Skip to contentSkip to conversation

ftrace: printk formatted recording

PATCH Applied Steven Rostedt 3 participants 13 messages Aug 1, 2008 – Aug 15, 2008

Subsystems (3)
lore
Share & tools
lore

Steven Rostedt

author maintainer
#0 lore raw

v2 cover ftrace: printk formatted recording

I previously sent out a patch to add printk formatted recording to the ftrace buffers. Andrew Morton suggested to get rid of my hacked trace_cont structure and do it properly with one structure at the top and two other structures to handle the difference between the fields and the continuation entry.

That is to have a top entry:

struct trace_entry {
	char type;
	union {
		struct trace_field field;
		struct trace_cont  cont;
	};
};

instead of having a separate trace_cont that was outside the trace_entry and would be typecasted when used. This would be a source of bugs with later modifications of the ftrace code if a developer did not relize the connection between trace_entry and trace_cont. Using this proper method keeps the two tighly coupled for new developers to see.

Unfortunatly, doing it the proper way caused a large code change since now all the users of trace_entry (and there are many) must now dereference the field item.

The first patch just does the struct update and changes all the users to do the dereferencing properly.

The second patch in this series adds the user of the trace_cont, which is the ftrace_printk.

Note: I can no longer use the unspecified array size buf[] because gcc complains about:

struct trace_cont {
	buf[];
};

I now use buf[sizeof(struct trace_field)] which I think is a better solution anyways.

-- Steve

author maintainer
#1 lore raw

PATCH v2 1/2 ftrace: new continue entry - separate out from trace_entry

Some tracers will need to work with more than one entry. In order to do this the trace_entry structure was split into two fields. One for the start of all entries, and one to continue an existing entry.

The trace_entry structure now has a "field" entry that consists of the previous content of the trace_entry, and a "cont" entry that is just a string buffer the size of the "field" entry.

Thanks to Andrew Morton for suggesting this idea.

---
 kernel/trace/trace.c           |  267 +++++++++++++++++++++--------------------
 kernel/trace/trace.h           |   17 ++
 kernel/trace/trace_mmiotrace.c |   12 -
 3 files changed, 160 insertions(+), 136 deletions(-)
kernel/trace/trace.c +140 -127
Index: linux-tip.git/kernel/trace/trace.c
===================================================================
@@ -826,10 +826,11 @@ tracing_generic_entry_update(struct trac
 
 	pc = preempt_count();
 
-	entry->preempt_count	= pc & 0xff;
-	entry->pid		= (tsk) ? tsk->pid : 0;
-	entry->t		= ftrace_now(raw_smp_processor_id());
-	entry->flags = (irqs_disabled_flags(flags) ? TRACE_FLAG_IRQS_OFF : 0) |
+	entry->field.preempt_count	= pc & 0xff;
+	entry->field.pid		= (tsk) ? tsk->pid : 0;
+	entry->field.t			= ftrace_now(raw_smp_processor_id());
+	entry->field.flags =
+		(irqs_disabled_flags(flags) ? TRACE_FLAG_IRQS_OFF : 0) |
 		((pc & HARDIRQ_MASK) ? TRACE_FLAG_HARDIRQ : 0) |
 		((pc & SOFTIRQ_MASK) ? TRACE_FLAG_SOFTIRQ : 0) |
 		(need_resched() ? TRACE_FLAG_NEED_RESCHED : 0);
@@ -844,11 +845,11 @@ trace_function(struct trace_array *tr, s
 
 	raw_local_irq_save(irq_flags);
 	__raw_spin_lock(&data->lock);
-	entry			= tracing_get_trace_entry(tr, data);
+	entry				= tracing_get_trace_entry(tr, data);
 	tracing_generic_entry_update(entry, flags);
-	entry->type		= TRACE_FN;
-	entry->fn.ip		= ip;
-	entry->fn.parent_ip	= parent_ip;
+	entry->type			= TRACE_FN;
+	entry->field.fn.ip		= ip;
+	entry->field.fn.parent_ip	= parent_ip;
 	__raw_spin_unlock(&data->lock);
 	raw_local_irq_restore(irq_flags);
 }
@@ -871,10 +872,10 @@ void __trace_mmiotrace_rw(struct trace_a
 	raw_local_irq_save(irq_flags);
 	__raw_spin_lock(&data->lock);
 
-	entry			= tracing_get_trace_entry(tr, data);
+	entry				= tracing_get_trace_entry(tr, data);
 	tracing_generic_entry_update(entry, 0);
-	entry->type		= TRACE_MMIO_RW;
-	entry->mmiorw		= *rw;
+	entry->type			= TRACE_MMIO_RW;
+	entry->field.mmiorw		= *rw;
 
 	__raw_spin_unlock(&data->lock);
 	raw_local_irq_restore(irq_flags);
@@ -891,10 +892,10 @@ void __trace_mmiotrace_map(struct trace_
 	raw_local_irq_save(irq_flags);
 	__raw_spin_lock(&data->lock);
 
-	entry			= tracing_get_trace_entry(tr, data);
+	entry				= tracing_get_trace_entry(tr, data);
 	tracing_generic_entry_update(entry, 0);
-	entry->type		= TRACE_MMIO_MAP;
-	entry->mmiomap		= *map;
+	entry->type			= TRACE_MMIO_MAP;
+	entry->field.mmiomap		= *map;
 
 	__raw_spin_unlock(&data->lock);
 	raw_local_irq_restore(irq_flags);
@@ -918,12 +919,12 @@ void __trace_stack(struct trace_array *t
 	tracing_generic_entry_update(entry, flags);
 	entry->type		= TRACE_STACK;
 
