git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCH 10/26] reftable: a generic binary tree implementation

From
Han-Wen Nienhuys via GitGitGadget <gitgitgadget@gmail.com>
Date
Jul 20, 2021, 17:04 UTC
Message-ID
<92970da9cb9105913753211c2b3bb98c1c577bc6.1626800687.git.gitgitgadget@gmail.com>
In-Reply-To
<pull.1054.git.git.1626800686.gitgitgadget@gmail.com>
From: Han-Wen Nienhuys <hanwen@google.com>

The reftable format includes support for an (OID => ref) map. This map can speed up visibility and reachability checks. In particular, various operations along the fetch/push path within Gerrit have ben sped up by using this structure.

The map is constructed with help of a binary tree. Object IDs are hashes, so they are uniformly distributed. Hence, the tree does not attempt forced rebalancing.

Signed-off-by: Han-Wen Nienhuys <hanwen@google.com>
---
 Makefile                 |  4 ++-
 reftable/tree.c          | 63 ++++++++++++++++++++++++++++++++++++++++
 reftable/tree.h          | 34 ++++++++++++++++++++++
 reftable/tree_test.c     | 61 ++++++++++++++++++++++++++++++++++++++
 t/helper/test-reftable.c |  1 +
 5 files changed, 162 insertions(+), 1 deletion(-)
 create mode 100644 reftable/tree.c
 create mode 100644 reftable/tree.h
 create mode 100644 reftable/tree_test.c
diff --git a/Makefile b/Makefile
index 91be4b9c27a..12bd12328b5 100644
--- a/Makefile
+++ b/Makefile
@@ -2454,11 +2454,13 @@ REFTABLE_OBJS += reftable/block.o
 REFTABLE_OBJS += reftable/blocksource.o
 REFTABLE_OBJS += reftable/publicbasics.o
 REFTABLE_OBJS += reftable/record.o
+REFTABLE_OBJS += reftable/tree.o
 
+REFTABLE_TEST_OBJS += reftable/basics_test.o
 REFTABLE_TEST_OBJS += reftable/block_test.o
 REFTABLE_TEST_OBJS += reftable/record_test.o
 REFTABLE_TEST_OBJS += reftable/test_framework.o
-REFTABLE_TEST_OBJS += reftable/basics_test.o
+REFTABLE_TEST_OBJS += reftable/tree_test.o
 
 TEST_OBJS := $(patsubst %$X,%.o,$(TEST_PROGRAMS)) $(patsubst %,t/helper/%,$(TEST_BUILTINS_OBJS))
 
