[PATCH 00/25] Fix some PM runtime issues at the media subsystem

[PATCH 00/25] Fix some PM runtime issues at the media subsystem

[Mauro Carvalho Chehab]

As part of an effort to cleanup pm_runtime*get* calls, I detected a number
of issues at the media subsystem.

Most of the patches here were submitted previously at:

	https://lore.kernel.org/linux-media/cover.1619621413.git.mchehab+huawei@kernel.org/

This series contain just the bug fixes and other related issues that are
present with the current code on media.

It address the points from the existing reviews. I also did my own
set of reviews, in order to avoid regressions.

Changes from v4 of the previous changeset:

- reworked i2c/css RPM get logic;
- dropped two patches that could cause regressions;
- am437x: keep using pm_runtime_get_sync on suspend/resume;
- atmel: fix the returned code and add a print on failures at start streaming;
- simplify some checks for return code > 0;
- mdk-vcodec: properly handle RPM errors at device on logic;
- venus: rework venus_sys_error_handler() logic;
- sti/delta: fix an issue at the error checking logic.

Mauro Carvalho Chehab (25):
  staging: media: rkvdec: fix pm_runtime_get_sync() usage count
  staging: media: imx7-mipi-csis: fix pm_runtime_get_sync() usage count
  media: venus: Rework error fail recover logic
  media: s5p_cec: decrement usage count if disabled
  media: i2c: ccs-core: return the right error code at suspend
  media: i2c: imx334: fix the pm runtime get logic
  media: exynos-gsc: don't resume at remove time
  media: atmel: properly get pm_runtime
  media: hantro: do a PM resume earlier
  media: marvel-ccic: fix some issues when getting pm_runtime
  media: mdk-mdp: fix pm_runtime_get_sync() usage count
  media: rcar_fdp1: simplify error check logic at fdp_open()
  media: rcar_fdp1: fix pm_runtime_get_sync() usage count
  media: renesas-ceu: Properly check for PM errors
  media: s5p: fix pm_runtime_get_sync() usage count
  media: am437x: fix pm_runtime_get_sync() usage count
  media: sh_vou: fix pm_runtime_get_sync() usage count
  media: mtk-vcodec: fix PM runtime get logic
  media: s5p-jpeg: fix pm_runtime_get_sync() usage count
  media: sti/delta: use pm_runtime_resume_and_get()
  media: sunxi: fix pm_runtime_get_sync() usage count
  media: sti/bdisp: fix pm_runtime_get_sync() usage count
  media: exynos4-is: fix pm_runtime_get_sync() usage count
  media: exynos-gsc: fix pm_runtime_get_sync() usage count
  media: i2c: ccs-core: fix pm_runtime_get_sync() usage count

 drivers/media/cec/platform/s5p/s5p_cec.c      |  7 ++-
 drivers/media/i2c/ccs/ccs-core.c              | 41 ++++++++-----
 drivers/media/i2c/imx334.c                    |  7 ++-
 drivers/media/platform/am437x/am437x-vpfe.c   | 15 ++++-
 drivers/media/platform/atmel/atmel-isc-base.c | 30 +++++++---
 drivers/media/platform/atmel/atmel-isi.c      | 19 ++++--
 drivers/media/platform/exynos-gsc/gsc-core.c  | 11 ++--
 drivers/media/platform/exynos-gsc/gsc-m2m.c   |  4 +-
 .../media/platform/exynos4-is/fimc-capture.c  |  6 +-
 drivers/media/platform/exynos4-is/fimc-is.c   |  4 +-
 .../platform/exynos4-is/fimc-isp-video.c      |  3 +-
 drivers/media/platform/exynos4-is/fimc-isp.c  |  7 +--
 drivers/media/platform/exynos4-is/fimc-lite.c |  5 +-
 drivers/media/platform/exynos4-is/fimc-m2m.c  |  5 +-
 drivers/media/platform/exynos4-is/media-dev.c |  9 +--
 drivers/media/platform/exynos4-is/mipi-csis.c | 10 ++--
 .../media/platform/marvell-ccic/mcam-core.c   |  9 ++-
 drivers/media/platform/mtk-mdp/mtk_mdp_m2m.c  |  6 +-
 .../platform/mtk-vcodec/mtk_vcodec_dec_drv.c  |  4 +-
 .../platform/mtk-vcodec/mtk_vcodec_dec_pm.c   |  8 ++-
 .../platform/mtk-vcodec/mtk_vcodec_dec_pm.h   |  2 +-
 drivers/media/platform/qcom/venus/core.c      | 59 +++++++++++++++----
 drivers/media/platform/rcar_fdp1.c            | 28 ++++++---
 drivers/media/platform/renesas-ceu.c          |  4 +-
 drivers/media/platform/s5p-jpeg/jpeg-core.c   |  5 +-
 drivers/media/platform/sh_vou.c               |  6 +-
 drivers/media/platform/sti/bdisp/bdisp-v4l2.c |  7 ++-
 drivers/media/platform/sti/delta/delta-v4l2.c |  8 +--
 .../sunxi/sun8i-rotate/sun8i_rotate.c         |  2 +-
 drivers/staging/media/hantro/hantro_drv.c     |  7 ++-
 drivers/staging/media/imx/imx7-mipi-csis.c    |  7 +--
 drivers/staging/media/rkvdec/rkvdec.c         |  2 +-
 32 files changed, 220 insertions(+), 127 deletions(-)