-	memset(&entry->stack, 0, sizeof(entry->stack));
+	memset(&entry->field.stack, 0, sizeof(entry->field.stack));
 
 	trace.nr_entries	= 0;
 	trace.max_entries	= FTRACE_STACK_ENTRIES;
 	trace.skip		= skip;
-	trace.entries		= entry->stack.caller;
+	trace.entries		= entry->field.stack.caller;
 
 	save_stack_trace(&trace);
 }
@@ -939,12 +940,12 @@ __trace_special(void *__tr, void *__data
 
 	raw_local_irq_save(irq_flags);
 	__raw_spin_lock(&data->lock);
-	entry			= tracing_get_trace_entry(tr, data);
+	entry				= tracing_get_trace_entry(tr, data);
 	tracing_generic_entry_update(entry, 0);
-	entry->type		= TRACE_SPECIAL;
-	entry->special.arg1	= arg1;
-	entry->special.arg2	= arg2;
-	entry->special.arg3	= arg3;
+	entry->type			= TRACE_SPECIAL;
+	entry->field.special.arg1	= arg1;
+	entry->field.special.arg2	= arg2;
+	entry->field.special.arg3	= arg3;
 	__trace_stack(tr, data, irq_flags, 4);
 	__raw_spin_unlock(&data->lock);
 	raw_local_irq_restore(irq_flags);
@@ -964,15 +965,15 @@ tracing_sched_switch_trace(struct trace_
 
 	raw_local_irq_save(irq_flags);
 	__raw_spin_lock(&data->lock);
-	entry			= tracing_get_trace_entry(tr, data);
+	entry				= tracing_get_trace_entry(tr, data);
 	tracing_generic_entry_update(entry, flags);
-	entry->type		= TRACE_CTX;
-	entry->ctx.prev_pid	= prev->pid;
-	entry->ctx.prev_prio	= prev->prio;
-	entry->ctx.prev_state	= prev->state;
-	entry->ctx.next_pid	= next->pid;
-	entry->ctx.next_prio	= next->prio;
-	entry->ctx.next_state	= next->state;
+	entry->type			= TRACE_CTX;
+	entry->field.ctx.prev_pid	= prev->pid;
+	entry->field.ctx.prev_prio	= prev->prio;
+	entry->field.ctx.prev_state	= prev->state;
+	entry->field.ctx.next_pid	= next->pid;
+	entry->field.ctx.next_prio	= next->prio;
+	entry->field.ctx.next_state	= next->state;
 	__trace_stack(tr, data, flags, 5);
 	__raw_spin_unlock(&data->lock);
 	raw_local_irq_restore(irq_flags);
@@ -993,12 +994,12 @@ tracing_sched_wakeup_trace(struct trace_
 	entry			= tracing_get_trace_entry(tr, data);
 	tracing_generic_entry_update(entry, flags);
 	entry->type		= TRACE_WAKE;
-	entry->ctx.prev_pid	= curr->pid;
-	entry->ctx.prev_prio	= curr->prio;
-	entry->ctx.prev_state	= curr->state;
-	entry->ctx.next_pid	= wakee->pid;
-	entry->ctx.next_prio	= wakee->prio;
-	entry->ctx.next_state	= wakee->state;
+	entry->field.ctx.prev_pid	= curr->pid;
+	entry->field.ctx.prev_prio	= curr->prio;
+	entry->field.ctx.prev_state	= curr->state;
+	entry->field.ctx.next_pid	= wakee->pid;
+	entry->field.ctx.next_prio	= wakee->prio;
+	entry->field.ctx.next_state	= wakee->state;
 	__trace_stack(tr, data, flags, 6);
 	__raw_spin_unlock(&data->lock);
 	raw_local_irq_restore(irq_flags);
@@ -1127,7 +1128,7 @@ find_next_entry(struct trace_iterator *i
 		/*
 		 * Pick the entry with the smallest timestamp:
 		 */
-		if (ent && (!next || ent->t < next->t)) {
+		if (ent && (!next || ent->field.t < next->field.t)) {
 			next = ent;
 			next_cpu = cpu;
 		}
@@ -1431,19 +1432,20 @@ print_trace_header(struct seq_file *m, s
 static void
 lat_print_generic(struct trace_seq *s, struct trace_entry *entry, int cpu)
 {
+	struct trace_field *field = &entry->field;
 	int hardirq, softirq;
 	char *comm;
 
-	comm = trace_find_cmdline(entry->pid);
+	comm = trace_find_cmdline(field->pid);
 
-	trace_seq_printf(s, "%8.8s-%-5d ", comm, entry->pid);
+	trace_seq_printf(s, "%8.8s-%-5d ", comm, field->pid);
 	trace_seq_printf(s, "%d", cpu);
 	trace_seq_printf(s, "%c%c",
-			(entry->flags & TRACE_FLAG_IRQS_OFF) ? 'd' : '.',
-			((entry->flags & TRACE_FLAG_NEED_RESCHED) ? 'N' : '.'));
+			(field->flags & TRACE_FLAG_IRQS_OFF) ? 'd' : '.',
+			((field->flags & TRACE_FLAG_NEED_RESCHED) ? 'N' : '.'));
 
-	hardirq = entry->flags & TRACE_FLAG_HARDIRQ;
-	softirq = entry->flags & TRACE_FLAG_SOFTIRQ;
+	hardirq = field->flags & TRACE_FLAG_HARDIRQ;
+	softirq = field->flags & TRACE_FLAG_SOFTIRQ;
 	if (hardirq && softirq) {
 		trace_seq_putc(s, 'H');
 	} else {
@@ -1457,8 +1459,8 @@ lat_print_generic(struct trace_seq *s, s
 		}
 	}
 
