2005-04-07 22:16:10 +00:00
|
|
|
/*
|
|
|
|
* GIT - The information manager from hell
|
|
|
|
*
|
|
|
|
* Copyright (C) Linus Torvalds, 2005
|
|
|
|
*/
|
2023-05-16 06:33:57 +00:00
|
|
|
#include "builtin.h"
|
2023-03-21 06:25:58 +00:00
|
|
|
#include "abspath.h"
|
2017-06-14 18:07:36 +00:00
|
|
|
#include "config.h"
|
2023-03-21 06:26:03 +00:00
|
|
|
#include "environment.h"
|
2023-03-21 06:25:54 +00:00
|
|
|
#include "gettext.h"
|
2023-04-11 07:41:53 +00:00
|
|
|
#include "object-file.h"
|
2009-07-12 10:24:32 +00:00
|
|
|
#include "parse-options.h"
|
2023-04-22 20:17:20 +00:00
|
|
|
#include "path.h"
|
2023-03-21 06:26:05 +00:00
|
|
|
#include "setup.h"
|
2023-05-16 06:34:02 +00:00
|
|
|
#include "strbuf.h"
|
2005-04-07 22:13:13 +00:00
|
|
|
|
2008-04-27 17:39:27 +00:00
|
|
|
static int guess_repository_type(const char *git_dir)
|
2007-08-27 07:58:06 +00:00
|
|
|
{
|
|
|
|
const char *slash;
|
2014-07-28 18:30:39 +00:00
|
|
|
char *cwd;
|
|
|
|
int cwd_is_git_dir;
|
2007-08-27 07:58:06 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* "GIT_DIR=. git init" is always bare.
|
|
|
|
* "GIT_DIR=`pwd` git init" too.
|
|
|
|
*/
|
|
|
|
if (!strcmp(".", git_dir))
|
2008-04-27 17:39:27 +00:00
|
|
|
return 1;
|
2014-07-28 18:30:39 +00:00
|
|
|
cwd = xgetcwd();
|
|
|
|
cwd_is_git_dir = !strcmp(git_dir, cwd);
|
|
|
|
free(cwd);
|
|
|
|
if (cwd_is_git_dir)
|
2008-04-27 17:39:27 +00:00
|
|
|
return 1;
|
2007-08-27 07:58:06 +00:00
|
|
|
/*
|
|
|
|
* "GIT_DIR=.git or GIT_DIR=something/.git is usually not.
|
|
|
|
*/
|
|
|
|
if (!strcmp(git_dir, ".git"))
|
2008-04-27 17:39:27 +00:00
|
|
|
return 0;
|
2007-08-27 07:58:06 +00:00
|
|
|
slash = strrchr(git_dir, '/');
|
|
|
|
if (slash && !strcmp(slash, "/.git"))
|
2008-04-27 17:39:27 +00:00
|
|
|
return 0;
|
2007-08-27 07:58:06 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Otherwise it is often bare. At this point
|
|
|
|
* we are just guessing.
|
|
|
|
*/
|
2008-04-27 17:39:27 +00:00
|
|
|
return 1;
|
2007-08-27 07:58:06 +00:00
|
|
|
}
|
|
|
|
|
2009-07-12 10:24:32 +00:00
|
|
|
static int shared_callback(const struct option *opt, const char *arg, int unset)
|
|
|
|
{
|
assert NOARG/NONEG behavior of parse-options callbacks
When we define a parse-options callback, the flags we put in the option
struct must match what the callback expects. For example, a callback
which does not handle the "unset" parameter should only be used with
PARSE_OPT_NONEG. But since the callback and the option struct are not
defined next to each other, it's easy to get this wrong (as earlier
patches in this series show).
Fortunately, the compiler can help us here: compiling with
-Wunused-parameters can show us which callbacks ignore their "unset"
parameters (and likewise, ones that ignore "arg" expect to be triggered
with PARSE_OPT_NOARG).
But after we've inspected a callback and determined that all of its
callers use the right flags, what do we do next? We'd like to silence
the compiler warning, but do so in a way that will catch any wrong calls
in the future.
