Cleanup: Move lttng-modules instrumentation headers
[lttng-modules.git] / include / instrumentation / events / block.h
diff --git a/include/instrumentation/events/block.h b/include/instrumentation/events/block.h
new file mode 100644 (file)
index 0000000..3b087d6
--- /dev/null
@@ -0,0 +1,1054 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM block
+
+#if !defined(LTTNG_TRACE_BLOCK_H) || defined(TRACE_HEADER_MULTI_READ)
+#define LTTNG_TRACE_BLOCK_H
+
+#include <lttng/tracepoint-event.h>
+#include <linux/blktrace_api.h>
+#include <linux/blkdev.h>
+#include <linux/trace_seq.h>
+#include <linux/version.h>
+
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0))
+#include <scsi/scsi_request.h>
+#endif /* (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0)) */
+
+#ifndef _TRACE_BLOCK_DEF_
+#define _TRACE_BLOCK_DEF_
+
+enum {
+       RWBS_FLAG_WRITE         = (1 << 0),
+       RWBS_FLAG_DISCARD       = (1 << 1),
+       RWBS_FLAG_READ          = (1 << 2),
+       RWBS_FLAG_RAHEAD        = (1 << 3),
+       RWBS_FLAG_BARRIER       = (1 << 4),
+       RWBS_FLAG_SYNC          = (1 << 5),
+       RWBS_FLAG_META          = (1 << 6),
+       RWBS_FLAG_SECURE        = (1 << 7),
+       RWBS_FLAG_FLUSH         = (1 << 8),
+       RWBS_FLAG_FUA           = (1 << 9),
+       RWBS_FLAG_PREFLUSH      = (1 << 10),
+};
+
+#endif /* _TRACE_BLOCK_DEF_ */
+
+LTTNG_TRACEPOINT_ENUM(block_rq_type,
+       TP_ENUM_VALUES(
+               ctf_enum_value("RWBS_FLAG_WRITE", RWBS_FLAG_WRITE)
+               ctf_enum_value("RWBS_FLAG_DISCARD", RWBS_FLAG_DISCARD)
+               ctf_enum_value("RWBS_FLAG_READ", RWBS_FLAG_READ)
+               ctf_enum_value("RWBS_FLAG_RAHEAD", RWBS_FLAG_RAHEAD)
+               ctf_enum_value("RWBS_FLAG_BARRIER", RWBS_FLAG_BARRIER)
+               ctf_enum_value("RWBS_FLAG_SYNC", RWBS_FLAG_SYNC)
+               ctf_enum_value("RWBS_FLAG_META", RWBS_FLAG_META)
+               ctf_enum_value("RWBS_FLAG_SECURE", RWBS_FLAG_SECURE)
+               ctf_enum_value("RWBS_FLAG_FLUSH", RWBS_FLAG_FLUSH)
+               ctf_enum_value("RWBS_FLAG_FUA", RWBS_FLAG_FUA)
+               ctf_enum_value("RWBS_FLAG_PREFLUSH", RWBS_FLAG_PREFLUSH)
+       )
+)
+
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,8,0) || \
+       LTTNG_SLE_KERNEL_RANGE(4,4,73,5,0,0, 4,4,73,6,0,0) || \
+       LTTNG_SLE_KERNEL_RANGE(4,4,82,6,0,0, 4,4,82,7,0,0) || \
+       LTTNG_SLE_KERNEL_RANGE(4,4,92,6,0,0, 4,4,92,7,0,0) || \
+       LTTNG_SLE_KERNEL_RANGE(4,4,103,6,0,0, 4,4,103,7,0,0) || \
+       LTTNG_SLE_KERNEL_RANGE(4,4,114,94,0,0, 4,4,114,95,0,0) || \
+       LTTNG_SLE_KERNEL_RANGE(4,4,120,94,0,0, 4,4,120,95,0,0) || \
+       LTTNG_SLE_KERNEL_RANGE(4,4,126,94,0,0, 4,5,0,0,0,0))
+
+#define lttng_req_op(rq)       req_op(rq)
+#define lttng_req_rw(rq)       ((rq)->cmd_flags)
+#define lttng_bio_op(bio)      bio_op(bio)
+#define lttng_bio_rw(bio)      ((bio)->bi_opf)
+
+#define blk_rwbs_ctf_integer(type, rwbs, op, rw, bytes)                              \
+               ctf_enum(block_rq_type, type, rwbs,                                           \
+                       (((op) == REQ_OP_WRITE || (op) == REQ_OP_WRITE_SAME) ? RWBS_FLAG_WRITE : \
+                       ( (op) == REQ_OP_DISCARD ? RWBS_FLAG_DISCARD :        \
+                       ( (op) == REQ_OP_SECURE_ERASE ? (RWBS_FLAG_DISCARD | RWBS_FLAG_SECURE) : \