-	if (entry->preempt_count)
-		trace_seq_printf(s, "%x", entry->preempt_count);
+	if (field->preempt_count)
+		trace_seq_printf(s, "%x", field->preempt_count);
 	else
 		trace_seq_puts(s, ".");
 }
@@ -1488,6 +1490,7 @@ print_lat_fmt(struct trace_iterator *ite
 	struct trace_entry *next_entry = find_next_entry(iter, NULL);
 	unsigned long verbose = (trace_flags & TRACE_ITER_VERBOSE);
 	struct trace_entry *entry = iter->ent;
+	struct trace_field *field = &entry->field;
 	unsigned long abs_usecs;
 	unsigned long rel_usecs;
 	char *comm;
@@ -1497,17 +1500,17 @@ print_lat_fmt(struct trace_iterator *ite
 
 	if (!next_entry)
 		next_entry = entry;
-	rel_usecs = ns2usecs(next_entry->t - entry->t);
-	abs_usecs = ns2usecs(entry->t - iter->tr->time_start);
+	rel_usecs = ns2usecs(next_entry->field.t - entry->field.t);
+	abs_usecs = ns2usecs(entry->field.t - iter->tr->time_start);
 
 	if (verbose) {
-		comm = trace_find_cmdline(entry->pid);
+		comm = trace_find_cmdline(field->pid);
 		trace_seq_printf(s, "%16s %5d %d %d %08x %08x [%08lx]"
 				 " %ld.%03ldms (+%ld.%03ldms): ",
 				 comm,
-				 entry->pid, cpu, entry->flags,
-				 entry->preempt_count, trace_idx,
-				 ns2usecs(entry->t),
+				 field->pid, cpu, field->flags,
+				 field->preempt_count, trace_idx,
+				 ns2usecs(field->t),
 				 abs_usecs/1000,
 				 abs_usecs % 1000, rel_usecs/1000,
 				 rel_usecs % 1000);
@@ -1517,41 +1520,42 @@ print_lat_fmt(struct trace_iterator *ite
 	}
 	switch (entry->type) {
 	case TRACE_FN:
-		seq_print_ip_sym(s, entry->fn.ip, sym_flags);
+		seq_print_ip_sym(s, field->fn.ip, sym_flags);
 		trace_seq_puts(s, " (");
-		if (kretprobed(entry->fn.parent_ip))
+		if (kretprobed(field->fn.parent_ip))
 			trace_seq_puts(s, KRETPROBE_MSG);
 		else
-			seq_print_ip_sym(s, entry->fn.parent_ip, sym_flags);
+			seq_print_ip_sym(s, field->fn.parent_ip, sym_flags);
 		trace_seq_puts(s, ")\n");
 		break;
 	case TRACE_CTX:
 	case TRACE_WAKE:
-		T = entry->ctx.next_state < sizeof(state_to_char) ?
-			state_to_char[entry->ctx.next_state] : 'X';
+		T = field->ctx.next_state < sizeof(state_to_char) ?
+			state_to_char[field->ctx.next_state] : 'X';
 
-		state = entry->ctx.prev_state ? __ffs(entry->ctx.prev_state) + 1 : 0;
+		state = field->ctx.prev_state ?
+			__ffs(field->ctx.prev_state) + 1 : 0;
 		S = state < sizeof(state_to_char) - 1 ? state_to_char[state] : 'X';
-		comm = trace_find_cmdline(entry->ctx.next_pid);
+		comm = trace_find_cmdline(field->ctx.next_pid);
 		trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d:%c %s\n",
-				 entry->ctx.prev_pid,
-				 entry->ctx.prev_prio,
+				 field->ctx.prev_pid,
+				 field->ctx.prev_prio,
 				 S, entry->type == TRACE_CTX ? "==>" : "  +",
-				 entry->ctx.next_pid,
-				 entry->ctx.next_prio,
+				 field->ctx.next_pid,
+				 field->ctx.next_prio,
 				 T, comm);
 		break;
 	case TRACE_SPECIAL:
 		trace_seq_printf(s, "# %ld %ld %ld\n",
-				 entry->special.arg1,
-				 entry->special.arg2,
-				 entry->special.arg3);
+				 field->special.arg1,
+				 field->special.arg2,
+				 field->special.arg3);
 		break;
 	case TRACE_STACK:
 		for (i = 0; i < FTRACE_STACK_ENTRIES; i++) {
 			if (i)
 				trace_seq_puts(s, " <= ");
-			seq_print_ip_sym(s, entry->stack.caller[i], sym_flags);
+			seq_print_ip_sym(s, field->stack.caller[i], sym_flags);
 		}
 		trace_seq_puts(s, "\n");
 		break;
@@ -1566,6 +1570,7 @@ static int print_trace_fmt(struct trace_
 	struct trace_seq *s = &iter->seq;
 	unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK);
 	struct trace_entry *entry;
+	struct trace_field *field;
 	unsigned long usec_rem;
 	unsigned long long t;
 	unsigned long secs;
@@ -1575,14 +1580,15 @@ static int print_trace_fmt(struct trace_
 	int i;
 
 	entry = iter->ent;
+	field = &entry->field;
 
-	comm = trace_find_cmdline(iter->ent->pid);
+	comm = trace_find_cmdline(iter->ent->field.pid);
 
