diff mbox series

[v3] clk: imx8qxp: Defer instead of failing probe

Message ID DU0PR01MB9382A2BBCCD8C786121975B89D922@DU0PR01MB9382.eurprd01.prod.exchangelabs.com (mailing list archive)
State Superseded
Headers show
Series [v3] clk: imx8qxp: Defer instead of failing probe | expand

Commit Message

Diogo Manuel Pais Silva Sept. 2, 2024, 2:18 p.m. UTC
When of_clk_parent_fill is ran without all the parent clocks having been
probed then the probe function will return -EINVAL, making it so that
the probe isn't attempted again. As fw_devlink is on by default this
does not usually happen, but if fw_devlink is disabled then it is very
possible that the parent clock will be probed after the lpcg first
attempt.

Signed-off-by: Diogo Silva <diogo.pais@ttcontrol.com>
---
v2: change from dev_warn to dev_err_probe
v3: refresh patch
---
 drivers/clk/imx/clk-imx8qxp-lpcg.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

--
2.34.1

Comments

Peng Fan Sept. 4, 2024, 2:03 a.m. UTC | #1
Hi Diogo,

> Subject: [PATCH v3] clk: imx8qxp: Defer instead of failing probe
> 
> When of_clk_parent_fill is ran without all the parent clocks having
> been probed then the probe function will return -EINVAL, making it so
> that the probe isn't attempted again. As fw_devlink is on by default
> this does not usually happen, but if fw_devlink is disabled then it is
> very possible that the parent clock will be probed after the lpcg first
> attempt.

Did you meet issue with fw_devlink disabled?

> 
> Signed-off-by: Diogo Silva <diogo.pais@ttcontrol.com>
> ---
> v2: change from dev_warn to dev_err_probe
> v3: refresh patch
> ---
>  drivers/clk/imx/clk-imx8qxp-lpcg.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/clk/imx/clk-imx8qxp-lpcg.c b/drivers/clk/imx/clk-
> imx8qxp-lpcg.c
> index d0ccaa040225..cae8f6060fe8 100644
> --- a/drivers/clk/imx/clk-imx8qxp-lpcg.c
> +++ b/drivers/clk/imx/clk-imx8qxp-lpcg.c
> @@ -225,8 +225,8 @@ static int imx_lpcg_parse_clks_from_dt(struct
> platform_device *pdev,
> 
>         ret = of_clk_parent_fill(np, parent_names, count);
>         if (ret != count) {
> -               dev_err(&pdev->dev, "failed to get clock parent names\n");
> -               return count;
> +               return dev_err_probe(&pdev->dev, -EPROBE_DEFER,
> +                                    "failed to get all clock parent
> + names\n");
>         }

The change is not enough, you also need to handle
        ret = imx_lpcg_parse_clks_from_dt(pdev, np);                                                
        if (!ret)                                                                                   
                return 0;
->
        ret = imx_lpcg_parse_clks_from_dt(pdev, np);                                                
        if (!ret)                                                                                   
                return 0;
        if (ret == -EPROBE_DEFER)
                return ret;

Regards,
Peng.

> 
>         ret = of_property_read_string_array(np, "clock-output-names",
> --
> 2.34.1
> ________________________________________
> From: Abel Vesa <abel.vesa@linaro.org>
> Sent: Wednesday, August 28, 2024 10:31 AM
> To: Diogo Manuel Pais Silva
> Cc: Peng Fan; abelvesa@kernel.org; linux-clk@vger.kernel.org;
> shawnguo@kernel.org; kernel@pengutronix.de;
> s.hauer@pengutronix.de; linux-arm-kernel@lists.infradead.org;
> sboyd@kernel.org; mturquette@baylibre.com; festevam@gmail.com;
> imx@lists.linux.dev; EMC: linux-kernel@vger.kernel.org
> Subject: Re: [PATCH v2] clk: imx8qxp: Defer instead of failing probe
> 
> On 24-07-02 08:10:44, Diogo Manuel Pais Silva wrote:
> > When of_clk_parent_fill is ran without all the parent clocks having
> been probed then the probe function will return -EINVAL, making it so
> that the probe isn't attempted again. As fw_devlink is on by default
> this does not usually happen, but if fw_devlink is disabled then it is
> very possible that the parent clock will be probed after the lpcg first
> attempt.
> >
> > Signed-off-by: Diogo Silva <diogo.pais@ttcontrol.com>
> 
> This patch doesn't apply cleanly.
> 
> Please respin.
> 
> Thanks!
> 
> > ---
> > v2: change from dev_warn to dev_err_probe
> > ---
> >  drivers/clk/imx/clk-imx8qxp-lpcg.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/clk/imx/clk-imx8qxp-lpcg.c
> > b/drivers/clk/imx/clk-imx8qxp-lpcg.c
> > index d0ccaa040225..7bd9b745edbe 100644
> > --- a/drivers/clk/imx/clk-imx8qxp-lpcg.c
> > +++ b/drivers/clk/imx/clk-imx8qxp-lpcg.c
> > @@ -225,8 +225,8 @@ static int
> imx_lpcg_parse_clks_from_dt(struct
> > platform_device *pdev,
> >
> >       ret = of_clk_parent_fill(np, parent_names, count);
> >       if (ret != count) {
> > -             dev_err(&pdev->dev, "failed to get clock parent names\n");
> > -             return count;
> > +             return dev_err_probe(&pdev->dev, -EPROBE_DEFER,
> > +                             "failed to get all clock parent
> > + names\n");
> >       }
> >
> >       ret = of_property_read_string_array(np, "clock-output-names",
> > --
> > 2.34.1
> CONFIDENTIALITY: The contents of this e-mail are confidential and
> intended only for the above addressee(s). If you are not the intended
> recipient, or the person responsible for delivering it to the intended
> recipient, copying or delivering it to anyone else or using it in any
> unauthorized manner is prohibited and may be unlawful. If you receive
> this e-mail by mistake, please notify the sender and the systems
> administrator at straymail@tttech.com immediately.
Diogo Manuel Pais Silva Sept. 4, 2024, 6:47 a.m. UTC | #2
Hi Peng,

Thanks for pointing that out, I missed it. I will submit a v4.
Yes, the issue only happens when fw_devlink is disabled. I believe this is because devlink generates the device dependency graphs and only probes this device when all the dependencies are met. When the devlink is disabled this dependency is not respect as far as probing order goes, so it can happen that the device is probed when the clock parent is yet to be probed.

Best regards,
Diogo Silva
diff mbox series

Patch

diff --git a/drivers/clk/imx/clk-imx8qxp-lpcg.c b/drivers/clk/imx/clk-imx8qxp-lpcg.c
index d0ccaa040225..cae8f6060fe8 100644
--- a/drivers/clk/imx/clk-imx8qxp-lpcg.c
+++ b/drivers/clk/imx/clk-imx8qxp-lpcg.c
@@ -225,8 +225,8 @@  static int imx_lpcg_parse_clks_from_dt(struct platform_device *pdev,

        ret = of_clk_parent_fill(np, parent_names, count);
        if (ret != count) {
-               dev_err(&pdev->dev, "failed to get clock parent names\n");
-               return count;
+               return dev_err_probe(&pdev->dev, -EPROBE_DEFER,
+                                    "failed to get all clock parent names\n");
        }

        ret = of_property_read_string_array(np, "clock-output-names",