dtrace: add SDT probes

This adds a variety of SDT probes.

XXX add documentation here from the commit messages

Signed-off-by: Nick Alcock <nick.alcock@oracle.com>
Signed-off-by: Kris Van Hees <kris.van.hees@oracle.com>
Signed-off-by: Tomas Jedlicka <tomas.jedlicka@oracle.com>
Signed-off-by: Eugene Loh <eugene.loh@oracle.com>
Signed-off-by: Alan Maguire <alan.maguire@oracle.com>
Signed-off-by: David Mc Lean <david.mclean@oracle.com>
Signed-off-by: Vincent Lim <vincent.lim@oracle.com>

block/bio.c

@@ -1247,6 +1247,8 @@ int submit_bio_wait(struct bio *bio)
 	bio->bi_opf |= REQ_SYNC;
 	submit_bio(bio);
 
+	DTRACE_IO(wait__start, struct bio * : (bufinfo_t *, devinfo_t *), bio,
+		  struct file * : fileinfo_t *, NULL);
 	/* Prevent hang_check timer from firing at us during very long I/O */
 	hang_check = sysctl_hung_task_timeout_secs;
 	if (hang_check)
@@ -1255,6 +1257,8 @@ int submit_bio_wait(struct bio *bio)
 			;
 	else
 		wait_for_completion_io(&done);
+	DTRACE_IO(wait__done, struct bio * : (bufinfo_t *, devinfo_t *), bio,
+		  struct file * : fileinfo_t *, NULL);
 
 	return blk_status_to_errno(bio->bi_status);
 }
@@ -1488,6 +1492,9 @@ void bio_endio(struct bio *bio)
 	}
 
 	blk_throtl_bio_endio(bio);
+	DTRACE_IO(done, struct bio * :
+		  (bufinfo_t *, devinfo_t *), bio,
+		  struct file * : fileinfo_t *, NULL);
 	/* release cgroup info */
 	bio_uninit(bio);
 	if (bio->bi_end_io)

block/blk-core.c

@@ -902,11 +902,15 @@ static noinline_for_stack bool submit_bio_checks(struct bio *bio)
 		 */
 		bio_set_flag(bio, BIO_TRACE_COMPLETION);
 	}
+	DTRACE_IO(start, struct bio * : (bufinfo_t *, devinfo_t *), bio,
+		  struct file * : fileinfo_t *, NULL);
 	return true;
 
 not_supported:
 	status = BLK_STS_NOTSUPP;
 end_io:
+	DTRACE_IO(start, struct bio * : (bufinfo_t *, devinfo_t *), bio,
+		  struct file * : fileinfo_t *, NULL);
 	bio->bi_status = status;
 	bio_endio(bio);
 	return false;

fs/exec.c

@@ -65,6 +65,7 @@
 #include <linux/vmalloc.h>
 #include <linux/io_uring.h>
 #include <linux/syscall_user_dispatch.h>
+#include <linux/sdt.h>
 #include <linux/dtrace_os.h>
 
 #include <linux/uaccess.h>
@@ -1809,6 +1810,7 @@ static int bprm_execve(struct linux_binprm *bprm,
 	current->in_execve = 1;
 
 	file = do_open_execat(fd, filename, flags);
+	DTRACE_PROC(exec, char *, filename->name);
 	retval = PTR_ERR(file);
 	if (IS_ERR(file))
 		goto out_unmark;
@@ -1847,6 +1849,8 @@ static int bprm_execve(struct linux_binprm *bprm,
 	rseq_execve(current);
 	acct_update_integrals(current);
 	task_numa_free(current, false);
+
+	DTRACE_PROC(exec__success);
 	return retval;
 
 out:
@@ -1932,6 +1936,8 @@ static int do_execveat_common(int fd, struct filename *filename,
 
 out_ret:
 	putname(filename);
+	if (retval < 0)
+		DTRACE_PROC(exec__failure, int, retval);
 	return retval;
 }
 
