git-commit-vandalism/builtin/submodule--helper.c

2376 lines
64 KiB
C
Raw Normal View History

#define USE_THE_INDEX_COMPATIBILITY_MACROS
#include "builtin.h"
#include "repository.h"
#include "cache.h"
#include "config.h"
#include "parse-options.h"
#include "quote.h"
#include "pathspec.h"
#include "dir.h"
#include "submodule.h"
#include "submodule-config.h"
#include "string-list.h"
#include "run-command.h"
#include "remote.h"
#include "refs.h"
#include "refspec.h"
#include "connect.h"
#include "revision.h"
#include "diffcore.h"
#include "diff.h"
#include "object-store.h"
clone --recurse-submodules: prevent name squatting on Windows In addition to preventing `.git` from being tracked by Git, on Windows we also have to prevent `git~1` from being tracked, as the default NTFS short name (also known as the "8.3 filename") for the file name `.git` is `git~1`, otherwise it would be possible for malicious repositories to write directly into the `.git/` directory, e.g. a `post-checkout` hook that would then be executed _during_ a recursive clone. When we implemented appropriate protections in 2b4c6efc821 (read-cache: optionally disallow NTFS .git variants, 2014-12-16), we had analyzed carefully that the `.git` directory or file would be guaranteed to be the first directory entry to be written. Otherwise it would be possible e.g. for a file named `..git` to be assigned the short name `git~1` and subsequently, the short name generated for `.git` would be `git~2`. Or `git~3`. Or even `~9999999` (for a detailed explanation of the lengths we have to go to protect `.gitmodules`, see the commit message of e7cb0b4455c (is_ntfs_dotgit: match other .git files, 2018-05-11)). However, by exploiting two issues (that will be addressed in a related patch series close by), it is currently possible to clone a submodule into a non-empty directory: - On Windows, file names cannot end in a space or a period (for historical reasons: the period separating the base name from the file extension was not actually written to disk, and the base name/file extension was space-padded to the full 8/3 characters, respectively). Helpfully, when creating a directory under the name, say, `sub.`, that trailing period is trimmed automatically and the actual name on disk is `sub`. This means that while Git thinks that the submodule names `sub` and `sub.` are different, they both access `.git/modules/sub/`. - While the backslash character is a valid file name character on Linux, it is not so on Windows. As Git tries to be cross-platform, it therefore allows backslash characters in the file names stored in tree objects. Which means that it is totally possible that a submodule `c` sits next to a file `c\..git`, and on Windows, during recursive clone a file called `..git` will be written into `c/`, of course _before_ the submodule is cloned. Note that the actual exploit is not quite as simple as having a submodule `c` next to a file `c\..git`, as we have to make sure that the directory `.git/modules/b` already exists when the submodule is checked out, otherwise a different code path is taken in `module_clone()` that does _not_ allow a non-empty submodule directory to exist already. Even if we will address both issues nearby (the next commit will disallow backslash characters in tree entries' file names on Windows, and another patch will disallow creating directories/files with trailing spaces or periods), it is a wise idea to defend in depth against this sort of attack vector: when submodules are cloned recursively, we now _require_ the directory to be empty, addressing CVE-2019-1349. Note: the code path we patch is shared with the code path of `git submodule update --init`, which must not expect, in general, that the directory is empty. Hence we have to introduce the new option `--force-init` and hand it all the way down from `git submodule` to the actual `git submodule--helper` process that performs the initial clone. Reported-by: Nicolas Joly <Nicolas.Joly@microsoft.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
2019-09-12 14:20:39 +02:00
#include "dir.h"
#include "advice.h"
#define OPT_QUIET (1 << 0)
#define OPT_CACHED (1 << 1)
#define OPT_RECURSIVE (1 << 2)
#define OPT_FORCE (1 << 3)
typedef void (*each_submodule_fn)(const struct cache_entry *list_item,
void *cb_data);
static char *get_default_remote(void)
{
char *dest = NULL, *ret;
struct strbuf sb = STRBUF_INIT;
const char *refname = resolve_ref_unsafe("HEAD", 0, NULL, NULL);
if (!refname)
die(_("No such ref: %s"), "HEAD");
/* detached HEAD */
if (!strcmp(refname, "HEAD"))
return xstrdup("origin");
if (!skip_prefix(refname, "refs/heads/", &refname))
die(_("Expecting a full ref name, got %s"), refname);
strbuf_addf(&sb, "branch.%s.remote", refname);
if (git_config_get_string(sb.buf, &dest))
ret = xstrdup("origin");
else
ret = dest;
strbuf_release(&sb);
return ret;
}
static int print_default_remote(int argc, const char **argv, const char *prefix)
{
char *remote;
if (argc != 1)
die(_("submodule--helper print-default-remote takes no arguments"));
remote = get_default_remote();
if (remote)
printf("%s\n", remote);
free(remote);
return 0;
}
static int starts_with_dot_slash(const char *str)
{
return str[0] == '.' && is_dir_sep(str[1]);
}
static int starts_with_dot_dot_slash(const char *str)
{
return str[0] == '.' && str[1] == '.' && is_dir_sep(str[2]);
}
/*
* Returns 1 if it was the last chop before ':'.
*/
static int chop_last_dir(char **remoteurl, int is_relative)
{
char *rfind = find_last_dir_sep(*remoteurl);
if (rfind) {
*rfind = '\0';
return 0;
}
rfind = strrchr(*remoteurl, ':');
if (rfind) {
*rfind = '\0';
return 1;
}
if (is_relative || !strcmp(".", *remoteurl))
die(_("cannot strip one component off url '%s'"),
*remoteurl);
free(*remoteurl);
*remoteurl = xstrdup(".");
return 0;
}
/*
* The `url` argument is the URL that navigates to the submodule origin
* repo. When relative, this URL is relative to the superproject origin
* URL repo. The `up_path` argument, if specified, is the relative
* path that navigates from the submodule working tree to the superproject
* working tree. Returns the origin URL of the submodule.
*
* Return either an absolute URL or filesystem path (if the superproject
* origin URL is an absolute URL or filesystem path, respectively) or a
* relative file system path (if the superproject origin URL is a relative
* file system path).
*
* When the output is a relative file system path, the path is either
* relative to the submodule working tree, if up_path is specified, or to
* the superproject working tree otherwise.
*
* NEEDSWORK: This works incorrectly on the domain and protocol part.
* remote_url url outcome expectation
* http://a.com/b ../c http://a.com/c as is
* http://a.com/b/ ../c http://a.com/c same as previous line, but
* ignore trailing slash in url
* http://a.com/b ../../c http://c error out
* http://a.com/b ../../../c http:/c error out
* http://a.com/b ../../../../c http:c error out
* http://a.com/b ../../../../../c .:c error out
* NEEDSWORK: Given how chop_last_dir() works, this function is broken
* when a local part has a colon in its path component, too.
*/
static char *relative_url(const char *remote_url,
const char *url,
const char *up_path)
{
int is_relative = 0;
int colonsep = 0;
char *out;
char *remoteurl = xstrdup(remote_url);
struct strbuf sb = STRBUF_INIT;
size_t len = strlen(remoteurl);
if (is_dir_sep(remoteurl[len-1]))
remoteurl[len-1] = '\0';
if (!url_is_local_not_ssh(remoteurl) || is_absolute_path(remoteurl))
is_relative = 0;
else {
is_relative = 1;
/*
* Prepend a './' to ensure all relative
* remoteurls start with './' or '../'
*/
if (!starts_with_dot_slash(remoteurl) &&
!starts_with_dot_dot_slash(remoteurl)) {
strbuf_reset(&sb);
strbuf_addf(&sb, "./%s", remoteurl);
free(remoteurl);
remoteurl = strbuf_detach(&sb, NULL);
}
}
/*
* When the url starts with '../', remove that and the
* last directory in remoteurl.
*/
while (url) {
if (starts_with_dot_dot_slash(url)) {
url += 3;
colonsep |= chop_last_dir(&remoteurl, is_relative);
} else if (starts_with_dot_slash(url))
url += 2;
else
break;
}
strbuf_reset(&sb);
strbuf_addf(&sb, "%s%s%s", remoteurl, colonsep ? ":" : "/", url);
if (ends_with(url, "/"))
strbuf_setlen(&sb, sb.len - 1);
free(remoteurl);
if (starts_with_dot_slash(sb.buf))
out = xstrdup(sb.buf + 2);
else
out = xstrdup(sb.buf);
strbuf_reset(&sb);
if (!up_path || !is_relative)
return out;
strbuf_addf(&sb, "%s%s", up_path, out);
free(out);
return strbuf_detach(&sb, NULL);
}
static int resolve_relative_url(int argc, const char **argv, const char *prefix)
{
char *remoteurl = NULL;
char *remote = get_default_remote();
const char *up_path = NULL;
char *res;
const char *url;
struct strbuf sb = STRBUF_INIT;
if (argc != 2 && argc != 3)
die("resolve-relative-url only accepts one or two arguments");
url = argv[1];
strbuf_addf(&sb, "remote.%s.url", remote);
free(remote);
if (git_config_get_string(sb.buf, &remoteurl))
/* the repository is its own authoritative upstream */
remoteurl = xgetcwd();
if (argc == 3)
up_path = argv[2];
res = relative_url(remoteurl, url, up_path);
puts(res);
free(res);
free(remoteurl);
return 0;
}
static int resolve_relative_url_test(int argc, const char **argv, const char *prefix)
{
char *remoteurl, *res;
const char *up_path, *url;
if (argc != 4)
die("resolve-relative-url-test only accepts three arguments: <up_path> <remoteurl> <url>");
up_path = argv[1];
remoteurl = xstrdup(argv[2]);
url = argv[3];
if (!strcmp(up_path, "(null)"))
up_path = NULL;
res = relative_url(remoteurl, url, up_path);
puts(res);
free(res);
free(remoteurl);
return 0;
}
/* the result should be freed by the caller. */
static char *get_submodule_displaypath(const char *path, const char *prefix)
{
const char *super_prefix = get_super_prefix();
if (prefix && super_prefix) {
BUG("cannot have prefix '%s' and superprefix '%s'",
prefix, super_prefix);
} else if (prefix) {
struct strbuf sb = STRBUF_INIT;
char *displaypath = xstrdup(relative_path(path, prefix, &sb));
strbuf_release(&sb);
return displaypath;
} else if (super_prefix) {
return xstrfmt("%s%s", super_prefix, path);
} else {
return xstrdup(path);
}
}
static char *compute_rev_name(const char *sub_path, const char* object_id)
{
struct strbuf sb = STRBUF_INIT;
const char ***d;
static const char *describe_bare[] = { NULL };
static const char *describe_tags[] = { "--tags", NULL };
static const char *describe_contains[] = { "--contains", NULL };
static const char *describe_all_always[] = { "--all", "--always", NULL };
static const char **describe_argv[] = { describe_bare, describe_tags,
describe_contains,
describe_all_always, NULL };
for (d = describe_argv; *d; d++) {
struct child_process cp = CHILD_PROCESS_INIT;
prepare_submodule_repo_env(&cp.env_array);
cp.dir = sub_path;
cp.git_cmd = 1;
cp.no_stderr = 1;
strvec_push(&cp.args, "describe");
strvec_pushv(&cp.args, *d);
strvec_push(&cp.args, object_id);
if (!capture_command(&cp, &sb, 0)) {
strbuf_strip_suffix(&sb, "\n");
return strbuf_detach(&sb, NULL);
}
}
strbuf_release(&sb);
return NULL;
}
struct module_list {
const struct cache_entry **entries;
int alloc, nr;
};
#define MODULE_LIST_INIT { NULL, 0, 0 }
static int module_list_compute(int argc, const char **argv,
const char *prefix,
struct pathspec *pathspec,
struct module_list *list)
{
int i, result = 0;
char *ps_matched = NULL;
parse_pathspec(pathspec, 0,
PATHSPEC_PREFER_FULL,
prefix, argv);
if (pathspec->nr)
ps_matched = xcalloc(pathspec->nr, 1);
if (read_cache() < 0)
die(_("index file corrupt"));
for (i = 0; i < active_nr; i++) {
const struct cache_entry *ce = active_cache[i];
if (!match_pathspec(&the_index, pathspec, ce->name, ce_namelen(ce),
0, ps_matched, 1) ||
!S_ISGITLINK(ce->ce_mode))
continue;
ALLOC_GROW(list->entries, list->nr + 1, list->alloc);
list->entries[list->nr++] = ce;
while (i + 1 < active_nr &&
!strcmp(ce->name, active_cache[i + 1]->name))
/*
* Skip entries with the same name in different stages
* to make sure an entry is returned only once.
*/
i++;
}
if (ps_matched && report_path_error(ps_matched, pathspec))
result = -1;
free(ps_matched);
return result;
}
static void module_list_active(struct module_list *list)
{
int i;
struct module_list active_modules = MODULE_LIST_INIT;
for (i = 0; i < list->nr; i++) {
const struct cache_entry *ce = list->entries[i];
if (!is_submodule_active(the_repository, ce->name))
continue;
ALLOC_GROW(active_modules.entries,
active_modules.nr + 1,
active_modules.alloc);
active_modules.entries[active_modules.nr++] = ce;
}
free(list->entries);
*list = active_modules;
}
static char *get_up_path(const char *path)
{
int i;
struct strbuf sb = STRBUF_INIT;
for (i = count_slashes(path); i; i--)
strbuf_addstr(&sb, "../");
/*
* Check if 'path' ends with slash or not
* for having the same output for dir/sub_dir
* and dir/sub_dir/
*/
if (!is_dir_sep(path[strlen(path) - 1]))
strbuf_addstr(&sb, "../");
return strbuf_detach(&sb, NULL);
}
static int module_list(int argc, const char **argv, const char *prefix)
{
int i;
struct pathspec pathspec;
struct module_list list = MODULE_LIST_INIT;
struct option module_list_options[] = {
OPT_STRING(0, "prefix", &prefix,
N_("path"),
N_("alternative anchor for relative paths")),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
N_("git submodule--helper list [--prefix=<path>] [<path>...]"),
NULL
};
argc = parse_options(argc, argv, prefix, module_list_options,
git_submodule_helper_usage, 0);
if (module_list_compute(argc, argv, prefix, &pathspec, &list) < 0)
return 1;
for (i = 0; i < list.nr; i++) {
const struct cache_entry *ce = list.entries[i];
if (ce_stage(ce))
printf("%06o %s U\t", ce->ce_mode, oid_to_hex(&null_oid));
else
printf("%06o %s %d\t", ce->ce_mode,
oid_to_hex(&ce->oid), ce_stage(ce));
fprintf(stdout, "%s\n", ce->name);
}
return 0;
}
static void for_each_listed_submodule(const struct module_list *list,
each_submodule_fn fn, void *cb_data)
{
int i;
for (i = 0; i < list->nr; i++)
fn(list->entries[i], cb_data);
}
struct foreach_cb {
int argc;
const char **argv;
const char *prefix;
int quiet;
int recursive;
};
#define FOREACH_CB_INIT { 0 }
static void runcommand_in_submodule_cb(const struct cache_entry *list_item,
void *cb_data)
{
struct foreach_cb *info = cb_data;
const char *path = list_item->name;
const struct object_id *ce_oid = &list_item->oid;
const struct submodule *sub;
struct child_process cp = CHILD_PROCESS_INIT;
char *displaypath;
displaypath = get_submodule_displaypath(path, info->prefix);
sub = submodule_from_path(the_repository, &null_oid, path);
if (!sub)
die(_("No url found for submodule path '%s' in .gitmodules"),
displaypath);
if (!is_submodule_populated_gently(path, NULL))
goto cleanup;
prepare_submodule_repo_env(&cp.env_array);
/*
* For the purpose of executing <command> in the submodule,
* separate shell is used for the purpose of running the
* child process.
