Skip to content

DAOS-18484 container: debug IV issue #17969

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wangshilong wants to merge 1 commit into
base: master
Choose a base branch
from
+110 −1
Draft

DAOS-18484 container: debug IV issue #17969

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
106 changes: 105 additions & 1 deletion src/container/container_iv.c
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
/**
* (C) Copyright 2019-2023 Intel Corporation.
* (C) Copyright 2025 Hewlett Packard Enterprise Development LP
* (C) Copyright 2025-2026 Hewlett Packard Enterprise Development LP
*
* SPDX-License-Identifier: BSD-2-Clause-Patent
*/
Expand Down Expand Up @@ -31,6 +31,34 @@ key2priv(struct ds_iv_key *iv_key)
return (struct cont_iv_key *)iv_key->key_buf;
}

static uint32_t
cont_iv_prop_set_nr(daos_prop_t *prop)
{
uint32_t i;
uint32_t nr = 0;

if (prop == NULL || prop->dpp_entries == NULL)
return 0;

for (i = 0; i < prop->dpp_nr; i++) {
if (daos_prop_is_set(&prop->dpp_entries[i]))
nr++;
}

return nr;
}

static void
cont_iv_prop_log_invalid(const char *tag, uuid_t pool_uuid, uuid_t cont_uuid, d_rank_t src_rank,
d_rank_t master_rank, uint64_t bits, d_sg_list_t *src, daos_prop_t *prop)
{
D_ERROR(DF_CONT ": %s src_rank=%u master_rank=%u bits=" DF_X64 " sgl_nr=%u len=" DF_U64
" prop=%p dpp_nr=%u set_nr=%u\n",
DP_CONT(pool_uuid, cont_uuid), tag, src_rank, master_rank, bits,
src ? src->sg_nr : 0, (src && src->sg_nr > 0) ? src->sg_iovs[0].iov_len : 0, prop,
prop ? prop->dpp_nr : 0, cont_iv_prop_set_nr(prop));
}

static uint32_t
cont_iv_snap_ent_size(int nr)
{
Expand Down Expand Up @@ -419,6 +447,11 @@ cont_iv_prop_ent_create(struct ds_iv_entry *entry, struct ds_iv_key *key)
if (rc)
D_GOTO(out, rc);

if (prop == NULL || prop->dpp_nr == 0 || cont_iv_prop_set_nr(prop) == 0)
cont_iv_prop_log_invalid("miss-create got suspicious prop", entry->ns->iv_pool_uuid,
civ_key->cont_uuid, key->rank, entry->ns->iv_master_rank,
0, NULL, prop);

entry_size = cont_iv_prop_ent_size(DAOS_ACL_MAX_ACE_LEN);
D_ALLOC(iv_entry, entry_size);
if (iv_entry == NULL)
Expand All @@ -428,6 +461,11 @@ cont_iv_prop_ent_create(struct ds_iv_entry *entry, struct ds_iv_key *key)

uuid_copy(iv_entry->cont_uuid, civ_key->cont_uuid);
cont_iv_prop_l2g(prop, &iv_entry->iv_prop);
if (iv_entry->iv_prop.cip_valid_bits == 0)
cont_iv_prop_log_invalid("miss-create encoded zero-bit IV_CONT_PROP",
entry->ns->iv_pool_uuid, civ_key->cont_uuid, key->rank,
entry->ns->iv_master_rank,
iv_entry->iv_prop.cip_valid_bits, NULL, prop);
d_iov_set(&val_iov, iv_entry, entry_size);
d_iov_set(&key_iov, &civ_key->cont_uuid, sizeof(civ_key->cont_uuid));

Expand Down Expand Up @@ -582,6 +620,11 @@ cont_iv_ent_fetch(struct ds_iv_entry *entry, struct ds_iv_key *key,
}

src_iv = val_iov.iov_buf;
if (entry->iv_class->iv_class_id == IV_CONT_PROP && src_iv->iv_prop.cip_valid_bits == 0)
cont_iv_prop_log_invalid("fetch hit zero-bit IV_CONT_PROP in local tree",
entry->ns->iv_pool_uuid, civ_key->cont_uuid, key->rank,
entry->ns->iv_master_rank, src_iv->iv_prop.cip_valid_bits,
NULL, NULL);
rc = cont_iv_ent_copy(entry, civ_key, dst, src_iv);
out:
return rc;
Expand Down Expand Up @@ -654,12 +697,35 @@ cont_iv_ent_update(struct ds_iv_entry *entry, struct ds_iv_key *key,
} else if (entry->iv_class->iv_class_id == IV_CONT_PROP) {
daos_prop_t *prop = NULL;

if (src->sg_nr == 0 || src->sg_iovs == NULL ||
src->sg_iovs[0].iov_buf == NULL ||
src->sg_iovs[0].iov_len <
offsetof(struct cont_iv_entry, iv_prop.cip_valid_bits) +
sizeof(civ_ent->iv_prop.cip_valid_bits)) {
cont_iv_prop_log_invalid(
"ent_update got short/empty IV_CONT_PROP payload",
entry->ns->iv_pool_uuid, civ_key->cont_uuid, key->rank,
entry->ns->iv_master_rank, 0, src, NULL);
}