@@ -1985,6 +1991,8 @@ int kernel_execve(const char *kernel_filename,
 	free_bprm(bprm);
 out_ret:
 	putname(filename);
+	if (retval < 0)
+		DTRACE_PROC(exec__failure, int, retval);
 	return retval;
 }
 

fs/nfs/internal.h

@@ -10,6 +10,7 @@
 #include <linux/sunrpc/addr.h>
 #include <linux/nfs_page.h>
 #include <linux/wait_bit.h>
+#include <linux/sdt.h>
 
 #define NFS_SB_MASK (SB_RDONLY|SB_NOSUID|SB_NODEV|SB_NOEXEC|SB_SYNCHRONOUS)
 
@@ -834,3 +835,15 @@ static inline void nfs_set_port(struct sockaddr *sap, int *port,
 
 	rpc_set_port(sap, *port);
 }
+
+#define	DTRACE_IO_NFS(name, rw, size, inode)			\
+	if (DTRACE_IO_ENABLED(name)) {				\
+		struct bio bio __maybe_unused = {		\
+			.bi_opf = rw,				\
+			.bi_iter.bi_size = size,		\
+			.bi_iter.bi_sector = NFS_FILEID(inode),	\
+		};						\
+		DTRACE_IO(name, struct bio * : (bufinfo_t *,	\
+			  devinfo_t *), &bio,			\
+			  struct file * : fileinfo_t *, NULL);	\
+}

fs/nfs/read.c

@@ -197,6 +197,8 @@ static void nfs_initiate_read(struct nfs_pgio_header *hdr,
 	struct inode *inode = hdr->inode;
 	int swap_flags = IS_SWAPFILE(inode) ? NFS_RPC_SWAPFLAGS : 0;
 
+	DTRACE_IO_NFS(start, REQ_OP_READ, hdr->args.count, hdr->inode);
+
 	task_setup_data->flags |= swap_flags;
 	rpc_ops->read_setup(hdr, msg);
 	trace_nfs_initiate_read(hdr);
@@ -228,6 +230,7 @@ static int nfs_readpage_done(struct rpc_task *task,
 			     struct inode *inode)
 {
 	int status = NFS_PROTO(inode)->read_done(task, hdr);
+	DTRACE_IO_NFS(done, REQ_OP_READ, hdr->res.count, hdr->inode);
 	if (status != 0)
 		return status;
 

fs/nfs/write.c

@@ -1411,6 +1411,7 @@ static void nfs_initiate_write(struct nfs_pgio_header *hdr,
 	task_setup_data->priority = priority;
 	rpc_ops->write_setup(hdr, msg, &task_setup_data->rpc_client);
 	trace_nfs_initiate_write(hdr);
+	DTRACE_IO_NFS(start, REQ_OP_WRITE, hdr->args.count, hdr->inode);
 }
 
 /* If a nfs_flush_* function fails, it should remove reqs from @head and
@@ -1570,6 +1571,7 @@ static int nfs_writeback_done(struct rpc_task *task,
 	 * depend on tighter cache coherency when writing.
 	 */
 	status = NFS_PROTO(inode)->write_done(task, hdr);
+	DTRACE_IO_NFS(done, REQ_OP_WRITE, hdr->res.count, hdr->inode);
 	if (status != 0)
 		return status;
 

fs/xfs/xfs_buf.c

@@ -49,6 +49,20 @@ static kmem_zone_t *xfs_buf_zone;
  *	  b_lock (trylock due to inversion)
  */
 
+#define	DTRACE_IO_XFS_WAIT(name, bp, is_write)				\
+	if (DTRACE_IO_ENABLED(name)) {					\
+		struct bio bio __maybe_unused = {			\
+			.bi_iter.bi_sector = xfs_buf_daddr(bp),		\
+			.bi_iter.bi_size = (bp)->b_length,		\
+			.bi_opf = is_write ?				\
+				REQ_OP_WRITE : REQ_OP_READ,		\
+			.bi_bdev = (bp)->b_target->bt_bdev		\
+		};							\
+		DTRACE_IO(name, struct bio * : (bufinfo_t *,		\
+			  devinfo_t *), &bio,				\
+			  struct file * : fileinfo_t *, NULL);		\
+	}
+
 static int __xfs_buf_submit(struct xfs_buf *bp, bool wait);
 