*/
cp.use_shell = 1;
cp.dir = path;
/*
* NEEDSWORK: the command currently has access to the variables $name,
* $sm_path, $displaypath, $sha1 and $toplevel only when the command
* contains a single argument. This is done for maintaining a faithful
* translation from shell script.
*/
if (info->argc == 1) {
char *toplevel = xgetcwd();
struct strbuf sb = STRBUF_INIT;
strvec_pushf(&cp.env_array, "name=%s", sub->name);
strvec_pushf(&cp.env_array, "sm_path=%s", path);
strvec_pushf(&cp.env_array, "displaypath=%s", displaypath);
strvec_pushf(&cp.env_array, "sha1=%s",
oid_to_hex(ce_oid));
strvec_pushf(&cp.env_array, "toplevel=%s", toplevel);
/*
* Since the path variable was accessible from the script
* before porting, it is also made available after porting.
* The environment variable "PATH" has a very special purpose
* on windows. And since environment variables are
* case-insensitive in windows, it interferes with the
* existing PATH variable. Hence, to avoid that, we expose
* path via the args strvec and not via env_array.
*/
sq_quote_buf(&sb, path);
strvec_pushf(&cp.args, "path=%s; %s",
sb.buf, info->argv[0]);
strbuf_release(&sb);
free(toplevel);
} else {
strvec_pushv(&cp.args, info->argv);
}
if (!info->quiet)
printf(_("Entering '%s'\n"), displaypath);
if (info->argv[0] && run_command(&cp))
die(_("run_command returned non-zero status for %s\n."),
displaypath);
if (info->recursive) {
struct child_process cpr = CHILD_PROCESS_INIT;
cpr.git_cmd = 1;
cpr.dir = path;
prepare_submodule_repo_env(&cpr.env_array);
strvec_pushl(&cpr.args, "--super-prefix", NULL);
strvec_pushf(&cpr.args, "%s/", displaypath);
strvec_pushl(&cpr.args, "submodule--helper", "foreach", "--recursive",
NULL);
if (info->quiet)
strvec_push(&cpr.args, "--quiet");
strvec_push(&cpr.args, "--");
strvec_pushv(&cpr.args, info->argv);
if (run_command(&cpr))
die(_("run_command returned non-zero status while "
"recursing in the nested submodules of %s\n."),
displaypath);
}
cleanup:
free(displaypath);
}
static int module_foreach(int argc, const char **argv, const char *prefix)
{
struct foreach_cb info = FOREACH_CB_INIT;
struct pathspec pathspec;
struct module_list list = MODULE_LIST_INIT;
struct option module_foreach_options[] = {
OPT__QUIET(&info.quiet, N_("Suppress output of entering each submodule command")),
OPT_BOOL(0, "recursive", &info.recursive,
N_("Recurse into nested submodules")),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
submodule foreach: fix "<command> --quiet" not being respected Robin reported that git submodule foreach --quiet git pull --quiet origin is not really quiet anymore [1]. "git pull" behaves as if --quiet is not given. This happens because parseopt in submodule--helper will try to parse both --quiet options as if they are foreach's options, not git-pull's. The parsed options are removed from the command line. So when we do pull later, we execute just this git pull origin When calling submodule helper, adding "--" in front of "git pull" will stop parseopt for parsing options that do not really belong to submodule--helper foreach. PARSE_OPT_KEEP_UNKNOWN is removed as a safety measure. parseopt should never see unknown options or something has gone wrong. There are also a couple usage string update while I'm looking at them. While at it, I also add "--" to other subcommands that pass "$@" to submodule--helper. "$@" in these cases are paths and less likely to be --something-like-this. But the point still stands, git-submodule has parsed and classified what are options, what are paths. submodule--helper should never consider paths passed by git-submodule to be options even if they look like one. The test case is also contributed by Robin. [1] it should be quiet before fc1b9243cd (submodule: port submodule subcommand 'foreach' from shell to C, 2018-05-10) because parseopt can't accidentally eat options then. Reported-by: Robin H. Johnson <robbat2@gentoo.org> Tested-by: Robin H. Johnson <robbat2@gentoo.org> Signed-off-by: Robin H. Johnson <robbat2@gentoo.org> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-12 12:08:19 +02:00
N_("git submodule--helper foreach [--quiet] [--recursive] [--] <command>"),
NULL
};
argc = parse_options(argc, argv, prefix, module_foreach_options,
submodule foreach: fix "<command> --quiet" not being respected Robin reported that git submodule foreach --quiet git pull --quiet origin is not really quiet anymore [1]. "git pull" behaves as if --quiet is not given. This happens because parseopt in submodule--helper will try to parse both --quiet options as if they are foreach's options, not git-pull's. The parsed options are removed from the command line. So when we do pull later, we execute just this git pull origin When calling submodule helper, adding "--" in front of "git pull" will stop parseopt for parsing options that do not really belong to submodule--helper foreach. PARSE_OPT_KEEP_UNKNOWN is removed as a safety measure. parseopt should never see unknown options or something has gone wrong. There are also a couple usage string update while I'm looking at them. While at it, I also add "--" to other subcommands that pass "$@" to submodule--helper. "$@" in these cases are paths and less likely to be --something-like-this. But the point still stands, git-submodule has parsed and classified what are options, what are paths. submodule--helper should never consider paths passed by git-submodule to be options even if they look like one. The test case is also contributed by Robin. [1] it should be quiet before fc1b9243cd (submodule: port submodule subcommand 'foreach' from shell to C, 2018-05-10) because parseopt can't accidentally eat options then. Reported-by: Robin H. Johnson <robbat2@gentoo.org> Tested-by: Robin H. Johnson <robbat2@gentoo.org> Signed-off-by: Robin H. Johnson <robbat2@gentoo.org> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-12 12:08:19 +02:00
git_submodule_helper_usage, 0);
if (module_list_compute(0, NULL, prefix, &pathspec, &list) < 0)
return 1;
info.argc = argc;
info.argv = argv;
info.prefix = prefix;
for_each_listed_submodule(&list, runcommand_in_submodule_cb, &info);
return 0;
}
submodule helper: convert relative URL to absolute URL if needed The submodule helper update_clone called by "git submodule update", clones submodules if needed. As submodules used to have the URL indicating if they were active, the step to resolve relative URLs was done in the "submodule init" step. Nowadays submodules can be configured active without calling an explicit init, e.g. via configuring submodule.active. When trying to obtain submodules that are set active this way, we'll fallback to the URL found in the .gitmodules, which may be relative to the superproject, but we do not resolve it, yet: git clone https://gerrit.googlesource.com/gerrit cd gerrit && grep url .gitmodules url = ../plugins/codemirror-editor ... git config submodule.active . git submodule update fatal: repository '../plugins/codemirror-editor' does not exist fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor'. Retry scheduled [...] fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor' a second time, aborting [...] To resolve the issue, factor out the function that resolves the relative URLs in "git submodule init" (in the submodule helper in the init_submodule function) and call it at the appropriate place in the update_clone helper. Reported-by: Jaewoong Jung <jungjw@google.com> Signed-off-by: Stefan Beller <sbeller@google.com> Reviewed-by: Jonathan Nieder <jrnieder@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-10-16 19:27:03 +02:00
static char *compute_submodule_clone_url(const char *rel_url)
{
char *remoteurl, *relurl;
char *remote = get_default_remote();
struct strbuf remotesb = STRBUF_INIT;
strbuf_addf(&remotesb, "remote.%s.url", remote);
if (git_config_get_string(remotesb.buf, &remoteurl)) {
warning(_("could not look up configuration '%s'. Assuming this repository is its own authoritative upstream."), remotesb.buf);
remoteurl = xgetcwd();
}
relurl = relative_url(remoteurl, rel_url, NULL);
free(remote);
free(remoteurl);
strbuf_release(&remotesb);
return relurl;
}
struct init_cb {
const char *prefix;
unsigned int flags;
};
#define INIT_CB_INIT { NULL, 0 }
static void init_submodule(const char *path, const char *prefix,
unsigned int flags)
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
{
const struct submodule *sub;
struct strbuf sb = STRBUF_INIT;
char *upd = NULL, *url = NULL, *displaypath;
displaypath = get_submodule_displaypath(path, prefix);
sub = submodule_from_path(the_repository, &null_oid, path);
if (!sub)
die(_("No url found for submodule path '%s' in .gitmodules"),
displaypath);
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
/*
* NEEDSWORK: In a multi-working-tree world, this needs to be
* set in the per-worktree config.
*
* Set active flag for the submodule being initialized
*/
if (!is_submodule_active(the_repository, path)) {
strbuf_addf(&sb, "submodule.%s.active", sub->name);
git_config_set_gently(sb.buf, "true");
strbuf_reset(&sb);
}
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
/*
* Copy url setting when it is not set yet.
* To look up the url in .git/config, we must not fall back to
* .gitmodules, so look it up directly.
*/
strbuf_addf(&sb, "submodule.%s.url", sub->name);
if (git_config_get_string(sb.buf, &url)) {
if (!sub->url)
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
die(_("No url found for submodule path '%s' in .gitmodules"),
displaypath);
url = xstrdup(sub->url);
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
/* Possibly a url relative to parent */
if (starts_with_dot_dot_slash(url) ||
starts_with_dot_slash(url)) {
submodule helper: convert relative URL to absolute URL if needed The submodule helper update_clone called by "git submodule update", clones submodules if needed. As submodules used to have the URL indicating if they were active, the step to resolve relative URLs was done in the "submodule init" step. Nowadays submodules can be configured active without calling an explicit init, e.g. via configuring submodule.active. When trying to obtain submodules that are set active this way, we'll fallback to the URL found in the .gitmodules, which may be relative to the superproject, but we do not resolve it, yet: git clone https://gerrit.googlesource.com/gerrit cd gerrit && grep url .gitmodules url = ../plugins/codemirror-editor ... git config submodule.active . git submodule update fatal: repository '../plugins/codemirror-editor' does not exist fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor'. Retry scheduled [...] fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor' a second time, aborting [...] To resolve the issue, factor out the function that resolves the relative URLs in "git submodule init" (in the submodule helper in the init_submodule function) and call it at the appropriate place in the update_clone helper. Reported-by: Jaewoong Jung <jungjw@google.com> Signed-off-by: Stefan Beller <sbeller@google.com> Reviewed-by: Jonathan Nieder <jrnieder@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-10-16 19:27:03 +02:00
char *oldurl = url;
url = compute_submodule_clone_url(oldurl);
free(oldurl);
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
}
if (git_config_set_gently(sb.buf, url))
die(_("Failed to register url for submodule path '%s'"),
displaypath);
if (!(flags & OPT_QUIET))
fprintf(stderr,
_("Submodule '%s' (%s) registered for path '%s'\n"),
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
sub->name, url, displaypath);
}
strbuf_reset(&sb);
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
/* Copy "update" setting when it is not set yet */
strbuf_addf(&sb, "submodule.%s.update", sub->name);
if (git_config_get_string(sb.buf, &upd) &&
sub->update_strategy.type != SM_UPDATE_UNSPECIFIED) {
if (sub->update_strategy.type == SM_UPDATE_COMMAND) {
fprintf(stderr, _("warning: command update mode suggested for submodule '%s'\n"),
sub->name);
upd = xstrdup("none");
} else
upd = xstrdup(submodule_strategy_to_string(&sub->update_strategy));
if (git_config_set_gently(sb.buf, upd))
die(_("Failed to register update mode for submodule path '%s'"), displaypath);
}
strbuf_release(&sb);
free(displaypath);
free(url);
free(upd);
}
static void init_submodule_cb(const struct cache_entry *list_item, void *cb_data)
{
struct init_cb *info = cb_data;
init_submodule(list_item->name, info->prefix, info->flags);
}
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
static int module_init(int argc, const char **argv, const char *prefix)
{
struct init_cb info = INIT_CB_INIT;
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
struct pathspec pathspec;
struct module_list list = MODULE_LIST_INIT;
int quiet = 0;
struct option module_init_options[] = {
OPT__QUIET(&quiet, N_("Suppress output for initializing a submodule")),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
submodule foreach: fix "<command> --quiet" not being respected Robin reported that git submodule foreach --quiet git pull --quiet origin is not really quiet anymore [1]. "git pull" behaves as if --quiet is not given. This happens because parseopt in submodule--helper will try to parse both --quiet options as if they are foreach's options, not git-pull's. The parsed options are removed from the command line. So when we do pull later, we execute just this git pull origin When calling submodule helper, adding "--" in front of "git pull" will stop parseopt for parsing options that do not really belong to submodule--helper foreach. PARSE_OPT_KEEP_UNKNOWN is removed as a safety measure. parseopt should never see unknown options or something has gone wrong. There are also a couple usage string update while I'm looking at them. While at it, I also add "--" to other subcommands that pass "$@" to submodule--helper. "$@" in these cases are paths and less likely to be --something-like-this. But the point still stands, git-submodule has parsed and classified what are options, what are paths. submodule--helper should never consider paths passed by git-submodule to be options even if they look like one. The test case is also contributed by Robin. [1] it should be quiet before fc1b9243cd (submodule: port submodule subcommand 'foreach' from shell to C, 2018-05-10) because parseopt can't accidentally eat options then. Reported-by: Robin H. Johnson <robbat2@gentoo.org> Tested-by: Robin H. Johnson <robbat2@gentoo.org> Signed-off-by: Robin H. Johnson <robbat2@gentoo.org> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-12 12:08:19 +02:00
N_("git submodule--helper init [<options>] [<path>]"),
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
NULL
};
argc = parse_options(argc, argv, prefix, module_init_options,
git_submodule_helper_usage, 0);
if (module_list_compute(argc, argv, prefix, &pathspec, &list) < 0)
return 1;
/*
* If there are no path args and submodule.active is set then,
* by default, only initialize 'active' modules.
