Re: [RFC][PATCH] hugetlb: add information and interface in sysfs [Was Re: [RFC][PATCH 4/5] Documentation: add node files to sysfs ABI]

[Nishanth Aravamudan <nacc@us.ibm.com>]

On 28.04.2008 [10:29:51 -0700], Greg KH wrote:
> On Mon, Apr 28, 2008 at 10:22:39AM -0700, Nishanth Aravamudan wrote:
> > On 26.04.2008 [22:10:29 -0700], Greg KH wrote:
> > > On Sat, Apr 26, 2008 at 08:49:42PM -0700, Nishanth Aravamudan wrote:
> > > > 
> > > > [20:41:56]nacc@arkanoid:/sys/kernel/hugepages$ cat /sys/kernel/hugepages/hugepages-2MB/meminfo
> > > > HugePages_Total:     0
> > > > HugePages_Free:      0
> > > > HugePages_Rsvd:      0
> > > > HugePages_Surp:      0
> > > > Hugepagesize:     2048 kB
> > > > 
> > > > Greg, do you see any obvious violations of sysfs rules here? Well, beyond
> > > > meminfo itself, I guess, but given our previous snapshot discussion, I left it
> > > > simple and the same, rather than split it up.
> > > 
> > > Yeah, I don't like that file.  Why not just have 5 files, one for each
> > > value?  There isn't such a need for an immediate snapshot shere you
> > > can't just read all 5 values from 5 files?
> > 
> > Actually, we already have Total in nr_hugepages, so I only needed to add
> > 3 files. The size is implicit in the directory name?
> 
> Ah, good point.
> 
> > > Also, why use a "units" here, just always use the lowest unit, and
> > > userspace can convert from kB to GB if needed.
> > 
> > Agreed, so I changed the name of the directory from
> > 
> > hugepages-2M
> > 
> > to
> > 
> > hugpeages-2048
> > 
> > for instance. Userspace utilities can pretty-ize it :)
> 
> Exactly, that would be much better.

FWIW, here's the updated patch. Still needs more testing.

 include/linux/hugetlb.h |    9 +-
 mm/hugetlb.c            |  322 ++++++++++++++++++++++++++++++++++++-----------
 2 files changed, 256 insertions(+), 75 deletions(-)

diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index 7aa22e7..cac63bd 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -3,6 +3,9 @@
 
 #include <linux/fs.h>
 #include <linux/shm.h>
+#include <linux/mempolicy.h>
+#include <asm/tlbflush.h>
+#include <asm/hugetlb.h>
 
 #ifdef CONFIG_HUGETLBFS
 struct hugetlbfs_config {
@@ -69,10 +72,6 @@ static inline void set_file_hugepages(struct file *file)
 
 #ifdef CONFIG_HUGETLB_PAGE
 
-#include <linux/mempolicy.h>
-#include <asm/tlbflush.h>
-#include <asm/hugetlb.h>
-
 struct ctl_table;
 
 static inline int is_vm_hugetlb_page(struct vm_area_struct *vma)
@@ -131,6 +130,8 @@ struct hstate {
 	unsigned int nr_huge_pages_node[MAX_NUMNODES];
 	unsigned int free_huge_pages_node[MAX_NUMNODES];
 	unsigned int surplus_huge_pages_node[MAX_NUMNODES];
+	const char *name;
+	struct kobject kobj;
 };
 
 void __init huge_add_hstate(unsigned order);
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index de03a14..1d94a85 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -15,6 +15,7 @@
 #include <linux/cpuset.h>
 #include <linux/mutex.h>
 #include <linux/bootmem.h>
+#include <linux/sysfs.h>
 
 #include <asm/page.h>
 #include <asm/pgtable.h>
@@ -624,6 +625,8 @@ static void __init hugetlb_init_hstate(struct hstate *h)
 			break;
 	}
 	h->max_huge_pages = i;
+	h->name = kasprintf(GFP_KERNEL, "hugepages-%lu",
+						huge_page_size(h) / 1024);
 }
 
 static void __init hugetlb_init_hstates(void)
@@ -662,77 +665,6 @@ static void __init report_hugepages(void)
         }
 }
 
