slab allocators: Remove multiple alignment specifications.

slab allocators: Remove multiple alignment specifications.

[Christoph Lameter]

It is not necessary to tell the slab allocators to align to a cacheline
if an explicit alignment was already specified. It is rather confusing
to specify multiple alignments.

Make sure that the call sites only use one form of alignment.

Signed-off-by: Christoph Lameter <clameter@sgi.com>

Index: linux-2.6.21-rc6/arch/powerpc/mm/hugetlbpage.c
===================================================================
--- linux-2.6.21-rc6.orig/arch/powerpc/mm/hugetlbpage.c	2007-04-20 19:25:14.000000000 -0700
+++ linux-2.6.21-rc6/arch/powerpc/mm/hugetlbpage.c	2007-04-20 19:25:40.000000000 -0700
@@ -1063,7 +1063,7 @@ static int __init hugetlbpage_init(void)
 	huge_pgtable_cache = kmem_cache_create("hugepte_cache",
 					       HUGEPTE_TABLE_SIZE,
 					       HUGEPTE_TABLE_SIZE,
-					       SLAB_HWCACHE_ALIGN,
+					       0,
 					       zero_ctor, NULL);
 	if (! huge_pgtable_cache)
 		panic("hugetlbpage_init(): could not create hugepte cache\n");
Index: linux-2.6.21-rc6/arch/powerpc/mm/init_64.c
===================================================================
--- linux-2.6.21-rc6.orig/arch/powerpc/mm/init_64.c	2007-04-20 19:25:14.000000000 -0700
+++ linux-2.6.21-rc6/arch/powerpc/mm/init_64.c	2007-04-20 19:25:40.000000000 -0700
@@ -183,7 +183,7 @@ void pgtable_cache_init(void)
 		    "for size: %08x...\n", name, i, size);
 		pgtable_cache[i] = kmem_cache_create(name,
 						     size, size,
-						     SLAB_HWCACHE_ALIGN,
+						     0,
 						     zero_ctor,
 						     NULL);
 		if (! pgtable_cache[i])
Index: linux-2.6.21-rc6/arch/sparc64/mm/init.c
===================================================================
--- linux-2.6.21-rc6.orig/arch/sparc64/mm/init.c	2007-04-20 19:25:14.000000000 -0700
+++ linux-2.6.21-rc6/arch/sparc64/mm/init.c	2007-04-20 19:25:40.000000000 -0700
@@ -191,7 +191,7 @@ void pgtable_cache_init(void)
 {
 	pgtable_cache = kmem_cache_create("pgtable_cache",
 					  PAGE_SIZE, PAGE_SIZE,
-					  SLAB_HWCACHE_ALIGN,
+					  0,
 					  zero_ctor,
 					  NULL);
 	if (!pgtable_cache) {
Index: linux-2.6.21-rc6/arch/sparc64/mm/tsb.c
===================================================================
--- linux-2.6.21-rc6.orig/arch/sparc64/mm/tsb.c	2007-04-20 19:25:14.000000000 -0700
+++ linux-2.6.21-rc6/arch/sparc64/mm/tsb.c	2007-04-20 19:25:40.000000000 -0700
@@ -262,7 +262,7 @@ void __init tsb_cache_init(void)
 
 		tsb_caches[i] = kmem_cache_create(name,
 						  size, size,
-						  SLAB_HWCACHE_ALIGN,
+						  0,
 						  NULL, NULL);
 		if (!tsb_caches[i]) {
 			prom_printf("Could not create %s cache\n", name);

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Andrew Morton]

On Fri, 20 Apr 2007 22:12:52 -0700 (PDT) Christoph Lameter <clameter@sgi.com> wrote:

> --- linux-2.6.21-rc6.orig/arch/sparc64/mm/init.c	2007-04-20 19:25:14.000000000 -0700
> +++ linux-2.6.21-rc6/arch/sparc64/mm/init.c	2007-04-20 19:25:40.000000000 -0700
> @@ -191,7 +191,7 @@ void pgtable_cache_init(void)
>  {
>  	pgtable_cache = kmem_cache_create("pgtable_cache",
>  					  PAGE_SIZE, PAGE_SIZE,
> -					  SLAB_HWCACHE_ALIGN,
> +					  0,
>  					  zero_ctor,
>  					  NULL);

You're patching code which your earlier patches deleted.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Christoph Lameter]

On Fri, 20 Apr 2007, Andrew Morton wrote:

> You're patching code which your earlier patches deleted.

