[PATCH] mailmap.c: fix a memory leak in free_mailap_{info,entry}()

From: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date: Aug 31, 2021, 13:42 UTC
Message-ID: <patch-1.1-f11eb44e4c5-20210831T134023Z-avarab@gmail.com>
In-Reply-To: <patch-4.4-ad8680f529-20210714T172251Z-avarab@gmail.com>

In the free_mailmap_entry() code added in 0925ce4d49 (Add map_user() and clear_mailmap() to mailmap, 2009-02-08) the intent was clearly to clear the "me" structure, but while we freed parts of the mailmap_entry structure, we didn't free the structure itself. The same goes for the "mailmap_info" structure.

This brings us from 50 failures when running t4203-mailmap.sh to
49. Not really progress as far as the number of failures is concerned,
but as far as I can tell this fixes all leaks in mailmap.c
itself. There's still users of it such as builtin/log.c that call
read_mailmap() without a clear_mailmap(), but that's on them.

Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
---

This was originally submitted as part of the SANITIZE=leak series as https://lore.kernel.org/git/patch-4.4-ad8680f529-20210714T172251Z-avarab@gmail.com/

In its v3 I stopped doing these leak fixes & test changes, let's just consider this separately. We'll eventually want to add SANITIZE=leak whitelisting to the relevant test if and when my SANITIZE=leak series goes in, but we can just do that then along with adding various other tests.

Range-diff:
1:  80edda308c9 ! 1:  f11eb44e4c5 SANITIZE tests: fix leak in mailmap.c
    @@ Metadata
     Author: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
     
      ## Commit message ##
    -    SANITIZE tests: fix leak in mailmap.c
    -
    -    Get closer to being able to run t4203-mailmap.sh by fixing a couple of
    -    memory leak in mailmap.c.
    +    mailmap.c: fix a memory leak in free_mailap_{info,entry}()
     
         In the free_mailmap_entry() code added in 0925ce4d49 (Add map_user()
         and clear_mailmap() to mailmap, 2009-02-08) the intent was clearly to
    @@ Commit message
         mailmap_entry structure, we didn't free the structure itself. The same
         goes for the "mailmap_info" structure.
     
    +    This brings us from 50 failures when running t4203-mailmap.sh to
    +    49. Not really progress as far as the number of failures is concerned,
    +    but as far as I can tell this fixes all leaks in mailmap.c
    +    itself. There's still users of it such as builtin/log.c that call
    +    read_mailmap() without a clear_mailmap(), but that's on them.
    +
         Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
     
      ## mailmap.c ##
    @@ mailmap.c: static void free_mailmap_entry(void *p, const char *s)
      }
      
      /*
    -
    - ## t/t4203-mailmap.sh ##
    -@@ t/t4203-mailmap.sh: test_expect_success 'check-mailmap bogus contact --stdin' '
    - 	test_must_fail git check-mailmap --stdin bogus </dev/null
    - '
    - 
    -+if test_have_prereq SANITIZE_LEAK
    -+then
    -+	skip_all='skipping the rest of mailmap tests under SANITIZE_LEAK'
    -+	test_done
    -+fi
    -+
    - test_expect_success 'No mailmap' '
    - 	cat >expect <<-EOF &&
    - 	$GIT_AUTHOR_NAME (1):

 mailmap.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/mailmap.c b/mailmap.c
index 462b3956340..40ce152024d 100644
--- a/mailmap.c
+++ b/mailmap.c
@@ -37,6 +37,7 @@ static void free_mailmap_info(void *p, const char *s)
 		 s, debug_str(mi->name), debug_str(mi->email));
 	free(mi->name);
 	free(mi->email);
+	free(mi);
 }
 
 static void free_mailmap_entry(void *p, const char *s)
@@ -52,6 +53,7 @@ static void free_mailmap_entry(void *p, const char *s)
 
 	me->namemap.strdup_strings = 1;
 	string_list_clear_func(&me->namemap, free_mailmap_info);
+	free(me);
 }
 
 /*

-- 
2.33.0.805.g739b16c2189