+                       ( (op) == REQ_OP_FLUSH ? RWBS_FLAG_FLUSH :            \
+                       ( (op) == REQ_OP_READ ? RWBS_FLAG_READ :              \
+                       ( 0 ))))))                                            \
+                       | ((rw) & REQ_RAHEAD ? RWBS_FLAG_RAHEAD : 0)          \
+                       | ((rw) & REQ_SYNC ? RWBS_FLAG_SYNC : 0)              \
+                       | ((rw) & REQ_META ? RWBS_FLAG_META : 0)              \
+                       | ((rw) & REQ_PREFLUSH ? RWBS_FLAG_PREFLUSH : 0)      \
+                       | ((rw) & REQ_FUA ? RWBS_FLAG_FUA : 0))
+
+#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,1,0))
+
+#define lttng_req_op(rq)
+#define lttng_req_rw(rq)       ((rq)->cmd_flags)
+#define lttng_bio_op(bio)
+#define lttng_bio_rw(bio)      ((bio)->bi_rw)
+
+#define blk_rwbs_ctf_integer(type, rwbs, op, rw, bytes)                              \
+               ctf_enum(block_rq_type, type, rwbs, ((rw) & WRITE ? RWBS_FLAG_WRITE : \
+                       ( (rw) & REQ_DISCARD ? RWBS_FLAG_DISCARD :            \
+                       ( (bytes) ? RWBS_FLAG_READ :                          \
+                       ( 0 ))))                                              \
+                       | ((rw) & REQ_RAHEAD ? RWBS_FLAG_RAHEAD : 0)          \
+                       | ((rw) & REQ_SYNC ? RWBS_FLAG_SYNC : 0)              \
+                       | ((rw) & REQ_META ? RWBS_FLAG_META : 0)              \
+                       | ((rw) & REQ_SECURE ? RWBS_FLAG_SECURE : 0)          \
+                       | ((rw) & REQ_FLUSH ? RWBS_FLAG_FLUSH : 0)            \
+                       | ((rw) & REQ_FUA ? RWBS_FLAG_FUA : 0))
+
+#else
+
+#define lttng_req_op(rq)
+#define lttng_req_rw(rq)       ((rq)->cmd_flags)
+#define lttng_bio_op(bio)
+#define lttng_bio_rw(bio)      ((bio)->bi_rw)
+
+#define blk_rwbs_ctf_integer(type, rwbs, op, rw, bytes)                              \
+               ctf_enum(block_rq_type, type, rwbs, ((rw) & WRITE ? RWBS_FLAG_WRITE :     \
+                       ( (rw) & REQ_DISCARD ? RWBS_FLAG_DISCARD :            \
+                       ( (bytes) ? RWBS_FLAG_READ :                          \
+                       ( 0 ))))                                              \
+                       | ((rw) & REQ_RAHEAD ? RWBS_FLAG_RAHEAD : 0)          \
+                       | ((rw) & REQ_SYNC ? RWBS_FLAG_SYNC : 0)              \
+                       | ((rw) & REQ_META ? RWBS_FLAG_META : 0)              \
+                       | ((rw) & REQ_SECURE ? RWBS_FLAG_SECURE : 0))
+
+#endif
+
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,9,0))
+LTTNG_TRACEPOINT_EVENT_CLASS(block_buffer,
+
+       TP_PROTO(struct buffer_head *bh),
+
+       TP_ARGS(bh),
+
+       TP_FIELDS (
+               ctf_integer(dev_t, dev, bh->b_bdev->bd_dev)
+               ctf_integer(sector_t, sector, bh->b_blocknr)
+               ctf_integer(size_t, size, bh->b_size)
+       )
+)
+
+/**
+ * block_touch_buffer - mark a buffer accessed
+ * @bh: buffer_head being touched
+ *
+ * Called from touch_buffer().
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_buffer, block_touch_buffer,
+
+       TP_PROTO(struct buffer_head *bh),
+
+       TP_ARGS(bh)
+)
+
+/**
+ * block_dirty_buffer - mark a buffer dirty
+ * @bh: buffer_head being dirtied
+ *
+ * Called from mark_buffer_dirty().
