Messages in this thread | | | Subject | Re: [PATCH] ARM: Kconfig: improve ARM_SINGLE_ARMV7M description with Cortex M7 | From | Vladimir Murzin <> | Date | Fri, 3 Dec 2021 17:01:05 +0000 |
| |
On 12/3/21 4:41 PM, Giulio Benetti wrote: > Hi Vladimir, > > On 03/12/21 10:30, Vladimir Murzin wrote: >> On 12/2/21 10:20 PM, Giulio Benetti wrote: >>> ARM_SINGLE_ARMV7M implies Arm Cortex M7 too, so let's add it to >>> description with M0/M3/M4. >> >> Well it also implies M33 and M55. I'd suggest drop specific implementations >> from description - so we do not need to update it each time new compatible >> core is released. To account newest cores like M33 and M55 you can rephrase >> to: >> - ARMv7-M/v8-M > > This ^^^ is ok for me. Also, I don't see any M0 used at the moment so no ARMv6-M makes sense, if any in the future we can update. I see M3,M4,M7 for the moment, so ARMv7-M. But at the moment no ARMv8-M, or am I wrong?
We already support PMSAv8 which is ARMv8-M, M33/M55 is in flight [1].
> Maybe it would be better to add ARMv8-M when there will really be the support, considering also that ARM_SINGLE_ARMV7M is a bit ambiguous to be used with ARMv8-M. So maybe that could be modified too when an ARMv8-M is added. > > What do you think about this?
Well, I prefer to touch this as rare as possible, and if we touch it then ensure we do not need to touch it for a long time ;)
Look at A-class cores,
config ARCH_MULTI_V7 bool "ARMv7 based platforms (Cortex-A, PJ4, Scorpion, Krait)"
it doesn't mention ARMv8, yet it is pretty much possible to run it on some cores either bare-metal or as a guest (just look under arch/arm/include/asm/cputype.h)
These are just config options, not a legal deeds...
[1] https://lore.kernel.org/linux-arm-kernel/20211201132908.106711-1-vladimir.murzin@arm.com/T/
Cheers Vladimir
> > Best regards
| |