files_transaction_prepare(): fix handling of ref lock failure
authorMichael Haggerty <mhagger@alum.mit.edu>
Tue, 24 Oct 2017 15:16:25 +0000 (17:16 +0200)
committerJunio C Hamano <gitster@pobox.com>
Wed, 25 Oct 2017 06:08:26 +0000 (15:08 +0900)
Since dc39e09942 (files_ref_store: use a transaction to update packed
refs, 2017-09-08), failure to lock a reference has been handled
incorrectly by `files_transaction_prepare()`. If
`lock_ref_for_update()` fails in the lock-acquisition loop of that
function, it sets `ret` then breaks out of that loop. Prior to
dc39e09942, that was OK, because the only thing following the loop was
the cleanup code. But dc39e09942 added another blurb of code between
the loop and the cleanup. That blurb sometimes resets `ret` to zero,
making the cleanup code think that the locking was successful.

Specifically, whenever

* One or more reference deletions have been processed successfully in
the lock-acquisition loop. (Processing the first such reference
causes a packed-ref transaction to be initialized.)

* Then `lock_ref_for_update()` fails for a subsequent reference. Such
a failure can happen for a number of reasons, such as the old SHA-1
not being correct, lock contention, etc. This causes a `break` out
of the lock-acquisition loop.

* The `packed-refs` lock is acquired successfully and
`ref_transaction_prepare()` succeeds for the packed-ref transaction.
This has the effect of resetting `ret` back to 0, and making the
cleanup code think that lock acquisition was successful.

In that case, any reference updates that were processed prior to
breaking out of the loop would be carried out (loose and packed), but
the reference that couldn't be locked and any subsequent references
would silently be ignored.

This can easily cause data loss if, for example, the user was trying
to push a new name for an existing branch while deleting the old name.
After the push, the branch could be left unreachable, and could even
subsequently be garbage-collected.

This problem was noticed in the context of deleting one reference and
creating another in a single transaction, when the two references D/F
conflict with each other, like

git update-ref --stdin <<EOF
delete refs/foo
create refs/foo/bar HEAD
EOF

This triggers the above bug because the deletion is processed
successfully for `refs/foo`, then the D/F conflict causes
`lock_ref_for_update()` to fail when `refs/foo/bar` is processed. In
this case the transaction *should* fail, but instead it causes
`refs/foo` to be deleted without creating `refs/foo`. This could
easily result in data loss.

The fix is simple: instead of just breaking out of the loop, jump
directly to the cleanup code. This fixes some tests in t1404 that were
added in the previous commit.

Signed-off-by: Michael Haggerty <mhagger@alum.mit.edu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
refs/files-backend.c
t/t1404-update-ref-errors.sh
index 961424a4ea7e9d08f65e053243ad2def93456feb..a8c29b7865e327e9a09dc832fb0ea7994ce82ce5 100644 (file)
@@ -2527,7 +2527,7 @@ static int files_transaction_prepare(struct ref_store *ref_store,
                ret = lock_ref_for_update(refs, update, transaction,
                                          head_ref, &affected_refnames, err);
                if (ret)
-                       break;
+                       goto cleanup;
 
                if (update->flags & REF_DELETING &&
                    !(update->flags & REF_LOG_ONLY) &&
index f26a17df33eeebad30f90e9f224340ff14d24b05..3a887b511340b50a0cadcf1968a8c0986d67d66c 100755 (executable)
@@ -271,11 +271,11 @@ test_expect_success 'D/F conflict prevents add short + delete long' '
        df_test refs/df-as-dl --add-del foo foo/bar
 '
 
-test_expect_failure 'D/F conflict prevents delete long + add short' '
+test_expect_success 'D/F conflict prevents delete long + add short' '
        df_test refs/df-dl-as --del-add foo/bar foo
 '
 
-test_expect_failure 'D/F conflict prevents delete short + add long' '
+test_expect_success 'D/F conflict prevents delete short + add long' '
        df_test refs/df-ds-al --del-add foo foo/bar
 '
 
@@ -287,17 +287,17 @@ test_expect_success 'D/F conflict prevents add short + delete long packed' '
        df_test refs/df-as-dlp --pack --add-del foo foo/bar
 '
 
-test_expect_failure 'D/F conflict prevents delete long packed + add short' '
+test_expect_success 'D/F conflict prevents delete long packed + add short' '
        df_test refs/df-dlp-as --pack --del-add foo/bar foo
 '
 
-test_expect_failure 'D/F conflict prevents delete short packed + add long' '
+test_expect_success 'D/F conflict prevents delete short packed + add long' '
        df_test refs/df-dsp-al --pack --del-add foo foo/bar
 '
 
 # Try some combinations involving symbolic refs...
 
-test_expect_failure 'D/F conflict prevents indirect add long + delete short' '
+test_expect_success 'D/F conflict prevents indirect add long + delete short' '
        df_test refs/df-ial-ds --sym-add --add-del foo/bar foo
 '
 
@@ -309,11 +309,11 @@ test_expect_success 'D/F conflict prevents indirect add short + indirect delete
        df_test refs/df-ias-idl --sym-add --sym-del --add-del foo foo/bar
 '
 
-test_expect_failure 'D/F conflict prevents indirect delete long + indirect add short' '
+test_expect_success 'D/F conflict prevents indirect delete long + indirect add short' '
        df_test refs/df-idl-ias --sym-add --sym-del --del-add foo/bar foo
 '
 
-test_expect_failure 'D/F conflict prevents indirect add long + delete short packed' '
+test_expect_success 'D/F conflict prevents indirect add long + delete short packed' '
        df_test refs/df-ial-dsp --sym-add --pack --add-del foo/bar foo
 '
 
@@ -325,7 +325,7 @@ test_expect_success 'D/F conflict prevents add long + indirect delete short pack
        df_test refs/df-al-idsp --sym-del --pack --add-del foo/bar foo
 '
 
-test_expect_failure 'D/F conflict prevents indirect delete long packed + indirect add short' '
+test_expect_success 'D/F conflict prevents indirect delete long packed + indirect add short' '
        df_test refs/df-idlp-ias --sym-add --sym-del --pack --del-add foo/bar foo
 '