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_buffer, block_dirty_buffer,
+
+       TP_PROTO(struct buffer_head *bh),
+
+       TP_ARGS(bh)
+)
+#endif
+
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,12,0))
+/* block_rq_with_error event class removed in kernel 4.12 */
+#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0))
+LTTNG_TRACEPOINT_EVENT_CLASS_CODE(block_rq_with_error,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq),
+
+       TP_locvar(
+               sector_t sector;
+               unsigned int nr_sector;
+               unsigned char *cmd;
+               size_t cmd_len;
+       ),
+
+       TP_code_pre(
+               if (blk_rq_is_scsi(rq)) {
+                       struct scsi_request *scsi_rq = scsi_req(rq);
+                       tp_locvar->sector = 0;
+                       tp_locvar->nr_sector = 0;
+                       tp_locvar->cmd = scsi_rq->cmd;
+                       tp_locvar->cmd_len = scsi_rq->cmd_len;
+               } else {
+                       tp_locvar->sector = blk_rq_pos(rq);
+                       tp_locvar->nr_sector = blk_rq_sectors(rq);
+                       tp_locvar->cmd = NULL;
+                       tp_locvar->cmd_len = 0;
+               }
+       ),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, tp_locvar->sector)
+               ctf_integer(unsigned int, nr_sector, tp_locvar->nr_sector)
+               ctf_integer(int, errors, rq->errors)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), blk_rq_bytes(rq))
+               ctf_sequence_hex(unsigned char, cmd,
+                       tp_locvar->cmd, size_t, tp_locvar->cmd_len)
+       ),
+
+       TP_code_post()
+)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0)) */
+LTTNG_TRACEPOINT_EVENT_CLASS_CODE(block_rq_with_error,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq),
+
+       TP_locvar(
+               sector_t sector;
+               unsigned int nr_sector;
+               unsigned char *cmd;
+               size_t cmd_len;
+       ),
+
+       TP_code_pre(
+
+               if (rq->cmd_type == REQ_TYPE_BLOCK_PC) {
+                       tp_locvar->sector = 0;
+                       tp_locvar->nr_sector = 0;
+                       tp_locvar->cmd = rq->cmd;
+                       tp_locvar->cmd_len = rq->cmd_len;
+               } else {
+                       tp_locvar->sector = blk_rq_pos(rq);
+                       tp_locvar->nr_sector = blk_rq_sectors(rq);
+                       tp_locvar->cmd = NULL;
+                       tp_locvar->cmd_len = 0;
+               }
+       ),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, tp_locvar->sector)
+               ctf_integer(unsigned int, nr_sector, tp_locvar->nr_sector)
+               ctf_integer(int, errors, rq->errors)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), blk_rq_bytes(rq))
+               ctf_sequence_hex(unsigned char, cmd,
+                       tp_locvar->cmd, size_t, tp_locvar->cmd_len)
+       ),
+
+       TP_code_post()
+)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0)) */
+
+#if (LINUX_VERSION_CODE < KERNEL_VERSION(4,12,0))
+/**
+ * block_rq_abort - abort block operation request
+ * @q: queue containing the block operation request
+ * @rq: block IO operation request
+ *
+ * Called immediately after pending block IO operation request @rq in
+ * queue @q is aborted. The fields in the operation request @rq
+ * can be examined to determine which device and sectors the pending
+ * operation would access.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_rq_with_error, block_rq_abort,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq)
+)
+#endif
+
+/**
+ * block_rq_requeue - place block IO request back on a queue
+ * @q: queue holding operation
+ * @rq: block IO operation request
+ *
+ * The block operation request @rq is being placed back into queue
+ * @q.  For some reason the request was not completed and needs to be
+ * put back in the queue.
+ */
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,12,0))
+LTTNG_TRACEPOINT_EVENT(block_rq_requeue,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, blk_rq_trace_sector(rq))
+               ctf_integer(unsigned int, nr_sector, blk_rq_trace_nr_sectors(rq))
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), blk_rq_bytes(rq))
+       )
+)
+#else
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_rq_with_error, block_rq_requeue,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq)
+)
+#endif
+
+/**
+ * block_rq_complete - block IO operation completed by device driver
+ * @q: queue containing the block operation request
+ * @rq: block operations request
+ * @nr_bytes: number of completed bytes
+ *
+ * The block_rq_complete tracepoint event indicates that some portion
+ * of operation request has been completed by the device driver.  If
+ * the @rq->bio is %NULL, then there is absolutely no additional work to
+ * do for the request. If @rq->bio is non-NULL then there is
+ * additional work required to complete the request.
+ */
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,12,0))
+LTTNG_TRACEPOINT_EVENT(block_rq_complete,
+
+       TP_PROTO(struct request *rq, int error, unsigned int nr_bytes),
+
+       TP_ARGS(rq, error, nr_bytes),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, blk_rq_pos(rq))
+               ctf_integer(unsigned int, nr_sector, nr_bytes >> 9)
+               ctf_integer(int, error, error)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), nr_bytes)
+       )
+)
+#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0))
+LTTNG_TRACEPOINT_EVENT_CODE(block_rq_complete,
+
+       TP_PROTO(struct request_queue *q, struct request *rq,
+                unsigned int nr_bytes),
+
+       TP_ARGS(q, rq, nr_bytes),
+
+       TP_locvar(
+               unsigned char *cmd;
+               size_t cmd_len;
+       ),
+
+       TP_code_pre(
+               if (blk_rq_is_scsi(rq)) {
+                       struct scsi_request *scsi_rq = scsi_req(rq);
+                       tp_locvar->cmd = scsi_rq->cmd;
+                       tp_locvar->cmd_len = scsi_rq->cmd_len;
+               } else {
+                       tp_locvar->cmd = NULL;
+                       tp_locvar->cmd_len = 0;
+               }
+       ),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, blk_rq_pos(rq))
+               ctf_integer(unsigned int, nr_sector, nr_bytes >> 9)
+               ctf_integer(int, errors, rq->errors)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), nr_bytes)
+               ctf_sequence_hex(unsigned char, cmd,
+                       tp_locvar->cmd, size_t, tp_locvar->cmd_len)
+       ),
+
+       TP_code_post()
+)
+#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,5)    \
+       || LTTNG_KERNEL_RANGE(3,12,21, 3,13,0)          \
+       || LTTNG_KERNEL_RANGE(3,10,41, 3,11,0)          \
+       || LTTNG_KERNEL_RANGE(3,4,91, 3,5,0)            \
+       || LTTNG_KERNEL_RANGE(3,2,58, 3,3,0)            \
+       || LTTNG_UBUNTU_KERNEL_RANGE(3,13,11,28, 3,14,0,0)      \
+       || LTTNG_RHEL_KERNEL_RANGE(3,10,0,229,0,0, 3,11,0,0,0,0))
+
+LTTNG_TRACEPOINT_EVENT_CODE(block_rq_complete,
+
+       TP_PROTO(struct request_queue *q, struct request *rq,
+                unsigned int nr_bytes),
+
+       TP_ARGS(q, rq, nr_bytes),
+
+       TP_locvar(
+               unsigned char *cmd;
+               size_t cmd_len;
+       ),
+
+       TP_code_pre(
+               if (rq->cmd_type == REQ_TYPE_BLOCK_PC) {
+                       tp_locvar->cmd = rq->cmd;
+                       tp_locvar->cmd_len = rq->cmd_len;
+               } else {
+                       tp_locvar->cmd = NULL;
+                       tp_locvar->cmd_len = 0;
+               }
+       ),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, blk_rq_pos(rq))
+               ctf_integer(unsigned int, nr_sector, nr_bytes >> 9)
+               ctf_integer(int, errors, rq->errors)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), nr_bytes)
+               ctf_sequence_hex(unsigned char, cmd,
+                       tp_locvar->cmd, size_t, tp_locvar->cmd_len)
+       ),
+
+       TP_code_post()
+)
+
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,15,0)) */
+
+/**
+ * block_rq_complete - block IO operation completed by device driver
+ * @q: queue containing the block operation request
+ * @rq: block operations request
+ *
+ * The block_rq_complete tracepoint event indicates that some portion
+ * of operation request has been completed by the device driver.  If
+ * the @rq->bio is %NULL, then there is absolutely no additional work to
+ * do for the request. If @rq->bio is non-NULL then there is
+ * additional work required to complete the request.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_rq_with_error, block_rq_complete,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq)
+)
+
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,15,0)) */
+
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,12,0))
+LTTNG_TRACEPOINT_EVENT_CLASS(block_rq,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, blk_rq_trace_sector(rq))
+               ctf_integer(unsigned int, nr_sector, blk_rq_trace_nr_sectors(rq))
+               ctf_integer(unsigned int, bytes, blk_rq_bytes(rq))
+               ctf_integer(pid_t, tid, current->pid)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), blk_rq_bytes(rq))
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       )
+)
+#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0))
+LTTNG_TRACEPOINT_EVENT_CLASS_CODE(block_rq,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq),
+
+       TP_locvar(
