Re: [PATCH 10/12] [media] move i2c files into drivers/media/i2c

2012-09-06 Thread Prabhakar Lad
Hi Mauro,

Thanks for the patch.

On Thu, Sep 6, 2012 at 7:37 PM, Mauro Carvalho Chehab
 wrote:
> Em 24-08-2012 20:44, Sylwester Nawrocki escreveu:
>> From: Sylwester Nawrocki 
>> Date: Sat, 25 Aug 2012 01:23:14 +0200
>> Subject: [PATCH] [media] Fix link order of the V4L2 bridge and I2C modules
>>
>> All I2C modules must be linked first to ensure proper module
>> initialization order. With platform devices linked before I2C
>> modules I2C subdev registration fails as the subdev drivers
>> are not yet initialized during bridge driver's probing.
>>
>> This fixes regression introduced with commmit cb7a01ac324bf2ee2,
>> "[media] move i2c files into drivers/media/i2c".
>>
>> Signed-off-by: Sylwester Nawrocki 
>> ---
>>  drivers/media/Makefile |7 ---
>>  1 files changed, 4 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/media/Makefile b/drivers/media/Makefile
>> index b0b0193..92a8bcf 100644
>> --- a/drivers/media/Makefile
>> +++ b/drivers/media/Makefile
>> @@ -8,8 +8,9 @@ ifeq ($(CONFIG_MEDIA_CONTROLLER),y)
>>obj-$(CONFIG_MEDIA_SUPPORT) += media.o
>>  endif
>>
>> -obj-y += tuners/ common/ rc/ platform/
>> -obj-y += i2c/ pci/ usb/ mmc/ firewire/ parport/
>> +obj-$(CONFIG_VIDEO_DEV) += v4l2-core/
>> +obj-y += common/ rc/ i2c/
>> +obj-y += tuners/ platform/ pci/ usb/ mmc/ firewire/ parport/
>>
>> -obj-$(CONFIG_VIDEO_DEV) += radio/ v4l2-core/
>> +obj-$(CONFIG_VIDEO_DEV) += radio/
>>  obj-$(CONFIG_DVB_CORE)  += dvb-core/ dvb-frontends/
>> -- 1.7.4.1
>
> Hmm... This change seems incomplete on my eyes: tuners and dvb-frontends
> are also I2C drivers. So, while this fixes the issue for platform drivers,
> other drivers will still suffer this issue, at least on drivers that doesn't
> depend on drivers located outside the media subsystem[1]
>
> [1] thankfully, staging compiles after media, so the drivers there
> shouldn't be affected. Also, drivers that use alsa won't be affected, as
> alsa core (with is compiled after media) uses subsys_initcall().
>
> IMO, the correct fix is the one below. Could you please test it?
>
> Regards,
> Mauro
>
> -
>
> [media] Fix init order for I2C drivers
>
> Based on a patch from Sylvester Nawrocki
>
> This fixes regression introduced with commmit cb7a01ac324bf2ee2,
> "[media] move i2c files into drivers/media/i2c".
>
> The linked order affect what drivers will be initialized first, when
> they're built-in at Kernel. While there are macros that allow changing
> the init order, like subsys_initcall(), late_initcall() & friends,
> when all drivers  linked belong to the same subsystem, it is easier
> to change the order at the Makefile.
>
> All I2C modules must be linked before any drivers that actually use it,
> in order to ensure proper module initialization order.
>
> Also, the core drivers should be initialized before the drivers that use
> them.
>
> This patch reorders the drivers init, in order to fulfill the above
> requirements.
>
> Reported-by: Sylwester Nawrocki 
> Signed-off-by: Mauro Carvalho Chehab 
>
Acked-by: Prabhakar Lad 