We can do that by actually checking those variables and asserting that
they match our expectations. Because this is such a common pattern,
we'll introduce some helper macros. The resulting messages aren't
as descriptive as we could make them, but the file/line information from
BUG() is enough to identify the problem (and anyway, the point is that
these should never be seen).
Each of the annotated callbacks in this patch triggers
-Wunused-parameters, and was manually inspected to make sure all callers
use the correct options (so none of these BUGs should be triggerable).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-11-05 06:45:42 +00:00
|
|
|
BUG_ON_OPT_NEG(unset);
|
2009-07-12 10:24:32 +00:00
|
|
|
*((int *) opt->value) = (arg) ? git_config_perm("arg", arg) : PERM_GROUP;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const char *const init_db_usage[] = {
|
2022-10-13 15:39:02 +00:00
|
|
|
N_("git init [-q | --quiet] [--bare] [--template=<template-directory>]\n"
|
2022-10-13 15:39:18 +00:00
|
|
|
" [--separate-git-dir <git-dir>] [--object-format=<format>]\n"
|
|
|
|
" [-b <branch-name> | --initial-branch=<branch-name>]\n"
|
2022-10-13 15:39:02 +00:00
|
|
|
" [--shared[=<permissions>]] [<directory>]"),
|
2009-07-12 10:24:32 +00:00
|
|
|
NULL
|
|
|
|
};
|
2005-08-06 19:50:14 +00:00
|
|
|
|
2005-04-20 04:48:15 +00:00
|
|
|
/*
|
|
|
|
* If you want to, you can share the DB area with any number of branches.
|
|
|
|
* That has advantages: you can save space by sharing all the SHA1 objects.
|
|
|
|
* On the other hand, it might just make lookup slower and messier. You
|
|
|
|
* be the judge. The default case is to have one DB per managed directory.
|
|
|
|
*/
|
2006-07-29 05:44:25 +00:00
|
|
|
int cmd_init_db(int argc, const char **argv, const char *prefix)
|
2005-04-07 22:13:13 +00:00
|
|
|
{
|
2005-05-30 17:20:44 +00:00
|
|
|
const char *git_dir;
|
2011-03-19 15:16:56 +00:00
|
|
|
const char *real_git_dir = NULL;
|
2010-11-26 15:32:40 +00:00
|
|
|
const char *work_tree;
|
2006-05-19 10:03:57 +00:00
|
|
|
const char *template_dir = NULL;
|
2008-04-27 17:39:27 +00:00
|
|
|
unsigned int flags = 0;
|
2020-02-22 20:17:38 +00:00
|
|
|
const char *object_format = NULL;
|
2020-06-24 14:46:32 +00:00
|
|
|
const char *initial_branch = NULL;
|
2020-02-22 20:17:38 +00:00
|
|
|
int hash_algo = GIT_HASH_UNKNOWN;
|
2023-05-16 06:33:43 +00:00
|
|
|
int init_shared_repository = -1;
|
2009-07-12 10:24:32 +00:00
|
|
|
const struct option init_db_options[] = {
|
2012-08-20 12:32:18 +00:00
|
|
|
OPT_STRING(0, "template", &template_dir, N_("template-directory"),
|
|
|
|
N_("directory from which templates will be used")),
|
2009-07-12 10:24:32 +00:00
|
|
|
OPT_SET_INT(0, "bare", &is_bare_repository_cfg,
|
2012-08-20 12:32:18 +00:00
|
|
|
N_("create a bare repository"), 1),
|
2009-07-12 10:24:32 +00:00
|
|
|
{ OPTION_CALLBACK, 0, "shared", &init_shared_repository,
|
2012-08-20 12:32:18 +00:00
|
|
|
N_("permissions"),
|
|
|
|
N_("specify that the git repository is to be shared amongst several users"),
|
2009-07-12 10:24:32 +00:00
|
|
|
PARSE_OPT_OPTARG | PARSE_OPT_NONEG, shared_callback, 0},
|
2012-08-20 12:32:18 +00:00
|
|
|
OPT_BIT('q', "quiet", &flags, N_("be quiet"), INIT_DB_QUIET),
|
|
|
|
OPT_STRING(0, "separate-git-dir", &real_git_dir, N_("gitdir"),
|
|
|
|
N_("separate git dir from working tree")),
|
2020-06-24 14:46:32 +00:00
|
|
|
OPT_STRING('b', "initial-branch", &initial_branch, N_("name"),
|
|
|
|
N_("override the name of the initial branch")),
|
2020-02-22 20:17:38 +00:00
|
|
|
OPT_STRING(0, "object-format", &object_format, N_("hash"),
|
|
|
|
N_("specify the hash algorithm to use")),
|
2009-07-12 10:24:32 +00:00
|
|
|
OPT_END()
|
|
|
|
};
|
|
|
|
|
2009-08-05 19:39:33 +00:00
|
|
|
argc = parse_options(argc, argv, prefix, init_db_options, init_db_usage, 0);
|
|
|
|
|
init: disallow --separate-git-dir with bare repository
The purpose of "git init --separate-git-dir" is to separate the
repository from the worktree. This is true even when --separate-git-dir
is used on an existing worktree, in which case, it moves the .git/
subdirectory to a new location outside the worktree.
