Merge pull request #5396 from lhchavez/mwindow-file-limit

mwindow: set limit on number of open files

4 files changed, 281 insertions, 54 deletions

diff --git a/include/git2/common.h b/include/git2/common.h
index d6696061d..8dd30d506 100644
--- a/include/git2/common.h
+++ b/include/git2/common.h
@@ -205,7 +205,9 @@ typedef enum {
 	GIT_OPT_GET_PACK_MAX_OBJECTS,
 	GIT_OPT_SET_PACK_MAX_OBJECTS,
 	GIT_OPT_DISABLE_PACK_KEEP_FILE_CHECKS,
-	GIT_OPT_ENABLE_HTTP_EXPECT_CONTINUE
+	GIT_OPT_ENABLE_HTTP_EXPECT_CONTINUE,
+	GIT_OPT_GET_MWINDOW_FILE_LIMIT,
+	GIT_OPT_SET_MWINDOW_FILE_LIMIT
 } git_libgit2_opt_t;
 
 /**
@@ -227,8 +229,18 @@ typedef enum {
  *
  *	* opts(GIT_OPT_SET_MWINDOW_MAPPED_LIMIT, size_t):
  *
- *		>Set the maximum amount of memory that can be mapped at any time
- *		by the library
+ *		> Set the maximum amount of memory that can be mapped at any time
+ *		> by the library
+ *
+ *	* opts(GIT_OPT_GET_MWINDOW_FILE_LIMIT, size_t *):
+ *
+ *		> Get the maximum number of files that will be mapped at any time by the
+ *		> library
+ *
+ *	* opts(GIT_OPT_SET_MWINDOW_FILE_LIMIT, size_t):
+ *
+ *		> Set the maximum number of files that can be mapped at any time
+ *		> by the library. The default (0) is unlimited.
  *
  *	* opts(GIT_OPT_GET_SEARCH_PATH, int level, git_buf *buf)
  *
diff --git a/src/mwindow.c b/src/mwindow.c
index 262786a5f..c257f0c71 100644
--- a/src/mwindow.c
+++ b/src/mwindow.c
@@ -22,11 +22,15 @@
 #define DEFAULT_MAPPED_LIMIT \
 	((1024 * 1024) * (sizeof(void*) >= 8 ? 8192ULL : 256UL))
 
+/* default is unlimited */
+#define DEFAULT_FILE_LIMIT 0
+
 size_t git_mwindow__window_size = DEFAULT_WINDOW_SIZE;
 size_t git_mwindow__mapped_limit = DEFAULT_MAPPED_LIMIT;
+size_t git_mwindow__file_limit = DEFAULT_FILE_LIMIT;
 
 /* Whenever you want to read or modify this, grab git__mwindow_mutex */
-static git_mwindow_ctl mem_ctl;
+git_mwindow_ctl git_mwindow__mem_ctl;
 
 /* Global list of mwindow files, to open packs once across repos */
 git_strmap *git__pack_cache = NULL;
@@ -132,7 +136,7 @@ void git_mwindow_free_all(git_mwindow_file *mwf)
  */
 void git_mwindow_free_all_locked(git_mwindow_file *mwf)
 {
-	git_mwindow_ctl *ctl = &mem_ctl;
+	git_mwindow_ctl *ctl = &git_mwindow__mem_ctl;
 	size_t i;
 
 	/*
@@ -174,82 +178,143 @@ int git_mwindow_contains(git_mwindow *win, off64_t offset)
 		&& offset <= (off64_t)(win_off + win->window_map.len);
 }
 
