diff options
author | Ævar Arnfjörð Bjarmason <avarab@gmail.com> | 2021-10-07 12:01:35 +0200 |
---|---|---|
committer | Junio C Hamano <gitster@pobox.com> | 2021-10-07 15:40:15 -0700 |
commit | eab4ac6a233e505e78fc8b04df03d58628d5ff3e (patch) | |
tree | b78169da7e63ce82c038dceeaf9212c61d35a80d /t/t9126/follow-deleted-readded.dump | |
parent | tests: fix test-oid-array leak, test in SANITIZE=leak (diff) | |
download | tgif-eab4ac6a233e505e78fc8b04df03d58628d5ff3e.tar.xz |
ls-files: fix a trivial dir_clear() leak
Fix an edge case that was missed when the dir_clear() call was added
in eceba532141 (dir: fix problematic API to avoid memory leaks,
2020-08-18), we need to also clean up when we're about to exit with
non-zero.
That commit says, on the topic of the dir_clear() API and UNLEAK():
[...]two of them clearly thought about leaks since they had an
UNLEAK(dir) directive, which to me suggests that the method to
free the data was too unclear.
I think that 0e5bba53af7 (add UNLEAK annotation for reducing leak
false positives, 2017-09-08) which added the UNLEAK() makes it clear
that that wasn't the case, rather it was the desire to avoid the
complexity of freeing the memory at the end of the program.
This does add a bit of complexity, but I think it's worth it to just
fix these leaks when it's easy in built-ins. It allows them to serve
as canaries for underlying APIs that shouldn't be leaking, it
encourages us to make those freeing APIs nicer for all their users,
and it prevents other leaking regressions by being able to mark the
entire test as TEST_PASSES_SANITIZE_LEAK=true.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Diffstat (limited to 't/t9126/follow-deleted-readded.dump')
0 files changed, 0 insertions, 0 deletions