However, an outright bare repository (such as one created by "git init
--bare"), has no worktree, so using --separate-git-dir to separate it
from its non-existent worktree is nonsensical. Therefore, make it an
error to use --separate-git-dir on a bare repository.
Implementation note: "git init" considers a repository bare if told so
explicitly via --bare or if it guesses it to be so based upon
heuristics. In the explicit --bare case, a conflict with
--separate-git-dir is easy to detect early. In the guessed case,
however, the conflict can only be detected once "bareness" is guessed,
which happens after "git init" has begun creating the repository.
Technically, we can get by with a single late check which would cover
both cases, however, erroring out early, when possible, without leaving
detritus provides a better user experience.
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-08-09 22:53:16 +00:00
|
|
|
if (real_git_dir && is_bare_repository_cfg == 1)
|
2022-01-05 20:02:14 +00:00
|
|
|
die(_("options '%s' and '%s' cannot be used together"), "--separate-git-dir", "--bare");
|
init: disallow --separate-git-dir with bare repository
The purpose of "git init --separate-git-dir" is to separate the
repository from the worktree. This is true even when --separate-git-dir
is used on an existing worktree, in which case, it moves the .git/
subdirectory to a new location outside the worktree.
However, an outright bare repository (such as one created by "git init
--bare"), has no worktree, so using --separate-git-dir to separate it
from its non-existent worktree is nonsensical. Therefore, make it an
error to use --separate-git-dir on a bare repository.
Implementation note: "git init" considers a repository bare if told so
explicitly via --bare or if it guesses it to be so based upon
heuristics. In the explicit --bare case, a conflict with
--separate-git-dir is easy to detect early. In the guessed case,
however, the conflict can only be detected once "bareness" is guessed,
which happens after "git init" has begun creating the repository.
Technically, we can get by with a single late check which would cover
both cases, however, erroring out early, when possible, without leaving
detritus provides a better user experience.
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-08-09 22:53:16 +00:00
|
|
|
|
2011-03-19 15:16:56 +00:00
|
|
|
if (real_git_dir && !is_absolute_path(real_git_dir))
|
2017-03-08 15:43:40 +00:00
|
|
|
real_git_dir = real_pathdup(real_git_dir, 1);
|
2011-03-19 15:16:56 +00:00
|
|
|
|
2021-03-14 18:47:39 +00:00
|
|
|
if (template_dir && *template_dir && !is_absolute_path(template_dir)) {
|
2019-05-10 10:46:57 +00:00
|
|
|
template_dir = absolute_pathdup(template_dir);
|
2021-03-14 18:47:39 +00:00
|
|
|
UNLEAK(template_dir);
|
|
|
|
}
|
2019-05-10 10:46:57 +00:00
|
|
|
|
2009-08-05 19:39:33 +00:00
|
|
|
if (argc == 1) {
|
2009-07-24 21:59:28 +00:00
|
|
|
int mkdir_tried = 0;
|
|
|
|
retry:
|
2009-08-05 19:39:33 +00:00
|
|
|
if (chdir(argv[0]) < 0) {
|
2009-07-24 21:59:28 +00:00
|
|
|
if (!mkdir_tried) {
|
|
|
|
int saved;
|
|
|
|
/*
|
|
|
|
* At this point we haven't read any configuration,
|
|
|
|
* and we know shared_repository should always be 0;
|
|
|
|
* but just in case we play safe.
