[PATCH v2 0/6] fsck: better "invalid object" error reporting

["Ævar Arnfjörð Bjarmason" <avarab@gmail.com>]

A re-roll of improved error reporting for fsck-ing bad loose
objects. See [1] for v1.

This is no longer based on the series to add a bug() function, since
as noted in the re-roll of that[2] that function is gone. This version
uses a plain BUG() for that condition.

Other than that the only change is improved commit messages, and I
added a trivial patch to move read_loose_object() around in
object-store.h so I wouldn't need a forward declaration, and updated
the comment for that function.

1. https://lore.kernel.org/git/cover-0.6-00000000000-20210328T025618Z-avarab@gmail.com/
2. https://lore.kernel.org/git/cover-0.3-00000000000-20210413T090603Z-avarab@gmail.com


Ævar Arnfjörð Bjarmason (6):
  cache.h: move object functions to object-store.h
  fsck tests: refactor one test to use a sub-repo
  fsck: don't hard die on invalid object types
  object-store.h: move read_loose_object() below 'struct object_info'
  fsck: report invalid types recorded in objects
  fsck: report invalid object type-path combinations

 builtin/cat-file.c    |  7 +++--
 builtin/fast-export.c |  2 +-
 builtin/fsck.c        | 28 +++++++++++++++---
 builtin/index-pack.c  |  2 +-
 builtin/mktag.c       |  3 +-
 cache.h               | 10 -------
 object-file.c         | 66 +++++++++++++++++++++----------------------
 object-store.h        | 39 ++++++++++++++++---------
 object.c              |  4 +--
 pack-check.c          |  3 +-
 streaming.c           |  5 +++-
 t/t1450-fsck.sh       | 64 ++++++++++++++++++++++++++++++-----------
 12 files changed, 146 insertions(+), 87 deletions(-)

Range-diff against v1:
1:  f8f00db8d31 = 1:  37c323a2410 cache.h: move object functions to object-store.h
2:  3e547289408 = 2:  5a2cd6cca9c fsck tests: refactor one test to use a sub-repo
3:  74654a01ba3 = 3:  d0d9cb33315 fsck: don't hard die on invalid object types
-:  ----------- > 4:  81fffefcf99 object-store.h: move read_loose_object() below 'struct object_info'
4:  d23fb5cd039 ! 5:  5fb6ac4faee fsck: improve the error on invalid object types
    @@ Metadata
     Author: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
     
      ## Commit message ##
    -    fsck: improve the error on invalid object types
    +    fsck: report invalid types recorded in objects
     
         Continue the work in the preceding commit and improve the error on:
     
    @@ object-file.c: int read_loose_object(const char *path,
      			free(*contents);
     
      ## object-store.h ##
    -@@ object-store.h: int force_object_loose(const struct object_id *oid, time_t mtime);
    +@@ object-store.h: int oid_object_info_extended(struct repository *r,
    + 
    + /*
    +  * Open the loose object at path, check its hash, and return the contents,
    ++ * use the "oi" argument to assert things about the object, or e.g. populate its
    +  * type, and size. If the object is a blob, then "contents" may return NULL,
    +  * to allow streaming of large blobs.
       *
    -  * Returns 0 on success, negative on error (details may be written to stderr).
    +@@ object-store.h: int oid_object_info_extended(struct repository *r,
       */
    -+struct object_info;
      int read_loose_object(const char *path,
      		      const struct object_id *expected_oid,
     -		      enum object_type *type,
    @@ object-store.h: int force_object_loose(const struct object_id *oid, time_t mtime
     +		      struct object_info *oi,
      		      unsigned int oi_flags);
      
    - /* Retry packed storage after checking packed and loose storage */
    + /*
     
      ## t/t1450-fsck.sh ##
     @@ t/t1450-fsck.sh: test_expect_success 'object with hash mismatch' '
5:  bcec536b0f6 ! 6:  226d2031bcf fsck: improve error on loose object hash mismatch
    @@ Metadata
     Author: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
     
      ## Commit message ##
    +    fsck: report invalid object type-path combinations
    +
         fsck: improve error on loose object hash mismatch
     
         Improve the error that's emitted in cases where we find a loose object
    @@ object-file.c: int read_loose_object(const char *path,
     -			      oid_to_hex(expected_oid));
     +					   *contents, *size, oi->type_name->buf, real_oid)) {
     +			if (oideq(real_oid, &null_oid))
    -+				/*
    -+				 * Not a plain BUG() because if it
    -+				 * does happen we're in the middle of
    -+				 * an fsck we'd like to see to the
    -+				 * end.
    -+				 */
    -+				bug("BUG trying to compute hash for object at %s (expected %s)",
    -+				    path, oid_to_hex(expected_oid));
    ++				BUG("should only get OID mismatch errors with mapped contents");
      			free(*contents);
      			goto out;
      		}
     
      ## object-store.h ##
    -@@ object-store.h: int force_object_loose(const struct object_id *oid, time_t mtime);
    - struct object_info;
    +@@ object-store.h: int oid_object_info_extended(struct repository *r,
    +  */
      int read_loose_object(const char *path,
      		      const struct object_id *expected_oid,
     +		      struct object_id *real_oid,
-- 
2.31.1.645.g989d83ea6a6