diff mbox series

[v7,2/7] mtk-mdp: add driver to probe mdp components

Message ID 20210825163247.v7.2.Ie6d1e6e39cf9b5d6b2108ae1096af34c3d55880b@changeid (mailing list archive)
State New, archived
Headers show
Series Refactor MTK MDP driver into core/components | expand

Commit Message

Eizan Miyamoto Aug. 25, 2021, 6:33 a.m. UTC
Broadly, this patch (1) adds a driver for various MTK MDP components to
go alongside the main MTK MDP driver, and (2) hooks them all together
using the component framework.

(1) Up until now, the MTK MDP driver controls 8 devices in the device
tree on its own. When running tests for the hardware video decoder, we
found that the iommus and LARBs were not being properly configured. To
configure them, a driver for each be added to mtk_mdp_comp so that
mtk_iommu_add_device() can (eventually) be called from dma_configure()
inside really_probe().

(2) The integration into the component framework allows us to defer the
registration with the v4l2 subsystem until all the MDP-related devices
have been probed, so that the relevant device node does not become
available until initialization of all the components is complete.

Some notes about how the component framework has been integrated:

- The driver for the rdma0 component serves double duty as the "master"
  (aggregate) driver as well as a component driver. This is a non-ideal
  compromise until a better solution is developed. This device is
  differentiated from the rest by checking for a "mediatek,vpu" property
  in the device node.

- The list of mdp components remains hard-coded as mtk_mdp_comp_dt_ids[]
  in mtk_mdp_core.c, and as mtk_mdp_comp_driver_dt_match[] in
  mtk_mdp_comp.c. This unfortunate duplication of information is
  addressed in a following patch in this series.

- The component driver calls component_add() for each device that is
  probed.

- In mtk_mdp_probe (the "master" device), we scan the device tree for
  any matching nodes against mtk_mdp_comp_dt_ids, and add component
  matches for them. The match criteria is a matching device node
  pointer.

- When the set of components devices that have been probed corresponds
  with the list that is generated by the "master", the callback to
  mtk_mdp_master_bind() is made, which then calls the component bind
  functions.

- Inside mtk_mdp_master_bind(), once all the component bind functions
  have been called, we can then register our device to the v4l2
  subsystem.

- The call to pm_runtime_enable() in the master device is called after
  all the components have been registered by their bind() functions
  called by mtk_mtp_master_bind(). As a result, the list of components
  will not change while power management callbacks mtk_mdp_suspend()/
  resume() are accessing the list of components.

Signed-off-by: Eizan Miyamoto <eizan@chromium.org>
Reviewed-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
Reviewed-by: Houlong Wei <houlong.wei@mediatek.com>
---

(no changes since v1)

 drivers/media/platform/mtk-mdp/mtk_mdp_comp.c | 143 ++++++++++++--
 drivers/media/platform/mtk-mdp/mtk_mdp_comp.h |  25 +--
 drivers/media/platform/mtk-mdp/mtk_mdp_core.c | 174 +++++++++++++-----
 drivers/media/platform/mtk-mdp/mtk_mdp_core.h |   1 +
 4 files changed, 252 insertions(+), 91 deletions(-)

Comments

Ezequiel Garcia Sept. 4, 2021, 12:34 p.m. UTC | #1
Hi Eizan,

Sorry for seeing this series so late.

On Wed, 25 Aug 2021 at 03:35, Eizan Miyamoto <eizan@chromium.org> wrote:
>
> Broadly, this patch (1) adds a driver for various MTK MDP components to
> go alongside the main MTK MDP driver, and (2) hooks them all together
> using the component framework.
>
> (1) Up until now, the MTK MDP driver controls 8 devices in the device
> tree on its own. When running tests for the hardware video decoder, we
> found that the iommus and LARBs were not being properly configured.

Why were not being properly configured? What was the problem?
Why not fixing that instead?

Does this mean the driver is currently broken and unusable?

> To
> configure them, a driver for each be added to mtk_mdp_comp so that
> mtk_iommu_add_device() can (eventually) be called from dma_configure()
> inside really_probe().
>
> (2) The integration into the component framework allows us to defer the
> registration with the v4l2 subsystem until all the MDP-related devices
> have been probed, so that the relevant device node does not become
> available until initialization of all the components is complete.
>
> Some notes about how the component framework has been integrated:
>
> - The driver for the rdma0 component serves double duty as the "master"
>   (aggregate) driver as well as a component driver. This is a non-ideal
>   compromise until a better solution is developed. This device is
>   differentiated from the rest by checking for a "mediatek,vpu" property
>   in the device node.
>

As I have stated in Yunfei, I am not convinced you need an async framework
at all. It seems all these devices could have been linked together
in the device tree, and then have a master device to tie them.

I.e. something like

mdp {
  mdp_rdma0 {
  }
  mdp_rsz0 {
  }
  mdp_rsz1 {
  }
}

All this async games seem like making the driver really obfuscated,
which will mean harder to debug and maintain.
I am not sure we want that burden.

Even if we are all fully convinced that you absolutely need
an async framework, then what's wrong with v4l2-async?

I would start by addressing what is wrong with the IOMMUs
in the current design.

Thanks,
Ezequiel

> - The list of mdp components remains hard-coded as mtk_mdp_comp_dt_ids[]
>   in mtk_mdp_core.c, and as mtk_mdp_comp_driver_dt_match[] in
>   mtk_mdp_comp.c. This unfortunate duplication of information is
>   addressed in a following patch in this series.
>
> - The component driver calls component_add() for each device that is
>   probed.
>
> - In mtk_mdp_probe (the "master" device), we scan the device tree for
>   any matching nodes against mtk_mdp_comp_dt_ids, and add component
>   matches for them. The match criteria is a matching device node
>   pointer.
>
> - When the set of components devices that have been probed corresponds
>   with the list that is generated by the "master", the callback to
>   mtk_mdp_master_bind() is made, which then calls the component bind
>   functions.
>
> - Inside mtk_mdp_master_bind(), once all the component bind functions
>   have been called, we can then register our device to the v4l2
>   subsystem.
>
> - The call to pm_runtime_enable() in the master device is called after
>   all the components have been registered by their bind() functions
>   called by mtk_mtp_master_bind(). As a result, the list of components
>   will not change while power management callbacks mtk_mdp_suspend()/
>   resume() are accessing the list of components.
>
> Signed-off-by: Eizan Miyamoto <eizan@chromium.org>
> Reviewed-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
> Reviewed-by: Houlong Wei <houlong.wei@mediatek.com>
> ---
>
houlong.wei Sept. 5, 2021, 4:23 p.m. UTC | #2
Hi Ezequiel,

Thank you for your attention to this series of patches. I answer partial of your questions below.
Regards,
Houlong

On Sat, 2021-09-04 at 20:34 +0800, Ezequiel Garcia wrote:
> Hi Eizan,
> 
> Sorry for seeing this series so late.
> 
> On Wed, 25 Aug 2021 at 03:35, Eizan Miyamoto <eizan@chromium.org>
> wrote:
> > 
> > Broadly, this patch (1) adds a driver for various MTK MDP
> > components to
> > go alongside the main MTK MDP driver, and (2) hooks them all
> > together
> > using the component framework.
> > 
> > (1) Up until now, the MTK MDP driver controls 8 devices in the
> > device
> > tree on its own. When running tests for the hardware video decoder,
> > we
> > found that the iommus and LARBs were not being properly configured.
> 
> Why were not being properly configured? What was the problem?
> Why not fixing that instead?
> 
> Does this mean the driver is currently broken and unusable?

