@@ -1246,6 +1246,9 @@ void __flush_tlb_all(void)
}
EXPORT_SYMBOL_GPL(__flush_tlb_all);
+extern void migrc_flush_start(void);
+extern void migrc_flush_end(struct arch_tlbflush_unmap_batch *arch);
+
void arch_tlbbatch_flush(struct arch_tlbflush_unmap_batch *batch)
{
struct flush_tlb_info *info;
@@ -1254,6 +1257,9 @@ void arch_tlbbatch_flush(struct arch_tlbflush_unmap_batch *batch)
info = get_flush_tlb_info(NULL, 0, TLB_FLUSH_ALL, 0, false,
TLB_GENERATION_INVALID);
+
+ migrc_flush_start();
+
/*
* flush_tlb_multi() is not optimized for the common case in which only
* a local TLB flush is needed. Optimize this use-case by calling
@@ -1268,6 +1274,7 @@ void arch_tlbbatch_flush(struct arch_tlbflush_unmap_batch *batch)
local_irq_enable();
}
+ migrc_flush_end(batch);
cpumask_clear(&batch->cpumask);
put_flush_tlb_info();
@@ -1002,6 +1002,11 @@ struct zone {
/* Zone statistics */
atomic_long_t vm_stat[NR_VM_ZONE_STAT_ITEMS];
atomic_long_t vm_numa_event[NR_VM_NUMA_EVENT_ITEMS];
+
+ /*
+ * the number of folios pending for TLB flush in the zone
+ */
+ atomic_t migrc_pending_nr;
} ____cacheline_internodealigned_in_smp;
enum pgdat_flags {
@@ -1330,6 +1330,12 @@ struct task_struct {
struct tlbflush_unmap_batch tlb_ubc;
struct tlbflush_unmap_batch tlb_ubc_ro;
+ /*
+ * whether all the mappings of a folio during unmap are read-only
+ * so that migrc can work on the folio
+ */
+ bool can_migrc;
+
/* Cache last used pipe for splice(): */
struct pipe_inode_info *splice_pipe;
@@ -925,6 +925,13 @@ void try_to_unmap_flush(void);
void try_to_unmap_flush_dirty(void);
void flush_tlb_batched_pending(struct mm_struct *mm);
void fold_ubc(struct tlbflush_unmap_batch *dst, struct tlbflush_unmap_batch *src);
+
+static inline void init_tlb_ubc(struct tlbflush_unmap_batch *ubc)
+{
+ arch_tlbbatch_clear(&ubc->arch);
+ ubc->flush_required = false;
+ ubc->writable = false;
+}
#else
static inline void try_to_unmap_flush(void)
{
@@ -938,6 +945,9 @@ static inline void flush_tlb_batched_pending(struct mm_struct *mm)
static inline void fold_ubc(struct tlbflush_unmap_batch *dst, struct tlbflush_unmap_batch *src)
{
}
+static inline void init_tlb_ubc(struct tlbflush_unmap_batch *ubc)
+{
+}
#endif /* CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH */
extern const struct trace_print_flags pageflag_names[];
@@ -1284,4 +1294,53 @@ static inline void shrinker_debugfs_remove(struct dentry *debugfs_entry,
}
#endif /* CONFIG_SHRINKER_DEBUG */
+#ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
+/*
+ * Reset the indicator indicating there are no writable mappings at the
+ * beginning of every rmap traverse for unmap. Migrc can work only when
+ * all the mappings are read-only.
+ */
+static inline void can_migrc_init(void)
+{
+ current->can_migrc = true;
+}
+
+/*
+ * Mark the folio is not applicable to migrc, once it found a writble or
+ * dirty pte during rmap traverse for unmap.
+ */
+static inline void can_migrc_fail(void)
+{
+ current->can_migrc = false;
+}
+
+/*
+ * Check if all the mappings are read-only and read-only mappings even
+ * exist.
+ */
+static inline bool can_migrc_test(void)
+{
+ return current->can_migrc && current->tlb_ubc_ro.flush_required;
+}
+
+/*
+ * Return the number of folios pending TLB flush that have yet to get
+ * freed in the zone.