+               sector_t sector;
+               unsigned int nr_sector;
+               unsigned int bytes;
+               unsigned char *cmd;
+               size_t cmd_len;
+       ),
+
+       TP_code_pre(
+               if (blk_rq_is_scsi(rq)) {
+                       struct scsi_request *scsi_rq = scsi_req(rq);
+                       tp_locvar->sector = 0;
+                       tp_locvar->nr_sector = 0;
+                       tp_locvar->bytes = scsi_rq->resid_len;
+                       tp_locvar->cmd = scsi_rq->cmd;
+                       tp_locvar->cmd_len = scsi_rq->cmd_len;
+               } else {
+                       tp_locvar->sector = blk_rq_pos(rq);
+                       tp_locvar->nr_sector = blk_rq_sectors(rq);
+                       tp_locvar->bytes = 0;
+                       tp_locvar->cmd = NULL;
+                       tp_locvar->cmd_len = 0;
+               }
+       ),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, tp_locvar->sector)
+               ctf_integer(unsigned int, nr_sector, tp_locvar->nr_sector)
+               ctf_integer(unsigned int, bytes, tp_locvar->bytes)
+               ctf_integer(pid_t, tid, current->pid)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), blk_rq_bytes(rq))
+               ctf_sequence_hex(unsigned char, cmd,
+                       tp_locvar->cmd, size_t, tp_locvar->cmd_len)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       ),
+
+       TP_code_post()
+)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0)) */
+LTTNG_TRACEPOINT_EVENT_CLASS_CODE(block_rq,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq),
+
+       TP_locvar(
+               sector_t sector;
+               unsigned int nr_sector;
+               unsigned int bytes;
+               unsigned char *cmd;
+               size_t cmd_len;
+       ),
+
+       TP_code_pre(
+               if (rq->cmd_type == REQ_TYPE_BLOCK_PC) {
+                       tp_locvar->sector = 0;
+                       tp_locvar->nr_sector = 0;
+                       tp_locvar->bytes = blk_rq_bytes(rq);
+                       tp_locvar->cmd = rq->cmd;
+                       tp_locvar->cmd_len = rq->cmd_len;
+               } else {
+                       tp_locvar->sector = blk_rq_pos(rq);
+                       tp_locvar->nr_sector = blk_rq_sectors(rq);
+                       tp_locvar->bytes = 0;
+                       tp_locvar->cmd = NULL;
+                       tp_locvar->cmd_len = 0;
+               }
+       ),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev,
+                       rq->rq_disk ? disk_devt(rq->rq_disk) : 0)
+               ctf_integer(sector_t, sector, tp_locvar->sector)
+               ctf_integer(unsigned int, nr_sector, tp_locvar->nr_sector)
+               ctf_integer(unsigned int, bytes, tp_locvar->bytes)
+               ctf_integer(pid_t, tid, current->pid)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), blk_rq_bytes(rq))
+               ctf_sequence_hex(unsigned char, cmd,
+                       tp_locvar->cmd, size_t, tp_locvar->cmd_len)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       ),
+
+       TP_code_post()
+)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,11,0)) */
+
+/**
+ * block_rq_insert - insert block operation request into queue
+ * @q: target queue
+ * @rq: block IO operation request
+ *
+ * Called immediately before block operation request @rq is inserted
+ * into queue @q.  The fields in the operation request @rq struct can
+ * be examined to determine which device and sectors the pending
+ * operation would access.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_rq, block_rq_insert,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq)
+)
+
+/**
+ * block_rq_issue - issue pending block IO request operation to device driver
+ * @q: queue holding operation
+ * @rq: block IO operation operation request
+ *
+ * Called when block operation request @rq from queue @q is sent to a
+ * device driver for processing.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_rq, block_rq_issue,
+
+       TP_PROTO(struct request_queue *q, struct request *rq),
+
+       TP_ARGS(q, rq)
+)
+
+/**
+ * block_bio_bounce - used bounce buffer when processing block operation
+ * @q: queue holding the block operation
+ * @bio: block operation
+ *
+ * A bounce buffer was used to handle the block operation @bio in @q.
+ * This occurs when hardware limitations prevent a direct transfer of
+ * data between the @bio data memory area and the IO device.  Use of a
+ * bounce buffer requires extra copying of data and decreases
+ * performance.
