Merge branch 'sp/mmap'

* sp/mmap: (27 commits)
  Spell default packedgitlimit slightly differently
  Increase packedGit{Limit,WindowSize} on 64 bit systems.
  Update packedGit config option documentation.
  mmap: set FD_CLOEXEC for file descriptors we keep open for mmap()
  pack-objects: fix use of use_pack().
  Fix random segfaults in pack-objects.
  Cleanup read_cache_from error handling.
  Replace mmap with xmmap, better handling MAP_FAILED.
  Release pack windows before reporting out of memory.
  Default core.packdGitWindowSize to 1 MiB if NO_MMAP.
  Test suite for sliding window mmap implementation.
  Create pack_report() as a debugging aid.
  Support unmapping windows on 'temporary' packfiles.
  Improve error message when packfile mmap fails.
  Ensure core.packedGitWindowSize cannot be less than 2 pages.
  Load core configuration in git-verify-pack.
  Fully activate the sliding window pack access.
  Unmap individual windows rather than entire files.
  Document why header parsing won't exceed a window.
  Loop over pack_windows when inflating/accessing data.
  ...

Conflicts:

	cache.h
	pack-check.c

Documentation/config.txt

@@ -118,6 +118,34 @@ core.legacyheaders::
 	database directly (where the "http://" and "rsync://" protocols
 	count as direct access).
 
+core.packedGitWindowSize::
+	Number of bytes of a pack file to map into memory in a
+	single mapping operation.  Larger window sizes may allow
+	your system to process a smaller number of large pack files
+	more quickly.  Smaller window sizes will negatively affect
+	performance due to increased calls to the operating system's
+	memory manager, but may improve performance when accessing
+	a large number of large pack files.
++
+Default is 1 MiB if NO_MMAP was set at compile time, otherwise 32
+MiB on 32 bit platforms and 1 GiB on 64 bit platforms.  This should
+be reasonable for all users/operating systems.  You probably do
+not need to adjust this value.
++
+Common unit suffixes of 'k', 'm', or 'g' are supported.
+
+core.packedGitLimit::
+	Maximum number of bytes to map simultaneously into memory
+	from pack files.  If Git needs to access more than this many
+	bytes at once to complete an operation it will unmap existing
+	regions to reclaim virtual address space within the process.
++
+Default is 256 MiB on 32 bit platforms and 8 GiB on 64 bit platforms.
+This should be reasonable for all users/operating systems, except on
+the largest projects.  You probably do not need to adjust this value.
++
+Common unit suffixes of 'k', 'm', or 'g' are supported.
+
 alias.*::
 	Command aliases for the gitlink:git[1] command wrapper - e.g.
 	after defining "alias.last = cat-file commit HEAD", the invocation

builtin-pack-objects.c

@@ -276,7 +276,52 @@ static int encode_header(enum object_type type, unsigned long size, unsigned cha
  * we are going to reuse the existing object data as is.  make
  * sure it is not corrupt.
  */
-static int check_inflate(unsigned char *data, unsigned long len, unsigned long expect)
+static int check_pack_inflate(struct packed_git *p,
+		struct pack_window **w_curs,
+		unsigned long offset,
+		unsigned long len,
+		unsigned long expect)
+{
+	z_stream stream;
+	unsigned char fakebuf[4096], *in;
+	int st;
+
+	memset(&stream, 0, sizeof(stream));
+	inflateInit(&stream);
+	do {
+		in = use_pack(p, w_curs, offset, &stream.avail_in);
+		stream.next_in = in;
+		stream.next_out = fakebuf;
+		stream.avail_out = sizeof(fakebuf);
+		st = inflate(&stream, Z_FINISH);
+		offset += stream.next_in - in;
+	} while (st == Z_OK || st == Z_BUF_ERROR);
+	inflateEnd(&stream);
+	return (st == Z_STREAM_END &&
+		stream.total_out == expect &&
+		stream.total_in == len) ? 0 : -1;
+}
+
+static void copy_pack_data(struct sha1file *f,
+		struct packed_git *p,
+		struct pack_window **w_curs,
+		unsigned long offset,
+		unsigned long len)
+{
+	unsigned char *in;
+	unsigned int avail;
+
+	while (len) {
+		in = use_pack(p, w_curs, offset, &avail);
+		if (avail > len)
+			avail = len;
+		sha1write(f, in, avail);
+		offset += avail;
+		len -= avail;
+	}
+}
+
+static int check_loose_inflate(unsigned char *data, unsigned long len, unsigned long expect)
 {
 	z_stream stream;
 	unsigned char fakebuf[4096];
@@ -323,7 +368,7 @@ static int revalidate_loose_object(struct object_entry *entry,
 		return -1;
 	map += used;
 	mapsize -= used;
-	return check_inflate(map, mapsize, size);
+	return check_loose_inflate(map, mapsize, size);
 }
 
 static unsigned long write_object(struct sha1file *f,
@@ -416,6 +461,8 @@ static unsigned long write_object(struct sha1file *f,
 	}
 	else {
 		struct packed_git *p = entry->in_pack;
+		struct pack_window *w_curs = NULL;
+		unsigned long offset;
 
 		if (entry->delta) {
 			obj_type = (allow_ofs_delta && entry->delta->offset) ?
@@ -437,16 +484,14 @@ static unsigned long write_object(struct sha1file *f,
 			hdrlen += 20;
 		}
 
