# [PATCH] multi-pack-index: fix potential segfault without sub-command 4 messages from 2021-07-19 to 2021-07-23. Participants: Taylor Blau, Jeff King. Thread: https://gitlist.dev/t/56125 ## Taylor Blau, 2021-07-19 17:18 Subject: [PATCH] multi-pack-index: fix potential segfault without sub-command Message-ID: <8c0bb3e0dc121bd68f7014000fbb60b28750a0fe.1626715096.git.me@ttaylorr.com> URL: https://gitlist.dev/e/8c0bb3e0dc121bd68f7014000fbb60b28750a0fe.1626715096.git.me%40ttaylorr.com ``` Since cd57bc41bb (builtin/multi-pack-index.c: display usage on unrecognized command, 2021-03-30) we have used a "usage" label to avoid having two separate callers of usage_with_options (one when no arguments are given, and another for unrecognized sub-commands). But the first caller has been broken since cd57bc41bb, since it will happily jump to usage without arguments, and then pass argv[0] to the "unrecognized subcommand" error. Many compilers will save us from a segfault here, but the end result is ugly, since it mentions an unrecognized subcommand when we didn't even pass one, and (on GCC) includes "(null)" in its output. Move the "usage" label down past the error about unrecognized subcommands so that it is only triggered when it should be. While we're at it, bulk up our test coverage in this area, too. Signed-off-by: Taylor Blau --- Noticed this while I was reading code in a similar area while reviewing one of Ævar's series. builtin/multi-pack-index.c | 2 +- t/t5319-multi-pack-index.sh | 5 +++++ 2 files changed, 6 insertions(+), 1 deletion(-) diff --git a/builtin/multi-pack-index.c b/builtin/multi-pack-index.c index 5d3ea445fd..8ff0dee2ec 100644 --- a/builtin/multi-pack-index.c +++ b/builtin/multi-pack-index.c @@ -176,8 +176,8 @@ int cmd_multi_pack_index(int argc, const char **argv, else if (!strcmp(argv[0], "expire")) return cmd_multi_pack_index_expire(argc, argv); else { -usage: error(_("unrecognized subcommand: %s"), argv[0]); +usage: usage_with_options(builtin_multi_pack_index_usage, builtin_multi_pack_index_options); } diff --git a/t/t5319-multi-pack-index.sh b/t/t5319-multi-pack-index.sh index 5641d158df..dab7123b3a 100755 --- a/t/t5319-multi-pack-index.sh +++ b/t/t5319-multi-pack-index.sh @@ -824,4 +824,9 @@ test_expect_success 'load reverse index when missing .idx, .pack' ' ) ' +test_expect_success 'usage shown without sub-command' ' + test_expect_code 129 git multi-pack-index 2>err && + ! test_i18ngrep "unrecognized subcommand" err +' + test_done -- 2.31.1.163.ga65ce7f831 ``` ## Jeff King, 2021-07-21 08:10 Subject: Re: [PATCH] multi-pack-index: fix potential segfault without sub-command Message-ID: URL: https://gitlist.dev/e/YPfWkzRtQKthOgZx%40coredump.intra.peff.net In-Reply-To: <8c0bb3e0dc121bd68f7014000fbb60b28750a0fe.1626715096.git.me@ttaylorr.com> ``` On Mon, Jul 19, 2021 at 01:18:49PM -0400, Taylor Blau wrote: > Since cd57bc41bb (builtin/multi-pack-index.c: display usage on > unrecognized command, 2021-03-30) we have used a "usage" label to avoid > having two separate callers of usage_with_options (one when no arguments > are given, and another for unrecognized sub-commands). > > But the first caller has been broken since cd57bc41bb, since it will > happily jump to usage without arguments, and then pass argv[0] to the > "unrecognized subcommand" error. > > Many compilers will save us from a segfault here, but the end result is > ugly, since it mentions an unrecognized subcommand when we didn't even > pass one, and (on GCC) includes "(null)" in its output. > > Move the "usage" label down past the error about unrecognized > subcommands so that it is only triggered when it should be. While we're > at it, bulk up our test coverage in this area, too. Good find. The code change seems obviously correct. > +test_expect_success 'usage shown without sub-command' ' > + test_expect_code 129 git