Sorry. Trees overloaded with series of patches. Any change to get 
a new tree?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Andrew Morton]

On Fri, 20 Apr 2007 22:44:39 -0700 (PDT) Christoph Lameter <clameter@sgi.com> wrote:

> On Fri, 20 Apr 2007, Andrew Morton wrote:
> 
> > You're patching code which your earlier patches deleted.
> 
> Sorry. Trees overloaded with series of patches. Any change to get 
> a new tree?

rofl.

I'm still recovering from that dang Itanium conference.  Since rc6-mm1 I
have added 684 patches and removed 164.  It's simply idiotic.

http://userweb.kernel.org/~akpm/cl.bz2 is the current rollup against rc7. 
I haven't tried compiling it for nearly a week.  Good luck ;)

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Christoph Lameter]

On Fri, 20 Apr 2007, Andrew Morton wrote:

> > Sorry. Trees overloaded with series of patches. Any change to get 
> > a new tree?
> 
> rofl.
> 
> I'm still recovering from that dang Itanium conference.  Since rc6-mm1 I
> have added 684 patches and removed 164.  It's simply idiotic.
> 
> http://userweb.kernel.org/~akpm/cl.bz2 is the current rollup against rc7. 
> I haven't tried compiling it for nearly a week.  Good luck ;)

Well xpmem is broke and readahead is failing all over the place. Some 
patches missing?

Hmmmm... Revoke.c has another copy of these fs constructor flag checks 
that I fixed earlier.

Index: linux-2.6.21-rc7/fs/revoke.c
===================================================================
--- linux-2.6.21-rc7.orig/fs/revoke.c	2007-04-20 23:30:11.000000000 -0700
+++ linux-2.6.21-rc7/fs/revoke.c	2007-04-20 23:30:33.000000000 -0700
@@ -709,8 +709,7 @@ static void revokefs_init_inode(void *ob
 {
 	struct revokefs_inode_info *info = obj;
 
-	if ((flags & (SLAB_CTOR_VERIFY | SLAB_CTOR_CONSTRUCTOR)) ==
-	    SLAB_CTOR_CONSTRUCTOR) {
+	if (flags & SLAB_CTOR_CONSTRUCTOR) {
 		info->owner = NULL;
 		inode_init_once(&info->vfs_inode);
 	}

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Christoph Lameter]

On Fri, 20 Apr 2007, Christoph Lameter wrote:

> Hmmmm... Revoke.c has another copy of these fs constructor flag checks 
> that I fixed earlier.

And another one

Index: linux-2.6.21-rc7/fs/proc/inode.c
===================================================================
--- linux-2.6.21-rc7.orig/fs/proc/inode.c	2007-04-20 23:35:07.000000000 -0700
+++ linux-2.6.21-rc7/fs/proc/inode.c	2007-04-20 23:35:19.000000000 -0700
@@ -109,8 +109,7 @@ static void init_once(void * foo, struct
 {
 	struct proc_inode *ei = (struct proc_inode *) foo;
 
-	if ((flags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) ==
-	    SLAB_CTOR_CONSTRUCTOR)
+	if (flags & SLAB_CTOR_CONSTRUCTOR)
 		inode_init_once(&ei->vfs_inode);
 }
  

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Christoph Lameter]

Cannot get around the readahead problems:

 CC      mm/truncate.o
mm/readahead.c: In function 'page_cache_readahead':
mm/readahead.c:586: error: 'struct file_ra_state' has no member named 
'prev_page'
mm/readahead.c:587: error: 'struct file_ra_state' has no member named 
'prev_page'
mm/readahead.c: In function 'try_context_based_readahead':
mm/readahead.c:1414: error: 'struct file_ra_state' has no member named 
'prev_page'
mm/readahead.c: In function 'try_backward_prefetching':
mm/readahead.c:1534: error: 'struct file_ra_state' has no member named 
'prev_page'
mm/readahead.c: In function 'page_cache_readahead_adaptive':
mm/readahead.c:1687: error: 'struct file_ra_state' has no member named 
'prev_page'
mm/readahead.c:1693: error: 'struct file_ra_state' has no member named 
'prev_page'
mm/readahead.c:1738: error: 'struct file_ra_state' has no member named 
'prev_page'
make[3]: *** [mm/readahead.o] Error 1
make[3]: *** Waiting for unfinished jobs....
make[2]: *** [mm] Error 2

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Andrew Morton]

On Fri, 20 Apr 2007 23:32:48 -0700 (PDT) Christoph Lameter <clameter@sgi.com> wrote:

> On Fri, 20 Apr 2007, Andrew Morton wrote:
> 
> > > Sorry. Trees overloaded with series of patches. Any change to get 
> > > a new tree?
> > 
> > rofl.
> > 
> > I'm still recovering from that dang Itanium conference.  Since rc6-mm1 I
> > have added 684 patches and removed 164.  It's simply idiotic.
> > 
> > http://userweb.kernel.org/~akpm/cl.bz2 is the current rollup against rc7. 
> > I haven't tried compiling it for nearly a week.  Good luck ;)
> 
> Well xpmem is broke

What's xpmem?

