2021-11-02 18:21:42

by Tim Harvey

[permalink] [raw]
Subject: Re: [PATCH] arm64: dts: imx8m: add syscon node for display_blk_ctrl module regs

On Tue, Nov 2, 2021 at 9:08 AM Adam Ford <[email protected]> wrote:
>
> On Tue, Nov 2, 2021 at 10:47 AM Tommaso Merciai <[email protected]> wrote:
> >
> > On Tue, Nov 02, 2021 at 07:23:06AM -0500, Adam Ford wrote:
> > > The upcoming 5.16 kernel will have a new blk-ctrl driver which will
> > > work in conjunction with the GPC. You can see it in linux-next [1],
> > > and I would expect it to be present in 5.16-rc1 once the merge is
> > > done.
> > >
> > > In [1], Look for :
> > >
> > > disp_blk_ctrl: blk-ctrl@32e28000 {
> > > compatible = "fsl,imx8mm-disp-blk-ctrl", "syscon";
> > >
> > > It creates a bunch of virtual power domains which are effectively the
> > > resets for the VPU, CSI, DSI, and LCDIF [2].
> > >
> > > Basically, to pull the respective device out of reset, you'd reference
> > > them using power-domains. I have an RFC patch for the CSI located [3]
> > > which should bring the GPC power domain up, then take the CSI bridge
> > > and MIPI_CSI out of reset using the blk-ctrl. A few of us are still
> > > investigating the CSI bridge and mipi_csi drivers to determine what's
> > > going wrong, but inside that patch, you'll see that we reference
> > > "power-domains = <&disp_blk_ctrl IMX8MM_DISPBLK_PD_CSI_BRIDGE>;" and
> > > "power-domains = <&disp_blk_ctrl IMX8MM_DISPBLK_PD_MIPI_CSI>;" which
> > > are part of the new blk-ctrl driver @32e2800. Other peripherals like
> > > LCD, DSI, and the VPU's should be able to reference their respective
> > > power domains to activate the corresponding resets after enabling the
> > > proper GPC power domain.
> >
> > Hi Adam,
> > Then is all done right. Using this this new driver/dts node eLCDIF/mipi_dsi
> > module are out of reset. Thanks for the tips. I'm trying to get eLCDIF/mipi_dsi
> > work on mainline. I try to get work
> >
> > - eLCDIF using: mxsfb_drv.c
> > - mipi_dsi using: nwl-dsi.c
> >
> > What do you think about? You think that can be a good way ( taking
> > imx8mq as reference )?
>
> The DSI controller for the 8MM and 8MN is not the same as the DSI
> controller on the 8MQ, but the LCDIF controller should be compatible.
>
> There have been several attempts to support the 8MM DSI, but none of
> them have been accepted for various reasons.
>
> The latest was found here [1], but others [2] and [3] , when used
> together, do something similar.
>
> If memory serves, the main issue has to do with the fact that the DSIM
> controller in the 8MM and 8MN is also present in one of the Samsung
> processors, and the goal is to rework those drivers so we'll have one
> driver that supports both Samsung progressors and NXP instead of
> having two duplicate drivers doing the same thing. When whatever
> driver is chosen is ready, it'll be likely that the LCDIF will use
> power-domains = <&disp_blk_ctrl MX8MM_DISPBLK_PD_LCDIF> and the DSI
> node will use power-domains = <&disp_blk_ctrl
> IMX8MM_DISPBLK_PD_MIPI_DSI> to pull their respective devices out of
> reset and enable the gpc.
>
>
> [1] - https://patchwork.kernel.org/project/linux-arm-kernel/list/?series=510489&archive=both&state=*
> [2] - https://patchwork.kernel.org/project/dri-devel/list/?series=347439&archive=both&state=*
> [3] - https://patchwork.kernel.org/project/linux-arm-kernel/list/?series=359775&archive=both&state=*

Adam,

Thanks for the good summary... I was just putting this info together
as well. I'm also interested to see if anyone has made progress on
IMX8MM MIPI DSI display. Now that blk-ctl and most of the dt bindings
have been merged for 5.16 I think we are just down to the drm/exynos
driver issue.

Added Frieder, Jagan, and Michael to the thread.

