mm, swap: VMA based swap readahead

The swap readahead is an important mechanism to reduce the swap in
latency.  Although pure sequential memory access pattern isn't very
popular for anonymous memory, the space locality is still considered
valid.

In the original swap readahead implementation, the consecutive blocks in
swap device are readahead based on the global space locality estimation.
But the consecutive blocks in swap device just reflect the order of page
reclaiming, don't necessarily reflect the access pattern in virtual
memory.  And the different tasks in the system may have different access
patterns, which makes the global space locality estimation incorrect.

In this patch, when page fault occurs, the virtual pages near the fault
address will be readahead instead of the swap slots near the fault swap
slot in swap device.  This avoid to readahead the unrelated swap slots.
At the same time, the swap readahead is changed to work on per-VMA from
globally.  So that the different access patterns of the different VMAs
could be distinguished, and the different readahead policy could be
applied accordingly.  The original core readahead detection and scaling
algorithm is reused, because it is an effect algorithm to detect the
space locality.

The test and result is as follow,

Common test condition
=====================

Test Machine: Xeon E5 v3 (2 sockets, 72 threads, 32G RAM) Swap device:
NVMe disk

Micro-benchmark with combined access pattern
============================================

vm-scalability, sequential swap test case, 4 processes to eat 50G
virtual memory space, repeat the sequential memory writing until 300
seconds.  The first round writing will trigger swap out, the following
rounds will trigger sequential swap in and out.

At the same time, run vm-scalability random swap test case in
background, 8 processes to eat 30G virtual memory space, repeat the
random memory write until 300 seconds.  This will trigger random swap-in
in the background.

This is a combined workload with sequential and random memory accessing
at the same time.  The result (for sequential workload) is as follow,

			Base		Optimized
			----		---------
throughput		345413 KB/s	414029 KB/s (+19.9%)
latency.average		97.14 us	61.06 us (-37.1%)
latency.50th		2 us		1 us
latency.60th		2 us		1 us
latency.70th		98 us		2 us
latency.80th		160 us		2 us
latency.90th		260 us		217 us
latency.95th		346 us		369 us
latency.99th		1.34 ms		1.09 ms
ra_hit%			52.69%		99.98%

The original swap readahead algorithm is confused by the background
random access workload, so readahead hit rate is lower.  The VMA-base
readahead algorithm works much better.

Linpack
=======

The test memory size is bigger than RAM to trigger swapping.

			Base		Optimized
			----		---------
elapsed_time		393.49 s	329.88 s (-16.2%)
ra_hit%			86.21%		98.82%

The score of base and optimized kernel hasn't visible changes.  But the
elapsed time reduced and readahead hit rate improved, so the optimized
kernel runs better for startup and tear down stages.  And the absolute
value of readahead hit rate is high, shows that the space locality is
still valid in some practical workloads.

Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: "Huang, Ying" <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Minchan Kim <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: Shaohua Li <[email protected]>
Cc: Hugh Dickins <[email protected]>
Cc: Fengguang Wu <[email protected]>
Cc: Tim Chen <[email protected]>
Cc: Dave Hansen <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

include/linux/mm_types.h

@@ -335,6 +335,7 @@ struct vm_area_struct {
 	struct file * vm_file;		/* File we map to (can be NULL). */
 	void * vm_private_data;		/* was vm_pte (shared mem) */
 
+	atomic_long_t swap_readahead_info;
 #ifndef CONFIG_MMU
 	struct vm_region *vm_region;	/* NOMMU mapping region */
 #endif

include/linux/swap.h

@@ -251,6 +251,25 @@ struct swap_info_struct {
 	struct swap_cluster_list discard_clusters; /* discard clusters list */
 };
 
+#ifdef CONFIG_64BIT
+#define SWAP_RA_ORDER_CEILING	5
+#else
+/* Avoid stack overflow, because we need to save part of page table */
+#define SWAP_RA_ORDER_CEILING	3
+#define SWAP_RA_PTE_CACHE_SIZE	(1 << SWAP_RA_ORDER_CEILING)
+#endif
+
+struct vma_swap_readahead {
+	unsigned short win;
+	unsigned short offset;
+	unsigned short nr_pte;
+#ifdef CONFIG_64BIT
+	pte_t *ptes;
+#else
+	pte_t ptes[SWAP_RA_PTE_CACHE_SIZE];
+#endif
+};
+
 /* linux/mm/workingset.c */
 void *workingset_eviction(struct address_space *mapping, struct page *page);
 bool workingset_refault(void *shadow);
@@ -350,6 +369,7 @@ int generic_swapfile_activate(struct swap_info_struct *, struct file *,
 #define SWAP_ADDRESS_SPACE_SHIFT	14
 #define SWAP_ADDRESS_SPACE_PAGES	(1 << SWAP_ADDRESS_SPACE_SHIFT)
 extern struct address_space *swapper_spaces[];
+extern bool swap_vma_readahead;
 #define swap_address_space(entry)			    \
 	(&swapper_spaces[swp_type(entry)][swp_offset(entry) \
 		>> SWAP_ADDRESS_SPACE_SHIFT])
@@ -362,7 +382,9 @@ extern void __delete_from_swap_cache(struct page *);
 extern void delete_from_swap_cache(struct page *);
 extern void free_page_and_swap_cache(struct page *);
 extern void free_pages_and_swap_cache(struct page **, int);
-extern struct page *lookup_swap_cache(swp_entry_t);
+extern struct page *lookup_swap_cache(swp_entry_t entry,
+				      struct vm_area_struct *vma,
+				      unsigned long addr);
 extern struct page *read_swap_cache_async(swp_entry_t, gfp_t,
 			struct vm_area_struct *vma, unsigned long addr,
 			bool do_poll);
@@ -372,6 +394,17 @@ extern struct page *__read_swap_cache_async(swp_entry_t, gfp_t,
 extern struct page *swapin_readahead(swp_entry_t, gfp_t,
 			struct vm_area_struct *vma, unsigned long addr);
 