> and readahead is failing all over the place.

runtime or compiletime?  If the latter, what are the symptoms?

(I'd be shocked if it compiled)

> Some 
> patches missing?

dude, this is why it takes ten-odd hours for me to get a -mm release out.

> Hmmmm... Revoke.c has another copy of these fs constructor flag checks 
> that I fixed earlier.

Yeah, I tossed that hunk because it was miles out of order.  I'll put this
bit into the right place.


> Index: linux-2.6.21-rc7/fs/revoke.c
> ===================================================================
> --- linux-2.6.21-rc7.orig/fs/revoke.c	2007-04-20 23:30:11.000000000 -0700
> +++ linux-2.6.21-rc7/fs/revoke.c	2007-04-20 23:30:33.000000000 -0700
> @@ -709,8 +709,7 @@ static void revokefs_init_inode(void *ob
>  {
>  	struct revokefs_inode_info *info = obj;
>  
> -	if ((flags & (SLAB_CTOR_VERIFY | SLAB_CTOR_CONSTRUCTOR)) ==
> -	    SLAB_CTOR_CONSTRUCTOR) {
> +	if (flags & SLAB_CTOR_CONSTRUCTOR) {
>  		info->owner = NULL;
>  		inode_init_once(&info->vfs_inode);
>  	}

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Andrew Morton]

On Fri, 20 Apr 2007 23:38:43 -0700 (PDT) Christoph Lameter <clameter@sgi.com> wrote:

> Cannot get around the readahead problems:
> 
>  CC      mm/truncate.o
> mm/readahead.c: In function 'page_cache_readahead':
> mm/readahead.c:586: error: 'struct file_ra_state' has no member named 
> 'prev_page'
> mm/readahead.c:587: error: 'struct file_ra_state' has no member named 
> 'prev_page'
> mm/readahead.c: In function 'try_context_based_readahead':
> mm/readahead.c:1414: error: 'struct file_ra_state' has no member named 
> 'prev_page'
> mm/readahead.c: In function 'try_backward_prefetching':
> mm/readahead.c:1534: error: 'struct file_ra_state' has no member named 
> 'prev_page'
> mm/readahead.c: In function 'page_cache_readahead_adaptive':
> mm/readahead.c:1687: error: 'struct file_ra_state' has no member named 
> 'prev_page'
> mm/readahead.c:1693: error: 'struct file_ra_state' has no member named 
> 'prev_page'
> mm/readahead.c:1738: error: 'struct file_ra_state' has no member named 
> 'prev_page'

prev_page got renamed to prev_index.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Christoph Lameter]

On Fri, 20 Apr 2007, Andrew Morton wrote:

> > Well xpmem is broke
> 
> What's xpmem?

Its an SSI->SSI memory mapping / communication driver.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Robin Holt]

On Fri, Apr 20, 2007 at 11:32:48PM -0700, Christoph Lameter wrote:
> Well xpmem is broke and readahead is failing all over the place. Some 
> patches missing?

Which xpmem are you compiling?  Did you let Dean Nelson know about this?
Has anything been submitted to the community yet?

Thanks,
Robin

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Christoph Lameter]

On Mon, 23 Apr 2007, Robin Holt wrote:

> On Fri, Apr 20, 2007 at 11:32:48PM -0700, Christoph Lameter wrote:
> > Well xpmem is broke and readahead is failing all over the place. Some 
> > patches missing?
> 
> Which xpmem are you compiling?  Did you let Dean Nelson know about this?
> Has anything been submitted to the community yet?

I think this was just a an artifact of an inconsistent ia64 mix of 
patches in a temporary tree by Andrew.

Here is the hack that I used to compile more of the temp tree.


