object-file: fix SEGV on free() regression in v2.34.0-rc2
commit168a937bbcf74370267fe04f5368035948745baa
authorÆvar Arnfjörð Bjarmason <avarab@gmail.com>
Thu, 11 Nov 2021 05:18:55 +0000 (11 06:18 +0100)
committerJunio C Hamano <gitster@pobox.com>
Thu, 11 Nov 2021 18:41:54 +0000 (11 10:41 -0800)
tree6c558744b02d9a6d0ab1e10aaee842ba0feea45a
parent96e41f58fe1a5aeadf2bf1c1850c53a1c1144bbc
object-file: fix SEGV on free() regression in v2.34.0-rc2

Fix a regression introduced in my 96e41f58fe1 (fsck: report invalid
object type-path combinations, 2021-10-01). When fsck-ing blobs larger
than core.bigFileThreshold, we'd free() a pointer to uninitialized
memory.

This issue would have been caught by SANITIZE=address, but since it
involves core.bigFileThreshold, none of the existing tests in our test
suite covered it.

Running them with the "big_file_threshold" in "environment.c" changed
to say "6" would have shown this failure, but let's add a dedicated
test for this scenario based on Han Xin's report[1].

The bug was introduced between v9 and v10[2] of the fsck series merged
in 061a21d36d8 (Merge branch 'ab/fsck-unexpected-type', 2021-10-25).

1. https://lore.kernel.org/git/20211111030302.75694-1-hanxin.hx@alibaba-inc.com/
2. https://lore.kernel.org/git/cover-v10-00.17-00000000000-20211001T091051Z-avarab@gmail.com/

Reported-by: Han Xin <chiyutianyi@gmail.com>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
object-file.c
t/t1050-large.sh