if (civ_ent->iv_prop.cip_valid_bits == 0)
cont_iv_prop_log_invalid(
"ent_update got zero-bit IV_CONT_PROP payload",
entry->ns->iv_pool_uuid, civ_key->cont_uuid, key->rank,
entry->ns->iv_master_rank, civ_ent->iv_prop.cip_valid_bits, src,
NULL);

rc = cont_iv_prop_g2l(&civ_ent->iv_prop, &prop);
if (rc) {
D_ERROR("cont_iv_prop_g2l failed "DF_RC"\n",
DP_RC(rc));
D_GOTO(out, rc);
}
if (prop == NULL || prop->dpp_nr == 0)
cont_iv_prop_log_invalid(
"ent_update g2l produced empty prop", entry->ns->iv_pool_uuid,
civ_key->cont_uuid, key->rank, entry->ns->iv_master_rank,
civ_ent->iv_prop.cip_valid_bits, src, prop);

rc = ds_cont_tgt_prop_update(entry->ns->iv_pool_uuid,
civ_ent->cont_uuid, prop);
Expand Down Expand Up @@ -748,6 +814,34 @@ cont_iv_ent_refresh(struct ds_iv_entry *entry, struct ds_iv_key *key,
return ref_rc;
}

if (entry->iv_class->iv_class_id == IV_CONT_PROP) {
struct cont_iv_key *civ_key = key2priv(key);

if (src == NULL || src->sg_nr == 0 || src->sg_iovs == NULL ||
src->sg_iovs[0].iov_buf == NULL) {
cont_iv_prop_log_invalid(
"refresh got empty IV_CONT_PROP src", entry->ns->iv_pool_uuid,
civ_key->cont_uuid, key->rank, entry->ns->iv_master_rank, 0, src, NULL);
} else {
struct cont_iv_entry *civ_ent = src->sg_iovs[0].iov_buf;

if (src->sg_iovs[0].iov_len <
offsetof(struct cont_iv_entry, iv_prop.cip_valid_bits) +
sizeof(civ_ent->iv_prop.cip_valid_bits)) {
cont_iv_prop_log_invalid("refresh got short IV_CONT_PROP payload",
entry->ns->iv_pool_uuid,
civ_key->cont_uuid, key->rank,
entry->ns->iv_master_rank, 0, src, NULL);
} else if (civ_ent->iv_prop.cip_valid_bits == 0) {
cont_iv_prop_log_invalid(
"refresh got zero-bit IV_CONT_PROP payload",
entry->ns->iv_pool_uuid, civ_key->cont_uuid, key->rank,
entry->ns->iv_master_rank, civ_ent->iv_prop.cip_valid_bits, src,
NULL);
}
}
}

return cont_iv_ent_update(entry, key, src, priv);
}

Expand Down Expand Up @@ -1592,12 +1686,22 @@ cont_iv_prop_fetch_ult(void *data)
D_GOTO(out, rc);
}

if (iv_entry->iv_prop.cip_valid_bits == 0)
cont_iv_prop_log_invalid("cont_iv_fetch returned zero-bit IV_CONT_PROP",
pool->sp_uuid, arg->cont_uuid, 0,
pool->sp_iv_ns->iv_master_rank,
iv_entry->iv_prop.cip_valid_bits, NULL, NULL);

rc = cont_iv_prop_g2l(&iv_entry->iv_prop, &prop_fetch);
if (rc) {
DL_ERROR(rc, DF_CONT ": cont_iv_prop_g2l failed",
DP_CONT(pool->sp_uuid, arg->cont_uuid));
D_GOTO(out, rc);
}
if (prop_fetch == NULL || prop_fetch->dpp_nr == 0)
cont_iv_prop_log_invalid("fetch g2l produced empty prop", pool->sp_uuid,
arg->cont_uuid, 0, pool->sp_iv_ns->iv_master_rank,
iv_entry->iv_prop.cip_valid_bits, NULL, prop_fetch);

rc = daos_prop_copy(prop, prop_fetch);
if (rc) {
Expand Down
5 changes: 5 additions & 0 deletions src/container/srv_container.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6420,6 +6420,11 @@ ds_cont_get_prop(uuid_t pool_uuid, uuid_t cont_uuid, daos_prop_t **prop_out)
if (rc != 0)
D_GOTO(out_lock, rc);

if (prop == NULL || prop->dpp_nr == 0) {
D_ERROR(DF_CONT ": ds_cont_get_prop returned empty prop ptr=%p\n",
DP_CONT(pool_uuid, cont_uuid), prop);
}

D_ASSERT(prop != NULL);
D_ASSERT(prop->dpp_nr <= CONT_PROP_NUM);

Expand Down
Loading