+#define GIT_MWINDOW__LRU -1
+#define GIT_MWINDOW__MRU 1
+
 /*
- * Find the least-recently-used window in a file
+ * Find the least- or most-recently-used window in a file that is not currently
+ * being used. The 'only_unused' flag controls whether the caller requires the
+ * file to only have unused windows.
+ *
+ * Returns whether such a window was found in the file.
  */
-static void git_mwindow_scan_lru(
-	git_mwindow_file *mwf,
-	git_mwindow **lru_w,
-	git_mwindow **lru_l)
+static bool git_mwindow_scan_recently_used(
+		git_mwindow_file *mwf,
+		git_mwindow **out_window,
+		git_mwindow **out_last,
+		bool only_unused,
+		int comparison_sign)
 {
-	git_mwindow *w, *w_l;
-
-	for (w_l = NULL, w = mwf->windows; w; w = w->next) {
-		if (!w->inuse_cnt) {
-			/*
-			 * If the current one is more recent than the last one,
-			 * store it in the output parameter. If lru_w is NULL,
-			 * it's the first loop, so store it as well.
-			 */
-			if (!*lru_w || w->last_used < (*lru_w)->last_used) {
-				*lru_w = w;
-				*lru_l = w_l;
-			}
+	git_mwindow *w, *w_last;
+	git_mwindow *lru_window = NULL, *lru_last = NULL;
+
+	assert(mwf);
+	assert(out_window);
+
+	lru_window = *out_window;
+	if (out_last)
+		lru_last = *out_last;
+
+	for (w_last = NULL, w = mwf->windows; w; w_last = w, w = w->next) {
+		if (w->inuse_cnt) {
+			if (only_unused)
+				return false;
+			/* This window is currently being used. Skip it. */
+			continue;
+		}
+
+		/*
+		 * If the current one is more (or less) recent than the last one,
+		 * store it in the output parameter. If lru_window is NULL,
+		 * it's the first loop, so store it as well.
+		 */
+		if (!lru_window ||
+				(comparison_sign == GIT_MWINDOW__LRU && lru_window->last_used > w->last_used) ||
+				(comparison_sign == GIT_MWINDOW__MRU && lru_window->last_used < w->last_used)) {
+			lru_window = w;
+			lru_last = w_last;
 		}
-		w_l = w;
 	}
+
+	if (!lru_window && !lru_last)
+		return false;
+
+	*out_window = lru_window;
+	if (out_last)
+		*out_last = lru_last;
+	return true;
 }
 
 /*
- * Close the least recently used window. You should check to see if
- * the file descriptors need closing from time to time. Called under
- * lock from new_window.
+ * Close the least recently used window (that is currently not being used) out
+ * of all the files. Called under lock from new_window.
  */
-static int git_mwindow_close_lru(git_mwindow_file *mwf)
+static int git_mwindow_close_lru_window(void)
 {
-	git_mwindow_ctl *ctl = &mem_ctl;
+	git_mwindow_ctl *ctl = &git_mwindow__mem_ctl;
+	git_mwindow_file *cur;
 	size_t i;
-	git_mwindow *lru_w = NULL, *lru_l = NULL, **list = &mwf->windows;
-
-	/* FIXME: Does this give us any advantage? */
-	if(mwf->windows)
-		git_mwindow_scan_lru(mwf, &lru_w, &lru_l);
+	git_mwindow *lru_window = NULL, *lru_last = NULL, **list = NULL;
 
-	for (i = 0; i < ctl->windowfiles.length; ++i) {
-		git_mwindow *last = lru_w;
-		git_mwindow_file *cur = git_vector_get(&ctl->windowfiles, i);
-		git_mwindow_scan_lru(cur, &lru_w, &lru_l);
-		if (lru_w != last)
+	git_vector_foreach(&ctl->windowfiles, i, cur) {
+		if (git_mwindow_scan_recently_used(
+				cur, &lru_window, &lru_last, false, GIT_MWINDOW__LRU)) {
 			list = &cur->windows;
+		}
 	}
 
-	if (!lru_w) {
+	if (!lru_window) {
 		git_error_set(GIT_ERROR_OS, "failed to close memory window; couldn't find LRU");
 		return -1;
 	}
 
-	ctl->mapped -= lru_w->window_map.len;
-	git_futils_mmap_free(&lru_w->window_map);
+	ctl->mapped -= lru_window->window_map.len;
+	git_futils_mmap_free(&lru_window->window_map);
 
-	if (lru_l)
-		lru_l->next = lru_w->next;
+	if (lru_last)
+		lru_last->next = lru_window->next;
 	else
-		*list = lru_w->next;
+		*list = lru_window->next;
 