This series of patches are supplements to another series, please refer
to  
https://patchwork.kernel.org/project/linux-mediatek/list/?series=515129c
, which add device link between the mtk-iommu consumer and the mtk-larb 
devices. Without that series of patches, the mtk-mdp driver can work
well so far.
But with that series, it seems the device link only can be established
for the device which is registered as a platform driver. That's why
Eizan adds this series of patches to make all mdp components to be
registered as platform drivers.

> 
> > To
> > configure them, a driver for each be added to mtk_mdp_comp so that
> > mtk_iommu_add_device() can (eventually) be called from
> > dma_configure()
> > inside really_probe().
> > 
> > (2) The integration into the component framework allows us to defer
> > the
> > registration with the v4l2 subsystem until all the MDP-related
> > devices
> > have been probed, so that the relevant device node does not become
> > available until initialization of all the components is complete.
> > 
> > Some notes about how the component framework has been integrated:
> > 
> > - The driver for the rdma0 component serves double duty as the
> > "master"
> >   (aggregate) driver as well as a component driver. This is a non-
> > ideal
> >   compromise until a better solution is developed. This device is
> >   differentiated from the rest by checking for a "mediatek,vpu"
> > property
> >   in the device node.
> > 
> 
> As I have stated in Yunfei, I am not convinced you need an async
> framework
> at all. It seems all these devices could have been linked together
> in the device tree, and then have a master device to tie them.
> 
> I.e. something like
> 
> mdp {
>   mdp_rdma0 {
>   }
>   mdp_rsz0 {
>   }
>   mdp_rsz1 {
>   }
> }
> 

The commit message of the patch below explains that " If the mdp_*
nodes are under an mdp sub-node, their corresponding platform device
does not automatically get its iommu assigned properly."
Please refer to 
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/arch/arm64/boot/dts/mediatek/mt8173.dtsi?h=v5.14.1&id=8127881f741dbbf9a1da9e9bc59133820160b217

> All this async games seem like making the driver really obfuscated,
> which will mean harder to debug and maintain.
> I am not sure we want that burden.
> 
> Even if we are all fully convinced that you absolutely need
> an async framework, then what's wrong with v4l2-async?
> 
> I would start by addressing what is wrong with the IOMMUs
> in the current design.
> 
> Thanks,
> Ezequiel
> 
> > - The list of mdp components remains hard-coded as
> > mtk_mdp_comp_dt_ids[]
> >   in mtk_mdp_core.c, and as mtk_mdp_comp_driver_dt_match[] in
> >   mtk_mdp_comp.c. This unfortunate duplication of information is
> >   addressed in a following patch in this series.
> > 
> > - The component driver calls component_add() for each device that
> > is
> >   probed.
> > 
> > - In mtk_mdp_probe (the "master" device), we scan the device tree
> > for
> >   any matching nodes against mtk_mdp_comp_dt_ids, and add component
> >   matches for them. The match criteria is a matching device node
> >   pointer.
> > 
> > - When the set of components devices that have been probed
> > corresponds
> >   with the list that is generated by the "master", the callback to
> >   mtk_mdp_master_bind() is made, which then calls the component
> > bind
> >   functions.
> > 
> > - Inside mtk_mdp_master_bind(), once all the component bind
> > functions
> >   have been called, we can then register our device to the v4l2
> >   subsystem.
> > 
> > - The call to pm_runtime_enable() in the master device is called
> > after
> >   all the components have been registered by their bind() functions
> >   called by mtk_mtp_master_bind(). As a result, the list of
> > components
> >   will not change while power management callbacks
> > mtk_mdp_suspend()/
> >   resume() are accessing the list of components.
> > 
> > Signed-off-by: Eizan Miyamoto <eizan@chromium.org>
> > Reviewed-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
> > Reviewed-by: Houlong Wei <houlong.wei@mediatek.com>
> > ---
> >
houlong.wei Sept. 5, 2021, 4:39 p.m. UTC | #3
Sorry, I fix a link error below.

On Mon, 2021-09-06 at 00:23 +0800, houlong wei wrote:
> Hi Ezequiel,
> 
> Thank you for your attention to this series of patches. I answer
> partial of your questions below.
> Regards,
> Houlong
> 
> On Sat, 2021-09-04 at 20:34 +0800, Ezequiel Garcia wrote:
> > Hi Eizan,
> > 
> > Sorry for seeing this series so late.
> > 
> > On Wed, 25 Aug 2021 at 03:35, Eizan Miyamoto <eizan@chromium.org>
> > wrote:
> > > 
> > > Broadly, this patch (1) adds a driver for various MTK MDP
> > > components to
> > > go alongside the main MTK MDP driver, and (2) hooks them all
> > > together
> > > using the component framework.
> > > 
> > > (1) Up until now, the MTK MDP driver controls 8 devices in the
> > > device
> > > tree on its own. When running tests for the hardware video
> > > decoder,
> > > we
> > > found that the iommus and LARBs were not being properly
> > > configured.
> > 
> > Why were not being properly configured? What was the problem?
> > Why not fixing that instead?
> > 
> > Does this mean the driver is currently broken and unusable?
> 
> This series of patches are supplements to another series, please
> refer
> to  
> 
https://patchwork.kernel.org/project/linux-mediatek/list/?series=515129c

Fix the link address: 
https://patchwork.kernel.org/project/linux-mediatek/list/?series=523771