diff --git a/reftable/tree.c b/reftable/tree.c
new file mode 100644
index 00000000000..82db7995dd6
--- /dev/null
+++ b/reftable/tree.c
@@ -0,0 +1,63 @@
+/*
+Copyright 2020 Google LLC
+
+Use of this source code is governed by a BSD-style
+license that can be found in the LICENSE file or at
+https://developers.google.com/open-source/licenses/bsd
+*/
+
+#include "tree.h"
+
+#include "basics.h"
+#include "system.h"
+
+struct tree_node *tree_search(void *key, struct tree_node **rootp,
+			      int (*compare)(const void *, const void *),
+			      int insert)
+{
+	int res;
+	if (*rootp == NULL) {
+		if (!insert) {
+			return NULL;
+		} else {
+			struct tree_node *n =
+				reftable_calloc(sizeof(struct tree_node));
+			n->key = key;
+			*rootp = n;
+			return *rootp;
+		}
+	}
+
+	res = compare(key, (*rootp)->key);
+	if (res < 0)
+		return tree_search(key, &(*rootp)->left, compare, insert);
+	else if (res > 0)
+		return tree_search(key, &(*rootp)->right, compare, insert);
+	return *rootp;
+}
+
+void infix_walk(struct tree_node *t, void (*action)(void *arg, void *key),
+		void *arg)
+{
+	if (t->left) {
+		infix_walk(t->left, action, arg);
+	}
+	action(arg, t->key);
+	if (t->right) {
+		infix_walk(t->right, action, arg);
+	}
+}
+
+void tree_free(struct tree_node *t)
+{
+	if (t == NULL) {
+		return;
+	}
+	if (t->left) {
+		tree_free(t->left);
+	}
+	if (t->right) {
+		tree_free(t->right);
+	}
+	reftable_free(t);
+}
diff --git a/reftable/tree.h b/reftable/tree.h
new file mode 100644
index 00000000000..fbdd002e23a
--- /dev/null
+++ b/reftable/tree.h
@@ -0,0 +1,34 @@
+/*
+Copyright 2020 Google LLC
+
+Use of this source code is governed by a BSD-style
+license that can be found in the LICENSE file or at
+https://developers.google.com/open-source/licenses/bsd
+*/
+
+#ifndef TREE_H
+#define TREE_H
+
+/* tree_node is a generic binary search tree. */
+struct tree_node {
+	void *key;
+	struct tree_node *left, *right;
+};
+
+/* looks for `key` in `rootp` using `compare` as comparison function. If insert
+ * is set, insert the key if it's not found. Else, return NULL.
+ */
+struct tree_node *tree_search(void *key, struct tree_node **rootp,
+			      int (*compare)(const void *, const void *),
+			      int insert);
+
+/* performs an infix walk of the tree. */
+void infix_walk(struct tree_node *t, void (*action)(void *arg, void *key),
+		void *arg);
+
+/*
+ * deallocates the tree nodes recursively. Keys should be deallocated separately
+ * by walking over the tree. */
+void tree_free(struct tree_node *t);
+
+#endif
diff --git a/reftable/tree_test.c b/reftable/tree_test.c
new file mode 100644
index 00000000000..09a970e17b9
--- /dev/null
+++ b/reftable/tree_test.c
@@ -0,0 +1,61 @@
+/*
+Copyright 2020 Google LLC
+
+Use of this source code is governed by a BSD-style
+license that can be found in the LICENSE file or at
+https://developers.google.com/open-source/licenses/bsd
+*/
+
+#include "tree.h"
+
+#include "basics.h"
+#include "record.h"
+#include "test_framework.h"
+#include "reftable-tests.h"
+
+static int test_compare(const void *a, const void *b)
+{
+	return (char *)a - (char *)b;
+}
+
+struct curry {
+	void *last;
+};
+
+static void check_increasing(void *arg, void *key)
+{
+	struct curry *c = arg;
+	if (c->last) {
+		assert(test_compare(c->last, key) < 0);
+	}
+	c->last = key;
+}
+
+static void test_tree(void)
+{
+	struct tree_node *root = NULL;
+
+	void *values[11] = { NULL };
+	struct tree_node *nodes[11] = { NULL };
+	int i = 1;
+	struct curry c = { NULL };
+	do {
+		nodes[i] = tree_search(values + i, &root, &test_compare, 1);
+		i = (i * 7) % 11;
+	} while (i != 1);
+
+	for (i = 1; i < ARRAY_SIZE(nodes); i++) {
+		assert(values + i == nodes[i]->key);
+		assert(nodes[i] ==
+		       tree_search(values + i, &root, &test_compare, 0));
+	}
+
+	infix_walk(root, check_increasing, &c);
+	tree_free(root);
+}
+
+int tree_test_main(int argc, const char *argv[])
+{
+	RUN_TEST(test_tree);
+	return 0;
+}
diff --git a/t/helper/test-reftable.c b/t/helper/test-reftable.c
index c9deeaf08c7..050551fa698 100644
--- a/t/helper/test-reftable.c
+++ b/t/helper/test-reftable.c
@@ -6,5 +6,6 @@ int cmd__reftable(int argc, const char **argv)
 	basics_test_main(argc, argv);
 	block_test_main(argc, argv);
 	record_test_main(argc, argv);
+	tree_test_main(argc, argv);
 	return 0;
 }
-- 
gitgitgadget
Previous: Han-Wen Nienhuys via GitGitGadgetNext: Han-Wen Nienhuys via GitGitGadget
Message 10 of 132 in “Support reftable ref backend for Git”
  1. 00/26 Support reftable ref backend for GitHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  2. 01/26 hash.h: provide constants for the hash IDsHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  3. 02/26 init-db: set the_repository->hash_algo early onHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  4. 03/26 reftable: RFC: add LICENSEHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  5. 06/26 reftable: add blocksource, an abstraction for random access readsHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  6. 04/26 reftable: add error related functionalityHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  7. 05/26 reftable: utility functionsHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  8. 08/26 Provide zlib's uncompress2 from compat/zlib-compat.cHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  9. 07/26 reftable: (de)serialization for the polymorphic record type.Han-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  10. 10/26 reftable: a generic binary tree implementationHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  11. 