| Message ID | 604160e79cef94fd8e03fe025990c999bb795395.1669839849.git.jonathantanmy@google.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Don't lazy-fetch commits when parsing them | expand |
diff --git a/object-file.c b/object-file.c index 26290554bb..1cde477267 100644 --- a/object-file.c +++ b/object-file.c @@ -1621,7 +1621,7 @@ static int do_oid_object_info_extended(struct repository *r, rtype = packed_object_info(r, e.p, e.offset, oi); if (rtype < 0) { mark_bad_packed_object(e.p, real); - return do_oid_object_info_extended(r, real, oi, 0); + return do_oid_object_info_extended(r, oid, oi, flags); } else if (oi->whence == OI_PACKED) { oi->u.packed.offset = e.offset; oi->u.packed.pack = e.p;
When an object in do_oid_object_info_extended() is found in a packfile, but corrupt, that packfile entry is marked as bad and the read is retried. Currently, this is done by invoking the function again but with the replace target of the object and with no flags. This currently works, but will be clumsy when a later patch modifies this function to also return the "real" object being read (that is, the replace target). It does not make sense to pass a pointer in order to receive this information when no replace lookups are requested, which is exactly what the reinvocation does. Therefore, change this reinvocation to pass exactly the arguments which were originally passed. This also makes us forwards compatible with future flags that may change the behavior of this function. This does slow down the case when packfile corruption is detected, but that is expected to be a very rare case. Signed-off-by: Jonathan Tan <jonathantanmy@google.com> --- object-file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)