> , which add device link between the mtk-iommu consumer and the mtk-
> larb 
> devices. Without that series of patches, the mtk-mdp driver can work
> well so far.
> But with that series, it seems the device link only can be
> established
> for the device which is registered as a platform driver. That's why
> Eizan adds this series of patches to make all mdp components to be
> registered as platform drivers.
> 
> > 
> > > To
> > > configure them, a driver for each be added to mtk_mdp_comp so
> > > that
> > > mtk_iommu_add_device() can (eventually) be called from
> > > dma_configure()
> > > inside really_probe().
> > > 
> > > (2) The integration into the component framework allows us to
> > > defer
> > > the
> > > registration with the v4l2 subsystem until all the MDP-related
> > > devices
> > > have been probed, so that the relevant device node does not
> > > become
> > > available until initialization of all the components is complete.
> > > 
> > > Some notes about how the component framework has been integrated:
> > > 
> > > - The driver for the rdma0 component serves double duty as the
> > > "master"
> > >   (aggregate) driver as well as a component driver. This is a
> > > non-
> > > ideal
> > >   compromise until a better solution is developed. This device is
> > >   differentiated from the rest by checking for a "mediatek,vpu"
> > > property
> > >   in the device node.
> > > 
> > 
> > As I have stated in Yunfei, I am not convinced you need an async
> > framework
> > at all. It seems all these devices could have been linked together
> > in the device tree, and then have a master device to tie them.
> > 
> > I.e. something like
> > 
> > mdp {
> >   mdp_rdma0 {
> >   }
> >   mdp_rsz0 {
> >   }
> >   mdp_rsz1 {
> >   }
> > }
> > 
> 
> The commit message of the patch below explains that " If the mdp_*
> nodes are under an mdp sub-node, their corresponding platform device
> does not automatically get its iommu assigned properly."
> Please refer to 
> 
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/arch/arm64/boot/dts/mediatek/mt8173.dtsi?h=v5.14.1&id=8127881f741dbbf9a1da9e9bc59133820160b217
> 
> > All this async games seem like making the driver really obfuscated,
> > which will mean harder to debug and maintain.
> > I am not sure we want that burden.
> > 
> > Even if we are all fully convinced that you absolutely need
> > an async framework, then what's wrong with v4l2-async?
> > 
> > I would start by addressing what is wrong with the IOMMUs
> > in the current design.
> > 
> > Thanks,
> > Ezequiel
> > 
> > > - The list of mdp components remains hard-coded as
> > > mtk_mdp_comp_dt_ids[]
> > >   in mtk_mdp_core.c, and as mtk_mdp_comp_driver_dt_match[] in
> > >   mtk_mdp_comp.c. This unfortunate duplication of information is
> > >   addressed in a following patch in this series.
> > > 
> > > - The component driver calls component_add() for each device that
> > > is
> > >   probed.
> > > 
> > > - In mtk_mdp_probe (the "master" device), we scan the device tree
> > > for
> > >   any matching nodes against mtk_mdp_comp_dt_ids, and add
> > > component
> > >   matches for them. The match criteria is a matching device node
> > >   pointer.
> > > 
> > > - When the set of components devices that have been probed
> > > corresponds
> > >   with the list that is generated by the "master", the callback
> > > to
> > >   mtk_mdp_master_bind() is made, which then calls the component
> > > bind
> > >   functions.
> > > 
> > > - Inside mtk_mdp_master_bind(), once all the component bind
> > > functions
> > >   have been called, we can then register our device to the v4l2
> > >   subsystem.
> > > 
> > > - The call to pm_runtime_enable() in the master device is called
> > > after
> > >   all the components have been registered by their bind()
> > > functions
> > >   called by mtk_mtp_master_bind(). As a result, the list of
> > > components
> > >   will not change while power management callbacks
> > > mtk_mdp_suspend()/
> > >   resume() are accessing the list of components.
> > > 
> > > Signed-off-by: Eizan Miyamoto <eizan@chromium.org>
> > > Reviewed-by: Enric Balletbo i Serra <enric.balletbo@collabora.com
> > > >
> > > Reviewed-by: Houlong Wei <houlong.wei@mediatek.com>
> > > ---
> > > 
> 
>
Hans Verkuil Dec. 7, 2021, 12:37 p.m. UTC | #4
On 9/5/21 6:23 PM, houlong wei wrote:
> Hi Ezequiel,
> 
> Thank you for your attention to this series of patches. I answer partial of your questions below.
> Regards,
> Houlong
> 
> On Sat, 2021-09-04 at 20:34 +0800, Ezequiel Garcia wrote:
>> Hi Eizan,
>>
>> Sorry for seeing this series so late.
>>
>> On Wed, 25 Aug 2021 at 03:35, Eizan Miyamoto <eizan@chromium.org>
>> wrote:
>>>
>>> Broadly, this patch (1) adds a driver for various MTK MDP
>>> components to
>>> go alongside the main MTK MDP driver, and (2) hooks them all
>>> together
>>> using the component framework.
>>>
>>> (1) Up until now, the MTK MDP driver controls 8 devices in the
>>> device
>>> tree on its own. When running tests for the hardware video decoder,
>>> we
>>> found that the iommus and LARBs were not being properly configured.
>>
>> Why were not being properly configured? What was the problem?
>> Why not fixing that instead?
>>
>> Does this mean the driver is currently broken and unusable?
> 
> This series of patches are supplements to another series, please refer
> to  
> https://patchwork.kernel.org/project/linux-mediatek/list/?series=515129c
> , which add device link between the mtk-iommu consumer and the mtk-larb 
> devices. Without that series of patches, the mtk-mdp driver can work
> well so far.
> But with that series, it seems the device link only can be established
> for the device which is registered as a platform driver. That's why
> Eizan adds this series of patches to make all mdp components to be
> registered as platform drivers.

Hold on, so this means that if that iommu device-link series is merged,
then the mtk-mdp driver breaks? I posted a PR for that iommu series, but
I've just withdrawn that PR until this issue is clarified.

Is it only mtk-mdp that is affected by this iommu device-link series, or
others as well?

