[PATCH v3 00/16] block: add blktrace support for zoned block devi

[PATCH v3 00/16] block: add blktrace support for zoned block devi

[Johannes Thumshirn]

This patch series extends the kernel blktrace infrastructure to support
tracing zoned block device commands. Zoned block devices (e.g., ZAC/ZBC and
ZNS) introduce command types such as zone open, close, reset, finish, and zone
append. These are currently not visible in blktrace, making it difficult to
debug and analyze I/O workloads on zoned devices.

The patches in this series utilize the new trace points for these zone
management operations, and propagate the necessary context to the blktrace
logging path. These additions are designed to be backward-compatible, and are
only active when zoned devices are in use.

In order to not break the user-space ABI, a new ioctl was introduced to
request the new version of the blk_io_trace with extended 'action' field.

The user-space tools in the blktrace suite are updated in a separate patch
series to interpret and display the new trace events.This patch series extends
the kernel blktrace infrastructure to support tracing zoned block device
commands. Zoned block devices (e.g., ZAC/ZBC and ZNS) introduce command types
such as zone open, close, reset, finish, and zone append. These are currently
not visible in blktrace, making it difficult to debug and analyze I/O
workloads on zoned devices.

The patches in this series utilize the new trace points for these zone
management operations, and propagate the necessary context to the blktrace
logging path. These additions are designed to be backward-compatible, and are
only active when zoned devices are in use.

In order to not break the user-space ABI, a new ioctl was introduced to
request the new version of the blk_io_trace with extended 'action' field.

The user-space tools in the blktrace suite are updated in a separate patch
series to interpret and display the new trace events.

I've tested on SMR drives, TCMU emulated SMR drives and zloop, with both XFS
and BTRFS as filesystems on top of the hardwdare. Testing of different
hardware/setups is highly encouraged.

Changes to v2:
- Collect Reviewed-bys
- Convert trace_note and ftrace's blk_io_tracer to blk_io_trace2
- Don't play games with the layout of the 'action' field
- Fix structure alignments
- Drop Zone Management trace action, it's a command not an action

Johannes Thumshirn (16):
  blktrace: only calculate trace length once
  blktrace: factor out recording a blktrace event
  blktrace: split out relaying a blktrace event
  blktrace: untangle if/else sequence in __blk_add_trace
  blktrace: change the internal action to 64bit
  blktrace: split do_blk_trace_setup into two functions
  blktrace: add definitions for blk_user_trace_setup2
  blktrace: pass blk_user_trace2 to setup functions
  blktrace: add definitions for struct blk_io_trace2
  blktrace: differentiate between blk_io_trace versions
  blktrace: move trace_note to blk_io_trace2
  blktrace: move ftrace blk_io_tracer to blk_io_trace2
  blktrace: add block trace commands for zone operations
  blktrace: expose ZONE APPEND completions to blktrace
  blktrace: trace zone write plugging operations
  blktrace: handle BLKTRACESETUP2 ioctl

 block/ioctl.c                     |   1 +
 include/linux/blktrace_api.h      |   3 +-
 include/uapi/linux/blktrace_api.h |  53 +++-
 include/uapi/linux/fs.h           |   1 +
 kernel/trace/blktrace.c           | 463 ++++++++++++++++++++++--------
 5 files changed, 400 insertions(+), 121 deletions(-)

-- 
2.51.0

[PATCH v3 01/16] blktrace: only calculate trace length once

[Johannes Thumshirn]

De-duplicate the calculation of the trace length instead of doing the
calculation twice, once for calling trace_buffer_lock_reserve() and once
for calling relay_reserve().

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 6941145b5058..bc4b885f2cec 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -76,13 +76,14 @@ static void trace_note(struct blk_trace *bt, pid_t pid, int action,
 	int cpu = smp_processor_id();
 	bool blk_tracer = blk_tracer_enabled;
 	ssize_t cgid_len = cgid ? sizeof(cgid) : 0;
+	size_t trace_len;
 
+	trace_len = sizeof(*t) + cgid_len + len;
 	if (blk_tracer) {
 		buffer = blk_tr->array_buffer.buffer;
 		trace_ctx = tracing_gen_ctx_flags(0);
 		event = trace_buffer_lock_reserve(buffer, TRACE_BLK,
-						  sizeof(*t) + len + cgid_len,
-						  trace_ctx);
+						  trace_len, trace_ctx);
 		if (!event)
 			return;
 		t = ring_buffer_event_data(event);
@@ -92,7 +93,7 @@ static void trace_note(struct blk_trace *bt, pid_t pid, int action,
 	if (!bt->rchan)
 		return;
 
-	t = relay_reserve(bt->rchan, sizeof(*t) + len + cgid_len);
+	t = relay_reserve(bt->rchan, trace_len);
 	if (t) {
 		t->magic = BLK_IO_TRACE_MAGIC | BLK_IO_TRACE_VERSION;
 		t->time = ktime_to_ns(ktime_get());
@@ -228,6 +229,7 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 	bool blk_tracer = blk_tracer_enabled;
 	ssize_t cgid_len = cgid ? sizeof(cgid) : 0;
 	const enum req_op op = opf & REQ_OP_MASK;
+	size_t trace_len;
 
 	if (unlikely(bt->trace_state != Blktrace_running && !blk_tracer))
 		return;
@@ -250,14 +252,14 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 		return;
 	cpu = raw_smp_processor_id();
 
+	trace_len = sizeof(*t) + pdu_len + cgid_len;
 	if (blk_tracer) {
 		tracing_record_cmdline(current);
 
 		buffer = blk_tr->array_buffer.buffer;
 		trace_ctx = tracing_gen_ctx_flags(0);
 		event = trace_buffer_lock_reserve(buffer, TRACE_BLK,
-						  sizeof(*t) + pdu_len + cgid_len,
-						  trace_ctx);
+						  trace_len, trace_ctx);
 		if (!event)
 			return;
 		t = ring_buffer_event_data(event);
@@ -273,7 +275,7 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 	 * from coming in and stepping on our toes.
 	 */
 	local_irq_save(flags);
-	t = relay_reserve(bt->rchan, sizeof(*t) + pdu_len + cgid_len);
+	t = relay_reserve(bt->rchan, trace_len);
 	if (t) {
 		sequence = per_cpu_ptr(bt->sequence, cpu);
 
-- 
2.51.0

[PATCH v3 02/16] blktrace: factor out recording a blktrace event

[Johannes Thumshirn]

Factor out the recording of a blktrace event into its own function,
deduplicating the code.

This also enables recording different versions of the blktrace protocol
later on.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 89 +++++++++++++++++++++++------------------
 1 file changed, 49 insertions(+), 40 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index bc4b885f2cec..25a0a1b09747 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -63,6 +63,34 @@ static int blk_probes_ref;
 static void blk_register_tracepoints(void);
 static void blk_unregister_tracepoints(void);
 
+static void record_blktrace_event(struct blk_io_trace *t, pid_t pid, int cpu,
+				  sector_t sector, int bytes, u32 what,
+				  dev_t dev, int error, u64 cgid,
+				  ssize_t cgid_len, void *pdu_data, int pdu_len)
+
+{
+	/*
+	 * These two are not needed in ftrace as they are in the
+	 * generic trace_entry, filled by tracing_generic_entry_update,
+	 * but for the trace_event->bin() synthesizer benefit we do it
+	 * here too.
+	 */
+	t->cpu = cpu;
+	t->pid = pid;
+
+	t->sector = sector;
+	t->bytes = bytes;
+	t->action = what;
+	t->device = dev;
+	t->error = error;
+	t->pdu_len = pdu_len + cgid_len;
+
+	if (cgid_len)
+		memcpy((void *)t + sizeof(*t), &cgid, cgid_len);
+	if (pdu_len)
+		memcpy((void *)t + sizeof(*t) + cgid_len, pdu_data, pdu_len);
+}
+
 /*
  * Send out a notify message.
  */
@@ -87,7 +115,12 @@ static void trace_note(struct blk_trace *bt, pid_t pid, int action,
 		if (!event)
 			return;
 		t = ring_buffer_event_data(event);
-		goto record_it;
+		record_blktrace_event(t, pid, cpu, 0, 0,
+				      action | (cgid ? __BLK_TN_CGROUP : 0),
+				      bt->dev, 0, cgid, cgid_len, (void *)data,
+				      len);
+		trace_buffer_unlock_commit(blk_tr, buffer, event, trace_ctx);
+		return;
 	}
 
 	if (!bt->rchan)
@@ -97,18 +130,11 @@ static void trace_note(struct blk_trace *bt, pid_t pid, int action,
 	if (t) {
 		t->magic = BLK_IO_TRACE_MAGIC | BLK_IO_TRACE_VERSION;
 		t->time = ktime_to_ns(ktime_get());
-record_it:
-		t->device = bt->dev;
-		t->action = action | (cgid ? __BLK_TN_CGROUP : 0);
-		t->pid = pid;
-		t->cpu = cpu;
-		t->pdu_len = len + cgid_len;
-		if (cgid_len)
-			memcpy((void *)t + sizeof(*t), &cgid, cgid_len);
-		memcpy((void *) t + sizeof(*t) + cgid_len, data, len);
-
-		if (blk_tracer)
-			trace_buffer_unlock_commit(blk_tr, buffer, event, trace_ctx);
+
+		record_blktrace_event(t, pid, cpu, 0, 0,
+				      action | (cgid ? __BLK_TN_CGROUP : 0),
+				      bt->dev, 0, cgid, cgid_len, (void *)data,
+				      len);
 	}
 }
 