-static int __init hugetlb_init(void)
-{
-	BUILD_BUG_ON(HPAGE_SHIFT == 0);
-
-	if (!size_to_hstate(HPAGE_SIZE)) {
-		huge_add_hstate(HUGETLB_PAGE_ORDER);
-		parsed_hstate->max_huge_pages = default_hstate_resv;
-	}
-
-	hugetlb_init_hstates();
-
-	gather_bootmem_prealloc();
-
-	report_hugepages();
-
-	return 0;
-}
-module_init(hugetlb_init);
-
-/* Should be called on processing a hugepagesz=... option */
-void __init huge_add_hstate(unsigned order)
-{
-	struct hstate *h;
-	if (size_to_hstate(PAGE_SIZE << order)) {
-		printk("hugepagesz= specified twice, ignoring\n");
-		return;
-	}
-	BUG_ON(max_hstate >= HUGE_MAX_HSTATE);
-	BUG_ON(order < HPAGE_SHIFT - PAGE_SHIFT);
-	h = &hstates[max_hstate++];
-	h->order = order;
-	h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1);
-	hugetlb_init_hstate(h);
-	parsed_hstate = h;
-}
-
-static int __init hugetlb_setup(char *s)
-{
-	unsigned long *mhp;
-
-	if (!max_hstate)
-		mhp = &default_hstate_resv;
-	else
-		mhp = &parsed_hstate->max_huge_pages;
-
-	if (sscanf(s, "%lu", mhp) <= 0)
-		*mhp = 0;
-
-	/*
-	 * Global state is always initialized later in hugetlb_init.
-	 * But we need to allocate >= MAX_ORDER hstates here early to still
-	 * use the bootmem allocator.
-	 */
-	if (max_hstate > 0 && parsed_hstate->order >= MAX_ORDER)
-		hugetlb_init_hstate(parsed_hstate);
-
-	return 1;
-}
-__setup("hugepages=", hugetlb_setup);
-
-static unsigned int cpuset_mems_nr(unsigned int *array)
-{
-	int node;
-	unsigned int nr = 0;
-
-	for_each_node_mask(node, cpuset_current_mems_allowed)
-		nr += array[node];
-
-	return nr;
-}
-
 #ifdef CONFIG_SYSCTL
 #ifdef CONFIG_HIGHMEM
 static void try_to_free_low(struct hstate *h, unsigned long count)
@@ -843,6 +775,254 @@ out:
 	return ret;
 }
 
+#ifdef CONFIG_SYSFS
+#define to_hstate_attr(n) container_of(n, struct hstate_attribute, attr)
+#define to_hstate(n) container_of(n, struct hstate, kobj)
+
+struct hstate_attribute {
+	struct attribute attr;
+	ssize_t (*show)(struct hstate *h, char *buf);
+	ssize_t (*store)(struct hstate *h, const char *buf, size_t count);
+};
+
+#define HSTATE_ATTR_RO(_name) \
+	static struct hstate_attribute _name##_attr = __ATTR_RO(_name)
+
+#define HSTATE_ATTR(_name) \
+	static struct hstate_attribute _name##_attr = \
+		__ATTR(_name, 0644, _name##_show, _name##_store)
+
+static ssize_t nr_hugepages_show(struct hstate *h, char *buf)
+{
+	return sprintf(buf, "%lu\n", h->nr_huge_pages);
+}
+static ssize_t nr_hugepages_store(struct hstate *h, const char *buf,
+								size_t count)
+{
+	int tmp, err;
+	unsigned long input;
+	err = strict_strtoul(buf, 10, &input);
+	if (err)
+		return 0;
+
+	h->max_huge_pages = set_max_huge_pages(h, input, &tmp);
+	max_huge_pages[h - hstates] = h->max_huge_pages;
+	return count;
+}
+HSTATE_ATTR(nr_hugepages);
+
+static ssize_t nr_overcommit_hugepages_show(struct hstate *h, char *buf)
+{
+	return sprintf(buf, "%lu\n", h->nr_overcommit_huge_pages);
+}
+static ssize_t nr_overcommit_hugepages_store(struct hstate *h, const char *buf,
+								size_t count)
+{
+	unsigned long input;
+	int err;
+
+	err = strict_strtoul(buf, 10, &input);
+	if (err)
+		return 0;
+
+	spin_lock(&hugetlb_lock);
+	h->nr_overcommit_huge_pages = input;
+	sysctl_overcommit_huge_pages[h - hstates] = h->nr_overcommit_huge_pages;
+	spin_unlock(&hugetlb_lock);
+	return count;
+}
+HSTATE_ATTR(nr_overcommit_hugepages);
+
+static ssize_t free_hugepages_show(struct hstate *h, char *buf)
+{
+	return sprintf(buf, "%lu\n", h->free_huge_pages);
+}
+HSTATE_ATTR_RO(free_hugepages);
+
+static ssize_t resv_hugepages_show(struct hstate *h, char *buf)
+{
+	return sprintf(buf, "%lu\n", h->resv_huge_pages);
+}
+HSTATE_ATTR_RO(resv_hugepages);
+
+static ssize_t surplus_hugepages_show(struct hstate *h, char *buf)
+{
+	return sprintf(buf, "%lu\n", h->surplus_huge_pages);
+}
+HSTATE_ATTR_RO(surplus_hugepages);
+
+static struct kset *hstate_kset;
+
+static struct attribute *hstate_attrs[] = {
+	&nr_hugepages_attr.attr,
+	&nr_overcommit_hugepages_attr.attr,
+	&free_hugepages_attr.attr,
+	&resv_hugepages_attr.attr,
+	&surplus_hugepages_attr.attr,
+};
+
+static struct attribute_group hstate_attr_group = {
+	.attrs = hstate_attrs,
+};
+
+static ssize_t hstate_attr_show(struct kobject *kobj,
+					struct attribute *attr,
+					char *buf)
+{
+	struct hstate_attribute *attribute;
+	struct hstate *h;
+	int err;
+
+	attribute = to_hstate_attr(attr);
+	h = to_hstate(kobj);
+
+	if (!attribute->show)
+		return -EIO;
+
+	err = attribute->show(h, buf);
+
+	return err;
+}
+
+static ssize_t hstate_attr_store(struct kobject *kobj,
+					struct attribute *attr,
+					const char *buf, size_t len)
+{
+	struct hstate_attribute *attribute;
+	struct hstate *h;
+	int err;
+
+	attribute = to_hstate_attr(attr);
+	h = to_hstate(kobj);
+
+	if (!attribute->store)
+		return -EIO;
+
+	err = attribute->store(h, buf, len);
+
+	return err;
+}
+
+static struct sysfs_ops hstate_sysfs_ops = {
+	.show = hstate_attr_show,
+	.store = hstate_attr_store,
+};
+
+static struct kobj_type hstate_ktype = {
+	.sysfs_ops = &hstate_sysfs_ops,
+};
+
+static int __init hugetlb_sysfs_add_hstate(struct hstate *h)
+{
+	int err;
+	h->kobj.kset = hstate_kset;
+	err = kobject_init_and_add(&h->kobj, &hstate_ktype, NULL, h->name);
+	if (err) {
+		kobject_put(&h->kobj);
+		return err;
+	}
+	err = sysfs_create_group(&h->kobj, &hstate_attr_group);
+	if (err)
+		return err;
+	return 0;
+}
+
+static void __init hugetlb_sysfs_init(void)
+{
+	struct hstate *h;
+	int err;
+
+	hstate_kset = kset_create_and_add("hugepages", NULL, kernel_kobj);
+	if (!hstate_kset)
+		return;
+
+	for_each_hstate(h) {
+		err = hugetlb_sysfs_add_hstate(h);
+		if (err)
+			printk(KERN_ERR "Hugetlb: Unable to add hstate %s",
+								h->name);
+	}
+}
+#else
+static void __init hugetlb_sysfs_init(void)
+{
+}
+#endif
+
+static int __init hugetlb_init(void)
+{
+	BUILD_BUG_ON(HPAGE_SHIFT == 0);
+
+	if (!size_to_hstate(HPAGE_SIZE)) {
+		huge_add_hstate(HUGETLB_PAGE_ORDER);
+		parsed_hstate->max_huge_pages = default_hstate_resv;
+	}
+
+	hugetlb_init_hstates();
+
+	gather_bootmem_prealloc();
+
+	report_hugepages();
+
+	hugetlb_sysfs_init();
+
+	return 0;
+}
+module_init(hugetlb_init);
+
+/* Should be called on processing a hugepagesz=... option */
+void __init huge_add_hstate(unsigned order)
+{
+	struct hstate *h;
+	if (size_to_hstate(PAGE_SIZE << order)) {
+		printk("hugepagesz= specified twice, ignoring\n");
+		return;
+	}
+	BUG_ON(max_hstate >= HUGE_MAX_HSTATE);
+	BUG_ON(order < HPAGE_SHIFT - PAGE_SHIFT);
+	h = &hstates[max_hstate++];
+	h->order = order;
+	h->mask = ~((1ULL << (order + PAGE_SHIFT)) - 1);
+	hugetlb_init_hstate(h);
+	parsed_hstate = h;
+}
+
+static int __init hugetlb_setup(char *s)
+{
+	unsigned long *mhp;
+
+	if (!max_hstate)
+		mhp = &default_hstate_resv;
+	else
+		mhp = &parsed_hstate->max_huge_pages;
+
+	if (sscanf(s, "%lu", mhp) <= 0)
+		*mhp = 0;
+
+	/*
+	 * Global state is always initialized later in hugetlb_init.
+	 * But we need to allocate >= MAX_ORDER hstates here early to still
+	 * use the bootmem allocator.
+	 */
+	if (max_hstate > 0 && parsed_hstate->order >= MAX_ORDER)
+		hugetlb_init_hstate(parsed_hstate);
+
+	return 1;
+}
+__setup("hugepages=", hugetlb_setup);
+
+static unsigned int cpuset_mems_nr(unsigned int *array)
+{
+	int node;
+	unsigned int nr = 0;
+
+	for_each_node_mask(node, cpuset_current_mems_allowed)
+		nr += array[node];
+
+	return nr;
+}
+
+
 int hugetlb_sysctl_handler(struct ctl_table *table, int write,
 			   struct file *file, void __user *buffer,
 			   size_t *length, loff_t *ppos)

-- 
Nishanth Aravamudan <nacc@us.ibm.com>
IBM Linux Technology Center

--
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>