*/
if (!argc && git_config_get_value_multi("submodule.active"))
module_list_active(&list);
info.prefix = prefix;
if (quiet)
info.flags |= OPT_QUIET;
for_each_listed_submodule(&list, init_submodule_cb, &info);
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
return 0;
}
struct status_cb {
const char *prefix;
unsigned int flags;
};
#define STATUS_CB_INIT { NULL, 0 }
static void print_status(unsigned int flags, char state, const char *path,
const struct object_id *oid, const char *displaypath)
{
if (flags & OPT_QUIET)
return;
printf("%c%s %s", state, oid_to_hex(oid), displaypath);
if (state == ' ' || state == '+') {
const char *name = compute_rev_name(path, oid_to_hex(oid));
if (name)
printf(" (%s)", name);
}
printf("\n");
}
static int handle_submodule_head_ref(const char *refname,
const struct object_id *oid, int flags,
void *cb_data)
{
struct object_id *output = cb_data;
if (oid)
oidcpy(output, oid);
return 0;
}
static void status_submodule(const char *path, const struct object_id *ce_oid,
unsigned int ce_flags, const char *prefix,
unsigned int flags)
{
char *displaypath;
struct strvec diff_files_args = STRVEC_INIT;
struct rev_info rev;
int diff_files_result;
struct strbuf buf = STRBUF_INIT;
const char *git_dir;
if (!submodule_from_path(the_repository, &null_oid, path))
die(_("no submodule mapping found in .gitmodules for path '%s'"),
path);
displaypath = get_submodule_displaypath(path, prefix);
if ((CE_STAGEMASK & ce_flags) >> CE_STAGESHIFT) {
print_status(flags, 'U', path, &null_oid, displaypath);
goto cleanup;
}
strbuf_addf(&buf, "%s/.git", path);
git_dir = read_gitfile(buf.buf);
if (!git_dir)
git_dir = buf.buf;
if (!is_submodule_active(the_repository, path) ||
!is_git_directory(git_dir)) {
print_status(flags, '-', path, ce_oid, displaypath);
strbuf_release(&buf);
goto cleanup;
}
strbuf_release(&buf);
strvec_pushl(&diff_files_args, "diff-files",
"--ignore-submodules=dirty", "--quiet", "--",
path, NULL);
git_config(git_diff_basic_config, NULL);
repo_init_revisions(the_repository, &rev, NULL);
rev.abbrev = 0;
diff_files_args.argc = setup_revisions(diff_files_args.argc,
diff_files_args.argv,
&rev, NULL);
diff_files_result = run_diff_files(&rev, 0);
if (!diff_result_code(&rev.diffopt, diff_files_result)) {
print_status(flags, ' ', path, ce_oid,
displaypath);
} else if (!(flags & OPT_CACHED)) {
struct object_id oid;
struct ref_store *refs = get_submodule_ref_store(path);
if (!refs) {
print_status(flags, '-', path, ce_oid, displaypath);
goto cleanup;
}
if (refs_head_ref(refs, handle_submodule_head_ref, &oid))
die(_("could not resolve HEAD ref inside the "
"submodule '%s'"), path);
print_status(flags, '+', path, &oid, displaypath);
} else {
print_status(flags, '+', path, ce_oid, displaypath);
}
if (flags & OPT_RECURSIVE) {
struct child_process cpr = CHILD_PROCESS_INIT;
cpr.git_cmd = 1;
cpr.dir = path;
prepare_submodule_repo_env(&cpr.env_array);
strvec_push(&cpr.args, "--super-prefix");
strvec_pushf(&cpr.args, "%s/", displaypath);
strvec_pushl(&cpr.args, "submodule--helper", "status",
"--recursive", NULL);
if (flags & OPT_CACHED)
strvec_push(&cpr.args, "--cached");
if (flags & OPT_QUIET)
strvec_push(&cpr.args, "--quiet");
if (run_command(&cpr))
die(_("failed to recurse into submodule '%s'"), path);
}
cleanup:
strvec_clear(&diff_files_args);
free(displaypath);
}
static void status_submodule_cb(const struct cache_entry *list_item,
void *cb_data)
{
struct status_cb *info = cb_data;
status_submodule(list_item->name, &list_item->oid, list_item->ce_flags,
info->prefix, info->flags);
}
static int module_status(int argc, const char **argv, const char *prefix)
{
struct status_cb info = STATUS_CB_INIT;
struct pathspec pathspec;
struct module_list list = MODULE_LIST_INIT;
int quiet = 0;
struct option module_status_options[] = {
OPT__QUIET(&quiet, N_("Suppress submodule status output")),
OPT_BIT(0, "cached", &info.flags, N_("Use commit stored in the index instead of the one stored in the submodule HEAD"), OPT_CACHED),
OPT_BIT(0, "recursive", &info.flags, N_("recurse into nested submodules"), OPT_RECURSIVE),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
N_("git submodule status [--quiet] [--cached] [--recursive] [<path>...]"),
NULL
};
argc = parse_options(argc, argv, prefix, module_status_options,
git_submodule_helper_usage, 0);
if (module_list_compute(argc, argv, prefix, &pathspec, &list) < 0)
return 1;
info.prefix = prefix;
if (quiet)
info.flags |= OPT_QUIET;
for_each_listed_submodule(&list, status_submodule_cb, &info);
submodule: port init from shell to C By having the `submodule init` functionality in C, we can reference it easier from other parts in the code in later patches. The code is split up to have one function to initialize one submodule and a calling function that takes care of the rest, such as argument handling and translating the arguments to the paths of the submodules. This is the first submodule subcommand that is fully converted to C except for the usage string, so this is actually removing a call to the `submodule--helper list` function, which is supposed to be used in this transition. Instead we'll make a direct call to `module_list_compute`. An explanation why we need to edit the prefixes in cmd_update in git-submodule.sh in this patch: By having no processing in the shell part, we need to convey the notion of wt_prefix and prefix to the C parts, which former patches punted on and did the processing of displaying path in the shell. `wt_prefix` used to hold the path from the repository root to the current directory, e.g. wt_prefix would be t/ if the user invoked the `git submodule` command in ~/repo/t and ~repo is the GIT_DIR. `prefix` used to hold the relative path from the repository root to the operation, e.g. if you have recursive submodules, the shell script would modify the `prefix` in each recursive step by adding the submodule path. We will pass `wt_prefix` into the C helper via `git -C <dir>` as that will setup git in the directory the user actually called git-submodule.sh from. The `prefix` will be passed in via the `--prefix` option. Having `prefix` and `wt_prefix` relative to the GIT_DIR of the calling superproject is unfortunate with this patch as the C code doesn't know about a possible recursion from a superproject via `submodule update --init --recursive`. To fix this, we change the meaning of `wt_prefix` to point to the current project instead of the superproject and `prefix` to include any relative paths issues in the superproject. That way `prefix` will become the leading part for displaying paths and `wt_prefix` will be empty in recursive calls for now. The new notion of `wt_prefix` and `prefix` still allows us to reconstruct the calling directory in the superproject by just traveling reverse of `prefix`. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-04-16 02:50:13 +02:00
return 0;
}
static int module_name(int argc, const char **argv, const char *prefix)
{
const struct submodule *sub;
if (argc != 2)
usage(_("git submodule--helper name <path>"));
sub = submodule_from_path(the_repository, &null_oid, argv[1]);
if (!sub)
die(_("no submodule mapping found in .gitmodules for path '%s'"),
argv[1]);
printf("%s\n", sub->name);
return 0;
}
struct sync_cb {
const char *prefix;
unsigned int flags;
};
#define SYNC_CB_INIT { NULL, 0 }
static void sync_submodule(const char *path, const char *prefix,
unsigned int flags)
{
const struct submodule *sub;
char *remote_key = NULL;
char *sub_origin_url, *super_config_url, *displaypath;
struct strbuf sb = STRBUF_INIT;
struct child_process cp = CHILD_PROCESS_INIT;
char *sub_config_path = NULL;
if (!is_submodule_active(the_repository, path))
return;
sub = submodule_from_path(the_repository, &null_oid, path);
if (sub && sub->url) {
if (starts_with_dot_dot_slash(sub->url) ||
starts_with_dot_slash(sub->url)) {
char *remote_url, *up_path;
char *remote = get_default_remote();
strbuf_addf(&sb, "remote.%s.url", remote);
if (git_config_get_string(sb.buf, &remote_url))
remote_url = xgetcwd();
up_path = get_up_path(path);
sub_origin_url = relative_url(remote_url, sub->url, up_path);
super_config_url = relative_url(remote_url, sub->url, NULL);
free(remote);
free(up_path);
free(remote_url);
} else {
sub_origin_url = xstrdup(sub->url);
super_config_url = xstrdup(sub->url);
}
} else {
sub_origin_url = xstrdup("");
super_config_url = xstrdup("");
}
displaypath = get_submodule_displaypath(path, prefix);
if (!(flags & OPT_QUIET))
printf(_("Synchronizing submodule url for '%s'\n"),
displaypath);
strbuf_reset(&sb);
strbuf_addf(&sb, "submodule.%s.url", sub->name);
if (git_config_set_gently(sb.buf, super_config_url))
die(_("failed to register url for submodule path '%s'"),
displaypath);
if (!is_submodule_populated_gently(path, NULL))
goto cleanup;
prepare_submodule_repo_env(&cp.env_array);
cp.git_cmd = 1;
cp.dir = path;
strvec_pushl(&cp.args, "submodule--helper",
"print-default-remote", NULL);
strbuf_reset(&sb);
if (capture_command(&cp, &sb, 0))
die(_("failed to get the default remote for submodule '%s'"),
path);
strbuf_strip_suffix(&sb, "\n");
remote_key = xstrfmt("remote.%s.url", sb.buf);
strbuf_reset(&sb);
submodule_to_gitdir(&sb, path);
strbuf_addstr(&sb, "/config");
if (git_config_set_in_file_gently(sb.buf, remote_key, sub_origin_url))
die(_("failed to update remote for submodule '%s'"),
path);
if (flags & OPT_RECURSIVE) {
struct child_process cpr = CHILD_PROCESS_INIT;
cpr.git_cmd = 1;
cpr.dir = path;
prepare_submodule_repo_env(&cpr.env_array);
strvec_push(&cpr.args, "--super-prefix");
strvec_pushf(&cpr.args, "%s/", displaypath);
strvec_pushl(&cpr.args, "submodule--helper", "sync",
"--recursive", NULL);
if (flags & OPT_QUIET)
strvec_push(&cpr.args, "--quiet");
if (run_command(&cpr))
die(_("failed to recurse into submodule '%s'"),
path);
}
cleanup:
free(super_config_url);
free(sub_origin_url);
strbuf_release(&sb);
free(remote_key);
free(displaypath);
free(sub_config_path);
}
static void sync_submodule_cb(const struct cache_entry *list_item, void *cb_data)
{
struct sync_cb *info = cb_data;
sync_submodule(list_item->name, info->prefix, info->flags);
}
static int module_sync(int argc, const char **argv, const char *prefix)
{
struct sync_cb info = SYNC_CB_INIT;
struct pathspec pathspec;
struct module_list list = MODULE_LIST_INIT;
int quiet = 0;
int recursive = 0;
struct option module_sync_options[] = {
OPT__QUIET(&quiet, N_("Suppress output of synchronizing submodule url")),
OPT_BOOL(0, "recursive", &recursive,
N_("Recurse into nested submodules")),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
N_("git submodule--helper sync [--quiet] [--recursive] [<path>]"),
NULL
};
argc = parse_options(argc, argv, prefix, module_sync_options,
git_submodule_helper_usage, 0);
if (module_list_compute(argc, argv, prefix, &pathspec, &list) < 0)
return 1;
info.prefix = prefix;
if (quiet)
info.flags |= OPT_QUIET;
if (recursive)
info.flags |= OPT_RECURSIVE;
for_each_listed_submodule(&list, sync_submodule_cb, &info);
return 0;
}
struct deinit_cb {
const char *prefix;
unsigned int flags;
};
#define DEINIT_CB_INIT { NULL, 0 }
static void deinit_submodule(const char *path, const char *prefix,
unsigned int flags)
{
const struct submodule *sub;
char *displaypath = NULL;
struct child_process cp_config = CHILD_PROCESS_INIT;
struct strbuf sb_config = STRBUF_INIT;
char *sub_git_dir = xstrfmt("%s/.git", path);
sub = submodule_from_path(the_repository, &null_oid, path);
if (!sub || !sub->name)
goto cleanup;
displaypath = get_submodule_displaypath(path, prefix);
/* remove the submodule work tree (unless the user already did it) */
if (is_directory(path)) {
struct strbuf sb_rm = STRBUF_INIT;
const char *format;
/*
* protect submodules containing a .git directory
* NEEDSWORK: instead of dying, automatically call
* absorbgitdirs and (possibly) warn.
