From: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
To: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
Cc: Linus Walleij <linusw@kernel.org>,
Bartosz Golaszewski <brgl@kernel.org>,
Kent Gibson <warthog618@gmail.com>,
Adrian Ambrozewicz <aambrozewicz@nvidia.com>,
linux-gpio@vger.kernel.org, linux-kernel@vger.kernel.org,
stable@vger.kernel.org
Subject: Re: [PATCH v2 5/5] gpio: cdev: don't overwrite the whole flags field when configuring lines
Date: Mon, 9 Nov 2026 23:58:41 +0200 [thread overview]
Message-ID: <avJCETUZidpmICKR@ashevche-desk.local> (raw)
In-Reply-To: <20261109-gpio-desc-flags-atomic-v2-5-c4c54b476628@oss.qualcomm.com>
On Mon, Nov 09, 2026 at 11:27:09AM +0100, Bartosz Golaszewski wrote:
> linehandle_flags_to_desc_flags() and
> gpio_v2_line_config_flags_to_desc_flags() read the flags word with
> READ_ONCE(), modify the local copy and write it back with WRITE_ONCE().
>
> The per-request config_mutex does not exclude sysfs, the irq core or
> drivers touching the same descriptor, so bits they set or clear
> concurrently can be lost.
>
> Build the mask of bits owned by the line configuration and their new
> values locally and apply them with gpiod_update_flags(). The whole
> group changes in a single atomic step.
>
> While at it: rename the functions whose semantics changed from
> "conversion" to "flags update" accordingly.
...
> @@ -3,6 +3,7 @@
> #include <linux/anon_inodes.h>
> #include <linux/atomic.h>
> #include <linux/bitmap.h>
> +#include <linux/bitops.h>
> #include <linux/build_bug.h>
> #include <linux/cdev.h>
> #include <linux/cleanup.h>
Unneeded hunk as explained in the previous reply.
...
> -static void linehandle_flags_to_desc_flags(u32 lflags, unsigned long *flagsp)
> +static void linehandle_update_desc_flags(u32 lflags, struct gpio_desc *desc)
> {
> - unsigned long flags = READ_ONCE(*flagsp);
> + const unsigned long mask = BIT(GPIOD_FLAG_ACTIVE_LOW) |
> + BIT(GPIOD_FLAG_OPEN_DRAIN) |
> + BIT(GPIOD_FLAG_OPEN_SOURCE) |
> + BIT(GPIOD_FLAG_PULL_UP) |
> + BIT(GPIOD_FLAG_PULL_DOWN) |
> + BIT(GPIOD_FLAG_BIAS_DISABLE);
Usually we don't do such an assignments in the definition blocks.
I believe the split will look better.
> + unsigned long flags = 0;
>
mask = ...;
(and with longer line you may combine close bits, like OD/OS or PD,PU, and
DISABLE).
> - assign_bit(GPIOD_FLAG_ACTIVE_LOW, &flags,
> - lflags & GPIOHANDLE_REQUEST_ACTIVE_LOW);
> - assign_bit(GPIOD_FLAG_OPEN_DRAIN, &flags,
> - lflags & GPIOHANDLE_REQUEST_OPEN_DRAIN);
> - assign_bit(GPIOD_FLAG_OPEN_SOURCE, &flags,
> - lflags & GPIOHANDLE_REQUEST_OPEN_SOURCE);
> - assign_bit(GPIOD_FLAG_PULL_UP, &flags,
> - lflags & GPIOHANDLE_REQUEST_BIAS_PULL_UP);
> - assign_bit(GPIOD_FLAG_PULL_DOWN, &flags,
> - lflags & GPIOHANDLE_REQUEST_BIAS_PULL_DOWN);
> - assign_bit(GPIOD_FLAG_BIAS_DISABLE, &flags,
> - lflags & GPIOHANDLE_REQUEST_BIAS_DISABLE);
> + __assign_bit(GPIOD_FLAG_ACTIVE_LOW, &flags,
> + lflags & GPIOHANDLE_REQUEST_ACTIVE_LOW);
> + __assign_bit(GPIOD_FLAG_OPEN_DRAIN, &flags,
> + lflags & GPIOHANDLE_REQUEST_OPEN_DRAIN);
> + __assign_bit(GPIOD_FLAG_OPEN_SOURCE, &flags,
> + lflags & GPIOHANDLE_REQUEST_OPEN_SOURCE);
> + __assign_bit(GPIOD_FLAG_PULL_UP, &flags,
> + lflags & GPIOHANDLE_REQUEST_BIAS_PULL_UP);
> + __assign_bit(GPIOD_FLAG_PULL_DOWN, &flags,
> + lflags & GPIOHANDLE_REQUEST_BIAS_PULL_DOWN);
> + __assign_bit(GPIOD_FLAG_BIAS_DISABLE, &flags,
> + lflags & GPIOHANDLE_REQUEST_BIAS_DISABLE);
...
> +static void linereq_update_desc_flags(u64 lflags,
> + struct gpio_desc *desc)
This is ugliest part of this patch in my opinion. Just make it a single line.
`clang-format` is in use I assume?
...
> + const unsigned long mask = BIT(GPIOD_FLAG_ACTIVE_LOW) |
> + BIT(GPIOD_FLAG_EDGE_RISING) |
> + BIT(GPIOD_FLAG_EDGE_FALLING) |
> + BIT(GPIOD_FLAG_OPEN_DRAIN) |
> + BIT(GPIOD_FLAG_OPEN_SOURCE) |
> + BIT(GPIOD_FLAG_PULL_UP) |
> + BIT(GPIOD_FLAG_PULL_DOWN) |
> + BIT(GPIOD_FLAG_BIAS_DISABLE) |
> + BIT(GPIOD_FLAG_EVENT_CLOCK_REALTIME) |
> + BIT(GPIOD_FLAG_EVENT_CLOCK_HTE);
Same comment as before.
> + unsigned long flags = 0;
--
With Best Regards,
Andy Shevchenko
prev parent reply other threads:[~2026-11-09 21:58 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-09 10:27 [PATCH v2 0/5] gpiolib: improve concurrent access to the flags field of struct gpio_desc Bartosz Golaszewski
2026-11-09 10:27 ` [PATCH v2 1/5] gpiolib: don't overwrite the entire flags field in gpiod_get_direction() Bartosz Golaszewski
2026-11-09 10:27 ` [PATCH v2 2/5] gpio: shared: use atomic set_bit() for descriptor flags Bartosz Golaszewski
2026-11-09 10:27 ` [PATCH v2 3/5] gpiolib: clear debounce period before the "requested" flag Bartosz Golaszewski
2026-11-09 10:27 ` [PATCH v2 4/5] gpiolib: don't overwrite the entire flags field in gpiod_free_commit() Bartosz Golaszewski
2026-11-09 21:53 ` Andy Shevchenko
2026-11-09 10:27 ` [PATCH v2 5/5] gpio: cdev: don't overwrite the whole flags field when configuring lines Bartosz Golaszewski
2026-11-09 21:58 ` Andy Shevchenko [this message]
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=avJCETUZidpmICKR@ashevche-desk.local \
--to=andriy.shevchenko@linux.intel.com \
--cc=aambrozewicz@nvidia.com \
--cc=bartosz.golaszewski@oss.qualcomm.com \
--cc=brgl@kernel.org \
--cc=linusw@kernel.org \
--cc=linux-gpio@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=stable@vger.kernel.org \
--cc=warthog618@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
all inboxes | Powered by JetHome®