-	git__free(lru_w);
+	git__free(lru_window);
 	ctl->open_windows--;
 
 	return 0;
 }
 
+/*
+ * Close the file that does not have any open windows AND whose
+ * most-recently-used window is the least-recently used one across all
+ * currently open files.
+ *
+ * Called under lock from new_window.
+ */
+static int git_mwindow_close_lru_file(void)
+{
+	git_mwindow_ctl *ctl = &git_mwindow__mem_ctl;
+	git_mwindow_file *lru_file = NULL, *current_file = NULL;
+	git_mwindow *lru_window = NULL;
+	size_t i;
+
+	git_vector_foreach(&ctl->windowfiles, i, current_file) {
+		git_mwindow *mru_window = NULL;
+		if (!git_mwindow_scan_recently_used(
+				current_file, &mru_window, NULL, true, GIT_MWINDOW__MRU)) {
+			continue;
+		}
+		if (!lru_window || lru_window->last_used > mru_window->last_used)
+			lru_file = current_file;
+	}
+
+	if (!lru_file) {
+		git_error_set(GIT_ERROR_OS, "failed to close memory window file; couldn't find LRU");
+		return -1;
+	}
+
+	git_mwindow_free_all_locked(lru_file);
+	p_close(lru_file->fd);
+	lru_file->fd = -1;
+
+	return 0;
+}
+
 /* This gets called under lock from git_mwindow_open */
 static git_mwindow *new_window(
-	git_mwindow_file *mwf,
 	git_file fd,
 	off64_t size,
 	off64_t offset)
 {
-	git_mwindow_ctl *ctl = &mem_ctl;
+	git_mwindow_ctl *ctl = &git_mwindow__mem_ctl;
 	size_t walign = git_mwindow__window_size / 2;
 	off64_t len;
 	git_mwindow *w;
@@ -269,7 +334,7 @@ static git_mwindow *new_window(
 	ctl->mapped += (size_t)len;
 
 	while (git_mwindow__mapped_limit < ctl->mapped &&
-			git_mwindow_close_lru(mwf) == 0) /* nop */;
+			git_mwindow_close_lru_window() == 0) /* nop */;
 
 	/*
 	 * We treat `mapped_limit` as a soft limit. If we can't find a
@@ -283,7 +348,7 @@ static git_mwindow *new_window(
 		 * we're below our soft limits, so free up what we can and try again.
 		 */
 
-		while (git_mwindow_close_lru(mwf) == 0)
+		while (git_mwindow_close_lru_window() == 0)
 			/* nop */;
 
 		if (git_futils_mmap_ro(&w->window_map, fd, w->offset, (size_t)len) < 0) {
@@ -315,7 +380,7 @@ unsigned char *git_mwindow_open(
 	size_t extra,
 	unsigned int *left)
 {
-	git_mwindow_ctl *ctl = &mem_ctl;
+	git_mwindow_ctl *ctl = &git_mwindow__mem_ctl;
 	git_mwindow *w = *cursor;
 
 	if (git_mutex_lock(&git__mwindow_mutex)) {
@@ -339,7 +404,7 @@ unsigned char *git_mwindow_open(
 		 * one.
 		 */
 		if (!w) {
-			w = new_window(mwf, mwf->fd, mwf->size, offset);
+			w = new_window(mwf->fd, mwf->size, offset);
 			if (w == NULL) {
 				git_mutex_unlock(&git__mwindow_mutex);
 				return NULL;
@@ -367,7 +432,7 @@ unsigned char *git_mwindow_open(
 
 int git_mwindow_file_register(git_mwindow_file *mwf)
 {
-	git_mwindow_ctl *ctl = &mem_ctl;
+	git_mwindow_ctl *ctl = &git_mwindow__mem_ctl;
 	int ret;
 
 	if (git_mutex_lock(&git__mwindow_mutex)) {
@@ -381,6 +446,11 @@ int git_mwindow_file_register(git_mwindow_file *mwf)
 		return -1;
 	}
 
+	if (git_mwindow__file_limit) {
+		while (git_mwindow__file_limit <= ctl->windowfiles.length &&
+				git_mwindow_close_lru_file() == 0) /* nop */;
+	}
+
 	ret = git_vector_insert(&ctl->windowfiles, mwf);
 	git_mutex_unlock(&git__mwindow_mutex);
 