Best Regards,

Tim


2021-11-02 22:44:10

by Tommaso Merciai

[permalink] [raw]
Subject: Re: [PATCH] arm64: dts: imx8m: add syscon node for display_blk_ctrl module regs

On Tue, Nov 02, 2021 at 11:18:53AM -0700, Tim Harvey wrote:
> On Tue, Nov 2, 2021 at 9:08 AM Adam Ford <[email protected]> wrote:
> >
> > On Tue, Nov 2, 2021 at 10:47 AM Tommaso Merciai <[email protected]> wrote:
> > >
> > > On Tue, Nov 02, 2021 at 07:23:06AM -0500, Adam Ford wrote:
> > > > The upcoming 5.16 kernel will have a new blk-ctrl driver which will
> > > > work in conjunction with the GPC. You can see it in linux-next [1],
> > > > and I would expect it to be present in 5.16-rc1 once the merge is
> > > > done.
> > > >
> > > > In [1], Look for :
> > > >
> > > > disp_blk_ctrl: blk-ctrl@32e28000 {
> > > > compatible = "fsl,imx8mm-disp-blk-ctrl", "syscon";
> > > >
> > > > It creates a bunch of virtual power domains which are effectively the
> > > > resets for the VPU, CSI, DSI, and LCDIF [2].
> > > >
> > > > Basically, to pull the respective device out of reset, you'd reference
> > > > them using power-domains. I have an RFC patch for the CSI located [3]
> > > > which should bring the GPC power domain up, then take the CSI bridge
> > > > and MIPI_CSI out of reset using the blk-ctrl. A few of us are still
> > > > investigating the CSI bridge and mipi_csi drivers to determine what's
> > > > going wrong, but inside that patch, you'll see that we reference
> > > > "power-domains = <&disp_blk_ctrl IMX8MM_DISPBLK_PD_CSI_BRIDGE>;" and
> > > > "power-domains = <&disp_blk_ctrl IMX8MM_DISPBLK_PD_MIPI_CSI>;" which
> > > > are part of the new blk-ctrl driver @32e2800. Other peripherals like
> > > > LCD, DSI, and the VPU's should be able to reference their respective
> > > > power domains to activate the corresponding resets after enabling the
> > > > proper GPC power domain.
> > >
> > > Hi Adam,
> > > Then is all done right. Using this this new driver/dts node eLCDIF/mipi_dsi
> > > module are out of reset. Thanks for the tips. I'm trying to get eLCDIF/mipi_dsi
> > > work on mainline. I try to get work
> > >
> > > - eLCDIF using: mxsfb_drv.c
> > > - mipi_dsi using: nwl-dsi.c
> > >
> > > What do you think about? You think that can be a good way ( taking
> > > imx8mq as reference )?
> >
> > The DSI controller for the 8MM and 8MN is not the same as the DSI
> > controller on the 8MQ, but the LCDIF controller should be compatible.
> >
> > There have been several attempts to support the 8MM DSI, but none of
> > them have been accepted for various reasons.
> >
> > The latest was found here [1], but others [2] and [3] , when used
> > together, do something similar.
> >
> > If memory serves, the main issue has to do with the fact that the DSIM
> > controller in the 8MM and 8MN is also present in one of the Samsung
> > processors, and the goal is to rework those drivers so we'll have one
> > driver that supports both Samsung progressors and NXP instead of
> > having two duplicate drivers doing the same thing. When whatever
> > driver is chosen is ready, it'll be likely that the LCDIF will use
> > power-domains = <&disp_blk_ctrl MX8MM_DISPBLK_PD_LCDIF> and the DSI
> > node will use power-domains = <&disp_blk_ctrl
> > IMX8MM_DISPBLK_PD_MIPI_DSI> to pull their respective devices out of
> > reset and enable the gpc.
> >
> >
> > [1] - https://patchwork.kernel.org/project/linux-arm-kernel/list/?series=510489&archive=both&state=*
> > [2] - https://patchwork.kernel.org/project/dri-devel/list/?series=347439&archive=both&state=*
> > [3] - https://patchwork.kernel.org/project/linux-arm-kernel/list/?series=359775&archive=both&state=*
>
> Adam,
>
> Thanks for the good summary... I was just putting this info together
> as well. I'm also interested to see if anyone has made progress on
> IMX8MM MIPI DSI display. Now that blk-ctl and most of the dt bindings
> have been merged for 5.16 I think we are just down to the drm/exynos
> driver issue.
>
> Added Frieder, Jagan, and Michael to the thread.
>
> Best Regards,
>
> Tim

Hi Adam,
Thanks again for your explanation. Then, now the main goal is refactoring
exynos dsim driver as bridge driver in order to support both imx8mm and exynos
SOC. I'll investigate on it.

Regards,
Thanks