From: Glen Choo <chooglen@google.com>
To: "Ævar Arnfjörð Bjarmason" <avarab@gmail.com>, git@vger.kernel.org
Cc: "Junio C Hamano" <gitster@pobox.com>,
"Jonas Bernoulli" <jonas@bernoul.li>, "Jeff King" <peff@peff.net>,
"Emily Shaffer" <emilyshaffer@google.com>,
"Ævar Arnfjörð Bjarmason" <avarab@gmail.com>
Subject: Re: [PATCH 10/10] submodule: don't use a subprocess to invoke "submodule--helper"
Date: Thu, 20 Oct 2022 16:18:53 -0700 [thread overview]
Message-ID: <kl6lk04uxew2.fsf@chooglen-macbookpro.roam.corp.google.com> (raw)
In-Reply-To: <patch-10.10-81f138e460c-20221017T115544Z-avarab@gmail.com>
Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:
> In a preceding commit we created "builtin/submodule.c" and faithfully
> tried to reproduce every aspect of "git-submodule.sh", including its
> invocation of "git submodule--helper" as a sub-process.
>
> Let's do away with the sub-process and invoke
> "cmd_submodule__helper()" directly. Eventually we'll want to do away
> with "builtin/submodule--helper.c" altogether, but let's not do that
> for now to avoid conflicts with other in-flight topics. Even without
> those conflicts the resulting diff would be large. We can leave that
> for a later cleanup.
Thanks! e.g. this managed to avoid conflicts with my newly sent out
"submodule clone with branches" v2 [1]
[1] https://lore.kernel.org/git/pull.1321.v2.git.git.1666297238.gitgitgadget@gmail.com
> It's also worth noting that some users were using e.g. "git
> submodule--helper list" directly for performance reasons[2]. With
> 31955475d1c (submodule--helper: remove unused "list" helper,
> 2022-09-01) released with v2.38.0 the "list" command was no longer
> provided.
[...]
> I think it would make sense to implement a "--format" option for "git
> submodule foreach" to help anyone who cares about that remaining
> performance (and to improve the API, e.g. by supporting "-z"), but as
> far as performance goes this makes the runtime acceptable again.
FWIW, I don't think that dropping "git submodule--helper list" was a
mistake, since all of "git submodule--helper" was meant to be internal
anyway. But if users find it useful, we could add actually supported
functionality like what you propose here.
> The pattern in "cmd_submodule_builtin()" of saving "struct strvec"
> arguments to a "struct string_list" and free()-ing them after the
> "argv" has been modified by "cmd_submodule__helper()" is new, without
> it we'd get various already-passing tests failing under SANITIZE=leak.
[...]
> +static int cmd_submodule_builtin(struct strvec *args, const char *prefix)
> +{
> + size_t i;
> + struct string_list to_free = STRING_LIST_INIT_DUP;
> + int ret;
> +
> + /*
> + * The cmd_submodule__helper() will treat the argv as
> + * its own and modify it, so e.g. for "git submodule
> + * add" the "add" argument will be removed, and we'll
> + * thus leak from the strvec_push()'s in
> + * setup_helper_args().
> + *
> + * So in lieu of some generic "snapshot for a free"
> + * API for "struct strvec" squirrel away the pointers
> + * to free with string_list_clear() later.
> + */
> + for (i = 0; i < args->nr; i++)
> + string_list_append_nodup(&to_free, (char *)args->v[i]);
> +
> + ret = cmd_submodule__helper(args->nr, args->v, prefix);
> +
> + string_list_clear(&to_free, 0);
Hm, so this trick works because we init STRING_LIST_INIT_DUP to make the
string_list think that it owns its strings, but when we append, we use
string_list_append_nodup(), which doesn't dup the strings. This
'moves' the strings into the string_list and causes them to be freed
when we call string_list_clear().
Sounds reasonable enough to me, but I'm not an expert on leaks :)
> + free(strvec_detach(args));
> +
> + return ret;
> +}
> +
> int cmd_submodule(int argc, const char **argv, const char *prefix)
> {
> int opt_quiet = 0;
> int opt_cached = 0;
> int opt_recursive = 0;
> - struct child_process cp = CHILD_PROCESS_INIT;
> + struct strvec args = STRVEC_INIT;
> struct option options[] = {
> OPT__QUIET(&opt_quiet, N_("be quiet")),
> OPT_BOOL(0, "cached", &opt_cached,
> @@ -141,13 +169,10 @@ int cmd_submodule(int argc, const char **argv, const char *prefix)
> * Tell the rest of git that any URLs we get don't come
> * directly from the user, so it can apply policy as appropriate.
> */
> - strvec_push(&cp.env, "GIT_PROTOCOL_FROM_USER=0");
> - setup_helper_args(argc, argv, prefix, opt_quiet, opt_cached,
> - opt_recursive, &cp.args, options);
> + xsetenv("GIT_PROTOCOL_FROM_USER", "0", 1);
>
> - cp.git_cmd = 1;
> - cp.no_stdin = 0; /* for git submodule foreach */
> - cp.dir = startup_info->original_cwd;
> + setup_helper_args(argc, argv, prefix, opt_quiet, opt_cached,
> + opt_recursive, &args, options);
>
> - return run_command(&cp);
> + return cmd_submodule_builtin(&args, prefix);
> }
> --
> 2.38.0.1091.gf9d18265e59
prev parent reply other threads:[~2022-10-20 23:19 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-17 12:09 [PATCH 00/10] submodule: make it a built-in, remove git-submodule.sh Ævar Arnfjörð Bjarmason
2022-10-17 12:09 ` [PATCH 01/10] git-submodule.sh: create a "case" dispatch statement Ævar Arnfjörð Bjarmason
2022-10-17 12:09 ` [PATCH 02/10] git-submodule.sh: dispatch "sync" to helper Ævar Arnfjörð Bjarmason
2022-10-20 20:42 ` Glen Choo
2022-10-17 12:09 ` [PATCH 03/10] git-submodule.sh: dispatch directly " Ævar Arnfjörð Bjarmason
2022-10-17 12:09 ` [PATCH 04/10] git-submodule.sh: dispatch "foreach" " Ævar Arnfjörð Bjarmason
2022-10-20 21:14 ` Glen Choo
2022-10-17 12:09 ` [PATCH 05/10] git-submodule.sh: dispatch "update" " Ævar Arnfjörð Bjarmason
2022-10-20 21:50 ` Glen Choo
2022-10-17 12:09 ` [PATCH 06/10] git-submodule.sh: don't support top-level "--cached" Ævar Arnfjörð Bjarmason
2022-10-20 22:14 ` Glen Choo
2022-10-17 12:09 ` [PATCH 07/10] submodule: make it a built-in, remove git-submodule.sh Ævar Arnfjörð Bjarmason
2022-10-20 22:49 ` Glen Choo
2022-10-17 12:09 ` [PATCH 08/10] submodule: support "--" with no other arguments Ævar Arnfjörð Bjarmason
2022-10-17 12:09 ` [PATCH 09/10] submodule: support sub-command-less "--recursive" option Ævar Arnfjörð Bjarmason
2022-10-20 23:05 ` Glen Choo
2022-10-17 12:09 ` [PATCH 10/10] submodule: don't use a subprocess to invoke "submodule--helper" Ævar Arnfjörð Bjarmason
2022-10-20 23:18 ` Glen Choo [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=kl6lk04uxew2.fsf@chooglen-macbookpro.roam.corp.google.com \
--to=chooglen@google.com \
--cc=avarab@gmail.com \
--cc=emilyshaffer@google.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=jonas@bernoul.li \
--cc=peff@peff.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).