+ */
+static inline int migrc_pending_nr_in_zone(struct zone *z)
+{
+ return atomic_read(&z->migrc_pending_nr);
+}
+
+/*
+ * Perform TLB flush needed and free the folios under migrc's control.
+ */
+bool migrc_flush_free_folios(void);
+#else /* CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH */
+static inline void can_migrc_init(void) {}
+static inline void can_migrc_fail(void) {}
+static inline bool can_migrc_test(void) { return false; }
+static inline int migrc_pending_nr_in_zone(struct zone *z) { return 0; }
+static inline bool migrc_flush_free_folios(void) { return false; }
+#endif
#endif /* __MM_INTERNAL_H */
@@ -3462,6 +3462,14 @@ static vm_fault_t do_wp_page(struct vm_fault *vmf)
if (vmf->page)
folio = page_folio(vmf->page);
+ /*
+ * The folio may or may not be one that is under migrc's control
+ * and about to change its permission from read-only to writable.
+ * Conservatively give up deferring TLB flush just in case.
+ */
+ if (folio)
+ migrc_flush_free_folios();
+
/*
* Shared mapping: we are guaranteed to have VM_WRITE and
* FAULT_FLAG_WRITE set at this point.
@@ -57,6 +57,162 @@
#include "internal.h"
+#ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
+static struct tlbflush_unmap_batch migrc_ubc;
+static LIST_HEAD(migrc_folios);
+static DEFINE_SPINLOCK(migrc_lock);
+
+/*
+ * Need to synchronize between TLB flush and managing pending CPUs in
+ * migrc_ubc. Take a look at the following scenario:
+ *
+ * CPU0 CPU1
+ * ---- ----
+ * TLB flush
+ * Unmap folios (needing TLB flush)
+ * Add pending CPUs to migrc_ubc
+ * Clear the CPUs from migrc_ubc
+ *
+ * The pending CPUs added in CPU1 should not be cleared from migrc_ubc
+ * in CPU0 because the TLB flush for migrc_ubc added in CPU1 has not
+ * been performed this turn. To avoid this, using 'migrc_flushing'
+ * variable, prevent adding pending CPUs to migrc_ubc and give up migrc
+ * mechanism if others are in the middle of TLB flush, like:
+ *
+ * CPU0 CPU1
+ * ---- ----
+ * migrc_flushing++
+ * TLB flush
+ * Unmap folios (needing TLB flush)
+ * If migrc_flushing == 0:
+ * Add pending CPUs to migrc_ubc
+ * Else: <--- hit
+ * Give up migrc mechanism
+ * Clear the CPUs from migrc_ubc
+ * migrc_flush--
+ *
+ * Only the following case would be allowed for migrc mechanism to work:
+ *
+ * CPU0 CPU1
+ * ---- ----
+ * Unmap folios (needing TLB flush)
+ * If migrc_flushing == 0: <--- hit
+ * Add pending CPUs to migrc_ubc
+ * Else:
+ * Give up migrc mechanism
+ * migrc_flushing++
+ * TLB flush
+ * Clear the CPUs from migrc_ubc
+ * migrc_flush--
+ */
+static int migrc_flushing;
+
+static bool migrc_add_pending_ubc(struct tlbflush_unmap_batch *ubc)
+{
+ struct tlbflush_unmap_batch *tlb_ubc = ¤t->tlb_ubc;
+ unsigned long flags;
+
+ spin_lock_irqsave(&migrc_lock, flags);
+ if (migrc_flushing) {
+ spin_unlock_irqrestore(&migrc_lock, flags);
+
+ /*
+ * Give up migrc mechanism. Just let TLB flush needed
+ * handled by try_to_unmap_flush() at the caller side.
+ */
+ fold_ubc(tlb_ubc, ubc);
+ return false;
+ }
+ fold_ubc(&migrc_ubc, ubc);
+ spin_unlock_irqrestore(&migrc_lock, flags);
+ return true;
+}
+
+static bool migrc_add_pending_folios(struct list_head *folios)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&migrc_lock, flags);
+ if (migrc_flushing) {
+ spin_unlock_irqrestore(&migrc_lock, flags);
+
+ /*
+ * Give up migrc mechanism. The caller should perform
+ * TLB flush needed using migrc_flush_free_folios() and
+ * undo some on the folios e.g. restore folios'
+ * reference count increased by migrc and more.