 static inline int
@@ -1567,10 +1581,14 @@ static int
 xfs_buf_iowait(
 	struct xfs_buf	*bp)
 {
+	int orig_flags __attribute__((unused)) = bp->b_flags;
+
 	ASSERT(!(bp->b_flags & XBF_ASYNC));
 
 	trace_xfs_buf_iowait(bp, _RET_IP_);
+	DTRACE_IO_XFS_WAIT(wait__start, bp, orig_flags & XBF_WRITE);
 	wait_for_completion(&bp->b_iowait);
+	DTRACE_IO_XFS_WAIT(wait__done, bp, orig_flags & XBF_WRITE);
 	trace_xfs_buf_iowait_done(bp, _RET_IP_);
 
 	return bp->b_error;

include/linux/rwlock_api_smp.h

@@ -5,6 +5,8 @@
 # error "please don't include this file directly"
 #endif
 
+#include <linux/sdt.h>
+
 /*
  * include/linux/rwlock_api_smp.h
  *
@@ -119,6 +121,8 @@ static inline int __raw_read_trylock(rwlock_t *lock)
 	preempt_disable();
 	if (do_raw_read_trylock(lock)) {
 		rwlock_acquire_read(&lock->dep_map, 0, 1, _RET_IP_);
+		DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+				DTRACE_LOCKSTAT_RW_READER);
 		return 1;
 	}
 	preempt_enable();
@@ -130,6 +134,8 @@ static inline int __raw_write_trylock(rwlock_t *lock)
 	preempt_disable();
 	if (do_raw_write_trylock(lock)) {
 		rwlock_acquire(&lock->dep_map, 0, 1, _RET_IP_);
+		DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+				DTRACE_LOCKSTAT_RW_WRITER);
 		return 1;
 	}
 	preempt_enable();
@@ -148,6 +154,8 @@ static inline void __raw_read_lock(rwlock_t *lock)
 	preempt_disable();
 	rwlock_acquire_read(&lock->dep_map, 0, 0, _RET_IP_);
 	LOCK_CONTENDED(lock, do_raw_read_trylock, do_raw_read_lock);
+	DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_READER);
 }
 
 static inline unsigned long __raw_read_lock_irqsave(rwlock_t *lock)
@@ -159,6 +167,8 @@ static inline unsigned long __raw_read_lock_irqsave(rwlock_t *lock)
 	rwlock_acquire_read(&lock->dep_map, 0, 0, _RET_IP_);
 	LOCK_CONTENDED_FLAGS(lock, do_raw_read_trylock, do_raw_read_lock,
 			     do_raw_read_lock_flags, &flags);
+	DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_READER);
 	return flags;
 }
 
@@ -168,13 +178,17 @@ static inline void __raw_read_lock_irq(rwlock_t *lock)
 	preempt_disable();
 	rwlock_acquire_read(&lock->dep_map, 0, 0, _RET_IP_);
 	LOCK_CONTENDED(lock, do_raw_read_trylock, do_raw_read_lock);
+	DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_READER);
 }
 
 static inline void __raw_read_lock_bh(rwlock_t *lock)
 {
 	__local_bh_disable_ip(_RET_IP_, SOFTIRQ_LOCK_OFFSET);
 	rwlock_acquire_read(&lock->dep_map, 0, 0, _RET_IP_);
 	LOCK_CONTENDED(lock, do_raw_read_trylock, do_raw_read_lock);
+	DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_READER);
 }
 
 static inline unsigned long __raw_write_lock_irqsave(rwlock_t *lock)
@@ -186,6 +200,8 @@ static inline unsigned long __raw_write_lock_irqsave(rwlock_t *lock)
 	rwlock_acquire(&lock->dep_map, 0, 0, _RET_IP_);
 	LOCK_CONTENDED_FLAGS(lock, do_raw_write_trylock, do_raw_write_lock,
 			     do_raw_write_lock_flags, &flags);
+	DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_WRITER);
 	return flags;
 }
 