@@ -263,7 +289,12 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 		if (!event)
 			return;
 		t = ring_buffer_event_data(event);
-		goto record_it;
+
+		record_blktrace_event(t, pid, cpu, sector, bytes, what, bt->dev,
+				      error, cgid, cgid_len, pdu_data, pdu_len);
+
+		trace_buffer_unlock_commit(blk_tr, buffer, event, trace_ctx);
+		return;
 	}
 
 	if (unlikely(tsk->btrace_seq != blktrace_seq))
@@ -282,32 +313,10 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 		t->magic = BLK_IO_TRACE_MAGIC | BLK_IO_TRACE_VERSION;
 		t->sequence = ++(*sequence);
 		t->time = ktime_to_ns(ktime_get());
-record_it:
-		/*
-		 * These two are not needed in ftrace as they are in the
-		 * generic trace_entry, filled by tracing_generic_entry_update,
-		 * but for the trace_event->bin() synthesizer benefit we do it
-		 * here too.
-		 */
-		t->cpu = cpu;
-		t->pid = pid;
-
-		t->sector = sector;
-		t->bytes = bytes;
-		t->action = what;
-		t->device = bt->dev;
-		t->error = error;
-		t->pdu_len = pdu_len + cgid_len;
-
-		if (cgid_len)
-			memcpy((void *)t + sizeof(*t), &cgid, cgid_len);
-		if (pdu_len)
-			memcpy((void *)t + sizeof(*t) + cgid_len, pdu_data, pdu_len);
-
-		if (blk_tracer) {
-			trace_buffer_unlock_commit(blk_tr, buffer, event, trace_ctx);
-			return;
-		}
+
+		record_blktrace_event(t, pid, cpu, sector, bytes, what,
+				      bt->dev, error, cgid, cgid_len,
+				      pdu_data, pdu_len);
 	}
 
 	local_irq_restore(flags);
-- 
2.51.0

[PATCH v3 03/16] blktrace: split out relaying a blktrace event

[Johannes Thumshirn]

Split out the code relaying a blktrace event to user-space using relayfs.

This enables adding a second version supporting a new version of the
protocol.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 60 ++++++++++++++++++++++-------------------
 1 file changed, 32 insertions(+), 28 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 25a0a1b09747..51745832c713 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -91,6 +91,26 @@ static void record_blktrace_event(struct blk_io_trace *t, pid_t pid, int cpu,
 		memcpy((void *)t + sizeof(*t) + cgid_len, pdu_data, pdu_len);
 }
 
+static void relay_blktrace_event(struct blk_trace *bt, unsigned long sequence,
+				 pid_t pid, int cpu, sector_t sector, int bytes,
+				 u32 what, int error, u64 cgid,
+				 ssize_t cgid_len, void *pdu_data, int pdu_len)
+{
+	struct blk_io_trace *t;
+	size_t trace_len = sizeof(*t) + pdu_len + cgid_len;
+
+	t = relay_reserve(bt->rchan, trace_len);
+	if (!t)
+		return;
+
+	t->magic = BLK_IO_TRACE_MAGIC | BLK_IO_TRACE_VERSION;
+	t->sequence = sequence;
+	t->time = ktime_to_ns(ktime_get());
+
+	record_blktrace_event(t, pid, cpu, sector, bytes, what, bt->dev, error,
+			      cgid, cgid_len, pdu_data, pdu_len);
+}
+
 /*
  * Send out a notify message.
  */
@@ -126,16 +146,9 @@ static void trace_note(struct blk_trace *bt, pid_t pid, int action,
 	if (!bt->rchan)
 		return;
 
-	t = relay_reserve(bt->rchan, trace_len);
-	if (t) {
-		t->magic = BLK_IO_TRACE_MAGIC | BLK_IO_TRACE_VERSION;
-		t->time = ktime_to_ns(ktime_get());
-
-		record_blktrace_event(t, pid, cpu, 0, 0,
-				      action | (cgid ? __BLK_TN_CGROUP : 0),
-				      bt->dev, 0, cgid, cgid_len, (void *)data,
-				      len);
-	}
+	relay_blktrace_event(bt, 0, pid, cpu, 0, 0,
+			     action | (cgid ? __BLK_TN_CGROUP : 0), 0, cgid,
+			     cgid_len, (void *)data, len);
 }
 
 /*
@@ -246,7 +259,6 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 	struct task_struct *tsk = current;
 	struct ring_buffer_event *event = NULL;
 	struct trace_buffer *buffer = NULL;
-	struct blk_io_trace *t;
 	unsigned long flags = 0;
 	unsigned long *sequence;
 	unsigned int trace_ctx = 0;
@@ -278,20 +290,21 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 		return;
 	cpu = raw_smp_processor_id();
 
-	trace_len = sizeof(*t) + pdu_len + cgid_len;
 	if (blk_tracer) {
 		tracing_record_cmdline(current);
 
 		buffer = blk_tr->array_buffer.buffer;
 		trace_ctx = tracing_gen_ctx_flags(0);
+		trace_len = sizeof(struct blk_io_trace) + pdu_len + cgid_len;
 		event = trace_buffer_lock_reserve(buffer, TRACE_BLK,
 						  trace_len, trace_ctx);
 		if (!event)
 			return;
-		t = ring_buffer_event_data(event);
 
-		record_blktrace_event(t, pid, cpu, sector, bytes, what, bt->dev,
-				      error, cgid, cgid_len, pdu_data, pdu_len);
+		record_blktrace_event(ring_buffer_event_data(event),
+				      pid, cpu, sector, bytes, what, bt->dev,
+				      error, cgid, cgid_len, pdu_data,
+				      pdu_len);
 
 		trace_buffer_unlock_commit(blk_tr, buffer, event, trace_ctx);
 		return;
@@ -306,19 +319,10 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 	 * from coming in and stepping on our toes.
 	 */
 	local_irq_save(flags);
-	t = relay_reserve(bt->rchan, trace_len);
-	if (t) {
-		sequence = per_cpu_ptr(bt->sequence, cpu);
-
-		t->magic = BLK_IO_TRACE_MAGIC | BLK_IO_TRACE_VERSION;
-		t->sequence = ++(*sequence);
-		t->time = ktime_to_ns(ktime_get());
-
-		record_blktrace_event(t, pid, cpu, sector, bytes, what,
-				      bt->dev, error, cgid, cgid_len,
-				      pdu_data, pdu_len);
-	}
-
+	sequence = per_cpu_ptr(bt->sequence, cpu);
+	(*sequence)++;
+	relay_blktrace_event(bt, *sequence, pid, cpu, sector, bytes, what,
+			     error, cgid, cgid_len, pdu_data, pdu_len);
 	local_irq_restore(flags);
 }
 
-- 
2.51.0

[PATCH v3 04/16] blktrace: untangle if/else sequence in __blk_add_trace

[Johannes Thumshirn]

Untangle the if/else sequence setting the trace action in
__blk_add_trace() and turn it into a switch statement for better
extensibility.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 51745832c713..11e264f67851 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -278,10 +278,19 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 	what |= MASK_TC_BIT(opf, META);
 	what |= MASK_TC_BIT(opf, PREFLUSH);
 	what |= MASK_TC_BIT(opf, FUA);
-	if (op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE)
+
+	switch (op) {
+	case REQ_OP_DISCARD:
+	case REQ_OP_SECURE_ERASE:
 		what |= BLK_TC_ACT(BLK_TC_DISCARD);
-	if (op == REQ_OP_FLUSH)
+		break;
+	case REQ_OP_FLUSH:
 		what |= BLK_TC_ACT(BLK_TC_FLUSH);
+		break;
+	default:
+		break;
+	}
+
 	if (cgid)
 		what |= __BLK_TA_CGROUP;
 
-- 
2.51.0

[PATCH v3 05/16] blktrace: change the internal action to 64bit

[Johannes Thumshirn]

Change the internal use of the action in blktrace to 64bit. Although for
now only the lower 32bits will be used.

