Historically, we exported shared pages to userspace via sysinfo(2) sharedram
and /proc/meminfo's "MemShared" fields. With the advent of tmpfs, from kernel
v2.4 onward, that old way for accounting shared mem was deemed inaccurate and
we started to export a hard-coded 0 for sysinfo.sharedram. Later on, during
the 2.6 timeframe, "MemShared" got re-introduced to /proc/meminfo re-branded
as "Shmem", but we're still reporting sysinfo.sharedmem as that old hard-coded
zero, which makes the "shared memory" report inconsistent across interfaces.
This patch leverages the addition of explicit accounting for pages used by
shmem/tmpfs -- "4b02108 mm: oom analysis: add shmem vmstat" -- in order to
make the users of sysinfo(2) and si_meminfo*() friends aware of that
vmstat entry and make them report it consistently across the interfaces,
as well to make sysinfo(2) returned data consistent with our current API
documentation states.
Signed-off-by: Rafael Aquini <[email protected]>
---
Changelog from v1:
- updated commit log message to include historical context (kosaki-san)
drivers/base/node.c | 2 +-
fs/proc/meminfo.c | 2 +-
mm/page_alloc.c | 3 ++-
3 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/drivers/base/node.c b/drivers/base/node.c
index 8f7ed99..c6d3ae0 100644
--- a/drivers/base/node.c
+++ b/drivers/base/node.c
@@ -126,7 +126,7 @@ static ssize_t node_read_meminfo(struct device *dev,
nid, K(node_page_state(nid, NR_FILE_PAGES)),
nid, K(node_page_state(nid, NR_FILE_MAPPED)),
nid, K(node_page_state(nid, NR_ANON_PAGES)),
- nid, K(node_page_state(nid, NR_SHMEM)),
+ nid, K(i.sharedram),
nid, node_page_state(nid, NR_KERNEL_STACK) *
THREAD_SIZE / 1024,
nid, K(node_page_state(nid, NR_PAGETABLE)),
diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c
index 7445af0..aa1eee0 100644
--- a/fs/proc/meminfo.c
+++ b/fs/proc/meminfo.c
@@ -168,7 +168,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
K(global_page_state(NR_WRITEBACK)),
K(global_page_state(NR_ANON_PAGES)),
K(global_page_state(NR_FILE_MAPPED)),
- K(global_page_state(NR_SHMEM)),
+ K(i.sharedram),
K(global_page_state(NR_SLAB_RECLAIMABLE) +
global_page_state(NR_SLAB_UNRECLAIMABLE)),
K(global_page_state(NR_SLAB_RECLAIMABLE)),
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 20d17f8..f72ea38 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -3040,7 +3040,7 @@ static inline void show_node(struct zone *zone)
void si_meminfo(struct sysinfo *val)
{
val->totalram = totalram_pages;
- val->sharedram = 0;
+ val->sharedram = global_page_state(NR_SHMEM);
val->freeram = global_page_state(NR_FREE_PAGES);
val->bufferram = nr_blockdev_pages();
val->totalhigh = totalhigh_pages;
@@ -3060,6 +3060,7 @@ void si_meminfo_node(struct sysinfo *val, int nid)
for (zone_type = 0; zone_type < MAX_NR_ZONES; zone_type++)
managed_pages += pgdat->node_zones[zone_type].managed_pages;
val->totalram = managed_pages;
+ val->sharedram = node_page_state(nid, NR_SHMEM);
val->freeram = node_page_state(nid, NR_FREE_PAGES);
#ifdef CONFIG_HIGHMEM
val->totalhigh = pgdat->node_zones[ZONE_HIGHMEM].managed_pages;
--
1.9.3
On 06/26/2014 04:00 PM, Rafael Aquini wrote:
> Historically, we exported shared pages to userspace via sysinfo(2) sharedram
> and /proc/meminfo's "MemShared" fields. With the advent of tmpfs, from kernel
> v2.4 onward, that old way for accounting shared mem was deemed inaccurate and
> we started to export a hard-coded 0 for sysinfo.sharedram. Later on, during
> the 2.6 timeframe, "MemShared" got re-introduced to /proc/meminfo re-branded
> as "Shmem", but we're still reporting sysinfo.sharedmem as that old hard-coded
> zero, which makes the "shared memory" report inconsistent across interfaces.
>
> This patch leverages the addition of explicit accounting for pages used by
> shmem/tmpfs -- "4b02108 mm: oom analysis: add shmem vmstat" -- in order to
> make the users of sysinfo(2) and si_meminfo*() friends aware of that
> vmstat entry and make them report it consistently across the interfaces,
> as well to make sysinfo(2) returned data consistent with our current API
> documentation states.
>
> Signed-off-by: Rafael Aquini <[email protected]>
Acked-by: Rik van Riel <[email protected]>