[PATCH 4/4] clk: samsung: exynos850: fix propagation of SPI IPCLK rate

Tudor Ambarus tudor.ambarus at linaro.org
Fri Mar 22 02:39:28 PDT 2024


Hi, Sam!

On 3/1/24 00:13, Sam Protsenko wrote:
> I fail to see how this patch fixes anything. Instead it looks to me it
> replaces the (already) correctly implemented logic with incorrect one.

I opened another thread asking for feedback on whether it's safe to
re-parent the USI MUX to OSCCLK at run-time, find it here:
https://lore.kernel.org/linux-samsung-soc/71df1d6b-f40b-4896-a672-c5f0f526fb1f@linaro.org/T/#m588abb87eb5fd8817d71d06b94c91eb84928e06b

Jaewon came up with the idea on verifying what the downstream clock
driver does. I added some prints in the driver, and indeed the USI MUX
re-parents to OSCCLK on low SPI clock rates in the GS101 case.

Thus I'll respin this patch set fixing GS101 on low USI clock rates by
re-parenting the USI MUX to OSCCLK. I'll leave exynos850 out if I don't
hear back from you, but I think it deserves the same fix. Allowing SPI
to modify the clock rate of HSI2C/I3C at run-time is bad IMO.
Re-parenting the USI MUX to OSCCLK fixes this problem, HSI2C/I3C will no
longer be affected on low SPI clock rates.

Cheers,
ta



More information about the linux-arm-kernel mailing list