Message ID | 20240102081402.1226795-1-treapking@chromium.org (mailing list archive) |
---|---|
State | Changes Requested, archived |
Headers | show |
Series | [v2,1/2] clk: mediatek: Introduce need_pm_runtime to mtk_clk_desc | expand |
On Tue, Jan 2, 2024 at 4:14 PM Pin-yen Lin <treapking@chromium.org> wrote: > > Introduce a new need_pm_runtime variable to mtk_clk_desc to indicate > this clock needs a runtime PM get on the clock controller during the > probing stage. No. The flag indicates that the clock controller needs runtime PM for its operation, likely because it needs some power domain enabled. The runtime PM get during the probe phase is a workaround to prevent a deadlock in clk_register. These are two separate things. The second part also should be documented in the code with a comment, i.e. a comment should be placed before pm_runtime_resume_and_get(). ChenYu > Signed-off-by: Pin-yen Lin <treapking@chromium.org> > --- > > Changes in v2: > - Fix the order of error handling > - Update the commit message and add a comment before the runtime PM call > > drivers/clk/mediatek/clk-mtk.c | 15 +++++++++++++++ > drivers/clk/mediatek/clk-mtk.h | 2 ++ > 2 files changed, 17 insertions(+) > > diff --git a/drivers/clk/mediatek/clk-mtk.c b/drivers/clk/mediatek/clk-mtk.c > index 2e55368dc4d8..c31e535909c8 100644 > --- a/drivers/clk/mediatek/clk-mtk.c > +++ b/drivers/clk/mediatek/clk-mtk.c > @@ -13,6 +13,7 @@ > #include <linux/of.h> > #include <linux/of_address.h> > #include <linux/platform_device.h> > +#include <linux/pm_runtime.h> > #include <linux/slab.h> > > #include "clk-mtk.h" > @@ -494,6 +495,14 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > return IS_ERR(base) ? PTR_ERR(base) : -ENOMEM; > } > > + > + if (mcd->need_runtime_pm) { > + devm_pm_runtime_enable(&pdev->dev); > + r = pm_runtime_resume_and_get(&pdev->dev); > + if (r) > + return r; > + } > + > /* Calculate how many clk_hw_onecell_data entries to allocate */ > num_clks = mcd->num_clks + mcd->num_composite_clks; > num_clks += mcd->num_fixed_clks + mcd->num_factor_clks; > @@ -574,6 +583,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > goto unregister_clks; > } > > + if (mcd->need_runtime_pm) > + pm_runtime_put(&pdev->dev); > + > return r; > > unregister_clks: > @@ -604,6 +616,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > free_base: > if (mcd->shared_io && base) > iounmap(base); > + > + if (mcd->need_runtime_pm) > + pm_runtime_put(&pdev->dev); > return r; > } > > diff --git a/drivers/clk/mediatek/clk-mtk.h b/drivers/clk/mediatek/clk-mtk.h > index 22096501a60a..c17fe1c2d732 100644 > --- a/drivers/clk/mediatek/clk-mtk.h > +++ b/drivers/clk/mediatek/clk-mtk.h > @@ -237,6 +237,8 @@ struct mtk_clk_desc { > > int (*clk_notifier_func)(struct device *dev, struct clk *clk); > unsigned int mfg_clk_idx; > + > + bool need_runtime_pm; > }; > > int mtk_clk_pdev_probe(struct platform_device *pdev); > -- > 2.43.0.472.g3155946c3a-goog >
Il 02/01/24 09:12, Pin-yen Lin ha scritto: > Introduce a new need_pm_runtime variable to mtk_clk_desc to indicate > this clock needs a runtime PM get on the clock controller during the > probing stage. > > Signed-off-by: Pin-yen Lin <treapking@chromium.org> Hello Pin-yen, We have experienced something similar, but it was really hard to reproduce after some changes. In an effort to try to solve this issue (but again, reproducing is really hard), Eugen has sent a commit in the hope that someone else found a way to easily reproduce. Please look at [1]. I'm also adding Eugen to the Cc's of this email. Cheers, Angelo [1]: https://patchwork.kernel.org/project/linux-pm/patch/20231225133615.78993-1-eugen.hristev@collabora.com/ > --- > > Changes in v2: > - Fix the order of error handling > - Update the commit message and add a comment before the runtime PM call > > drivers/clk/mediatek/clk-mtk.c | 15 +++++++++++++++ > drivers/clk/mediatek/clk-mtk.h | 2 ++ > 2 files changed, 17 insertions(+) > > diff --git a/drivers/clk/mediatek/clk-mtk.c b/drivers/clk/mediatek/clk-mtk.c > index 2e55368dc4d8..c31e535909c8 100644 > --- a/drivers/clk/mediatek/clk-mtk.c > +++ b/drivers/clk/mediatek/clk-mtk.c > @@ -13,6 +13,7 @@ > #include <linux/of.h> > #include <linux/of_address.h> > #include <linux/platform_device.h> > +#include <linux/pm_runtime.h> > #include <linux/slab.h> > > #include "clk-mtk.h" > @@ -494,6 +495,14 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > return IS_ERR(base) ? PTR_ERR(base) : -ENOMEM; > } > > + > + if (mcd->need_runtime_pm) { > + devm_pm_runtime_enable(&pdev->dev); > + r = pm_runtime_resume_and_get(&pdev->dev); > + if (r) > + return r; > + } > + > /* Calculate how many clk_hw_onecell_data entries to allocate */ > num_clks = mcd->num_clks + mcd->num_composite_clks; > num_clks += mcd->num_fixed_clks + mcd->num_factor_clks; > @@ -574,6 +583,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > goto unregister_clks; > } > > + if (mcd->need_runtime_pm) > + pm_runtime_put(&pdev->dev); > + > return r; > > unregister_clks: > @@ -604,6 +616,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > free_base: > if (mcd->shared_io && base) > iounmap(base); > + > + if (mcd->need_runtime_pm) > + pm_runtime_put(&pdev->dev); > return r; > } > > diff --git a/drivers/clk/mediatek/clk-mtk.h b/drivers/clk/mediatek/clk-mtk.h > index 22096501a60a..c17fe1c2d732 100644 > --- a/drivers/clk/mediatek/clk-mtk.h > +++ b/drivers/clk/mediatek/clk-mtk.h > @@ -237,6 +237,8 @@ struct mtk_clk_desc { > > int (*clk_notifier_func)(struct device *dev, struct clk *clk); > unsigned int mfg_clk_idx; > + > + bool need_runtime_pm; > }; > > int mtk_clk_pdev_probe(struct platform_device *pdev);
On 1/3/24 14:19, AngeloGioacchino Del Regno wrote: > Il 02/01/24 09:12, Pin-yen Lin ha scritto: >> Introduce a new need_pm_runtime variable to mtk_clk_desc to indicate >> this clock needs a runtime PM get on the clock controller during the >> probing stage. >> >> Signed-off-by: Pin-yen Lin <treapking@chromium.org> > > Hello Pin-yen, > > We have experienced something similar, but it was really hard to reproduce after > some changes. > > In an effort to try to solve this issue (but again, reproducing is really hard), > Eugen has sent a commit in the hope that someone else found a way to easily > reproduce. Please look at [1]. > > I'm also adding Eugen to the Cc's of this email. > > Cheers, > Angelo > > [1]: > https://patchwork.kernel.org/project/linux-pm/patch/20231225133615.78993-1-eugen.hristev@collabora.com/ Hello Pin-yen, Can you try my patch and let me know if this changes anything for you ? If it does not change anything, can you also try this one as well ? It's another attempt to fix the synchronization with genpd. https://lore.kernel.org/linux-arm-kernel/20231129113120.4907-1-eugen.hristev@collabora.com/ Thanks, Eugen > >> --- >> >> Changes in v2: >> - Fix the order of error handling >> - Update the commit message and add a comment before the runtime PM call >> >> drivers/clk/mediatek/clk-mtk.c | 15 +++++++++++++++ >> drivers/clk/mediatek/clk-mtk.h | 2 ++ >> 2 files changed, 17 insertions(+) >> >> diff --git a/drivers/clk/mediatek/clk-mtk.c b/drivers/clk/mediatek/clk-mtk.c >> index 2e55368dc4d8..c31e535909c8 100644 >> --- a/drivers/clk/mediatek/clk-mtk.c >> +++ b/drivers/clk/mediatek/clk-mtk.c >> @@ -13,6 +13,7 @@ >> #include <linux/of.h> >> #include <linux/of_address.h> >> #include <linux/platform_device.h> >> +#include <linux/pm_runtime.h> >> #include <linux/slab.h> >> >> #include "clk-mtk.h" >> @@ -494,6 +495,14 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, >> return IS_ERR(base) ? PTR_ERR(base) : -ENOMEM; >> } >> >> + >> + if (mcd->need_runtime_pm) { >> + devm_pm_runtime_enable(&pdev->dev); >> + r = pm_runtime_resume_and_get(&pdev->dev); >> + if (r) >> + return r; >> + } >> + >> /* Calculate