[PATCH] x86: Fix booting with DEBUG_PAGE_ALLOC with more than 512G RAM

[PATCH] x86: Fix booting with DEBUG_PAGE_ALLOC with more than 512G RAM

[Yinghai Lu]

Dave reported that system have early crash if DEBUG_PAGEALLOC is selected,
when system have between 500G and 600G.

> [    0.000000] init_memory_mapping: [mem 0x00000000-0x000fffff]
> [    0.000000]  [mem 0x00000000-0x000fffff] page 4k
> [    0.000000] BRK [0x02086000, 0x02086fff] PGTABLE
> [    0.000000] BRK [0x02087000, 0x02087fff] PGTABLE
> [    0.000000] BRK [0x02088000, 0x02088fff] PGTABLE
> [    0.000000] init_memory_mapping: [mem 0xe80ee00000-0xe80effffff]
> [    0.000000]  [mem 0xe80ee00000-0xe80effffff] page 4k
> [    0.000000] BRK [0x02089000, 0x02089fff] PGTABLE
> [    0.000000] BRK [0x0208a000, 0x0208afff] PGTABLE
> [    0.000000] Kernel panic - not syncing: alloc_low_page: ran out of memory

It turns out that we missed increasing needed pages in BRK to mapping
initial 2M and [0,1M) when we switch to use #PF handler set mem mapping.

> commit 8170e6bed465b4b0c7687f93e9948aca4358a33b
> Author: H. Peter Anvin <hpa@zytor.com>
> Date:   Thu Jan 24 12:19:52 2013 -0800
>
>     x86, 64bit: Use a #PF handler to materialize early mappings on demand

Before that, we have maping from [0,512M) in head_64.S, and we can
spare two pages [0-1M).  After that change, we can not reuse pages anymore.

When we have more than 512M ram, we need extra page for pgd page with
[512G, 1024g).

Increase pages in BRK for page table to solve the booting problem.

Reported-by: Dave Hansen <dave.hansen@intel.com>
Bisected-by: Dave Hansen <dave.hansen@intel.com>
Tested-by: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: <stable@vger.kernel.org> 3.9+

---
 arch/x86/mm/init.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

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
@@ -78,8 +78,8 @@ __ref void *alloc_low_pages(unsigned int
 	return __va(pfn << PAGE_SHIFT);
 }
 
-/* need 4 4k for initial PMD_SIZE, 4k for 0-ISA_END_ADDRESS */
-#define INIT_PGT_BUF_SIZE	(5 * PAGE_SIZE)
+/* need 3 4k for initial PMD_SIZE,  3 4k for 0-ISA_END_ADDRESS */
+#define INIT_PGT_BUF_SIZE	(6 * PAGE_SIZE)
 RESERVE_BRK(early_pgt_alloc, INIT_PGT_BUF_SIZE);
 void  __init early_alloc_pgt_buf(void)
 {

Re: [PATCH] x86: Fix booting with DEBUG_PAGE_ALLOC with more than 512G RAM

[Dave Hansen]

On 08/12/2013 04:43 PM, Yinghai Lu wrote:
>> > commit 8170e6bed465b4b0c7687f93e9948aca4358a33b
>> > Author: H. Peter Anvin <hpa@zytor.com>
>> > Date:   Thu Jan 24 12:19:52 2013 -0800
>> >
>> >     x86, 64bit: Use a #PF handler to materialize early mappings on demand
> Before that, we have maping from [0,512M) in head_64.S, and we can
> spare two pages [0-1M).  After that change, we can not reuse pages anymore.
> 
> When we have more than 512M ram, we need extra page for pgd page with
> [512G, 1024g).
> 
> Increase pages in BRK for page table to solve the booting problem.

So how much does this get us up to?  1TB?  That's actually _fairly_
small today.  I've got a fairly old machine with that much in it, and
it's only half full of DIMMs.

It's also a bit worrying that this is completely disconnected from the
other code in the kernel that is concerned with the amount of total
address space in the system: MAX_PHYSADDR_BITS.

Re: [PATCH] x86: Fix booting with DEBUG_PAGE_ALLOC with more than 512G RAM

[Yinghai Lu]

On Mon, Aug 12, 2013 at 4:50 PM, Dave Hansen <dave.hansen@intel.com> wrote:
> On 08/12/2013 04:43 PM, Yinghai Lu wrote:
>>> > commit 8170e6bed465b4b0c7687f93e9948aca4358a33b
>>> > Author: H. Peter Anvin <hpa@zytor.com>
>>> > Date:   Thu Jan 24 12:19:52 2013 -0800
>>> >
>>> >     x86, 64bit: Use a #PF handler to materialize early mappings on demand
>> Before that, we have maping from [0,512M) in head_64.S, and we can
>> spare two pages [0-1M).  After that change, we can not reuse pages anymore.
>>
>> When we have more than 512M ram, we need extra page for pgd page with
>> [512G, 1024g).
>>
>> Increase pages in BRK for page table to solve the booting problem.
>
> So how much does this get us up to?  1TB?  That's actually _fairly_
> small today.  I've got a fairly old machine with that much in it, and
> it's only half full of DIMMs.
>
> It's also a bit worrying that this is completely disconnected from the
> other code in the kernel that is concerned with the amount of total
> address space in the system: MAX_PHYSADDR_BITS.

3 pages for [0,1M)
3 pages for initial 2M. ( it is 2M alignment).
are enough.

one page for PGD page (cover 512g), one page for PUD page (cover 1G)
and one page for PMD page (cover 2M).

After initial 2M is mapped, we will use that mapped 2M for other
memory range page table buffer.

Thanks

Yinghai