With the upcoming version 2 of the blktrace user-space protocol the upper
32bit will also be utilized.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 38 +++++++++++++++++++-------------------
 1 file changed, 19 insertions(+), 19 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 11e264f67851..15d6788700ca 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -64,7 +64,7 @@ static void blk_register_tracepoints(void);
 static void blk_unregister_tracepoints(void);
 
 static void record_blktrace_event(struct blk_io_trace *t, pid_t pid, int cpu,
-				  sector_t sector, int bytes, u32 what,
+				  sector_t sector, int bytes, u64 what,
 				  dev_t dev, int error, u64 cgid,
 				  ssize_t cgid_len, void *pdu_data, int pdu_len)
 
@@ -80,7 +80,7 @@ static void record_blktrace_event(struct blk_io_trace *t, pid_t pid, int cpu,
 
 	t->sector = sector;
 	t->bytes = bytes;
-	t->action = what;
+	t->action = lower_32_bits(what);
 	t->device = dev;
 	t->error = error;
 	t->pdu_len = pdu_len + cgid_len;
@@ -93,7 +93,7 @@ static void record_blktrace_event(struct blk_io_trace *t, pid_t pid, int cpu,
 
 static void relay_blktrace_event(struct blk_trace *bt, unsigned long sequence,
 				 pid_t pid, int cpu, sector_t sector, int bytes,
-				 u32 what, int error, u64 cgid,
+				 u64 what, int error, u64 cgid,
 				 ssize_t cgid_len, void *pdu_data, int pdu_len)
 {
 	struct blk_io_trace *t;
@@ -114,7 +114,7 @@ static void relay_blktrace_event(struct blk_trace *bt, unsigned long sequence,
 /*
  * Send out a notify message.
  */
-static void trace_note(struct blk_trace *bt, pid_t pid, int action,
+static void trace_note(struct blk_trace *bt, pid_t pid, u64 action,
 		       const void *data, size_t len, u64 cgid)
 {
 	struct blk_io_trace *t;
@@ -127,6 +127,7 @@ static void trace_note(struct blk_trace *bt, pid_t pid, int action,
 	size_t trace_len;
 
 	trace_len = sizeof(*t) + cgid_len + len;
+	action = lower_32_bits(action | (cgid ? __BLK_TN_CGROUP : 0));
 	if (blk_tracer) {
 		buffer = blk_tr->array_buffer.buffer;
 		trace_ctx = tracing_gen_ctx_flags(0);
@@ -136,9 +137,8 @@ static void trace_note(struct blk_trace *bt, pid_t pid, int action,
 			return;
 		t = ring_buffer_event_data(event);
 		record_blktrace_event(t, pid, cpu, 0, 0,
-				      action | (cgid ? __BLK_TN_CGROUP : 0),
-				      bt->dev, 0, cgid, cgid_len, (void *)data,
-				      len);
+				      action, bt->dev, 0, cgid, cgid_len,
+				      (void *)data, len);
 		trace_buffer_unlock_commit(blk_tr, buffer, event, trace_ctx);
 		return;
 	}
@@ -146,8 +146,7 @@ static void trace_note(struct blk_trace *bt, pid_t pid, int action,
 	if (!bt->rchan)
 		return;
 
-	relay_blktrace_event(bt, 0, pid, cpu, 0, 0,
-			     action | (cgid ? __BLK_TN_CGROUP : 0), 0, cgid,
+	relay_blktrace_event(bt, 0, pid, cpu, 0, 0, action, 0, cgid,
 			     cgid_len, (void *)data, len);
 }
 
@@ -222,7 +221,7 @@ void __blk_trace_note_message(struct blk_trace *bt,
 }
 EXPORT_SYMBOL_GPL(__blk_trace_note_message);
 
-static int act_log_check(struct blk_trace *bt, u32 what, sector_t sector,
+static int act_log_check(struct blk_trace *bt, u64 what, sector_t sector,
 			 pid_t pid)
 {
 	if (((bt->act_mask << BLK_TC_SHIFT) & what) == 0)
@@ -253,7 +252,7 @@ static const u32 ddir_act[2] = { BLK_TC_ACT(BLK_TC_READ),
  * blk_io_trace structure and places it in a per-cpu subbuffer.
  */
 static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
-			    const blk_opf_t opf, u32 what, int error,
+			    const blk_opf_t opf, u64 what, int error,
 			    int pdu_len, void *pdu_data, u64 cgid)
 {
 	struct task_struct *tsk = current;
@@ -311,9 +310,9 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 			return;
 
 		record_blktrace_event(ring_buffer_event_data(event),
-				      pid, cpu, sector, bytes, what, bt->dev,
-				      error, cgid, cgid_len, pdu_data,
-				      pdu_len);
+				      pid, cpu, sector, bytes,
+				      what, bt->dev, error, cgid, cgid_len,
+				      pdu_data, pdu_len);
 
 		trace_buffer_unlock_commit(blk_tr, buffer, event, trace_ctx);
 		return;
@@ -330,8 +329,9 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 	local_irq_save(flags);
 	sequence = per_cpu_ptr(bt->sequence, cpu);
 	(*sequence)++;
-	relay_blktrace_event(bt, *sequence, pid, cpu, sector, bytes, what,
-			     error, cgid, cgid_len, pdu_data, pdu_len);
+	relay_blktrace_event(bt, *sequence, pid, cpu, sector, bytes,
+			     lower_32_bits(what), error, cgid, cgid_len,
+			     pdu_data, pdu_len);
 	local_irq_restore(flags);
 }
 
@@ -818,7 +818,7 @@ blk_trace_request_get_cgid(struct request *rq)
  *
  **/
 static void blk_add_trace_rq(struct request *rq, blk_status_t error,
-			     unsigned int nr_bytes, u32 what, u64 cgid)
+			     unsigned int nr_bytes, u64 what, u64 cgid)
 {
 	struct blk_trace *bt;
 
@@ -882,7 +882,7 @@ static void blk_add_trace_rq_complete(void *ignore, struct request *rq,
  *
  **/
 static void blk_add_trace_bio(struct request_queue *q, struct bio *bio,
-			      u32 what, int error)
+			      u64 what, int error)
 {
 	struct blk_trace *bt;
 
@@ -948,7 +948,7 @@ static void blk_add_trace_unplug(void *ignore, struct request_queue *q,
 	bt = rcu_dereference(q->blk_trace);
 	if (bt) {
 		__be64 rpdu = cpu_to_be64(depth);
-		u32 what;
+		u64 what;
 
 		if (explicit)
 			what = BLK_TA_UNPLUG_IO;
-- 
2.51.0

[PATCH v3 06/16] blktrace: split do_blk_trace_setup into two functions

[Johannes Thumshirn]

Split do_blk_trace_setup into two functions, this is done to prepare for
an incoming new BLKTRACESETUP2 ioctl(2) which can receive extended
parameters from user-space.

Also move the size verification logic to the callers in preparation for
using a new internal structure later.

Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 94 ++++++++++++++++++++++++-----------------
 1 file changed, 56 insertions(+), 38 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 15d6788700ca..df90422ae613 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -518,9 +518,10 @@ static void blk_trace_setup_lba(struct blk_trace *bt,
 /*
  * Setup everything required to start tracing
  */
-static int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
-			      struct block_device *bdev,
-			      struct blk_user_trace_setup *buts)
+static struct blk_trace *blk_trace_setup_prepare(struct request_queue *q,
+						 char *name, dev_t dev,
+						 u32 buf_size, u32 buf_nr,
+						 struct block_device *bdev)
 {
 	struct blk_trace *bt = NULL;
 	struct dentry *dir = NULL;
@@ -528,31 +529,19 @@ static int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 
 	lockdep_assert_held(&q->debugfs_mutex);
 
-	if (!buts->buf_size || !buts->buf_nr)
-		return -EINVAL;
-
-	strscpy_pad(buts->name, name, BLKTRACE_BDEV_SIZE);
-
-	/*
-	 * some device names have larger paths - convert the slashes
-	 * to underscores for this to work as expected
-	 */
-	strreplace(buts->name, '/', '_');
-
 	/*
 	 * bdev can be NULL, as with scsi-generic, this is a helpful as
 	 * we can be.
 	 */
 	if (rcu_dereference_protected(q->blk_trace,
 				      lockdep_is_held(&q->debugfs_mutex))) {
-		pr_warn("Concurrent blktraces are not allowed on %s\n",
-			buts->name);
-		return -EBUSY;
+		pr_warn("Concurrent blktraces are not allowed on %s\n", name);
+		return ERR_PTR(-EBUSY);
 	}
 
 	bt = kzalloc(sizeof(*bt), GFP_KERNEL);
 	if (!bt)
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 
 	ret = -ENOMEM;
 	bt->sequence = alloc_percpu(unsigned long);
@@ -572,7 +561,7 @@ static int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	if (bdev && !bdev_is_partition(bdev))
 		dir = q->debugfs_dir;
 	else
-		bt->dir = dir = debugfs_create_dir(buts->name, blk_debugfs_root);
+		bt->dir = dir = debugfs_create_dir(name, blk_debugfs_root);
 
 	/*
 	 * As blktrace relies on debugfs for its interface the debugfs directory
@@ -580,8 +569,7 @@ static int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	 * files or directories.
 	 */
 	if (IS_ERR_OR_NULL(dir)) {
-		pr_warn("debugfs_dir not present for %s so skipping\n",
-			buts->name);
+		pr_warn("debugfs_dir not present for %s so skipping\n", name);
 		ret = -ENOENT;
 		goto err;
 	}
@@ -593,17 +581,38 @@ static int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	debugfs_create_file("dropped", 0444, dir, bt, &blk_dropped_fops);
 	debugfs_create_file("msg", 0222, dir, bt, &blk_msg_fops);
 