-- 
2.30.2

[PATCH 12/25] media: rcar_fdp1: simplify error check logic at fdp_open()

[Mauro Carvalho Chehab]

Avoid some code duplication by moving the common error path
logit at fdp_open().

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
---
 drivers/media/platform/rcar_fdp1.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/drivers/media/platform/rcar_fdp1.c b/drivers/media/platform/rcar_fdp1.c
index 01c1fbb97bf6..d26413fa5205 100644
--- a/drivers/media/platform/rcar_fdp1.c
+++ b/drivers/media/platform/rcar_fdp1.c
@@ -2117,9 +2117,7 @@ static int fdp1_open(struct file *file)
 
 	if (ctx->hdl.error) {
 		ret = ctx->hdl.error;
-		v4l2_ctrl_handler_free(&ctx->hdl);
-		kfree(ctx);
-		goto done;
+		goto error_ctx;
 	}
 
 	ctx->fh.ctrl_handler = &ctx->hdl;
@@ -2133,10 +2131,7 @@ static int fdp1_open(struct file *file)
 
 	if (IS_ERR(ctx->fh.m2m_ctx)) {
 		ret = PTR_ERR(ctx->fh.m2m_ctx);
-
-		v4l2_ctrl_handler_free(&ctx->hdl);
-		kfree(ctx);
-		goto done;
+		goto error_ctx;
 	}
 
 	/* Perform any power management required */
@@ -2147,6 +2142,12 @@ static int fdp1_open(struct file *file)
 	dprintk(fdp1, "Created instance: %p, m2m_ctx: %p\n",
 		ctx, ctx->fh.m2m_ctx);
 
+	mutex_unlock(&fdp1->dev_mutex);
+	return 0;
+
+error_ctx:
+	v4l2_ctrl_handler_free(&ctx->hdl);
+	kfree(ctx);
 done:
 	mutex_unlock(&fdp1->dev_mutex);
 	return ret;
-- 
2.30.2

[PATCH 13/25] media: rcar_fdp1: fix pm_runtime_get_sync() usage count

[Mauro Carvalho Chehab]

The pm_runtime_get_sync() internally increments the
dev->power.usage_count without decrementing it, even on errors.
Replace it by the new pm_runtime_resume_and_get(), introduced by:
commit dd8088d5a896 ("PM: runtime: Add pm_runtime_resume_and_get to deal with usage counter")
in order to properly decrement the usage counter, avoiding
a potential PM usage counter leak.

Also, right now, the driver is ignoring any troubles when
trying to do PM resume. So, add the proper error handling
for the code.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
---
 drivers/media/platform/rcar_fdp1.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/drivers/media/platform/rcar_fdp1.c b/drivers/media/platform/rcar_fdp1.c
index d26413fa5205..89aac60066d9 100644
--- a/drivers/media/platform/rcar_fdp1.c
+++ b/drivers/media/platform/rcar_fdp1.c
@@ -2135,7 +2135,9 @@ static int fdp1_open(struct file *file)
 	}
 
 	/* Perform any power management required */
-	pm_runtime_get_sync(fdp1->dev);
+	ret = pm_runtime_resume_and_get(fdp1->dev);
+	if (ret < 0)
+		goto error_pm;
 
 	v4l2_fh_add(&ctx->fh);
 
@@ -2145,6 +2147,8 @@ static int fdp1_open(struct file *file)
 	mutex_unlock(&fdp1->dev_mutex);
 	return 0;
 