@@ -195,20 +211,26 @@ static inline void __raw_write_lock_irq(rwlock_t *lock)
 	preempt_disable();
 	rwlock_acquire(&lock->dep_map, 0, 0, _RET_IP_);
 	LOCK_CONTENDED(lock, do_raw_write_trylock, do_raw_write_lock);
+	DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_WRITER);
 }
 
 static inline void __raw_write_lock_bh(rwlock_t *lock)
 {
 	__local_bh_disable_ip(_RET_IP_, SOFTIRQ_LOCK_OFFSET);
 	rwlock_acquire(&lock->dep_map, 0, 0, _RET_IP_);
 	LOCK_CONTENDED(lock, do_raw_write_trylock, do_raw_write_lock);
+	DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_WRITER);
 }
 
 static inline void __raw_write_lock(rwlock_t *lock)
 {
 	preempt_disable();
 	rwlock_acquire(&lock->dep_map, 0, 0, _RET_IP_);
 	LOCK_CONTENDED(lock, do_raw_write_trylock, do_raw_write_lock);
+	DTRACE_LOCKSTAT(rw__acquire, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_WRITER);
 }
 
 #endif /* !CONFIG_GENERIC_LOCKBREAK || CONFIG_DEBUG_LOCK_ALLOC */
@@ -217,13 +239,17 @@ static inline void __raw_write_unlock(rwlock_t *lock)
 {
 	rwlock_release(&lock->dep_map, _RET_IP_);
 	do_raw_write_unlock(lock);
+	DTRACE_LOCKSTAT(rw__release, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_WRITER);
 	preempt_enable();
 }
 
 static inline void __raw_read_unlock(rwlock_t *lock)
 {
 	rwlock_release(&lock->dep_map, _RET_IP_);
 	do_raw_read_unlock(lock);
+	DTRACE_LOCKSTAT(rw__release, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_READER);
 	preempt_enable();
 }
 
@@ -232,6 +258,8 @@ __raw_read_unlock_irqrestore(rwlock_t *lock, unsigned long flags)
 {
 	rwlock_release(&lock->dep_map, _RET_IP_);
 	do_raw_read_unlock(lock);
+	DTRACE_LOCKSTAT(rw__release, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_READER);
 	local_irq_restore(flags);
 	preempt_enable();
 }
@@ -240,6 +268,8 @@ static inline void __raw_read_unlock_irq(rwlock_t *lock)
 {
 	rwlock_release(&lock->dep_map, _RET_IP_);
 	do_raw_read_unlock(lock);
+	DTRACE_LOCKSTAT(rw__release, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_READER);
 	local_irq_enable();
 	preempt_enable();
 }
@@ -248,6 +278,8 @@ static inline void __raw_read_unlock_bh(rwlock_t *lock)
 {
 	rwlock_release(&lock->dep_map, _RET_IP_);
 	do_raw_read_unlock(lock);
+	DTRACE_LOCKSTAT(rw__release, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_READER);
 	__local_bh_enable_ip(_RET_IP_, SOFTIRQ_LOCK_OFFSET);
 }
 
@@ -256,6 +288,8 @@ static inline void __raw_write_unlock_irqrestore(rwlock_t *lock,
 {
 	rwlock_release(&lock->dep_map, _RET_IP_);
 	do_raw_write_unlock(lock);
+	DTRACE_LOCKSTAT(rw__release, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_WRITER);
 	local_irq_restore(flags);
 	preempt_enable();
 }
@@ -264,6 +298,8 @@ static inline void __raw_write_unlock_irq(rwlock_t *lock)
 {
 	rwlock_release(&lock->dep_map, _RET_IP_);
 	do_raw_write_unlock(lock);
+	DTRACE_LOCKSTAT(rw__release, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_WRITER);
 	local_irq_enable();
 	preempt_enable();
 }
@@ -272,6 +308,8 @@ static inline void __raw_write_unlock_bh(rwlock_t *lock)
 {
 	rwlock_release(&lock->dep_map, _RET_IP_);
 	do_raw_write_unlock(lock);
+	DTRACE_LOCKSTAT(rw__release, struct rwlock *, lock, int,
+			DTRACE_LOCKSTAT_RW_WRITER);
 	__local_bh_enable_ip(_RET_IP_, SOFTIRQ_LOCK_OFFSET);
 }