how many clk_hw_onecell_data entries to allocate */ >> num_clks = mcd->num_clks + mcd->num_composite_clks; >> num_clks += mcd->num_fixed_clks + mcd->num_factor_clks; >> @@ -574,6 +583,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, >> goto unregister_clks; >> } >> >> + if (mcd->need_runtime_pm) >> + pm_runtime_put(&pdev->dev); >> + >> return r; >> >> unregister_clks: >> @@ -604,6 +616,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, >> free_base: >> if (mcd->shared_io && base) >> iounmap(base); >> + >> + if (mcd->need_runtime_pm) >> + pm_runtime_put(&pdev->dev); >> return r; >> } >> >> diff --git a/drivers/clk/mediatek/clk-mtk.h b/drivers/clk/mediatek/clk-mtk.h >> index 22096501a60a..c17fe1c2d732 100644 >> --- a/drivers/clk/mediatek/clk-mtk.h >> +++ b/drivers/clk/mediatek/clk-mtk.h >> @@ -237,6 +237,8 @@ struct mtk_clk_desc { >> >> int (*clk_notifier_func)(struct device *dev, struct clk *clk); >> unsigned int mfg_clk_idx; >> + >> + bool need_runtime_pm; >> }; >> >> int mtk_clk_pdev_probe(struct platform_device *pdev); > > > >
Hi Eugen and Angelo, Unfortunately, I don't have a way to reliably reproduce this either. We notice this issue from the automatic crash reports sent from the users, but we are still not able to reproduce this locally. So our plan is to ship this patch to the users and see if the crash rate goes down after a month or so. Regards, Pin-yen On Wed, Jan 3, 2024 at 9:20 PM Eugen Hristev <eugen.hristev@collabora.com> wrote: > > On 1/3/24 14:19, AngeloGioacchino Del Regno wrote: > > Il 02/01/24 09:12, Pin-yen Lin ha scritto: > >> Introduce a new need_pm_runtime variable to mtk_clk_desc to indicate > >> this clock needs a runtime PM get on the clock controller during the > >> probing stage. > >> > >> Signed-off-by: Pin-yen Lin <treapking@chromium.org> > > > > Hello Pin-yen, > > > > We have experienced something similar, but it was really hard to reproduce after > > some changes. > > > > In an effort to try to solve this issue (but again, reproducing is really hard), > > Eugen has sent a commit in the hope that someone else found a way to easily > > reproduce. Please look at [1]. > > > > I'm also adding Eugen to the Cc's of this email. > > > > Cheers, > > Angelo > > > > [1]: > > https://patchwork.kernel.org/project/linux-pm/patch/20231225133615.78993-1-eugen.hristev@collabora.com/ > > Hello Pin-yen, > > Can you try my patch and let me know if this changes anything for you ? > > If it does not change anything, can you also try this one as well ? It's another > attempt to fix the synchronization with genpd. > > https://lore.kernel.org/linux-arm-kernel/20231129113120.4907-1-eugen.hristev@collabora.com/ > > Thanks, > Eugen > > > > >> --- > >> > >> Changes in v2: > >> - Fix the order of error handling > >> - Update the commit message and add a comment before the runtime PM call > >> > >> drivers/clk/mediatek/clk-mtk.c | 15 +++++++++++++++ > >> drivers/clk/mediatek/clk-mtk.h | 2 ++ > >> 2 files changed, 17 insertions(+) > >> > >> diff --git a/drivers/clk/mediatek/clk-mtk.c b/drivers/clk/mediatek/clk-mtk.c > >> index 2e55368dc4d8..c31e535909c8 100644 > >> --- a/drivers/clk/mediatek/clk-mtk.c > >> +++ b/drivers/clk/mediatek/clk-mtk.c > >> @@ -13,6 +13,7 @@ > >> #include <linux/of.h> > >> #include <linux/of_address.h> > >> #include <linux/platform_device.h> > >> +#include <linux/pm_runtime.h> > >> #include <linux/slab.h> > >> > >> #include "clk-mtk.h" > >> @@ -494,6 +495,14 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > >> return IS_ERR(base) ? PTR_ERR(base) : -ENOMEM; > >> } > >> > >> + > >> + if (mcd->need_runtime_pm) { > >> + devm_pm_runtime_enable(&pdev->dev); > >> + r = pm_runtime_resume_and_get(&pdev->dev); > >> + if (r) > >> + return r; > >> + } > >> + > >> /* Calculate how many clk_hw_onecell_data entries to allocate */ > >> num_clks = mcd->num_clks + mcd->num_composite_clks; > >> num_clks += mcd->num_fixed_clks + mcd->num_factor_clks; > >> @@ -574,6 +583,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > >> goto unregister_clks; > >> } > >> > >> + if (mcd->need_runtime_pm) > >> + pm_runtime_put(&pdev->dev); > >> + > >> return r; > >> > >> unregister_clks: > >> @@ -604,6 +616,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > >> free_base: > >> if (mcd->shared_io && base) > >> iounmap(base); > >> + > >> + if (mcd->need_runtime_pm) > >> + pm_runtime_put(&pdev->dev); > >> return r; > >> } > >> > >> diff --git a/drivers/clk/mediatek/clk-mtk.h b/drivers/clk/mediatek/clk-mtk.h > >> index 22096501a60a..c17fe1c2d732 100644 > >> --- a/drivers/clk/mediatek/clk-mtk.h > >> +++ b/drivers/clk/mediatek/clk-mtk.h > >> @@ -237,6 +237,8 @@ struct mtk_clk_desc { > >> > >> int (*clk_notifier_func)(struct device *dev, struct clk *clk); > >> unsigned int mfg_clk_idx; > >> + > >> + bool need_runtime_pm; > >> }; > >> > >> int mtk_clk_pdev_probe(struct platform_device *pdev); > > > > > > > > >
On Thu, Jan 4, 2024 at 10:51 PM Pin-yen Lin <treapking@chromium.org> wrote: > > Hi Eugen and Angelo, > > Unfortunately, I don't have a way to reliably reproduce this either. > > We notice this issue from the automatic crash reports sent from the > users, but we are still not able to reproduce this locally. So our > plan is to ship this patch to the users and see if the crash rate goes > down after a month or so. > > Regards, > Pin-yen > > On Wed, Jan 3, 2024 at 9:20 PM Eugen Hristev > <eugen.hristev@collabora.com> wrote: > > > > On 1/3/24 14:19, AngeloGioacchino Del Regno wrote: > > > Il 02/01/24 09:12, Pin-yen Lin ha scritto: > > >> Introduce a new need_pm_runtime variable to mtk_clk_desc to indicate > > >> this clock needs a runtime PM get on the clock controller during the > > >> probing stage. > > >> > > >> Signed-off-by: Pin-yen Lin <treapking@chromium.org> > > > > > > Hello Pin-yen, > > > > > > We have experienced something similar, but it was really hard to reproduce after > > > some changes. > > > > > > In an effort to try to solve this issue (but again, reproducing is really hard), > > > Eugen has sent a commit in the hope that someone else found a way to easily > > > reproduce. Please look at [1]. > > > > > > I'm also adding Eugen to the Cc's of this email. > > > > > > Cheers, > > > Angelo > > > > > > [1]: > > > https://patchwork.kernel.org/project/linux-pm/patch/20231225133615.78993-1-eugen.hristev@collabora.com/ > > > > Hello Pin-yen, > > > > Can you try my patch and let me know if this changes anything for you ? > > > > If it does not change anything, can you also try this one as well ? It's another > > attempt to fix the synchronization with genpd. > > > > https://lore.kernel.org/linux-arm-kernel/20231129113120.4907-1-eugen.hristev@collabora.com/ > > > > Thanks, > > Eugen Hi Eugen and Angelo, After the offline discussion with Chen-yu, we think this series is solving a different issue from the patches you mentioned. This one is trying to resolve a deadlock in the probe stage, and more details can be found in the commit message of the next patch. The referenced patches seem to be fixing other race conditions on powering on/off the power domain. Sorry for adding the wrong commit message and maybe leading to incorrect understanding on this series. By the way, sorry for the top posting in the previous mail. Regards, Pin-yen > > > > > > > >> --- > > >> > > >> Changes in v2: > > >> - Fix the order of error handling > > >> - Update the commit message and add a comment before the runtime PM call > > >> > > >> drivers/clk/mediatek/clk-mtk.c | 15 +++++++++++++++ > > >> drivers/clk/mediatek/clk-mtk.h | 2 ++ > > >> 2 files changed, 17 insertions(+) > > >> > > >> diff --git a/drivers/clk/mediatek/clk-mtk.c b/drivers/clk/mediatek/clk-mtk.c > > >> index 2e55368dc4d8..c31e535909c8 100644 > > >> --- a/drivers/clk/mediatek/clk-mtk.c > > >> +++ b/drivers/clk/mediatek/clk-mtk.c > > >> @@ -13,6 +13,7 @@ > > >> #include <linux/of.h> > > >> #include <linux/of_address.h> > > >> #include <linux/platform_device.h> > > >> +#include <linux/pm_runtime.h> > > >> #include <linux/slab.h> > > >> > > >> #include "clk-mtk.h" > > >> @@ -494,6 +495,14 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > > >> return IS_ERR(base) ? PTR_ERR(base) : -ENOMEM; > > >> } > > >> > > >> + > > >> + if (mcd->need_runtime_pm) { > > >> + devm_pm_runtime_enable(&pdev->dev); > > >> + r = pm_runtime_resume_and_get(&pdev->dev); > > >> + if (r) > > >> + return r; > > >> + } > > >> + > > >> /* Calculate how many clk_hw_onecell_data entries to allocate */ > > >> num_clks = mcd->num_clks + mcd->num_composite_clks; > > >> num_clks += mcd->num_fixed_clks + mcd->num_factor_clks; > > >> @@ -574,6 +583,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > > >> goto unregister_clks; > > >> } > > >> > > >> + if (mcd->need_runtime_pm) > > >> + pm_runtime_put(&pdev->dev); > > >> + > > >> return r; > > >> > > >> unregister_clks: > > >> @@ -604,6 +616,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, > > >> free_base: > > >> if (mcd->shared_io && base) > > >> iounmap(base); > > >> + > > >> + if (mcd->need_runtime_pm) > > >> + pm_runtime_put(&pdev->dev); > > >> return r; > > >> } > > >> > > >> diff --git a/drivers/clk/mediatek/clk-mtk.h b/drivers/clk/mediatek/clk-mtk.h > > >> index 22096501a60a..c17fe1c2d732 100644 > > >> --- a/drivers/clk/mediatek/clk-mtk.h > > >> +++ b/drivers/clk/mediatek/clk-mtk.h > > >> @@ -237,6 +237,8 @@ struct mtk_clk_desc { > > >> > > >> int (*clk_notifier_func)(struct device *dev, struct clk *clk); > > >> unsigned int mfg_clk_idx; > > >> + > > >> + bool need_runtime_pm; > > >> }; > > >> > > >> int mtk_clk_pdev_probe(struct platform_device *pdev); > > > > > > > > > > > > > >
diff --git a/drivers/clk/mediatek/clk-mtk.c b/drivers/clk/mediatek/clk-mtk.c index 2e55368dc4d8..c31e535909c8 100644 --- a/drivers/clk/mediatek/clk-mtk.c +++ b/drivers/clk/mediatek/clk-mtk.c @@ -13,6 +13,7 @@ #include <linux/of.h> #include <linux/of_address.h> #include <linux/platform_device.h> +#include <linux/pm_runtime.h> #include <linux/slab.h> #include "clk-mtk.h" @@ -494,6 +495,14 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, return IS_ERR(base) ? PTR_ERR(base) : -ENOMEM; } + + if (mcd->need_runtime_pm) { + devm_pm_runtime_enable(&pdev->dev); + r = pm_runtime_resume_and_get(&pdev->dev); + if (r) + return r; + } + /* Calculate how many clk_hw_onecell_data entries to allocate */ num_clks = mcd->num_clks + mcd->num_composite_clks; num_clks += mcd->num_fixed_clks + mcd->num_factor_clks; @@ -574,6 +583,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, goto unregister_clks; } + if (mcd->need_runtime_pm) + pm_runtime_put(&pdev->dev); + return r; unregister_clks: @@ -604,6 +616,9 @@ static int __mtk_clk_simple_probe(struct platform_device *pdev, free_base: if (mcd->shared_io && base) iounmap(base); + + if (mcd->need_runtime_pm) + pm_runtime_put(&pdev->dev); return r; } diff --git a/drivers/clk/mediatek/clk-mtk.h b/drivers/clk/mediatek/clk-mtk.h index 22096501a60a..c17fe1c2d732 100644 --- a/drivers/clk/mediatek/clk-mtk.h +++ b/drivers/clk/mediatek/clk-mtk.h @@ -237,6 +237,8 @@ struct mtk_clk_desc { int (*clk_notifier_func)(struct device *dev, struct clk *clk); unsigned int mfg_clk_idx; + + bool need_runtime_pm; }; int mtk_clk_pdev_probe(struct platform_device *pdev);
Introduce a new need_pm_runtime variable to mtk_clk_desc to indicate this clock needs a runtime PM get on the clock controller during the probing stage. Signed-off-by: Pin-yen Lin <treapking@chromium.org> --- Changes in v2: - Fix the order of error handling - Update the commit message and add a comment before the runtime PM call drivers/clk/mediatek/clk-mtk.c | 15 +++++++++++++++ drivers/clk/mediatek/clk-mtk.h | 2 ++ 2 files changed, 17 insertions(+)