[tip:x86/mm] x86/mm: Fix the size calculation of mapping tables

[tip:x86/mm] x86/mm: Fix the size calculation of mapping tables

[tip-bot for WANG Cong]

Commit-ID:  722bc6b16771ed80871e1fd81c86d3627dda2ac8
Gitweb:     http://git.kernel.org/tip/722bc6b16771ed80871e1fd81c86d3627dda2ac8
Author:     WANG Cong <xiyou.wangcong@gmail.com>
AuthorDate: Mon, 5 Mar 2012 15:05:13 -0800
Committer:  Ingo Molnar <mingo@elte.hu>
CommitDate: Tue, 6 Mar 2012 09:38:26 +0100

x86/mm: Fix the size calculation of mapping tables

For machines that enable PSE, the first 2/4M memory region still uses
4K pages, so needs more PTEs in this case, but
find_early_table_space() doesn't count this.

This patch fixes it.

The bug was found via code review, no misbehavior of the kernel
was observed.

Signed-off-by: WANG Cong <xiyou.wangcong@gmail.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: <ianfang.cn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/n/tip-kq6a00qe33h7c7ais2xsywnh@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
---
 arch/x86/mm/init.c |   21 ++++++++++++---------
 1 files changed, 12 insertions(+), 9 deletions(-)

diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c
index 6cabf65..2e92fdc 100644
--- a/arch/x86/mm/init.c
+++ b/arch/x86/mm/init.c
@@ -30,8 +30,14 @@ int direct_gbpages
 #endif
 ;
 
-static void __init find_early_table_space(unsigned long end, int use_pse,
-					  int use_gbpages)
+struct map_range {
+	unsigned long start;
+	unsigned long end;
+	unsigned page_size_mask;
+};
+
+static void __init find_early_table_space(struct map_range *mr, unsigned long end,
+					  int use_pse, int use_gbpages)
 {
 	unsigned long puds, pmds, ptes, tables, start = 0, good_end = end;
 	phys_addr_t base;
@@ -56,6 +62,9 @@ static void __init find_early_table_space(unsigned long end, int use_pse,
 #ifdef CONFIG_X86_32
 		extra += PMD_SIZE;
 #endif
+		/* The first 2/4M doesn't use large pages. */
+		extra += mr->end - mr->start;
+
 		ptes = (extra + PAGE_SIZE - 1) >> PAGE_SHIFT;
 	} else
 		ptes = (end + PAGE_SIZE - 1) >> PAGE_SHIFT;
@@ -85,12 +94,6 @@ void __init native_pagetable_reserve(u64 start, u64 end)
 	memblock_reserve(start, end - start);
 }
 
-struct map_range {
-	unsigned long start;
-	unsigned long end;
-	unsigned page_size_mask;
-};
-
 #ifdef CONFIG_X86_32
 #define NR_RANGE_MR 3
 #else /* CONFIG_X86_64 */
@@ -262,7 +265,7 @@ unsigned long __init_refok init_memory_mapping(unsigned long start,
 	 * nodes are discovered.
 	 */
 	if (!after_bootmem)
-		find_early_table_space(end, use_pse, use_gbpages);
+		find_early_table_space(&mr[0], end, use_pse, use_gbpages);
 
 	for (i = 0; i < nr_range; i++)
 		ret = kernel_physical_mapping_init(mr[i].start, mr[i].end,

Re: [tip:x86/mm] x86/mm: Fix the size calculation of mapping tables

[Yinghai Lu]

On Tue, Mar 6, 2012 at 8:16 AM, tip-bot for WANG Cong
<xiyou.wangcong@gmail.com> wrote:
> Commit-ID:  722bc6b16771ed80871e1fd81c86d3627dda2ac8
> Gitweb:     http://git.kernel.org/tip/722bc6b16771ed80871e1fd81c86d3627dda2ac8
> Author:     WANG Cong <xiyou.wangcong@gmail.com>
> AuthorDate: Mon, 5 Mar 2012 15:05:13 -0800
> Committer:  Ingo Molnar <mingo@elte.hu>
> CommitDate: Tue, 6 Mar 2012 09:38:26 +0100
>
> x86/mm: Fix the size calculation of mapping tables
>
> For machines that enable PSE, the first 2/4M memory region still uses
> 4K pages, so needs more PTEs in this case, but
> find_early_table_space() doesn't count this.
>
> This patch fixes it.
>
> The bug was found via code review, no misbehavior of the kernel
> was observed.
>
> Signed-off-by: WANG Cong <xiyou.wangcong@gmail.com>
> Cc: Yinghai Lu <yinghai@kernel.org>
> Cc: Tejun Heo <tj@kernel.org>
> Cc: <ianfang.cn@gmail.com>
> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
> Link: http://lkml.kernel.org/n/tip-kq6a00qe33h7c7ais2xsywnh@git.kernel.org
> Signed-off-by: Ingo Molnar <mingo@elte.hu>
> ---
>  arch/x86/mm/init.c |   21 ++++++++++++---------
>  1 files changed, 12 insertions(+), 9 deletions(-)
>
> diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c
> index 6cabf65..2e92fdc 100644
> --- a/arch/x86/mm/init.c
> +++ b/arch/x86/mm/init.c
> @@ -30,8 +30,14 @@ int direct_gbpages
>  #endif
>  ;
>
> -static void __init find_early_table_space(unsigned long end, int use_pse,
> -                                         int use_gbpages)
> +struct map_range {
> +       unsigned long start;
> +       unsigned long end;
> +       unsigned page_size_mask;
> +};
> +
> +static void __init find_early_table_space(struct map_range *mr, unsigned long end,
> +                                         int use_pse, int use_gbpages)
>  {
>        unsigned long puds, pmds, ptes, tables, start = 0, good_end = end;
>        phys_addr_t base;
> @@ -56,6 +62,9 @@ static void __init find_early_table_space(unsigned long end, int use_pse,
>  #ifdef CONFIG_X86_32
>                extra += PMD_SIZE;
>  #endif
> +               /* The first 2/4M doesn't use large pages. */
> +               extra += mr->end - mr->start;
> +

Something wrong here, you need to check if this MR is really cover
first 2/4M ...
something like
Index: linux-2.6/arch/x86/mm/init.c
===================================================================
--- linux-2.6.orig/arch/x86/mm/init.c
+++ linux-2.6/arch/x86/mm/init.c
@@ -63,7 +63,8 @@ static void __init find_early_table_spac
 		extra += PMD_SIZE;
 #endif
 		/* The first 2/4M doesn't use large pages. */
-		extra += mr->end - mr->start;
+		if (mr->start < PMD_SIZE)
+			extra += mr->end - mr->start;

 		ptes = (extra + PAGE_SIZE - 1) >> PAGE_SHIFT;
 	} else

otherwise, for system with 512G or more. page table for [4G, 512g) may
have preallocate more
but only use little.

Yinghai