+ */
+LTTNG_TRACEPOINT_EVENT(block_bio_bounce,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio),
+
+       TP_ARGS(q, bio),
+
+       TP_FIELDS(
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,14,0))
+               ctf_integer(dev_t, dev, bio_dev(bio))
+#else
+               ctf_integer(dev_t, dev, bio->bi_bdev ? bio->bi_bdev->bd_dev : 0)
+#endif
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0))
+               ctf_integer(sector_t, sector, bio->bi_iter.bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio_sectors(bio))
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio),
+                       bio->bi_iter.bi_size)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(sector_t, sector, bio->bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio->bi_size >> 9)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio),
+                       bio->bi_size)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(pid_t, tid, current->pid)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       )
+)
+
+/**
+ * block_bio_complete - completed all work on the block operation
+ * @q: queue holding the block operation
+ * @bio: block operation completed
+ * @error: io error value
+ *
+ * This tracepoint indicates there is no further work to do on this
+ * block IO operation @bio.
+ */
+LTTNG_TRACEPOINT_EVENT(block_bio_complete,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio, int error),
+
+       TP_ARGS(q, bio, error),
+
+       TP_FIELDS(
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,14,0))
+               ctf_integer(dev_t, dev, bio_dev(bio))
+#else
+               ctf_integer(dev_t, dev, bio->bi_bdev->bd_dev)
+#endif
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0))
+               ctf_integer(sector_t, sector, bio->bi_iter.bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio_sectors(bio))
+               ctf_integer(int, error, error)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio),
+                       bio->bi_iter.bi_size)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(sector_t, sector, bio->bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio->bi_size >> 9)
+               ctf_integer(int, error, error)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio), bio->bi_size)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+       )
+)
+
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,9,0))
+LTTNG_TRACEPOINT_EVENT_CLASS(block_bio_merge,
+
+       TP_PROTO(struct request_queue *q, struct request *rq, struct bio *bio),
+
+       TP_ARGS(q, rq, bio),
+
+       TP_FIELDS(
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,14,0))
+               ctf_integer(dev_t, dev, bio_dev(bio))
+#else
+               ctf_integer(dev_t, dev, bio->bi_bdev->bd_dev)
+#endif
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0))
+               ctf_integer(sector_t, sector, bio->bi_iter.bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio_sectors(bio))
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio),
+                       bio->bi_iter.bi_size)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(sector_t, sector, bio->bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio->bi_size >> 9)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio), bio->bi_size)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(pid_t, tid, current->pid)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       )
+)
+
+/**
+ * block_bio_backmerge - merging block operation to the end of an existing operation
+ * @q: queue holding operation
+ * @bio: new block operation to merge
+ *
+ * Merging block request @bio to the end of an existing block request
+ * in queue @q.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_bio_merge, block_bio_backmerge,
+
+       TP_PROTO(struct request_queue *q, struct request *rq, struct bio *bio),
+
+       TP_ARGS(q, rq, bio)
+)
+
+/**
+ * block_bio_frontmerge - merging block operation to the beginning of an existing operation
+ * @q: queue holding operation
+ * @bio: new block operation to merge
+ *
+ * Merging block IO operation @bio to the beginning of an existing block
+ * operation in queue @q.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_bio_merge, block_bio_frontmerge,
+
+       TP_PROTO(struct request_queue *q, struct request *rq, struct bio *bio),
+
+       TP_ARGS(q, rq, bio)
+)
+
+/**
+ * block_bio_queue - putting new block IO operation in queue
+ * @q: queue holding operation
+ * @bio: new block operation
+ *
+ * About to place the block IO operation @bio into queue @q.