multi-pack-index 2>err && > + ! test_i18ngrep "unrecognized subcommand" err > +' I think we're avoiding test_i18ngrep in new code these days. -Peff ``` ## Taylor Blau, 2021-07-21 16:52 Subject: [PATCH v2] multi-pack-index: fix potential segfault without sub-command Message-ID: <12804130427fd71e88ce10aed283c1f9cdd330ef.1626886330.git.me@ttaylorr.com> URL: https://gitlist.dev/e/12804130427fd71e88ce10aed283c1f9cdd330ef.1626886330.git.me%40ttaylorr.com In-Reply-To: ``` Since cd57bc41bb (builtin/multi-pack-index.c: display usage on unrecognized command, 2021-03-30) we have used a "usage" label to avoid having two separate callers of usage_with_options (one when no arguments are given, and another for unrecognized sub-commands). But the first caller has been broken since cd57bc41bb, since it will happily jump to usage without arguments, and then pass argv[0] to the "unrecognized subcommand" error. Many compilers will save us from a segfault here, but the end result is ugly, since it mentions an unrecognized subcommand when we didn't even pass one, and (on GCC) includes "(null)" in its output. Move the "usage" label down past the error about unrecognized subcommands so that it is only triggered when it should be. While we're at it, bulk up our test coverage in this area, too. Signed-off-by: Taylor Blau --- I didn't realize that we are veering away from test_i18ngrep in new tests, so here's a version of this trivial patch with grep instead. Range-diff against v1: 1: 8c0bb3e0dc ! 1: 1280413042 multi-pack-index: fix potential segfault without sub-command @@ t/t5319-multi-pack-index.sh: test_expect_success 'load reverse index when missin +test_expect_success 'usage shown without sub-command' ' + test_expect_code 129 git multi-pack-index 2>err && -+ ! test_i18ngrep "unrecognized subcommand" err ++ ! grep "unrecognized subcommand" err +' + test_done builtin/multi-pack-index.c | 2 +- t/t5319-multi-pack-index.sh | 5 +++++ 2 files changed, 6 insertions(+), 1 deletion(-) diff --git a/builtin/multi-pack-index.c b/builtin/multi-pack-index.c index 5d3ea445fd..8ff0dee2ec 100644 --- a/builtin/multi-pack-index.c +++ b/builtin/multi-pack-index.c @@ -176,8 +176,8 @@ int cmd_multi_pack_index(int argc, const char **argv, else if (!strcmp(argv[0], "expire")) return cmd_multi_pack_index_expire(argc, argv); else { -usage: error(_("unrecognized subcommand: %s"), argv[0]); +usage: usage_with_options(builtin_multi_pack_index_usage, builtin_multi_pack_index_options); } diff --git a/t/t5319-multi-pack-index.sh b/t/t5319-multi-pack-index.sh index 7609f1ea64..c0ec5a6fd3 100755 --- a/t/t5319-multi-pack-index.sh +++ b/t/t5319-multi-pack-index.sh @@ -837,4 +837,9 @@ test_expect_success 'load reverse index when missing .idx, .pack' ' ) ' +test_expect_success 'usage shown without sub-command' ' + test_expect_code 129 git multi-pack-index 2>err && + ! grep "unrecognized subcommand" err +' + test_done -- 2.31.1.163.ga65ce7f831 ``` ## Jeff King, 2021-07-23 07:34 Subject: Re: [PATCH v2] multi-pack-index: fix potential segfault without sub-command Message-ID: URL: https://gitlist.dev/e/YPpxA1XWw65pRr0P%40coredump.intra.peff.net In-Reply-To: <12804130427fd71e88ce10aed283c1f9cdd330ef.1626886330.git.me@ttaylorr.com> ``` On Wed, Jul 21, 2021 at 12:52:37PM -0400, Taylor Blau wrote: > I didn't realize that we are veering away from test_i18ngrep in new > tests, so here's a version of this trivial patch with grep instead. This looks fine to me, but it looks like Junio may have already merged the original into 'next' via 995cb54b5b (Merge branch 'tb/reverse-midx' into next, 2021-07-20). So we'd have to update it on top. I think we can just leave it be, though. There's nothing incorrect about using test_i18ngrep for now, and there are a ton of existing instances to handle if we choose to do a bulk cleanup. One more isn't a big deal. -Peff ```