Regards,

	Hans

> 
>>
>>> To
>>> configure them, a driver for each be added to mtk_mdp_comp so that
>>> mtk_iommu_add_device() can (eventually) be called from
>>> dma_configure()
>>> inside really_probe().
>>>
>>> (2) The integration into the component framework allows us to defer
>>> the
>>> registration with the v4l2 subsystem until all the MDP-related
>>> devices
>>> have been probed, so that the relevant device node does not become
>>> available until initialization of all the components is complete.
>>>
>>> Some notes about how the component framework has been integrated:
>>>
>>> - The driver for the rdma0 component serves double duty as the
>>> "master"
>>>   (aggregate) driver as well as a component driver. This is a non-
>>> ideal
>>>   compromise until a better solution is developed. This device is
>>>   differentiated from the rest by checking for a "mediatek,vpu"
>>> property
>>>   in the device node.
>>>
>>
>> As I have stated in Yunfei, I am not convinced you need an async
>> framework
>> at all. It seems all these devices could have been linked together
>> in the device tree, and then have a master device to tie them.
>>
>> I.e. something like
>>
>> mdp {
>>   mdp_rdma0 {
>>   }
>>   mdp_rsz0 {
>>   }
>>   mdp_rsz1 {
>>   }
>> }
>>
> 
> The commit message of the patch below explains that " If the mdp_*
> nodes are under an mdp sub-node, their corresponding platform device
> does not automatically get its iommu assigned properly."
> Please refer to 
> https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/arch/arm64/boot/dts/mediatek/mt8173.dtsi?h=v5.14.1&id=8127881f741dbbf9a1da9e9bc59133820160b217
> 
>> All this async games seem like making the driver really obfuscated,
>> which will mean harder to debug and maintain.
>> I am not sure we want that burden.
>>
>> Even if we are all fully convinced that you absolutely need
>> an async framework, then what's wrong with v4l2-async?
>>
>> I would start by addressing what is wrong with the IOMMUs
>> in the current design.
>>
>> Thanks,
>> Ezequiel
>>
>>> - The list of mdp components remains hard-coded as
>>> mtk_mdp_comp_dt_ids[]
>>>   in mtk_mdp_core.c, and as mtk_mdp_comp_driver_dt_match[] in
>>>   mtk_mdp_comp.c. This unfortunate duplication of information is
>>>   addressed in a following patch in this series.
>>>
>>> - The component driver calls component_add() for each device that
>>> is
>>>   probed.
>>>
>>> - In mtk_mdp_probe (the "master" device), we scan the device tree
>>> for
>>>   any matching nodes against mtk_mdp_comp_dt_ids, and add component
>>>   matches for them. The match criteria is a matching device node
>>>   pointer.
>>>
>>> - When the set of components devices that have been probed
>>> corresponds
>>>   with the list that is generated by the "master", the callback to
>>>   mtk_mdp_master_bind() is made, which then calls the component
>>> bind
>>>   functions.
>>>
>>> - Inside mtk_mdp_master_bind(), once all the component bind
>>> functions
>>>   have been called, we can then register our device to the v4l2
>>>   subsystem.
>>>
>>> - The call to pm_runtime_enable() in the master device is called
>>> after
>>>   all the components have been registered by their bind() functions
>>>   called by mtk_mtp_master_bind(). As a result, the list of
>>> components
>>>   will not change while power management callbacks
>>> mtk_mdp_suspend()/
>>>   resume() are accessing the list of components.
>>>
>>> Signed-off-by: Eizan Miyamoto <eizan@chromium.org>
>>> Reviewed-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
>>> Reviewed-by: Houlong Wei <houlong.wei@mediatek.com>
>>> ---
>>>
>
Ezequiel Garcia Dec. 7, 2021, 12:46 p.m. UTC | #5
Hi everyone,

On Tue, 7 Dec 2021 at 09:37, Hans Verkuil <hverkuil@xs4all.nl> wrote:
>
> On 9/5/21 6:23 PM, houlong wei wrote:
> > Hi Ezequiel,
> >
> > Thank you for your attention to this series of patches. I answer partial of your questions below.
> > Regards,
> > Houlong
> >
> > On Sat, 2021-09-04 at 20:34 +0800, Ezequiel Garcia wrote:
> >> Hi Eizan,
> >>
> >> Sorry for seeing this series so late.
> >>
> >> On Wed, 25 Aug 2021 at 03:35, Eizan Miyamoto <eizan@chromium.org>
> >> wrote:
> >>>
> >>> Broadly, this patch (1) adds a driver for various MTK MDP
> >>> components to
> >>> go alongside the main MTK MDP driver, and (2) hooks them all
> >>> together
> >>> using the component framework.
> >>>
> >>> (1) Up until now, the MTK MDP driver controls 8 devices in the
> >>> device
> >>> tree on its own. When running tests for the hardware video decoder,
> >>> we
> >>> found that the iommus and LARBs were not being properly configured.
> >>
> >> Why were not being properly configured? What was the problem?
> >> Why not fixing that instead?
> >>
> >> Does this mean the driver is currently broken and unusable?
> >
> > This series of patches are supplements to another series, please refer
> > to
> > https://patchwork.kernel.org/project/linux-mediatek/list/?series=515129c
> > , which add device link between the mtk-iommu consumer and the mtk-larb
> > devices. Without that series of patches, the mtk-mdp driver can work
> > well so far.
> > But with that series, it seems the device link only can be established
> > for the device which is registered as a platform driver. That's why
> > Eizan adds this series of patches to make all mdp components to be
> > registered as platform drivers.
>
> Hold on, so this means that if that iommu device-link series is merged,
> then the mtk-mdp driver breaks? I posted a PR for that iommu series, but
> I've just withdrawn that PR until this issue is clarified.
>
> Is it only mtk-mdp that is affected by this iommu device-link series, or
> others as well?
>

Like I said before, I think it's a mistake to introduce the component
framework in V4L2. This whole idea looks like a hack to me.

If we merge this, then we validate using the component framework
as a way to avoid fixing things properly.

Regards,
Ezequiel
Hans Verkuil Dec. 7, 2021, 1:02 p.m. UTC | #6
On 12/7/21 1:46 PM, Ezequiel Garcia wrote:
> Hi everyone,
> 
> On Tue, 7 Dec 2021 at 09:37, Hans Verkuil <hverkuil@xs4all.nl> wrote:
>>
>> On 9/5/21 6:23 PM, houlong wei wrote:
>>> Hi Ezequiel,
>>>
>>> Thank you for your attention to this series of patches. I answer partial of your questions below.
>>> Regards,
>>> Houlong
>>>
>>> On Sat, 2021-09-04 at 20:34 +0800, Ezequiel Garcia wrote:
>>>> Hi Eizan,
>>>>
>>>> Sorry for seeing this series so late.
>>>>
>>>> On Wed, 25 Aug 2021 at 03:35, Eizan Miyamoto <eizan@chromium.org>
>>>> wrote:
>>>>>
>>>>> Broadly, this patch (1) adds a driver for various MTK MDP
>>>>> components to
>>>>> go alongside the main MTK MDP driver, and (2) hooks them all
>>>>> together
>>>>> using the component framework.
>>>>>
>>>>> (1) Up until now, the MTK MDP driver controls 8 devices in the
>>>>> device
>>>>> tree on its own. When running tests for the hardware video decoder,
>>>>> we
>>>>> found that the iommus and LARBs were not being properly configured.
>>>>
>>>> Why were not being properly configured? What was the problem?
>>>> Why not fixing that instead?
>>>>
>>>> Does this mean the driver is currently broken and unusable?
>>>
>>> This series of patches are supplements to another series, please refer
>>> to
>>> https://patchwork.kernel.org/project/linux-mediatek/list/?series=515129c
>>> , which add device link between the mtk-iommu consumer and the mtk-larb
>>> devices. Without that series of patches, the mtk-mdp driver can work
>>> well so far.
>>> But with that series, it seems the device link only can be established
>>> for the device which is registered as a platform driver. That's why
>>> Eizan adds this series of patches to make all mdp components to be
>>> registered as platform drivers.
>>
>> Hold on, so this means that if that iommu device-link series is merged,
>> then the mtk-mdp driver breaks? I posted a PR for that iommu series, but
>> I've just withdrawn that PR until this issue is clarified.
>>
>> Is it only mtk-mdp that is affected by this iommu device-link series, or
>> others as well?
>>
> 
> Like I said before, I think it's a mistake to introduce the component
> framework in V4L2. This whole idea looks like a hack to me.
> 
> If we merge this, then we validate using the component framework
> as a way to avoid fixing things properly.

