Re: [PATCH 1/2] ARM: Remove redundant ARCH_SPARSEMEM_DEFAULT setting

[Mike Rapoport <rppt@linux.ibm.com>]

On Thu, May 07, 2020 at 11:30:39AM +0100, Russell King - ARM Linux admin wrote:
> On Wed, May 06, 2020 at 04:50:08PM -0700, Florian Fainelli wrote:
> > From: Kevin Cernekee <cernekee@gmail.com>
> > 
> > If ARCH_SPARSEMEM_ENABLE=y and ARCH_{FLATMEM,DISCONTIGMEM}_ENABLE=n,
> > then the logic in mm/Kconfig already makes CONFIG_SPARSEMEM the only
> > choice.  This is true for all of the existing ARM users of
> > ARCH_SPARSEMEM_ENABLE.
> > 
> > Forcing ARCH_SPARSEMEM_DEFAULT=y if ARCH_SPARSEMEM_ENABLE=y prevents
> > us from ever defaulting to FLATMEM, so we should remove this setting.
> 
> No explanation why that is desirable.
> 
> > -config ARCH_SPARSEMEM_DEFAULT
> > -	def_bool ARCH_SPARSEMEM_ENABLE
> > -
> 
> What this basically says is ARCH_SPARSEMEM_ENABLE=ARCH_SPARSEMEM_DEFAULT,
> which is exactly what we want for the non-multiplatform boards that
> select ARCH_SPARSEMEM_ENABLE - we _want_ them to default to sparsemem
> because that is what the platform requires.

The ARCH_SPARSEMEM_DEFAULT knob has no effect unless manual selection
of the memory model is enabled, i.e. arch Kconfig sets
ARCH_SELECT_MEMORY_MODEL to 'y'.

Otherwise, ARCH_SPARSEMEM_ENABLE is sufficient to select SPARSEMEM over
FLATMEM:

config SPARSEMEM
	def_bool y
	depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL

config FLATMEM
	def_bool y
	depends on (!DISCONTIGMEM && !SPARSEMEM) || FLATMEM_MANUAL

> For example, with RiscPC, which selects ARCH_SPARSEMEM_ENABLE, we have
> four banks of memory at 0x10000000, 0x14000000, 0x18000000 and
> 0x1c000000.  These correspond with the two memory slots - the first two
> for the first slot, and the second two for the second slot.  Each slot
> has two banks.  The size of each memory bank depends on the size of the
> module.
>
> Flatmem is completely unable to work with this setup if all banks are
> populated, and the first bank does not contain enough memory to allocate
> the struct page array.  So, sparsemem is the only option there.
> 
> Hence, for these platforms, we want sparsemem and only sparsemem, not
> flatmem.
> 
> So, this patch which makes it possible to select flatmem is completely
> out of the question for these platforms.

This patch alone won't make it possible to select flatmem for these
platforms, actually, as of now it changes nothing.

The removal of ARCH_SPARSEMEM_DEFAULT will have an effect only with
addtion of ARCH_SELECT_MEMORY_MODEL.

If I understood Florian's intention correctly, the goal was to allow
manual selection of the memory model for multiplatform builds that today
implicitly use flatmem.

I think that the patch below would achieve that without changing the
current defaults and without forcing flatmem on the platforms that
already explicitly select sparsemem.

diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index 66a04f6f4775..b6eb1a28ca27 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
@@ -312,6 +312,9 @@ choice
 config ARCH_MULTIPLATFORM
 	bool "Allow multiple platforms to be selected"
 	depends on MMU
+	select ARCH_FLATMEM_ENABLE
+	select ARCH_SPARSEMEM_ENABLE
+	select ARCH_SELECT_MEMORY_MODEL
 	select ARM_HAS_SG_CHAIN
 	select ARM_PATCH_PHYS_VIRT
 	select AUTO_ZRELADDR
@@ -1515,11 +1518,14 @@ config OABI_COMPAT
 config ARCH_HAS_HOLES_MEMORYMODEL
 	bool
 
+config ARCH_FLATMEM_ENABLE
+	bool
+
 config ARCH_SPARSEMEM_ENABLE
 	bool
 
-config ARCH_SPARSEMEM_DEFAULT
-	def_bool ARCH_SPARSEMEM_ENABLE
+config ARCH_SELECT_MEMORY_MODEL
+	bool
 
 config HAVE_ARCH_PFN_VALID
 	def_bool ARCH_HAS_HOLES_MEMORYMODEL || !SPARSEMEM

> -- 
> RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
> FTTC broadband for 0.8mile line in suburbia: sync at 10.2Mbps down 587kbps up

-- 
Sincerely yours,
Mike.

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel