Recent "Found the commit that causes the OOMs" discussion notice us that kernel
stack usage should be showed in OOM log.
At least, I think ;)
this patch provide it.
========
Subject: [PATCH] Show kernel stack usage to /proc/meminfo and OOM log
if the system have a lot of thread, kernel stack consume unignorable large size
memory.
IOW, it make a lot of unaccountable memory.
Tons unaccountable memory bring to harder analyse memory related trouble.
Then, kernel stack account is useful.
---
fs/proc/meminfo.c | 2 ++
include/linux/mmzone.h | 3 ++-
kernel/fork.c | 13 +++++++++++++
mm/page_alloc.c | 6 ++++--
mm/vmstat.c | 1 +
5 files changed, 22 insertions(+), 3 deletions(-)
diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c
index d5c410d..1fbf8c0 100644
--- a/fs/proc/meminfo.c
+++ b/fs/proc/meminfo.c
@@ -85,6 +85,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
"SReclaimable: %8lu kB\n"
"SUnreclaim: %8lu kB\n"
"PageTables: %8lu kB\n"
+ "KernelStack %8lu kB\n"
#ifdef CONFIG_QUICKLIST
"Quicklists: %8lu kB\n"
#endif
@@ -129,6 +130,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
K(global_page_state(NR_SLAB_RECLAIMABLE)),
K(global_page_state(NR_SLAB_UNRECLAIMABLE)),
K(global_page_state(NR_PAGETABLE)),
+ K(global_page_state(NR_KERNEL_STACK)),
#ifdef CONFIG_QUICKLIST
K(quicklist_total_size()),
#endif
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 8895985..d9335b8 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -94,10 +94,11 @@ enum zone_stat_item {
NR_SLAB_RECLAIMABLE,
NR_SLAB_UNRECLAIMABLE,
NR_PAGETABLE, /* used for pagetables */
+ NR_KERNEL_STACK,
+ /* Second 128 byte cacheline */
NR_UNSTABLE_NFS, /* NFS unstable pages */
NR_BOUNCE,
NR_VMSCAN_WRITE,
- /* Second 128 byte cacheline */
NR_WRITEBACK_TEMP, /* Writeback using temporary buffers */
#ifdef CONFIG_NUMA
NUMA_HIT, /* allocated in intended node */
diff --git a/kernel/fork.c b/kernel/fork.c
index 467746b..21cd4aa 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -137,9 +137,19 @@ struct kmem_cache *vm_area_cachep;
/* SLAB cache for mm_struct structures (tsk->mm) */
static struct kmem_cache *mm_cachep;
+static void account_kernel_stack(struct thread_info *ti, int on)
+{
+ struct zone* zone = page_zone(virt_to_page(ti));
+ int sign = on ? 1 : -1;
+ long acct = sign * (THREAD_SIZE / PAGE_SIZE);
+
+ mod_zone_page_state(zone, NR_KERNEL_STACK, acct);
+}
+
void free_task(struct task_struct *tsk)
{
prop_local_destroy_single(&tsk->dirties);
+ account_kernel_stack(tsk->stack, 0);
free_thread_info(tsk->stack);
rt_mutex_debug_task_free(tsk);
ftrace_graph_exit_task(tsk);
@@ -255,6 +265,9 @@ static struct task_struct *dup_task_struct(struct task_struct *orig)
tsk->btrace_seq = 0;
#endif
tsk->splice_pipe = NULL;
+
+ account_kernel_stack(ti, 1);
+
return tsk;
out:
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 30d5093..0edec1c 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2119,7 +2119,8 @@ void show_free_areas(void)
" inactive_file:%lu"
" unevictable:%lu"
" dirty:%lu writeback:%lu unstable:%lu\n"
- " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n",
+ " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n"
+ " kernel_stack:%lu\n",
global_page_state(NR_ACTIVE_ANON),
global_page_state(NR_ACTIVE_FILE),
global_page_state(NR_INACTIVE_ANON),
@@ -2133,7 +2134,8 @@ void show_free_areas(void)
global_page_state(NR_SLAB_UNRECLAIMABLE),
global_page_state(NR_FILE_MAPPED),
global_page_state(NR_PAGETABLE),
- global_page_state(NR_BOUNCE));
+ global_page_state(NR_BOUNCE),
+ global_page_state(NR_KERNEL_STACK));
for_each_populated_zone(zone) {
int i;
diff --git a/mm/vmstat.c b/mm/vmstat.c
index 138bed5..ceda39b 100644
--- a/mm/vmstat.c
+++ b/mm/vmstat.c
@@ -639,6 +639,7 @@ static const char * const vmstat_text[] = {
"nr_slab_reclaimable",
"nr_slab_unreclaimable",
"nr_page_table_pages",
+ "nr_kernel_stack",
"nr_unstable",
"nr_bounce",
"nr_vmscan_write",
--
1.6.0.GIT
> Recent "Found the commit that causes the OOMs" discussion notice us that kernel
> stack usage should be showed in OOM log.
>
> At least, I think ;)
>
> this patch provide it.
>
>
> ========
> Subject: [PATCH] Show kernel stack usage to /proc/meminfo and OOM log
>
> if the system have a lot of thread, kernel stack consume unignorable large size
> memory.
> IOW, it make a lot of unaccountable memory.
>
> Tons unaccountable memory bring to harder analyse memory related trouble.
>
> Then, kernel stack account is useful.
I forgot to insert most important one line ;-)
Signed-off-by: KOSAKI Motohiro <[email protected]>
> ---
> fs/proc/meminfo.c | 2 ++
> include/linux/mmzone.h | 3 ++-
> kernel/fork.c | 13 +++++++++++++
> mm/page_alloc.c | 6 ++++--
> mm/vmstat.c | 1 +
> 5 files changed, 22 insertions(+), 3 deletions(-)
>
> diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c
> index d5c410d..1fbf8c0 100644
> --- a/fs/proc/meminfo.c
> +++ b/fs/proc/meminfo.c
> @@ -85,6 +85,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
> "SReclaimable: %8lu kB\n"
> "SUnreclaim: %8lu kB\n"
> "PageTables: %8lu kB\n"
> + "KernelStack %8lu kB\n"
> #ifdef CONFIG_QUICKLIST
> "Quicklists: %8lu kB\n"
> #endif
> @@ -129,6 +130,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
> K(global_page_state(NR_SLAB_RECLAIMABLE)),
> K(global_page_state(NR_SLAB_UNRECLAIMABLE)),
> K(global_page_state(NR_PAGETABLE)),
> + K(global_page_state(NR_KERNEL_STACK)),
> #ifdef CONFIG_QUICKLIST
> K(quicklist_total_size()),
> #endif
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index 8895985..d9335b8 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -94,10 +94,11 @@ enum zone_stat_item {
> NR_SLAB_RECLAIMABLE,
> NR_SLAB_UNRECLAIMABLE,
> NR_PAGETABLE, /* used for pagetables */
> + NR_KERNEL_STACK,
> + /* Second 128 byte cacheline */
> NR_UNSTABLE_NFS, /* NFS unstable pages */
> NR_BOUNCE,
> NR_VMSCAN_WRITE,
> - /* Second 128 byte cacheline */
> NR_WRITEBACK_TEMP, /* Writeback using temporary buffers */
> #ifdef CONFIG_NUMA
> NUMA_HIT, /* allocated in intended node */
> diff --git a/kernel/fork.c b/kernel/fork.c
> index 467746b..21cd4aa 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -137,9 +137,19 @@ struct kmem_cache *vm_area_cachep;
> /* SLAB cache for mm_struct structures (tsk->mm) */
> static struct kmem_cache *mm_cachep;
>
> +static void account_kernel_stack(struct thread_info *ti, int on)
> +{
> + struct zone* zone = page_zone(virt_to_page(ti));
> + int sign = on ? 1 : -1;
> + long acct = sign * (THREAD_SIZE / PAGE_SIZE);
> +
> + mod_zone_page_state(zone, NR_KERNEL_STACK, acct);
> +}
> +
> void free_task(struct task_struct *tsk)
> {
> prop_local_destroy_single(&tsk->dirties);
> + account_kernel_stack(tsk->stack, 0);
> free_thread_info(tsk->stack);
> rt_mutex_debug_task_free(tsk);
> ftrace_graph_exit_task(tsk);
> @@ -255,6 +265,9 @@ static struct task_struct *dup_task_struct(struct task_struct *orig)
> tsk->btrace_seq = 0;
> #endif
> tsk->splice_pipe = NULL;
> +
> + account_kernel_stack(ti, 1);
> +
> return tsk;
>
> out:
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 30d5093..0edec1c 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -2119,7 +2119,8 @@ void show_free_areas(void)
> " inactive_file:%lu"
> " unevictable:%lu"
> " dirty:%lu writeback:%lu unstable:%lu\n"
> - " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n",
> + " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n"
> + " kernel_stack:%lu\n",
> global_page_state(NR_ACTIVE_ANON),
> global_page_state(NR_ACTIVE_FILE),
> global_page_state(NR_INACTIVE_ANON),
> @@ -2133,7 +2134,8 @@ void show_free_areas(void)
> global_page_state(NR_SLAB_UNRECLAIMABLE),
> global_page_state(NR_FILE_MAPPED),
> global_page_state(NR_PAGETABLE),
> - global_page_state(NR_BOUNCE));
> + global_page_state(NR_BOUNCE),
> + global_page_state(NR_KERNEL_STACK));
>
> for_each_populated_zone(zone) {
> int i;
> diff --git a/mm/vmstat.c b/mm/vmstat.c
> index 138bed5..ceda39b 100644
> --- a/mm/vmstat.c
> +++ b/mm/vmstat.c
> @@ -639,6 +639,7 @@ static const char * const vmstat_text[] = {
> "nr_slab_reclaimable",
> "nr_slab_unreclaimable",
> "nr_page_table_pages",
> + "nr_kernel_stack",
> "nr_unstable",
> "nr_bounce",
> "nr_vmscan_write",
> --
> 1.6.0.GIT
>
>
>
>
On Tue, 30 Jun 2009, KOSAKI Motohiro wrote:
> +static void account_kernel_stack(struct thread_info *ti, int on)
static inline?
> +{
> + struct zone* zone = page_zone(virt_to_page(ti));
> + int sign = on ? 1 : -1;
> + long acct = sign * (THREAD_SIZE / PAGE_SIZE);
int pages = THREAD_SIZE / PAGE_SIZE;
?
> +
> + mod_zone_page_state(zone, NR_KERNEL_STACK, acct);
mod_zone_page_state(zone, NR_KERNEL_STACK, on ? pages : -pages);
> On Tue, 30 Jun 2009, KOSAKI Motohiro wrote:
>
> > +static void account_kernel_stack(struct thread_info *ti, int on)
>
> static inline?
gcc automatically inlined, IMHO.
> > +{
> > + struct zone* zone = page_zone(virt_to_page(ti));
> > + int sign = on ? 1 : -1;
> > + long acct = sign * (THREAD_SIZE / PAGE_SIZE);
>
> int pages = THREAD_SIZE / PAGE_SIZE;
>
> ?
Will fix. thanks cleaner code advise.
>
> > +
> > + mod_zone_page_state(zone, NR_KERNEL_STACK, acct);
>
> mod_zone_page_state(zone, NR_KERNEL_STACK, on ? pages : -pages);
yes, will fix.
Subject: [PATCH] Show kernel stack usage to /proc/meminfo and OOM log
if the system have a lot of thread, kernel stack consume unignorable large size
memory.
IOW, it make a lot of unaccountable memory.
Tons unaccountable memory bring to harder analyse memory related trouble.
Then, kernel stack account is useful.
Signed-off-by: KOSAKI Motohiro <[email protected]>
---
fs/proc/meminfo.c | 2 ++
include/linux/mmzone.h | 3 ++-
kernel/fork.c | 12 ++++++++++++
mm/page_alloc.c | 6 ++++--
mm/vmstat.c | 1 +
5 files changed, 21 insertions(+), 3 deletions(-)
Index: b/fs/proc/meminfo.c
===================================================================
--- a/fs/proc/meminfo.c
+++ b/fs/proc/meminfo.c
@@ -85,6 +85,7 @@ static int meminfo_proc_show(struct seq_
"SReclaimable: %8lu kB\n"
"SUnreclaim: %8lu kB\n"
"PageTables: %8lu kB\n"
+ "KernelStack %8lu kB\n"
#ifdef CONFIG_QUICKLIST
"Quicklists: %8lu kB\n"
#endif
@@ -129,6 +130,7 @@ static int meminfo_proc_show(struct seq_
K(global_page_state(NR_SLAB_RECLAIMABLE)),
K(global_page_state(NR_SLAB_UNRECLAIMABLE)),
K(global_page_state(NR_PAGETABLE)),
+ K(global_page_state(NR_KERNEL_STACK)),
#ifdef CONFIG_QUICKLIST
K(quicklist_total_size()),
#endif
Index: b/include/linux/mmzone.h
===================================================================
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -94,10 +94,11 @@ enum zone_stat_item {
NR_SLAB_RECLAIMABLE,
NR_SLAB_UNRECLAIMABLE,
NR_PAGETABLE, /* used for pagetables */
+ NR_KERNEL_STACK,
+ /* Second 128 byte cacheline */
NR_UNSTABLE_NFS, /* NFS unstable pages */
NR_BOUNCE,
NR_VMSCAN_WRITE,
- /* Second 128 byte cacheline */
NR_WRITEBACK_TEMP, /* Writeback using temporary buffers */
#ifdef CONFIG_NUMA
NUMA_HIT, /* allocated in intended node */
Index: b/kernel/fork.c
===================================================================
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -137,9 +137,18 @@ struct kmem_cache *vm_area_cachep;
/* SLAB cache for mm_struct structures (tsk->mm) */
static struct kmem_cache *mm_cachep;
+static void account_kernel_stack(struct thread_info *ti, int on)
+{
+ struct zone *zone = page_zone(virt_to_page(ti));
+ int pages = THREAD_SIZE / PAGE_SIZE;
+
+ mod_zone_page_state(zone, NR_KERNEL_STACK, on ? pages : -pages);
+}
+
void free_task(struct task_struct *tsk)
{
prop_local_destroy_single(&tsk->dirties);
+ account_kernel_stack(tsk->stack, 0);
free_thread_info(tsk->stack);
rt_mutex_debug_task_free(tsk);
ftrace_graph_exit_task(tsk);
@@ -255,6 +264,9 @@ static struct task_struct *dup_task_stru
tsk->btrace_seq = 0;
#endif
tsk->splice_pipe = NULL;
+
+ account_kernel_stack(ti, 1);
+
return tsk;
out:
Index: b/mm/page_alloc.c
===================================================================
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2119,7 +2119,8 @@ void show_free_areas(void)
" inactive_file:%lu"
" unevictable:%lu"
" dirty:%lu writeback:%lu unstable:%lu\n"
- " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n",
+ " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n"
+ " kernel_stack:%lu\n",
global_page_state(NR_ACTIVE_ANON),
global_page_state(NR_ACTIVE_FILE),
global_page_state(NR_INACTIVE_ANON),
@@ -2133,7 +2134,8 @@ void show_free_areas(void)
global_page_state(NR_SLAB_UNRECLAIMABLE),
global_page_state(NR_FILE_MAPPED),
global_page_state(NR_PAGETABLE),
- global_page_state(NR_BOUNCE));
+ global_page_state(NR_BOUNCE),
+ global_page_state(NR_KERNEL_STACK));
for_each_populated_zone(zone) {
int i;
Index: b/mm/vmstat.c
===================================================================
--- a/mm/vmstat.c
+++ b/mm/vmstat.c
@@ -639,6 +639,7 @@ static const char * const vmstat_text[]
"nr_slab_reclaimable",
"nr_slab_unreclaimable",
"nr_page_table_pages",
+ "nr_kernel_stack",
"nr_unstable",
"nr_bounce",
"nr_vmscan_write",
On Wed, 1 Jul 2009, KOSAKI Motohiro wrote:
> Subject: [PATCH] Show kernel stack usage to /proc/meminfo and OOM log
>
> if the system have a lot of thread, kernel stack consume unignorable large size
> memory.
> IOW, it make a lot of unaccountable memory.
>
> Tons unaccountable memory bring to harder analyse memory related trouble.
>
> Then, kernel stack account is useful.
>
>
I know this is the second revision of the patch, apologies for not
responding to the first.
> Signed-off-by: KOSAKI Motohiro <[email protected]>
> ---
> fs/proc/meminfo.c | 2 ++
> include/linux/mmzone.h | 3 ++-
> kernel/fork.c | 12 ++++++++++++
> mm/page_alloc.c | 6 ++++--
> mm/vmstat.c | 1 +
> 5 files changed, 21 insertions(+), 3 deletions(-)
>
> Index: b/fs/proc/meminfo.c
> ===================================================================
> --- a/fs/proc/meminfo.c
> +++ b/fs/proc/meminfo.c
> @@ -85,6 +85,7 @@ static int meminfo_proc_show(struct seq_
> "SReclaimable: %8lu kB\n"
> "SUnreclaim: %8lu kB\n"
> "PageTables: %8lu kB\n"
> + "KernelStack %8lu kB\n"
Missing :.
> #ifdef CONFIG_QUICKLIST
> "Quicklists: %8lu kB\n"
> #endif
> @@ -129,6 +130,7 @@ static int meminfo_proc_show(struct seq_
> K(global_page_state(NR_SLAB_RECLAIMABLE)),
> K(global_page_state(NR_SLAB_UNRECLAIMABLE)),
> K(global_page_state(NR_PAGETABLE)),
> + K(global_page_state(NR_KERNEL_STACK)),
> #ifdef CONFIG_QUICKLIST
> K(quicklist_total_size()),
> #endif
> Index: b/include/linux/mmzone.h
> ===================================================================
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -94,10 +94,11 @@ enum zone_stat_item {
> NR_SLAB_RECLAIMABLE,
> NR_SLAB_UNRECLAIMABLE,
> NR_PAGETABLE, /* used for pagetables */
> + NR_KERNEL_STACK,
> + /* Second 128 byte cacheline */
> NR_UNSTABLE_NFS, /* NFS unstable pages */
> NR_BOUNCE,
> NR_VMSCAN_WRITE,
> - /* Second 128 byte cacheline */
> NR_WRITEBACK_TEMP, /* Writeback using temporary buffers */
> #ifdef CONFIG_NUMA
> NUMA_HIT, /* allocated in intended node */
> Index: b/kernel/fork.c
> ===================================================================
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -137,9 +137,18 @@ struct kmem_cache *vm_area_cachep;
> /* SLAB cache for mm_struct structures (tsk->mm) */
> static struct kmem_cache *mm_cachep;
>
> +static void account_kernel_stack(struct thread_info *ti, int on)
> +{
> + struct zone *zone = page_zone(virt_to_page(ti));
> + int pages = THREAD_SIZE / PAGE_SIZE;
> +
> + mod_zone_page_state(zone, NR_KERNEL_STACK, on ? pages : -pages);
> +}
> +
> void free_task(struct task_struct *tsk)
> {
> prop_local_destroy_single(&tsk->dirties);
> + account_kernel_stack(tsk->stack, 0);
I think it would be better to do
#define THREAD_PAGES (THREAD_SIZE / PAGE_SIZE)
since it's currently unused and then
struct zone *zone = page_zone(virt_to_page(tsk->stack));
mod_zone_page_state(zone, NR_KERNEL_STACK, THREAD_PAGES);
in free_task() and
struct zone *zone = page_zone(virt_to_page(ti));
mod_zone_page_state(zone, NR_KERNEL_STACK, -THREAD_PAGES);
in dup_task_struct().
> free_thread_info(tsk->stack);
> rt_mutex_debug_task_free(tsk);
> ftrace_graph_exit_task(tsk);
> @@ -255,6 +264,9 @@ static struct task_struct *dup_task_stru
> tsk->btrace_seq = 0;
> #endif
> tsk->splice_pipe = NULL;
> +
> + account_kernel_stack(ti, 1);
> +
> return tsk;
>
> out:
> Index: b/mm/page_alloc.c
> ===================================================================
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -2119,7 +2119,8 @@ void show_free_areas(void)
> " inactive_file:%lu"
> " unevictable:%lu"
> " dirty:%lu writeback:%lu unstable:%lu\n"
> - " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n",
> + " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n"
> + " kernel_stack:%lu\n",
Does kernel_stack really need to be printed on its own line?
> On Wed, 1 Jul 2009, KOSAKI Motohiro wrote:
>
> > Subject: [PATCH] Show kernel stack usage to /proc/meminfo and OOM log
> >
> > if the system have a lot of thread, kernel stack consume unignorable large size
> > memory.
> > IOW, it make a lot of unaccountable memory.
> >
> > Tons unaccountable memory bring to harder analyse memory related trouble.
> >
> > Then, kernel stack account is useful.
> >
> >
>
> I know this is the second revision of the patch, apologies for not
> responding to the first.
Thanks, good review.
>
> > Signed-off-by: KOSAKI Motohiro <[email protected]>
> > ---
> > fs/proc/meminfo.c | 2 ++
> > include/linux/mmzone.h | 3 ++-
> > kernel/fork.c | 12 ++++++++++++
> > mm/page_alloc.c | 6 ++++--
> > mm/vmstat.c | 1 +
> > 5 files changed, 21 insertions(+), 3 deletions(-)
> >
> > Index: b/fs/proc/meminfo.c
> > ===================================================================
> > --- a/fs/proc/meminfo.c
> > +++ b/fs/proc/meminfo.c
> > @@ -85,6 +85,7 @@ static int meminfo_proc_show(struct seq_
> > "SReclaimable: %8lu kB\n"
> > "SUnreclaim: %8lu kB\n"
> > "PageTables: %8lu kB\n"
> > + "KernelStack %8lu kB\n"
>
> Missing :.
Grr, thanks. Will fix.
>
> > #ifdef CONFIG_QUICKLIST
> > "Quicklists: %8lu kB\n"
> > #endif
> > @@ -129,6 +130,7 @@ static int meminfo_proc_show(struct seq_
> > K(global_page_state(NR_SLAB_RECLAIMABLE)),
> > K(global_page_state(NR_SLAB_UNRECLAIMABLE)),
> > K(global_page_state(NR_PAGETABLE)),
> > + K(global_page_state(NR_KERNEL_STACK)),
> > #ifdef CONFIG_QUICKLIST
> > K(quicklist_total_size()),
> > #endif
> > Index: b/include/linux/mmzone.h
> > ===================================================================
> > --- a/include/linux/mmzone.h
> > +++ b/include/linux/mmzone.h
> > @@ -94,10 +94,11 @@ enum zone_stat_item {
> > NR_SLAB_RECLAIMABLE,
> > NR_SLAB_UNRECLAIMABLE,
> > NR_PAGETABLE, /* used for pagetables */
> > + NR_KERNEL_STACK,
> > + /* Second 128 byte cacheline */
> > NR_UNSTABLE_NFS, /* NFS unstable pages */
> > NR_BOUNCE,
> > NR_VMSCAN_WRITE,
> > - /* Second 128 byte cacheline */
> > NR_WRITEBACK_TEMP, /* Writeback using temporary buffers */
> > #ifdef CONFIG_NUMA
> > NUMA_HIT, /* allocated in intended node */
> > Index: b/kernel/fork.c
> > ===================================================================
> > --- a/kernel/fork.c
> > +++ b/kernel/fork.c
> > @@ -137,9 +137,18 @@ struct kmem_cache *vm_area_cachep;
> > /* SLAB cache for mm_struct structures (tsk->mm) */
> > static struct kmem_cache *mm_cachep;
> >
> > +static void account_kernel_stack(struct thread_info *ti, int on)
> > +{
> > + struct zone *zone = page_zone(virt_to_page(ti));
> > + int pages = THREAD_SIZE / PAGE_SIZE;
> > +
> > + mod_zone_page_state(zone, NR_KERNEL_STACK, on ? pages : -pages);
> > +}
> > +
> > void free_task(struct task_struct *tsk)
> > {
> > prop_local_destroy_single(&tsk->dirties);
> > + account_kernel_stack(tsk->stack, 0);
>
> I think it would be better to do
>
> #define THREAD_PAGES (THREAD_SIZE / PAGE_SIZE)
>
> since it's currently unused and then
>
> struct zone *zone = page_zone(virt_to_page(tsk->stack));
> mod_zone_page_state(zone, NR_KERNEL_STACK, THREAD_PAGES);
>
> in free_task() and
>
> struct zone *zone = page_zone(virt_to_page(ti));
> mod_zone_page_state(zone, NR_KERNEL_STACK, -THREAD_PAGES);
>
> in dup_task_struct().
maybe, gcc makes same code. then I keep current code. because
"struct zone *zone = page_zone(virt_to_page(tsk->stack))" line is a bit
complicate statement and I don't hope sprinkle it.
>
> > free_thread_info(tsk->stack);
> > rt_mutex_debug_task_free(tsk);
> > ftrace_graph_exit_task(tsk);
> > @@ -255,6 +264,9 @@ static struct task_struct *dup_task_stru
> > tsk->btrace_seq = 0;
> > #endif
> > tsk->splice_pipe = NULL;
> > +
> > + account_kernel_stack(ti, 1);
> > +
> > return tsk;
> >
> > out:
> > Index: b/mm/page_alloc.c
> > ===================================================================
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -2119,7 +2119,8 @@ void show_free_areas(void)
> > " inactive_file:%lu"
> > " unevictable:%lu"
> > " dirty:%lu writeback:%lu unstable:%lu\n"
> > - " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n",
> > + " free:%lu slab:%lu mapped:%lu pagetables:%lu bounce:%lu\n"
> > + " kernel_stack:%lu\n",
>
> Does kernel_stack really need to be printed on its own line?
Well, my another patch (Makes slab pages field in show_free_areas() separate two field)
already used full space of previous line. new line is really needed.
KOSAKI Motohiro <[email protected]> wrote:
> + int pages = THREAD_SIZE / PAGE_SIZE;
Bad assumption. On FRV, for example, THREAD_SIZE is 8K and PAGE_SIZE is 16K.
David
On Wed, 1 Jul 2009, David Howells wrote:
> KOSAKI Motohiro <[email protected]> wrote:
>
> > + int pages = THREAD_SIZE / PAGE_SIZE;
>
> Bad assumption. On FRV, for example, THREAD_SIZE is 8K and PAGE_SIZE is 16K.
Guess that means we need arch specific accounting for this counter.
Christoph Lameter wrote:
> On Wed, 1 Jul 2009, David Howells wrote:
>
>> KOSAKI Motohiro <[email protected]> wrote:
>>
>>> + int pages = THREAD_SIZE / PAGE_SIZE;
>> Bad assumption. On FRV, for example, THREAD_SIZE is 8K and PAGE_SIZE is 16K.
>
> Guess that means we need arch specific accounting for this counter.
Or we count the number of stacks internally and only
convert to pages whenever we display the value.
--
All rights reversed.
> Christoph Lameter wrote:
> > On Wed, 1 Jul 2009, David Howells wrote:
> >
> >> KOSAKI Motohiro <[email protected]> wrote:
> >>
> >>> + int pages = THREAD_SIZE / PAGE_SIZE;
> >> Bad assumption. On FRV, for example, THREAD_SIZE is 8K and PAGE_SIZE is 16K.
> >
> > Guess that means we need arch specific accounting for this counter.
>
> Or we count the number of stacks internally and only
> convert to pages whenever we display the value.
Thanks good idea. I'll implement this today (or tommorow).