[patch] Memory Binding API v0.2

[patch] Memory Binding API v0.2

[Matthew Dobson]

[-- Attachment #1: Type: text/plain, Size: 538 bytes --]

Here is the latest version of the Mem Binding API.  It's a follow-up to the 
patch posted a week or so ago.  It incorporates some changes, and should be a 
bit more efficient, readable, and functional.  Bigger, better, faster, eh?  It 
needs to be patched on top of the Simple binding API that I posted a minute ago..

This API as well as the Simple Binding were both originally part of a NUMA 
Binding API discussed at length on LSE-Tech, and met with approval there.  I 
hope that these can be included in the mainline.

Cheers!

-Matt

[-- Attachment #2: mem_api-v0.2-2.5.28.patch --]
[-- Type: text/plain, Size: 11165 bytes --]

diff -Nur linux-2.5.27-vanilla/arch/i386/config.in linux-2.5.27-api/arch/i386/config.in
--- linux-2.5.27-vanilla/arch/i386/config.in	Wed Jul 24 17:33:41 2002
+++ linux-2.5.27-api/arch/i386/config.in	Wed Jul 24 17:38:34 2002
@@ -168,6 +168,8 @@
    bool 'Multiquad NUMA system' CONFIG_X86_NUMAQ
    if [ "$CONFIG_X86_NUMAQ" = y ]; then
       define_bool CONFIG_MULTIQUAD y
+      bool 'Memory Binding API Support' CONFIG_MEMBIND
+      bool 'NUMA Memory Allocation Support' CONFIG_NUMA
    fi
 fi
 
diff -Nur linux-2.5.27-vanilla/arch/i386/kernel/entry.S linux-2.5.27-api/arch/i386/kernel/entry.S
--- linux-2.5.27-vanilla/arch/i386/kernel/entry.S	Wed Jul 24 17:33:41 2002
+++ linux-2.5.27-api/arch/i386/kernel/entry.S	Wed Jul 24 17:38:34 2002
@@ -754,6 +754,8 @@
 	.long sys_sched_setaffinity
 	.long sys_sched_getaffinity
 	.long sys_check_topology
+	.long sys_mem_setbinding
+	.long sys_mem_getbinding
 
 	.rept NR_syscalls-(.-sys_call_table)/4
 		.long sys_ni_syscall
diff -Nur linux-2.5.27-vanilla/include/asm-i386/unistd.h linux-2.5.27-api/include/asm-i386/unistd.h
--- linux-2.5.27-vanilla/include/asm-i386/unistd.h	Wed Jul 24 17:33:41 2002
+++ linux-2.5.27-api/include/asm-i386/unistd.h	Wed Jul 24 17:38:34 2002
@@ -248,6 +248,8 @@
 #define __NR_sched_setaffinity	241
 #define __NR_sched_getaffinity	242
 #define __NR_check_topology	243
+#define __NR_mem_setbinding	244
+#define __NR_mem_getbinding	245
 
 /* user-visible error numbers are in the range -1 - -124: see <asm-i386/errno.h> */
 