+ */
+LTTNG_TRACEPOINT_EVENT(block_bio_queue,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio),
+
+       TP_ARGS(q, bio),
+
+       TP_FIELDS(
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,14,0))
+               ctf_integer(dev_t, dev, bio_dev(bio))
+#else
+               ctf_integer(dev_t, dev, bio->bi_bdev->bd_dev)
+#endif
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0))
+               ctf_integer(sector_t, sector, bio->bi_iter.bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio_sectors(bio))
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio),
+                       bio->bi_iter.bi_size)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(sector_t, sector, bio->bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio->bi_size >> 9)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio), bio->bi_size)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(pid_t, tid, current->pid)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       )
+)
+#else /* if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,9,0)) */
+LTTNG_TRACEPOINT_EVENT_CLASS(block_bio,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio),
+
+       TP_ARGS(q, bio),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev, bio->bi_bdev ? bio->bi_bdev->bd_dev : 0)
+               ctf_integer(sector_t, sector, bio->bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio->bi_size >> 9)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio), bio->bi_size)
+               ctf_integer(pid_t, tid, current->pid)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       )
+)
+
+/**
+ * block_bio_backmerge - merging block operation to the end of an existing operation
+ * @q: queue holding operation
+ * @bio: new block operation to merge
+ *
+ * Merging block request @bio to the end of an existing block request
+ * in queue @q.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_bio, block_bio_backmerge,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio),
+
+       TP_ARGS(q, bio)
+)
+
+/**
+ * block_bio_frontmerge - merging block operation to the beginning of an existing operation
+ * @q: queue holding operation
+ * @bio: new block operation to merge
+ *
+ * Merging block IO operation @bio to the beginning of an existing block
+ * operation in queue @q.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_bio, block_bio_frontmerge,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio),
+
+       TP_ARGS(q, bio)
+)
+
+/**
+ * block_bio_queue - putting new block IO operation in queue
+ * @q: queue holding operation
+ * @bio: new block operation
+ *
+ * About to place the block IO operation @bio into queue @q.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_bio, block_bio_queue,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio),
+
+       TP_ARGS(q, bio)
+)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,9,0)) */
+
+LTTNG_TRACEPOINT_EVENT_CLASS(block_get_rq,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio, int rw),
+
+       TP_ARGS(q, bio, rw),
+
+       TP_FIELDS(
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,14,0))
+               ctf_integer(dev_t, dev, bio ? bio_dev(bio) : 0)
+#else
+               ctf_integer(dev_t, dev, bio ? bio->bi_bdev->bd_dev : 0)
+#endif
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0))
+               ctf_integer(sector_t, sector, bio ? bio->bi_iter.bi_sector : 0)
+               ctf_integer(unsigned int, nr_sector,
+                       bio ? bio_sectors(bio) : 0)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       bio ? lttng_bio_op(bio) : 0,
+                       bio ? lttng_bio_rw(bio) : 0,
+                       bio ? bio->bi_iter.bi_size : 0)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(sector_t, sector, bio ? bio->bi_sector : 0)
+               ctf_integer(unsigned int, nr_sector,
+                       bio ? bio->bi_size >> 9 : 0)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       bio ? lttng_bio_op(bio) : 0,
+                       bio ? lttng_bio_rw(bio) : 0,
+                       bio ? bio->bi_size : 0)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(pid_t, tid, current->pid)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+        )
+)
+
+/**
+ * block_getrq - get a free request entry in queue for block IO operations
+ * @q: queue for operations
+ * @bio: pending block IO operation (can be %NULL)
+ * @rw: low bit indicates a read (%0) or a write (%1)
+ *
+ * A request struct for queue @q has been allocated to handle the
+ * block IO operation @bio.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_get_rq, block_getrq,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio, int rw),
+
+       TP_ARGS(q, bio, rw)
+)
+
+/**
+ * block_sleeprq - waiting to get a free request entry in queue for block IO operation
+ * @q: queue for operation
+ * @bio: pending block IO operation (can be %NULL)
+ * @rw: low bit indicates a read (%0) or a write (%1)
+ *
+ * In the case where a request struct cannot be provided for queue @q
+ * the process needs to wait for an request struct to become
+ * available.  This tracepoint event is generated each time the
+ * process goes to sleep waiting for request struct become available.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_get_rq, block_sleeprq,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio, int rw),
+
+       TP_ARGS(q, bio, rw)
+)
+
+/**
+ * block_plug - keep operations requests in request queue
+ * @q: request queue to plug
+ *
+ * Plug the request queue @q.  Do not allow block operation requests
+ * to be sent to the device driver. Instead, accumulate requests in
+ * the queue to improve throughput performance of the block device.
+ */
+LTTNG_TRACEPOINT_EVENT(block_plug,
+
+       TP_PROTO(struct request_queue *q),
+
+       TP_ARGS(q),
+
+       TP_FIELDS(
+               ctf_integer(pid_t, tid, current->pid)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       )
+)
+
+LTTNG_TRACEPOINT_EVENT_CLASS(block_unplug,
+
+       TP_PROTO(struct request_queue *q, unsigned int depth, bool explicit),
+
+       TP_ARGS(q, depth, explicit),
+
+       TP_FIELDS(
+               ctf_integer(int, nr_rq, depth)
+               ctf_integer(pid_t, tid, current->pid)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       )
+)
+
+/**
+ * block_unplug - release of operations requests in request queue
+ * @q: request queue to unplug
+ * @depth: number of requests just added to the queue
+ * @explicit: whether this was an explicit unplug, or one from schedule()
+ *
+ * Unplug request queue @q because device driver is scheduled to work
+ * on elements in the request queue.