|
|
|
|
*/
|
2016-03-11 22:36:49 +00:00
|
|
|
saved = get_shared_repository();
|
|
|
|
set_shared_repository(0);
|
2009-08-05 19:39:33 +00:00
|
|
|
switch (safe_create_leading_directories_const(argv[0])) {
|
2014-01-06 13:45:26 +00:00
|
|
|
case SCLD_OK:
|
|
|
|
case SCLD_PERMS:
|
|
|
|
break;
|
2014-01-06 13:45:25 +00:00
|
|
|
case SCLD_EXISTS:
|
2009-07-24 21:59:28 +00:00
|
|
|
errno = EEXIST;
|
|
|
|
/* fallthru */
|
|
|
|
default:
|
2014-01-06 13:45:26 +00:00
|
|
|
die_errno(_("cannot mkdir %s"), argv[0]);
|
2009-07-24 21:59:28 +00:00
|
|
|
break;
|
|
|
|
}
|
2016-03-11 22:36:49 +00:00
|
|
|
set_shared_repository(saved);
|
2009-08-05 19:39:33 +00:00
|
|
|
if (mkdir(argv[0], 0777) < 0)
|
2011-02-22 23:41:24 +00:00
|
|
|
die_errno(_("cannot mkdir %s"), argv[0]);
|
2009-07-24 21:59:28 +00:00
|
|
|
mkdir_tried = 1;
|
|
|
|
goto retry;
|
|
|
|
}
|
2011-02-22 23:41:24 +00:00
|
|
|
die_errno(_("cannot chdir to %s"), argv[0]);
|
2009-07-24 21:59:28 +00:00
|
|
|
}
|
2009-08-05 19:39:33 +00:00
|
|
|
} else if (0 < argc) {
|
|
|
|
usage(init_db_usage[0]);
|
2005-08-06 19:50:14 +00:00
|
|
|
}
|
2009-08-05 19:39:33 +00:00
|
|
|
if (is_bare_repository_cfg == 1) {
|
2014-07-28 18:31:57 +00:00
|
|
|
char *cwd = xgetcwd();
|
|
|
|
setenv(GIT_DIR_ENVIRONMENT, cwd, argc > 0);
|
|
|
|
free(cwd);
|
2005-08-06 19:50:14 +00:00
|
|
|
}
|
|
|
|
|
2020-02-22 20:17:38 +00:00
|
|
|
if (object_format) {
|
|
|
|
hash_algo = hash_algo_by_name(object_format);
|
|
|
|
if (hash_algo == GIT_HASH_UNKNOWN)
|
|
|
|
die(_("unknown hash algorithm '%s'"), object_format);
|
|
|
|
}
|
|
|
|
|
2009-03-25 23:19:36 +00:00
|
|
|
if (init_shared_repository != -1)
|
2016-03-11 22:36:49 +00:00
|
|
|
set_shared_repository(init_shared_repository);
|
2009-03-25 23:19:36 +00:00
|
|
|
|
2007-08-27 07:58:06 +00:00
|
|
|
/*
|
|
|
|
* GIT_WORK_TREE makes sense only in conjunction with GIT_DIR
|
|
|
|
* without --bare. Catch the error early.