*/
if (is_directory(sub_git_dir))
die(_("Submodule work tree '%s' contains a .git "
"directory (use 'rm -rf' if you really want "
"to remove it including all of its history)"),
displaypath);
if (!(flags & OPT_FORCE)) {
struct child_process cp_rm = CHILD_PROCESS_INIT;
cp_rm.git_cmd = 1;
strvec_pushl(&cp_rm.args, "rm", "-qn",
path, NULL);
if (run_command(&cp_rm))
die(_("Submodule work tree '%s' contains local "
"modifications; use '-f' to discard them"),
displaypath);
}
strbuf_addstr(&sb_rm, path);
if (!remove_dir_recursively(&sb_rm, 0))
format = _("Cleared directory '%s'\n");
else
format = _("Could not remove submodule work tree '%s'\n");
if (!(flags & OPT_QUIET))
printf(format, displaypath);
submodule_unset_core_worktree(sub);
strbuf_release(&sb_rm);
}
if (mkdir(path, 0777))
printf(_("could not create empty submodule directory %s"),
displaypath);
cp_config.git_cmd = 1;
strvec_pushl(&cp_config.args, "config", "--get-regexp", NULL);
strvec_pushf(&cp_config.args, "submodule.%s\\.", sub->name);
/* remove the .git/config entries (unless the user already did it) */
if (!capture_command(&cp_config, &sb_config, 0) && sb_config.len) {
char *sub_key = xstrfmt("submodule.%s", sub->name);
/*
* remove the whole section so we have a clean state when
* the user later decides to init this submodule again
*/
git_config_rename_section_in_file(NULL, sub_key, NULL);
if (!(flags & OPT_QUIET))
printf(_("Submodule '%s' (%s) unregistered for path '%s'\n"),
sub->name, sub->url, displaypath);
free(sub_key);
}
cleanup:
free(displaypath);
free(sub_git_dir);
strbuf_release(&sb_config);
}
static void deinit_submodule_cb(const struct cache_entry *list_item,
void *cb_data)
{
struct deinit_cb *info = cb_data;
deinit_submodule(list_item->name, info->prefix, info->flags);
}
static int module_deinit(int argc, const char **argv, const char *prefix)
{
struct deinit_cb info = DEINIT_CB_INIT;
struct pathspec pathspec;
struct module_list list = MODULE_LIST_INIT;
int quiet = 0;
int force = 0;
int all = 0;
struct option module_deinit_options[] = {
OPT__QUIET(&quiet, N_("Suppress submodule status output")),
OPT__FORCE(&force, N_("Remove submodule working trees even if they contain local changes"), 0),
OPT_BOOL(0, "all", &all, N_("Unregister all submodules")),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
N_("git submodule deinit [--quiet] [-f | --force] [--all | [--] [<path>...]]"),
NULL
};
argc = parse_options(argc, argv, prefix, module_deinit_options,
git_submodule_helper_usage, 0);
if (all && argc) {
error("pathspec and --all are incompatible");
usage_with_options(git_submodule_helper_usage,
module_deinit_options);
}
if (!argc && !all)
die(_("Use '--all' if you really want to deinitialize all submodules"));
if (module_list_compute(argc, argv, prefix, &pathspec, &list) < 0)
return 1;
info.prefix = prefix;
if (quiet)
info.flags |= OPT_QUIET;
if (force)
info.flags |= OPT_FORCE;
for_each_listed_submodule(&list, deinit_submodule_cb, &info);
return 0;
}
static int clone_submodule(const char *path, const char *gitdir, const char *url,
const char *depth, struct string_list *reference, int dissociate,
int quiet, int progress, int single_branch)
{
struct child_process cp = CHILD_PROCESS_INIT;
strvec_push(&cp.args, "clone");
strvec_push(&cp.args, "--no-checkout");
if (quiet)
strvec_push(&cp.args, "--quiet");
clone: pass --progress decision to recursive submodules When cloning with "--recursive", we'd generally expect submodules to show progress reports if the main clone did, too. In older versions of git, this mostly worked out of the box. Since we show progress by default when stderr is a tty, and since the child clones inherit the parent stderr, then both processes would come to the same decision by default. If the parent clone was asked for "--quiet", we passed down "--quiet" to the child. However, if stderr was not a tty and the user specified "--progress", we did not propagate this to the child. That's a minor bug, but things got much worse when we switched recently to submodule--helper's update_clone command. With that change, the stderr of the child clones are always connected to a pipe, and we never output progress at all. This patch teaches git-submodule and git-submodule--helper how to pass down an explicit "--progress" flag when cloning. The clone command then decides to propagate that flag based on the cloning decision made earlier (which takes into account isatty(2) of the parent process, existing --progress or --quiet flags, etc). Since the child processes always run without a tty on stderr, we don't have to worry about passing an explicit "--no-progress"; it's the default for them. This fixes the recent loss of progress during recursive clones. And as a bonus, it makes: git clone --recursive --progress ... 2>&1 | cat work by triggering progress explicitly in the children. Signed-off-by: Jeff King <peff@peff.net> Acked-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-22 07:24:46 +02:00
if (progress)
strvec_push(&cp.args, "--progress");
if (depth && *depth)
strvec_pushl(&cp.args, "--depth", depth, NULL);
if (reference->nr) {
struct string_list_item *item;
for_each_string_list_item(item, reference)
strvec_pushl(&cp.args, "--reference",
item->string, NULL);
}
if (dissociate)
strvec_push(&cp.args, "--dissociate");
if (gitdir && *gitdir)
strvec_pushl(&cp.args, "--separate-git-dir", gitdir, NULL);
if (single_branch >= 0)
strvec_push(&cp.args, single_branch ?
"--single-branch" :
"--no-single-branch");
strvec_push(&cp.args, "--");
strvec_push(&cp.args, url);
strvec_push(&cp.args, path);
cp.git_cmd = 1;
prepare_submodule_repo_env(&cp.env_array);
cp.no_stdin = 1;
return run_command(&cp);
}
struct submodule_alternate_setup {
const char *submodule_name;
enum SUBMODULE_ALTERNATE_ERROR_MODE {
SUBMODULE_ALTERNATE_ERROR_DIE,
SUBMODULE_ALTERNATE_ERROR_INFO,
SUBMODULE_ALTERNATE_ERROR_IGNORE
} error_mode;
struct string_list *reference;
};
#define SUBMODULE_ALTERNATE_SETUP_INIT { NULL, \
SUBMODULE_ALTERNATE_ERROR_IGNORE, NULL }
static const char alternate_error_advice[] = N_(
"An alternate computed from a superproject's alternate is invalid.\n"
"To allow Git to clone without an alternate in such a case, set\n"
"submodule.alternateErrorStrategy to 'info' or, equivalently, clone with\n"
"'--reference-if-able' instead of '--reference'."
);
static int add_possible_reference_from_superproject(
struct object_directory *odb, void *sas_cb)
{
struct submodule_alternate_setup *sas = sas_cb;
size_t len;
/*
* If the alternate object store is another repository, try the
* standard layout with .git/(modules/<name>)+/objects
*/
if (strip_suffix(odb->path, "/objects", &len)) {
char *sm_alternate;
struct strbuf sb = STRBUF_INIT;
struct strbuf err = STRBUF_INIT;
strbuf_add(&sb, odb->path, len);
/*
* We need to end the new path with '/' to mark it as a dir,
* otherwise a submodule name containing '/' will be broken
* as the last part of a missing submodule reference would
* be taken as a file name.
*/
strbuf_addf(&sb, "/modules/%s/", sas->submodule_name);
sm_alternate = compute_alternate_path(sb.buf, &err);
if (sm_alternate) {
string_list_append(sas->reference, xstrdup(sb.buf));
free(sm_alternate);
} else {
switch (sas->error_mode) {
case SUBMODULE_ALTERNATE_ERROR_DIE:
if (advice_submodule_alternate_error_strategy_die)
advise(_(alternate_error_advice));
die(_("submodule '%s' cannot add alternate: %s"),
sas->submodule_name, err.buf);
case SUBMODULE_ALTERNATE_ERROR_INFO:
fprintf_ln(stderr, _("submodule '%s' cannot add alternate: %s"),
sas->submodule_name, err.buf);
case SUBMODULE_ALTERNATE_ERROR_IGNORE:
; /* nothing */
}
}
strbuf_release(&sb);
}
return 0;
}
static void prepare_possible_alternates(const char *sm_name,
struct string_list *reference)
{
char *sm_alternate = NULL, *error_strategy = NULL;
struct submodule_alternate_setup sas = SUBMODULE_ALTERNATE_SETUP_INIT;
git_config_get_string("submodule.alternateLocation", &sm_alternate);
if (!sm_alternate)
return;
git_config_get_string("submodule.alternateErrorStrategy", &error_strategy);
if (!error_strategy)
error_strategy = xstrdup("die");
sas.submodule_name = sm_name;
sas.reference = reference;
if (!strcmp(error_strategy, "die"))
sas.error_mode = SUBMODULE_ALTERNATE_ERROR_DIE;
else if (!strcmp(error_strategy, "info"))
sas.error_mode = SUBMODULE_ALTERNATE_ERROR_INFO;
else if (!strcmp(error_strategy, "ignore"))
sas.error_mode = SUBMODULE_ALTERNATE_ERROR_IGNORE;
else
die(_("Value '%s' for submodule.alternateErrorStrategy is not recognized"), error_strategy);
if (!strcmp(sm_alternate, "superproject"))
foreach_alt_odb(add_possible_reference_from_superproject, &sas);
else if (!strcmp(sm_alternate, "no"))
; /* do nothing */
else
die(_("Value '%s' for submodule.alternateLocation is not recognized"), sm_alternate);
free(sm_alternate);
free(error_strategy);
}
static int module_clone(int argc, const char **argv, const char *prefix)
{
const char *name = NULL, *url = NULL, *depth = NULL;
int quiet = 0;
clone: pass --progress decision to recursive submodules When cloning with "--recursive", we'd generally expect submodules to show progress reports if the main clone did, too. In older versions of git, this mostly worked out of the box. Since we show progress by default when stderr is a tty, and since the child clones inherit the parent stderr, then both processes would come to the same decision by default. If the parent clone was asked for "--quiet", we passed down "--quiet" to the child. However, if stderr was not a tty and the user specified "--progress", we did not propagate this to the child. That's a minor bug, but things got much worse when we switched recently to submodule--helper's update_clone command. With that change, the stderr of the child clones are always connected to a pipe, and we never output progress at all. This patch teaches git-submodule and git-submodule--helper how to pass down an explicit "--progress" flag when cloning. The clone command then decides to propagate that flag based on the cloning decision made earlier (which takes into account isatty(2) of the parent process, existing --progress or --quiet flags, etc). Since the child processes always run without a tty on stderr, we don't have to worry about passing an explicit "--no-progress"; it's the default for them. This fixes the recent loss of progress during recursive clones. And as a bonus, it makes: git clone --recursive --progress ... 2>&1 | cat work by triggering progress explicitly in the children. Signed-off-by: Jeff King <peff@peff.net> Acked-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-22 07:24:46 +02:00
int progress = 0;
char *p, *path = NULL, *sm_gitdir;
struct strbuf sb = STRBUF_INIT;
struct string_list reference = STRING_LIST_INIT_NODUP;
int dissociate = 0, require_init = 0;
char *sm_alternate = NULL, *error_strategy = NULL;
int single_branch = -1;
struct option module_clone_options[] = {
OPT_STRING(0, "prefix", &prefix,
N_("path"),
N_("alternative anchor for relative paths")),
OPT_STRING(0, "path", &path,
N_("path"),
N_("where the new submodule will be cloned to")),
OPT_STRING(0, "name", &name,
N_("string"),
N_("name of the new submodule")),
OPT_STRING(0, "url", &url,
N_("string"),
N_("url where to clone the submodule from")),
OPT_STRING_LIST(0, "reference", &reference,
N_("repo"),
N_("reference repository")),
OPT_BOOL(0, "dissociate", &dissociate,
N_("use --reference only while cloning")),
OPT_STRING(0, "depth", &depth,
N_("string"),
N_("depth for shallow clones")),
OPT__QUIET(&quiet, "Suppress output for cloning a submodule"),
clone: pass --progress decision to recursive submodules When cloning with "--recursive", we'd generally expect submodules to show progress reports if the main clone did, too. In older versions of git, this mostly worked out of the box. Since we show progress by default when stderr is a tty, and since the child clones inherit the parent stderr, then both processes would come to the same decision by default. If the parent clone was asked for "--quiet", we passed down "--quiet" to the child. However, if stderr was not a tty and the user specified "--progress", we did not propagate this to the child. That's a minor bug, but things got much worse when we switched recently to submodule--helper's update_clone command. With that change, the stderr of the child clones are always connected to a pipe, and we never output progress at all. This patch teaches git-submodule and git-submodule--helper how to pass down an explicit "--progress" flag when cloning. The clone command then decides to propagate that flag based on the cloning decision made earlier (which takes into account isatty(2) of the parent process, existing --progress or --quiet flags, etc). Since the child processes always run without a tty on stderr, we don't have to worry about passing an explicit "--no-progress"; it's the default for them. This fixes the recent loss of progress during recursive clones. And as a bonus, it makes: git clone --recursive --progress ... 2>&1 | cat work by triggering progress explicitly in the children. Signed-off-by: Jeff King <peff@peff.net> Acked-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-22 07:24:46 +02:00
OPT_BOOL(0, "progress", &progress,
N_("force cloning progress")),
clone --recurse-submodules: prevent name squatting on Windows In addition to preventing `.git` from being tracked by Git, on Windows we also have to prevent `git~1` from being tracked, as the default NTFS short name (also known as the "8.3 filename") for the file name `.git` is `git~1`, otherwise it would be possible for malicious repositories to write directly into the `.git/` directory, e.g. a `post-checkout` hook that would then be executed _during_ a recursive clone. When we implemented appropriate protections in 2b4c6efc821 (read-cache: optionally disallow NTFS .git variants, 2014-12-16), we had analyzed carefully that the `.git` directory or file would be guaranteed to be the first directory entry to be written. Otherwise it would be possible e.g. for a file named `..git` to be assigned the short name `git~1` and subsequently, the short name generated for `.git` would be `git~2`. Or `git~3`. Or even `~9999999` (for a detailed explanation of the lengths we have to go to protect `.gitmodules`, see the commit message of e7cb0b4455c (is_ntfs_dotgit: match other .git files, 2018-05-11)). However, by exploiting two issues (that will be addressed in a related patch series close by), it is currently possible to clone a submodule into a non-empty directory: - On Windows, file names cannot end in a space or a period (for historical reasons: the period separating the base name from the file extension was not actually written to disk, and the base name/file extension was space-padded to the full 8/3 characters, respectively). Helpfully, when creating a directory under the name, say, `sub.`, that trailing period is trimmed automatically and the actual name on disk is `sub`. This means that while Git thinks that the submodule names `sub` and `sub.` are different, they both access `.git/modules/sub/`. - While the backslash character is a valid file name character on Linux, it is not so on Windows. As Git tries to be cross-platform, it therefore allows backslash characters in the file names stored in tree objects. Which means that it is totally possible that a submodule `c` sits next to a file `c\..git`, and on Windows, during recursive clone a file called `..git` will be written into `c/`, of course _before_ the submodule is cloned. Note that the actual exploit is not quite as simple as having a submodule `c` next to a file `c\..git`, as we have to make sure that the directory `.git/modules/b` already exists when the submodule is checked out, otherwise a different code path is taken in `module_clone()` that does _not_ allow a non-empty submodule directory to exist already. Even if we will address both issues nearby (the next commit will disallow backslash characters in tree entries' file names on Windows, and another patch will disallow creating directories/files with trailing spaces or periods), it is a wise idea to defend in depth against this sort of attack vector: when submodules are cloned recursively, we now _require_ the directory to be empty, addressing CVE-2019-1349. Note: the code path we patch is shared with the code path of `git submodule update --init`, which must not expect, in general, that the directory is empty. Hence we have to introduce the new option `--force-init` and hand it all the way down from `git submodule` to the actual `git submodule--helper` process that performs the initial clone. Reported-by: Nicolas Joly <Nicolas.Joly@microsoft.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