I agree with Ezequiel.

	Hans
Yong Wu (吴勇) Dec. 8, 2021, 2 a.m. UTC | #7
On Mon, 2021-09-06 at 00:23 +0800, houlong wei wrote:
> Hi Ezequiel,
> 
> Thank you for your attention to this series of patches. I answer
> partial of your questions below.
> Regards,
> Houlong
> 
> On Sat, 2021-09-04 at 20:34 +0800, Ezequiel Garcia wrote:
> > Hi Eizan,
> > 
> > Sorry for seeing this series so late.
> > 
> > On Wed, 25 Aug 2021 at 03:35, Eizan Miyamoto <eizan@chromium.org>
> > wrote:
> > > 
> > > Broadly, this patch (1) adds a driver for various MTK MDP
> > > components to
> > > go alongside the main MTK MDP driver, and (2) hooks them all
> > > together
> > > using the component framework.
> > > 
> > > (1) Up until now, the MTK MDP driver controls 8 devices in the
> > > device
> > > tree on its own. When running tests for the hardware video
> > > decoder,
> > > we
> > > found that the iommus and LARBs were not being properly
> > > configured.
> > 
> > Why were not being properly configured? What was the problem?
> > Why not fixing that instead?
> > 
> > Does this mean the driver is currently broken and unusable?
> 
> This series of patches are supplements to another series, please
> refer
> to  
> 
https://patchwork.kernel.org/project/linux-mediatek/list/?series=515129c
> , which add device link between the mtk-iommu consumer and the mtk-
> larb 
> devices. Without that series of patches, the mtk-mdp driver can work
> well so far.
> But with that series, it seems the device link only can be
> established
> for the device which is registered as a platform driver. That's why
> Eizan adds this series of patches to make all mdp components to be
> registered as platform drivers.

The mt8173 mdp has several devices:
   mediatek,mt8173-mdp-rdma, mediatek,mt8173-mdp  
   mediatek,mt8173-mdp-rsz
   mediatek,mt8173-mdp-wdma
   mediatek,mt8173-mdp-wrot

Except the first one, the last three devices are not the standard
platform devices. Thus, they should not be the iommu consumer devices.

Question 1: The last three device don't work actually in mt8173 chrome,
right? or they access continuous buffers?

Question 2: The IOMMU device-link patchset just replaces the pm runtime
interfaces. It don't improve the mdp flow, also should not introduce
regression. thus, my v8 don't rebase this mdp patches. Does the iommu
patchset introduce regression for mdp?

@Eizan, @houlong, Could you help confirm this?
Thanks.

> 
> > 
> > > To
> > > configure them, a driver for each be added to mtk_mdp_comp so
> > > that
> > > mtk_iommu_add_device() can (eventually) be called from
> > > dma_configure()
> > > inside really_probe().
> > > 
> > > (2) The integration into the component framework allows us to
> > > defer
> > > the
> > > registration with the v4l2 subsystem until all the MDP-related
> > > devices
> > > have been probed, so that the relevant device node does not
> > > become
> > > available until initialization of all the components is complete.
> > > 
> > > Some notes about how the component framework has been integrated:
> > > 
> > > - The driver for the rdma0 component serves double duty as the
> > > "master"
> > >   (aggregate) driver as well as a component driver. This is a
> > > non-
> > > ideal
> > >   compromise until a better solution is developed. This device is
> > >   differentiated from the rest by checking for a "mediatek,vpu"
> > > property
> > >   in the device node.
> > > 
> > 
> > As I have stated in Yunfei, I am not convinced you need an async
> > framework
> > at all. It seems all these devices could have been linked together
> > in the device tree, and then have a master device to tie them.
> > 
> > I.e. something like
> > 
> > mdp {
> >   mdp_rdma0 {
> >   }
> >   mdp_rsz0 {
> >   }
> >   mdp_rsz1 {
> >   }
> > }
> > 
> 
> The commit message of the patch below explains that " If the mdp_*
> nodes are under an mdp sub-node, their corresponding platform device
> does not automatically get its iommu assigned properly."
> Please refer to 
> 
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/arch/arm64/boot/dts/mediatek/mt8173.dtsi?h=v5.14.1&id=8127881f741dbbf9a1da9e9bc59133820160b217
> 
[snip]
diff mbox series

Patch

diff --git a/drivers/media/platform/mtk-mdp/mtk_mdp_comp.c b/drivers/media/platform/mtk-mdp/mtk_mdp_comp.c
index 76e295c8d9bc..7a0e3acffab9 100644
--- a/drivers/media/platform/mtk-mdp/mtk_mdp_comp.c
+++ b/drivers/media/platform/mtk-mdp/mtk_mdp_comp.c
@@ -5,13 +5,50 @@ 
  */
 
 #include <linux/clk.h>
+#include <linux/component.h>
 #include <linux/device.h>
 #include <linux/of.h>
+#include <linux/module.h>
 #include <linux/of_address.h>
 #include <linux/of_platform.h>
 #include <soc/mediatek/smi.h>
 
 #include "mtk_mdp_comp.h"
+#include "mtk_mdp_core.h"
+
+/**
+ * enum mtk_mdp_comp_type - the MDP component
+ * @MTK_MDP_RDMA:		Read DMA
+ * @MTK_MDP_RSZ:		Reszer
+ * @MTK_MDP_WDMA:		Write DMA
+ * @MTK_MDP_WROT:		Write DMA with rotation
+ * @MTK_MDP_COMP_TYPE_MAX:	Placeholder for num elems in this enum
+ */
+enum mtk_mdp_comp_type {
+	MTK_MDP_RDMA,
+	MTK_MDP_RSZ,
+	MTK_MDP_WDMA,
+	MTK_MDP_WROT,
+	MTK_MDP_COMP_TYPE_MAX,
+};
+
+static const struct of_device_id mtk_mdp_comp_driver_dt_match[] = {
+	{
+		.compatible = "mediatek,mt8173-mdp-rdma",
+		.data = (void *)MTK_MDP_RDMA
+	}, {
+		.compatible = "mediatek,mt8173-mdp-rsz",
+		.data = (void *)MTK_MDP_RSZ
+	}, {
+		.compatible = "mediatek,mt8173-mdp-wdma",
+		.data = (void *)MTK_MDP_WDMA
+	}, {
+		.compatible = "mediatek,mt8173-mdp-wrot",
+		.data = (void *)MTK_MDP_WROT
+	},
+	{ }
+};
+MODULE_DEVICE_TABLE(of, mtk_mdp_comp_driver_dt_match);
 
 int mtk_mdp_comp_clock_on(struct device *dev, struct mtk_mdp_comp *comp)
 {
@@ -20,9 +57,7 @@  int mtk_mdp_comp_clock_on(struct device *dev, struct mtk_mdp_comp *comp)
 	if (comp->larb_dev) {
 		err = mtk_smi_larb_get(comp->larb_dev);
 		if (err)
-			dev_err(dev,
-				"failed to get larb, err %d. type:%d\n",
-				err, comp->type);
+			dev_err(dev, "failed to get larb, err %d.\n", err);
 	}
 
 	for (i = 0; i < ARRAY_SIZE(comp->clk); i++) {
@@ -62,17 +97,41 @@  void mtk_mdp_comp_clock_off(struct device *dev, struct mtk_mdp_comp *comp)
 		mtk_smi_larb_put(comp->larb_dev);
 }
 