+ */
+ return false;
+ }
+ list_splice(folios, &migrc_folios);
+ spin_unlock_irqrestore(&migrc_lock, flags);
+ return true;
+}
+
+void migrc_flush_start(void)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&migrc_lock, flags);
+ migrc_flushing++;
+ spin_unlock_irqrestore(&migrc_lock, flags);
+}
+
+void migrc_flush_end(struct arch_tlbflush_unmap_batch *arch)
+{
+ LIST_HEAD(folios);
+ struct folio *f, *f2;
+ unsigned long flags;
+
+ spin_lock_irqsave(&migrc_lock, flags);
+ if (!arch_tlbbatch_done(&migrc_ubc.arch, arch)) {
+ list_splice_init(&migrc_folios, &folios);
+ migrc_ubc.flush_required = false;
+ migrc_ubc.writable = false;
+ }
+ migrc_flushing--;
+ spin_unlock_irqrestore(&migrc_lock, flags);
+
+ list_for_each_entry_safe(f, f2, &folios, lru) {
+ folio_put_small_nopcp(f);
+ atomic_dec(&folio_zone(f)->migrc_pending_nr);
+ }
+}
+
+bool migrc_flush_free_folios(void)
+{
+ struct tlbflush_unmap_batch *tlb_ubc = ¤t->tlb_ubc;
+ LIST_HEAD(folios);
+ struct folio *f, *f2;
+ unsigned long flags;
+ bool ret = true;
+
+ spin_lock_irqsave(&migrc_lock, flags);
+ list_splice_init(&migrc_folios, &folios);
+ fold_ubc(tlb_ubc, &migrc_ubc);
+ spin_unlock_irqrestore(&migrc_lock, flags);
+
+ if (list_empty(&folios))
+ ret = false;
+
+ try_to_unmap_flush();
+ list_for_each_entry_safe(f, f2, &folios, lru) {
+ folio_put_small_nopcp(f);
+ atomic_dec(&folio_zone(f)->migrc_pending_nr);
+ }
+ return ret;
+}
+#else /* CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH */
+static bool migrc_add_pending_ubc(struct tlbflush_unmap_batch *ubc)
+{
+ return false;
+}
+static bool migrc_add_pending_folios(struct list_head *folios)
+{
+ return false;
+}
+#endif
+
bool isolate_movable_page(struct page *page, isolate_mode_t mode)
{
struct folio *folio = folio_get_nontail_page(page);
@@ -1274,7 +1430,7 @@ static int migrate_folio_unmap(new_folio_t get_new_folio,
static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
struct folio *src, struct folio *dst,
enum migrate_mode mode, enum migrate_reason reason,
- struct list_head *ret)
+ struct list_head *ret, struct list_head *move_succ)
{
int rc;
int old_page_state = 0;
@@ -1321,9 +1477,15 @@ static int migrate_folio_move(free_folio_t put_new_folio, unsigned long private,
/*
* A folio that has been migrated has all references removed
- * and will be freed.
+ * and will be freed, unless it's under migrc's control.