2019-09-12 14:20:39 +02:00
OPT_BOOL(0, "require-init", &require_init,
N_("disallow cloning into non-empty directory")),
OPT_BOOL(0, "single-branch", &single_branch,
N_("clone only one branch, HEAD or --branch")),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
N_("git submodule--helper clone [--prefix=<path>] [--quiet] "
"[--reference <repository>] [--name <name>] [--depth <depth>] "
"[--single-branch] "
"--url <url> --path <path>"),
NULL
};
argc = parse_options(argc, argv, prefix, module_clone_options,
git_submodule_helper_usage, 0);
if (argc || !url || !path || !*path)
usage_with_options(git_submodule_helper_usage,
module_clone_options);
strbuf_addf(&sb, "%s/modules/%s", get_git_dir(), name);
sm_gitdir = absolute_pathdup(sb.buf);
strbuf_reset(&sb);
if (!is_absolute_path(path)) {
strbuf_addf(&sb, "%s/%s", get_git_work_tree(), path);
path = strbuf_detach(&sb, NULL);
} else
path = xstrdup(path);
if (validate_submodule_git_dir(sm_gitdir, name) < 0)
die(_("refusing to create/use '%s' in another submodule's "
"git dir"), sm_gitdir);
if (!file_exists(sm_gitdir)) {
if (safe_create_leading_directories_const(sm_gitdir) < 0)
die(_("could not create directory '%s'"), sm_gitdir);
prepare_possible_alternates(name, &reference);
if (clone_submodule(path, sm_gitdir, url, depth, &reference, dissociate,
quiet, progress, single_branch))
die(_("clone of '%s' into submodule path '%s' failed"),
url, path);
} else {
clone --recurse-submodules: prevent name squatting on Windows In addition to preventing `.git` from being tracked by Git, on Windows we also have to prevent `git~1` from being tracked, as the default NTFS short name (also known as the "8.3 filename") for the file name `.git` is `git~1`, otherwise it would be possible for malicious repositories to write directly into the `.git/` directory, e.g. a `post-checkout` hook that would then be executed _during_ a recursive clone. When we implemented appropriate protections in 2b4c6efc821 (read-cache: optionally disallow NTFS .git variants, 2014-12-16), we had analyzed carefully that the `.git` directory or file would be guaranteed to be the first directory entry to be written. Otherwise it would be possible e.g. for a file named `..git` to be assigned the short name `git~1` and subsequently, the short name generated for `.git` would be `git~2`. Or `git~3`. Or even `~9999999` (for a detailed explanation of the lengths we have to go to protect `.gitmodules`, see the commit message of e7cb0b4455c (is_ntfs_dotgit: match other .git files, 2018-05-11)). However, by exploiting two issues (that will be addressed in a related patch series close by), it is currently possible to clone a submodule into a non-empty directory: - On Windows, file names cannot end in a space or a period (for historical reasons: the period separating the base name from the file extension was not actually written to disk, and the base name/file extension was space-padded to the full 8/3 characters, respectively). Helpfully, when creating a directory under the name, say, `sub.`, that trailing period is trimmed automatically and the actual name on disk is `sub`. This means that while Git thinks that the submodule names `sub` and `sub.` are different, they both access `.git/modules/sub/`. - While the backslash character is a valid file name character on Linux, it is not so on Windows. As Git tries to be cross-platform, it therefore allows backslash characters in the file names stored in tree objects. Which means that it is totally possible that a submodule `c` sits next to a file `c\..git`, and on Windows, during recursive clone a file called `..git` will be written into `c/`, of course _before_ the submodule is cloned. Note that the actual exploit is not quite as simple as having a submodule `c` next to a file `c\..git`, as we have to make sure that the directory `.git/modules/b` already exists when the submodule is checked out, otherwise a different code path is taken in `module_clone()` that does _not_ allow a non-empty submodule directory to exist already. Even if we will address both issues nearby (the next commit will disallow backslash characters in tree entries' file names on Windows, and another patch will disallow creating directories/files with trailing spaces or periods), it is a wise idea to defend in depth against this sort of attack vector: when submodules are cloned recursively, we now _require_ the directory to be empty, addressing CVE-2019-1349. Note: the code path we patch is shared with the code path of `git submodule update --init`, which must not expect, in general, that the directory is empty. Hence we have to introduce the new option `--force-init` and hand it all the way down from `git submodule` to the actual `git submodule--helper` process that performs the initial clone. Reported-by: Nicolas Joly <Nicolas.Joly@microsoft.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
2019-09-12 14:20:39 +02:00
if (require_init && !access(path, X_OK) && !is_empty_dir(path))
die(_("directory not empty: '%s'"), path);
if (safe_create_leading_directories_const(path) < 0)
die(_("could not create directory '%s'"), path);
strbuf_addf(&sb, "%s/index", sm_gitdir);
unlink_or_warn(sb.buf);
strbuf_reset(&sb);
}
connect_work_tree_and_git_dir(path, sm_gitdir, 0);
p = git_pathdup_submodule(path, "config");
if (!p)
die(_("could not get submodule directory for '%s'"), path);
/* setup alternateLocation and alternateErrorStrategy in the cloned submodule if needed */
git_config_get_string("submodule.alternateLocation", &sm_alternate);
if (sm_alternate)
git_config_set_in_file(p, "submodule.alternateLocation",
sm_alternate);
git_config_get_string("submodule.alternateErrorStrategy", &error_strategy);
if (error_strategy)
git_config_set_in_file(p, "submodule.alternateErrorStrategy",
error_strategy);
free(sm_alternate);
free(error_strategy);
strbuf_release(&sb);
free(sm_gitdir);
free(path);
free(p);
return 0;
}
static void determine_submodule_update_strategy(struct repository *r,
int just_cloned,
const char *path,
const char *update,
struct submodule_update_strategy *out)
{
const struct submodule *sub = submodule_from_path(r, &null_oid, path);
char *key;
const char *val;
key = xstrfmt("submodule.%s.update", sub->name);
if (update) {
if (parse_submodule_update_strategy(update, out) < 0)
die(_("Invalid update mode '%s' for submodule path '%s'"),
update, path);
} else if (!repo_config_get_string_const(r, key, &val)) {
if (parse_submodule_update_strategy(val, out) < 0)
die(_("Invalid update mode '%s' configured for submodule path '%s'"),
val, path);
} else if (sub->update_strategy.type != SM_UPDATE_UNSPECIFIED) {
if (sub->update_strategy.type == SM_UPDATE_COMMAND)
BUG("how did we read update = !command from .gitmodules?");
out->type = sub->update_strategy.type;
out->command = sub->update_strategy.command;
} else
out->type = SM_UPDATE_CHECKOUT;
if (just_cloned &&
(out->type == SM_UPDATE_MERGE ||
out->type == SM_UPDATE_REBASE ||
out->type == SM_UPDATE_NONE))
out->type = SM_UPDATE_CHECKOUT;
free(key);
}
static int module_update_module_mode(int argc, const char **argv, const char *prefix)
{
const char *path, *update = NULL;
int just_cloned;
struct submodule_update_strategy update_strategy = { .type = SM_UPDATE_CHECKOUT };
if (argc < 3 || argc > 4)
die("submodule--helper update-module-clone expects <just-cloned> <path> [<update>]");
just_cloned = git_config_int("just_cloned", argv[1]);
path = argv[2];
if (argc == 4)
update = argv[3];
determine_submodule_update_strategy(the_repository,
just_cloned, path, update,
&update_strategy);
fputs(submodule_strategy_to_string(&update_strategy), stdout);
return 0;
}
struct update_clone_data {
const struct submodule *sub;
struct object_id oid;
unsigned just_cloned;
};
struct submodule_update_clone {
/* index into 'list', the list of submodules to look into for cloning */
int current;
struct module_list list;
unsigned warn_if_uninitialized : 1;
/* update parameter passed via commandline */
struct submodule_update_strategy update;
/* configuration parameters which are passed on to the children */
clone: pass --progress decision to recursive submodules When cloning with "--recursive", we'd generally expect submodules to show progress reports if the main clone did, too. In older versions of git, this mostly worked out of the box. Since we show progress by default when stderr is a tty, and since the child clones inherit the parent stderr, then both processes would come to the same decision by default. If the parent clone was asked for "--quiet", we passed down "--quiet" to the child. However, if stderr was not a tty and the user specified "--progress", we did not propagate this to the child. That's a minor bug, but things got much worse when we switched recently to submodule--helper's update_clone command. With that change, the stderr of the child clones are always connected to a pipe, and we never output progress at all. This patch teaches git-submodule and git-submodule--helper how to pass down an explicit "--progress" flag when cloning. The clone command then decides to propagate that flag based on the cloning decision made earlier (which takes into account isatty(2) of the parent process, existing --progress or --quiet flags, etc). Since the child processes always run without a tty on stderr, we don't have to worry about passing an explicit "--no-progress"; it's the default for them. This fixes the recent loss of progress during recursive clones. And as a bonus, it makes: git clone --recursive --progress ... 2>&1 | cat work by triggering progress explicitly in the children. Signed-off-by: Jeff King <peff@peff.net> Acked-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-22 07:24:46 +02:00
int progress;
int quiet;
int recommend_shallow;
struct string_list references;
int dissociate;
clone --recurse-submodules: prevent name squatting on Windows In addition to preventing `.git` from being tracked by Git, on Windows we also have to prevent `git~1` from being tracked, as the default NTFS short name (also known as the "8.3 filename") for the file name `.git` is `git~1`, otherwise it would be possible for malicious repositories to write directly into the `.git/` directory, e.g. a `post-checkout` hook that would then be executed _during_ a recursive clone. When we implemented appropriate protections in 2b4c6efc821 (read-cache: optionally disallow NTFS .git variants, 2014-12-16), we had analyzed carefully that the `.git` directory or file would be guaranteed to be the first directory entry to be written. Otherwise it would be possible e.g. for a file named `..git` to be assigned the short name `git~1` and subsequently, the short name generated for `.git` would be `git~2`. Or `git~3`. Or even `~9999999` (for a detailed explanation of the lengths we have to go to protect `.gitmodules`, see the commit message of e7cb0b4455c (is_ntfs_dotgit: match other .git files, 2018-05-11)). However, by exploiting two issues (that will be addressed in a related patch series close by), it is currently possible to clone a submodule into a non-empty directory: - On Windows, file names cannot end in a space or a period (for historical reasons: the period separating the base name from the file extension was not actually written to disk, and the base name/file extension was space-padded to the full 8/3 characters, respectively). Helpfully, when creating a directory under the name, say, `sub.`, that trailing period is trimmed automatically and the actual name on disk is `sub`. This means that while Git thinks that the submodule names `sub` and `sub.` are different, they both access `.git/modules/sub/`. - While the backslash character is a valid file name character on Linux, it is not so on Windows. As Git tries to be cross-platform, it therefore allows backslash characters in the file names stored in tree objects. Which means that it is totally possible that a submodule `c` sits next to a file `c\..git`, and on Windows, during recursive clone a file called `..git` will be written into `c/`, of course _before_ the submodule is cloned. Note that the actual exploit is not quite as simple as having a submodule `c` next to a file `c\..git`, as we have to make sure that the directory `.git/modules/b` already exists when the submodule is checked out, otherwise a different code path is taken in `module_clone()` that does _not_ allow a non-empty submodule directory to exist already. Even if we will address both issues nearby (the next commit will disallow backslash characters in tree entries' file names on Windows, and another patch will disallow creating directories/files with trailing spaces or periods), it is a wise idea to defend in depth against this sort of attack vector: when submodules are cloned recursively, we now _require_ the directory to be empty, addressing CVE-2019-1349. Note: the code path we patch is shared with the code path of `git submodule update --init`, which must not expect, in general, that the directory is empty. Hence we have to introduce the new option `--force-init` and hand it all the way down from `git submodule` to the actual `git submodule--helper` process that performs the initial clone. Reported-by: Nicolas Joly <Nicolas.Joly@microsoft.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
2019-09-12 14:20:39 +02:00
unsigned require_init;
const char *depth;
const char *recursive_prefix;
const char *prefix;
int single_branch;
/* to be consumed by git-submodule.sh */
struct update_clone_data *update_clone;
int update_clone_nr; int update_clone_alloc;
/* If we want to stop as fast as possible and return an error */
unsigned quickstop : 1;
/* failed clones to be retried again */
const struct cache_entry **failed_clones;
int failed_clones_nr, failed_clones_alloc;
int max_jobs;
};
#define SUBMODULE_UPDATE_CLONE_INIT { \
.list = MODULE_LIST_INIT, \
.update = SUBMODULE_UPDATE_STRATEGY_INIT, \
.recommend_shallow = -1, \
.references = STRING_LIST_INIT_DUP, \
.single_branch = -1, \
.max_jobs = 1, \
}
static void next_submodule_warn_missing(struct submodule_update_clone *suc,
struct strbuf *out, const char *displaypath)
{
/*
* Only mention uninitialized submodules when their
* paths have been specified.
*/
if (suc->warn_if_uninitialized) {
strbuf_addf(out,
_("Submodule path '%s' not initialized"),
displaypath);
strbuf_addch(out, '\n');
strbuf_addstr(out,
_("Maybe you want to use 'update --init'?"));
strbuf_addch(out, '\n');
}
}
/**
* Determine whether 'ce' needs to be cloned. If so, prepare the 'child' to
* run the clone. Returns 1 if 'ce' needs to be cloned, 0 otherwise.