|
|
|
|
*/
|
2019-01-11 22:16:31 +00:00
|
|
|
git_dir = xstrdup_or_null(getenv(GIT_DIR_ENVIRONMENT));
|
|
|
|
work_tree = xstrdup_or_null(getenv(GIT_WORK_TREE_ENVIRONMENT));
|
2010-11-26 15:32:40 +00:00
|
|
|
if ((!git_dir || is_bare_repository_cfg == 1) && work_tree)
|
2011-02-22 23:41:24 +00:00
|
|
|
die(_("%s (or --work-tree=<directory>) not allowed without "
|
|
|
|
"specifying %s (or --git-dir=<directory>)"),
|
2007-08-27 07:58:06 +00:00
|
|
|
GIT_WORK_TREE_ENVIRONMENT,
|
|
|
|
GIT_DIR_ENVIRONMENT);
|
|
|
|
|
2005-05-30 17:20:44 +00:00
|
|
|
/*
|
|
|
|
* Set up the default .git directory contents
|
|
|
|
*/
|
2006-12-15 05:44:58 +00:00
|
|
|
if (!git_dir)
|
2005-05-30 17:20:44 +00:00
|
|
|
git_dir = DEFAULT_GIT_DIR_ENVIRONMENT;
|
2005-12-22 22:19:37 +00:00
|
|
|
|
2020-08-31 06:58:00 +00:00
|
|
|
/*
|
|
|
|
* When --separate-git-dir is used inside a linked worktree, take
|
|
|
|
* care to ensure that the common .git/ directory is relocated, not
|
|
|
|
* the worktree-specific .git/worktrees/<id>/ directory.
|
|
|
|
*/
|
|
|
|
if (real_git_dir) {
|
|
|
|
int err;
|
|
|
|
const char *p;
|
|
|
|
struct strbuf sb = STRBUF_INIT;
|
|
|
|
|
|
|
|
p = read_gitfile_gently(git_dir, &err);
|
|
|
|
if (p && get_common_dir(&sb, p)) {
|
|
|
|
struct strbuf mainwt = STRBUF_INIT;
|
|
|
|
|
|
|
|
strbuf_addbuf(&mainwt, &sb);
|
|
|
|
strbuf_strip_suffix(&mainwt, "/.git");
|
|
|
|
if (chdir(mainwt.buf) < 0)
|
|
|
|
die_errno(_("cannot chdir to %s"), mainwt.buf);
|
|
|
|
strbuf_release(&mainwt);
|
|
|
|
git_dir = strbuf_detach(&sb, NULL);
|
|
|
|
}
|
|
|
|
strbuf_release(&sb);
|
|
|
|
}
|
|
|
|
|
2008-04-27 17:39:27 +00:00
|
|
|
if (is_bare_repository_cfg < 0)
|
|
|
|
is_bare_repository_cfg = guess_repository_type(git_dir);
|
|
|
|
|
|
|
|
if (!is_bare_repository_cfg) {
|
2011-03-03 12:34:51 +00:00
|
|
|
const char *git_dir_parent = strrchr(git_dir, '/');
|
|
|
|
if (git_dir_parent) {
|
|
|
|
char *rel = xstrndup(git_dir, git_dir_parent - git_dir);
|
2017-03-08 15:43:40 +00:00
|
|
|
git_work_tree_cfg = real_pathdup(rel, 1);
|
2011-03-03 12:34:51 +00:00
|
|
|
free(rel);
|
2008-04-27 17:39:27 +00:00
|
|
|
}
|
2014-07-28 18:30:39 +00:00
|
|
|
if (!git_work_tree_cfg)
|
|
|
|
git_work_tree_cfg = xgetcwd();
|
2010-11-26 15:32:40 +00:00
|
|
|
if (work_tree)
|
2014-07-28 18:42:05 +00:00
|
|
|
set_git_work_tree(work_tree);
|
2010-11-26 15:32:40 +00:00
|
|
|
else
|
|
|
|
set_git_work_tree(git_work_tree_cfg);
|
2008-04-27 17:39:27 +00:00
|
|
|
if (access(get_git_work_tree(), X_OK))
|
2011-02-22 23:41:24 +00:00
|
|
|
die_errno (_("Cannot access work tree '%s'"),
|
2009-06-27 15:58:47 +00:00
|
|
|
get_git_work_tree());
|
2006-06-10 06:09:49 +00:00
|
|
|
}
|
2010-11-26 15:32:40 +00:00
|
|
|
else {
|
init: disallow --separate-git-dir with bare repository
The purpose of "git init --separate-git-dir" is to separate the
repository from the worktree. This is true even when --separate-git-dir
is used on an existing worktree, in which case, it moves the .git/
subdirectory to a new location outside the worktree.