-		use_packed_git(p);
-		buf = (char *) p->pack_base
-			+ entry->in_pack_offset
-			+ entry->in_pack_header_size;
+		offset = entry->in_pack_offset + entry->in_pack_header_size;
 		datalen = find_packed_object_size(p, entry->in_pack_offset)
 				- entry->in_pack_header_size;
-		if (!pack_to_stdout && check_inflate(buf, datalen, entry->size))
+		if (!pack_to_stdout && check_pack_inflate(p, &w_curs,
+				offset, datalen, entry->size))
 			die("corrupt delta in pack %s", sha1_to_hex(entry->sha1));
-		sha1write(f, buf, datalen);
-		unuse_packed_git(p);
+		copy_pack_data(f, p, &w_curs, offset, datalen);
+		unuse_pack(&w_curs);
 		reused++;
 	}
 	if (entry->delta)
@@ -937,22 +982,19 @@ static void check_object(struct object_entry *entry)
 
 	if (entry->in_pack && !entry->preferred_base) {
 		struct packed_git *p = entry->in_pack;
+		struct pack_window *w_curs = NULL;
 		unsigned long left = p->pack_size - entry->in_pack_offset;
 		unsigned long size, used;
 		unsigned char *buf;
 		struct object_entry *base_entry = NULL;
 
-		use_packed_git(p);
-		buf = p->pack_base;
-		buf += entry->in_pack_offset;
+		buf = use_pack(p, &w_curs, entry->in_pack_offset, NULL);
 
 		/* We want in_pack_type even if we do not reuse delta.
 		 * There is no point not reusing non-delta representations.
 		 */
 		used = unpack_object_header_gently(buf, left,
 						   &entry->in_pack_type, &size);
-		if (!used || left - used <= 20)
-			die("corrupt pack for %s", sha1_to_hex(entry->sha1));
 
 		/* Check if it is delta, and the base is also an object
 		 * we are going to pack.  If so we will reuse the existing
@@ -961,36 +1003,42 @@ static void check_object(struct object_entry *entry)
 		if (!no_reuse_delta) {
 			unsigned char c, *base_name;
 			unsigned long ofs;
+			unsigned long used_0;
 			/* there is at least 20 bytes left in the pack */
 			switch (entry->in_pack_type) {
 			case OBJ_REF_DELTA:
-				base_name = buf + used;
+				base_name = use_pack(p, &w_curs,
+					entry->in_pack_offset + used, NULL);
 				used += 20;
 				break;
 			case OBJ_OFS_DELTA:
-				c = buf[used++];
+				buf = use_pack(p, &w_curs,
+					entry->in_pack_offset + used, NULL);
+				used_0 = 0;
+				c = buf[used_0++];
 				ofs = c & 127;
 				while (c & 128) {
 					ofs += 1;
 					if (!ofs || ofs & ~(~0UL >> 7))
 						die("delta base offset overflow in pack for %s",
 						    sha1_to_hex(entry->sha1));
-					c = buf[used++];
+					c = buf[used_0++];
 					ofs = (ofs << 7) + (c & 127);
 				}
 				if (ofs >= entry->in_pack_offset)
 					die("delta base offset out of bound for %s",
 					    sha1_to_hex(entry->sha1));
 				ofs = entry->in_pack_offset - ofs;
 				base_name = find_packed_object_name(p, ofs);
+				used += used_0;
 				break;
 			default:
 				base_name = NULL;
 			}
 			if (base_name)
 				base_entry = locate_object_entry(base_name);
 		}
-		unuse_packed_git(p);
+		unuse_pack(&w_curs);
 		entry->in_pack_header_size = used;
 
 		if (base_entry) {

builtin-verify-pack.c

@@ -55,6 +55,7 @@ int cmd_verify_pack(int argc, const char **argv, const char *prefix)
 	int no_more_options = 0;
 	int nothing_done = 1;
 
+	git_config(git_default_config);
 	while (1 < argc) {
 		if (!no_more_options && argv[1][0] == '-') {
 			if (!strcmp("-v", argv[1]))

cache.h

@@ -197,6 +197,8 @@ extern int warn_ambiguous_refs;
 extern int shared_repository;
 extern const char *apply_default_whitespace;
 extern int zlib_compression_level;
+extern size_t packed_git_window_size;
+extern size_t packed_git_limit;
 
