Re: [PATCH RESEND 3/4] ARM: BCM: Add SMP support for Broadcom NSP

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 





On 11/5/2015 3:14 PM, Linus Walleij wrote:
> On Thu, Nov 5, 2015 at 6:51 AM, Kapil Hali <kapilh@xxxxxxxxxxxx> wrote:
> 
>> Add SMP support for Broadcom's Northstar Plus SoC,
>> cpu enable method and pen_release procedures. This
>> changes also consolidates iProc family's - BCM NSP
>> and BCM Kona, SMP handling in a common file.
>>
>> Northstar Plus SoC is based on ARM Cortex-A9
>> revision r3p0 which requires configuration for ARM
>> Errata 764369 for SMP. This change adds the needed
>> configuration option.
>>
>> Signed-off-by: Kapil Hali <kapilh@xxxxxxxxxxxx>
> (...)
>> +/*
>> + * iProc specific entry point for secondary CPUs.  This provides
>> + * a "holding pen" into which all secondary cores are held until
>> + * we are ready for them to initialise.
>> + */
> 
> Do you *REALLY* need the complex pen hold/release semaphore
> stuff?
> 
> Consult for example commit
> c00def71efd919e8ae835a25f4f4c80a4b2d36d3
> "ARM: ux500: simplify secondary CPU boot"
> 
> I realized the pen hold/release stuff was totally surplus on the
> ux500 platform, it was probably just copied from the Vexpress
> or RealView reference design and kept around because ARM
> did things that way. And you are copying the same code from
> other platforms again. Do you *really* understand the pen hold/release
> code? (I don't.)
> 
> In our case it turned out that all that was really needed was:
> 
> - Map and enable SCU at .smp_prepare_cpus()
> 
> - Write the start address for the secondary CPU(s) in their
>   magic registers/addresses i.e what you do in
>   nsp_write_lut() and then call
>   arch_send_wakeup_ipi_mask(cpumask_of(cpu)); in
>   .smp_boot_secondary()
>   It seems you may not even need the IPI, I can't see
>   that call in your patch so I guess your secondary
>   CPUs aren't in WFI at boot.
> 
There is a dsb_sev() to get secondary core out of standby. However,
I think arch_send_wakeup_ipi_mask() is a better way as it is per
core. I will make the required changes in the next patch set.
> No pen hold/release magic! Works like a charm for us.
> So see of you really need this horror story with copied
> assembly code from realview etc.
> 
> Yours,
> Linus Walleij
> 
Thanks,
Kapil Hali


--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html



[Index of Archives]     [Device Tree Compilter]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux PCI Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Yosemite Backpacking]
  Powered by Linux