However, an outright bare repository (such as one created by "git init
--bare"), has no worktree, so using --separate-git-dir to separate it
from its non-existent worktree is nonsensical. Therefore, make it an
error to use --separate-git-dir on a bare repository.
Implementation note: "git init" considers a repository bare if told so
explicitly via --bare or if it guesses it to be so based upon
heuristics. In the explicit --bare case, a conflict with
--separate-git-dir is easy to detect early. In the guessed case,
however, the conflict can only be detected once "bareness" is guessed,
which happens after "git init" has begun creating the repository.
Technically, we can get by with a single late check which would cover
both cases, however, erroring out early, when possible, without leaving
detritus provides a better user experience.
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-08-09 22:53:16 +00:00
|
|
|
if (real_git_dir)
|
|
|
|
die(_("--separate-git-dir incompatible with bare repository"));
|
2010-11-26 15:32:40 +00:00
|
|
|
if (work_tree)
|
2014-07-28 18:42:05 +00:00
|
|
|
set_git_work_tree(work_tree);
|
2010-11-26 15:32:40 +00:00
|
|
|
}
|
2005-12-22 22:19:37 +00:00
|
|
|
|
add UNLEAK annotation for reducing leak false positives
It's a common pattern in git commands to allocate some
memory that should last for the lifetime of the program and
then not bother to free it, relying on the OS to throw it
away.
This keeps the code simple, and it's fast (we don't waste
time traversing structures or calling free at the end of the
program). But it also triggers warnings from memory-leak
checkers like valgrind or LSAN. They know that the memory
was still allocated at program exit, but they don't know
_when_ the leaked memory stopped being useful. If it was
early in the program, then it's probably a real and
important leak. But if it was used right up until program
exit, it's not an interesting leak and we'd like to suppress
it so that we can see the real leaks.
This patch introduces an UNLEAK() macro that lets us do so.
To understand its design, let's first look at some of the
alternatives.
Unfortunately the suppression systems offered by
leak-checking tools don't quite do what we want. A
leak-checker basically knows two things:
1. Which blocks were allocated via malloc, and the
callstack during the allocation.
2. Which blocks were left un-freed at the end of the
program (and which are unreachable, but more on that
later).
Their suppressions work by mentioning the function or
callstack of a particular allocation, and marking it as OK
to leak. So imagine you have code like this:
int cmd_foo(...)
{
/* this allocates some memory */
char *p = some_function();
printf("%s", p);
return 0;
}
You can say "ignore allocations from some_function(),
they're not leaks". But that's not right. That function may
be called elsewhere, too, and we would potentially want to
know about those leaks.
So you can say "ignore the callstack when main calls
some_function". That works, but your annotations are
brittle. In this case it's only two functions, but you can
imagine that the actual allocation is much deeper. If any of
the intermediate code changes, you have to update the
suppression.
What we _really_ want to say is that "the value assigned to
p at the end of the function is not a real leak". But
leak-checkers can't understand that; they don't know about
"p" in the first place.
However, we can do something a little bit tricky if we make
some assumptions about how leak-checkers work. They
generally don't just report all un-freed blocks. That would
report even globals which are still accessible when the
leak-check is run. Instead they take some set of memory
(like BSS) as a root and mark it as "reachable". Then they
scan the reachable blocks for anything that looks like a
pointer to a malloc'd block, and consider that block
reachable. And then they scan those blocks, and so on,
transitively marking anything reachable from a global as
"not leaked" (or at least leaked in a different category).
So we can mark the value of "p" as reachable by putting it
into a variable with program lifetime. One way to do that is
to just mark "p" as static. But that actually affects the
run-time behavior if the function is called twice (you
aren't likely to call main() twice, but some of our cmd_*()
functions are called from other commands).
Instead, we can trick the leak-checker by putting the value
into _any_ reachable bytes. This patch keeps a global
linked-list of bytes copied from "unleaked" variables. That
list is reachable even at program exit, which confers
recursive reachability on whatever values we unleak.
In other words, you can do:
int cmd_foo(...)
