LU-18435 lod: recover layout generation from replay

The offset of the layout generation is different between struct
lov_mds_md_v1/v3.lmm_layout_gen and lov_comp_md.lcm_layout_gen.
When checking/setting layout gen, we must use layout-specific field.

Otherwise layout generation can be set to 0 (or other random value)
after replay and client can't apply new layout during later update.

Fixes: 13557aa86904 ("LU-15300 mdt: refresh LOVEA with LL granted")
Signed-off-by: Alex Zhuravlev <bzzz@whamcloud.com>
Change-Id: I5e4a63cd097d157317e0e8d1a0fca4a46817d118
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/56950
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Reviewed-by: Qian Yingjin <qian@ddn.com>
Reviewed-by: Alexey Lyashkov <alexey.lyashkov@hpe.com>
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>

diff --git a/lustre/lod/lod_qos.c b/lustre/lod/lod_qos.c
index aa8c20b..3290642 100644 (file)
--- a/lustre/lod/lod_qos.c
+++ b/lustre/lod/lod_qos.c
@@ -2174,7 +2174,14 @@ int lod_use_defined_striping(const struct lu_env *env,
            magic != LOV_MAGIC_COMP_V1 && magic != LOV_MAGIC_FOREIGN)
                GOTO(unlock, rc = -EINVAL);
 
-       if (magic == LOV_MAGIC_COMP_V1) {
+       /* layout generation must be preserved, otherwise client won't
+        * be able to refresh it having a more recent generation */
+       if (magic == LOV_MAGIC_V1 || magic == LOV_MAGIC_V3) {
+               mo->ldo_is_composite = 0;
+               comp_cnt = 1;
+               mirror_cnt = 0;
+               mo->ldo_layout_gen = le16_to_cpu(v1->lmm_layout_gen);
+       } else if (magic == LOV_MAGIC_COMP_V1) {
                comp_v1 = buf->lb_buf;
                comp_cnt = le16_to_cpu(comp_v1->lcm_entry_count);
                if (comp_cnt == 0)
@@ -2183,6 +2190,7 @@ int lod_use_defined_striping(const struct lu_env *env,
                mo->ldo_flr_state = le16_to_cpu(comp_v1->lcm_flags) &
                                        LCM_FL_FLR_MASK;
                mo->ldo_is_composite = 1;
+               mo->ldo_layout_gen = le32_to_cpu(comp_v1->lcm_layout_gen);
        } else if (magic == LOV_MAGIC_FOREIGN) {
                struct lov_foreign_md *foreign;
                size_t length;
@@ -2210,11 +2218,8 @@ int lod_use_defined_striping(const struct lu_env *env,
                memcpy(mo->ldo_foreign_lov, buf->lb_buf, length);
                GOTO(out, rc);
        } else {
-               mo->ldo_is_composite = 0;
-               comp_cnt = 1;
-               mirror_cnt = 0;
+               GOTO(out, rc = -EINVAL);
        }
-       mo->ldo_layout_gen = le16_to_cpu(v1->lmm_layout_gen);
 
        rc = lod_alloc_comp_entries(mo, mirror_cnt, comp_cnt);
        if (rc)

diff --git a/lustre/tests/replay-single.sh b/lustre/tests/replay-single.sh
index a3212e4..3695f2b 100755 (executable)
--- a/lustre/tests/replay-single.sh
+++ b/lustre/tests/replay-single.sh
@@ -5342,6 +5342,28 @@ test_201() {
 }
 run_test 201 "MDT umount cascading disconnects timeouts"
 
+test_202() {
+       local td=$DIR/$tdir
+       local tf=$td/$tfile
+
+       (( $MDS1_VERSION >= $(version_code 2.16.0) )) ||
+          (( $MDS1_VERSION < $(version_code v2_15_55-64-g13557aa869) &&
+          $MDS1_VERSION >= $(version_code 2.14.0-ddn178) )) ||
+          (( $MDS1_VERSION < $(version_code 2.14.0-ddn87-14-gf1bd967799) )) ||
+               skip "need MDS with LU-18416 fix for layout version"
+
+       mkdir_on_mdt0 $td || error "can't mkdir"
+       $LFS setstripe -E128M -c1 -Eeof -c2 $td || error "can't setstripe"
+       replay_barrier mds1
+       touch $tf
+       local before=$($LFS getstripe -v $tf|awk '/lcm_layout_gen:/{print $2}')
+       fail mds1
+       cancel_lru_locks mdc
+       local after=$($LFS getstripe -v $tf|awk '/lcm_layout_gen:/{print $2}')
+       (( $before == $after )) || error "layout gen changed: $before -> $after"
+}
+run_test 202 "pfl replay should recovery layout generation"
+
 
 complete_test $SECONDS
 check_and_cleanup_lustre