 #define GIT_REPO_VERSION 0
 extern int repository_format_version;
@@ -336,14 +338,22 @@ extern struct alternate_object_database {
 } *alt_odb_list;
 extern void prepare_alt_odb(void);
 
+struct pack_window {
+	struct pack_window *next;
+	unsigned char *base;
+	off_t offset;
+	size_t len;
+	unsigned int last_used;
+	unsigned int inuse_cnt;
+};
+
 extern struct packed_git {
 	struct packed_git *next;
-	unsigned long index_size;
-	unsigned long pack_size;
+	struct pack_window *windows;
 	unsigned int *index_base;
-	void *pack_base;
-	unsigned int pack_last_used;
-	unsigned int pack_use_cnt;
+	off_t index_size;
+	off_t pack_size;
+	int pack_fd;
 	int pack_local;
 	unsigned char sha1[20];
 	/* something like ".git/objects/pack/xxxxx.pack" */
@@ -389,13 +399,14 @@ extern void install_packed_git(struct packed_git *pack);
 extern struct packed_git *find_sha1_pack(const unsigned char *sha1, 
 					 struct packed_git *packs);
 
-extern int use_packed_git(struct packed_git *);
-extern void unuse_packed_git(struct packed_git *);
+extern void pack_report();
+extern unsigned char* use_pack(struct packed_git *, struct pack_window **, unsigned long, unsigned int *);
+extern void unuse_pack(struct pack_window **);
 extern struct packed_git *add_packed_git(char *, int, int);
 extern int num_packed_objects(const struct packed_git *p);
 extern int nth_packed_object_sha1(const struct packed_git *, int, unsigned char*);
 extern unsigned long find_pack_entry_one(const unsigned char *, struct packed_git *);
-extern void *unpack_entry_gently(struct packed_git *, unsigned long, char *, unsigned long *);
+extern void *unpack_entry(struct packed_git *, unsigned long, char *, unsigned long *);
 extern unsigned long unpack_object_header_gently(const unsigned char *buf, unsigned long len, enum object_type *type, unsigned long *sizep);
 extern void packed_object_info_detail(struct packed_git *, unsigned long, char *, unsigned long *, unsigned long *, unsigned int *, unsigned char *);
 
@@ -421,6 +432,7 @@ extern char *git_commit_encoding;
 extern char *git_log_output_encoding;
 
 extern int copy_fd(int ifd, int ofd);
+extern void read_or_die(int fd, void *buf, size_t count);
 extern int write_in_full(int fd, const void *buf, size_t count, const char *);
 extern void write_or_die(int fd, const void *buf, size_t count);
 extern int write_or_whine(int fd, const void *buf, size_t count, const char *msg);

config.c

@@ -304,6 +304,21 @@ int git_default_config(const char *var, const char *value)
 		return 0;
 	}
 
+	if (!strcmp(var, "core.packedgitwindowsize")) {
+		int pgsz = getpagesize();
+		packed_git_window_size = git_config_int(var, value);
+		packed_git_window_size /= pgsz;
+		if (packed_git_window_size < 2)
+			packed_git_window_size = 2;
+		packed_git_window_size *= pgsz;
+		return 0;
+	}
+
+	if (!strcmp(var, "core.packedgitlimit")) {
+		packed_git_limit = git_config_int(var, value);
+		return 0;
+	}
+
 	if (!strcmp(var, "user.name")) {
 		strlcpy(git_default_name, value, sizeof(git_default_name));
 		return 0;
@@ -695,7 +710,7 @@ int git_config_set_multivar(const char* key, const char* value,
 		}
 
 		fstat(in_fd, &st);
-		contents = mmap(NULL, st.st_size, PROT_READ,
+		contents = xmmap(NULL, st.st_size, PROT_READ,
 			MAP_PRIVATE, in_fd, 0);
 		close(in_fd);
 

diff.c

@@ -1341,10 +1341,8 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
 		fd = open(s->path, O_RDONLY);
 		if (fd < 0)
 			goto err_empty;
-		s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
+		s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
 		close(fd);
-		if (s->data == MAP_FAILED)
-			goto err_empty;
 		s->should_munmap = 1;
 	}
 	else {

environment.c

@@ -23,6 +23,8 @@ char *git_log_output_encoding;
 int shared_repository = PERM_UMASK;
 const char *apply_default_whitespace;
 int zlib_compression_level = Z_DEFAULT_COMPRESSION;
+size_t packed_git_window_size = DEFAULT_PACKED_GIT_WINDOW_SIZE;
+size_t packed_git_limit = DEFAULT_PACKED_GIT_LIMIT;
 int pager_in_use;
 int pager_use_color = 1;