13/26 reftable: read reftable filesHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  12. 15/26 reftable: add a heap-based priority queue for reftable recordsHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  13. 09/26 reftable: reading/writing blocksHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  14. 12/26 reftable: generic interface to tablesHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  15. 14/26 reftable: reftable file level testsHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  16. 11/26 reftable: write reftable filesHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  17. 19/26 reftable: add dump utilityHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  18. 17/26 reftable: implement refname validationHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  19. 16/26 reftable: add merged table viewHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  20. 20/26 refs: RFC: Reftable support for git-coreHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  21. 24/26 t1401,t2011: parameterize HEAD.lock for REFFILESHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  22. 23/26 t1301: document what needs to be done for reftableHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  23. 26/26 t7004: avoid direct filesystem accessHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  24. 25/26 t1404: annotate test cases with REFFILESHan-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  25. 21/26 git-prompt: prepare for reftable refs backendSZEDER Gábor via GitGitGadget, Jul 20, 2021
  26. 18/26 reftable: implement stack, a mutable database of reftable files.Han-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  27. 22/26 Add "test-tool dump-reftable" command.Han-Wen Nienhuys via GitGitGadget, Jul 20, 2021
  28. 00/25 Support reftable ref backend for GitHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  29. 01/25 hash.h: provide constants for the hash IDsHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  30. 02/25 init-db: set the_repository->hash_algo early onHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  31. 03/25 reftable: RFC: add LICENSEHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  32. 04/25 reftable: add error related functionalityHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  33. 05/25 reftable: utility functionsHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  34. 06/25 reftable: add blocksource, an abstraction for random access readsHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  35. 08/25 Provide zlib's uncompress2 from compat/zlib-compat.cHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  36. 07/25 reftable: (de)serialization for the polymorphic record type.Han-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  37. Carlo Marcelo Arenas BelónAug 16, 2021
  38. Han-Wen NienhuysAug 17, 2021
  39. 09/25 reftable: reading/writing blocksHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  40. 10/25 reftable: a generic binary tree implementationHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  41. 11/25 reftable: write reftable filesHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  42. 12/25 reftable: generic interface to tablesHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  43. 13/25 reftable: read reftable filesHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  44. 14/25 reftable: reftable file level testsHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  45. 15/25 reftable: add a heap-based priority queue for reftable recordsHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  46. 16/25 reftable: add merged table viewHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  47. 17/25 reftable: implement refname validationHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  48. 18/25 reftable: implement stack, a mutable database of reftable files.Han-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  49. 19/25 reftable: add dump utilityHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  50. 21/25 git-prompt: prepare for reftable refs backendSZEDER Gábor via GitGitGadget, Aug 16, 2021
  51. 20/25 refs: RFC: Reftable support for git-coreHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  52. 22/25 Add "test-tool dump-reftable" command.Han-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  53. 23/25 t1301: document what needs to be done for reftableHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  54. 24/25 t1401,t2011: parameterize HEAD.lock for REFFILESHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  55. 25/25 t1404: annotate test cases with REFFILESHan-Wen Nienhuys via GitGitGadget, Aug 16, 2021
  56. Junio C HamanoAug 16, 2021
  57. Han-Wen NienhuysAug 17, 2021
  58. 00/25 Support reftable ref backend for GitHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  59. 01/25 hash.h: provide constants for the hash IDsHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  60. Ævar Arnfjörð BjarmasonAug 23, 2021
  61. 06/25 reftable: add blocksource, an abstraction for random access readsHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  62. 08/25 Provide zlib's uncompress2 from compat/zlib-compat.cHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  63. Philip OakleyAug 18, 2021
  64. Han-Wen NienhuysAug 18, 2021
  65. Philip OakleyAug 18, 2021
