@@ -50,3 +50,16 @@ config CEPH_FS_SECURITY_LABEL
If you are not using a security module that requires using
extended attributes for file security labels, say N.
+
+config CEPH_FS_DEBUG
+ bool "Ceph client debugging"
+ depends on CEPH_FS
+ default n
+ help
+ If you say Y here, this option enables additional pre-
condition
+ and post-condition checks in functions. Also it could enable
+ BUG_ON() instead of returning the error code. This option
could
+ save more messages in system log and execute additional
computation.
+
+ If you are going to debug the code, then chose Y here.
+ If unsure, say N.
b/include/linux/ceph/messenger.h
@@ -532,7 +532,7 @@ u32 ceph_get_global_seq(struct ceph_messenger
*msgr, u32 gt);
void ceph_con_discard_sent(struct ceph_connection *con, u64 ack_seq);
void ceph_con_discard_requeued(struct ceph_connection *con, u64
reconnect_seq);
-void ceph_msg_data_cursor_init(struct ceph_msg_data_cursor *cursor,
+int ceph_msg_data_cursor_init(struct ceph_msg_data_cursor *cursor,
struct ceph_msg *msg, size_t length);
struct page *ceph_msg_data_next(struct ceph_msg_data_cursor *cursor,
size_t *page_offset, size_t *length);
@@ -45,3 +45,16 @@ config CEPH_LIB_USE_DNS_RESOLVER
Documentation/networking/dns_resolver.rst
If unsure, say N.
+
+config CEPH_LIB_DEBUG
+ bool "Ceph core library debugging"
+ depends on CEPH_LIB
+ default n
+ help
+ If you say Y here, this option enables additional pre-
condition
+ and post-condition checks in functions. Also it could enable
+ BUG_ON() instead of returning the error code. This option
could
+ save more messages in system log and execute additional
computation.
+
+ If you are going to debug the code, then chose Y here.
+ If unsure, say N.
@@ -1063,18 +1063,30 @@ static void __ceph_msg_data_cursor_init(struct
ceph_msg_data_cursor *cursor)
cursor->need_crc = true;
}
-void ceph_msg_data_cursor_init(struct ceph_msg_data_cursor *cursor,
- struct ceph_msg *msg, size_t length)
+int ceph_msg_data_cursor_init(struct ceph_msg_data_cursor *cursor,
+ struct ceph_msg *msg, size_t length)
{
+#ifdef CONFIG_CEPH_LIB_DEBUG
BUG_ON(!length);
BUG_ON(length > msg->data_length);
BUG_ON(!msg->num_data_items);
+#else
+ if (!length)
+ return -EINVAL;
+
+ if (length > msg->data_length)
+ return -EINVAL;
+
+ if (!msg->num_data_items)
+ return -EINVAL;
+#endif /* CONFIG_CEPH_LIB_DEBUG */
cursor->total_resid = length;
cursor->data = msg->data;
cursor->sr_resid = 0;
__ceph_msg_data_cursor_init(cursor);
+ return 0;
}
/*
@@ -157,12 +157,12 @@ static size_t sizeof_footer(struct
ceph_connection *con)
sizeof(struct ceph_msg_footer_old);
}
-static void prepare_message_data(struct ceph_msg *msg, u32 data_len)
+static int prepare_message_data(struct ceph_msg *msg, u32 data_len)
{
/* Initialize data cursor if it's not a sparse read */
u64 len = msg->sparse_read_total ? : data_len;
- ceph_msg_data_cursor_init(&msg->cursor, msg, len);
+ return ceph_msg_data_cursor_init(&msg->cursor, msg, len);
}
/*
@@ -192,10 +192,11 @@ static void prepare_write_message_footer(struct
ceph_connection *con)
/*
* Prepare headers for the next outgoing message.