*/
- list_del(&src->lru);
+ if (move_succ) {
+ folio_get(src);
+ atomic_inc(&folio_zone(src)->migrc_pending_nr);
+ list_move_tail(&src->lru, move_succ);
+ } else
+ list_del(&src->lru);
+
/* Drop an anon_vma reference if we took one */
if (anon_vma)
put_anon_vma(anon_vma);
@@ -1618,7 +1780,7 @@ static void migrate_folios_move(struct list_head *src_folios,
struct list_head *ret_folios,
struct migrate_pages_stats *stats,
int *retry, int *thp_retry, int *nr_failed,
- int *nr_retry_pages)
+ int *nr_retry_pages, struct list_head *move_succ)
{
struct folio *folio, *folio2, *dst, *dst2;
bool is_thp;
@@ -1635,7 +1797,7 @@ static void migrate_folios_move(struct list_head *src_folios,
rc = migrate_folio_move(put_new_folio, private,
folio, dst, mode,
- reason, ret_folios);
+ reason, ret_folios, move_succ);
/*
* The rules are:
* Success: folio will be freed
@@ -1712,17 +1874,33 @@ static int migrate_pages_batch(struct list_head *from,
int rc, rc_saved = 0, nr_pages;
LIST_HEAD(unmap_folios);
LIST_HEAD(dst_folios);
+ LIST_HEAD(unmap_folios_migrc);
+ LIST_HEAD(dst_folios_migrc);
+ LIST_HEAD(move_succ);
bool nosplit = (reason == MR_NUMA_MISPLACED);
+ struct tlbflush_unmap_batch pending_ubc;
+ struct tlbflush_unmap_batch *tlb_ubc = ¤t->tlb_ubc;
+ struct tlbflush_unmap_batch *tlb_ubc_ro = ¤t->tlb_ubc_ro;
+ bool do_migrc;
+ bool migrc_ubc_succ;
VM_WARN_ON_ONCE(mode != MIGRATE_ASYNC &&
!list_empty(from) && !list_is_singular(from));
+ /*
+ * Apply migrc only to numa migration for now.
+ */
+ init_tlb_ubc(&pending_ubc);
+ do_migrc = (reason == MR_DEMOTION || reason == MR_NUMA_MISPLACED);
+
for (pass = 0; pass < nr_pass && retry; pass++) {
retry = 0;
thp_retry = 0;
nr_retry_pages = 0;
list_for_each_entry_safe(folio, folio2, from, lru) {
+ bool can_migrc;
+
is_large = folio_test_large(folio);
is_thp = is_large && folio_test_pmd_mappable(folio);
nr_pages = folio_nr_pages(folio);
@@ -1752,9 +1930,12 @@ static int migrate_pages_batch(struct list_head *from,
continue;
}
+ can_migrc_init();
rc = migrate_folio_unmap(get_new_folio, put_new_folio,
private, folio, &dst, mode, reason,
ret_folios);
+ can_migrc = do_migrc && can_migrc_test() && !is_large;
+
/*
* The rules are:
* Success: folio will be freed
@@ -1800,7 +1981,8 @@ static int migrate_pages_batch(struct list_head *from,
/* nr_failed isn't updated for not used */
stats->nr_thp_failed += thp_retry;
rc_saved = rc;
- if (list_empty(&unmap_folios))
+ if (list_empty(&unmap_folios) &&
+ list_empty(&unmap_folios_migrc))
goto out;
else
goto move;
@@ -1814,8 +1996,19 @@ static int migrate_pages_batch(struct list_head *from,
stats->nr_thp_succeeded += is_thp;
break;
case MIGRATEPAGE_UNMAP:
- list_move_tail(&folio->lru, &unmap_folios);
- list_add_tail(&dst->lru, &dst_folios);
+ if (can_migrc) {
+ list_move_tail(&folio->lru, &unmap_folios_migrc);
+ list_add_tail(&dst->lru, &dst_folios_migrc);
+
+ /*
+ * Gather ro batch data to add
+ * to migrc_ubc after unmap.
+ */
+ fold_ubc(&pending_ubc, tlb_ubc_ro);
+ } else {
+ list_move_tail(&folio->lru, &unmap_folios);
+ list_add_tail(&dst->lru, &dst_folios);
+ }
break;
default:
/*
@@ -1829,12 +2022,19 @@ static int migrate_pages_batch(struct list_head *from,
stats->nr_failed_pages += nr_pages;
break;
}
+ /*
+ * Done with the current folio. Fold the ro
+ * batch data gathered, to the normal batch.