Thanks and Regards,
--Prabhakar Lad

> diff --git a/drivers/media/Makefile b/drivers/media/Makefile
> index b0b0193..620f275 100644
> --- a/drivers/media/Makefile
> +++ b/drivers/media/Makefile
> @@ -4,12 +4,30 @@
>
>  media-objs := media-device.o media-devnode.o media-entity.o
>
> +#
> +# I2C drivers should come before other drivers, otherwise they'll fail
> +# when compiled as builtin drivers
> +#
> +obj-y += i2c/ tuners/
> +obj-$(CONFIG_DVB_CORE)  += dvb-frontends/
> +
> +#
> +# Now, let's link-in the media core
> +#
>  ifeq ($(CONFIG_MEDIA_CONTROLLER),y)
>obj-$(CONFIG_MEDIA_SUPPORT) += media.o
>  endif
>
> -obj-y += tuners/ common/ rc/ platform/
> -obj-y += i2c/ pci/ usb/ mmc/ firewire/ parport/
> +obj-$(CONFIG_VIDEO_DEV) += v4l2-core/
> +obj-$(CONFIG_DVB_CORE)  += dvb-core/
> +
> +# There are both core and drivers at RC subtree - merge before drivers
> +obj-y += rc/
> +
> +#
> +# Finally, merge the drivers that require the core
> +#
> +
> +obj-y += common/ platform/ pci/ usb/ mmc/ firewire/ parport/
> +obj-$(CONFIG_VIDEO_DEV) += radio/
>
> -obj-$(CONFIG_VIDEO_DEV) += radio/ v4l2-core/
> -obj-$(CONFIG_DVB_CORE)  += dvb-core/ dvb-frontends/
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


Re: [PATCH 10/12] [media] move i2c files into drivers/media/i2c

2012-09-06 Thread Sylwester Nawrocki
On 09/06/2012 04:07 PM, Mauro Carvalho Chehab wrote:
> Em 24-08-2012 20:44, Sylwester Nawrocki escreveu:
>> From: Sylwester Nawrocki
>> Date: Sat, 25 Aug 2012 01:23:14 +0200
>> Subject: [PATCH] [media] Fix link order of the V4L2 bridge and I2C modules
>>
>> All I2C modules must be linked first to ensure proper module
>> initialization order. With platform devices linked before I2C
>> modules I2C subdev registration fails as the subdev drivers
>> are not yet initialized during bridge driver's probing.
>>
>> This fixes regression introduced with commmit cb7a01ac324bf2ee2,
>> "[media] move i2c files into drivers/media/i2c".
>>
>> Signed-off-by: Sylwester Nawrocki
>> ---
>>   drivers/media/Makefile |7 ---
>>   1 files changed, 4 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/media/Makefile b/drivers/media/Makefile
>> index b0b0193..92a8bcf 100644
>> --- a/drivers/media/Makefile
>> +++ b/drivers/media/Makefile
>> @@ -8,8 +8,9 @@ ifeq ($(CONFIG_MEDIA_CONTROLLER),y)
>> obj-$(CONFIG_MEDIA_SUPPORT) += media.o
>>   endif
>>
>> -obj-y += tuners/ common/ rc/ platform/
>> -obj-y += i2c/ pci/ usb/ mmc/ firewire/ parport/
>> +obj-$(CONFIG_VIDEO_DEV) += v4l2-core/
>> +obj-y += common/ rc/ i2c/
>> +obj-y += tuners/ platform/ pci/ usb/ mmc/ firewire/ parport/
>>
>> -obj-$(CONFIG_VIDEO_DEV) += radio/ v4l2-core/
>> +obj-$(CONFIG_VIDEO_DEV) += radio/
>>   obj-$(CONFIG_DVB_CORE)  += dvb-core/ dvb-frontends/
>> -- 1.7.4.1
> 
> Hmm... This change seems incomplete on my eyes: tuners and dvb-frontends
> are also I2C drivers. So, while this fixes the issue for platform drivers,
> other drivers will still suffer this issue, at least on drivers that doesn't
> depend on drivers located outside the media subsystem[1]

Yeah, that's possible. I wasn't very confident about anything but platform 
and i2c drivers. Thanks for improving it. I tested the patch with a camera 
host and I2C sensor driver and it also fixed the problem.