*/
-static void prepare_write_message(struct ceph_connection *con)
+static int prepare_write_message(struct ceph_connection *con)
{
struct ceph_msg *m;
u32 crc;
+ int ret;
con_out_kvec_reset(con);
con->v1.out_msg_done = false;
@@ -251,7 +252,10 @@ static void prepare_write_message(struct
ceph_connection *con)
/* is there a data payload? */
con->out_msg->footer.data_crc = 0;
if (m->data_length) {
- prepare_message_data(con->out_msg, m->data_length);
+ ret = prepare_message_data(con->out_msg, m-
>data_length);
+ if (ret)
+ return ret;
+
con->v1.out_more = 1; /* data + footer will follow */
} else {
/* no, queue up footer too and be done */
@@ -259,6 +263,7 @@ static void prepare_write_message(struct
ceph_connection *con)
}
ceph_con_flag_set(con, CEPH_CON_F_WRITE_PENDING);
+ return 0;
}
/*
@@ -1230,8 +1235,11 @@ static int read_partial_message(struct
ceph_connection *con)
/* prepare for data payload, if any */
- if (data_len)
- prepare_message_data(con->in_msg, data_len);
+ if (data_len) {
+ ret = prepare_message_data(con->in_msg,
data_len);
+ if (ret)
+ return ret;
+ }
}
/* front */
@@ -1546,8 +1554,11 @@ int ceph_con_v1_try_write(struct ceph_connection
*con)
}
/* is anything else pending? */
if (!list_empty(&con->out_queue)) {
- prepare_write_message(con);
- goto more;
+ ret = prepare_write_message(con);
+ if (ret)
+ goto out;
+ else
+ goto more;
}
if (con->in_seq > con->in_seq_acked) {
prepare_write_ack(con);
@@ -1026,7 +1026,10 @@ static int setup_message_sgs(struct sg_table
*sgt, struct ceph_msg *msg,
if (need_padding(dlen))
sg_cnt++;
} else {
- ceph_msg_data_cursor_init(&cursor, msg, dlen);
+ ret = ceph_msg_data_cursor_init(&cursor, msg,
dlen);
+ if (ret)
+ return ret;
+
sg_cnt += calc_sg_cnt_cursor(&cursor);
Hello, There are multiple cases of using BUG_ON() in the main logic of CephFS kernel code. For example, ceph_msg_data_cursor_init() is one of the example: void ceph_msg_data_cursor_init(struct ceph_msg_data_cursor *cursor, struct ceph_msg *msg, size_t length) { BUG_ON(!length); BUG_ON(length > msg->data_length); BUG_ON(!msg->num_data_items); <skipped> } Such approach is good for the case of debugging an issue. But it is not user friendly approach because returning and processing an error is more preferable than crashing the kernel. This patch introduces a special debug configuration option for CephFS subsystems with the goal of error processing in the case of release build and kernel crash in the case of debug build: if CONFIG_CEPH_LIB_DEBUG BUG_ON(); else return <error code>; endif Signed-off-by: Viacheslav Dubeyko <Slava.Dubeyko@ibm.com> --- fs/ceph/Kconfig | 13 +++++++++++ include/linux/ceph/messenger.h | 2 +- net/ceph/Kconfig | 13 +++++++++++ net/ceph/messenger.c | 16 +++++++++++-- net/ceph/messenger_v1.c | 27 +++++++++++++++------- net/ceph/messenger_v2.c | 41 +++++++++++++++++++++++++--------- 6 files changed, 90 insertions(+), 22 deletions(-) } } @@ -1046,7 +1049,10 @@ static int setup_message_sgs(struct sg_table *sgt, struct ceph_msg *msg, if (pages) { init_sgs_pages(&cur_sg, pages, dpos, dlen, data_pad); } else { - ceph_msg_data_cursor_init(&cursor, msg, dlen); + ret = ceph_msg_data_cursor_init(&cursor, msg, dlen); + if (ret) + return ret; + init_sgs_cursor(&cur_sg, &cursor, data_pad); } } @@ -1860,10 +1866,13 @@ static int prepare_read_control_remainder(struct ceph_connection *con) static int prepare_read_data(struct ceph_connection *con) { struct bio_vec bv; + int ret; con->in_data_crc = -1; - ceph_msg_data_cursor_init(&con->v2.in_cursor, con->in_msg, - data_len(con->in_msg)); + ret = ceph_msg_data_cursor_init(&con->v2.in_cursor, con- >in_msg, + data_len(con->in_msg)); + if (ret) + return ret; get_bvec_at(&con->v2.in_cursor, &bv); if (ceph_test_opt(from_msgr(con->msgr), RXBOUNCE)) { @@ -2025,6 +2034,7 @@ static int prepare_sparse_read_cont(struct ceph_connection *con) static int prepare_sparse_read_data(struct ceph_connection *con) { struct ceph_msg *msg = con->in_msg; + int ret; dout("%s: starting sparse read\n", __func__); @@ -2034,8 +2044,10 @@ static int prepare_sparse_read_data(struct ceph_connection *con) if (!con_secure(con)) con->in_data_crc = -1; - ceph_msg_data_cursor_init(&con->v2.in_cursor, msg, - msg->sparse_read_total); + ret = ceph_msg_data_cursor_init(&con->v2.in_cursor, msg, + msg->sparse_read_total); + if (ret) + return ret; reset_in_kvecs(con); con->v2.in_state = IN_S_PREPARE_SPARSE_DATA_CONT; @@ -3184,17 +3196,21 @@ int ceph_con_v2_try_read(struct ceph_connection *con) } } -static void queue_data(struct ceph_connection *con) +static int queue_data(struct ceph_connection *con) { struct bio_vec bv; + int ret; con->v2.out_epil.data_crc = -1; - ceph_msg_data_cursor_init(&con->v2.out_cursor, con->out_msg, - data_len(con->out_msg)); + ret = ceph_msg_data_cursor_init(&con->v2.out_cursor, con- >out_msg, + data_len(con->out_msg)); + if (ret) + return ret; get_bvec_at(&con->v2.out_cursor, &bv); set_out_bvec(con, &bv, true); con->v2.out_state = OUT_S_QUEUE_DATA_CONT; + return 0; } static void queue_data_cont(struct ceph_connection *con) @@ -3309,8 +3325,11 @@ static int populate_out_iter(struct ceph_connection *con) switch (con->v2.out_state) { case OUT_S_QUEUE_DATA: WARN_ON(!con->out_msg); - queue_data(con); - goto populated; + ret = queue_data(con); + if (ret) + return ret; + else + goto populated; case OUT_S_QUEUE_DATA_CONT: WARN_ON(!con->out_msg); queue_data_cont(con);