-	t = ns2usecs(entry->t);
+	t = ns2usecs(field->t);
 	usec_rem = do_div(t, 1000000ULL);
 	secs = (unsigned long)t;
 
-	ret = trace_seq_printf(s, "%16s-%-5d ", comm, entry->pid);
+	ret = trace_seq_printf(s, "%16s-%-5d ", comm, field->pid);
 	if (!ret)
 		return 0;
 	ret = trace_seq_printf(s, "[%02d] ", iter->cpu);
@@ -1594,18 +1600,19 @@ static int print_trace_fmt(struct trace_
 
 	switch (entry->type) {
 	case TRACE_FN:
-		ret = seq_print_ip_sym(s, entry->fn.ip, sym_flags);
+		ret = seq_print_ip_sym(s, field->fn.ip, sym_flags);
 		if (!ret)
 			return 0;
 		if ((sym_flags & TRACE_ITER_PRINT_PARENT) &&
-						entry->fn.parent_ip) {
+						field->fn.parent_ip) {
 			ret = trace_seq_printf(s, " <-");
 			if (!ret)
 				return 0;
-			if (kretprobed(entry->fn.parent_ip))
+			if (kretprobed(field->fn.parent_ip))
 				ret = trace_seq_puts(s, KRETPROBE_MSG);
 			else
-				ret = seq_print_ip_sym(s, entry->fn.parent_ip,
+				ret = seq_print_ip_sym(s,
+						       field->fn.parent_ip,
 						       sym_flags);
 			if (!ret)
 				return 0;
@@ -1616,26 +1623,26 @@ static int print_trace_fmt(struct trace_
 		break;
 	case TRACE_CTX:
 	case TRACE_WAKE:
-		S = entry->ctx.prev_state < sizeof(state_to_char) ?
-			state_to_char[entry->ctx.prev_state] : 'X';
-		T = entry->ctx.next_state < sizeof(state_to_char) ?
-			state_to_char[entry->ctx.next_state] : 'X';
+		S = field->ctx.prev_state < sizeof(state_to_char) ?
+			state_to_char[field->ctx.prev_state] : 'X';
+		T = field->ctx.next_state < sizeof(state_to_char) ?
+			state_to_char[field->ctx.next_state] : 'X';
 		ret = trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d:%c\n",
-				       entry->ctx.prev_pid,
-				       entry->ctx.prev_prio,
+				       field->ctx.prev_pid,
+				       field->ctx.prev_prio,
 				       S,
 				       entry->type == TRACE_CTX ? "==>" : "  +",
-				       entry->ctx.next_pid,
-				       entry->ctx.next_prio,
+				       field->ctx.next_pid,
+				       field->ctx.next_prio,
 				       T);
 		if (!ret)
 			return 0;
 		break;
 	case TRACE_SPECIAL:
 		ret = trace_seq_printf(s, "# %ld %ld %ld\n",
-				 entry->special.arg1,
-				 entry->special.arg2,
-				 entry->special.arg3);
+				 field->special.arg1,
+				 field->special.arg2,
+				 field->special.arg3);
 		if (!ret)
 			return 0;
 		break;
@@ -1646,7 +1653,7 @@ static int print_trace_fmt(struct trace_
 				if (!ret)
 					return 0;
 			}
-			ret = seq_print_ip_sym(s, entry->stack.caller[i],
+			ret = seq_print_ip_sym(s, field->stack.caller[i],
 					       sym_flags);
 			if (!ret)
 				return 0;
@@ -1663,37 +1670,40 @@ static int print_raw_fmt(struct trace_it
 {
 	struct trace_seq *s = &iter->seq;
 	struct trace_entry *entry;
+	struct trace_field *field;
 	int ret;
 	int S, T;
 
 	entry = iter->ent;
+	field = &entry->field;
 
 	ret = trace_seq_printf(s, "%d %d %llu ",
-		entry->pid, iter->cpu, entry->t);
+		field->pid, iter->cpu, field->t);
 	if (!ret)
 		return 0;
 
 	switch (entry->type) {
 	case TRACE_FN:
 		ret = trace_seq_printf(s, "%x %x\n",
-					entry->fn.ip, entry->fn.parent_ip);
+					field->fn.ip,
+					field->fn.parent_ip);
 		if (!ret)
 			return 0;
 		break;
 	case TRACE_CTX:
 	case TRACE_WAKE:
-		S = entry->ctx.prev_state < sizeof(state_to_char) ?
-			state_to_char[entry->ctx.prev_state] : 'X';
-		T = entry->ctx.next_state < sizeof(state_to_char) ?
-			state_to_char[entry->ctx.next_state] : 'X';
+		S = field->ctx.prev_state < sizeof(state_to_char) ?
+			state_to_char[field->ctx.prev_state] : 'X';
+		T = field->ctx.next_state < sizeof(state_to_char) ?
+			state_to_char[field->ctx.next_state] : 'X';
 		if (entry->type == TRACE_WAKE)
 			S = '+';
 		ret = trace_seq_printf(s, "%d %d %c %d %d %c\n",
-				       entry->ctx.prev_pid,
-				       entry->ctx.prev_prio,
+				       field->ctx.prev_pid,
+				       field->ctx.prev_prio,
 				       S,
-				       entry->ctx.next_pid,
-				       entry->ctx.next_prio,
+				       field->ctx.next_pid,
+				       field->ctx.next_prio,
 				       T);
 		if (!ret)
 			return 0;
@@ -1701,9 +1711,9 @@ static int print_raw_fmt(struct trace_it
 	case TRACE_SPECIAL:
 	case TRACE_STACK:
 		ret = trace_seq_printf(s, "# %ld %ld %ld\n",
-				 entry->special.arg1,
-				 entry->special.arg2,
-				 entry->special.arg3);
+				 field->special.arg1,
+				 field->special.arg2,
+				 field->special.arg3);
 		if (!ret)
 			return 0;
 		break;
@@ -1728,40 +1738,41 @@ static int print_hex_fmt(struct trace_it
 	struct trace_seq *s = &iter->seq;
 	unsigned char newline = '\n';
 	struct trace_entry *entry;
+	struct trace_field *field;
 	int S, T;
 
 	entry = iter->ent;
+	field = &entry->field;
 
