# [PATCH 0/3] fix memory leak in git-am 5 messages from 2017-04-27 to 2017-04-27. Participants: Jeff King, Junio C Hamano. Thread: https://gitlist.dev/t/45815 ## Jeff King, 2017-04-27 03:25 Subject: [PATCH 0/3] fix memory leak in git-am Message-ID: <20170427032503.rgz3bnctnxdn3hin@sigill.intra.peff.net> URL: https://gitlist.dev/e/20170427032503.rgz3bnctnxdn3hin%40sigill.intra.peff.net ``` This was noticed by Coverity. The leak isn't new, but I think was "re-found" by Coverity because some nearby code did an unrelated s/sha1/oid/ change, throwing off its heuristics. I also checked whether this was in Dscho's big pack of Coverity fixups from earlier today, and it's not. The first one is the actual fix. The second one is pure cleanup, but I think worth doing. The final one is arguably just churn, and maybe people even like the end result less. I'm OK to drop it. [1/3]: am: fix commit buffer leak in get_commit_info() [2/3]: am: simplify allocations in get_commit_info() [3/3]: am: shorten ident_split variable name in get_commit_info() builtin/am.c | 34 ++++++++++++++-------------------- 1 file changed, 14 insertions(+), 20 deletions(-) -Peff ``` ## Jeff King, 2017-04-27 03:25 Subject: [PATCH 1/3] am: fix commit buffer leak in get_commit_info() Message-ID: <20170427032555.imjr5to4uys2cz4k@sigill.intra.peff.net> URL: https://gitlist.dev/e/20170427032555.imjr5to4uys2cz4k%40sigill.intra.peff.net In-Reply-To: <20170427032503.rgz3bnctnxdn3hin@sigill.intra.peff.net> ``` Calling logmsg_reencode() may allocate a buffer for the commit message (because we need to load it from disk, or because it needs re-encoded). We must "unuse" it afterwards to free it. Signed-off-by: Jeff King --- builtin/am.c | 1 + 1 file changed, 1 insertion(+) diff --git a/builtin/am.c b/builtin/am.c index 805f56cec..baab951ed 100644 --- a/builtin/am.c +++ b/builtin/am.c @@ -1421,6 +1421,7 @@ static void get_commit_info(struct am_state *state, struct commit *commit) die(_("unable to parse commit %s"), oid_to_hex(&commit->object.oid)); state->msg = xstrdup(msg + 2); state->msg_len = strlen(state->msg); + unuse_commit_buffer(commit, buffer); } /** -- 2.13.0.rc0.465.ga1e654229 ``` ## Jeff King, 2017-04-27 03:27 Subject: [PATCH 2/3] am: simplify allocations in get_commit_info() Message-ID: <20170427032717.hzs3og4ql5vceqlf@sigill.intra.peff.net> URL: https://gitlist.dev/e/20170427032717.hzs3og4ql5vceqlf%40sigill.intra.peff.net In-Reply-To: <20170427032503.rgz3bnctnxdn3hin@sigill.intra.peff.net> ``` After we call split_ident_line(), we have several begin/end pairs for various parts of the ident. We then copy each into a strbuf to create a single string, and then detach that string. We can instead skip the strbuf entirely and just duplicate the strings directly. This is shorter, and it makes it more obvious that we are not leaking the strbuf (we were not before, because every code path either died or hit a strbuf_detach). Signed-off-by: Jeff King --- builtin/am.c | 25 ++++++++++--------------- 1 file changed, 10 insertions(+), 15 deletions(-) diff --git a/builtin/am.c b/builtin/am.c index baab951ed..e08f03326 100644 --- a/builtin/am.c +++ b/builtin/am.c @@ -1380,40 +1380,35 @@ static int get_mail_commit_oid(struct object_id *commit_id, const char *mail) */ static void get_commit_info(struct am_state *state, struct commit *commit) { - const char *buffer, *ident_line, *author_date, *msg; + const char *buffer, *ident_line, *msg; size_t ident_len; struct ident_split ident_split; - struct strbuf sb = STRBUF_INIT; buffer = logmsg_reencode(commit, NULL, get_commit_output_encoding()); ident_line = find_commit_header(buffer, "author", &ident_len); - if (split_ident_line(&ident_split, ident_line, ident_len) < 0) { - strbuf_add(&sb, ident_line, ident_len); - die(_("invalid ident line: %s"), sb.buf); - } + if (split_ident_line(&ident_split, ident_line, ident_len) < 0) + die(_("invalid ident line: %.