+error_pm:
+       v4l2_m2m_ctx_release(ctx->fh.m2m_ctx);
 error_ctx:
 	v4l2_ctrl_handler_free(&ctx->hdl);
 	kfree(ctx);
@@ -2352,7 +2356,9 @@ static int fdp1_probe(struct platform_device *pdev)
 
 	/* Power up the cells to read HW */
 	pm_runtime_enable(&pdev->dev);
-	pm_runtime_get_sync(fdp1->dev);
+	ret = pm_runtime_resume_and_get(fdp1->dev);
+	if (ret < 0)
+		goto disable_pm;
 
 	hw_version = fdp1_read(fdp1, FD1_IP_INTDATA);
 	switch (hw_version) {
@@ -2381,6 +2387,9 @@ static int fdp1_probe(struct platform_device *pdev)
 
 	return 0;
 
+disable_pm:
+	pm_runtime_disable(fdp1->dev);
+
 release_m2m:
 	v4l2_m2m_release(fdp1->m2m_dev);
 
-- 
2.30.2

[PATCH 14/25] media: renesas-ceu: Properly check for PM errors

[Mauro Carvalho Chehab]

Right now, the driver just assumes that PM runtime resume
worked, but it may fail.

Well, the pm_runtime_get_sync() internally increments the
dev->power.usage_count without decrementing it, even on errors.

So, using it is tricky. Let's replace it by the new
pm_runtime_resume_and_get(), introduced by:
commit dd8088d5a896 ("PM: runtime: Add pm_runtime_resume_and_get to deal with usage counter")
and return an error if something bad happens.

This should ensure that the PM runtime usage_count will be
properly decremented if an error happens at open time.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
---
 drivers/media/platform/renesas-ceu.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/media/platform/renesas-ceu.c b/drivers/media/platform/renesas-ceu.c
index cd137101d41e..17f01b6e3fe0 100644
--- a/drivers/media/platform/renesas-ceu.c
+++ b/drivers/media/platform/renesas-ceu.c
@@ -1099,10 +1099,10 @@ static int ceu_open(struct file *file)
 
 	mutex_lock(&ceudev->mlock);
 	/* Causes soft-reset and sensor power on on first open */
-	pm_runtime_get_sync(ceudev->dev);
+	ret = pm_runtime_resume_and_get(ceudev->dev);
 	mutex_unlock(&ceudev->mlock);
 
-	return 0;
+	return ret;
 }
 
 static int ceu_release(struct file *file)
-- 
2.30.2

Re: [PATCH 12/25] media: rcar_fdp1: simplify error check logic at fdp_open()

[Sergei Shtylyov]

Hello!

On 05.05.2021 12:42, Mauro Carvalho Chehab wrote:

> Avoid some code duplication by moving the common error path
> logit at fdp_open().

    Logic?

> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
[...]

MBR, Sergei

Re: [PATCH 14/25] media: renesas-ceu: Properly check for PM errors

[Jacopo Mondi]

Hi Mauro,

On Wed, May 05, 2021 at 11:42:04AM +0200, Mauro Carvalho Chehab wrote:
> Right now, the driver just assumes that PM runtime resume
> worked, but it may fail.
>
> Well, the pm_runtime_get_sync() internally increments the
> dev->power.usage_count without decrementing it, even on errors.
>
> So, using it is tricky. Let's replace it by the new
> pm_runtime_resume_and_get(), introduced by:
> commit dd8088d5a896 ("PM: runtime: Add pm_runtime_resume_and_get to deal with usage counter")
> and return an error if something bad happens.
>
> This should ensure that the PM runtime usage_count will be
> properly decremented if an error happens at open time.
>
> Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

Looks good!
Acked-by: Jacopo Mondi <jacopo@jmondi.org>

Thanks
  j

> ---
>  drivers/media/platform/renesas-ceu.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/media/platform/renesas-ceu.c b/drivers/media/platform/renesas-ceu.c
> index cd137101d41e..17f01b6e3fe0 100644
> --- a/drivers/media/platform/renesas-ceu.c
> +++ b/drivers/media/platform/renesas-ceu.c
> @@ -1099,10 +1099,10 @@ static int ceu_open(struct file *file)
>
>  	mutex_lock(&ceudev->mlock);
>  	/* Causes soft-reset and sensor power on on first open */
> -	pm_runtime_get_sync(ceudev->dev);
> +	ret = pm_runtime_resume_and_get(ceudev->dev);
>  	mutex_unlock(&ceudev->mlock);
>
> -	return 0;
> +	return ret;
>  }
>
>  static int ceu_release(struct file *file)
> --
> 2.30.2
>