-	bt->rchan = relay_open("trace", dir, buts->buf_size,
-				buts->buf_nr, &blk_relay_callbacks, bt);
+	bt->rchan = relay_open("trace", dir, buf_size, buf_nr,
+			       &blk_relay_callbacks, bt);
 	if (!bt->rchan)
 		goto err;
 
+	blk_trace_setup_lba(bt, bdev);
+
+	return bt;
+
+err:
+	blk_trace_free(q, bt);
+
+	return ERR_PTR(ret);
+}
+
+static void blk_trace_setup_finalize(struct request_queue *q,
+				     char *name, struct blk_trace *bt,
+				     struct blk_user_trace_setup *buts)
+
+{
+	strscpy_pad(buts->name, name, BLKTRACE_BDEV_SIZE);
+
+	/*
+	 * some device names have larger paths - convert the slashes
+	 * to underscores for this to work as expected
+	 */
+	strreplace(buts->name, '/', '_');
+
 	bt->act_mask = buts->act_mask;
 	if (!bt->act_mask)
 		bt->act_mask = (u16) -1;
 
-	blk_trace_setup_lba(bt, bdev);
-
 	/* overwrite with user settings */
 	if (buts->start_lba)
 		bt->start_lba = buts->start_lba;
@@ -615,12 +624,6 @@ static int do_blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 
 	rcu_assign_pointer(q->blk_trace, bt);
 	get_probe_ref();
-
-	ret = 0;
-err:
-	if (ret)
-		blk_trace_free(q, bt);
-	return ret;
 }
 
 int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
@@ -628,17 +631,25 @@ int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 		    char __user *arg)
 {
 	struct blk_user_trace_setup buts;
+	struct blk_trace *bt;
 	int ret;
 
 	ret = copy_from_user(&buts, arg, sizeof(buts));
 	if (ret)
 		return -EFAULT;
 
+	if (!buts.buf_size || !buts.buf_nr)
+		return -EINVAL;
+
 	mutex_lock(&q->debugfs_mutex);
-	ret = do_blk_trace_setup(q, name, dev, bdev, &buts);
+	bt = blk_trace_setup_prepare(q, name, dev, buts.buf_size, buts.buf_nr,
+				     bdev);
+	if (IS_ERR(bt)) {
+		mutex_unlock(&q->debugfs_mutex);
+		return PTR_ERR(bt);
+	}
+	blk_trace_setup_finalize(q, name, bt, &buts);
 	mutex_unlock(&q->debugfs_mutex);
-	if (ret)
-		return ret;
 
 	if (copy_to_user(arg, &buts, sizeof(buts))) {
 		blk_trace_remove(q);
@@ -655,11 +666,14 @@ static int compat_blk_trace_setup(struct request_queue *q, char *name,
 {
 	struct blk_user_trace_setup buts;
 	struct compat_blk_user_trace_setup cbuts;
-	int ret;
+	struct blk_trace *bt;
 
 	if (copy_from_user(&cbuts, arg, sizeof(cbuts)))
 		return -EFAULT;
 
+	if (!cbuts.buf_size || !cbuts.buf_nr)
+		return -EINVAL;
+
 	buts = (struct blk_user_trace_setup) {
 		.act_mask = cbuts.act_mask,
 		.buf_size = cbuts.buf_size,
@@ -670,10 +684,14 @@ static int compat_blk_trace_setup(struct request_queue *q, char *name,
 	};
 
 	mutex_lock(&q->debugfs_mutex);
-	ret = do_blk_trace_setup(q, name, dev, bdev, &buts);
+	bt = blk_trace_setup_prepare(q, name, dev, buts.buf_size, buts.buf_nr,
+				     bdev);
+	if (IS_ERR(bt)) {
+		mutex_unlock(&q->debugfs_mutex);
+		return PTR_ERR(bt);
+	}
+	blk_trace_setup_finalize(q, name, bt, &buts);
 	mutex_unlock(&q->debugfs_mutex);
-	if (ret)
-		return ret;
 
 	if (copy_to_user(arg, &buts.name, ARRAY_SIZE(buts.name))) {
 		blk_trace_remove(q);
-- 
2.51.0

[PATCH v3 07/16] blktrace: add definitions for blk_user_trace_setup2

[Johannes Thumshirn]

Add definitions for a version 2 of the blk_user_trace_setup ioctl. This
new ioctl will enable a different struct layout of the binary data passed
to user-space when using a new version of the blktrace utility requesting
the new struct layout.

Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 include/uapi/linux/blktrace_api.h | 16 ++++++++++++++++
 include/uapi/linux/fs.h           |  1 +
 kernel/trace/blktrace.c           |  3 +++
 3 files changed, 20 insertions(+)

diff --git a/include/uapi/linux/blktrace_api.h b/include/uapi/linux/blktrace_api.h
index 1bfb635e309b..a6958708d477 100644
--- a/include/uapi/linux/blktrace_api.h
+++ b/include/uapi/linux/blktrace_api.h
@@ -129,6 +129,7 @@ enum {
 };
 
 #define BLKTRACE_BDEV_SIZE	32
+#define BLKTRACE_BDEV_SIZE2	64
 
 /*
  * User setup structure passed with BLKTRACESETUP
@@ -143,4 +144,19 @@ struct blk_user_trace_setup {
 	__u32 pid;
 };
 
+/*
+ * User setup structure passed with BLKTRACESETUP2
+ */
+struct blk_user_trace_setup2 {
+	char name[BLKTRACE_BDEV_SIZE2];		/* output */
+	__u64 act_mask;				/* input */
+	__u32 buf_size;				/* input */
+	__u32 buf_nr;				/* input */
+	__u64 start_lba;
+	__u64 end_lba;
+	__u32 pid;
+	__u32 flags;		/* currently unused */
+	__u64 reserved[11];
+};
+
 #endif /* _UAPIBLKTRACE_H */
diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
index beb4c2d1e41c..957ce3343a4f 100644
--- a/include/uapi/linux/fs.h
+++ b/include/uapi/linux/fs.h
@@ -300,6 +300,7 @@ struct file_attr {
 #define BLKGETDISKSEQ _IOR(0x12,128,__u64)
 /* 130-136 are used by zoned block device ioctls (uapi/linux/blkzoned.h) */
 /* 137-141 are used by blk-crypto ioctls (uapi/linux/blk-crypto.h) */
+#define BLKTRACESETUP2 _IOWR(0x12, 142, struct blk_user_trace_setup2)
 
 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index df90422ae613..c31b8f433116 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -1601,6 +1601,9 @@ static int __init init_blk_tracer(void)
 		return 1;
 	}
 
+	BUILD_BUG_ON(__alignof__(struct blk_user_trace_setup2) %
+		     __alignof__(long));
+
 	return 0;
 }
 
-- 
2.51.0

[PATCH v3 08/16] blktrace: pass blk_user_trace2 to setup functions

[Johannes Thumshirn]

Pass struct blk_user_trace_setup2 to blktrace_setup_finalize(). This
prepares for the incoming extension of the blktrace protocol with a 64bit
act_mask.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 include/linux/blktrace_api.h |  3 ++-
 kernel/trace/blktrace.c      | 31 ++++++++++++++++++++++---------
 2 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/include/linux/blktrace_api.h b/include/linux/blktrace_api.h
index 122c62e561fc..05c8754456aa 100644
--- a/include/linux/blktrace_api.h
+++ b/include/linux/blktrace_api.h
@@ -14,11 +14,12 @@
 #include <linux/sysfs.h>
 
 struct blk_trace {
+	int version;
 	int trace_state;
 	struct rchan *rchan;
 	unsigned long __percpu *sequence;
 	unsigned char __percpu *msg_data;
-	u16 act_mask;
+	u64 act_mask;
 	u64 start_lba;
 	u64 end_lba;
 	u32 pid;
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index c31b8f433116..d1532df84cc8 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -597,11 +597,12 @@ static struct blk_trace *blk_trace_setup_prepare(struct request_queue *q,
 }
 
 static void blk_trace_setup_finalize(struct request_queue *q,
-				     char *name, struct blk_trace *bt,
-				     struct blk_user_trace_setup *buts)
+				     char *name, int version,
+				     struct blk_trace *bt,
+				     struct blk_user_trace_setup2 *buts)
 
 {
-	strscpy_pad(buts->name, name, BLKTRACE_BDEV_SIZE);
+	strscpy_pad(buts->name, name, BLKTRACE_BDEV_SIZE2);
 
 	/*
 	 * some device names have larger paths - convert the slashes
@@ -609,6 +610,7 @@ static void blk_trace_setup_finalize(struct request_queue *q,
 	 */
 	strreplace(buts->name, '/', '_');
 
+	bt->version = version;
 	bt->act_mask = buts->act_mask;
 	if (!bt->act_mask)
 		bt->act_mask = (u16) -1;
@@ -630,6 +632,7 @@ int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 		    struct block_device *bdev,
 		    char __user *arg)
 {
+	struct blk_user_trace_setup2 buts2;
 	struct blk_user_trace_setup buts;
 	struct blk_trace *bt;
 	int ret;
@@ -641,6 +644,15 @@ int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 	if (!buts.buf_size || !buts.buf_nr)
 		return -EINVAL;
 
