Re: [PATCH] clk: mediatek: mt8186: Change I2C 4/5/6 ap clocks parent to infra

[Yu-chang Lee (李禹璋)]

On Wed, 2023-10-25 at 13:29 +0200, AngeloGioacchino Del Regno wrote:
> Il 25/10/23 11:50, Yu-chang Lee (李禹璋) ha scritto:
> > On Tue, 2023-10-24 at 17:20 +0800, Chen-Yu Tsai wrote:
> > >   	
> > > External email : Please do not click links or open attachments
> > > until
> > > you have verified the sender or the content.
> > >   On Tue, Oct 24, 2023 at 3:47 PM Yu-chang Lee (李禹璋)
> > > <Yu-chang.Lee@xxxxxxxxxxxx> wrote:
> > > > 
> > > > On Tue, 2023-10-24 at 10:58 +0800, Chen-Yu Tsai wrote:
> > > > > On Tue, Oct 24, 2023 at 10:52 AM Stephen Boyd <
> > > > > sboyd@xxxxxxxxxx>
> > > > > wrote:
> > > > > > 
> > > > > > Quoting Chen-Yu Tsai (2023-10-19 22:06:35)
> > > > > > > On Thu, Oct 19, 2023 at 8:49 PM AngeloGioacchino Del
> > > > > > > Regno
> > > > > > > <angelogioacchino.delregno@xxxxxxxxxxxxx> wrote:
> > > > > > > > 
> > > > > > > > Fix the parenting of clocks
> > > > > > > > imp_iic_wrap_ap_clock_i2c{4-6},
> > > 
> > > as
> > > > > > > > those
> > > > > > > > are effectively parented to infra_ao_i2c{4-6} and not
> > > > > > > > to
> > > 
> > > the
> > > > > > > > I2C_AP.
> > > > > > > > This permits the correct (and full) enablement and
> > > 
> > > disablement
> > > > > > > > of the
> > > > > > > > I2C4, I2C5 and I2C6 bus clocks, satisfying the whole
> > > > > > > > clock
> > > 
> > > tree
> > > > > > > > of
> > > > > > > > those.
> > > > > > > > 
> > > > > > > > As an example, when requesting to enable
> > > > > > > > imp_iic_wrap_ap_clock_i2c4:
> > > > > > > > 
> > > > > > > > Before: infra_ao_i2c_ap -> imp_iic_wrap_ap_clock_i2c4
> > > > > > > > After:  infra_ao_i2c_ap -> infra_ao_i2c4 ->
> > > > > > > > imp_iic_wrap_ap_clock_i2c4
> > > > > > > > 
> > > > > > > > Fixes: 66cd0b4b0ce5 ("clk: mediatek: Add MT8186 imp i2c
> > > 
> > > wrapper
> > > > > > > > clock support")
> > > > > > > > Signed-off-by: AngeloGioacchino Del Regno <
> > > > > > > > angelogioacchino.delregno@xxxxxxxxxxxxx>
> > > > > > > 
> > > > > > > I'm curious about what led to discovering this error?
> > > > > > > 
> > > > > > 
> > > > > > Is that an acked-by?
> > > > > 
> > > > > MediaTek engineers are saying the original code already
> > > > > matches
> > > 
> > > the
> > > > > documentation provided by their hardware engineers. I'm
> > > > > trying to
> > > 
> > > get
> > > > > them to respond on the mailing list.
> > > > > 
> > > > > ChenYu
> > > > > 
> > > > 
> > > > After checking with I2C clock hardware designer there is no
> > > > infra_ao_i2c{4-6} clock gate in between. And the clock document
> > > > at
> > > 
> > > hand
> > > > aslo shows the same result. Generallly speaking, we would like
> > > > to
> > > 
> > > keep
> > > > sw setting align with the hardware design document. I would
> > > 
> > > recommand
> > > > not to change this part of code, but enable infra_ao_i2c{4-6}
> > > > prior
> > > 
> > > to
> > > > the usage of imp_iic_wrap_ap_clock_i2c clock.
> > > 
> > > Are infra_ao_i2c{4-6} actually used by the hardware? If so, for
> > > what
> > > purpose?
> > 
> > According to hardware designer it servers no purpose. Just a legacy
> > of
> > previous design...
> > 
> > > If it is actually needed by the hardware and it is not in the
> > > existing path,
> > > then it needs to be described in the device tree and handled by
> > > the
> > > driver.
> > > 
> > > ChenYu
> > 
> > After reviewing hardware design diagram, hardware designer
> > concludes
> > that the clock tree is indeed
> > 
> > top_i2c -> infra_ao_i2c{4-6}
> > top_i2c -> infra_ao_i2c_ap -> imp_iic_wrap_ap_clock_i2c{4-6}
> > 
> > so I think we should keep this clock relation unchanged.
> > 
> > Thanks
> > YuChang
> > 
> 
> Can you please also expand on CLK_INFRA_AO_I2C{1,2,5}_ARBITER clocks?
> Is the I2C arbiter also legacy of previous designs?
> 
> Please check [1], as I've sent a commit adding those in the
> devicetree.
> 
> Thanks,
> Angelo
> 
> [1]: 
> 
https://lore.kernel.org/all/20231020075540.15191-1-angelogioacchino.delregno@xxxxxxxxxxxxx/

According to Hardware designer this arbiter clock is also lagecy of
previous design and serve no function. And they are conneted to top_i2c
as well.

top_i2c-> CLK_INFRA_AO_I2C{1,2,5}_ARBITER

Also may I know the experiment that lead to the conclusion that you
need the ARBITER clock, and the clock tree is incorrect? I will bring
it back to discuss with our I2C owner. 

Thanks,
YuChang