*/
static int prepare_to_clone_next_submodule(const struct cache_entry *ce,
struct child_process *child,
struct submodule_update_clone *suc,
struct strbuf *out)
{
const struct submodule *sub = NULL;
const char *url = NULL;
const char *update_string;
enum submodule_update_type update_type;
char *key;
struct strbuf displaypath_sb = STRBUF_INIT;
struct strbuf sb = STRBUF_INIT;
const char *displaypath = NULL;
int needs_cloning = 0;
submodule helper: convert relative URL to absolute URL if needed The submodule helper update_clone called by "git submodule update", clones submodules if needed. As submodules used to have the URL indicating if they were active, the step to resolve relative URLs was done in the "submodule init" step. Nowadays submodules can be configured active without calling an explicit init, e.g. via configuring submodule.active. When trying to obtain submodules that are set active this way, we'll fallback to the URL found in the .gitmodules, which may be relative to the superproject, but we do not resolve it, yet: git clone https://gerrit.googlesource.com/gerrit cd gerrit && grep url .gitmodules url = ../plugins/codemirror-editor ... git config submodule.active . git submodule update fatal: repository '../plugins/codemirror-editor' does not exist fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor'. Retry scheduled [...] fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor' a second time, aborting [...] To resolve the issue, factor out the function that resolves the relative URLs in "git submodule init" (in the submodule helper in the init_submodule function) and call it at the appropriate place in the update_clone helper. Reported-by: Jaewoong Jung <jungjw@google.com> Signed-off-by: Stefan Beller <sbeller@google.com> Reviewed-by: Jonathan Nieder <jrnieder@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-10-16 19:27:03 +02:00
int need_free_url = 0;
if (ce_stage(ce)) {
if (suc->recursive_prefix)
strbuf_addf(&sb, "%s/%s", suc->recursive_prefix, ce->name);
else
strbuf_addstr(&sb, ce->name);
strbuf_addf(out, _("Skipping unmerged submodule %s"), sb.buf);
strbuf_addch(out, '\n');
goto cleanup;
}
sub = submodule_from_path(the_repository, &null_oid, ce->name);
if (suc->recursive_prefix)
displaypath = relative_path(suc->recursive_prefix,
ce->name, &displaypath_sb);
else
displaypath = ce->name;
if (!sub) {
next_submodule_warn_missing(suc, out, displaypath);
goto cleanup;
}
key = xstrfmt("submodule.%s.update", sub->name);
if (!repo_config_get_string_const(the_repository, key, &update_string)) {
update_type = parse_submodule_update_type(update_string);
} else {
update_type = sub->update_strategy.type;
}
free(key);
if (suc->update.type == SM_UPDATE_NONE
|| (suc->update.type == SM_UPDATE_UNSPECIFIED
&& update_type == SM_UPDATE_NONE)) {
strbuf_addf(out, _("Skipping submodule '%s'"), displaypath);
strbuf_addch(out, '\n');
goto cleanup;
}
/* Check if the submodule has been initialized. */
if (!is_submodule_active(the_repository, ce->name)) {
next_submodule_warn_missing(suc, out, displaypath);
goto cleanup;
}
strbuf_reset(&sb);
strbuf_addf(&sb, "submodule.%s.url", sub->name);
submodule helper: convert relative URL to absolute URL if needed The submodule helper update_clone called by "git submodule update", clones submodules if needed. As submodules used to have the URL indicating if they were active, the step to resolve relative URLs was done in the "submodule init" step. Nowadays submodules can be configured active without calling an explicit init, e.g. via configuring submodule.active. When trying to obtain submodules that are set active this way, we'll fallback to the URL found in the .gitmodules, which may be relative to the superproject, but we do not resolve it, yet: git clone https://gerrit.googlesource.com/gerrit cd gerrit && grep url .gitmodules url = ../plugins/codemirror-editor ... git config submodule.active . git submodule update fatal: repository '../plugins/codemirror-editor' does not exist fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor'. Retry scheduled [...] fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor' a second time, aborting [...] To resolve the issue, factor out the function that resolves the relative URLs in "git submodule init" (in the submodule helper in the init_submodule function) and call it at the appropriate place in the update_clone helper. Reported-by: Jaewoong Jung <jungjw@google.com> Signed-off-by: Stefan Beller <sbeller@google.com> Reviewed-by: Jonathan Nieder <jrnieder@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-10-16 19:27:03 +02:00
if (repo_config_get_string_const(the_repository, sb.buf, &url)) {
if (starts_with_dot_slash(sub->url) ||
starts_with_dot_dot_slash(sub->url)) {
url = compute_submodule_clone_url(sub->url);
need_free_url = 1;
} else
url = sub->url;
}
strbuf_reset(&sb);
strbuf_addf(&sb, "%s/.git", ce->name);
needs_cloning = !file_exists(sb.buf);
ALLOC_GROW(suc->update_clone, suc->update_clone_nr + 1,
suc->update_clone_alloc);
oidcpy(&suc->update_clone[suc->update_clone_nr].oid, &ce->oid);
suc->update_clone[suc->update_clone_nr].just_cloned = needs_cloning;
suc->update_clone[suc->update_clone_nr].sub = sub;
suc->update_clone_nr++;
if (!needs_cloning)
goto cleanup;
child->git_cmd = 1;
child->no_stdin = 1;
child->stdout_to_stderr = 1;
child->err = -1;
strvec_push(&child->args, "submodule--helper");
strvec_push(&child->args, "clone");
clone: pass --progress decision to recursive submodules When cloning with "--recursive", we'd generally expect submodules to show progress reports if the main clone did, too. In older versions of git, this mostly worked out of the box. Since we show progress by default when stderr is a tty, and since the child clones inherit the parent stderr, then both processes would come to the same decision by default. If the parent clone was asked for "--quiet", we passed down "--quiet" to the child. However, if stderr was not a tty and the user specified "--progress", we did not propagate this to the child. That's a minor bug, but things got much worse when we switched recently to submodule--helper's update_clone command. With that change, the stderr of the child clones are always connected to a pipe, and we never output progress at all. This patch teaches git-submodule and git-submodule--helper how to pass down an explicit "--progress" flag when cloning. The clone command then decides to propagate that flag based on the cloning decision made earlier (which takes into account isatty(2) of the parent process, existing --progress or --quiet flags, etc). Since the child processes always run without a tty on stderr, we don't have to worry about passing an explicit "--no-progress"; it's the default for them. This fixes the recent loss of progress during recursive clones. And as a bonus, it makes: git clone --recursive --progress ... 2>&1 | cat work by triggering progress explicitly in the children. Signed-off-by: Jeff King <peff@peff.net> Acked-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-22 07:24:46 +02:00
if (suc->progress)
strvec_push(&child->args, "--progress");
if (suc->quiet)
strvec_push(&child->args, "--quiet");
if (suc->prefix)
strvec_pushl(&child->args, "--prefix", suc->prefix, NULL);
if (suc->recommend_shallow && sub->recommend_shallow == 1)
strvec_push(&child->args, "--depth=1");
clone --recurse-submodules: prevent name squatting on Windows In addition to preventing `.git` from being tracked by Git, on Windows we also have to prevent `git~1` from being tracked, as the default NTFS short name (also known as the "8.3 filename") for the file name `.git` is `git~1`, otherwise it would be possible for malicious repositories to write directly into the `.git/` directory, e.g. a `post-checkout` hook that would then be executed _during_ a recursive clone. When we implemented appropriate protections in 2b4c6efc821 (read-cache: optionally disallow NTFS .git variants, 2014-12-16), we had analyzed carefully that the `.git` directory or file would be guaranteed to be the first directory entry to be written. Otherwise it would be possible e.g. for a file named `..git` to be assigned the short name `git~1` and subsequently, the short name generated for `.git` would be `git~2`. Or `git~3`. Or even `~9999999` (for a detailed explanation of the lengths we have to go to protect `.gitmodules`, see the commit message of e7cb0b4455c (is_ntfs_dotgit: match other .git files, 2018-05-11)). However, by exploiting two issues (that will be addressed in a related patch series close by), it is currently possible to clone a submodule into a non-empty directory: - On Windows, file names cannot end in a space or a period (for historical reasons: the period separating the base name from the file extension was not actually written to disk, and the base name/file extension was space-padded to the full 8/3 characters, respectively). Helpfully, when creating a directory under the name, say, `sub.`, that trailing period is trimmed automatically and the actual name on disk is `sub`. This means that while Git thinks that the submodule names `sub` and `sub.` are different, they both access `.git/modules/sub/`. - While the backslash character is a valid file name character on Linux, it is not so on Windows. As Git tries to be cross-platform, it therefore allows backslash characters in the file names stored in tree objects. Which means that it is totally possible that a submodule `c` sits next to a file `c\..git`, and on Windows, during recursive clone a file called `..git` will be written into `c/`, of course _before_ the submodule is cloned. Note that the actual exploit is not quite as simple as having a submodule `c` next to a file `c\..git`, as we have to make sure that the directory `.git/modules/b` already exists when the submodule is checked out, otherwise a different code path is taken in `module_clone()` that does _not_ allow a non-empty submodule directory to exist already. Even if we will address both issues nearby (the next commit will disallow backslash characters in tree entries' file names on Windows, and another patch will disallow creating directories/files with trailing spaces or periods), it is a wise idea to defend in depth against this sort of attack vector: when submodules are cloned recursively, we now _require_ the directory to be empty, addressing CVE-2019-1349. Note: the code path we patch is shared with the code path of `git submodule update --init`, which must not expect, in general, that the directory is empty. Hence we have to introduce the new option `--force-init` and hand it all the way down from `git submodule` to the actual `git submodule--helper` process that performs the initial clone. Reported-by: Nicolas Joly <Nicolas.Joly@microsoft.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
2019-09-12 14:20:39 +02:00
if (suc->require_init)
strvec_push(&child->args, "--require-init");
strvec_pushl(&child->args, "--path", sub->path, NULL);
strvec_pushl(&child->args, "--name", sub->name, NULL);
strvec_pushl(&child->args, "--url", url, NULL);
if (suc->references.nr) {
struct string_list_item *item;
for_each_string_list_item(item, &suc->references)
strvec_pushl(&child->args, "--reference", item->string, NULL);
}
if (suc->dissociate)
strvec_push(&child->args, "--dissociate");
if (suc->depth)
strvec_push(&child->args, suc->depth);
if (suc->single_branch >= 0)
strvec_push(&child->args, suc->single_branch ?
"--single-branch" :
"--no-single-branch");
cleanup:
strbuf_reset(&displaypath_sb);
strbuf_reset(&sb);
submodule helper: convert relative URL to absolute URL if needed The submodule helper update_clone called by "git submodule update", clones submodules if needed. As submodules used to have the URL indicating if they were active, the step to resolve relative URLs was done in the "submodule init" step. Nowadays submodules can be configured active without calling an explicit init, e.g. via configuring submodule.active. When trying to obtain submodules that are set active this way, we'll fallback to the URL found in the .gitmodules, which may be relative to the superproject, but we do not resolve it, yet: git clone https://gerrit.googlesource.com/gerrit cd gerrit && grep url .gitmodules url = ../plugins/codemirror-editor ... git config submodule.active . git submodule update fatal: repository '../plugins/codemirror-editor' does not exist fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor'. Retry scheduled [...] fatal: clone of '../plugins/codemirror-editor' into submodule path '/tmp/gerrit/plugins/codemirror-editor' failed Failed to clone 'plugins/codemirror-editor' a second time, aborting [...] To resolve the issue, factor out the function that resolves the relative URLs in "git submodule init" (in the submodule helper in the init_submodule function) and call it at the appropriate place in the update_clone helper. Reported-by: Jaewoong Jung <jungjw@google.com> Signed-off-by: Stefan Beller <sbeller@google.com> Reviewed-by: Jonathan Nieder <jrnieder@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-10-16 19:27:03 +02:00
if (need_free_url)
free((void*)url);
return needs_cloning;
}
static int update_clone_get_next_task(struct child_process *child,
struct strbuf *err,
void *suc_cb,
void **idx_task_cb)
{
struct submodule_update_clone *suc = suc_cb;
const struct cache_entry *ce;
int index;
for (; suc->current < suc->list.nr; suc->current++) {
ce = suc->list.entries[suc->current];
if (prepare_to_clone_next_submodule(ce, child, suc, err)) {
int *p = xmalloc(sizeof(*p));
*p = suc->current;
*idx_task_cb = p;
suc->current++;
return 1;
}
}
/*
* The loop above tried cloning each submodule once, now try the
* stragglers again, which we can imagine as an extension of the
* entry list.
*/
index = suc->current - suc->list.nr;
if (index < suc->failed_clones_nr) {
int *p;
ce = suc->failed_clones[index];
2016-08-09 23:29:13 +02:00
if (!prepare_to_clone_next_submodule(ce, child, suc, err)) {
suc->current ++;
strbuf_addstr(err, "BUG: submodule considered for "
"cloning, doesn't need cloning "
"any more?\n");
2016-08-09 23:29:13 +02:00
return 0;
}
p = xmalloc(sizeof(*p));
*p = suc->current;
*idx_task_cb = p;
suc->current ++;
return 1;
}
return 0;
}
static int update_clone_start_failure(struct strbuf *err,
void *suc_cb,
void *idx_task_cb)
{
struct submodule_update_clone *suc = suc_cb;
suc->quickstop = 1;
return 1;
}
static int update_clone_task_finished(int result,
struct strbuf *err,
void *suc_cb,
void *idx_task_cb)
{
const struct cache_entry *ce;
struct submodule_update_clone *suc = suc_cb;
int *idxP = idx_task_cb;
int idx = *idxP;
free(idxP);
if (!result)
return 0;
if (idx < suc->list.nr) {
ce = suc->list.entries[idx];
strbuf_addf(err, _("Failed to clone '%s'. Retry scheduled"),
ce->name);
strbuf_addch(err, '\n');
ALLOC_GROW(suc->failed_clones,
suc->failed_clones_nr + 1,
suc->failed_clones_alloc);
suc->failed_clones[suc->failed_clones_nr++] = ce;
return 0;
} else {
idx -= suc->list.nr;
ce = suc->failed_clones[idx];
strbuf_addf(err, _("Failed to clone '%s' a second time, aborting"),
ce->name);
strbuf_addch(err, '\n');
suc->quickstop = 1;
return 1;
}
return 0;
}
static int git_update_clone_config(const char *var, const char *value,
void *cb)
{
int *max_jobs = cb;
if (!strcmp(var, "submodule.fetchjobs"))
*max_jobs = parse_submodule_fetchjobs(var, value);
return 0;
}
static void update_submodule(struct update_clone_data *ucd)
{
fprintf(stdout, "dummy %s %d\t%s\n",
oid_to_hex(&ucd->oid),
ucd->just_cloned,
ucd->sub->path);
}
static int update_submodules(struct submodule_update_clone *suc)
{
int i;
run_processes_parallel_tr2(suc->max_jobs, update_clone_get_next_task,
update_clone_start_failure,
update_clone_task_finished, suc, "submodule",
"parallel/update");
/*
* We saved the output and put it out all at once now.