+	buts2 = (struct blk_user_trace_setup2) {
+		.act_mask = buts.act_mask,
+		.buf_size = buts.buf_size,
+		.buf_nr = buts.buf_nr,
+		.start_lba = buts.start_lba,
+		.end_lba = buts.end_lba,
+		.pid = buts.pid,
+	};
+
 	mutex_lock(&q->debugfs_mutex);
 	bt = blk_trace_setup_prepare(q, name, dev, buts.buf_size, buts.buf_nr,
 				     bdev);
@@ -648,7 +660,8 @@ int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 		mutex_unlock(&q->debugfs_mutex);
 		return PTR_ERR(bt);
 	}
-	blk_trace_setup_finalize(q, name, bt, &buts);
+	blk_trace_setup_finalize(q, name, 1, bt, &buts2);
+	strcpy(buts.name, buts2.name);
 	mutex_unlock(&q->debugfs_mutex);
 
 	if (copy_to_user(arg, &buts, sizeof(buts))) {
@@ -664,7 +677,7 @@ static int compat_blk_trace_setup(struct request_queue *q, char *name,
 				  dev_t dev, struct block_device *bdev,
 				  char __user *arg)
 {
-	struct blk_user_trace_setup buts;
+	struct blk_user_trace_setup2 buts2;
 	struct compat_blk_user_trace_setup cbuts;
 	struct blk_trace *bt;
 
@@ -674,7 +687,7 @@ static int compat_blk_trace_setup(struct request_queue *q, char *name,
 	if (!cbuts.buf_size || !cbuts.buf_nr)
 		return -EINVAL;
 
-	buts = (struct blk_user_trace_setup) {
+	buts2 = (struct blk_user_trace_setup2) {
 		.act_mask = cbuts.act_mask,
 		.buf_size = cbuts.buf_size,
 		.buf_nr = cbuts.buf_nr,
@@ -684,16 +697,16 @@ static int compat_blk_trace_setup(struct request_queue *q, char *name,
 	};
 
 	mutex_lock(&q->debugfs_mutex);
-	bt = blk_trace_setup_prepare(q, name, dev, buts.buf_size, buts.buf_nr,
+	bt = blk_trace_setup_prepare(q, name, dev, buts2.buf_size, buts2.buf_nr,
 				     bdev);
 	if (IS_ERR(bt)) {
 		mutex_unlock(&q->debugfs_mutex);
 		return PTR_ERR(bt);
 	}
-	blk_trace_setup_finalize(q, name, bt, &buts);
+	blk_trace_setup_finalize(q, name, 1, bt, &buts2);
 	mutex_unlock(&q->debugfs_mutex);
 
-	if (copy_to_user(arg, &buts.name, ARRAY_SIZE(buts.name))) {
+	if (copy_to_user(arg, &buts2.name, ARRAY_SIZE(buts2.name))) {
 		blk_trace_remove(q);
 		return -EFAULT;
 	}
-- 
2.51.0

[PATCH v3 09/16] blktrace: add definitions for struct blk_io_trace2

[Johannes Thumshirn]

Add definitions for the extended version of the blktrace protocol using a
wider action type to be able to record new actions in the kernel.

Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 include/uapi/linux/blktrace_api.h | 16 ++++++++++++++++
 kernel/trace/blktrace.c           |  1 +
 2 files changed, 17 insertions(+)

diff --git a/include/uapi/linux/blktrace_api.h b/include/uapi/linux/blktrace_api.h
index a6958708d477..3a771b9802aa 100644
--- a/include/uapi/linux/blktrace_api.h
+++ b/include/uapi/linux/blktrace_api.h
@@ -94,6 +94,7 @@ enum blktrace_notify {
 
 #define BLK_IO_TRACE_MAGIC	0x65617400
 #define BLK_IO_TRACE_VERSION	0x07
+#define BLK_IO_TRACE2_VERSION	0x08
 
 /*
  * The trace itself
@@ -113,6 +114,21 @@ struct blk_io_trace {
 	/* cgroup id will be stored here if exists */
 };
 
+struct blk_io_trace2 {
+	__u32 magic;		/* MAGIC << 8 | BLK_IO_TRACE2_VERSION */
+	__u32 sequence;		/* event number */
+	__u64 time;		/* in nanoseconds */
+	__u64 sector;		/* disk offset */
+	__u32 bytes;		/* transfer length */
+	__u32 pid;		/* who did it */
+	__u64 action;		/* what happened */
+	__u32 device;		/* device number */
+	__u32 cpu;		/* on what cpu did it happen */
+	__u16 error;		/* completion error */
+	__u16 pdu_len;		/* length of data after this trace */
+	__u8 pad[12];
+	/* cgroup id will be stored here if exists */
+};
 /*
  * The remap event
  */
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index d1532df84cc8..185f19c9f772 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -1616,6 +1616,7 @@ static int __init init_blk_tracer(void)
 
 	BUILD_BUG_ON(__alignof__(struct blk_user_trace_setup2) %
 		     __alignof__(long));
+	BUILD_BUG_ON(__alignof__(struct blk_io_trace2) % __alignof__(long));
 
 	return 0;
 }
-- 
2.51.0

[PATCH v3 10/16] blktrace: differentiate between blk_io_trace versions

[Johannes Thumshirn]

Differentiate between blk_io_trace and blk_io_trace2 when relaying to
user-space depending on which version has been requested by the blktrace
utility.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 62 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 59 insertions(+), 3 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 185f19c9f772..074a7d77158c 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -91,7 +91,29 @@ static void record_blktrace_event(struct blk_io_trace *t, pid_t pid, int cpu,
 		memcpy((void *)t + sizeof(*t) + cgid_len, pdu_data, pdu_len);
 }
 
-static void relay_blktrace_event(struct blk_trace *bt, unsigned long sequence,
+static void record_blktrace_event2(struct blk_io_trace2 *t2, pid_t pid, int cpu,
+				   sector_t sector, int bytes, u64 what,
+				   dev_t dev, int error, u64 cgid,
+				   ssize_t cgid_len, void *pdu_data,
+				   int pdu_len)
+{
+	t2->pid = pid;
+	t2->cpu = cpu;
+
+	t2->sector = sector;
+	t2->bytes = bytes;
+	t2->action = what;
+	t2->device = dev;
+	t2->error = error;
+	t2->pdu_len = pdu_len + cgid_len;
+
+	if (cgid_len)
+		memcpy((void *)t2 + sizeof(*t2), &cgid, cgid_len);
+	if (pdu_len)
+		memcpy((void *)t2 + sizeof(*t2) + cgid_len, pdu_data, pdu_len);
+}
+
+static void relay_blktrace_event1(struct blk_trace *bt, unsigned long sequence,
 				 pid_t pid, int cpu, sector_t sector, int bytes,
 				 u64 what, int error, u64 cgid,
 				 ssize_t cgid_len, void *pdu_data, int pdu_len)
@@ -111,6 +133,40 @@ static void relay_blktrace_event(struct blk_trace *bt, unsigned long sequence,
 			      cgid, cgid_len, pdu_data, pdu_len);
 }
 
+static void relay_blktrace_event2(struct blk_trace *bt, unsigned long sequence,
+				  pid_t pid, int cpu, sector_t sector,
+				  int bytes, u64 what, int error, u64 cgid,
+				  ssize_t cgid_len, void *pdu_data, int pdu_len)
+{
+	struct blk_io_trace2 *t;
+	size_t trace_len = sizeof(struct blk_io_trace2) + pdu_len + cgid_len;
+
+	t = relay_reserve(bt->rchan, trace_len);
+	if (!t)
+		return;
+
+	t->magic = BLK_IO_TRACE_MAGIC | BLK_IO_TRACE2_VERSION;
+	t->sequence = sequence;
+	t->time = ktime_to_ns(ktime_get());
+
+	record_blktrace_event2(t, pid, cpu, sector, bytes, what, bt->dev, error,
+			       cgid, cgid_len, pdu_data, pdu_len);
+}
+
+static void relay_blktrace_event(struct blk_trace *bt, unsigned long sequence,
+				 pid_t pid, int cpu, sector_t sector, int bytes,
+				 u64 what, int error, u64 cgid,
+				 ssize_t cgid_len, void *pdu_data, int pdu_len)
+{
+	if (bt->version == 2)
+		return relay_blktrace_event2(bt, sequence, pid, cpu, sector,
+					     bytes, what, error, cgid, cgid_len,
+					     pdu_data, pdu_len);
+	return relay_blktrace_event1(bt, sequence, pid, cpu, sector, bytes,
+				     lower_32_bits(what), error, cgid, cgid_len,
+				     pdu_data, pdu_len);
+}
+
 /*
  * Send out a notify message.
  */
@@ -146,8 +202,8 @@ static void trace_note(struct blk_trace *bt, pid_t pid, u64 action,
 	if (!bt->rchan)
 		return;
 
-	relay_blktrace_event(bt, 0, pid, cpu, 0, 0, action, 0, cgid,
-			     cgid_len, (void *)data, len);
+	relay_blktrace_event(bt, 0, pid, cpu, 0, 0, action, 0, cgid, cgid_len,
+			     (void *)data, len);
 }
 
 /*
-- 
2.51.0

[PATCH v3 11/16] blktrace: move trace_note to blk_io_trace2

[Johannes Thumshirn]

Move trace_note() to the new blk_io_trace2 infrastructure.

Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 074a7d77158c..756d2c7dfdf2 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -173,18 +173,18 @@ static void relay_blktrace_event(struct blk_trace *bt, unsigned long sequence,
 static void trace_note(struct blk_trace *bt, pid_t pid, u64 action,
 		       const void *data, size_t len, u64 cgid)
 {
-	struct blk_io_trace *t;
 	struct ring_buffer_event *event = NULL;
 	struct trace_buffer *buffer = NULL;
 	unsigned int trace_ctx = 0;
 	int cpu = smp_processor_id();
 	bool blk_tracer = blk_tracer_enabled;
 	ssize_t cgid_len = cgid ? sizeof(cgid) : 0;
-	size_t trace_len;
 
-	trace_len = sizeof(*t) + cgid_len + len;
 	action = lower_32_bits(action | (cgid ? __BLK_TN_CGROUP : 0));
 	if (blk_tracer) {
+		struct blk_io_trace2 *t;
+		size_t trace_len = sizeof(*t) + cgid_len + len;
+
 		buffer = blk_tr->array_buffer.buffer;
 		trace_ctx = tracing_gen_ctx_flags(0);
 		event = trace_buffer_lock_reserve(buffer, TRACE_BLK,
@@ -192,9 +192,9 @@ static void trace_note(struct blk_trace *bt, pid_t pid, u64 action,
 		if (!event)
 			return;
 		t = ring_buffer_event_data(event);
-		record_blktrace_event(t, pid, cpu, 0, 0,
-				      action, bt->dev, 0, cgid, cgid_len,
-				      (void *)data, len);
+		record_blktrace_event2(t, pid, cpu, 0, 0,
+				       action, bt->dev, 0, cgid, cgid_len,
+				       (void *)data, len);
 		trace_buffer_unlock_commit(blk_tr, buffer, event, trace_ctx);
 		return;
 	}
@@ -359,7 +359,7 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 
 		buffer = blk_tr->array_buffer.buffer;
 		trace_ctx = tracing_gen_ctx_flags(0);
-		trace_len = sizeof(struct blk_io_trace) + pdu_len + cgid_len;
+		trace_len = sizeof(struct blk_io_trace2) + pdu_len + cgid_len;
 		event = trace_buffer_lock_reserve(buffer, TRACE_BLK,
 						  trace_len, trace_ctx);
 		if (!event)
-- 
2.51.0

[PATCH v3 12/16] blktrace: move ftrace blk_io_tracer to blk_io_trace2

[Johannes Thumshirn]

Move ftrace's blk_io_tracer to the new blk_io_trace2 infrastructure.

Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 kernel/trace/blktrace.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 756d2c7dfdf2..8ffb218e9fb7 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -1224,7 +1224,7 @@ static void blk_unregister_tracepoints(void)
  * struct blk_io_tracer formatting routines
  */
 
-static void fill_rwbs(char *rwbs, const struct blk_io_trace *t)
+static void fill_rwbs(char *rwbs, const struct blk_io_trace2 *t)
 {
 	int i = 0;
 	int tc = t->action >> BLK_TC_SHIFT;
@@ -1259,9 +1259,9 @@ static void fill_rwbs(char *rwbs, const struct blk_io_trace *t)
 }
 
 static inline
-const struct blk_io_trace *te_blk_io_trace(const struct trace_entry *ent)
+const struct blk_io_trace2 *te_blk_io_trace(const struct trace_entry *ent)
 {
-	return (const struct blk_io_trace *)ent;
+	return (const struct blk_io_trace2 *)ent;
 }
 
 static inline const void *pdu_start(const struct trace_entry *ent, bool has_cg)
@@ -1320,7 +1320,7 @@ static void blk_log_action_classic(struct trace_iterator *iter, const char *act,
 	unsigned long long ts  = iter->ts;
 	unsigned long nsec_rem = do_div(ts, NSEC_PER_SEC);
 	unsigned secs	       = (unsigned long)ts;
-	const struct blk_io_trace *t = te_blk_io_trace(iter->ent);
+	const struct blk_io_trace2 *t = te_blk_io_trace(iter->ent);
 
 	fill_rwbs(rwbs, t);
 
@@ -1334,7 +1334,7 @@ static void blk_log_action(struct trace_iterator *iter, const char *act,
 	bool has_cg)
 {
 	char rwbs[RWBS_LEN];
-	const struct blk_io_trace *t = te_blk_io_trace(iter->ent);
+	const struct blk_io_trace2 *t = te_blk_io_trace(iter->ent);
 
 	fill_rwbs(rwbs, t);
 	if (has_cg) {
@@ -1555,7 +1555,7 @@ static enum print_line_t print_one_line(struct trace_iterator *iter,
 {
 	struct trace_array *tr = iter->tr;
 	struct trace_seq *s = &iter->seq;
-	const struct blk_io_trace *t;
+	const struct blk_io_trace2 *t;
 	u16 what;
 	bool long_act;
 	blk_log_action_t *log_action;
@@ -1592,8 +1592,8 @@ static enum print_line_t blk_trace_event_print(struct trace_iterator *iter,
 static void blk_trace_synthesize_old_trace(struct trace_iterator *iter)
 {
 	struct trace_seq *s = &iter->seq;
-	struct blk_io_trace *t = (struct blk_io_trace *)iter->ent;
-	const int offset = offsetof(struct blk_io_trace, sector);
+	struct blk_io_trace2 *t = (struct blk_io_trace2 *)iter->ent;
+	const int offset = offsetof(struct blk_io_trace2, sector);
 	struct blk_io_trace old = {
 		.magic	  = BLK_IO_TRACE_MAGIC | BLK_IO_TRACE_VERSION,
 		.time     = iter->ts,
-- 
2.51.0

[PATCH v3 13/16] blktrace: add block trace commands for zone operations

[Johannes Thumshirn]

Add block trace commands for zone operations.

Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 include/uapi/linux/blktrace_api.h | 13 +++++++++++--
 kernel/trace/blktrace.c           | 25 +++++++++++++++++++++----
 2 files changed, 32 insertions(+), 6 deletions(-)

diff --git a/include/uapi/linux/blktrace_api.h b/include/uapi/linux/blktrace_api.h
index 3a771b9802aa..925f78af939e 100644
--- a/include/uapi/linux/blktrace_api.h
+++ b/include/uapi/linux/blktrace_api.h
@@ -26,11 +26,20 @@ enum blktrace_cat {
 	BLK_TC_DRV_DATA	= 1 << 14,	/* binary per-driver data */
 	BLK_TC_FUA	= 1 << 15,	/* fua requests */
 
-	BLK_TC_END	= 1 << 15,	/* we've run out of bits! */
+	BLK_TC_END_V1	= 1 << 15,	/* we've run out of bits! */
+
+	BLK_TC_ZONE_APPEND	= 1ull << 16,  	/* zone append */
+	BLK_TC_ZONE_RESET	= 1ull << 17,	/* zone reset */
+	BLK_TC_ZONE_RESET_ALL	= 1ull << 18,	/* zone reset all */
+	BLK_TC_ZONE_FINISH	= 1ull << 19,	/* zone finish */
+	BLK_TC_ZONE_OPEN	= 1ull << 20,	/* zone open */
+	BLK_TC_ZONE_CLOSE	= 1ull << 21,	/* zone close */
+
+	BLK_TC_END_V2		= 1ull << 21,
 };
 
 #define BLK_TC_SHIFT		(16)
-#define BLK_TC_ACT(act)		((act) << BLK_TC_SHIFT)
+#define BLK_TC_ACT(act)		((u64)(act) << BLK_TC_SHIFT)
 
 /*
  * Basic trace actions
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 8ffb218e9fb7..90b225c5bad7 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -163,8 +163,8 @@ static void relay_blktrace_event(struct blk_trace *bt, unsigned long sequence,
 					     bytes, what, error, cgid, cgid_len,
 					     pdu_data, pdu_len);
 	return relay_blktrace_event1(bt, sequence, pid, cpu, sector, bytes,
-				     lower_32_bits(what), error, cgid, cgid_len,
-				     pdu_data, pdu_len);
+				     what, error, cgid, cgid_len, pdu_data,
+				     pdu_len);
 }
 
 /*
@@ -342,6 +342,24 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 	case REQ_OP_FLUSH:
 		what |= BLK_TC_ACT(BLK_TC_FLUSH);
 		break;
+	case REQ_OP_ZONE_APPEND:
+		what |= BLK_TC_ACT(BLK_TC_ZONE_APPEND);
+		break;
+	case REQ_OP_ZONE_RESET:
+		what |= BLK_TC_ACT(BLK_TC_ZONE_RESET);
+		break;
+	case REQ_OP_ZONE_RESET_ALL:
+		what |= BLK_TC_ACT(BLK_TC_ZONE_RESET_ALL);
+		break;
+	case REQ_OP_ZONE_FINISH:
+		what |= BLK_TC_ACT(BLK_TC_ZONE_FINISH);
+		break;
+	case REQ_OP_ZONE_OPEN:
+		what |= BLK_TC_ACT(BLK_TC_ZONE_OPEN);
+		break;
+	case REQ_OP_ZONE_CLOSE:
+		what |= BLK_TC_ACT(BLK_TC_ZONE_CLOSE);
+		break;
 	default:
 		break;
 	}
@@ -386,8 +404,7 @@ static void __blk_add_trace(struct blk_trace *bt, sector_t sector, int bytes,
 	sequence = per_cpu_ptr(bt->sequence, cpu);
 	(*sequence)++;
 	relay_blktrace_event(bt, *sequence, pid, cpu, sector, bytes,
-			     lower_32_bits(what), error, cgid, cgid_len,
-			     pdu_data, pdu_len);
+			     what, error, cgid, cgid_len, pdu_data, pdu_len);
 	local_irq_restore(flags);
 }
 
-- 
2.51.0

[PATCH v3 14/16] blktrace: expose ZONE APPEND completions to blktrace

[Johannes Thumshirn]

Expose ZONE APPEND completions as a block trace completion action to
blktrace.