diff -Nur linux-2.5.27-vanilla/include/linux/init_task.h linux-2.5.27-api/include/linux/init_task.h
--- linux-2.5.27-vanilla/include/linux/init_task.h	Sat Jul 20 12:11:07 2002
+++ linux-2.5.27-api/include/linux/init_task.h	Wed Jul 24 17:38:34 2002
@@ -59,6 +59,11 @@
     children:		LIST_HEAD_INIT(tsk.children),			\
     sibling:		LIST_HEAD_INIT(tsk.sibling),			\
     thread_group:	LIST_HEAD_INIT(tsk.thread_group),		\
+    memblk_binding:	{						\
+	bitmask:	MEMBLK_NO_BINDING,				\
+	behavior:	MPOL_STRICT,					\
+	lock:		SPIN_LOCK_UNLOCKED				\
+    },									\
     wait_chldexit:	__WAIT_QUEUE_HEAD_INITIALIZER(tsk.wait_chldexit),\
     real_timer:		{						\
 	function:		it_real_fn				\
diff -Nur linux-2.5.27-vanilla/include/linux/membind.h linux-2.5.27-api/include/linux/membind.h
--- linux-2.5.27-vanilla/include/linux/membind.h	Wed Dec 31 16:00:00 1969
+++ linux-2.5.27-api/include/linux/membind.h	Wed Jul 24 17:38:34 2002
@@ -0,0 +1,51 @@
+/*
+ * linux/include/linux/membind.h
+ *
+ * Written by: Matthew Dobson, IBM Corporation
+ *
+ * Copyright (C) 2002, IBM Corp.
+ *
+ * All rights reserved.          
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
+ * NON INFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ * Send feedback to <colpatch@us.ibm.com>
+ */
+#ifndef _LINUX_MEMBIND_H
+#define _LINUX_MEMBIND_H
+
+#include <linux/types.h>
+
+#define MEMBLK_NO_BINDING	(~0UL)
+
+typedef struct memblk_list {
+	unsigned long bitmask;
+	int behavior;
+	spinlock_t lock;
+} memblk_list_t;
+
+
+#define is_valid_memblk_behavior(x) (1) /* for now */
+#define is_memblk_subset(x, y) (!(~(x) & (y)))  /* test whether x is a subset of y */
+
+#define MPOL_STRICT	0   /* Memory MUST be allocated according to binding */
+#define MPOL_LOOSE	1   /* Memory will be allocated according to binding, but
+				can fall back to other memory blocks if necessary. */
+#define MPOL_FIRST	2   /* UNUSED FOR NOW */
+#define MPOL_STRIPE	4   /* UNUSED FOR NOW */
+#define MPOL_RR		8   /* UNUSED FOR NOW */
+
+#endif /* _LINUX_MEMBIND_H */
diff -Nur linux-2.5.27-vanilla/include/linux/mmzone.h linux-2.5.27-api/include/linux/mmzone.h
--- linux-2.5.27-vanilla/include/linux/mmzone.h	Wed Jul 24 17:33:41 2002
+++ linux-2.5.27-api/include/linux/mmzone.h	Wed Jul 24 17:38:34 2002
@@ -138,6 +138,7 @@
 	unsigned long node_start_mapnr;
 	unsigned long node_size;
 	int node_id;
+	int memblk_id; /* A unique ID for each memory block */
 	struct pglist_data *node_next;
 } pg_data_t;
 
diff -Nur linux-2.5.27-vanilla/include/linux/sched.h linux-2.5.27-api/include/linux/sched.h
--- linux-2.5.27-vanilla/include/linux/sched.h	Sat Jul 20 12:11:07 2002
+++ linux-2.5.27-api/include/linux/sched.h	Wed Jul 24 17:38:34 2002
@@ -27,6 +27,7 @@
 #include <linux/securebits.h>
 #include <linux/fs_struct.h>
 #include <linux/compiler.h>
+#include <linux/membind.h>
 
 struct exec_domain;
 
@@ -302,6 +303,9 @@
 	struct task_struct *pidhash_next;
 	struct task_struct **pidhash_pprev;
 
+	/* additional Memory Binding stuff */
+	memblk_list_t memblk_binding;
+	
 	wait_queue_head_t wait_chldexit;	/* for wait4() */
 	struct completion *vfork_done;		/* for vfork() */
 
diff -Nur linux-2.5.27-vanilla/kernel/sys.c linux-2.5.27-api/kernel/sys.c
--- linux-2.5.27-vanilla/kernel/sys.c	Wed Jul 24 17:33:41 2002
+++ linux-2.5.27-api/kernel/sys.c	Wed Jul 24 17:38:34 2002
@@ -1263,6 +1263,57 @@
 	return (long)ret;
 }
     