+ */
+ fold_ubc(tlb_ubc, tlb_ubc_ro);
}
}
nr_failed += retry;
stats->nr_thp_failed += thp_retry;
stats->nr_failed_pages += nr_retry_pages;
move:
+ /* Should be before try_to_unmap_flush() */
+ migrc_ubc_succ = do_migrc && migrc_add_pending_ubc(&pending_ubc);
/* Flush TLBs for all unmapped folios */
try_to_unmap_flush();
@@ -1848,7 +2048,30 @@ static int migrate_pages_batch(struct list_head *from,
migrate_folios_move(&unmap_folios, &dst_folios,
put_new_folio, private, mode, reason,
ret_folios, stats, &retry, &thp_retry,
- &nr_failed, &nr_retry_pages);
+ &nr_failed, &nr_retry_pages, NULL);
+ migrate_folios_move(&unmap_folios_migrc, &dst_folios_migrc,
+ put_new_folio, private, mode, reason,
+ ret_folios, stats, &retry, &thp_retry,
+ &nr_failed, &nr_retry_pages, migrc_ubc_succ ?
+ &move_succ : NULL);
+ }
+
+ /*
+ * In case that migrc_add_pending_ubc() has been added
+ * successfully but migrc_add_pending_folios() does not.
+ */
+ if (migrc_ubc_succ && !migrc_add_pending_folios(&move_succ)) {
+ migrc_flush_free_folios();
+
+ /*
+ * Undo src folios that have been successfully added to
+ * move_succ.
+ */
+ list_for_each_entry_safe(folio, folio2, &move_succ, lru) {
+ list_del(&folio->lru);
+ folio_put(folio);
+ atomic_dec(&folio_zone(folio)->migrc_pending_nr);
+ }
}
nr_failed += retry;
stats->nr_thp_failed += thp_retry;
@@ -1859,6 +2082,8 @@ static int migrate_pages_batch(struct list_head *from,
/* Cleanup remaining folios */
migrate_folios_undo(&unmap_folios, &dst_folios,
put_new_folio, private, ret_folios);
+ migrate_folios_undo(&unmap_folios_migrc, &dst_folios_migrc,
+ put_new_folio, private, ret_folios);
return rc;
}
@@ -2972,6 +2972,8 @@ bool __zone_watermark_ok(struct zone *z, unsigned int order, unsigned long mark,
long min = mark;
int o;
+ free_pages += migrc_pending_nr_in_zone(z);
+
/* free_pages may go negative - that's OK */
free_pages -= __zone_watermark_unusable_free(z, order, alloc_flags);
@@ -3066,7 +3068,7 @@ static inline bool zone_watermark_fast(struct zone *z, unsigned int order,
long usable_free;
long reserved;
- usable_free = free_pages;
+ usable_free = free_pages + migrc_pending_nr_in_zone(z);
reserved = __zone_watermark_unusable_free(z, 0, alloc_flags);
/* reserved may over estimate high-atomic reserves. */
@@ -3273,6 +3275,13 @@ get_page_from_freelist(gfp_t gfp_mask, unsigned int order, int alloc_flags,
gfp_mask)) {
int ret;
+ if (migrc_pending_nr_in_zone(zone) &&
+ migrc_flush_free_folios() &&
+ zone_watermark_fast(zone, order, mark,
+ ac->highest_zoneidx,
+ alloc_flags, gfp_mask))
+ goto try_this_zone;
+
if (has_unaccepted_memory()) {
if (try_to_accept_memory(zone, order))
goto try_this_zone;
@@ -676,9 +676,15 @@ static void set_tlb_ubc_flush_pending(struct mm_struct *mm, pte_t pteval,
if (!pte_accessible(mm, pteval))
return;
- if (pte_write(pteval) || writable)
+ if (pte_write(pteval) || writable) {
tlb_ubc = ¤t->tlb_ubc;
- else
+
+ /*
+ * Migrc cannot work with the folio, once it found a
+ * writable or dirty mapping on it.