*s"), (int)ident_len, ident_line); assert(!state->author_name); if (ident_split.name_begin) { - strbuf_add(&sb, ident_split.name_begin, - ident_split.name_end - ident_split.name_begin); - state->author_name = strbuf_detach(&sb, NULL); + state->author_name = + xmemdupz(ident_split.name_begin, + ident_split.name_end - ident_split.name_begin); } else state->author_name = xstrdup(""); assert(!state->author_email); if (ident_split.mail_begin) { - strbuf_add(&sb, ident_split.mail_begin, - ident_split.mail_end - ident_split.mail_begin); - state->author_email = strbuf_detach(&sb, NULL); + state->author_email = + xmemdupz(ident_split.mail_begin, + ident_split.mail_end - ident_split.mail_begin); } else state->author_email = xstrdup(""); - author_date = show_ident_date(&ident_split, DATE_MODE(NORMAL)); - strbuf_addstr(&sb, author_date); assert(!state->author_date); - state->author_date = strbuf_detach(&sb, NULL); + state->author_date = xstrdup(show_ident_date(&ident_split, DATE_MODE(NORMAL))); assert(!state->msg); msg = strstr(buffer, "\n\n"); -- 2.13.0.rc0.465.ga1e654229 ``` ## Jeff King, 2017-04-27 03:28 Subject: [PATCH 3/3] am: shorten ident_split variable name in get_commit_info() Message-ID: <20170427032830.sushxlvqb44px7lx@sigill.intra.peff.net> URL: https://gitlist.dev/e/20170427032830.sushxlvqb44px7lx%40sigill.intra.peff.net In-Reply-To: <20170427032503.rgz3bnctnxdn3hin@sigill.intra.peff.net> ``` The local ident_split variable is often mentioned three times per line when dealing with its begin/end pointer pairs. Let's use a shorter name which lets us get rid of some long lines. Since this is a short self-contained function, readability doesn't suffer. Signed-off-by: Jeff King --- builtin/am.c | 20 +++++++++----------- 1 file changed, 9 insertions(+), 11 deletions(-) diff --git a/builtin/am.c b/builtin/am.c index e08f03326..2aed3ef01 100644 --- a/builtin/am.c +++ b/builtin/am.c @@ -1382,33 +1382,31 @@ static void get_commit_info(struct am_state *state, struct commit *commit) { const char *buffer, *ident_line, *msg; size_t ident_len; - struct ident_split ident_split; + struct ident_split id; buffer = logmsg_reencode(commit, NULL, get_commit_output_encoding()); ident_line = find_commit_header(buffer, "author", &ident_len); - if (split_ident_line(&ident_split, ident_line, ident_len) < 0) + if (split_ident_line(&id, ident_line, ident_len) < 0) die(_("invalid ident line: %.*s"), (int)ident_len, ident_line); assert(!state->author_name); - if (ident_split.name_begin) { + if (id.name_begin) state->author_name = - xmemdupz(ident_split.name_begin, - ident_split.name_end - ident_split.name_begin); - } else + xmemdupz(id.name_begin, id.name_end - id.name_begin); + else state->author_name = xstrdup(""); assert(!state->author_email); - if (ident_split.mail_begin) { + if (id.mail_begin) state->author_email = - xmemdupz(ident_split.mail_begin, - ident_split.mail_end - ident_split.mail_begin); - } else + xmemdupz(id.mail_begin, id.mail_end - id.mail_begin); + else state->author_email = xstrdup(""); assert(!state->author_date); - state->author_date = xstrdup(show_ident_date(&ident_split, DATE_MODE(NORMAL))); + state->author_date = xstrdup(show_ident_date(&id, DATE_MODE(NORMAL))); assert(!state->msg); msg = strstr(buffer, "\n\n"); -- 2.13.0.rc0.465.ga1e654229 ``` ## Junio C Hamano, 2017-04-27 05:41 Subject: Re: [PATCH 0/3] fix memory leak in git-am Message-ID: URL: https://gitlist.dev/e/xmqqd1byo196.fsf%40gitster.mtv.corp.google.com In-Reply-To: <20170427032503.rgz3bnctnxdn3hin@sigill.intra.peff.net> ``` Jeff King writes: > This was noticed by Coverity. The leak isn't new, but I think was > "re-found" by Coverity because some nearby code did an unrelated > s/sha1/oid/ change, throwing off its heuristics. > > I also checked whether this was in Dscho's big pack of Coverity fixups > from earlier today, and it's not. > > The first one is the actual fix. The second one is pure cleanup, but I > think worth doing. The final one is arguably just churn, and maybe > people even like the end result less. I'm OK to drop it. > > [1/3]: am: fix commit buffer leak in get_commit_info() > [2/3]: am: simplify allocations in get_commit_info() > [3/3]: am: shorten ident_split variable name in get_commit_info() All looked sensible. Thanks. ```