As tracing of zoned block commands needs the upper 32bit of the widened
64bit action, only add traces to blktrace if user-space has requested
version 2 of the blktrace protocol.

Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 include/uapi/linux/blktrace_api.h |  3 +++
 kernel/trace/blktrace.c           | 21 +++++++++++++++++++++
 2 files changed, 24 insertions(+)

diff --git a/include/uapi/linux/blktrace_api.h b/include/uapi/linux/blktrace_api.h
index 925f78af939e..bf41e34df3c8 100644
--- a/include/uapi/linux/blktrace_api.h
+++ b/include/uapi/linux/blktrace_api.h
@@ -97,6 +97,9 @@ enum blktrace_notify {
 #define BLK_TA_ABORT		(__BLK_TA_ABORT | BLK_TC_ACT(BLK_TC_QUEUE))
 #define BLK_TA_DRV_DATA	(__BLK_TA_DRV_DATA | BLK_TC_ACT(BLK_TC_DRV_DATA))
 
+#define BLK_TA_ZONE_APPEND	(__BLK_TA_COMPLETE |\
+				 BLK_TC_ACT(BLK_TC_ZONE_APPEND))
+
 #define BLK_TN_PROCESS		(__BLK_TN_PROCESS | BLK_TC_ACT(BLK_TC_NOTIFY))
 #define BLK_TN_TIMESTAMP	(__BLK_TN_TIMESTAMP | BLK_TC_ACT(BLK_TC_NOTIFY))
 #define BLK_TN_MESSAGE		(__BLK_TN_MESSAGE | BLK_TC_ACT(BLK_TC_NOTIFY))
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 90b225c5bad7..e8d562ad21f9 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -974,6 +974,22 @@ static void blk_add_trace_rq_complete(void *ignore, struct request *rq,
 			 blk_trace_request_get_cgid(rq));
 }
 
+static void blk_add_trace_zone_update_request(void *ignore, struct request *rq)
+{
+	struct blk_trace *bt;
+
+	rcu_read_lock();
+	bt = rcu_dereference(rq->q->blk_trace);
+	if (likely(!bt) || bt->version < 2) {
+		rcu_read_unlock();
+		return;
+	}
+	rcu_read_unlock();
+
+	blk_add_trace_rq(rq, 0, blk_rq_bytes(rq), BLK_TA_ZONE_APPEND,
+			 blk_trace_request_get_cgid(rq));
+}
+
 /**
  * blk_add_trace_bio - Add a trace for a bio oriented action
  * @q:		queue the io is for
@@ -1204,6 +1220,9 @@ static void blk_register_tracepoints(void)
 	WARN_ON(ret);
 	ret = register_trace_block_getrq(blk_add_trace_getrq, NULL);
 	WARN_ON(ret);
+	ret = register_trace_blk_zone_append_update_request_bio(
+		blk_add_trace_zone_update_request, NULL);
+	WARN_ON(ret);
 	ret = register_trace_block_plug(blk_add_trace_plug, NULL);
 	WARN_ON(ret);
 	ret = register_trace_block_unplug(blk_add_trace_unplug, NULL);
@@ -1223,6 +1242,8 @@ static void blk_unregister_tracepoints(void)
 	unregister_trace_block_split(blk_add_trace_split, NULL);
 	unregister_trace_block_unplug(blk_add_trace_unplug, NULL);
 	unregister_trace_block_plug(blk_add_trace_plug, NULL);
+	unregister_trace_blk_zone_append_update_request_bio(
+		blk_add_trace_zone_update_request, NULL);
 	unregister_trace_block_getrq(blk_add_trace_getrq, NULL);
 	unregister_trace_block_bio_queue(blk_add_trace_bio_queue, NULL);
 	unregister_trace_block_bio_frontmerge(blk_add_trace_bio_frontmerge, NULL);
-- 
2.51.0

[PATCH v3 15/16] blktrace: trace zone write plugging operations

[Johannes Thumshirn]

Trace zone write plugging operations on block devices.

As tracing of zoned block commands needs the upper 32bit of the widened
64bit action, only add traces to blktrace if user-space has requested
version 2 of the blktrace protocol.

Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 include/uapi/linux/blktrace_api.h |  5 ++++
 kernel/trace/blktrace.c           | 39 +++++++++++++++++++++++++++++++
 2 files changed, 44 insertions(+)

diff --git a/include/uapi/linux/blktrace_api.h b/include/uapi/linux/blktrace_api.h
index bf41e34df3c8..63de5fb07553 100644
--- a/include/uapi/linux/blktrace_api.h
+++ b/include/uapi/linux/blktrace_api.h
@@ -62,6 +62,8 @@ enum blktrace_act {
 	__BLK_TA_REMAP,			/* bio was remapped */
 	__BLK_TA_ABORT,			/* request aborted */
 	__BLK_TA_DRV_DATA,		/* driver-specific binary data */
+	__BLK_TA_ZONE_PLUG,		/* zone write plug was plugged */
+	__BLK_TA_ZONE_UNPLUG,		/* zone write plug was unplugged */
 	__BLK_TA_CGROUP = 1 << 8,	/* from a cgroup*/
 };
 
@@ -99,6 +101,9 @@ enum blktrace_notify {
 
 #define BLK_TA_ZONE_APPEND	(__BLK_TA_COMPLETE |\
 				 BLK_TC_ACT(BLK_TC_ZONE_APPEND))
+#define BLK_TA_ZONE_PLUG	(__BLK_TA_ZONE_PLUG | BLK_TC_ACT(BLK_TC_QUEUE))
+#define BLK_TA_ZONE_UNPLUG	(__BLK_TA_ZONE_UNPLUG |\
+				 BLK_TC_ACT(BLK_TC_QUEUE))
 
 #define BLK_TN_PROCESS		(__BLK_TN_PROCESS | BLK_TC_ACT(BLK_TC_NOTIFY))
 #define BLK_TN_TIMESTAMP	(__BLK_TN_TIMESTAMP | BLK_TC_ACT(BLK_TC_NOTIFY))
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index e8d562ad21f9..d659fed0650d 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -1080,6 +1080,37 @@ static void blk_add_trace_unplug(void *ignore, struct request_queue *q,
 	rcu_read_unlock();
 }
 
+static void blk_add_trace_zone_plug(void *ignore, struct request_queue *q,
+				    unsigned int zno, sector_t sector,
+				    unsigned int sectors)
+{
+	struct blk_trace *bt;
+
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
+	if (bt && bt->version >= 2)
+		__blk_add_trace(bt, sector, sectors << SECTOR_SHIFT, 0,
+				BLK_TA_ZONE_PLUG, 0, 0, NULL, 0);
+	rcu_read_unlock();
+
+	return;
+}
+
+static void blk_add_trace_zone_unplug(void *ignore, struct request_queue *q,
+				      unsigned int zno, sector_t sector,
+				      unsigned int sectors)
+{
+	struct blk_trace *bt;
+
+	rcu_read_lock();
+	bt = rcu_dereference(q->blk_trace);
+	if (bt && bt->version >= 2)
+		__blk_add_trace(bt, sector, sectors << SECTOR_SHIFT, 0,
+				BLK_TA_ZONE_UNPLUG, 0, 0, NULL, 0);
+	rcu_read_unlock();
+	return;
+}
+
 static void blk_add_trace_split(void *ignore, struct bio *bio, unsigned int pdu)
 {
 	struct request_queue *q = bio->bi_bdev->bd_disk->queue;
@@ -1223,6 +1254,12 @@ static void blk_register_tracepoints(void)
 	ret = register_trace_blk_zone_append_update_request_bio(
 		blk_add_trace_zone_update_request, NULL);
 	WARN_ON(ret);
+	ret = register_trace_disk_zone_wplug_add_bio(blk_add_trace_zone_plug,
+						     NULL);
+	WARN_ON(ret);
+	ret = register_trace_blk_zone_wplug_bio(blk_add_trace_zone_unplug,
+						NULL);
+	WARN_ON(ret);
 	ret = register_trace_block_plug(blk_add_trace_plug, NULL);
 	WARN_ON(ret);
 	ret = register_trace_block_unplug(blk_add_trace_unplug, NULL);
@@ -1242,6 +1279,8 @@ static void blk_unregister_tracepoints(void)
 	unregister_trace_block_split(blk_add_trace_split, NULL);
 	unregister_trace_block_unplug(blk_add_trace_unplug, NULL);
 	unregister_trace_block_plug(blk_add_trace_plug, NULL);
+	unregister_trace_blk_zone_wplug_bio(blk_add_trace_zone_unplug, NULL);
+	unregister_trace_disk_zone_wplug_add_bio(blk_add_trace_zone_plug, NULL);
 	unregister_trace_blk_zone_append_update_request_bio(
 		blk_add_trace_zone_update_request, NULL);
 	unregister_trace_block_getrq(blk_add_trace_getrq, NULL);
-- 
2.51.0

[PATCH v3 16/16] blktrace: handle BLKTRACESETUP2 ioctl

[Johannes Thumshirn]

Handle the BLKTRACESETUP2 ioctl, requesting an extended version of the
blktrace protocol from user-space.

Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
 block/ioctl.c           |  1 +
 kernel/trace/blktrace.c | 35 +++++++++++++++++++++++++++++++++++
 2 files changed, 36 insertions(+)

diff --git a/block/ioctl.c b/block/ioctl.c
index d7489a56b33c..3927ca4707d0 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -691,6 +691,7 @@ long blkdev_ioctl(struct file *file, unsigned cmd, unsigned long arg)
 
 	/* Incompatible alignment on i386 */
 	case BLKTRACESETUP:
+	case BLKTRACESETUP2:
 		return blk_trace_ioctl(bdev, cmd, argp);
 	default:
 		break;
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index d659fed0650d..0719f9b76082 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -745,6 +745,37 @@ int blk_trace_setup(struct request_queue *q, char *name, dev_t dev,
 }
 EXPORT_SYMBOL_GPL(blk_trace_setup);
 
+static int blk_trace_setup2(struct request_queue *q, char *name, dev_t dev,
+			    struct block_device *bdev, char __user *arg)
+{
+	struct blk_user_trace_setup2 buts2;
+	struct blk_trace *bt;
+	int ret;
+
+	ret = copy_from_user(&buts2, arg, sizeof(buts2));
+	if (ret)
+		return -EFAULT;
+
+	if (!buts2.buf_size || !buts2.buf_nr)
+		return -EINVAL;
+
+	mutex_lock(&q->debugfs_mutex);
+	bt = blk_trace_setup_prepare(q, name, dev, buts2.buf_size, buts2.buf_nr,
+				     bdev);
+	if (IS_ERR(bt)) {
+		mutex_unlock(&q->debugfs_mutex);
+		return PTR_ERR(bt);
+	}
+	blk_trace_setup_finalize(q, name, 2, bt, &buts2);
+	mutex_unlock(&q->debugfs_mutex);
+
+	if (copy_to_user(arg, &buts2, sizeof(buts2))) {
+		blk_trace_remove(q);
+		return -EFAULT;
+	}
+	return 0;
+}
+
 #if defined(CONFIG_COMPAT) && defined(CONFIG_X86_64)
 static int compat_blk_trace_setup(struct request_queue *q, char *name,
 				  dev_t dev, struct block_device *bdev,
@@ -835,6 +866,10 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)
 	char b[BDEVNAME_SIZE];
 
 	switch (cmd) {
+	case BLKTRACESETUP2:
+		snprintf(b, sizeof(b), "%pg", bdev);
+		ret = blk_trace_setup2(q, b, bdev->bd_dev, bdev, arg);
+		break;
 	case BLKTRACESETUP:
 		snprintf(b, sizeof(b), "%pg", bdev);
 		ret = blk_trace_setup(q, b, bdev->bd_dev, bdev, arg);
-- 
2.51.0

Re: [PATCH v3 06/16] blktrace: split do_blk_trace_setup into two functions

[Christoph Hellwig]

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v3 07/16] blktrace: add definitions for blk_user_trace_setup2

[Christoph Hellwig]

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v3 09/16] blktrace: add definitions for struct blk_io_trace2

[Christoph Hellwig]

On Wed, Oct 15, 2025 at 12:54:28PM +0200, Johannes Thumshirn wrote:
> Add definitions for the extended version of the blktrace protocol using a
> wider action type to be able to record new actions in the kernel.

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v3 11/16] blktrace: move trace_note to blk_io_trace2

[Christoph Hellwig]

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v3 12/16] blktrace: move ftrace blk_io_tracer to blk_io_trace2

[Christoph Hellwig]

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v3 13/16] blktrace: add block trace commands for zone operations

[Christoph Hellwig]

On Wed, Oct 15, 2025 at 12:54:32PM +0200, Johannes Thumshirn wrote:
> Add block trace commands for zone operations.
> 
> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
> ---
>  include/uapi/linux/blktrace_api.h | 13 +++++++++++--
>  kernel/trace/blktrace.c           | 25 +++++++++++++++++++++----
>  2 files changed, 32 insertions(+), 6 deletions(-)
> 
> diff --git a/include/uapi/linux/blktrace_api.h b/include/uapi/linux/blktrace_api.h
> index 3a771b9802aa..925f78af939e 100644
> --- a/include/uapi/linux/blktrace_api.h
> +++ b/include/uapi/linux/blktrace_api.h
> @@ -26,11 +26,20 @@ enum blktrace_cat {
>  	BLK_TC_DRV_DATA	= 1 << 14,	/* binary per-driver data */
>  	BLK_TC_FUA	= 1 << 15,	/* fua requests */
>  
> -	BLK_TC_END	= 1 << 15,	/* we've run out of bits! */
> +	BLK_TC_END_V1	= 1 << 15,	/* we've run out of bits! */
> +
> +	BLK_TC_ZONE_APPEND	= 1ull << 16,  	/* zone append */

Should BLK_TC_ZONE_APPEND be BLK_TC_END_V1?  Or is BLK_TC_END_V1 going
ot be used as the catchall for everything not reported in v1?

Re: [PATCH v3 14/16] blktrace: expose ZONE APPEND completions to blktrace

[Christoph Hellwig]

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v3 15/16] blktrace: trace zone write plugging operations

[Christoph Hellwig]

Looks good:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH v3 16/16] blktrace: handle BLKTRACESETUP2 ioctl

[Christoph Hellwig]

> +static int blk_trace_setup2(struct request_queue *q, char *name, dev_t dev,
> +			    struct block_device *bdev, char __user *arg)
> +{
> +	struct blk_user_trace_setup2 buts2;
> +	struct blk_trace *bt;
> +	int ret;
> +
> +	ret = copy_from_user(&buts2, arg, sizeof(buts2));
> +	if (ret)
> +		return -EFAULT;

You can simplify this to the more usual:

	if (copy_from_user(&buts2, arg, sizeof(buts2)))
		return -EFAULT;

> +
> +	if (!buts2.buf_size || !buts2.buf_nr)
> +		return -EINVAL;

This should probably also check that nothing is set in the flags field
so that it can be used for backwards compatible extensions?

Re: [PATCH v3 13/16] blktrace: add block trace commands for zone operations

[Johannes Thumshirn]

On 10/16/25 9:09 AM, Christoph Hellwig wrote:
> On Wed, Oct 15, 2025 at 12:54:32PM +0200, Johannes Thumshirn wrote:
>> Add block trace commands for zone operations.
>>
>> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
>> ---
>>   include/uapi/linux/blktrace_api.h | 13 +++++++++++--
>>   kernel/trace/blktrace.c           | 25 +++++++++++++++++++++----
>>   2 files changed, 32 insertions(+), 6 deletions(-)
>>
>> diff --git a/include/uapi/linux/blktrace_api.h b/include/uapi/linux/blktrace_api.h
>> index 3a771b9802aa..925f78af939e 100644
>> --- a/include/uapi/linux/blktrace_api.h
>> +++ b/include/uapi/linux/blktrace_api.h
>> @@ -26,11 +26,20 @@ enum blktrace_cat {
>>   	BLK_TC_DRV_DATA	= 1 << 14,	/* binary per-driver data */
>>   	BLK_TC_FUA	= 1 << 15,	/* fua requests */
>>   
>> -	BLK_TC_END	= 1 << 15,	/* we've run out of bits! */
>> +	BLK_TC_END_V1	= 1 << 15,	/* we've run out of bits! */
>> +
>> +	BLK_TC_ZONE_APPEND	= 1ull << 16,  	/* zone append */
> Should BLK_TC_ZONE_APPEND be BLK_TC_END_V1?  Or is BLK_TC_END_V1 going
> ot be used as the catchall for everything not reported in v1?
BLK_TC_END_V1 is equal to BLK_TC_FUA, but at the moment we're not doing 
anything with it. I'll add a WARN_ON() in case bt->version 1 is 
requested and a action > BLK_TC_END_V1 is passed in.