+ */
+ can_migrc_fail();
+ } else
tlb_ubc = ¤t->tlb_ubc_ro;
arch_tlbbatch_add_pending(&tlb_ubc->arch, mm, uaddr);
Implementation of MIGRC mechanism that stands for 'Migration Read Copy'. We always face the migration overhead at either promotion or demotion, while working with tiered memory e.g. CXL memory and found out TLB shootdown is a quite big one that is needed to get rid of if possible. Fortunately, TLB flush can be defered if both source and destination of folios during migration are kept until all TLB flushes required will have been done, of course, only if the target PTE entries have read-only permission, more precisely speaking, don't have write permission. Otherwise, no doubt the folio might get messed up. To achieve that: 1. For the folios that map only to non-writable TLB entries, prevent TLB flush at migration by keeping both source and destination folios, which will be handled later at a better time. 2. When any non-writable TLB entry changes to writable e.g. through fault handler, give up migrc mechanism so as to perform TLB flush required right away. The following estimation using XSBench shows the improvement like: 1. itlb flush was reduced by 93.9%. 2. dtlb thread was reduced by 43.5%. 3. stlb flush was reduced by 24.9%. 4. dtlb store misses was reduced by 34.2%. 5. itlb load misses was reduced by 45.5%. 6. The runtime was reduced by 3.5%. --- The measurement result: Architecture - x86_64 QEMU - kvm enabled, host cpu Numa - 2 nodes (16 CPUs 1GB, no CPUs 8GB) Linux Kernel - v6.7, numa balancing tiering on, demotion enabled Benchmark - XSBench -p 100000000 (-p option makes the runtime longer) run 'perf stat' using events: 1) itlb.itlb_flush 2) tlb_flush.dtlb_thread 3) tlb_flush.stlb_any 4) dTLB-load-misses 5) dTLB-store-misses 6) iTLB-load-misses run 'cat /proc/vmstat' and pick: 1) numa_pages_migrated 2) pgmigrate_success 3) nr_tlb_remote_flush 4) nr_tlb_remote_flush_received 5) nr_tlb_local_flush_all 6) nr_tlb_local_flush_one BEFORE - mainline v6.7 ------------------------------------------ $ perf stat -a \ -e itlb.itlb_flush \ -e tlb_flush.dtlb_thread \ -e tlb_flush.stlb_any \ -e dTLB-load-misses \ -e dTLB-store-misses \ -e iTLB-load-misses \ ./XSBench -p 100000000 Performance counter stats for 'system wide': 85647229 itlb.itlb_flush 480981504 tlb_flush.dtlb_thread 323937200 tlb_flush.stlb_any 238381632579 dTLB-load-misses 601514255 dTLB-store-misses 2974157461 iTLB-load-misses 2252.883892112 seconds time elapsed $ cat /proc/vmstat ... numa_pages_migrated 12790664 pgmigrate_success 26835314 nr_tlb_remote_flush 3031412 nr_tlb_remote_flush_received 45234862 nr_tlb_local_flush_all 216584 nr_tlb_local_flush_one 740940 ... AFTER - mainline v6.7 + migrc ------------------------------------------ $ perf stat -a \ -e itlb.itlb_flush \ -e tlb_flush.dtlb_thread \ -e tlb_flush.stlb_any \ -e dTLB-load-misses \ -e dTLB-store-misses \ -e iTLB-load-misses \ ./XSBench -p 100000000 Performance counter stats for 'system wide': 5240261 itlb.itlb_flush 271581774 tlb_flush.dtlb_thread 243149389 tlb_flush.stlb_any 234502983364 dTLB-load-misses 395673680 dTLB-store-misses 1620215163 iTLB-load-misses 2172.283436287 seconds time elapsed $ cat /proc/vmstat ... numa_pages_migrated 14897064 pgmigrate_success 30825530 nr_tlb_remote_flush 198290 nr_tlb_remote_flush_received 2820156 nr_tlb_local_flush_all 92048 nr_tlb_local_flush_one 741401 ... Signed-off-by: Byungchul Park <byungchul@sk.com> --- arch/x86/mm/tlb.c | 7 ++ include/linux/mmzone.h | 5 + include/linux/sched.h | 6 + mm/internal.h | 59 ++++++++++ mm/memory.c | 8 ++ mm/migrate.c | 243 +++++++++++++++++++++++++++++++++++++++-- mm/page_alloc.c | 11 +- mm/rmap.c | 10 +- 8 files changed, 337 insertions(+), 12 deletions(-)