* That means:
* - the listener does not have to interleave their (checkout)
* work with our fetching. The writes involved in a
* checkout involve more straightforward sequential I/O.
* - the listener can avoid doing any work if fetching failed.
*/
if (suc->quickstop)
return 1;
for (i = 0; i < suc->update_clone_nr; i++)
update_submodule(&suc->update_clone[i]);
return 0;
}
static int update_clone(int argc, const char **argv, const char *prefix)
{
const char *update = NULL;
struct pathspec pathspec;
struct submodule_update_clone suc = SUBMODULE_UPDATE_CLONE_INIT;
struct option module_update_clone_options[] = {
OPT_STRING(0, "prefix", &prefix,
N_("path"),
N_("path into the working tree")),
OPT_STRING(0, "recursive-prefix", &suc.recursive_prefix,
N_("path"),
N_("path into the working tree, across nested "
"submodule boundaries")),
OPT_STRING(0, "update", &update,
N_("string"),
N_("rebase, merge, checkout or none")),
OPT_STRING_LIST(0, "reference", &suc.references, N_("repo"),
N_("reference repository")),
OPT_BOOL(0, "dissociate", &suc.dissociate,
N_("use --reference only while cloning")),
OPT_STRING(0, "depth", &suc.depth, "<depth>",
N_("Create a shallow clone truncated to the "
"specified number of revisions")),
OPT_INTEGER('j', "jobs", &suc.max_jobs,
N_("parallel jobs")),
OPT_BOOL(0, "recommend-shallow", &suc.recommend_shallow,
N_("whether the initial clone should follow the shallow recommendation")),
OPT__QUIET(&suc.quiet, N_("don't print cloning progress")),
clone: pass --progress decision to recursive submodules When cloning with "--recursive", we'd generally expect submodules to show progress reports if the main clone did, too. In older versions of git, this mostly worked out of the box. Since we show progress by default when stderr is a tty, and since the child clones inherit the parent stderr, then both processes would come to the same decision by default. If the parent clone was asked for "--quiet", we passed down "--quiet" to the child. However, if stderr was not a tty and the user specified "--progress", we did not propagate this to the child. That's a minor bug, but things got much worse when we switched recently to submodule--helper's update_clone command. With that change, the stderr of the child clones are always connected to a pipe, and we never output progress at all. This patch teaches git-submodule and git-submodule--helper how to pass down an explicit "--progress" flag when cloning. The clone command then decides to propagate that flag based on the cloning decision made earlier (which takes into account isatty(2) of the parent process, existing --progress or --quiet flags, etc). Since the child processes always run without a tty on stderr, we don't have to worry about passing an explicit "--no-progress"; it's the default for them. This fixes the recent loss of progress during recursive clones. And as a bonus, it makes: git clone --recursive --progress ... 2>&1 | cat work by triggering progress explicitly in the children. Signed-off-by: Jeff King <peff@peff.net> Acked-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-09-22 07:24:46 +02:00
OPT_BOOL(0, "progress", &suc.progress,
N_("force cloning progress")),
clone --recurse-submodules: prevent name squatting on Windows In addition to preventing `.git` from being tracked by Git, on Windows we also have to prevent `git~1` from being tracked, as the default NTFS short name (also known as the "8.3 filename") for the file name `.git` is `git~1`, otherwise it would be possible for malicious repositories to write directly into the `.git/` directory, e.g. a `post-checkout` hook that would then be executed _during_ a recursive clone. When we implemented appropriate protections in 2b4c6efc821 (read-cache: optionally disallow NTFS .git variants, 2014-12-16), we had analyzed carefully that the `.git` directory or file would be guaranteed to be the first directory entry to be written. Otherwise it would be possible e.g. for a file named `..git` to be assigned the short name `git~1` and subsequently, the short name generated for `.git` would be `git~2`. Or `git~3`. Or even `~9999999` (for a detailed explanation of the lengths we have to go to protect `.gitmodules`, see the commit message of e7cb0b4455c (is_ntfs_dotgit: match other .git files, 2018-05-11)). However, by exploiting two issues (that will be addressed in a related patch series close by), it is currently possible to clone a submodule into a non-empty directory: - On Windows, file names cannot end in a space or a period (for historical reasons: the period separating the base name from the file extension was not actually written to disk, and the base name/file extension was space-padded to the full 8/3 characters, respectively). Helpfully, when creating a directory under the name, say, `sub.`, that trailing period is trimmed automatically and the actual name on disk is `sub`. This means that while Git thinks that the submodule names `sub` and `sub.` are different, they both access `.git/modules/sub/`. - While the backslash character is a valid file name character on Linux, it is not so on Windows. As Git tries to be cross-platform, it therefore allows backslash characters in the file names stored in tree objects. Which means that it is totally possible that a submodule `c` sits next to a file `c\..git`, and on Windows, during recursive clone a file called `..git` will be written into `c/`, of course _before_ the submodule is cloned. Note that the actual exploit is not quite as simple as having a submodule `c` next to a file `c\..git`, as we have to make sure that the directory `.git/modules/b` already exists when the submodule is checked out, otherwise a different code path is taken in `module_clone()` that does _not_ allow a non-empty submodule directory to exist already. Even if we will address both issues nearby (the next commit will disallow backslash characters in tree entries' file names on Windows, and another patch will disallow creating directories/files with trailing spaces or periods), it is a wise idea to defend in depth against this sort of attack vector: when submodules are cloned recursively, we now _require_ the directory to be empty, addressing CVE-2019-1349. Note: the code path we patch is shared with the code path of `git submodule update --init`, which must not expect, in general, that the directory is empty. Hence we have to introduce the new option `--force-init` and hand it all the way down from `git submodule` to the actual `git submodule--helper` process that performs the initial clone. Reported-by: Nicolas Joly <Nicolas.Joly@microsoft.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
2019-09-12 14:20:39 +02:00
OPT_BOOL(0, "require-init", &suc.require_init,
N_("disallow cloning into non-empty directory")),
OPT_BOOL(0, "single-branch", &suc.single_branch,
N_("clone only one branch, HEAD or --branch")),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
N_("git submodule--helper update-clone [--prefix=<path>] [<path>...]"),
NULL
};
suc.prefix = prefix;
update_clone_config_from_gitmodules(&suc.max_jobs);
git_config(git_update_clone_config, &suc.max_jobs);
argc = parse_options(argc, argv, prefix, module_update_clone_options,
git_submodule_helper_usage, 0);
if (update)
if (parse_submodule_update_strategy(update, &suc.update) < 0)
die(_("bad value for update parameter"));
if (module_list_compute(argc, argv, prefix, &pathspec, &suc.list) < 0)
return 1;
if (pathspec.nr)
suc.warn_if_uninitialized = 1;
return update_submodules(&suc);
}
static int resolve_relative_path(int argc, const char **argv, const char *prefix)
{
struct strbuf sb = STRBUF_INIT;
if (argc != 3)
die("submodule--helper relative-path takes exactly 2 arguments, got %d", argc);
printf("%s", relative_path(argv[1], argv[2], &sb));
strbuf_release(&sb);
return 0;
}
static const char *remote_submodule_branch(const char *path)
{
const struct submodule *sub;
const char *branch = NULL;
char *key;
sub = submodule_from_path(the_repository, &null_oid, path);
if (!sub)
return NULL;
key = xstrfmt("submodule.%s.branch", sub->name);
if (repo_config_get_string_const(the_repository, key, &branch))
branch = sub->branch;
free(key);
if (!branch)
return "HEAD";
if (!strcmp(branch, ".")) {
const char *refname = resolve_ref_unsafe("HEAD", 0, NULL, NULL);
if (!refname)
die(_("No such ref: %s"), "HEAD");
/* detached HEAD */
if (!strcmp(refname, "HEAD"))
die(_("Submodule (%s) branch configured to inherit "
"branch from superproject, but the superproject "
"is not on any branch"), sub->name);
if (!skip_prefix(refname, "refs/heads/", &refname))
die(_("Expecting a full ref name, got %s"), refname);
return refname;
}
return branch;
}
static int resolve_remote_submodule_branch(int argc, const char **argv,
const char *prefix)
{
const char *ret;
struct strbuf sb = STRBUF_INIT;
if (argc != 2)
die("submodule--helper remote-branch takes exactly one arguments, got %d", argc);
ret = remote_submodule_branch(argv[1]);
if (!ret)
die("submodule %s doesn't exist", argv[1]);
printf("%s", ret);
strbuf_release(&sb);
return 0;
}
static int push_check(int argc, const char **argv, const char *prefix)
{
struct remote *remote;
const char *superproject_head;
char *head;
int detached_head = 0;
struct object_id head_oid;
if (argc < 3)
die("submodule--helper push-check requires at least 2 arguments");
/*
* superproject's resolved head ref.
* if HEAD then the superproject is in a detached head state, otherwise
* it will be the resolved head ref.
*/
superproject_head = argv[1];
argv++;
argc--;
/* Get the submodule's head ref and determine if it is detached */
head = resolve_refdup("HEAD", 0, &head_oid, NULL);
if (!head)
die(_("Failed to resolve HEAD as a valid ref."));
if (!strcmp(head, "HEAD"))
detached_head = 1;
/*
* The remote must be configured.
* This is to avoid pushing to the exact same URL as the parent.
*/
remote = pushremote_get(argv[1]);
if (!remote || remote->origin == REMOTE_UNCONFIGURED)
die("remote '%s' not configured", argv[1]);
/* Check the refspec */
if (argc > 2) {
int i;
struct ref *local_refs = get_local_heads();
struct refspec refspec = REFSPEC_INIT_PUSH;
refspec_appendn(&refspec, argv + 2, argc - 2);
for (i = 0; i < refspec.nr; i++) {
const struct refspec_item *rs = &refspec.items[i];
if (rs->pattern || rs->matching)
continue;
/* LHS must match a single ref */
switch (count_refspec_match(rs->src, local_refs, NULL)) {
case 1:
break;
case 0:
/*
* If LHS matches 'HEAD' then we need to ensure
* that it matches the same named branch
* checked out in the superproject.
*/
if (!strcmp(rs->src, "HEAD")) {
if (!detached_head &&
!strcmp(head, superproject_head))
break;
die("HEAD does not match the named branch in the superproject");
}
consistently use "fallthrough" comments in switches Gcc 7 adds -Wimplicit-fallthrough, which can warn when a switch case falls through to the next case. The general idea is that the compiler can't tell if this was intentional or not, so you should annotate any intentional fall-throughs as such, leaving it to complain about any unannotated ones. There's a GNU __attribute__ which can be used for annotation, but of course we'd have to #ifdef it away on non-gcc compilers. Gcc will also recognize specially-formatted comments, which matches our current practice. Let's extend that practice to all of the unannotated sites (which I did look over and verify that they were behaving as intended). Ideally in each case we'd actually give some reasons in the comment about why we're falling through, or what we're falling through to. And gcc does support that with -Wimplicit-fallthrough=2, which relaxes the comment pattern matching to anything that contains "fallthrough" (or a variety of spelling variants). However, this isn't the default for -Wimplicit-fallthrough, nor for -Wextra. In the name of simplicity, it's probably better for us to support the default level, which requires "fallthrough" to be the only thing in the comment (modulo some window dressing like "else" and some punctuation; see the gcc manual for the complete set of patterns). This patch suppresses all warnings due to -Wimplicit-fallthrough. We might eventually want to add that to the DEVELOPER Makefile knob, but we should probably wait until gcc 7 is more widely adopted (since earlier versions will complain about the unknown warning type). Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-09-21 08:25:41 +02:00
/* fallthrough */
default:
die("src refspec '%s' must name a ref",
rs->src);
}
}
refspec_clear(&refspec);
}
free(head);
return 0;
}
static int ensure_core_worktree(int argc, const char **argv, const char *prefix)
{
const struct submodule *sub;
const char *path;
char *cw;
struct repository subrepo;
if (argc != 2)
BUG("submodule--helper ensure-core-worktree <path>");
path = argv[1];
sub = submodule_from_path(the_repository, &null_oid, path);
if (!sub)
BUG("We could get the submodule handle before?");
if (repo_submodule_init(&subrepo, the_repository, sub))
die(_("could not get a repository handle for submodule '%s'"), path);
if (!repo_config_get_string(&subrepo, "core.worktree", &cw)) {
char *cfg_file, *abs_path;
const char *rel_path;
struct strbuf sb = STRBUF_INIT;
cfg_file = repo_git_path(&subrepo, "config");
abs_path = absolute_pathdup(path);
rel_path = relative_path(abs_path, subrepo.gitdir, &sb);
git_config_set_in_file(cfg_file, "core.worktree", rel_path);
free(cfg_file);
free(abs_path);
strbuf_release(&sb);
}
return 0;
}
static int absorb_git_dirs(int argc, const char **argv, const char *prefix)
{
int i;
struct pathspec pathspec;
struct module_list list = MODULE_LIST_INIT;
unsigned flags = ABSORB_GITDIR_RECURSE_SUBMODULES;
struct option embed_gitdir_options[] = {
OPT_STRING(0, "prefix", &prefix,
N_("path"),
N_("path into the working tree")),
OPT_BIT(0, "--recursive", &flags, N_("recurse into submodules"),
ABSORB_GITDIR_RECURSE_SUBMODULES),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
submodule foreach: fix "<command> --quiet" not being respected Robin reported that git submodule foreach --quiet git pull --quiet origin is not really quiet anymore [1]. "git pull" behaves as if --quiet is not given. This happens because parseopt in submodule--helper will try to parse both --quiet options as if they are foreach's options, not git-pull's. The parsed options are removed from the command line. So when we do pull later, we execute just this git pull origin When calling submodule helper, adding "--" in front of "git pull" will stop parseopt for parsing options that do not really belong to submodule--helper foreach. PARSE_OPT_KEEP_UNKNOWN is removed as a safety measure. parseopt should never see unknown options or something has gone wrong. There are also a couple usage string update while I'm looking at them. While at it, I also add "--" to other subcommands that pass "$@" to submodule--helper. "$@" in these cases are paths and less likely to be --something-like-this. But the point still stands, git-submodule has parsed and classified what are options, what are paths. submodule--helper should never consider paths passed by git-submodule to be options even if they look like one. The test case is also contributed by Robin. [1] it should be quiet before fc1b9243cd (submodule: port submodule subcommand 'foreach' from shell to C, 2018-05-10) because parseopt can't accidentally eat options then. Reported-by: Robin H. Johnson <robbat2@gentoo.org> Tested-by: Robin H. Johnson <robbat2@gentoo.org> Signed-off-by: Robin H. Johnson <robbat2@gentoo.org> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-12 12:08:19 +02:00
N_("git submodule--helper absorb-git-dirs [<options>] [<path>...]"),
NULL
};
argc = parse_options(argc, argv, prefix, embed_gitdir_options,
git_submodule_helper_usage, 0);
if (module_list_compute(argc, argv, prefix, &pathspec, &list) < 0)
return 1;
for (i = 0; i < list.nr; i++)
absorb_git_dir_into_superproject(list.entries[i]->name, flags);
return 0;
}
static int is_active(int argc, const char **argv, const char *prefix)
{
if (argc != 2)
die("submodule--helper is-active takes exactly 1 argument");
return !is_submodule_active(the_repository, argv[1]);
}
submodule-config: verify submodule names as paths Submodule "names" come from the untrusted .gitmodules file, but we blindly append them to $GIT_DIR/modules to create our on-disk repo paths. This means you can do bad things by putting "../" into the name (among other things). Let's sanity-check these names to avoid building a path that can be exploited. There are two main decisions: 1. What should the allowed syntax be? It's tempting to reuse verify_path(), since submodule names typically come from in-repo paths. But there are two reasons not to: a. It's technically more strict than what we need, as we really care only about breaking out of the $GIT_DIR/modules/ hierarchy. E.g., having a submodule named "foo/.git" isn't actually dangerous, and it's possible that somebody has manually given such a funny name. b. Since we'll eventually use this checking logic in fsck to prevent downstream repositories, it should be consistent across platforms. Because verify_path() relies on is_dir_sep(), it wouldn't block "foo\..\bar" on a non-Windows machine. 2. Where should we enforce it? These days most of the .gitmodules reads go through submodule-config.c, so I've put it there in the reading step. That should cover all of the C code. We also construct the name for "git submodule add" inside the git-submodule.sh script. This is probably not a big deal for security since the name is coming from the user anyway, but it would be polite to remind them if the name they pick is invalid (and we need to expose the name-checker to the shell anyway for our test scripts). This patch issues a warning when reading .gitmodules and just ignores the related config entry completely. This will generally end up producing a sensible error, as it works the same as a .gitmodules file which is missing a submodule entry (so "submodule update" will barf, but "git clone --recurse-submodules" will print an error but not abort the clone. There is one minor oddity, which is that we print the warning once per malformed config key (since that's how the config subsystem gives us the entries). So in the new test, for example, the user would see three warnings. That's OK, since the intent is that this case should never come up outside of malicious repositories (and then it might even benefit the user to see the message multiple times). Credit for finding this vulnerability and the proof of concept from which the test script was adapted goes to Etienne Stalmans. Signed-off-by: Jeff King <peff@peff.net>
2018-04-30 09:25:25 +02:00
/*
* Exit non-zero if any of the submodule names given on the command line is
* invalid. If no names are given, filter stdin to print only valid names
* (which is primarily intended for testing).
