LU-5871 lod: Do not return EAGAIN in lod_object_init

Convert EAGAIN to EIO if fld_client_rpc() fails in
lod_object_init(), otherwise it will confuse
lu_object_find_at(), and make it wait there for no
reason, which should only wait if the object is dying.
See call chain lu_object_find_at()-> lu_object_find_try()
->lu_object_alloc()->lod_object_init()->lod_fld_lookup()
->fld_client_rpc(), and even worse waitq is not being
initialized yet when the failure happened here.

Change-Id: Ieae434b34c239efea86a4a471fb01e397336a31c
Signed-off-by: Wang Di <di.wang@intel.com>
Reviewed-on: http://review.whamcloud.com/12586
Tested-by: Jenkins
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>

diff --git a/lustre/lod/lod_object.c b/lustre/lod/lod_object.c
index 8c2c8d1..4cb4d06 100644 (file)
--- a/lustre/lod/lod_object.c
+++ b/lustre/lod/lod_object.c
@@ -3984,8 +3984,16 @@ static int lod_object_init(const struct lu_env *env, struct lu_object *lo,
        ENTRY;
 
        rc = lod_fld_lookup(env, lod, lu_object_fid(lo), &idx, &type);
-       if (rc != 0)
+       if (rc != 0) {
+               /* Note: Sometimes, it will Return EAGAIN here, see
+                * ptrlpc_import_delay_req(), which might confuse
+                * lu_object_find_at() and make it wait there incorrectly.
+                * so we convert it to EIO here.*/
+               if (rc == -EAGAIN)
+                       rc = -EIO;
+
                RETURN(rc);
+       }
 
        if (type == LU_SEQ_RANGE_MDT &&
            idx == lu_site2seq(lo->lo_dev->ld_site)->ss_node_id) {