-int mtk_mdp_comp_init(struct device *dev, struct device_node *node,
-		      struct mtk_mdp_comp *comp,
-		      enum mtk_mdp_comp_type comp_type)
+static int mtk_mdp_comp_bind(struct device *dev, struct device *master, void *data)
+{
+	struct mtk_mdp_comp *comp = dev_get_drvdata(dev);
+	struct mtk_mdp_dev *mdp = data;
+
+	mtk_mdp_register_component(mdp, comp);
+
+	return 0;
+}
+
+static void mtk_mdp_comp_unbind(struct device *dev, struct device *master,
+			   void *data)
+{
+	struct mtk_mdp_comp *comp = dev_get_drvdata(dev);
+	struct mtk_mdp_dev *mdp = data;
+
+	mtk_mdp_unregister_component(mdp, comp);
+}
+
+static const struct component_ops mtk_mdp_component_ops = {
+	.bind   = mtk_mdp_comp_bind,
+	.unbind = mtk_mdp_comp_unbind,
+};
+
+int mtk_mdp_comp_init(struct mtk_mdp_comp *comp, struct device *dev)
 {
 	struct device_node *larb_node;
 	struct platform_device *larb_pdev;
 	int ret;
 	int i;
+	struct device_node *node = dev->of_node;
+	enum mtk_mdp_comp_type comp_type =
+		 (enum mtk_mdp_comp_type)of_device_get_match_data(dev);
 
-	comp->dev_node = of_node_get(node);
-	comp->type = comp_type;
+	INIT_LIST_HEAD(&comp->node);
 
 	for (i = 0; i < ARRAY_SIZE(comp->clk); i++) {
 		comp->clk[i] = of_clk_get(node, i);
@@ -80,19 +139,17 @@  int mtk_mdp_comp_init(struct device *dev, struct device_node *node,
 			if (PTR_ERR(comp->clk[i]) != -EPROBE_DEFER)
 				dev_err(dev, "Failed to get clock\n");
 			ret = PTR_ERR(comp->clk[i]);
-			goto put_dev;
+			goto err;
 		}
 
 		/* Only RDMA needs two clocks */
-		if (comp->type != MTK_MDP_RDMA)
+		if (comp_type != MTK_MDP_RDMA)
 			break;
 	}
 
 	/* Only DMA capable components need the LARB property */
 	comp->larb_dev = NULL;
-	if (comp->type != MTK_MDP_RDMA &&
-	    comp->type != MTK_MDP_WDMA &&
-	    comp->type != MTK_MDP_WROT)
+	if (comp_type != MTK_MDP_RDMA && comp_type != MTK_MDP_WDMA && comp_type != MTK_MDP_WROT)
 		return 0;
 
 	larb_node = of_parse_phandle(node, "mediatek,larb", 0);
@@ -100,7 +157,7 @@  int mtk_mdp_comp_init(struct device *dev, struct device_node *node,
 		dev_err(dev,
 			"Missing mediadek,larb phandle in %pOF node\n", node);
 		ret = -EINVAL;
-		goto put_dev;
+		goto err;
 	}
 
 	larb_pdev = of_find_device_by_node(larb_node);
@@ -108,7 +165,7 @@  int mtk_mdp_comp_init(struct device *dev, struct device_node *node,
 		dev_warn(dev, "Waiting for larb device %pOF\n", larb_node);
 		of_node_put(larb_node);
 		ret = -EPROBE_DEFER;
-		goto put_dev;
+		goto err;
 	}
 	of_node_put(larb_node);
 
@@ -116,13 +173,59 @@  int mtk_mdp_comp_init(struct device *dev, struct device_node *node,
 
 	return 0;
 
-put_dev:
-	of_node_put(comp->dev_node);
-
+err:
 	return ret;
 }
 
-void mtk_mdp_comp_deinit(struct device *dev, struct mtk_mdp_comp *comp)
+static int mtk_mdp_comp_probe(struct platform_device *pdev)
 {
-	of_node_put(comp->dev_node);
+	struct device *dev = &pdev->dev;
+	struct device_node *vpu_node;
+	int status;
+	struct mtk_mdp_comp *comp;
+
+	vpu_node = of_parse_phandle(dev->of_node, "mediatek,vpu", 0);
+	if (vpu_node) {
+		of_node_put(vpu_node);
+		/*
+		 * The device tree node with a mediatek,vpu property is deemed
+		 * the MDP "master" device, we don't want to add a component
+		 * for it in this function because the initialization for the
+		 * master is done elsewhere.
+		 */
+		dev_info(dev, "vpu node found, not probing\n");
+		return -ENODEV;
+	}
+
+	comp = devm_kzalloc(dev, sizeof(*comp), GFP_KERNEL);
+	if (!comp)
+		return -ENOMEM;
+
+	status = mtk_mdp_comp_init(comp, dev);
+	if (status) {
+		dev_err(dev, "Failed to initialize component: %d\n", status);
+		return status;
+	}
+
+	dev_set_drvdata(dev, comp);
+
+	return component_add(dev, &mtk_mdp_component_ops);
 }
+
+static int mtk_mdp_comp_remove(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+
+	component_del(dev, &mtk_mdp_component_ops);
+	return 0;
+}
+
+struct platform_driver mtk_mdp_component_driver = {
+	.probe          = mtk_mdp_comp_probe,
+	.remove         = mtk_mdp_comp_remove,
+	.driver         = {
+		.name   = "mediatek-mdp-comp",
+		.owner  = THIS_MODULE,
+		.of_match_table = mtk_mdp_comp_driver_dt_match,
+	},
+};
diff --git a/drivers/media/platform/mtk-mdp/mtk_mdp_comp.h b/drivers/media/platform/mtk-mdp/mtk_mdp_comp.h
index 92ab5249bcad..df5fc4c94f90 100644
--- a/drivers/media/platform/mtk-mdp/mtk_mdp_comp.h
+++ b/drivers/media/platform/mtk-mdp/mtk_mdp_comp.h
@@ -7,42 +7,23 @@ 
 #ifndef __MTK_MDP_COMP_H__
 #define __MTK_MDP_COMP_H__
 