Index: linux-2.6.21-rc7/arch/ia64/sn/kernel/xpc_main.c
===================================================================
--- linux-2.6.21-rc7.orig/arch/ia64/sn/kernel/xpc_main.c	2007-04-20 23:23:31.000000000 -0700
+++ linux-2.6.21-rc7/arch/ia64/sn/kernel/xpc_main.c	2007-04-20 23:25:32.000000000 -0700
@@ -811,6 +811,8 @@ xpc_create_kthreads(struct xpc_channel *
 	pid_t pid;
 	u64 args = XPC_PACK_ARGS(ch->partid, ch->number);
 	struct xpc_partition *part = &xpc_partitions[ch->partid];
+	struct task_struct *task;
+
 
 
 	while (needed-- > 0) {
@@ -839,7 +841,7 @@ xpc_create_kthreads(struct xpc_channel *
 		xpc_msgqueue_ref(ch);
 
 		task = kthread_run(xpc_daemonize_kthread, args,
-				   "xpc%02dc%d", partid, ch_number);
+				   "xpc%02dc%d", ch->partid, ch->number);
 		if (IS_ERR(task)) {
 			/* the fork failed */
 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Robin Holt]

On Mon, Apr 23, 2007 at 08:53:09AM -0700, Christoph Lameter wrote:
> On Mon, 23 Apr 2007, Robin Holt wrote:
> 
> > On Fri, Apr 20, 2007 at 11:32:48PM -0700, Christoph Lameter wrote:
> > > Well xpmem is broke and readahead is failing all over the place. Some 
> > > patches missing?
> > 
> > Which xpmem are you compiling?  Did you let Dean Nelson know about this?
> > Has anything been submitted to the community yet?
> 
> I think this was just a an artifact of an inconsistent ia64 mix of 
> patches in a temporary tree by Andrew.
> 
> Here is the hack that I used to compile more of the temp tree.
> 
> 
> Index: linux-2.6.21-rc7/arch/ia64/sn/kernel/xpc_main.c
> ===================================================================
> --- linux-2.6.21-rc7.orig/arch/ia64/sn/kernel/xpc_main.c	2007-04-20 23:23:31.000000000 -0700
> +++ linux-2.6.21-rc7/arch/ia64/sn/kernel/xpc_main.c	2007-04-20 23:25:32.000000000 -0700

Ah, xpc.  Much different from xpmem.  Now I understand...

Thanks,
Robin

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: slab allocators: Remove multiple alignment specifications.

[Dean Nelson]

On Mon, Apr 23, 2007 at 08:53:09AM -0700, Christoph Lameter wrote:
> On Mon, 23 Apr 2007, Robin Holt wrote:
> 
> > On Fri, Apr 20, 2007 at 11:32:48PM -0700, Christoph Lameter wrote:
> > > Well xpmem is broke and readahead is failing all over the place. Some 
> > > patches missing?
> > 
> > Which xpmem are you compiling?  Did you let Dean Nelson know about this?
> > Has anything been submitted to the community yet?
> 
> I think this was just a an artifact of an inconsistent ia64 mix of 
> patches in a temporary tree by Andrew.
> 
> Here is the hack that I used to compile more of the temp tree.
> 
> 
> Index: linux-2.6.21-rc7/arch/ia64/sn/kernel/xpc_main.c
> ===================================================================
> --- linux-2.6.21-rc7.orig/arch/ia64/sn/kernel/xpc_main.c	2007-04-20 23:23:31.000000000 -0700
> +++ linux-2.6.21-rc7/arch/ia64/sn/kernel/xpc_main.c	2007-04-20 23:25:32.000000000 -0700
> @@ -811,6 +811,8 @@ xpc_create_kthreads(struct xpc_channel *
>  	pid_t pid;
>  	u64 args = XPC_PACK_ARGS(ch->partid, ch->number);
>  	struct xpc_partition *part = &xpc_partitions[ch->partid];
> +	struct task_struct *task;
> +
>  
>  
>  	while (needed-- > 0) {
> @@ -839,7 +841,7 @@ xpc_create_kthreads(struct xpc_channel *
>  		xpc_msgqueue_ref(ch);
>  
>  		task = kthread_run(xpc_daemonize_kthread, args,
> -				   "xpc%02dc%d", partid, ch_number);
> +				   "xpc%02dc%d", ch->partid, ch->number);
>  		if (IS_ERR(task)) {
>  			/* the fork failed */
>  

Acked-by: Dean Nelson <dcn@sgi.com>

Andrew, this patch is a proper fix for a couple of compiler errors introduced
by a patch entitled ia64-sn-xpc-convert-to-use-kthread-api. You called the
following patch ia64-sn-xpc-convert-to-use-kthread-api-fix.

Thanks,
Dean

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>