-	SEQ_PUT_HEX_FIELD_RET(s, entry->pid);
+	SEQ_PUT_HEX_FIELD_RET(s, field->pid);
 	SEQ_PUT_HEX_FIELD_RET(s, iter->cpu);
-	SEQ_PUT_HEX_FIELD_RET(s, entry->t);
+	SEQ_PUT_HEX_FIELD_RET(s, field->t);
 
 	switch (entry->type) {
 	case TRACE_FN:
-		SEQ_PUT_HEX_FIELD_RET(s, entry->fn.ip);
-		SEQ_PUT_HEX_FIELD_RET(s, entry->fn.parent_ip);
+		SEQ_PUT_HEX_FIELD_RET(s, field->fn.ip);
+		SEQ_PUT_HEX_FIELD_RET(s, field->fn.parent_ip);
 		break;
 	case TRACE_CTX:
 	case TRACE_WAKE:
-		S = entry->ctx.prev_state < sizeof(state_to_char) ?
-			state_to_char[entry->ctx.prev_state] : 'X';
-		T = entry->ctx.next_state < sizeof(state_to_char) ?
-			state_to_char[entry->ctx.next_state] : 'X';
+		S = field->ctx.prev_state < sizeof(state_to_char) ?
+			state_to_char[field->ctx.prev_state] : 'X';
+		T = field->ctx.next_state < sizeof(state_to_char) ?
+			state_to_char[field->ctx.next_state] : 'X';
 		if (entry->type == TRACE_WAKE)
 			S = '+';
-		SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.prev_pid);
-		SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.prev_prio);
+		SEQ_PUT_HEX_FIELD_RET(s, field->ctx.prev_pid);
+		SEQ_PUT_HEX_FIELD_RET(s, field->ctx.prev_prio);
 		SEQ_PUT_HEX_FIELD_RET(s, S);
-		SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.next_pid);
-		SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.next_prio);
-		SEQ_PUT_HEX_FIELD_RET(s, entry->fn.parent_ip);
+		SEQ_PUT_HEX_FIELD_RET(s, field->ctx.next_pid);
+		SEQ_PUT_HEX_FIELD_RET(s, field->ctx.next_prio);
 		SEQ_PUT_HEX_FIELD_RET(s, T);
 		break;
 	case TRACE_SPECIAL:
 	case TRACE_STACK:
-		SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg1);
-		SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg2);
-		SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg3);
+		SEQ_PUT_HEX_FIELD_RET(s, field->special.arg1);
+		SEQ_PUT_HEX_FIELD_RET(s, field->special.arg2);
+		SEQ_PUT_HEX_FIELD_RET(s, field->special.arg3);
 		break;
 	}
 	SEQ_PUT_FIELD_RET(s, newline);
@@ -1773,31 +1784,33 @@ static int print_bin_fmt(struct trace_it
 {
 	struct trace_seq *s = &iter->seq;
 	struct trace_entry *entry;
+	struct trace_field *field;
 
 	entry = iter->ent;
+	field = &entry->field;
 
-	SEQ_PUT_FIELD_RET(s, entry->pid);
-	SEQ_PUT_FIELD_RET(s, entry->cpu);
-	SEQ_PUT_FIELD_RET(s, entry->t);
+	SEQ_PUT_FIELD_RET(s, field->pid);
+	SEQ_PUT_FIELD_RET(s, field->cpu);
+	SEQ_PUT_FIELD_RET(s, field->t);
 
 	switch (entry->type) {
 	case TRACE_FN:
-		SEQ_PUT_FIELD_RET(s, entry->fn.ip);
-		SEQ_PUT_FIELD_RET(s, entry->fn.parent_ip);
+		SEQ_PUT_FIELD_RET(s, field->fn.ip);
+		SEQ_PUT_FIELD_RET(s, field->fn.parent_ip);
 		break;
 	case TRACE_CTX:
-		SEQ_PUT_FIELD_RET(s, entry->ctx.prev_pid);
-		SEQ_PUT_FIELD_RET(s, entry->ctx.prev_prio);
-		SEQ_PUT_FIELD_RET(s, entry->ctx.prev_state);
-		SEQ_PUT_FIELD_RET(s, entry->ctx.next_pid);
-		SEQ_PUT_FIELD_RET(s, entry->ctx.next_prio);
-		SEQ_PUT_FIELD_RET(s, entry->ctx.next_state);
+		SEQ_PUT_FIELD_RET(s, field->ctx.prev_pid);
+		SEQ_PUT_FIELD_RET(s, field->ctx.prev_prio);
+		SEQ_PUT_FIELD_RET(s, field->ctx.prev_state);
+		SEQ_PUT_FIELD_RET(s, field->ctx.next_pid);
+		SEQ_PUT_FIELD_RET(s, field->ctx.next_prio);
+		SEQ_PUT_FIELD_RET(s, field->ctx.next_state);
 		break;
 	case TRACE_SPECIAL:
 	case TRACE_STACK:
-		SEQ_PUT_FIELD_RET(s, entry->special.arg1);
-		SEQ_PUT_FIELD_RET(s, entry->special.arg2);
-		SEQ_PUT_FIELD_RET(s, entry->special.arg3);
+		SEQ_PUT_FIELD_RET(s, field->special.arg1);
+		SEQ_PUT_FIELD_RET(s, field->special.arg2);
+		SEQ_PUT_FIELD_RET(s, field->special.arg3);
 		break;
 	}
 	return 1;
kernel/trace/trace.h +14 -3
Index: linux-tip.git/kernel/trace/trace.h
===================================================================
@@ -61,13 +61,12 @@ struct stack_entry {
 };
 
