[PATCH v2][POWERPC] Fix initial lmb add region with a non-zero base

[PATCH v2][POWERPC] Fix initial lmb add region with a non-zero base

[Kumar Gala]

If we add to an empty lmb region with a non-zero base we will not coalesce
the number of regions down to one.  This causes problems on ppc32 for the
memory region as its assumed to only have one region.

We can fix this easily by causing the initial add to replace the dummy
region.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
---

Fix a bug the initial patch introduced if we have a region that gets added
at the beginning of the list we wouldn't actually add it.

Dave can you replace the patch in you tree with this one.

 arch/powerpc/mm/lmb.c |   12 ++++++++++++
 1 files changed, 12 insertions(+), 0 deletions(-)

diff --git a/arch/powerpc/mm/lmb.c b/arch/powerpc/mm/lmb.c
index 4ce23bc..4bf8f19 100644
--- a/arch/powerpc/mm/lmb.c
+++ b/arch/powerpc/mm/lmb.c
@@ -141,6 +141,12 @@ static long __init lmb_add_region(struct lmb_region *rgn, unsigned long base,
 	unsigned long coalesced = 0;
 	long adjacent, i;

+	if ((rgn->cnt == 1) && (rgn->region[0].size == 0)) {
+		rgn->region[0].base = base;
+		rgn->region[0].size = size;
+		return 0;
+	}
+
 	/* First try and coalesce this LMB with another. */
 	for (i=0; i < rgn->cnt; i++) {
 		unsigned long rgnbase = rgn->region[i].base;
@@ -185,6 +191,12 @@ static long __init lmb_add_region(struct lmb_region *rgn, unsigned long base,
 			break;
 		}
 	}
+
+	if (base < rgn->region[0].base) {
+		rgn->region[0].base = base;
+		rgn->region[0].size = size;
+	}
+
 	rgn->cnt++;

 	return 0;
-- 
1.5.3.8

Re: [PATCH v2][POWERPC] Fix initial lmb add region with a non-zero base

[David Miller]

From: Kumar Gala <galak@kernel.crashing.org>
Date: Tue, 19 Feb 2008 13:51:37 -0600 (CST)

> If we add to an empty lmb region with a non-zero base we will not coalesce
> the number of regions down to one.  This causes problems on ppc32 for the
> memory region as its assumed to only have one region.
> 
> We can fix this easily by causing the initial add to replace the dummy
> region.
> 
> Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
> ---
> 
> Fix a bug the initial patch introduced if we have a region that gets added
> at the beginning of the list we wouldn't actually add it.
> 
> Dave can you replace the patch in you tree with this one.

I think my tree has already or will soon be pulled in so
I don't want to rebase it.

Why don't you simply send me the relative bug fix instead?

Thanks!

Re: [PATCH v2][POWERPC] Fix initial lmb add region with a non-zero base

[Kumar Gala]

On Feb 19, 2008, at 6:30 PM, David Miller wrote:

> From: Kumar Gala <galak@kernel.crashing.org>
> Date: Tue, 19 Feb 2008 13:51:37 -0600 (CST)
>
>> If we add to an empty lmb region with a non-zero base we will not  
>> coalesce
>> the number of regions down to one.  This causes problems on ppc32  
>> for the
>> memory region as its assumed to only have one region.
>>
>> We can fix this easily by causing the initial add to replace the  
>> dummy
>> region.
>>
>> Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
>> ---
>>
>> Fix a bug the initial patch introduced if we have a region that  
>> gets added
>> at the beginning of the list we wouldn't actually add it.
>>
>> Dave can you replace the patch in you tree with this one.
>
> I think my tree has already or will soon be pulled in so
> I don't want to rebase it.
>
> Why don't you simply send me the relative bug fix instead?

np.  Are we trying to get this into 2.6.25 or .26?

- k

Re: [PATCH v2][POWERPC] Fix initial lmb add region with a non-zero base

[David Miller]

From: Kumar Gala <galak@kernel.crashing.org>
Date: Tue, 19 Feb 2008 21:02:04 -0600

> np.  Are we trying to get this into 2.6.25 or .26?

I'm ambivalent but I would obviously prefer 2.6.25 because
it would allow me to proceed more easily with my sparc64
NUMA work as well as get your bug fixes in more smoothly.

Re: [PATCH v2][POWERPC] Fix initial lmb add region with a non-zero base

[Paul Mackerras]

David Miller writes:

> I'm ambivalent but I would obviously prefer 2.6.25 because
> it would allow me to proceed more easily with my sparc64
> NUMA work as well as get your bug fixes in more smoothly.

Sounds like we should get Stephen to put it in linux-next-stable once
we're convinced it's all OK.

Paul.

Re: [PATCH v2][POWERPC] Fix initial lmb add region with a non-zero base

[Paul Mackerras]

Kumar Gala writes:

> np.  Are we trying to get this into 2.6.25 or .26?

I was going to put it into my powerpc-next branch and put it in
2.6.26.  I don't see any need for it to go in 2.6.25.

Paul.

Re: [PATCH v2][POWERPC] Fix initial lmb add region with a non-zero base

[Josh Boyer]

On Thu, 21 Feb 2008 13:15:58 +1100
Paul Mackerras <paulus@samba.org> wrote:

> Kumar Gala writes:
> 
> > np.  Are we trying to get this into 2.6.25 or .26?
> 
> I was going to put it into my powerpc-next branch and put it in
> 2.6.26.  I don't see any need for it to go in 2.6.25.

Is that a new branch you're going to create, or the normal "for-2.6.xx"?

josh