+ */
+LTTNG_TRACEPOINT_EVENT_INSTANCE(block_unplug, block_unplug,
+
+       TP_PROTO(struct request_queue *q, unsigned int depth, bool explicit),
+
+       TP_ARGS(q, depth, explicit)
+)
+
+/**
+ * block_split - split a single bio struct into two bio structs
+ * @q: queue containing the bio
+ * @bio: block operation being split
+ * @new_sector: The starting sector for the new bio
+ *
+ * The bio request @bio in request queue @q needs to be split into two
+ * bio requests. The newly created @bio request starts at
+ * @new_sector. This split may be required due to hardware limitation
+ * such as operation crossing device boundaries in a RAID system.
+ */
+LTTNG_TRACEPOINT_EVENT(block_split,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio,
+                unsigned int new_sector),
+
+       TP_ARGS(q, bio, new_sector),
+
+       TP_FIELDS(
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,14,0))
+               ctf_integer(dev_t, dev, bio_dev(bio))
+#else
+               ctf_integer(dev_t, dev, bio->bi_bdev->bd_dev)
+#endif
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0))
+               ctf_integer(sector_t, sector, bio->bi_iter.bi_sector)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio),
+                       bio->bi_iter.bi_size)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(sector_t, sector, bio->bi_sector)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio), bio->bi_size)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(sector_t, new_sector, new_sector)
+               ctf_integer(pid_t, tid, current->pid)
+               ctf_array_text(char, comm, current->comm, TASK_COMM_LEN)
+       )
+)
+
+/**
+ * block_bio_remap - map request for a logical device to the raw device
+ * @q: queue holding the operation
+ * @bio: revised operation
+ * @dev: device for the operation
+ * @from: original sector for the operation
+ *
+ * An operation for a logical device has been mapped to the
+ * raw block device.
+ */
+LTTNG_TRACEPOINT_EVENT(block_bio_remap,
+
+       TP_PROTO(struct request_queue *q, struct bio *bio, dev_t dev,
+                sector_t from),
+
+       TP_ARGS(q, bio, dev, from),
+
+       TP_FIELDS(
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,14,0))
+               ctf_integer(dev_t, dev, bio_dev(bio))
+#else
+               ctf_integer(dev_t, dev, bio->bi_bdev->bd_dev)
+#endif
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0))
+               ctf_integer(sector_t, sector, bio->bi_iter.bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio_sectors(bio))
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio),
+                       bio->bi_iter.bi_size)
+#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(sector_t, sector, bio->bi_sector)
+               ctf_integer(unsigned int, nr_sector, bio->bi_size >> 9)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_bio_op(bio), lttng_bio_rw(bio), bio->bi_size)
+#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,14,0)) */
+               ctf_integer(dev_t, old_dev, dev)
+               ctf_integer(sector_t, old_sector, from)
+       )
+)
+
+/**
+ * block_rq_remap - map request for a block operation request
+ * @q: queue holding the operation
+ * @rq: block IO operation request
+ * @dev: device for the operation
+ * @from: original sector for the operation
+ *
+ * The block operation request @rq in @q has been remapped.  The block
+ * operation request @rq holds the current information and @from hold
+ * the original sector.
+ */
+LTTNG_TRACEPOINT_EVENT(block_rq_remap,
+
+       TP_PROTO(struct request_queue *q, struct request *rq, dev_t dev,
+                sector_t from),
+
+       TP_ARGS(q, rq, dev, from),
+
+       TP_FIELDS(
+               ctf_integer(dev_t, dev, disk_devt(rq->rq_disk))
+               ctf_integer(sector_t, sector, blk_rq_pos(rq))
+               ctf_integer(unsigned int, nr_sector, blk_rq_sectors(rq))
+               ctf_integer(dev_t, old_dev, dev)
+               ctf_integer(sector_t, old_sector, from)
+               blk_rwbs_ctf_integer(unsigned int, rwbs,
+                       lttng_req_op(rq), lttng_req_rw(rq), blk_rq_bytes(rq))
+       )
+)
+
+#undef __print_rwbs_flags
+#undef blk_fill_rwbs
+
+#endif /* LTTNG_TRACE_BLOCK_H */
+
+/* This part must be outside protection */
+#include <lttng/define_trace.h>
This page took 0.033099 seconds and 4 git commands to generate.