 /*
- * The trace entry - the most basic unit of tracing. This is what
+ * The trace field - the most basic unit of tracing. This is what
  * is printed in the end as a single line in the trace output, such as:
  *
  *     bash-15816 [01]   235.197585: idle_cpu <- irq_enter
  */
-struct trace_entry {
-	char			type;
+struct trace_field {
 	char			cpu;
 	char			flags;
 	char			preempt_count;
@@ -83,6 +82,18 @@ struct trace_entry {
 	};
 };
 
+struct trace_field_cont {
+	char				buf[sizeof(struct trace_field)];
+};
+
+struct trace_entry {
+	char 				type;
+	union {
+		struct trace_field	field;
+		struct trace_field_cont	cont;
+	};
+};
+
 #define TRACE_ENTRY_SIZE	sizeof(struct trace_entry)
 
 /*
kernel/trace/trace_mmiotrace.c +6 -6
Index: linux-tip.git/kernel/trace/trace_mmiotrace.c
===================================================================
@@ -174,14 +174,14 @@ print_out:
 static int mmio_print_rw(struct trace_iterator *iter)
 {
 	struct trace_entry *entry = iter->ent;
-	struct mmiotrace_rw *rw	= &entry->mmiorw;
+	struct mmiotrace_rw *rw	= &entry->field.mmiorw;
 	struct trace_seq *s	= &iter->seq;
-	unsigned long long t	= ns2usecs(entry->t);
+	unsigned long long t	= ns2usecs(entry->field.t);
 	unsigned long usec_rem	= do_div(t, 1000000ULL);
 	unsigned secs		= (unsigned long)t;
 	int ret = 1;
 
-	switch (entry->mmiorw.opcode) {
+	switch (entry->field.mmiorw.opcode) {
 	case MMIO_READ:
 		ret = trace_seq_printf(s,
 			"R %d %lu.%06lu %d 0x%llx 0x%lx 0x%lx %d\n",
@@ -216,14 +216,14 @@ static int mmio_print_rw(struct trace_it
 static int mmio_print_map(struct trace_iterator *iter)
 {
 	struct trace_entry *entry = iter->ent;
-	struct mmiotrace_map *m	= &entry->mmiomap;
+	struct mmiotrace_map *m	= &entry->field.mmiomap;
 	struct trace_seq *s	= &iter->seq;
-	unsigned long long t	= ns2usecs(entry->t);
+	unsigned long long t	= ns2usecs(entry->field.t);
 	unsigned long usec_rem	= do_div(t, 1000000ULL);
 	unsigned secs		= (unsigned long)t;
 	int ret = 1;
 
-	switch (entry->mmiorw.opcode) {
+	switch (entry->field.mmiorw.opcode) {
 	case MMIO_PROBE:
 		ret = trace_seq_printf(s,
 			"MAP %lu.%06lu %d 0x%llx 0x%lx 0x%lx 0x%lx %d\n",
author maintainer
#2 lore raw

PATCH v2 2/2 ftrace: printk formatting infrastructure

In reply to Steven Rostedt

This patch adds a feature that can help kernel developers debug their code using ftrace.

  int ftrace_printk(const char *fmt, ...);

This records into the ftrace buffer using printf formatting. The entry size in the buffers are still a fixed length. A new type has been added that allows for more entries to be used for a single recording.

The start of the print is still the same as the other entries.

It returns the number of characters written to the ftrace buffer.

For example:

Having a module with the following code:

static int __init ftrace_print_test(void)

{
        ftrace_printk("jiffies are %ld\n", jiffies);
        return 0;
}

Gives me:

insmod-5441 3...1 7569us : ftrace_print_test: jiffies are 4296626666

for the latency_trace file and:

          insmod-5441  [03]  1959.370498: ftrace_print_test jiffies are 4296626666

for the trace file.

Note: Only the infrastructure should go into the kernel. It is to help facilitate debugging for other kernel developers. Calls to ftrace_printk is not intended to be left in the kernel, and should be frowned upon just like scattering printks around in the code.

But having this easily at your fingertips helps the debugging go faster and bugs be solved quicker.

Maybe later on, we can hook this with markers and have their printf format be sucked into ftrace output.

