[PATCH v1 0/4] J721E DTS Sync with Kernel v6.5-rc1

[Neha Malcom Francis]

Hi Nishanth

On 31/08/23 17:05, Nishanth Menon wrote:
> On 13:56-20230831, Neha Malcom Francis wrote:
>> Hi Nishanth
>>
>> On 28/08/23 23:03, Nishanth Menon wrote:
>>> On 17:01-20230828, Neha Malcom Francis wrote:
>>>> This series aims to sync kernel.org v6.5-rc1 DTS with that of U-Boot. It
>>>> also includes cleanups where necessary along with certain changes to
>>>> ensure boot is unaffected.
>>>>
>>>> While it's mainly synced to the tag v6.5-rc1, one extra commit has been
>>>> added that cleans up duplicating wkup_i2c0 nodes [1] in Kernel.
>>>>
>>>> Same as with other board series that have taken up this effort, cleanup
>>>> of mcu_ringacc and mcu_udmap are dependent on MCU Ringacc [2] and DMA
>>>> [3] fixes. This will be taken up after their merge to Linux.
>>>>
>>>> [1] https://lore.kernel.org/all/20230721082344.1534094-1-u-kumar1@ti.com/
>>>> [2] https://lore.kernel.org/all/20230809175932.2553156-1-vigneshr@ti.com/
>>>> [3] https://lore.kernel.org/all/20230810174356.3322583-1-vigneshr@ti.com/
>>>>
>>>> Boot logs:
>>>> https://gist.github.com/nehamalcom/231426b784e0ba96ef3a0da84346e038
>>>>
>>>> Neha Malcom Francis (4):
>>>>     arm: dts: k3-j721e-r5: Clean up inclusion hierarchy
>>>>     arm: dts: k3-j721e: Sync with v6.5-rc1
>>>>     drivers: misc: k3_avs: Add linux compatible to maintain sync
>>>>     configs: j721e: Remove HBMC_AM654 config
>>> please change the order of patches:
>>> a) introduce the driver change for vtm
>>> b) cleanup config and drop the hbmc node stuff
>>> c) rest of dts sync
>>>
>>> NOTE:
>>> * J721e does'nt use SYSFW - it uses TIFS
>>> * for properties that dont use bootph or u-boot, properties, we
>>>     should'nt be including property changes in u-boot.dtsi or r5.dts
>>>     without explanation as to what.
>>> * There is a bunch of /delete-property/ plugged in for no clear reason.
>>>     I see that even for timer - but timer data could have been included in
>>>     arch/arm/mach-k3/j721e/dev-data.c and so could have been others too.
>>>     This needs clear explanation.
>>>
>>
>> All valid points, will factor these in. However regarding the timer, we have
>> already tried to do the same for J721E by adding in the data in dev-data.c
>> however this does not work since TIFS is in sysfw.itb which has not been
>> loaded yet. Will take action to debug and root cause that further as well as
>> try adding data for all the other existing /delete-property/ using nodes.
> 
> Thanks. What helped us on other platforms was to setup a breakpoints
> with openocd ti_clk_probe and omap_timer_probe using the debug
> technique[1] (board_init_f) and looking at the call stacks. in case of
> am62, ti_clk_probe is hit first and call stack looks like [2] followed
> by omap_timer_probe where the call stack looks like [3].
> 
> [1] https://u-boot.readthedocs.io/en/latest/board/ti/k3.html#code-modification
> [2] https://gist.github.com/nmenon/495f85267ecd0c833218b5054b568b00
> [3] https://gist.github.com/nmenon/c7a3a8c0de0a65b06334190ca9350ead
> 

Thanks for the detailed help on the debug! Will be useful!

-- 
Thanking You
Neha Malcom Francis