mdbx: add MDBX_opt_spill_max_denominator and MDBX_opt_spill_min_denominator runtime options.

More for https://github.com/erthink/libmdbx/issues/128

Change-Id: I6b1b00e30ac11e9c2e3d3d3a29cc73079b41d539

mdbx.h

@@ -1885,6 +1885,38 @@ enum MDBX_option_t {
   /** \brief Controls the in-process initial allocation size for dirty pages
    * list of a write transaction. Default is 1024. */
   MDBX_opt_txn_dp_initial,
+
+  /** \brief Controls the in-process how maximal part of the dirty pages may be
+   * spilled when necessary.
+   *
+   * \details The `MDBX_opt_spill_max_denominator` defines the denominator for
+   * limiting from the top for part of the current dirty pages may be spilled
+   * when the free room for a new dirty pages (i.e. distance to the
+   * `MDBX_opt_txn_dp_limit` threshold) is not enough to perform requested
+   * operation.
+   * Exactly `max_pages_to_spill = dirty_pages - dirty_pages / N`,
+   * where `N` is the value set by `MDBX_opt_spill_max_denominator`.
+   *
+   * Should be in the range 0..255, where zero means no limit, i.e. all dirty
+   * pages could be spilled. Default is 8, i.e. no more than 7/8 of the current
+   * dirty pages may be spilled when reached the condition described above. */
+  MDBX_opt_spill_max_denominator,
+
+  /** \brief Controls the in-process how minimal part of the dirty pages should
+   * be spilled when necessary.
+   *
+   * \details The `MDBX_opt_spill_min_denominator` defines the denominator for
+   * limiting from the bottom for part of the current dirty pages should be
+   * spilled when the free room for a new dirty pages (i.e. distance to the
+   * `MDBX_opt_txn_dp_limit` threshold) is not enough to perform requested
+   * operation.
+   * Exactly `min_pages_to_spill = dirty_pages / N`,
+   * where `N` is the value set by `MDBX_opt_spill_min_denominator`.
+   *
+   * Should be in the range 0..255, where zero means no restriction at the
+   * bottom. Default is 8, i.e. at least the 1/8 of the current dirty pages
+   * should be spilled when reached the condition described above. */
+  MDBX_opt_spill_min_denominator,
 };
 #ifndef __cplusplus
 /** \ingroup c_settings */

src/core.c

@@ -4532,21 +4532,33 @@ static void mdbx_txn_xkeep(MDBX_txn *txn, MDBX_cursor *m0,
  * the child hasn't committed yet, and we'd have no way to undo it if
  * the child aborted. */
 static int mdbx_txn_spill(MDBX_txn *txn, MDBX_cursor *m0, unsigned need) {
+#ifndef MDBX_DEBUG_SPILLING
   if (likely(txn->tw.dirtyroom > need))
     return MDBX_SUCCESS;
-
-  const unsigned spill_min =
-      (txn->tw.dirtylist->length / /* TODO: options */ 8);
-  const unsigned spill_max =
-      (txn->tw.dirtylist->length / /* TODO: options */ 1);
   unsigned spill = need - txn->tw.dirtyroom;
-  spill = (spill < spill_max) ? spill : spill_max;
+#else
+  /* spill at least one page if defined MDBX_DEBUG_SPILLING */
+  unsigned spill = (need > txn->tw.dirtyroom) ? need - txn->tw.dirtyroom : 1;
+#endif /* MDBX_DEBUG_SPILLING */
+
+  const unsigned dirty = txn->tw.dirtylist->length;
+  const unsigned spill_min =
+      txn->mt_env->me_options.spill_min_denominator
+          ? dirty / txn->mt_env->me_options.spill_min_denominator
+          : 0;
+  const unsigned spill_max =
+      dirty - (txn->mt_env->me_options.spill_max_denominator
+                   ? dirty / txn->mt_env->me_options.spill_max_denominator
+                   : 0);
   spill = (spill > spill_min) ? spill : spill_min;
+  spill = (spill < spill_max) ? spill : spill_max;
   if (!spill)
     return MDBX_SUCCESS;
 
   mdbx_notice("spilling %u dirty-entries (have %u dirty-room, need %u)", spill,
               txn->tw.dirtyroom, need);
+  mdbx_tassert(txn, txn->tw.dirtylist->length >= spill);
+
   int rc;
   if (!txn->tw.spill_pages) {
     txn->tw.spill_least_removed = INT_MAX;
@@ -9908,6 +9920,8 @@ __cold int mdbx_env_create(MDBX_env **penv) {
   env->me_options.dp_initial = MDBX_PNL_INITIAL;
   if (env->me_options.dp_initial > env->me_options.dp_limit)
     env->me_options.dp_initial = env->me_options.dp_limit;
+  env->me_options.spill_max_denominator = 8;
+  env->me_options.spill_min_denominator = 8;
 
   int rc;
   const size_t os_psize = mdbx_syspagesize();
@@ -20438,6 +20452,17 @@ __cold int mdbx_env_set_option(MDBX_env *env, const MDBX_option_t option,
     }
     break;
 
+  case MDBX_opt_spill_max_denominator:
+    if (unlikely(value > 255))
+      return MDBX_EINVAL;
+    env->me_options.spill_max_denominator = (uint8_t)value;
+    break;
+  case MDBX_opt_spill_min_denominator:
+    if (unlikely(value > 255))
+      return MDBX_EINVAL;
+    env->me_options.spill_min_denominator = (uint8_t)value;
+    break;
+
   default:
     return MDBX_EINVAL;
   }
@@ -20491,6 +20516,13 @@ __cold int mdbx_env_get_option(const MDBX_env *env, const MDBX_option_t option,
     *value = env->me_options.dp_initial;
     break;
 
+  case MDBX_opt_spill_max_denominator:
+    *value = env->me_options.spill_max_denominator;
+    break;
+  case MDBX_opt_spill_min_denominator:
+    *value = env->me_options.spill_min_denominator;
+    break;
+
   default:
     return MDBX_EINVAL;
   }

src/internals.h

@@ -983,6 +983,8 @@ struct MDBX_env {
     unsigned rp_augment_limit;
     unsigned dp_limit;
     unsigned dp_initial;
+    uint8_t spill_max_denominator;
+    uint8_t spill_min_denominator;
   } me_options;
   struct {
 #if MDBX_LOCKING > 0