Re: [PATCH 13/25] media: rcar_fdp1: fix pm_runtime_get_sync() usage count

[Jonathan Cameron]

On Wed, 5 May 2021 11:42:03 +0200
Mauro Carvalho Chehab <mchehab+huawei@kernel.org> wrote:

> The pm_runtime_get_sync() internally increments the
> dev->power.usage_count without decrementing it, even on errors.
> Replace it by the new pm_runtime_resume_and_get(), introduced by:
> commit dd8088d5a896 ("PM: runtime: Add pm_runtime_resume_and_get to deal with usage counter")
> in order to properly decrement the usage counter, avoiding
> a potential PM usage counter leak.
> 
> Also, right now, the driver is ignoring any troubles when
> trying to do PM resume. So, add the proper error handling
> for the code.
> 
> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> ---
>  drivers/media/platform/rcar_fdp1.c | 13 +++++++++++--
>  1 file changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/media/platform/rcar_fdp1.c b/drivers/media/platform/rcar_fdp1.c
> index d26413fa5205..89aac60066d9 100644
> --- a/drivers/media/platform/rcar_fdp1.c
> +++ b/drivers/media/platform/rcar_fdp1.c
> @@ -2135,7 +2135,9 @@ static int fdp1_open(struct file *file)
>  	}
>  
>  	/* Perform any power management required */
> -	pm_runtime_get_sync(fdp1->dev);
> +	ret = pm_runtime_resume_and_get(fdp1->dev);
> +	if (ret < 0)
> +		goto error_pm;
>  
>  	v4l2_fh_add(&ctx->fh);
>  
> @@ -2145,6 +2147,8 @@ static int fdp1_open(struct file *file)
>  	mutex_unlock(&fdp1->dev_mutex);
>  	return 0;
>  
> +error_pm:
> +       v4l2_m2m_ctx_release(ctx->fh.m2m_ctx);
>  error_ctx:
>  	v4l2_ctrl_handler_free(&ctx->hdl);
>  	kfree(ctx);
> @@ -2352,7 +2356,9 @@ static int fdp1_probe(struct platform_device *pdev)
>  
>  	/* Power up the cells to read HW */
>  	pm_runtime_enable(&pdev->dev);
> -	pm_runtime_get_sync(fdp1->dev);
> +	ret = pm_runtime_resume_and_get(fdp1->dev);
> +	if (ret < 0)
> +		goto disable_pm;
>  
>  	hw_version = fdp1_read(fdp1, FD1_IP_INTDATA);
>  	switch (hw_version) {
> @@ -2381,6 +2387,9 @@ static int fdp1_probe(struct platform_device *pdev)
>  
>  	return 0;
>  
> +disable_pm:
> +	pm_runtime_disable(fdp1->dev);
> +
>  release_m2m:
>  	v4l2_m2m_release(fdp1->m2m_dev);
>  

Re: [PATCH 00/25] Fix some PM runtime issues at the media subsystem

[Mauro Carvalho Chehab]

Em Wed,  5 May 2021 11:41:50 +0200
Mauro Carvalho Chehab <mchehab+huawei@kernel.org> escreveu:

> As part of an effort to cleanup pm_runtime*get* calls, I detected a number
> of issues at the media subsystem.
> 
> Most of the patches here were submitted previously at:
> 
> 	https://lore.kernel.org/linux-media/cover.1619621413.git.mchehab+huawei@kernel.org/
> 
> This series contain just the bug fixes and other related issues that are
> present with the current code on media.

Series merged on my stage tree, at:
	https://git.linuxtv.org/media_stage.git/log/

I'll be merging it at media_tree (either for 5.13 or 5.14) after the
end of the merge window (likely next week).

Please let me know if you find any problems on it.

PS.: please notice that my stage tree can be rebased.