{
char *p = some_function();
printf("%s", p);
UNLEAK(p);
return 0;
}
to annotate "p" and suppress the leak report.
But wait, couldn't we just say "free(p)"? In this toy
example, yes. But UNLEAK()'s byte-copying strategy has
several advantages over actually freeing the memory:
1. It's recursive across structures. In many cases our "p"
is not just a pointer, but a complex struct whose
fields may have been allocated by a sub-function. And
in some cases (e.g., dir_struct) we don't even have a
function which knows how to free all of the struct
members.
By marking the struct itself as reachable, that confers
reachability on any pointers it contains (including those
found in embedded structs, or reachable by walking
heap blocks recursively.
2. It works on cases where we're not sure if the value is
allocated or not. For example:
char *p = argc > 1 ? argv[1] : some_function();
It's safe to use UNLEAK(p) here, because it's not
freeing any memory. In the case that we're pointing to
argv here, the reachability checker will just ignore
our bytes.
3. Likewise, it works even if the variable has _already_
been freed. We're just copying the pointer bytes. If
the block has been freed, the leak-checker will skip
over those bytes as uninteresting.
4. Because it's not actually freeing memory, you can
UNLEAK() before we are finished accessing the variable.
This is helpful in cases like this:
char *p = some_function();
return another_function(p);
Writing this with free() requires:
int ret;
char *p = some_function();
ret = another_function(p);
free(p);
return ret;
But with unleak we can just write:
char *p = some_function();
UNLEAK(p);
return another_function(p);
This patch adds the UNLEAK() macro and enables it
automatically when Git is compiled with SANITIZE=leak. In
normal builds it's a noop, so we pay no runtime cost.
It also adds some UNLEAK() annotations to show off how the
feature works. On top of other recent leak fixes, these are
enough to get t0000 and t0001 to pass when compiled with
LSAN.
Note the case in commit.c which actually converts a
strbuf_release() into an UNLEAK. This code was already
non-leaky, but the free didn't do anything useful, since
we're exiting. Converting it to an annotation means that
non-leak-checking builds pay no runtime cost. The cost is
minimal enough that it's probably not worth going on a
crusade to convert these kinds of frees to UNLEAKS. I did it
here for consistency with the "sb" leak (though it would
have been equally correct to go the other way, and turn them
both into strbuf_release() calls).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-09-08 06:38:41 +00:00
|
|
|
UNLEAK(real_git_dir);
|
2019-01-11 22:16:31 +00:00
|
|
|
UNLEAK(git_dir);
|
|
|
|
UNLEAK(work_tree);
|
add UNLEAK annotation for reducing leak false positives
It's a common pattern in git commands to allocate some
memory that should last for the lifetime of the program and
then not bother to free it, relying on the OS to throw it
away.
This keeps the code simple, and it's fast (we don't waste
time traversing structures or calling free at the end of the
program). But it also triggers warnings from memory-leak
checkers like valgrind or LSAN. They know that the memory
was still allocated at program exit, but they don't know
_when_ the leaked memory stopped being useful. If it was
early in the program, then it's probably a real and
important leak. But if it was used right up until program
exit, it's not an interesting leak and we'd like to suppress
it so that we can see the real leaks.
This patch introduces an UNLEAK() macro that lets us do so.
To understand its design, let's first look at some of the
alternatives.
Unfortunately the suppression systems offered by
leak-checking tools don't quite do what we want. A
leak-checker basically knows two things:
1. Which blocks were allocated via malloc, and the
callstack during the allocation.
2. Which blocks were left un-freed at the end of the
program (and which are unreachable, but more on that
later).
Their suppressions work by mentioning the function or
callstack of a particular allocation, and marking it as OK
to leak. So imagine you have code like this:
int cmd_foo(...)
{
/* this allocates some memory */
char *p = some_function();
printf("%s", p);
return 0;
}
You can say "ignore allocations from some_function(),
they're not leaks". But that's not right. That function may
be called elsewhere, too, and we would potentially want to
know about those leaks.
So you can say "ignore the callstack when main calls
some_function". That works, but your annotations are
brittle. In this case it's only two functions, but you can
imagine that the actual allocation is much deeper. If any of
the intermediate code changes, you have to update the
suppression.
