[RFC] Question about contributing mt6577 support but SMP doesn't work

[Matthias Brugger]

Hi Boris,

On 30/01/2021 02:55, Boris Lysov wrote:
> Hello! I'm working on bringing the initial support for mt6577 into
> the mainline kernel. I've got the base set of hardware to work: timers,
> and few interrupt controllers, and all 4 UARTs, and PL310 cache
> controller. My test device boots mainline successfully and runs stable!
> 
> I'm having issues with getting SMP to work. mt6577 is a dual-core SoC,
> and broken SMP means half of the processing power is unavailable. I
> tried to use the existing SMP bring-up code located in
> arch/arm/mach-mediatek/platsmp.c , but it didn't work albeit all
> specified registers were correct. I also tried porting the SMP init
> code from the downstream mediatek kernel, but it never worked either.
> 
> My question is: could I still contribute my device support patches
> despite literally 50% of the CPU not working? I *strongly* doubt they
> would be accepted because of it, but on the other hand devices with
> mt6577 (and other compatible SoCs like mt8317) are still able to boot
> and run bare mainline kernel on a single core. I ran out of ideas on
> how to make that pesky 2nd core come up online with my poor knowledge.
> 

I'm happy to see mt6577 support coming into the mainline kernel. I added Yingjoe
to this thread, hopefully he can help to provide the info needed to enable SMP
in mainline.

In the meanwhile I'm happy with support for the chip without SMP, so please feel
free to provide patches :)

Regards,
Matthias