[PATCH 0/2] x86/Kconfig: deprecate DISCONTIGMEM support for 32-bit

[PATCH 0/2] x86/Kconfig: deprecate DISCONTIGMEM support for 32-bit

[Mike Rapoport]

Hi,

During the recent discussion [1] about DISCONTIGMEM being effectively
deprecated, Christoph suggested to mark it as BROKEN for x86-32.

These patches follow that suggestion and make SPARSEMEM the default for
X86_32 && NUMA and mark DISCONTIGMEM as BROKEN.

[1] https://lore.kernel.org/lkml/20190423071354.GB12114@infradead.org/

Mike Rapoport (2):
  x86/Kconfig: make SPARSEMEM default for 32-bit
  x86/Kconfig: deprecate DISCONTIGMEM support for 32-bit

 arch/x86/Kconfig | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)

-- 
2.7.4

[PATCH 1/2] x86/Kconfig: make SPARSEMEM default for 32-bit

[Mike Rapoport]

Sparsemem has been a default memory model for x86-64 for over a decade
since the commit b263295dbffd ("x86: 64-bit, make sparsemem vmemmap the
only memory model").

Make it the default for 32-bit NUMA systems (if there any left) as well.

Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
---
 arch/x86/Kconfig | 7 +------
 1 file changed, 1 insertion(+), 6 deletions(-)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 62fc3fd..5662a3e 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -1609,10 +1609,6 @@ config ARCH_DISCONTIGMEM_ENABLE
 	def_bool y
 	depends on NUMA && X86_32
 
-config ARCH_DISCONTIGMEM_DEFAULT
-	def_bool y
-	depends on NUMA && X86_32
-
 config ARCH_SPARSEMEM_ENABLE
 	def_bool y
 	depends on X86_64 || NUMA || X86_32 || X86_32_NON_STANDARD
@@ -1620,8 +1616,7 @@ config ARCH_SPARSEMEM_ENABLE
 	select SPARSEMEM_VMEMMAP_ENABLE if X86_64
 
 config ARCH_SPARSEMEM_DEFAULT
-	def_bool y
-	depends on X86_64
+	def_bool X86_64 || (NUMA && X86_32)
 
 config ARCH_SELECT_MEMORY_MODEL
 	def_bool y
-- 
2.7.4

[PATCH 2/2] x86/Kconfig: deprecate DISCONTIGMEM support for 32-bit

[Mike Rapoport]

Mel Gorman says:
  32-bit NUMA systems should be non-existent in practice.  The last NUMA
  system I'm aware of that was both NUMA and 32-bit only died somewhere
  between 2004 and 2007. If someone is running a 64-bit capable system in
  32-bit mode with NUMA, they really are just punishing themselves for fun.

Mark DISCONTIGMEM broken for now and remove it in a couple of releases.

Suggested-by: Christoph Hellwig <hch@infradead.org>
Suggested-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
---
 arch/x86/Kconfig | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 5662a3e..bd6f93c 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -1606,8 +1606,9 @@ config ARCH_FLATMEM_ENABLE
 	depends on X86_32 && !NUMA
 
 config ARCH_DISCONTIGMEM_ENABLE
-	def_bool y
+	def_bool n
 	depends on NUMA && X86_32
+	depends on BROKEN
 
 config ARCH_SPARSEMEM_ENABLE
 	def_bool y
-- 
2.7.4

Re: [PATCH 1/2] x86/Kconfig: make SPARSEMEM default for 32-bit

[Mel Gorman]

On Wed, Apr 24, 2019 at 04:24:11PM +0300, Mike Rapoport wrote:
> Sparsemem has been a default memory model for x86-64 for over a decade
> since the commit b263295dbffd ("x86: 64-bit, make sparsemem vmemmap the
> only memory model").
> 
> Make it the default for 32-bit NUMA systems (if there any left) as well.
> 
> Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>

Acked-by: Mel Gorman <mgorman@techsingularity.net>

-- 
Mel Gorman
SUSE Labs

Re: [PATCH 2/2] x86/Kconfig: deprecate DISCONTIGMEM support for 32-bit

[Mel Gorman]

On Wed, Apr 24, 2019 at 04:24:12PM +0300, Mike Rapoport wrote:
> Mel Gorman says:
>   32-bit NUMA systems should be non-existent in practice.  The last NUMA
>   system I'm aware of that was both NUMA and 32-bit only died somewhere
>   between 2004 and 2007. If someone is running a 64-bit capable system in
>   32-bit mode with NUMA, they really are just punishing themselves for fun.
> 
> Mark DISCONTIGMEM broken for now and remove it in a couple of releases.
> 
> Suggested-by: Christoph Hellwig <hch@infradead.org>
> Suggested-by: Mel Gorman <mgorman@techsingularity.net>
> Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>

It was really Christoph that suggested marking it broken but I do agree
that it should be marked broken to see if anyone complains and if not,
there is no real reason to keep discontig available on x86.

Acked-by: Mel Gorman <mgorman@techsingularity.net>

-- 
Mel Gorman
SUSE Labs