@@ -389,7 +459,7 @@ int git_mwindow_file_register(git_mwindow_file *mwf)
 
 void git_mwindow_file_deregister(git_mwindow_file *mwf)
 {
-	git_mwindow_ctl *ctl = &mem_ctl;
+	git_mwindow_ctl *ctl = &git_mwindow__mem_ctl;
 	git_mwindow_file *cur;
 	size_t i;
 
diff --git a/src/settings.c b/src/settings.c
index f9f6b8497..69ebcb7ab 100644
--- a/src/settings.c
+++ b/src/settings.c
@@ -59,6 +59,7 @@ int git_libgit2_features(void)
 /* Declarations for tuneable settings */
 extern size_t git_mwindow__window_size;
 extern size_t git_mwindow__mapped_limit;
+extern size_t git_mwindow__file_limit;
 extern size_t git_indexer__max_objects;
 extern bool git_disable_pack_keep_file_checks;
 
@@ -124,6 +125,14 @@ int git_libgit2_opts(int key, ...)
 		*(va_arg(ap, size_t *)) = git_mwindow__mapped_limit;
 		break;
 
+	case GIT_OPT_SET_MWINDOW_FILE_LIMIT:
+		git_mwindow__file_limit = va_arg(ap, size_t);
+		break;
+
+	case GIT_OPT_GET_MWINDOW_FILE_LIMIT:
+		*(va_arg(ap, size_t *)) = git_mwindow__file_limit;
+		break;
+
 	case GIT_OPT_GET_SEARCH_PATH:
 		if ((error = config_level_to_sysdir(va_arg(ap, int))) >= 0) {
 			git_buf *out = va_arg(ap, git_buf *);
diff --git a/tests/pack/filelimit.c b/tests/pack/filelimit.c
new file mode 100644
index 000000000..044679f3b
--- /dev/null
+++ b/tests/pack/filelimit.c
@@ -0,0 +1,136 @@
+#include "clar_libgit2.h"
+#include "mwindow.h"
+#include "global.h"
+
+#include <git2.h>
+#include "git2/sys/commit.h"
+#include "git2/sys/mempack.h"
+
+static size_t expected_open_mwindow_files = 0;
+static size_t original_mwindow_file_limit = 0;
+
+extern git_mwindow_ctl git_mwindow__mem_ctl;
+
+void test_pack_filelimit__initialize_tiny(void)
+{
+	expected_open_mwindow_files = 1;
+	cl_git_pass(git_libgit2_opts(GIT_OPT_GET_MWINDOW_FILE_LIMIT, &original_mwindow_file_limit));
+	cl_git_pass(git_libgit2_opts(GIT_OPT_SET_MWINDOW_FILE_LIMIT, expected_open_mwindow_files));
+}
+
+void test_pack_filelimit__initialize_medium(void)
+{
+	expected_open_mwindow_files = 10;
+	cl_git_pass(git_libgit2_opts(GIT_OPT_GET_MWINDOW_FILE_LIMIT, &original_mwindow_file_limit));
+	cl_git_pass(git_libgit2_opts(GIT_OPT_SET_MWINDOW_FILE_LIMIT, expected_open_mwindow_files));
+}
+
+void test_pack_filelimit__initialize_unlimited(void)
+{
+	expected_open_mwindow_files = 15;
+	cl_git_pass(git_libgit2_opts(GIT_OPT_GET_MWINDOW_FILE_LIMIT, &original_mwindow_file_limit));
+	cl_git_pass(git_libgit2_opts(GIT_OPT_SET_MWINDOW_FILE_LIMIT, 0));
+}
+
+void test_pack_filelimit__cleanup(void)
+{
+	git_buf path = GIT_BUF_INIT;
+	cl_git_pass(git_libgit2_opts(GIT_OPT_SET_MWINDOW_FILE_LIMIT, original_mwindow_file_limit));
+
+	cl_git_pass(git_buf_joinpath(&path, clar_sandbox_path(), "repo.git"));
+	cl_fixture_cleanup(path.ptr);
+	git_buf_dispose(&path);
+}
+
+/*
+ * Create a packfile with one commit, one tree, and two blobs. The first blob
+ * (README.md) has the same content in all commits, but the second one
+ * (file.txt) has a different content in each commit.
+ */
+void create_packfile_commit(