Regards,
Mauro

> 
> It address the points from the existing reviews. I also did my own
> set of reviews, in order to avoid regressions.
> 
> Changes from v4 of the previous changeset:
> 
> - reworked i2c/css RPM get logic;
> - dropped two patches that could cause regressions;
> - am437x: keep using pm_runtime_get_sync on suspend/resume;
> - atmel: fix the returned code and add a print on failures at start streaming;
> - simplify some checks for return code > 0;
> - mdk-vcodec: properly handle RPM errors at device on logic;
> - venus: rework venus_sys_error_handler() logic;
> - sti/delta: fix an issue at the error checking logic.
> 
> Mauro Carvalho Chehab (25):
>   staging: media: rkvdec: fix pm_runtime_get_sync() usage count
>   staging: media: imx7-mipi-csis: fix pm_runtime_get_sync() usage count
>   media: venus: Rework error fail recover logic
>   media: s5p_cec: decrement usage count if disabled
>   media: i2c: ccs-core: return the right error code at suspend
>   media: i2c: imx334: fix the pm runtime get logic
>   media: exynos-gsc: don't resume at remove time
>   media: atmel: properly get pm_runtime
>   media: hantro: do a PM resume earlier
>   media: marvel-ccic: fix some issues when getting pm_runtime
>   media: mdk-mdp: fix pm_runtime_get_sync() usage count
>   media: rcar_fdp1: simplify error check logic at fdp_open()
>   media: rcar_fdp1: fix pm_runtime_get_sync() usage count
>   media: renesas-ceu: Properly check for PM errors
>   media: s5p: fix pm_runtime_get_sync() usage count
>   media: am437x: fix pm_runtime_get_sync() usage count
>   media: sh_vou: fix pm_runtime_get_sync() usage count
>   media: mtk-vcodec: fix PM runtime get logic
>   media: s5p-jpeg: fix pm_runtime_get_sync() usage count
>   media: sti/delta: use pm_runtime_resume_and_get()
>   media: sunxi: fix pm_runtime_get_sync() usage count
>   media: sti/bdisp: fix pm_runtime_get_sync() usage count
>   media: exynos4-is: fix pm_runtime_get_sync() usage count
>   media: exynos-gsc: fix pm_runtime_get_sync() usage count
>   media: i2c: ccs-core: fix pm_runtime_get_sync() usage count
> 
>  drivers/media/cec/platform/s5p/s5p_cec.c      |  7 ++-
>  drivers/media/i2c/ccs/ccs-core.c              | 41 ++++++++-----
>  drivers/media/i2c/imx334.c                    |  7 ++-
>  drivers/media/platform/am437x/am437x-vpfe.c   | 15 ++++-
>  drivers/media/platform/atmel/atmel-isc-base.c | 30 +++++++---
>  drivers/media/platform/atmel/atmel-isi.c      | 19 ++++--
>  drivers/media/platform/exynos-gsc/gsc-core.c  | 11 ++--
>  drivers/media/platform/exynos-gsc/gsc-m2m.c   |  4 +-
>  .../media/platform/exynos4-is/fimc-capture.c  |  6 +-
>  drivers/media/platform/exynos4-is/fimc-is.c   |  4 +-
>  .../platform/exynos4-is/fimc-isp-video.c      |  3 +-
>  drivers/media/platform/exynos4-is/fimc-isp.c  |  7 +--
>  drivers/media/platform/exynos4-is/fimc-lite.c |  5 +-
>  drivers/media/platform/exynos4-is/fimc-m2m.c  |  5 +-
>  drivers/media/platform/exynos4-is/media-dev.c |  9 +--
>  drivers/media/platform/exynos4-is/mipi-csis.c | 10 ++--
>  .../media/platform/marvell-ccic/mcam-core.c   |  9 ++-
>  drivers/media/platform/mtk-mdp/mtk_mdp_m2m.c  |  6 +-
>  .../platform/mtk-vcodec/mtk_vcodec_dec_drv.c  |  4 +-
>  .../platform/mtk-vcodec/mtk_vcodec_dec_pm.c   |  8 ++-
>  .../platform/mtk-vcodec/mtk_vcodec_dec_pm.h   |  2 +-
>  drivers/media/platform/qcom/venus/core.c      | 59 +++++++++++++++----
>  drivers/media/platform/rcar_fdp1.c            | 28 ++++++---
>  drivers/media/platform/renesas-ceu.c          |  4 +-
>  drivers/media/platform/s5p-jpeg/jpeg-core.c   |  5 +-
>  drivers/media/platform/sh_vou.c               |  6 +-
>  drivers/media/platform/sti/bdisp/bdisp-v4l2.c |  7 ++-
>  drivers/media/platform/sti/delta/delta-v4l2.c |  8 +--
>  .../sunxi/sun8i-rotate/sun8i_rotate.c         |  2 +-
>  drivers/staging/media/hantro/hantro_drv.c     |  7 ++-
>  drivers/staging/media/imx/imx7-mipi-csis.c    |  7 +--
>  drivers/staging/media/rkvdec/rkvdec.c         |  2 +-
>  32 files changed, 220 insertions(+), 127 deletions(-)
> 



Thanks,
Mauro