*/
static int check_name(int argc, const char **argv, const char *prefix)
{
if (argc > 1) {
while (*++argv) {
if (check_submodule_name(*argv) < 0)
return 1;
}
} else {
struct strbuf buf = STRBUF_INIT;
while (strbuf_getline(&buf, stdin) != EOF) {
if (!check_submodule_name(buf.buf))
printf("%s\n", buf.buf);
}
strbuf_release(&buf);
}
return 0;
}
static int module_config(int argc, const char **argv, const char *prefix)
{
enum {
CHECK_WRITEABLE = 1,
DO_UNSET = 2
} command = 0;
struct option module_config_options[] = {
OPT_CMDMODE(0, "check-writeable", &command,
N_("check if it is safe to write to the .gitmodules file"),
CHECK_WRITEABLE),
OPT_CMDMODE(0, "unset", &command,
N_("unset the config in the .gitmodules file"),
DO_UNSET),
OPT_END()
};
const char *const git_submodule_helper_usage[] = {
N_("git submodule--helper config <name> [<value>]"),
N_("git submodule--helper config --unset <name>"),
N_("git submodule--helper config --check-writeable"),
NULL
};
argc = parse_options(argc, argv, prefix, module_config_options,
git_submodule_helper_usage, PARSE_OPT_KEEP_ARGV0);
if (argc == 1 && command == CHECK_WRITEABLE)
return is_writing_gitmodules_ok() ? 0 : -1;
/* Equivalent to ACTION_GET in builtin/config.c */
if (argc == 2 && command != DO_UNSET)
return print_config_from_gitmodules(the_repository, argv[1]);
/* Equivalent to ACTION_SET in builtin/config.c */
if (argc == 3 || (argc == 2 && command == DO_UNSET)) {
const char *value = (argc == 3) ? argv[2] : NULL;
submodule: support reading .gitmodules when it's not in the working tree When the .gitmodules file is not available in the working tree, try using the content from the index and from the current branch. This covers the case when the file is part of the repository but for some reason it is not checked out, for example because of a sparse checkout. This makes it possible to use at least the 'git submodule' commands which *read* the gitmodules configuration file without fully populating the working tree. Writing to .gitmodules will still require that the file is checked out, so check for that before calling config_set_in_gitmodules_file_gently. Add a similar check also in git-submodule.sh::cmd_add() to anticipate the eventual failure of the "git submodule add" command when .gitmodules is not safely writeable; this prevents the command from leaving the repository in a spurious state (e.g. the submodule repository was cloned but .gitmodules was not updated because config_set_in_gitmodules_file_gently failed). Moreover, since config_from_gitmodules() now accesses the global object store, it is necessary to protect all code paths which call the function against concurrent access to the global object store. Currently this only happens in builtin/grep.c::grep_submodules(), so call grep_read_lock() before invoking code involving config_from_gitmodules(). Finally, add t7418-submodule-sparse-gitmodules.sh to verify that reading from .gitmodules succeeds and that writing to it fails when the file is not checked out. NOTE: there is one rare case where this new feature does not work properly yet: nested submodules without .gitmodules in their working tree. This has been documented with a warning and a test_expect_failure item in t7814, and in this case the current behavior is not altered: no config is read. Signed-off-by: Antonio Ospite <ao2@ao2.it> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-10-25 18:18:12 +02:00
if (!is_writing_gitmodules_ok())
die(_("please make sure that the .gitmodules file is in the working tree"));
return config_set_in_gitmodules_file_gently(argv[1], value);
submodule: support reading .gitmodules when it's not in the working tree When the .gitmodules file is not available in the working tree, try using the content from the index and from the current branch. This covers the case when the file is part of the repository but for some reason it is not checked out, for example because of a sparse checkout. This makes it possible to use at least the 'git submodule' commands which *read* the gitmodules configuration file without fully populating the working tree. Writing to .gitmodules will still require that the file is checked out, so check for that before calling config_set_in_gitmodules_file_gently. Add a similar check also in git-submodule.sh::cmd_add() to anticipate the eventual failure of the "git submodule add" command when .gitmodules is not safely writeable; this prevents the command from leaving the repository in a spurious state (e.g. the submodule repository was cloned but .gitmodules was not updated because config_set_in_gitmodules_file_gently failed). Moreover, since config_from_gitmodules() now accesses the global object store, it is necessary to protect all code paths which call the function against concurrent access to the global object store. Currently this only happens in builtin/grep.c::grep_submodules(), so call grep_read_lock() before invoking code involving config_from_gitmodules(). Finally, add t7418-submodule-sparse-gitmodules.sh to verify that reading from .gitmodules succeeds and that writing to it fails when the file is not checked out. NOTE: there is one rare case where this new feature does not work properly yet: nested submodules without .gitmodules in their working tree. This has been documented with a warning and a test_expect_failure item in t7814, and in this case the current behavior is not altered: no config is read. Signed-off-by: Antonio Ospite <ao2@ao2.it> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-10-25 18:18:12 +02:00
}
usage_with_options(git_submodule_helper_usage, module_config_options);
}
static int module_set_url(int argc, const char **argv, const char *prefix)
{
int quiet = 0;
const char *newurl;
const char *path;
char *config_name;
struct option options[] = {
OPT__QUIET(&quiet, N_("Suppress output for setting url of a submodule")),
OPT_END()
};
const char *const usage[] = {
N_("git submodule--helper set-url [--quiet] <path> <newurl>"),
NULL
};
argc = parse_options(argc, argv, prefix, options, usage, 0);
if (argc != 2 || !(path = argv[0]) || !(newurl = argv[1]))
usage_with_options(usage, options);
config_name = xstrfmt("submodule.%s.url", path);
config_set_in_gitmodules_file_gently(config_name, newurl);
sync_submodule(path, prefix, quiet ? OPT_QUIET : 0);
free(config_name);
return 0;
}
static int module_set_branch(int argc, const char **argv, const char *prefix)
{
int opt_default = 0, ret;
const char *opt_branch = NULL;
const char *path;
char *config_name;
/*
* We accept the `quiet` option for uniformity across subcommands,
* though there is nothing to make less verbose in this subcommand.
*/
struct option options[] = {
OPT_NOOP_NOARG('q', "quiet"),
OPT_BOOL('d', "default", &opt_default,
N_("set the default tracking branch to master")),
OPT_STRING('b', "branch", &opt_branch, N_("branch"),
N_("set the default tracking branch")),
OPT_END()
};
const char *const usage[] = {
N_("git submodule--helper set-branch [-q|--quiet] (-d|--default) <path>"),
N_("git submodule--helper set-branch [-q|--quiet] (-b|--branch) <branch> <path>"),
NULL
};
argc = parse_options(argc, argv, prefix, options, usage, 0);
if (!opt_branch && !opt_default)
die(_("--branch or --default required"));
if (opt_branch && opt_default)
die(_("--branch and --default are mutually exclusive"));
if (argc != 1 || !(path = argv[0]))
usage_with_options(usage, options);
config_name = xstrfmt("submodule.%s.branch", path);
ret = config_set_in_gitmodules_file_gently(config_name, opt_branch);
free(config_name);
return !!ret;
}
#define SUPPORT_SUPER_PREFIX (1<<0)
struct cmd_struct {
const char *cmd;
int (*fn)(int, const char **, const char *);
unsigned option;
};
static struct cmd_struct commands[] = {
{"list", module_list, 0},
{"name", module_name, 0},
{"clone", module_clone, 0},
{"update-module-mode", module_update_module_mode, 0},
{"update-clone", update_clone, 0},
{"ensure-core-worktree", ensure_core_worktree, 0},
{"relative-path", resolve_relative_path, 0},
{"resolve-relative-url", resolve_relative_url, 0},
{"resolve-relative-url-test", resolve_relative_url_test, 0},
{"foreach", module_foreach, SUPPORT_SUPER_PREFIX},
{"init", module_init, SUPPORT_SUPER_PREFIX},
{"status", module_status, SUPPORT_SUPER_PREFIX},
{"print-default-remote", print_default_remote, 0},
{"sync", module_sync, SUPPORT_SUPER_PREFIX},
{"deinit", module_deinit, 0},
{"remote-branch", resolve_remote_submodule_branch, 0},
{"push-check", push_check, 0},
{"absorb-git-dirs", absorb_git_dirs, SUPPORT_SUPER_PREFIX},
{"is-active", is_active, 0},
submodule-config: verify submodule names as paths Submodule "names" come from the untrusted .gitmodules file, but we blindly append them to $GIT_DIR/modules to create our on-disk repo paths. This means you can do bad things by putting "../" into the name (among other things). Let's sanity-check these names to avoid building a path that can be exploited. There are two main decisions: 1. What should the allowed syntax be? It's tempting to reuse verify_path(), since submodule names typically come from in-repo paths. But there are two reasons not to: a. It's technically more strict than what we need, as we really care only about breaking out of the $GIT_DIR/modules/ hierarchy. E.g., having a submodule named "foo/.git" isn't actually dangerous, and it's possible that somebody has manually given such a funny name. b. Since we'll eventually use this checking logic in fsck to prevent downstream repositories, it should be consistent across platforms. Because verify_path() relies on is_dir_sep(), it wouldn't block "foo\..\bar" on a non-Windows machine. 2. Where should we enforce it? These days most of the .gitmodules reads go through submodule-config.c, so I've put it there in the reading step. That should cover all of the C code. We also construct the name for "git submodule add" inside the git-submodule.sh script. This is probably not a big deal for security since the name is coming from the user anyway, but it would be polite to remind them if the name they pick is invalid (and we need to expose the name-checker to the shell anyway for our test scripts). This patch issues a warning when reading .gitmodules and just ignores the related config entry completely. This will generally end up producing a sensible error, as it works the same as a .gitmodules file which is missing a submodule entry (so "submodule update" will barf, but "git clone --recurse-submodules" will print an error but not abort the clone. There is one minor oddity, which is that we print the warning once per malformed config key (since that's how the config subsystem gives us the entries). So in the new test, for example, the user would see three warnings. That's OK, since the intent is that this case should never come up outside of malicious repositories (and then it might even benefit the user to see the message multiple times). Credit for finding this vulnerability and the proof of concept from which the test script was adapted goes to Etienne Stalmans. Signed-off-by: Jeff King <peff@peff.net>
2018-04-30 09:25:25 +02:00
{"check-name", check_name, 0},
{"config", module_config, 0},
{"set-url", module_set_url, 0},
{"set-branch", module_set_branch, 0},
};
int cmd_submodule__helper(int argc, const char **argv, const char *prefix)
{
int i;
if (argc < 2 || !strcmp(argv[1], "-h"))
usage("git submodule--helper <command>");
for (i = 0; i < ARRAY_SIZE(commands); i++) {
if (!strcmp(argv[1], commands[i].cmd)) {
if (get_super_prefix() &&
!(commands[i].option & SUPPORT_SUPER_PREFIX))
die(_("%s doesn't support --super-prefix"),
commands[i].cmd);
return commands[i].fn(argc - 1, argv + 1, prefix);
}
}
die(_("'%s' is not a valid submodule--helper "
"subcommand"), argv[1]);
}