> [1] thankfully, staging compiles after media, so the drivers there
> shouldn't be affected. Also, drivers that use alsa won't be affected, as
> alsa core (with is compiled after media) uses subsys_initcall().
> 
> IMO, the correct fix is the one below. Could you please test it?
> 
> Regards,
> Mauro
> 
> -
> 
> [media] Fix init order for I2C drivers
> 
> Based on a patch from Sylvester Nawrocki
> 
> This fixes regression introduced with commmit cb7a01ac324bf2ee2,
> "[media] move i2c files into drivers/media/i2c".
> 
> The linked order affect what drivers will be initialized first, when
> they're built-in at Kernel. While there are macros that allow changing
> the init order, like subsys_initcall(), late_initcall()&  friends,
> when all drivers  linked belong to the same subsystem, it is easier
> to change the order at the Makefile.
> 
> All I2C modules must be linked before any drivers that actually use it,
> in order to ensure proper module initialization order.
> 
> Also, the core drivers should be initialized before the drivers that use
> them.
> 
> This patch reorders the drivers init, in order to fulfill the above
> requirements.
> 
> Reported-by: Sylwester Nawrocki
> Signed-off-by: Mauro Carvalho Chehab
 
Acked-by: Sylwester Nawrocki 
 
> diff --git a/drivers/media/Makefile b/drivers/media/Makefile
> index b0b0193..620f275 100644
> --- a/drivers/media/Makefile
> +++ b/drivers/media/Makefile
> @@ -4,12 +4,30 @@
> 
>   media-objs  := media-device.o media-devnode.o media-entity.o
> 
> +#
> +# I2C drivers should come before other drivers, otherwise they'll fail
> +# when compiled as builtin drivers
> +#
> +obj-y += i2c/ tuners/
> +obj-$(CONFIG_DVB_CORE)  += dvb-frontends/
> +
> +#
> +# Now, let's link-in the media core
> +#
>   ifeq ($(CONFIG_MEDIA_CONTROLLER),y)
> obj-$(CONFIG_MEDIA_SUPPORT) += media.o
>   endif
> 
> -obj-y += tuners/ common/ rc/ platform/
> -obj-y += i2c/ pci/ usb/ mmc/ firewire/ parport/
> +obj-$(CONFIG_VIDEO_DEV) += v4l2-core/
> +obj-$(CONFIG_DVB_CORE)  += dvb-core/
> +
> +# There are both core and drivers at RC subtree - merge before drivers
> +obj-y += rc/
> +
> +#
> +# Finally, merge the drivers that require the core
> +#
> +
> +obj-y += common/ platform/ pci/ usb/ mmc/ firewire/ parport/
> +obj-$(CONFIG_VIDEO_DEV) += radio/
> 
> -obj-$(CONFIG_VIDEO_DEV) += radio/ v4l2-core/
> -obj-$(CONFIG_DVB_CORE)  += dvb-core/ dvb-frontends/

--

Regards,
Sylwester
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


Re: [PATCH 10/12] [media] move i2c files into drivers/media/i2c

2012-09-06 Thread Mauro Carvalho Chehab
Em 24-08-2012 20:44, Sylwester Nawrocki escreveu:
> From: Sylwester Nawrocki 
> Date: Sat, 25 Aug 2012 01:23:14 +0200
> Subject: [PATCH] [media] Fix link order of the V4L2 bridge and I2C modules
> 
> All I2C modules must be linked first to ensure proper module
> initialization order. With platform devices linked before I2C
> modules I2C subdev registration fails as the subdev drivers
> are not yet initialized during bridge driver's probing.
> 
> This fixes regression introduced with commmit cb7a01ac324bf2ee2,
> "[media] move i2c files into drivers/media/i2c".
> 
> Signed-off-by: Sylwester Nawrocki 
> ---
>  drivers/media/Makefile |7 ---
>  1 files changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/media/Makefile b/drivers/media/Makefile
> index b0b0193..92a8bcf 100644
> --- a/drivers/media/Makefile
> +++ b/drivers/media/Makefile
> @@ -8,8 +8,9 @@ ifeq ($(CONFIG_MEDIA_CONTROLLER),y)
>obj-$(CONFIG_MEDIA_SUPPORT) += media.o
>  endif
>  
> -obj-y += tuners/ common/ rc/ platform/
> -obj-y += i2c/ pci/ usb/ mmc/ firewire/ parport/
> +obj-$(CONFIG_VIDEO_DEV) += v4l2-core/
> +obj-y += common/ rc/ i2c/
> +obj-y += tuners/ platform/ pci/ usb/ mmc/ firewire/ parport/
>  
> -obj-$(CONFIG_VIDEO_DEV) += radio/ v4l2-core/
> +obj-$(CONFIG_VIDEO_DEV) += radio/
>  obj-$(CONFIG_DVB_CORE)  += dvb-core/ dvb-frontends/
> -- 1.7.4.1

Hmm... This change seems incomplete on my eyes: tuners and dvb-frontends
are also I2C drivers. So, while this fixes the issue for platform drivers,
other drivers will still suffer this issue, at least on drivers that doesn't
depend on drivers located outside the media subsystem[1]

[1] thankfully, staging compiles after media, so the drivers there
shouldn't be affected. Also, drivers that use alsa won't be affected, as
alsa core (with is compiled after media) uses subsys_initcall().

