Here is a small series that came out of a corruption we noticed at GitHub. The
corruption is described in the third patch, but the gist is this:
- After writing a MIDX, we experience a disk failure that corrupts the
contents of the repository's MIDX.
- We later go to write another MIDX, notice that one is already there, and
then blindly reuse its data without checking its validity, resulting in a
MIDX that contains corrupt data, but whose trailing checksum matches the
preceding contents.
This series checks the validity of the trailing checksum when reusing a MIDX
while writing a new one, as well as a few other related fix-ups. That should be
sufficient to avoid propagating corrupt data forward while writing a MIDX.
We could go further here and run 'git multi-pack-index verify' on the existing
MIDX while writing a new one, but the point of this reuse is to avoid having to
re-compute/verify the contents of an existing MIDX. So a more lightweight option
is pursued here, but alternatives are discussed in 3/4.
Thanks in advance for your review.
Taylor Blau (4):
csum-file: introduce checksum_valid()
commit-graph: rewrite to use checksum_valid()
midx: don't reuse corrupt MIDXs when writing
midx: report checksum mismatches during 'verify'
commit-graph.c | 14 ++++++--------
csum-file.c | 16 ++++++++++++++++
csum-file.h | 3 +++
midx.c | 13 +++++++++++++
pack-check.c | 11 +----------
t/t5319-multi-pack-index.sh | 13 +++++++++++++
6 files changed, 52 insertions(+), 18 deletions(-)
--
2.31.1.163.ga65ce7f831
