@@ -243,6 +243,8 @@ enum node_stat_item {
NR_DIRTIED, /* page dirtyings since bootup */
NR_WRITTEN, /* page writings since bootup */
NR_KERNEL_MISC_RECLAIMABLE, /* reclaimable non-slab kernel pages */
+ NR_FOLL_PIN_REQUESTED, /* via: pin_user_page(), gup flag: FOLL_PIN */
+ NR_FOLL_PIN_RETURNED, /* pages returned via unpin_user_page() */
NR_VM_NODE_STAT_ITEMS
};
@@ -29,6 +29,19 @@ struct follow_page_context {
unsigned int page_mask;
};
+#ifdef CONFIG_DEBUG_VM
+static inline void __update_proc_vmstat(struct page *page,
+ enum node_stat_item item, int count)
+{
+ mod_node_page_state(page_pgdat(page), item, count);
+}
+#else
+static inline void __update_proc_vmstat(struct page *page,
+ enum node_stat_item item, int count)
+{
+}
+#endif
+
static void hpage_pincount_add(struct page *page, int refs)
{
VM_BUG_ON_PAGE(!hpage_pincount_available(page), page);
@@ -86,6 +99,8 @@ static __maybe_unused struct page *try_grab_compound_head(struct page *page,
if (flags & FOLL_GET)
return try_get_compound_head(page, refs);
else if (flags & FOLL_PIN) {
+ int orig_refs = refs;
+
/*
* When pinning a compound page of order > 1 (which is what
* hpage_pincount_available() checks for), use an exact count to
@@ -104,6 +119,7 @@ static __maybe_unused struct page *try_grab_compound_head(struct page *page,
if (hpage_pincount_available(page))
hpage_pincount_add(page, refs);
+ __update_proc_vmstat(page, NR_FOLL_PIN_REQUESTED, orig_refs);
return page;
}
@@ -159,6 +175,8 @@ bool __must_check try_grab_page(struct page *page, unsigned int flags)
* once, so that the page really is pinned.
*/
page_ref_add(page, refs);
+
+ __update_proc_vmstat(page, NR_FOLL_PIN_REQUESTED, 1);
}
return true;
@@ -179,6 +197,7 @@ static bool __unpin_devmap_managed_user_page(struct page *page)
count = page_ref_sub_return(page, refs);
+ __update_proc_vmstat(page, NR_FOLL_PIN_RETURNED, 1);
/*
* devmap page refcounts are 1-based, rather than 0-based: if
* refcount is 1, then the page is free and the refcount is
@@ -229,6 +248,8 @@ void unpin_user_page(struct page *page)
if (page_ref_sub_and_test(page, refs))
__put_page(page);
+
+ __update_proc_vmstat(page, NR_FOLL_PIN_RETURNED, 1);
}
EXPORT_SYMBOL(unpin_user_page);
@@ -1168,6 +1168,8 @@ const char * const vmstat_text[] = {
"nr_dirtied",
"nr_written",
"nr_kernel_misc_reclaimable",
+ "nr_foll_pin_requested",
+ "nr_foll_pin_returned",
/* enum writeback_stat_item counters */
"nr_dirty_threshold",
Now that pages are "DMA-pinned" via pin_user_page*(), and unpinned via unpin_user_pages*(), we need some visibility into whether all of this is working correctly. Add two new fields to /proc/vmstat: nr_foll_pin_requested nr_foll_pin_returned These are documented in Documentation/core-api/pin_user_pages.rst. They represent the number of pages (since boot time) that have been pinned ("nr_foll_pin_requested") and unpinned ("nr_foll_pin_returned"), via pin_user_pages*() and unpin_user_pages*(). In the absence of long-running DMA or RDMA operations that hold pages pinned, the above two fields will normally be equal to each other. Signed-off-by: John Hubbard <jhubbard@nvidia.com> --- include/linux/mmzone.h | 2 ++ mm/gup.c | 21 +++++++++++++++++++++ mm/vmstat.c | 2 ++ 3 files changed, 25 insertions(+)