---
 include/linux/ftrace.h        |   10 +
 kernel/trace/trace.c          |  273 +++++++++++++++++++++++++++++++++++++-----
 kernel/trace/trace.h          |   11 +
 kernel/trace/trace_selftest.c |   11 +
 4 files changed, 272 insertions(+), 33 deletions(-)
include/linux/ftrace.h +10 -0
Index: linux-tip.git/include/linux/ftrace.h
===================================================================
@@ -137,10 +137,20 @@ static inline void tracer_disable(void)
 extern void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3);
 extern void ftrace_dump(void);
+# define ftrace_printk(x...) __ftrace_printk(_THIS_IP_, x)
+extern int
+__ftrace_printk(unsigned long ip, const char *fmt, ...)
+	__attribute__ ((format (printf, 2, 3)));
 #else
 static inline void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) { }
 static inline void ftrace_dump(void) { }
+static inline int
+ftrace_printk(const char *fmt, ...) __attribute__ ((format (printf, 1, 0)))
+{
+	return 0;
+}
 #endif
 
+
 #endif /* _LINUX_FTRACE_H */
kernel/trace/trace.c +244 -29

387 lines not on this page · raw message

kernel/trace/trace.h +11 -0
Index: linux-tip.git/kernel/trace/trace.h
===================================================================
@@ -13,7 +13,9 @@ enum trace_type {
 	TRACE_FN,
 	TRACE_CTX,
 	TRACE_WAKE,
+	TRACE_CONT,
 	TRACE_STACK,
+	TRACE_PRINT,
 	TRACE_SPECIAL,
 	TRACE_MMIO_RW,
 	TRACE_MMIO_MAP,
@@ -61,6 +63,14 @@ struct stack_entry {
 };
 
 /*
+ * ftrace_printk entry:
+ */
+struct print_entry {
+	unsigned long		ip;
+	char			buf[];
+};
+
+/*
  * The trace field - the most basic unit of tracing. This is what
  * is printed in the end as a single line in the trace output, such as:
  *
@@ -77,6 +87,7 @@ struct trace_field {
 		struct ctx_switch_entry		ctx;
 		struct special_entry		special;
 		struct stack_entry		stack;
+		struct print_entry		print;
 		struct mmiotrace_rw		mmiorw;
 		struct mmiotrace_map		mmiomap;
 	};
kernel/trace/trace_selftest.c +7 -4
Index: linux-tip.git/kernel/trace/trace_selftest.c
===================================================================
@@ -9,7 +9,9 @@ static inline int trace_valid_entry(stru
 	case TRACE_FN:
 	case TRACE_CTX:
 	case TRACE_WAKE:
+	case TRACE_CONT:
 	case TRACE_STACK:
+	case TRACE_PRINT:
 	case TRACE_SPECIAL:
 		return 1;
 	}
@@ -120,11 +122,11 @@ int trace_selftest_startup_dynamic_traci
 					   struct trace_array *tr,
 					   int (*func)(void))
 {
-	unsigned long count;
-	int ret;
 	int save_ftrace_enabled = ftrace_enabled;
 	int save_tracer_enabled = tracer_enabled;
+	unsigned long count;
 	char *func_name;
+	int ret;
 
 	/* The ftrace test PASSED */
 	printk(KERN_CONT "PASSED\n");
@@ -157,6 +159,7 @@ int trace_selftest_startup_dynamic_traci
 	/* enable tracing */
 	tr->ctrl = 1;
 	trace->init(tr);
+
 	/* Sleep for a 1/10 of a second */
 	msleep(100);
 
@@ -212,10 +215,10 @@ int trace_selftest_startup_dynamic_traci
 int
 trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr)
 {
-	unsigned long count;
-	int ret;
 	int save_ftrace_enabled = ftrace_enabled;
 	int save_tracer_enabled = tracer_enabled;
+	unsigned long count;
+	int ret;
 
 	/* make sure msleep has been recorded */
 	msleep(1);
#3 lore raw

Re: [PATCH v2 2/2] ftrace: printk formatting infrastructure

Quoted 3 lines · On Fri, 01 Aug 2008 12:26:41 -0400 Steven Rostedt wrote:

+/**
+ * ftrace_printk - printf formatting in the ftrace buffer
+ * @fmt - the printf format for printing.

* @fmt: the printf format for printing
* @ip: <description>

Quoted 13 lines

+ *
+ * Note: __ftrace_printk is an internal function for ftrace_printk and
+ * the @ip is passed in via the ftrace_printk macro.
+ *
+ * This function allows a kernel developer to debug fast path sections
+ * that printk is not appropriate for. By scattering in various
+ * printk like tracing in the code, a developer can quickly see
+ * where problems are occurring.
+ *
+ * This is intended as a debugging tool for the developer only.
+ * Please reframe from leaving ftrace_printks scattered around in
+ * your code.
+ */

kernel/trace/trace.c void ftrace_dump(void) · line 3263

Quoted 2 lines
kernel/trace/trace.c +2 -0
+int __ftrace_printk(unsigned long ip, const char *fmt, ...)
+{

---
~Randy
Linux Plumbers Conference, 17-19 September 2008, Portland, Oregon USA
http://linuxplumbersconf.org/

author maintainer
#4 lore raw

Re: [PATCH v2 2/2] ftrace: printk formatting infrastructure

Quoted 12 lines · Randy Dunlap
Quoted 3 lines · On Fri, 01 Aug 2008 12:26:41 -0400 Steven Rostedt wrote:

+/**
+ * ftrace_printk - printf formatting in the ftrace buffer
+ * @fmt - the printf format for printing.

* @fmt: the printf format for printing
* @ip: <description>

Quoted 3 lines

+ *
+ * Note: __ftrace_printk is an internal function for ftrace_printk and
+ * the @ip is passed in via the ftrace_printk macro.

The above note actually explains this ;-)

Should I move the description to the ftrace_printk macro?

In include/linux/ftrace.h you see:

#ifdef CONFIG_TRACING
[...]

# define ftrace_printk(x...) __ftrace_printk(_THIS_IP_, x)
extern int
__ftrace_printk(unsigned long ip, const char *fmt, ...)
        __attribute__ ((format (printf, 2, 3)));
#else
[...]
static inline int
ftrace_printk(const char *fmt, ...) __attribute__ ((format (printf, 1, 
0)))
{
        return 0;
}
#endif

The true API is ftrace_printk(const char *fmt) and the __ftrace_printk version should not be used.

