This patch is a fix for 2.6.24.
Ordinarily the size of a pageblock is determined at compile-time based on the

hugepage size. On PPC64, the hugepage size is determined at runtime based on

what is supported by the machine. With legacy machines such as iSeries that do

not support hugepages, HPAGE_SHIFT becomes 0. This results in pageblock_order

being set to -PAGE_SHIFT and a crash results shortly afterwards.
This patch sets HUGETLB_PAGE_SIZE_VARIABLE for PPC64 and adds a function

to select a sensible value for pageblock order by default.  It checks that

HPAGE_SHIFT is a sensible value before using the hugepage size; if it is

not MAX_ORDER-1 is used.
Credit goes to Stephen Rothwell for identifying the bug and testing candidate

patches.  Additional credit goes to Andy Whitcroft for spotting a problem

with respects to IA-64 before releasing. Additional credit to David Gibson

for testing with the libhugetlbfs test suite.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>

Tested-by: Stephen Rothwell <sfr@canb.auug.org.au>
---

 arch/powerpc/Kconfig |    5 +++++

 mm/page_alloc.c      |   14 ++++++++++++--

 2 files changed, 17 insertions(+), 2 deletions(-)
diff -rup -X /usr/src/patchset-0.6/bin//dontdiff linux-2.6.24-rc2-mm1-clean/arch/powerpc/Kconfig linux-2.6.24-rc2-005_iSeries_fix/arch/powerpc/Kconfig

--- linux-2.6.24-rc2-mm1-clean/arch/powerpc/Kconfig	2007-11-14 11:38:05.000000000 +0000

+++ linux-2.6.24-rc2-005_iSeries_fix/arch/powerpc/Kconfig	2007-11-14 11:39:12.000000000 +0000

@@ -187,6 +187,11 @@ config FORCE_MAX_ZONEORDER

 	default "9" if PPC_64K_PAGES

 	default "13"
+config HUGETLB_PAGE_SIZE_VARIABLE

+	bool

+	depends on HUGETLB_PAGE

+	default y

+

 config MATH_EMULATION

 	bool "Math emulation"

 	depends on 4xx || 8xx || E200 || PPC_MPC832x || E500

diff -rup -X /usr/src/patchset-0.6/bin//dontdiff linux-2.6.24-rc2-mm1-clean/mm/page_alloc.c linux-2.6.24-rc2-005_iSeries_fix/mm/page_alloc.c

--- linux-2.6.24-rc2-mm1-clean/mm/page_alloc.c	2007-11-14 11:38:08.000000000 +...
[ message continues ]

-

As a #define, possibly but as a static inline - definitly not.
In this context, the define is not used because set_pageblock_order()

is a no-op when CONFIG_HUGETLB_PAGE_SIZE_VARIABLE is unset.

pageblock_default_order() is only defined for symmetry as set_pageblock_order()

is defined in both contexts. However, as a #define it might make more sense

to a casual reader to see HUGETLB_PAGE_ORDER even if it has no effect. I

can send a version of the patch that does this with a comment explaining

what is going on with set_pageblock_order() if you like.
However, in a follow-up fix, you make pageblock_default_order() a static

inline. If it tries to return HUGETLB_PAGE_ORDER, it will fail to compile

when CONFIG_HUGETLB_PAGE is not set.
--

--

Mel Gorman

Part-time Phd Student                          Linux Technology Center

University of Limerick                         IBM Dublin Software Lab

-

Don't care really.  Something which is fixed up ;)

-

--- a/mm/page_alloc.c~fix-boot-problem-with-iseries-lacking-hugepage-support-fix

+++ a/mm/page_alloc.c

@@ -3268,7 +3268,7 @@ static void inline setup_usemap(struct p

 #ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
 /* Return a sensible default order for the pageblock size. */

-static inline int __init pageblock_default_order(void)

+static inline int pageblock_default_order(void)

 {

 	if (HPAGE_SHIFT > PAGE_SHIFT)

 		return HUGETLB_PAGE_ORDER;

@@ -3291,7 +3291,11 @@ static inline void __init set_pageblock_

 }

 #else /* CONFIG_HUGETLB_PAGE_SIZE_VARIABLE */
-#define pageblock_default_order(x) (0)

+static inline int pageblock_default_order(void)

+{

+	return 0;

+}

+

 #define set_pageblock_order(x)	do {} while (0)
 #endif /* CONFIG_HUGETLB_PAGE_SIZE_VARIABLE */

_
please avoid adding macros when C could have been used.  C is nicer to look

at and has typechecking and stuff.

-

I know the __init is meaningless in this context. It is there as a guide

if someone decides to drop the inline for some reason that it should
It's never used so it could have been anything and still compiled. I admit

this is confusing.  I've posted a version below that changes this to a static

inline, returns MAX_ORDER-1 which is a sensible value even if unused and
Understood. Here is an updated version. It is functionally equivilant to

the earlier patch but may be easier on the eye
===
Ordinarily the size of a pageblock is determined at compile-time based on the

hugepage size. On PPC64, the hugepage size is determined at runtime based on

what is supported by the machine. With legacy machines such as iSeries that

do not support hugepages, HPAGE_SHIFT is 0. This results in pageblock_order

being set to -PAGE_SHIFT and a crash results shortly afterwards.
This patch adds a function to select a sensible value for pageblock order by

default when HUGETLB_PAGE_SIZE_VARIABLE is set. It checks that HPAGE_SHIFT

is a sensible value before using the hugepage size; if it is not MAX_ORDER-1

is used.
This is a fix for 2.6.24.
Credit goes to Stephen Rothwell for identifying the bug and testing candidate

patches.  Additional credit goes to Andy Whitcroft for spotting a problem

with respects to IA-64 before releasing. Additional credit to David Gibson

for testing with the libhugetlbfs test suite.
Signed-off-by: Mel Gorman <mel@csn.ul.ie>

Tested-by: Stephen Rothwell <sfr@canb.auug.org.au>
---

 arch/powerpc/Kconfig |    5 +++++

 mm/page_alloc.c      |   23 +++++++++++++++++++++--

 2 files changed, 26 insertions(+), 2 deletions(-)
diff -rup -X /usr/src/patchset-0.6/bin//dontdiff linux-2.6.24-rc2-mm1-clean/arch/powerpc/Kconfig linux-2.6.24-rc2-mm1-varpage-fix/arch/powerpc/Kconfig

--- linux-2.6.24-rc2-mm1-clean/arch/powerpc/Kconfig	2007-11-14 11:38:05.000000000 +0000

+++ linux-2.6.24-rc2-mm1-varpage-fix/arch/powerpc/Kconfig	2007-11-15 10:44:38.000000000 +0000

@@ -187,6 +1...
[ message continues ]