Linux-PHY Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Wesley Cheng" <wesley.cheng@oss.qualcomm.com>
Cc: linux-phy@lists.infradead.org, vkoul@kernel.org,
	neil.armstrong@linaro.org, olteanv@gmail.com
Subject: Re: [PATCH v2] phy: qualcomm: qmp-combo: Differentiate clocks related to the USB block
Date: Tue, 17 Nov 2026 23:15:03 +0000	[thread overview]
Message-ID: <sashiko-outbox-219787@kernel.org> (raw)
In-Reply-To: <20261117-qmp_phy_usb_clks_v1-v2-1-6e960853aa69@oss.qualcomm.com>

Thank you for your contribution! Sashiko AI review found 3 potential
issues to consider:
- [High] phy: qcom-qmp-combo: unbalanced usb_clks across USB3 and USB4
  init/exit
- [Medium] phy: qcom-qmp-combo: leak of usb_clks on
  qmp_combo_usb4_init() error paths
- [Medium] phy: qcom-qmp-combo: missing usb4_init_count check in runtime
  PM callbacks
--

commit b7cf67f2e40df5f2b187b7358099c33b24201c9f
Author: Wesley Cheng <wesley.cheng@oss.qualcomm.com>

phy: qualcomm: qmp-combo: Differentiate clocks related to the USB block

This commit splits the USB-specific clocks ("aux" and "com_aux") into a
separate clock bulk array so they are only enabled when the USB or USB4
paths are active, and gates autonomous mode transitions on USB activity.

> diff --git a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
> index 187f0cd11f6c3..bb9eee2f3f0b8 100644
> --- a/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
> +++ b/drivers/phy/qualcomm/phy-qcom-qmp-combo.c
[ ... ]
> @@ -4455,13 +4461,20 @@ static int qmp_combo_usb_init(struct phy *phy)
>  		return 0;
>  	}
>  
> -	ret = qmp_combo_com_init(qmp, false);
> +	ret = clk_bulk_prepare_enable(qmp->num_usb_clks, qmp->usb_clks);
>  	if (ret)
>  		return ret;
>  
> +	ret = qmp_combo_com_init(qmp, false);
> +	if (ret) {
> +		clk_bulk_disable_unprepare(qmp->num_usb_clks, qmp->usb_clks);
> +		return ret;
> +	}
> +
>  	ret = qmp_combo_usb_power_on(phy);
>  	if (ret) {
>  		qmp_combo_com_exit(qmp, false);
> +		clk_bulk_disable_unprepare(qmp->num_usb_clks, qmp->usb_clks);
>  		return ret;
>  	}
>  
> @@ -4492,6 +4505,8 @@ static int qmp_combo_usb_exit(struct phy *phy)
>  	if (ret)
>  		return ret;
>  
> +	clk_bulk_disable_unprepare(qmp->num_usb_clks, qmp->usb_clks);
> +
>  	qmp->usb_init_count--;
>  
>  	return 0;
> @@ -4547,6 +4562,10 @@ static int qmp_combo_usb4_init(struct phy *phy)
>  		return -EINVAL;
>  	}
>  
> +	ret = clk_bulk_prepare_enable(qmp->num_usb_clks, qmp->usb_clks);
> +	if (ret)
> +		return ret;
> +
>  	ret = clk_prepare_enable(qmp->p2rr2p_pipe_clk);
>  	if (ret) {
>  		dev_err(qmp->dev, "p2rr2p_pipe enable failed: %d\n", ret);

[Severity: Medium]
Does qmp_combo_usb4_init() leak qmp->usb_clks on failure?  If
clk_prepare_enable(qmp->p2rr2p_pipe_clk), qmp_combo_com_init(),
qmp_combo_reconfigure_phy(), or qmp_combo_usb_power_on() fails, none of
their error paths call clk_bulk_disable_unprepare(qmp->num_usb_clks,
qmp->usb_clks), leaving the USB clocks permanently prepared and enabled.

> @@ -4606,6 +4625,7 @@ static int qmp_combo_usb4_exit(struct phy *phy)
>  	qmp->usb4_init_count--;
>  
>  	clk_disable_unprepare(qmp->p2rr2p_pipe_clk);
> +	clk_bulk_disable_unprepare(qmp->num_usb_clks, qmp->usb_clks);
>  
>  	return 0;
>  }

[Severity: High]
As reported by Konrad Dybcio on v1 [1], can qmp->usb_clks become
unbalanced when USB3 and USB4 init/exit overlap?

