From: Sebastian Reichel <sebastian.reichel@collabora.com>
To: Myeonghun Pak <mhun512@gmail.com>
Cc: linux-pm@vger.kernel.org, linux-kernel@vger.kernel.org,
Ijae Kim <ae878000@gmail.com>
Subject: Re: [PATCH v2] power: supply: act8945a: Disable status work on removal
Date: Sun, 18 Oct 2026 12:19:41 +0200 [thread overview]
Message-ID: <atSacKtZ3e4YtVBy@venus> (raw)
In-Reply-To: <20261018020357.3845586-1-mhun512@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 4814 bytes --]
Hi,
On Sat, Oct 17, 2026 at 10:03:57PM -0400, Myeonghun Pak wrote:
> Clearing init_done does not synchronize with either status interrupt. A
> handler can observe init_done set, be delayed, and schedule the status
> work after cancel_work_sync() has returned. The work can then access the
> power supply after its managed cleanup.
>
> Use disable_work_sync() to reject subsequent queueing and drain any
> existing execution. The charger allocation remains valid until both
> managed IRQs have been released, so a late handler can safely attempt to
> queue the disabled work.
>
> Replace init_done with the work item's disable state. Initialize and
> disable the work before act8945a_charger_config(), which registers the
> LBO interrupt, and enable it only after probe initialization succeeds.
> This keeps both status IRQs from queueing work before the power supply
> is ready, including on probe failure.
>
> This issue was identified as part of our ongoing research while
> reviewing kernel code.
>
> Fixes: a09209acd6a8 ("power: supply: act8945a_charger: Add status change update support")
> Assisted-by: LLM
> Co-developed-by: Ijae Kim <ae878000@gmail.com>
> Signed-off-by: Ijae Kim <ae878000@gmail.com>
> Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
> ---
Having a closer look at how the work is being used: Instead of doing
all this complex stuff this should just use a threaded interrupt and
completley get rid of the worker. I.e.:
1. get rid of init_done (done in your version)
2. switch from devm_request_irq to devm_request_threaded_irq
3. move the act8945a_lbo_detect IRQ request after power-supply
registration to avoid problems with spurious interrupts
4. move act8945a_set_supply_type + power_supply_changed calls from
act8945a_work into act8945a_status_changed
5. completley remove charger->work
6. drop empty act8945a_charger_remove()
It results in less and simpler code.
Greetings,
-- Sebastian
> Changes in v2:
> - Remove init_done and all accesses to it, as requested by Sebastian
> Reichel.
> - Initialize the work before either IRQ is requested and keep it disabled
> until probe initialization succeeds, preserving the probe-time gate.
>
> Compile-tested with GCC 15, W=1 (x86_64), with OF, OF_IRQ and
> GPIOLIB enabled. No runtime testing.
>
> v1: https://lore.kernel.org/r/179211805459.3650807.8742309660995961596.G171@gmail.com/
> Review: https://lore.kernel.org/r/atPtc_jKLo8RhAMU@venus/
>
> drivers/power/supply/act8945a_charger.c | 16 ++++++++--------
> 1 file changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/power/supply/act8945a_charger.c b/drivers/power/supply/act8945a_charger.c
> index 7e4b087..7d47052 100644
> --- a/drivers/power/supply/act8945a_charger.c
> +++ b/drivers/power/supply/act8945a_charger.c
> @@ -14,6 +14,7 @@
> #include <linux/power_supply.h>
> #include <linux/regmap.h>
> #include <linux/gpio/consumer.h>
> +#include <linux/workqueue.h>
>
> static const char *act8945a_charger_model = "ACT8945A";
> static const char *act8945a_charger_manufacturer = "Active-semi";
> @@ -78,7 +79,6 @@ struct act8945a_charger {
> struct regmap *regmap;
> struct work_struct work;
>
> - bool init_done;
> struct gpio_desc *lbo_gpio;
> struct gpio_desc *chglev_gpio;
> };
> @@ -444,8 +444,7 @@ static irqreturn_t act8945a_status_changed(int irq, void *dev_id)
> {
> struct act8945a_charger *charger = dev_id;
>
> - if (charger->init_done)
> - schedule_work(&charger->work);
> + schedule_work(&charger->work);
>
> return IRQ_HANDLED;
> }
> @@ -587,6 +586,10 @@ static int act8945a_charger_probe(struct platform_device *pdev)
> return -EINVAL;
> }
>
> + /* Keep status IRQs from queueing work until probe has completed. */
> + INIT_WORK(&charger->work, act8945a_work);
> + disable_work(&charger->work);
> +
> ret = act8945a_charger_config(&pdev->dev, charger);
> if (ret)
> return ret;
> @@ -625,13 +628,11 @@ static int act8945a_charger_probe(struct platform_device *pdev)
>
> platform_set_drvdata(pdev, charger);
>
> - INIT_WORK(&charger->work, act8945a_work);
> -
> ret = act8945a_enable_interrupt(charger);
> if (ret)
> return -EIO;
>
> - charger->init_done = true;
> + enable_work(&charger->work);
>
> return 0;
> }
> @@ -640,8 +641,7 @@ static void act8945a_charger_remove(struct platform_device *pdev)
> {
> struct act8945a_charger *charger = platform_get_drvdata(pdev);
>
> - charger->init_done = false;
> - cancel_work_sync(&charger->work);
> + disable_work_sync(&charger->work);
> }
>
> static struct platform_driver act8945a_charger_driver = {
>
> base-commit: 238650ef6c7c7cca08e032527329424c9fbd70e5
> --
> 2.53.0
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
prev parent reply other threads:[~2026-10-18 10:19 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-18 2:03 [PATCH v2] power: supply: act8945a: Disable status work on removal Myeonghun Pak
2026-10-18 10:19 ` Sebastian Reichel [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=atSacKtZ3e4YtVBy@venus \
--to=sebastian.reichel@collabora.com \
--cc=ae878000@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pm@vger.kernel.org \
--cc=mhun512@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