-/**
- * enum mtk_mdp_comp_type - the MDP component
- * @MTK_MDP_RDMA:	Read DMA
- * @MTK_MDP_RSZ:	Riszer
- * @MTK_MDP_WDMA:	Write DMA
- * @MTK_MDP_WROT:	Write DMA with rotation
- */
-enum mtk_mdp_comp_type {
-	MTK_MDP_RDMA,
-	MTK_MDP_RSZ,
-	MTK_MDP_WDMA,
-	MTK_MDP_WROT,
-};
-
 /**
  * struct mtk_mdp_comp - the MDP's function component data
  * @node:	list node to track sibing MDP components
- * @dev_node:	component device node
  * @clk:	clocks required for component
  * @larb_dev:	SMI device required for component
- * @type:	component type
  */
 struct mtk_mdp_comp {
 	struct list_head	node;
-	struct device_node	*dev_node;
 	struct clk		*clk[2];
 	struct device		*larb_dev;
-	enum mtk_mdp_comp_type	type;
 };
 
-int mtk_mdp_comp_init(struct device *dev, struct device_node *node,
-		      struct mtk_mdp_comp *comp,
-		      enum mtk_mdp_comp_type comp_type);
-void mtk_mdp_comp_deinit(struct device *dev, struct mtk_mdp_comp *comp);
+int mtk_mdp_comp_init(struct mtk_mdp_comp *comp, struct device *dev);
+
 int mtk_mdp_comp_clock_on(struct device *dev, struct mtk_mdp_comp *comp);
 void mtk_mdp_comp_clock_off(struct device *dev, struct mtk_mdp_comp *comp);
 
+extern struct platform_driver mtk_mdp_component_driver;
 
 #endif /* __MTK_MDP_COMP_H__ */
diff --git a/drivers/media/platform/mtk-mdp/mtk_mdp_core.c b/drivers/media/platform/mtk-mdp/mtk_mdp_core.c
index 412bbec0f735..b813a822439a 100644
--- a/drivers/media/platform/mtk-mdp/mtk_mdp_core.c
+++ b/drivers/media/platform/mtk-mdp/mtk_mdp_core.c
@@ -6,6 +6,7 @@ 
  */
 
 #include <linux/clk.h>
+#include <linux/component.h>
 #include <linux/device.h>
 #include <linux/errno.h>
 #include <linux/interrupt.h>
@@ -19,6 +20,7 @@ 
 #include <linux/workqueue.h>
 #include <soc/mediatek/smi.h>
 
+#include "mtk_mdp_comp.h"
 #include "mtk_mdp_core.h"
 #include "mtk_mdp_m2m.h"
 #include "mtk_vpu.h"
@@ -32,16 +34,12 @@  module_param(mtk_mdp_dbg_level, int, 0644);
 static const struct of_device_id mtk_mdp_comp_dt_ids[] = {
 	{
 		.compatible = "mediatek,mt8173-mdp-rdma",
-		.data = (void *)MTK_MDP_RDMA
 	}, {
 		.compatible = "mediatek,mt8173-mdp-rsz",
-		.data = (void *)MTK_MDP_RSZ
 	}, {
 		.compatible = "mediatek,mt8173-mdp-wdma",
-		.data = (void *)MTK_MDP_WDMA
 	}, {
 		.compatible = "mediatek,mt8173-mdp-wrot",
-		.data = (void *)MTK_MDP_WROT
 	},
 	{ },
 };
@@ -106,6 +104,63 @@  static void mtk_mdp_reset_handler(void *priv)
 	queue_work(mdp->wdt_wq, &mdp->wdt_work);
 }
 
+static int compare_of(struct device *dev, void *data)
+{
+	return dev->of_node == data;
+}
+
+static void release_of(struct device *dev, void *data)
+{
+	of_node_put(data);
+}
+
+static int mtk_mdp_master_bind(struct device *dev)
+{
+	int status;
+	struct mtk_mdp_dev *mdp = dev_get_drvdata(dev);
+
+	mtk_mdp_register_component(mdp, &mdp->comp_self);
+
+	status = component_bind_all(dev, mdp);
+	if (status) {
+		dev_err(dev, "Failed to bind all components: %d\n", status);
+		goto err_component_bind_all;
+	}
+
+	status = mtk_mdp_register_m2m_device(mdp);
+	if (status) {
+		dev_err(dev, "Failed to register m2m device: %d\n", status);
+		goto err_mtk_mdp_register_m2m_device;
+	}
+
+	pm_runtime_enable(dev);
+
+	return 0;
+
+err_mtk_mdp_register_m2m_device:
+	component_unbind_all(dev, mdp);
+
+err_component_bind_all:
+	mtk_mdp_unregister_component(mdp, &mdp->comp_self);
+
+	return status;
+}
+
+static void mtk_mdp_master_unbind(struct device *dev)
+{
+	struct mtk_mdp_dev *mdp = dev_get_drvdata(dev);
+
+	pm_runtime_disable(dev);
+	mtk_mdp_unregister_m2m_device(mdp);
+	component_unbind_all(dev, mdp);
+	mtk_mdp_unregister_component(mdp, &mdp->comp_self);
+}
+
+static const struct component_master_ops mtk_mdp_com_ops = {
+	.bind		= mtk_mdp_master_bind,
+	.unbind		= mtk_mdp_master_unbind,
+};
+
 void mtk_mdp_register_component(struct mtk_mdp_dev *mdp,
 				struct mtk_mdp_comp *comp)
 {
@@ -123,8 +178,8 @@  static int mtk_mdp_probe(struct platform_device *pdev)
 	struct mtk_mdp_dev *mdp;
 	struct device *dev = &pdev->dev;
 	struct device_node *node, *parent;
-	struct mtk_mdp_comp *comp, *comp_temp;
-	int ret = 0;
+	int i, ret = 0;
+	struct component_match *match = NULL;
 
 	mdp = devm_kzalloc(dev, sizeof(*mdp), GFP_KERNEL);
 	if (!mdp)
@@ -149,36 +204,43 @@  static int mtk_mdp_probe(struct platform_device *pdev)
 	}
 
 	/* Iterate over sibling MDP function blocks */
+	i = 0;
 	for_each_child_of_node(parent, node) {
-		const struct of_device_id *of_id;
-		enum mtk_mdp_comp_type comp_type;
+		struct platform_device *pdev;
 
-		of_id = of_match_node(mtk_mdp_comp_dt_ids, node);
-		if (!of_id)
+		if (!of_match_node(mtk_mdp_comp_dt_ids, node))
 			continue;
 
-		if (!of_device_is_available(node)) {
-			dev_err(dev, "Skipping disabled component %pOF\n",
-				node);
+		if (!of_device_is_available(node))
 			continue;
-		}
-
-		comp_type = (enum mtk_mdp_comp_type)of_id->data;
 