  66. 09/25 reftable: reading/writing blocksHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  67. 05/25 reftable: utility functionsHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  68. 03/25 reftable: RFC: add LICENSEHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  69. 04/25 reftable: add error related functionalityHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  70. 12/25 reftable: generic interface to tablesHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  71. 07/25 reftable: (de)serialization for the polymorphic record type.Han-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  72. 11/25 reftable: write reftable filesHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  73. 10/25 reftable: a generic binary tree implementationHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  74. 21/25 git-prompt: prepare for reftable refs backendSZEDER Gábor via GitGitGadget, Aug 17, 2021
  75. 20/25 refs: RFC: Reftable support for git-coreHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  76. Ævar Arnfjörð BjarmasonAug 23, 2021
  77. Han-Wen NienhuysAug 30, 2021
  78. Ævar Arnfjörð BjarmasonAug 30, 2021
  79. 19/25 reftable: add dump utilityHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  80. 15/25 reftable: add a heap-based priority queue for reftable recordsHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  81. 13/25 reftable: read reftable filesHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  82. 25/25 t1404: annotate test cases with REFFILESHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  83. 16/25 reftable: add merged table viewHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  84. 18/25 reftable: implement stack, a mutable database of reftable files.Han-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  85. 24/25 t1401,t2011: parameterize HEAD.lock for REFFILESHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  86. 22/25 Add "test-tool dump-reftable" command.Han-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  87. 17/25 reftable: implement refname validationHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  88. 23/25 t1301: document what needs to be done for reftableHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  89. 14/25 reftable: reftable file level testsHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  90. 02/25 init-db: set the_repository->hash_algo early onHan-Wen Nienhuys via GitGitGadget, Aug 17, 2021
  91. Ævar Arnfjörð BjarmasonAug 23, 2021
  92. Ævar Arnfjörð BjarmasonAug 26, 2021
  93. 00/28 Support reftable ref backend for GitÆvar Arnfjörð Bjarmason, Aug 23, 2021
  94. 01/28 hash.h: provide constants for the hash IDsÆvar Arnfjörð Bjarmason, Aug 23, 2021
  95. 02/28 init-db: set the_repository->hash_algo early onÆvar Arnfjörð Bjarmason, Aug 23, 2021
  96. 03/28 reftable: RFC: add LICENSEÆvar Arnfjörð Bjarmason, Aug 23, 2021
  97. 04/28 reftable: add error related functionalityÆvar Arnfjörð Bjarmason, Aug 23, 2021
  98. 05/28 reftable: utility functionsÆvar Arnfjörð Bjarmason, Aug 23, 2021
  99. 06/28 reftable: add blocksource, an abstraction for random access readsÆvar Arnfjörð Bjarmason, Aug 23, 2021
  100. 07/28 reftable: (de)serialization for the polymorphic record type.Ævar Arnfjörð Bjarmason, Aug 23, 2021
  101. 08/28 Provide zlib's uncompress2 from compat/zlib-compat.cÆvar Arnfjörð Bjarmason, Aug 23, 2021
  102. 09/28 reftable: reading/writing blocksÆvar Arnfjörð Bjarmason, Aug 23, 2021
  103. 10/28 reftable: a generic binary tree implementationÆvar Arnfjörð Bjarmason, Aug 23, 2021
  104. 12/28 reftable: generic interface to tablesÆvar Arnfjörð Bjarmason, Aug 23, 2021
  105. 13/28 reftable: read reftable filesÆvar Arnfjörð Bjarmason, Aug 23, 2021
  106. 14/28 reftable: reftable file level testsÆvar Arnfjörð Bjarmason, Aug 23, 2021
  107. 15/28 reftable: add a heap-based priority queue for reftable recordsÆvar Arnfjörð Bjarmason, Aug 23, 2021
  108. 16/28 reftable: add merged table viewÆvar Arnfjörð Bjarmason, Aug 23, 2021
  109. 17/28 reftable: implement refname validationÆvar Arnfjörð Bjarmason, Aug 23, 2021
  110. 19/28 reftable: add dump utilityÆvar Arnfjörð Bjarmason, Aug 23, 2021
  111. 18/28 reftable: implement stack, a mutable database of reftable files.Ævar Arnfjörð Bjarmason, Aug 23, 2021
  112. Junio C HamanoAug 27, 2021
  113. reftable: fixup for broken __FUNCTION__ useÆvar Arnfjörð Bjarmason, Aug 27, 2021
  114. Carlo ArenasAug 27, 2021
  115. Han-Wen NienhuysAug 30, 2021
  116. 21/28 git-prompt: prepare for reftable refs backendÆvar Arnfjörð Bjarmason, Aug 23, 2021
  117. 20/28 refs: RFC: Reftable support for git-coreÆvar Arnfjörð Bjarmason, Aug 23, 2021
  118. 22/28 Add "test-tool dump-reftable" command.Ævar Arnfjörð Bjarmason, Aug 23, 2021
  119. 23/28 t1301: document what needs to be done for reftableÆvar Arnfjörð Bjarmason, Aug 23, 2021
  120. 26/28 reftable: fixup for new base topic 1/3Ævar Arnfjörð Bjarmason, Aug 23, 2021
  121. 25/28 t1404: annotate test cases with REFFILESÆvar Arnfjörð Bjarmason, Aug 23, 2021
  122. 24/28 t1401,t2011: parameterize HEAD.lock for REFFILESÆvar Arnfjörð Bjarmason, Aug 23, 2021
  123. 27/28 reftable: fixup for new base topic 2/3Ævar Arnfjörð Bjarmason, Aug 23, 2021
  124. Han-Wen NienhuysAug 30, 2021
  125. Ævar Arnfjörð BjarmasonAug 30, 2021
  126. Han-Wen NienhuysAug 30, 2021
  127. Ævar Arnfjörð BjarmasonAug 30, 2021
  128. 28/28 reftable: fixup for new base topic 3/3Ævar Arnfjörð Bjarmason, Aug 23, 2021
  129. 11/28 reftable: write reftable filesÆvar Arnfjörð Bjarmason, Aug 23, 2021
  130. Junio C HamanoAug 26, 2021
  131. Han-Wen NienhuysAug 26, 2021
  132. Junio C HamanoAug 26, 2021

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.