+		git_repository *repo,
+		git_oid *out_commit_id,
+		git_oid *parent_id,
+		size_t commit_index,
+		size_t commit_count)
+{
+	git_buf file_contents = GIT_BUF_INIT;
+	git_treebuilder *treebuilder;
+	git_packbuilder *packbuilder;
+	git_signature *s;
+	git_oid oid, tree_id, commit_id;
+	const git_oid *parents[] = { parent_id };
+	size_t parent_count = parent_id ? 1 : 0;
+
+	cl_git_pass(git_treebuilder_new(&treebuilder, repo, NULL));
+
+	cl_git_pass(git_blob_create_from_buffer(&oid, repo, "", 0));
+	cl_git_pass(git_treebuilder_insert(NULL, treebuilder, "README.md", &oid, 0100644));
+
+	cl_git_pass(git_buf_printf(&file_contents, "Commit %zd/%zd", commit_index, commit_count));
+	cl_git_pass(git_blob_create_from_buffer(&oid, repo, file_contents.ptr, file_contents.size));
+	cl_git_pass(git_treebuilder_insert(NULL, treebuilder, "file.txt", &oid, 0100644));
+
+	cl_git_pass(git_treebuilder_write(&tree_id, treebuilder));
+	cl_git_pass(git_signature_now(&s, "alice", "alice@example.com"));
+	cl_git_pass(git_commit_create_from_ids(&commit_id, repo, "refs/heads/master", s, s,
+			NULL, file_contents.ptr, &tree_id, parent_count, parents));
+
+	cl_git_pass(git_packbuilder_new(&packbuilder, repo));
+	cl_git_pass(git_packbuilder_insert_commit(packbuilder, &commit_id));
+	cl_git_pass(git_packbuilder_write(packbuilder, NULL, 0, NULL, NULL));
+
+	cl_git_pass(git_oid_cpy(out_commit_id, &commit_id));
+
+	git_buf_dispose(&file_contents);
+	git_treebuilder_free(treebuilder);
+	git_packbuilder_free(packbuilder);
+	git_signature_free(s);
+}
+
+void test_pack_filelimit__open_repo_with_multiple_packfiles(void)
+{
+	git_buf path = GIT_BUF_INIT;
+	git_mwindow_ctl *ctl = &git_mwindow__mem_ctl;
+	git_repository *repo;
+	git_revwalk *walk;
+	git_oid id, *parent_id = NULL;
+	size_t i;
+	const size_t commit_count = 16;
+	unsigned int open_windows;
+
+	/*
+	 * Create a repository and populate it with 16 commits, each in its own
+	 * packfile.
+	 */
+	cl_git_pass(git_buf_joinpath(&path, clar_sandbox_path(), "repo.git"));
+	cl_git_pass(git_repository_init(&repo, path.ptr, true));
+	for (i = 0; i < commit_count; ++i) {
+		create_packfile_commit(repo, &id, parent_id, i + 1, commit_count);
+		parent_id = &id;
+	}
+
+	cl_git_pass(git_revwalk_new(&walk, repo));
+	cl_git_pass(git_revwalk_sorting(walk, GIT_SORT_TOPOLOGICAL));
+	cl_git_pass(git_revwalk_push_ref(walk, "refs/heads/master"));
+
+	/* Walking the repository requires eventually opening each of the packfiles. */
+	i = 0;
+	while (git_revwalk_next(&id, walk) == 0)
+		++i;
+	cl_assert_equal_i(commit_count, i);
+
+	cl_git_pass(git_mutex_lock(&git__mwindow_mutex));
+	/*
+	 * Adding an assert while holding a lock will cause the whole process to
+	 * deadlock. Copy the value and do the assert after releasing the lock.
+	 */
+	open_windows = ctl->open_windows;
+	cl_git_pass(git_mutex_unlock(&git__mwindow_mutex));
+
+	cl_assert_equal_i(expected_open_mwindow_files, open_windows);
+
+	git_buf_dispose(&path);
+	git_revwalk_free(walk);
+	git_repository_free(repo);
+}