[patch] fix bootmem regression on Altix

[patch] fix bootmem regression on Altix

[Russ Anderson]

[patch] fix bootmem regression on Altix

A recent change prevents SGI Altix from booting.
This patch fixes the problem.

The regresson was introduced in commit 434d53b00d6bb7be0a1d3dcc0d0d5df6c042e164

Signed-off-by: Russ Anderson (rja@sgi.com)

---
 kernel/sched.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: test/kernel/sched.c
=================================--- test.orig/kernel/sched.c	2008-04-23 16:00:58.000000000 -0500
+++ test/kernel/sched.c	2008-04-23 16:01:33.047453963 -0500
@@ -8127,7 +8127,7 @@ void __init sched_init(void)
 	 * we use alloc_bootmem().
 	 */
 	if (alloc_size) {
-		ptr = (unsigned long)alloc_bootmem_low(alloc_size);
+		ptr = (unsigned long)alloc_bootmem(alloc_size);
 
 #ifdef CONFIG_FAIR_GROUP_SCHED
 		init_task_group.se = (struct sched_entity **)ptr;
-- 
Russ Anderson, OS RAS/Partitioning Project Lead  
SGI - Silicon Graphics Inc          rja@sgi.com

Re: [patch] fix bootmem regression on Altix

[Petr Tesarik]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Russ Anderson wrote:
> [patch] fix bootmem regression on Altix
> 
> A recent change prevents SGI Altix from booting.
> This patch fixes the problem.
> 
> The regresson was introduced in commit 434d53b00d6bb7be0a1d3dcc0d0d5df6c042e164

Could we get this into mainline? My Altix box won't boot without this
patch either and applying it on top of all trees after the
above-mentioned commit is annoying when bisecting. :(

Anyway, thank you for your patch, Russ!

Petr Tesarik

> Signed-off-by: Russ Anderson (rja@sgi.com)
> 
> ---
>  kernel/sched.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> Index: test/kernel/sched.c
> =================================> --- test.orig/kernel/sched.c	2008-04-23 16:00:58.000000000 -0500
> +++ test/kernel/sched.c	2008-04-23 16:01:33.047453963 -0500
> @@ -8127,7 +8127,7 @@ void __init sched_init(void)
>  	 * we use alloc_bootmem().
>  	 */
>  	if (alloc_size) {
> -		ptr = (unsigned long)alloc_bootmem_low(alloc_size);
> +		ptr = (unsigned long)alloc_bootmem(alloc_size);
>  
>  #ifdef CONFIG_FAIR_GROUP_SCHED
>  		init_task_group.se = (struct sched_entity **)ptr;

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIRQcqjpY2ODFi2ogRArSnAKCLqsJldLKNMWKgA5KFB2K1S0kOoQCeKRe4
ARtwQ/mpL0BrFVo+h2IpdJQ=vvtw
-----END PGP SIGNATURE-----

Re: [patch] fix bootmem regression on Altix

[Russ Anderson]

On Tue, Jun 03, 2008 at 10:56:10AM +0200, Petr Tesarik wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Russ Anderson wrote:
> > [patch] fix bootmem regression on Altix
> > 
> > A recent change prevents SGI Altix from booting.
> > This patch fixes the problem.
> > 
> > The regresson was introduced in commit 434d53b00d6bb7be0a1d3dcc0d0d5df6c042e164
> 
> Could we get this into mainline?

Which mainline tree?  Tony pushed the patch to Linus back on 4/24
and I see it in Linus' tree.

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;hG2613b961affef0c73f1c797993678312e7c666


>                                    My Altix box won't boot without this
> patch either and applying it on top of all trees after the
> above-mentioned commit is annoying when bisecting. :(
> 
> Anyway, thank you for your patch, Russ!
> 
> Petr Tesarik
> 
> > Signed-off-by: Russ Anderson (rja@sgi.com)
> > 
> > ---
> >  kernel/sched.c |    2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > Index: test/kernel/sched.c
> > =================================> > --- test.orig/kernel/sched.c	2008-04-23 16:00:58.000000000 -0500
> > +++ test/kernel/sched.c	2008-04-23 16:01:33.047453963 -0500
> > @@ -8127,7 +8127,7 @@ void __init sched_init(void)
> >  	 * we use alloc_bootmem().
> >  	 */
> >  	if (alloc_size) {
> > -		ptr = (unsigned long)alloc_bootmem_low(alloc_size);
> > +		ptr = (unsigned long)alloc_bootmem(alloc_size);
> >  
> >  #ifdef CONFIG_FAIR_GROUP_SCHED
> >  		init_task_group.se = (struct sched_entity **)ptr;
> 
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.6 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> 
> iD8DBQFIRQcqjpY2ODFi2ogRArSnAKCLqsJldLKNMWKgA5KFB2K1S0kOoQCeKRe4
> ARtwQ/mpL0BrFVo+h2IpdJQ> =vvtw
> -----END PGP SIGNATURE-----

-- 
Russ Anderson, OS RAS/Partitioning Project Lead  
SGI - Silicon Graphics Inc          rja@sgi.com

RE: [patch] fix bootmem regression on Altix

[Luck, Tony]

> Could we get this into mainline? My Altix box won't boot without this
> patch either and applying it on top of all trees after the
> above-mentioned commit is annoying when bisecting. :(

It (well an equivalent patch) is in mainline.  When you are bisecting,
chances are that you'll step into the time period after this was
broken, but before this patch was applied ... so you'll have to look
at sched.c and apply this patch manually when bisect picks a commit
in this zone. But I don't see anyway around that (history is immutable).

-Tony

See this commit:

commit 5a9d3225a0d7060bdf3a18018992dc8cef958425
Author: David Miller <davem@davemloft.net>
Date:   Thu Apr 24 20:46:20 2008 -0700

    sched: use alloc_bootmem() instead of alloc_bootmem_low()
    
    There is no guarantee that there is physical ram below 4GB, and in
    fact many boxes don't have exactly that.
    
    Signed-off-by: David S. Miller <davem@davemloft.net>
    Signed-off-by: Ingo Molnar <mingo@elte.hu>

diff --git a/kernel/sched.c b/kernel/sched.c
index 0014b03..09ca69b 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -8128,7 +8128,7 @@ void __init sched_init(void)
 	 * we use alloc_bootmem().
 	 */
 	if (alloc_size) {
-		ptr = (unsigned long)alloc_bootmem_low(alloc_size);
+		ptr = (unsigned long)alloc_bootmem(alloc_size);
 
 #ifdef CONFIG_FAIR_GROUP_SCHED
 		init_task_group.se = (struct sched_entity **)ptr;

Re: [patch] fix bootmem regression on Altix

[Petr Tesarik]

Russ Anderson wrote:
> On Tue, Jun 03, 2008 at 10:56:10AM +0200, Petr Tesarik wrote:
> Russ Anderson wrote:
>>>> [patch] fix bootmem regression on Altix
>>>>
>>>> A recent change prevents SGI Altix from booting.
>>>> This patch fixes the problem.
>>>>
>>>> The regresson was introduced in commit 434d53b00d6bb7be0a1d3dcc0d0d5df6c042e164
> Could we get this into mainline?
> 
>> Which mainline tree?  Tony pushed the patch to Linus back on 4/24
>> and I see it in Linus' tree.
> 
>> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;hG2613b961affef0c73f1c797993678312e7c666
>

Ooops, forget it, I screwed up the bisect branch and the change was
apparently no longer there. After git reset --hard, I can see it correctly.

Sorry for the fuss,
Petr Tesarik