IMO, the correct fix is the one below. Could you please test it?

Regards,
Mauro

-

[media] Fix init order for I2C drivers

Based on a patch from Sylvester Nawrocki

This fixes regression introduced with commmit cb7a01ac324bf2ee2,
"[media] move i2c files into drivers/media/i2c".

The linked order affect what drivers will be initialized first, when
they're built-in at Kernel. While there are macros that allow changing
the init order, like subsys_initcall(), late_initcall() & friends,
when all drivers  linked belong to the same subsystem, it is easier
to change the order at the Makefile.

All I2C modules must be linked before any drivers that actually use it,
in order to ensure proper module initialization order.

Also, the core drivers should be initialized before the drivers that use
them.

This patch reorders the drivers init, in order to fulfill the above
requirements.

Reported-by: Sylwester Nawrocki 
Signed-off-by: Mauro Carvalho Chehab 

diff --git a/drivers/media/Makefile b/drivers/media/Makefile
index b0b0193..620f275 100644
--- a/drivers/media/Makefile
+++ b/drivers/media/Makefile
@@ -4,12 +4,30 @@
 
 media-objs := media-device.o media-devnode.o media-entity.o
 
+#
+# I2C drivers should come before other drivers, otherwise they'll fail
+# when compiled as builtin drivers
+#
+obj-y += i2c/ tuners/
+obj-$(CONFIG_DVB_CORE)  += dvb-frontends/
+
+#
+# Now, let's link-in the media core
+#
 ifeq ($(CONFIG_MEDIA_CONTROLLER),y)
   obj-$(CONFIG_MEDIA_SUPPORT) += media.o
 endif
 
-obj-y += tuners/ common/ rc/ platform/
-obj-y += i2c/ pci/ usb/ mmc/ firewire/ parport/
+obj-$(CONFIG_VIDEO_DEV) += v4l2-core/
+obj-$(CONFIG_DVB_CORE)  += dvb-core/
+
+# There are both core and drivers at RC subtree - merge before drivers
+obj-y += rc/
+
+#
+# Finally, merge the drivers that require the core
+#
+
+obj-y += common/ platform/ pci/ usb/ mmc/ firewire/ parport/
+obj-$(CONFIG_VIDEO_DEV) += radio/
 
-obj-$(CONFIG_VIDEO_DEV) += radio/ v4l2-core/
-obj-$(CONFIG_DVB_CORE)  += dvb-core/ dvb-frontends/


--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


Re: [PATCH 10/12] [media] move i2c files into drivers/media/i2c

2012-08-24 Thread Sylwester Nawrocki
Hi Mauro,

On 08/15/2012 03:48 PM, Mauro Carvalho Chehab wrote:
> Move ancillary I2C drivers into drivers/media/i2c, in order to
> better organize them.
> 
> Signed-off-by: Mauro Carvalho Chehab
> ---
>   drivers/media/Kconfig  |   9 +-
>   drivers/media/Makefile |   2 +-
>   drivers/media/i2c/Kconfig  | 566 
> 
>   drivers/media/i2c/Makefile |  63 +++
>   drivers/media/{video =>  i2c}/adp1653.c |   2 +-
...
>   rename drivers/media/{video =>  i2c}/wm8775.c (100%)
> 
> diff --git a/drivers/media/Kconfig b/drivers/media/Kconfig
> index c9cdc61..26f3de5 100644
> --- a/drivers/media/Kconfig
> +++ b/drivers/media/Kconfig
> @@ -151,18 +151,15 @@ source "drivers/media/rc/Kconfig"
> 
>   source "drivers/media/tuners/Kconfig"
> 
> +source "drivers/media/i2c/Kconfig"
> +
>   #
> -# Video/Radio/Hybrid adapters
> +# V4L platform/mem2mem drivers
>   #
> -
>   source "drivers/media/video/Kconfig"
> 
>   source "drivers/media/radio/Kconfig"
> 
> -#
> -# DVB adapters
> -#
> -
>   source "drivers/media/pci/Kconfig"
>   source "drivers/media/usb/Kconfig"
>   source "drivers/media/mmc/Kconfig"
> diff --git a/drivers/media/Makefile b/drivers/media/Makefile
> index 360c44d..e1be196 100644
> --- a/drivers/media/Makefile
> +++ b/drivers/media/Makefile
> @@ -9,7 +9,7 @@ ifeq ($(CONFIG_MEDIA_CONTROLLER),y)
>   endif
> 
>   obj-y += tuners/ common/ rc/ video/
> -obj-y += pci/ usb/ mmc/ firewire/ parport/
> +obj-y += i2c/ pci/ usb/ mmc/ firewire/ parport/

