[PATCH] fix valid_memory_range region merging

[PATCH] fix valid_memory_range region merging

[Michael Neuling]

Check the next region we are including is type RANGE_RAM as well.

Signed-off-by: Michael Neuling <mikey@neuling.org>
---
I've not actually seen this cause a problem, but it looks wrong.  We
should probably merge regions properly and solve this problem for real. 

 kexec/kexec.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Index: kexec-tools-testing/kexec/kexec.c
===================================================================
--- kexec-tools-testing.orig/kexec/kexec.c
+++ kexec-tools-testing/kexec/kexec.c
@@ -96,7 +96,8 @@ int valid_memory_range(struct kexec_info
 		mstart = info->memory_range[i].start;
 		mend = info->memory_range[i].end;
 		if (i < info->memory_ranges - 1
-		    && mend == info->memory_range[i+1].start)
+		    && mend == info->memory_range[i+1].start
+		    && info->memory_range[i+1].type == RANGE_RAM)
 			mend = info->memory_range[i+1].end;
 
 		/* Check to see if we are fully contained */

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH] fix valid_memory_range region merging

[Simon Horman]

On Thu, Apr 26, 2007 at 04:50:34PM +1000, Michael Neuling wrote:
> Check the next region we are including is type RANGE_RAM as well.
> 
> Signed-off-by: Michael Neuling <mikey@neuling.org>
> ---
> I've not actually seen this cause a problem, but it looks wrong.  We
> should probably merge regions properly and solve this problem for real. 

I think that Magnus was working on some code do do just that...

-- 
Horms
  H: http://www.vergenet.net/~horms/
  W: http://www.valinux.co.jp/en/


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH] fix valid_memory_range region merging

[Simon Horman]

On Thu, Apr 26, 2007 at 04:50:34PM +1000, Michael Neuling wrote:
> Check the next region we are including is type RANGE_RAM as well.
> 
> Signed-off-by: Michael Neuling <mikey@neuling.org>
> ---
> I've not actually seen this cause a problem, but it looks wrong.  We
> should probably merge regions properly and solve this problem for real. 

Hi Michael,

appologies for this blast-from-the-past. This mail somehow
ended up filed in the wrong place and I chanced upon it just now.

This fix does seem correct to me, though I haven't seen it cause
any problems either. I'm happy to merge it (better late than never,
right?) if you are still happy with it.

> 
>  kexec/kexec.c |    3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> Index: kexec-tools-testing/kexec/kexec.c
> ===================================================================
> --- kexec-tools-testing.orig/kexec/kexec.c
> +++ kexec-tools-testing/kexec/kexec.c
> @@ -96,7 +96,8 @@ int valid_memory_range(struct kexec_info
>  		mstart = info->memory_range[i].start;
>  		mend = info->memory_range[i].end;
>  		if (i < info->memory_ranges - 1
> -		    && mend == info->memory_range[i+1].start)
> +		    && mend == info->memory_range[i+1].start
> +		    && info->memory_range[i+1].type == RANGE_RAM)
>  			mend = info->memory_range[i+1].end;
>  
>  		/* Check to see if we are fully contained */

-- 
Horms


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH] fix valid_memory_range region merging

[Michael Neuling]

In message <20080226022933.GA12270@verge.net.au> you wrote:
> On Thu, Apr 26, 2007 at 04:50:34PM +1000, Michael Neuling wrote:
> > Check the next region we are including is type RANGE_RAM as well.
> > 
> > Signed-off-by: Michael Neuling <mikey@neuling.org>
> > ---
> > I've not actually seen this cause a problem, but it looks wrong.  We
> > should probably merge regions properly and solve this problem for real. 
> 
> Hi Michael,
> 
> appologies for this blast-from-the-past. This mail somehow
> ended up filed in the wrong place and I chanced upon it just now.
> 
> This fix does seem correct to me, though I haven't seen it cause
> any problems either. I'm happy to merge it (better late than never,
> right?) if you are still happy with it.

Sweet thanks, I'd well and truly forgotten about it.  I think it was
just a random code cleanup, so if it still applies it should be fine to
merge.

Mikey

> 
> > 
> >  kexec/kexec.c |    3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> > 
> > Index: kexec-tools-testing/kexec/kexec.c
> > ===================================================================
> > --- kexec-tools-testing.orig/kexec/kexec.c
> > +++ kexec-tools-testing/kexec/kexec.c
> > @@ -96,7 +96,8 @@ int valid_memory_range(struct kexec_info
> >  		mstart = info->memory_range[i].start;
> >  		mend = info->memory_range[i].end;
> >  		if (i < info->memory_ranges - 1
> > -		    && mend == info->memory_range[i+1].start)
> > +		    && mend == info->memory_range[i+1].start
> > +		    && info->memory_range[i+1].type == RANGE_RAM)
> >  			mend = info->memory_range[i+1].end;
> >  
> >  		/* Check to see if we are fully contained */
> 
> -- 
> Horms
> 

_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec

Re: [PATCH] fix valid_memory_range region merging

[Simon Horman]

On Tue, Feb 26, 2008 at 10:29:28AM +0100, Michael Neuling wrote:
> In message <20080226022933.GA12270@verge.net.au> you wrote:
> > On Thu, Apr 26, 2007 at 04:50:34PM +1000, Michael Neuling wrote:
> > > Check the next region we are including is type RANGE_RAM as well.
> > > 
> > > Signed-off-by: Michael Neuling <mikey@neuling.org>
> > > ---
> > > I've not actually seen this cause a problem, but it looks wrong.  We
> > > should probably merge regions properly and solve this problem for real. 
> > 
> > Hi Michael,
> > 
> > appologies for this blast-from-the-past. This mail somehow
> > ended up filed in the wrong place and I chanced upon it just now.
> > 
> > This fix does seem correct to me, though I haven't seen it cause
> > any problems either. I'm happy to merge it (better late than never,
> > right?) if you are still happy with it.
> 
> Sweet thanks, I'd well and truly forgotten about it.  I think it was
> just a random code cleanup, so if it still applies it should be fine to
> merge.

Cool, it seems to apply just fine.
I've queued it up to be merged after the release.

-- 
Horms


_______________________________________________
kexec mailing list
kexec@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kexec