+/*
+ * sys_mem_setbinding(): Sets up a new MemBlk Binding
+ */
+asmlinkage long sys_mem_setbinding(unsigned long memblks, int behavior)
+{
+	long ret;
+	unsigned long flags;
+	struct task_struct *curr = current;
+
+	ret = -ENODEV;
+	/* Make sure that at least one of the memblks in the new binding set is online. */
+	if (!(memblks & memblk_online_map))
+		goto out;
+
+	ret = -EINVAL;
+	/* Test to make sure the behavior argument is valid. */
+	if (!is_valid_memblk_behavior(behavior))
+		goto out;
+
+	ret = -EPERM;
+	spin_lock_irqsave(&curr->memblk_binding.lock, flags);
+	/* If the new binding expands upon the old binding, the caller
+	   must have CAP_SYS_NICE. */
+	if (is_memblk_subset(memblks, curr->memblk_binding.bitmask) ||
+	    capable(CAP_SYS_NICE)){
+		curr->memblk_binding.bitmask = memblks;
+		curr->memblk_binding.behavior = behavior;
+		ret = 0;
+	}
+	spin_unlock_irqrestore(&curr->memblk_binding.lock, flags);
+
+ out:
+	return ret;
+}
+
+/*
+ * sys_mem_getbinding(): Returns the current MemBlk Binding
+ */
+asmlinkage long sys_mem_getbinding(void)
+{
+	long flags;
+	unsigned long memblk_binding;
+	struct task_struct *curr = current;
+
+	spin_lock_irqsave(&curr->memblk_binding.lock, flags);
+	memblk_binding = curr->memblk_binding.bitmask;
+	spin_unlock_irqrestore(&curr->memblk_binding.lock, flags);
+
+	return memblk_binding;
+}
+
 asmlinkage long sys_prctl(int option, unsigned long arg2, unsigned long arg3,
 			  unsigned long arg4, unsigned long arg5)
 {
diff -Nur linux-2.5.27-vanilla/mm/numa.c linux-2.5.27-api/mm/numa.c
--- linux-2.5.27-vanilla/mm/numa.c	Sat Jul 20 12:11:12 2002
+++ linux-2.5.27-api/mm/numa.c	Wed Jul 24 17:38:34 2002
@@ -8,6 +8,7 @@
 #include <linux/bootmem.h>
 #include <linux/mmzone.h>
 #include <linux/spinlock.h>
+#include <linux/membind.h>
 
 int numnodes = 1;	/* Initialized for UMA platforms */
 
@@ -27,6 +28,9 @@
 {
 	free_area_init_core(0, &contig_page_data, &mem_map, zones_size, 
 				zone_start_paddr, zholes_size, pmap);
+	contig_page_data.node_id = 0;
+	contig_page_data.memblk_id = 0;
+	memblk_online_map = 1UL;
 }
 
 #endif /* !CONFIG_DISCONTIGMEM */
@@ -71,6 +75,11 @@
 	free_area_init_core(nid, pgdat, &discard, zones_size, zone_start_paddr,
 					zholes_size, pmap);
 	pgdat->node_id = nid;
+	pgdat->memblk_id = num_online_memblks();
+	if (test_and_set_bit(num_online_memblks() + 1, &memblk_online_map)){
+		printk("memblk alread counted?!?!\n");
+		BUG();
+	}
 
 	/*
 	 * Get space for the valid bitmap.
@@ -88,6 +97,8 @@
 	return __alloc_pages(gfp_mask, order, pgdat->node_zonelists + (gfp_mask & GFP_ZONEMASK));
 }
 
+#ifdef CONFIG_NUMA
+
 /*
  * This can be refined. Currently, tries to do round robin, instead
  * should do concentratic circle search, starting from current node.
@@ -96,23 +107,67 @@
 {
 	struct page *ret = 0;
 	pg_data_t *start, *temp;
-#ifndef CONFIG_NUMA
+	int search_twice = 0;
+	unsigned long memblk_mask;
+	struct task_struct *curr = current;
 	unsigned long flags;
-	static pg_data_t *next = 0;
-#endif
 
