <kirk@...sers.ca>, speakup@...ux-speakup.org,
devel@...verdev.osuosl.org
Subject: Re: [patch v2 1/3] tty: add function to convert device name to number
On Sun, Jun 18, 2017 at 11:58 AM, Okash Khawaja <okash.khawaja@...il.com> wrote:
> The function converts strings like ttyS0 and ttyUSB0 to dev_t like
> (4, 64) and (188, 0). It does this by scanning tty_drivers list for
> corresponding device name and index. If the driver is not registered,
> this function returns -ENODEV. It also acquires tty_mutex.
Nice!
Few comments below.
> +/**
> + * tty_dev_name_to_number - return dev_t for device name
> + * @device_name: user space name of device under /dev
This doesn't have actual parameter name.
Btw, I would drop dev_ suffix completely from parameter (you have it
in function name).
> + * @dev_no: pointer to dev_t that this function will populate
Here I'm not sure if dev_ prefix is good to have or not.
> + *
> + * This function converts device names like ttyS0 or ttyUSB1 into dev_t
> + * like (4, 64) or (188, 1). If no corresponding driver is registered then
> + * the function returns -ENODEV.
> + *
> + * Locking: this acquires tty_mutex
...and releases.
Perhaps it makes sense to describe what it protects (like it's done in
some functions around).
> + */
> +int tty_dev_name_to_number(char *dev_name, dev_t *dev_no)
const char *name, right?
> +{
> + struct tty_driver *p;
> + int rv, index, prefix_length = 0;
I would keep returned variable on a separate line and name it like
other functions do in this file, i.e. ret.
int ret;
> + while (!isdigit(*(dev_name + prefix_length)) && prefix_length <
> + strlen(dev_name) )
> + prefix_length++;
> +
> + if (prefix_length == strlen(dev_name))
> + return -EINVAL;
Basically, what you need is to get tailing digits, right?
Moreover, there is quite similar piece of code in
tty_find_polling_driver() you may share.
> + mutex_lock(&tty_mutex);
> +
> + list_for_each_entry(p, &tty_drivers, tty_drivers)
> + if (prefix_length == strlen(p->name) && strncmp(dev_name,
> + p->name, prefix_length) == 0) {
> + rv = kstrtoint(dev_name + prefix_length, 10, &index);
> + if (rv) {
> + mutex_unlock(&tty_mutex);
> + return rv;
I would go with goto style in this function (since it has locking involved).
> + }
All together kstrtoint() is invariant here as far as I can see and can
be done out of locking. Also see above comment how to get line index.
> + if (index < p->num) {
> + *dev_no = MKDEV(p->major, p->minor_start + index);
> + mutex_unlock(&tty_mutex);
> + return 0;
> + }
> + }
> +
> + mutex_unlock(&tty_mutex);
> + return -ENODEV;
> +}
> +EXPORT_SYMBOL_GPL(tty_dev_name_to_number);
--
With Best Regards,
Andy Shevchenko
Powered by blists - more mailing lists