In qmp_combo_usb_init() and qmp_combo_usb_exit(), when qmp->usb4_init_count
is non-zero, both functions update qmp->usb_init_count and return early
before touching qmp->usb_clks:

drivers/phy/qualcomm/phy-qcom-qmp-combo.c:qmp_combo_usb_init() {
	/* USB4 mode takes precedence, do not reprogram the PHY in that case */
	if (qmp->usb4_init_count) {
		qmp->init_count++;
		qmp->usb_init_count++;
		return 0;
	}

	ret = clk_bulk_prepare_enable(qmp->num_usb_clks, qmp->usb_clks);

If qmp_combo_usb4_init() runs first (enabling qmp->usb_clks once),
qmp_combo_usb_init() runs next (skipping clk_bulk_prepare_enable()), and
qmp_combo_usb4_exit() then runs, qmp_combo_usb4_exit() disables
qmp->usb_clks down to zero while qmp->usb_init_count is still 1.  Any
subsequent USB3 register access in qmp_combo_typec_mux_set(),
qmp_combo_runtime_suspend(), or qmp_combo_usb_exit() happens with
qmp->usb_clks disabled, and qmp_combo_usb_exit() then calls
clk_bulk_disable_unprepare() a second time on already-disabled clocks.

Conversely, if qmp_combo_usb_init() runs before qmp_combo_usb4_init() (so
both enable qmp->usb_clks) and qmp_combo_usb_exit() is called while
qmp->usb4_init_count is still non-zero, qmp_combo_usb_exit() returns early
without disabling qmp->usb_clks, leaking a clock reference.

> @@ -4722,7 +4742,10 @@ static int __maybe_unused qmp_combo_runtime_suspend(struct device *dev)
>  		return 0;
>  	}
>  
> -	qmp_combo_enable_autonomous_mode(qmp);
> +	if (qmp->usb_init_count) {
> +		qmp_combo_enable_autonomous_mode(qmp);
> +		clk_bulk_disable_unprepare(qmp->num_usb_clks, qmp->usb_clks);
> +	}
>  
>  	clk_disable_unprepare(qmp->pipe_clk);
>  	clk_bulk_disable_unprepare(qmp->num_clks, qmp->clks);
> @@ -4749,13 +4772,27 @@ static int __maybe_unused qmp_combo_runtime_resume(struct device *dev)
>  	ret = clk_prepare_enable(qmp->pipe_clk);
>  	if (ret) {
>  		dev_err(dev, "pipe_clk enable failed, err=%d\n", ret);
> -		clk_bulk_disable_unprepare(qmp->num_clks, qmp->clks);
> -		return ret;
> +		goto disable_clks;
>  	}
>  
> -	qmp_combo_disable_autonomous_mode(qmp);
> +	if (qmp->usb_init_count) {
> +		ret = clk_bulk_prepare_enable(qmp->num_usb_clks, qmp->usb_clks);
> +		if (ret) {
> +			dev_err(dev, "usb_clks enable failed, err=%d\n", ret);
> +			goto disable_pipe_clk;
> +		}
> +
> +		qmp_combo_disable_autonomous_mode(qmp);
> +	}
>  
>  	return 0;
> +
> +disable_pipe_clk:
> +	clk_disable_unprepare(qmp->pipe_clk);
> +disable_clks:
> +	clk_bulk_disable_unprepare(qmp->num_clks, qmp->clks);
> +
> +	return ret;
>  }

[Severity: Medium]
Should qmp_combo_runtime_suspend() and qmp_combo_runtime_resume() also
account for qmp->usb4_init_count?

qmp_combo_usb4_init() enables qmp->usb_clks and increments
qmp->usb4_init_count without incrementing qmp->usb_init_count, and
qmp_combo_enable_autonomous_mode() / qmp_combo_disable_autonomous_mode()
explicitly handle QMPPHY_MODE_USB4 via QPHY_PCS_USB4_CLAMP_ENABLE.
Checking only qmp->usb_init_count skips autonomous mode and leaves
qmp->usb_clks enabled during runtime suspend when only USB4 is initialized.

[1] https://lore.kernel.org/r/c1abd9d4-98ef-4469-b4f3-94a8dcad9d0b@oss.qualcomm.com

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117-qmp_phy_usb_clks_v1-v2-1-6e960853aa69@oss.qualcomm.com?part=1

-- 
linux-phy mailing list
linux-phy@lists.infradead.org
https://lists.infradead.org/mailman/listinfo/linux-phy

      reply	other threads:[~2026-11-17 23:15 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 23:02 [PATCH v2] phy: qualcomm: qmp-combo: Differentiate clocks related to the USB block Wesley Cheng
2026-11-17 23:15 ` sashiko-bot [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=sashiko-outbox-219787@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=linux-phy@lists.infradead.org \
    --cc=neil.armstrong@linaro.org \
    --cc=olteanv@gmail.com \
    --cc=sashiko-reviews@lists.linux.dev \
    --cc=vkoul@kernel.org \
    --cc=wesley.cheng@oss.qualcomm.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