That way all i2c drivers won't be linked before bridge drivers, which
causes failure of sensor subdev registration. There was a comment
about it in the original Makefile, please see further below.

This issue have already shown up for me in a real system,

[1.075000] s3c_camif_driver_init:633
[1.08] s3c-camif s3c2440-camif: sensor clock frequency: 1200
[1.09] s3c-camif: failed to acquire subdev OV9650
[1.095000] platform s3c2440-camif: Driver s3c-camif requests probe deferral
[1.10] ov965x_i2c_driver_init:820

I've attached a patch to fix this issue.

>   obj-$(CONFIG_VIDEO_DEV) += radio/ v4l2-core/
>   obj-$(CONFIG_DVB_CORE)  += dvb-core/ dvb-frontends/

[snip]

> diff --git a/drivers/media/video/Makefile b/drivers/media/video/Makefile
> index a0c6692..52a04fa 100644
> --- a/drivers/media/video/Makefile
> +++ b/drivers/media/video/Makefile
> @@ -2,73 +2,9 @@
>   # Makefile for the video capture/playback device drivers.
>   #
> 
> -msp3400-objs :=  msp3400-driver.o msp3400-kthreads.o
> -
>   omap2cam-objs   :=  omap24xxcam.o omap24xxcam-dma.o
> 
> -# Helper modules
> -
> -obj-$(CONFIG_VIDEO_APTINA_PLL) += aptina-pll.o
> -
> -# All i2c modules must come first:
^
|
This is currently not preserved +
   
> -obj-$(CONFIG_VIDEO_TVAUDIO) += tvaudio.o
> -obj-$(CONFIG_VIDEO_TDA7432) += tda7432.o
> -obj-$(CONFIG_VIDEO_SAA6588) += saa6588.o

--

Regards,
Sylwester
>From ca02e4cba9f400a0066f2875c2fe4284bc8e06dc Mon Sep 17 00:00:00 2001
From: Sylwester Nawrocki 
Date: Sat, 25 Aug 2012 01:23:14 +0200
Subject: [PATCH] [media] Fix link order of the V4L2 bridge and I2C modules

All I2C modules must be linked first to ensure proper module
initialization order. With platform devices linked before I2C
modules I2C subdev registration fails as the subdev drivers
are not yet initialized during bridge driver's probing.

This fixes regression introduced with commmit cb7a01ac324bf2ee2,
"[media] move i2c files into drivers/media/i2c".

Signed-off-by: Sylwester Nawrocki 
---
 drivers/media/Makefile |7 ---
 1 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/media/Makefile b/drivers/media/Makefile
index b0b0193..92a8bcf 100644
--- a/drivers/media/Makefile
+++ b/drivers/media/Makefile
@@ -8,8 +8,9 @@ ifeq ($(CONFIG_MEDIA_CONTROLLER),y)
   obj-$(CONFIG_MEDIA_SUPPORT) += media.o
 endif
 
-obj-y += tuners/ common/ rc/ platform/
-obj-y += i2c/ pci/ usb/ mmc/ firewire/ parport/
+obj-$(CONFIG_VIDEO_DEV) += v4l2-core/
+obj-y += common/ rc/ i2c/
+obj-y += tuners/ platform/ pci/ usb/ mmc/ firewire/ parport/
 
-obj-$(CONFIG_VIDEO_DEV) += radio/ v4l2-core/
+obj-$(CONFIG_VIDEO_DEV) += radio/
 obj-$(CONFIG_DVB_CORE)  += dvb-core/ dvb-frontends/
-- 
1.7.4.1