-		comp = devm_kzalloc(dev, sizeof(*comp), GFP_KERNEL);
-		if (!comp) {
-			ret = -ENOMEM;
-			of_node_put(node);
-			goto err_comp;
+		pdev = of_find_device_by_node(node);
+		if (!pdev) {
+			dev_warn(dev, "Unable to find comp device %s\n",
+				 node->full_name);
+			continue;
 		}
 
-		ret = mtk_mdp_comp_init(dev, node, comp, comp_type);
-		if (ret) {
-			of_node_put(node);
-			goto err_comp;
+		/*
+		 * Do not add a match for my own (rdma0) device node.
+		 * I will be managing it directly instead using comp_self.
+		 */
+		if (&pdev->dev != dev) {
+			dev_dbg(dev, "adding match %d for: %pOF\n", i++, node);
+			component_match_add_release(dev, &match, release_of,
+						    compare_of,
+						    of_node_get(node));
 		}
+	}
 
-		mtk_mdp_register_component(mdp, comp);
+	/*
+	 * Create a component for myself so that clocks can be toggled in
+	 * clock_on().
+	 */
+	ret = mtk_mdp_comp_init(&mdp->comp_self, dev);
+	if (ret) {
+		dev_err(dev, "Failed to initialize component\n");
+		goto err_comp;
 	}
 
 	mdp->job_wq = create_singlethread_workqueue(MTK_MDP_MODULE_NAME);
@@ -203,18 +265,12 @@  static int mtk_mdp_probe(struct platform_device *pdev)
 		goto err_dev_register;
 	}
 
-	ret = mtk_mdp_register_m2m_device(mdp);
-	if (ret) {
-		v4l2_err(&mdp->v4l2_dev, "Failed to init mem2mem device\n");
-		goto err_m2m_register;
-	}
-
 	mdp->vpu_dev = vpu_get_plat_device(pdev);
 	ret = vpu_wdt_reg_handler(mdp->vpu_dev, mtk_mdp_reset_handler, mdp,
 				  VPU_RST_MDP);
 	if (ret) {
 		dev_err(&pdev->dev, "Failed to register reset handler\n");
-		goto err_m2m_register;
+		goto err_wdt_reg;
 	}
 
 	platform_set_drvdata(pdev, mdp);
@@ -222,15 +278,25 @@  static int mtk_mdp_probe(struct platform_device *pdev)
 	ret = vb2_dma_contig_set_max_seg_size(&pdev->dev, DMA_BIT_MASK(32));
 	if (ret) {
 		dev_err(&pdev->dev, "Failed to set vb2 dma mag seg size\n");
-		goto err_m2m_register;
+		goto err_set_max_seg_size;
+	}
+
+	ret = component_master_add_with_match(dev, &mtk_mdp_com_ops, match);
+	if (ret) {
+		dev_err(dev, "Component master add failed\n");
+		goto err_component_master_add;
 	}
 
-	pm_runtime_enable(dev);
 	dev_dbg(dev, "mdp-%d registered successfully\n", mdp->id);
 
 	return 0;
 
-err_m2m_register:
+err_component_master_add:
+	vb2_dma_contig_clear_max_seg_size(&pdev->dev);
+
+err_set_max_seg_size:
+
+err_wdt_reg:
 	v4l2_device_unregister(&mdp->v4l2_dev);
 
 err_dev_register:
@@ -242,11 +308,6 @@  static int mtk_mdp_probe(struct platform_device *pdev)
 err_alloc_job_wq:
 
 err_comp:
-	list_for_each_entry_safe(comp, comp_temp, &mdp->comp_list, node) {
-		mtk_mdp_unregister_component(mdp, comp);
-		mtk_mdp_comp_deinit(dev, comp);
-	}
-
 	dev_dbg(dev, "err %d\n", ret);
 	return ret;
 }
@@ -254,11 +315,10 @@  static int mtk_mdp_probe(struct platform_device *pdev)
 static int mtk_mdp_remove(struct platform_device *pdev)
 {
 	struct mtk_mdp_dev *mdp = platform_get_drvdata(pdev);
-	struct mtk_mdp_comp *comp, *comp_temp;
 
-	pm_runtime_disable(&pdev->dev);
+	component_master_del(&pdev->dev, &mtk_mdp_com_ops);
+
 	vb2_dma_contig_clear_max_seg_size(&pdev->dev);
-	mtk_mdp_unregister_m2m_device(mdp);
 	v4l2_device_unregister(&mdp->v4l2_dev);
 
 	flush_workqueue(mdp->wdt_wq);
@@ -267,10 +327,8 @@  static int mtk_mdp_remove(struct platform_device *pdev)
 	flush_workqueue(mdp->job_wq);
 	destroy_workqueue(mdp->job_wq);
 
-	list_for_each_entry_safe(comp, comp_temp, &mdp->comp_list, node) {
-		mtk_mdp_unregister_component(mdp, comp);
-		mtk_mdp_comp_deinit(&pdev->dev, comp);
-	}
+	if (!list_empty(&mdp->comp_list))
+		dev_warn(&pdev->dev, "not all components removed\n");
 
 	dev_dbg(&pdev->dev, "%s driver unloaded\n", pdev->name);
 	return 0;
@@ -323,7 +381,25 @@  static struct platform_driver mtk_mdp_driver = {
 	}
 };
 
-module_platform_driver(mtk_mdp_driver);
+static struct platform_driver * const mtk_mdp_drivers[] = {
+	&mtk_mdp_driver,
+	&mtk_mdp_component_driver,
+};
+
+static int __init mtk_mdp_init(void)
+{
+	return platform_register_drivers(mtk_mdp_drivers,
+					 ARRAY_SIZE(mtk_mdp_drivers));
+}
+
+static void __exit mtk_mdp_exit(void)
+{
+	platform_unregister_drivers(mtk_mdp_drivers,
+				    ARRAY_SIZE(mtk_mdp_drivers));
+}
+
+module_init(mtk_mdp_init);
+module_exit(mtk_mdp_exit);
 
 MODULE_AUTHOR("Houlong Wei <houlong.wei@mediatek.com>");
 MODULE_DESCRIPTION("Mediatek image processor driver");
diff --git a/drivers/media/platform/mtk-mdp/mtk_mdp_core.h b/drivers/media/platform/mtk-mdp/mtk_mdp_core.h
index a6e6dc36307b..8a52539b15d4 100644
--- a/drivers/media/platform/mtk-mdp/mtk_mdp_core.h
+++ b/drivers/media/platform/mtk-mdp/mtk_mdp_core.h
@@ -155,6 +155,7 @@  struct mtk_mdp_dev {
 	struct mtk_mdp_variant		*variant;
 	u16				id;
 	struct list_head		comp_list;
+	struct mtk_mdp_comp		comp_self;
 	struct v4l2_m2m_dev		*m2m_dev;
 	struct list_head		ctx_list;
 	struct video_device		*vdev;