What we _really_ want to say is that "the value assigned to
p at the end of the function is not a real leak". But
leak-checkers can't understand that; they don't know about
"p" in the first place.
However, we can do something a little bit tricky if we make
some assumptions about how leak-checkers work. They
generally don't just report all un-freed blocks. That would
report even globals which are still accessible when the
leak-check is run. Instead they take some set of memory
(like BSS) as a root and mark it as "reachable". Then they
scan the reachable blocks for anything that looks like a
pointer to a malloc'd block, and consider that block
reachable. And then they scan those blocks, and so on,
transitively marking anything reachable from a global as
"not leaked" (or at least leaked in a different category).
So we can mark the value of "p" as reachable by putting it
into a variable with program lifetime. One way to do that is
to just mark "p" as static. But that actually affects the
run-time behavior if the function is called twice (you
aren't likely to call main() twice, but some of our cmd_*()
functions are called from other commands).
Instead, we can trick the leak-checker by putting the value
into _any_ reachable bytes. This patch keeps a global
linked-list of bytes copied from "unleaked" variables. That
list is reachable even at program exit, which confers
recursive reachability on whatever values we unleak.
In other words, you can do:
int cmd_foo(...)
{
char *p = some_function();
printf("%s", p);
UNLEAK(p);
return 0;
}
to annotate "p" and suppress the leak report.
But wait, couldn't we just say "free(p)"? In this toy
example, yes. But UNLEAK()'s byte-copying strategy has
several advantages over actually freeing the memory:
1. It's recursive across structures. In many cases our "p"
is not just a pointer, but a complex struct whose
fields may have been allocated by a sub-function. And
in some cases (e.g., dir_struct) we don't even have a
function which knows how to free all of the struct
members.
By marking the struct itself as reachable, that confers
reachability on any pointers it contains (including those
found in embedded structs, or reachable by walking
heap blocks recursively.
2. It works on cases where we're not sure if the value is
allocated or not. For example:
char *p = argc > 1 ? argv[1] : some_function();
It's safe to use UNLEAK(p) here, because it's not
freeing any memory. In the case that we're pointing to
argv here, the reachability checker will just ignore
our bytes.
3. Likewise, it works even if the variable has _already_
been freed. We're just copying the pointer bytes. If
the block has been freed, the leak-checker will skip
over those bytes as uninteresting.
4. Because it's not actually freeing memory, you can
UNLEAK() before we are finished accessing the variable.
This is helpful in cases like this:
char *p = some_function();
return another_function(p);
Writing this with free() requires:
int ret;
char *p = some_function();
ret = another_function(p);
free(p);
return ret;
But with unleak we can just write:
char *p = some_function();
UNLEAK(p);
return another_function(p);
This patch adds the UNLEAK() macro and enables it
automatically when Git is compiled with SANITIZE=leak. In
normal builds it's a noop, so we pay no runtime cost.
It also adds some UNLEAK() annotations to show off how the
feature works. On top of other recent leak fixes, these are
enough to get t0000 and t0001 to pass when compiled with
LSAN.
Note the case in commit.c which actually converts a
strbuf_release() into an UNLEAK. This code was already
non-leaky, but the free didn't do anything useful, since
we're exiting. Converting it to an annotation means that
non-leak-checking builds pay no runtime cost. The cost is
minimal enough that it's probably not worth going on a
crusade to convert these kinds of frees to UNLEAKS. I did it
here for consistency with the "sb" leak (though it would
have been equally correct to go the other way, and turn them
both into strbuf_release() calls).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-09-08 06:38:41 +00:00
|
|
|
|
2016-09-25 03:14:37 +00:00
|
|
|
flags |= INIT_DB_EXIST_OK;
|
2020-06-24 14:46:32 +00:00
|
|
|
return init_db(git_dir, real_git_dir, template_dir, hash_algo,
|
2023-05-16 06:33:43 +00:00
|
|
|
initial_branch, init_shared_repository, flags);
|
2005-04-07 22:13:13 +00:00
|
|
|
}
|