+extern struct page *swap_readahead_detect(struct vm_fault *vmf,
+					  struct vma_swap_readahead *swap_ra);
+extern struct page *do_swap_page_readahead(swp_entry_t fentry, gfp_t gfp_mask,
+					   struct vm_fault *vmf,
+					   struct vma_swap_readahead *swap_ra);
+
+static inline bool swap_use_vma_readahead(void)
+{
+	return READ_ONCE(swap_vma_readahead);
+}
+
 /* linux/mm/swapfile.c */
 extern atomic_long_t nr_swap_pages;
 extern long total_swap_pages;
@@ -466,12 +499,32 @@ static inline struct page *swapin_readahead(swp_entry_t swp, gfp_t gfp_mask,
 	return NULL;
 }
 
+static inline bool swap_use_vma_readahead(void)
+{
+	return false;
+}
+
+static inline struct page *swap_readahead_detect(
+	struct vm_fault *vmf, struct vma_swap_readahead *swap_ra)
+{
+	return NULL;
+}
+
+static inline struct page *do_swap_page_readahead(
+	swp_entry_t fentry, gfp_t gfp_mask,
+	struct vm_fault *vmf, struct vma_swap_readahead *swap_ra)
+{
+	return NULL;
+}
+
 static inline int swap_writepage(struct page *p, struct writeback_control *wbc)
 {
 	return 0;
 }
 
-static inline struct page *lookup_swap_cache(swp_entry_t swp)
+static inline struct page *lookup_swap_cache(swp_entry_t swp,
+					     struct vm_area_struct *vma,
+					     unsigned long addr)
 {
 	return NULL;
 }

mm/memory.c

@@ -2752,16 +2752,23 @@ EXPORT_SYMBOL(unmap_mapping_range);
 int do_swap_page(struct vm_fault *vmf)
 {
 	struct vm_area_struct *vma = vmf->vma;
-	struct page *page, *swapcache;
+	struct page *page = NULL, *swapcache;
 	struct mem_cgroup *memcg;
+	struct vma_swap_readahead swap_ra;
 	swp_entry_t entry;
 	pte_t pte;
 	int locked;
 	int exclusive = 0;
 	int ret = 0;
+	bool vma_readahead = swap_use_vma_readahead();
 
-	if (!pte_unmap_same(vma->vm_mm, vmf->pmd, vmf->pte, vmf->orig_pte))
+	if (vma_readahead)
+		page = swap_readahead_detect(vmf, &swap_ra);
+	if (!pte_unmap_same(vma->vm_mm, vmf->pmd, vmf->pte, vmf->orig_pte)) {
+		if (page)
+			put_page(page);
 		goto out;
+	}
 
 	entry = pte_to_swp_entry(vmf->orig_pte);
 	if (unlikely(non_swap_entry(entry))) {
@@ -2777,10 +2784,16 @@ int do_swap_page(struct vm_fault *vmf)
 		goto out;
 	}
 	delayacct_set_flag(DELAYACCT_PF_SWAPIN);
-	page = lookup_swap_cache(entry);
+	if (!page)
+		page = lookup_swap_cache(entry, vma_readahead ? vma : NULL,
+					 vmf->address);
 	if (!page) {
-		page = swapin_readahead(entry, GFP_HIGHUSER_MOVABLE, vma,
-					vmf->address);
+		if (vma_readahead)
+			page = do_swap_page_readahead(entry,
+				GFP_HIGHUSER_MOVABLE, vmf, &swap_ra);
+		else
+			page = swapin_readahead(entry,
+				GFP_HIGHUSER_MOVABLE, vma, vmf->address);
 		if (!page) {
 			/*
 			 * Back out if somebody else faulted in this pte

mm/shmem.c

@@ -1650,7 +1650,7 @@ static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
 
 	if (swap.val) {
 		/* Look it up and read it in.. */
-		page = lookup_swap_cache(swap);
+		page = lookup_swap_cache(swap, NULL, 0);
 		if (!page) {
 			/* Or update major stats only when swapin succeeds?? */
 			if (fault_type) {