 	if (order >= MAX_ORDER)
 		return NULL;
-#ifdef CONFIG_NUMA
+
+	spin_lock_irqsave(&curr->memblk_binding.lock, flags);
+	memblk_mask = curr->memblk_binding.bitmask;
+	/* if it is a loose binding, remember to search other memblks */
+	if ((curr->memblk_binding.behavior == MPOL_LOOSE) &&
+	    (curr->memblk_binding.bitmask != MEMBLK_NO_BINDING))
+		 search_twice = 1;
+	spin_unlock_irqrestore(&curr->memblk_binding.lock, flags);
+
+search_through_memblks: 
 	temp = NODE_DATA(numa_node_id());
-#else
-	spin_lock_irqsave(&node_lock, flags);
+	start = temp;
+	while (temp) {
+		if (memblk_mask & (1 << temp->memblk_id))
+			if ((ret = alloc_pages_pgdat(temp, gfp_mask, order)))
+				return(ret);
+		temp = temp->node_next;
+	}
+	temp = pgdat_list;
+	while (temp != start) {
+		if (!(memblk_mask & (1 << temp->memblk_id)))
+			if ((ret = alloc_pages_pgdat(temp, gfp_mask, order)))
+				return(ret);
+		temp = temp->node_next;
+	}
+
+	if (search_twice) {
+		/* 
+		 * If we failed to find a "preferred" memblk, try again 
+		 * looking for anything we haven't checked yet.
+		 */
+		search_twice = 0; /* no infinite loops, please */
+		memblk_mask = ~memblk_mask;
+		goto search_through_memblks;
+	}
+	return(0);
+}
+
+#else /* !CONFIG_NUMA */
+
+struct page * _alloc_pages(unsigned int gfp_mask, unsigned int order)
+{
+	struct page *ret = 0;
+	pg_data_t *start, *temp;
+	static pg_data_t *next = 0;
+	unsigned long flags;
+
+	if (order >= MAX_ORDER)
+		return NULL;
+
 	if (!next) next = pgdat_list;
-	temp = next;
+	temp = start = next;
 	next = next->node_next;
-	spin_unlock_irqrestore(&node_lock, flags);
-#endif
-	start = temp;
+
 	while (temp) {
 		if ((ret = alloc_pages_pgdat(temp, gfp_mask, order)))
 			return(ret);
@@ -127,4 +182,6 @@
 	return(0);
 }
 
+#endif /* CONFIG_NUMA */
+
 #endif /* CONFIG_DISCONTIGMEM */
diff -Nur linux-2.5.27-vanilla/mm/page_alloc.c linux-2.5.27-api/mm/page_alloc.c
--- linux-2.5.27-vanilla/mm/page_alloc.c	Sat Jul 20 12:11:07 2002
+++ linux-2.5.27-api/mm/page_alloc.c	Wed Jul 24 17:38:34 2002
@@ -42,6 +42,8 @@
 static int zone_balance_min[MAX_NR_ZONES] __initdata = { 20 , 20, 20, };
 static int zone_balance_max[MAX_NR_ZONES] __initdata = { 255 , 255, 255, };
 
+extern unsigned long memblk_online_map;
+
 /*
  * Temporary debugging check for pages not lying within a given zone.
  */
@@ -927,6 +929,9 @@
 void __init free_area_init(unsigned long *zones_size)
 {
 	free_area_init_core(0, &contig_page_data, &mem_map, zones_size, 0, 0, 0);
+	contig_page_data.node_id = 0;
+	contig_page_data.memblk_id = 0;
+	memblk_online_map = 1UL;
 }
 
 static int __init setup_mem_frac(char *str)

Re: [Lse-tech] [patch] Memory Binding API v0.2

[Erich Focht]

Hi Matt,

On Friday 26 July 2002 00:40, Matthew Dobson wrote:
> Here is the latest version of the Mem Binding API.  It's a follow-up to the
> patch posted a week or so ago.  It incorporates some changes, and should be
> a bit more efficient, readable, and functional.  Bigger, better, faster,
> eh?  It needs to be patched on top of the Simple binding API that I posted
> a minute ago..

the patch is a good start for introducing the NUMA API which we definitely
need for some coherence among NUMA developments. Also I think stuff like
memory hot-add/remove will absolutely need a concept like memblk.

I'm just having some small technical comments:

You're using spinlocks for protecting the memblk structure. Mostly these
structures would be just read during the lifetime of a task. Besides: the
current syscalls allow changing them only from the current context, there
is no danger of getting the wrong values here, because current isn't
allocating pages while it changes its memblk_list. Even if you change
the memblk variables from another task, it isn't that bad if they're
unprotected. After all just before changing the values you might have
allocated memory by using the old values... I think we can live without
the spinlocks.

The implemented syscalls only change the memblk_list of the current task.
Would you please consider extending them to arbitrary PIDs? With
reasonable protection, of course.

The include file include/linux/membind.h is kind of small and isn't
supposed to grow too much. How about putting this into
include/linux/numa.h where we could gather some more NUMA stuff and more
things to come with the NUMA API?

In _alloc_pages() you might want to check for online memblks when
initialising the mask:
	memblk_mask = curr->memblk_binding.bitmask & memblk_online_map;
In the search_twice branch I'd reset the mask to memblk_online_map. There
are chances that the desired memblk has been freed in the mean time.

Do you have plans for adding the setlaunch() part of the NUMA API? Guess
that should take care of both memblk_list and cpus_allowed...

Thanks,
best regards,

Erich