Do you still think I should document the @ip?

I could also just remove the passing of _THIS_IP_ and use _RET_IP_ inside of ftrace_printk but the _THIS_IP_ seems to be more accurate especially when we have stack unwinding returns.

-- Steve

Quoted 19 lines

kernel/trace/trace.c void ftrace_dump(void) · line 3263

Quoted 12 lines

+ *
+ * This function allows a kernel developer to debug fast path sections
+ * that printk is not appropriate for. By scattering in various
+ * printk like tracing in the code, a developer can quickly see
+ * where problems are occurring.
+ *
+ * This is intended as a debugging tool for the developer only.
+ * Please reframe from leaving ftrace_printks scattered around in
+ * your code.
+ */

kernel/trace/trace.c +2 -0
+int __ftrace_printk(unsigned long ip, const char *fmt, ...)
+{

---
~Randy
Linux Plumbers Conference, 17-19 September 2008, Portland, Oregon USA
http://linuxplumbersconf.org/

#5 lore raw

Re: [PATCH v2 2/2] ftrace: printk formatting infrastructure

Quoted 9 lines · On Fri, 1 Aug 2008 12:53:17 -0400 (EDT) Steven Rostedt wrote:
Quoted 6 lines · Randy Dunlap
Quoted 3 lines · On Fri, 01 Aug 2008 12:26:41 -0400 Steven Rostedt wrote:

+/**
+ * ftrace_printk - printf formatting in the ftrace buffer
+ * @fmt - the printf format for printing.

Quoted 8 lines
Quoted 6 lines

* @fmt: the printf format for printing
* @ip: <description>

Quoted 3 lines

+ *
+ * Note: __ftrace_printk is an internal function for ftrace_printk and
+ * the @ip is passed in via the ftrace_printk macro.

The above note actually explains this ;-)

I read that.

Quoted 1 line

Should I move the description to the ftrace_printk macro?

Probably.

Quoted 18 lines

In include/linux/ftrace.h you see:

#ifdef CONFIG_TRACING
[...]

# define ftrace_printk(x...) __ftrace_printk(_THIS_IP_, x)
extern int
__ftrace_printk(unsigned long ip, const char *fmt, ...)
        __attribute__ ((format (printf, 2, 3)));
#else
[...]
static inline int
ftrace_printk(const char *fmt, ...) __attribute__ ((format (printf, 1, 
0)))
{
        return 0;
}
#endif
Quoted 4 lines

The true API is ftrace_printk(const char *fmt) and the __ftrace_printk version should not be used.

Do you still think I should document the @ip?

Just move the kernel-doc to the macro then.

Quoted 8 lines

I could also just remove the passing of _THIS_IP_ and use _RET_IP_ inside of ftrace_printk but the _THIS_IP_ seems to be more accurate especially when we have stack unwinding returns.

-- Steve

Quoted 12 lines
Quoted 4 lines
Quoted 4 lines

+ *
+ * This function allows a kernel developer to debug fast path sections
+ * that printk is not appropriate for. By scattering in various
+ * printk like tracing in the code, a developer can quickly see

Quoted 8 lines

kernel/trace/trace.c void ftrace_dump(void) · line 3263

Quoted 8 lines

+ * where problems are occurring.
+ *
+ * This is intended as a debugging tool for the developer only.
+ * Please reframe from leaving ftrace_printks scattered around in
+ * your code.
+ */

kernel/trace/trace.c +2 -0
+int __ftrace_printk(unsigned long ip, const char *fmt, ...)
+{

---
~Randy
Linux Plumbers Conference, 17-19 September 2008, Portland, Oregon USA
http://linuxplumbersconf.org/

author maintainer
#6 lore raw

[PATCH v2 3/2] ftrace: ftrace_printk doc moved

Based on Randy Dunlap's suggestion, the ftrace_printk kernel-doc belongs with the ftrace_printk macro that should be used. Not with the __ftrace_printk internal function.

---
 include/linux/ftrace.h |   19 ++++++++++++++++++-
 kernel/trace/trace.c   |   16 ----------------
 2 files changed, 18 insertions(+), 17 deletions(-)
include/linux/ftrace.h +18 -1
Index: linux-tip.git/include/linux/ftrace.h
===================================================================
@@ -137,7 +137,24 @@ static inline void tracer_disable(void)
 extern void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3);
 extern void ftrace_dump(void);
-# define ftrace_printk(x...) __ftrace_printk(_THIS_IP_, x)
+
+/**
+ * ftrace_printk - printf formatting in the ftrace buffer
+ * @fmt - the printf format for printing.
+ *
+ * Note: __ftrace_printk is an internal function for ftrace_printk and
+ *       the @ip is passed in via the ftrace_printk macro.
+ *
+ * This function allows a kernel developer to debug fast path sections
+ * that printk is not appropriate for. By scattering in various
+ * printk like tracing in the code, a developer can quickly see
+ * where problems are occurring.
+ *
+ * This is intended as a debugging tool for the developer only.
+ * Please reframe from leaving ftrace_printks scattered around in
+ * your code.
+ */
+# define ftrace_printk(fmt...) __ftrace_printk(_THIS_IP_, fmt)
 extern int
 __ftrace_printk(unsigned long ip, const char *fmt, ...)
 	__attribute__ ((format (printf, 2, 3)));
kernel/trace/trace.c +0 -16
Index: linux-tip.git/kernel/trace/trace.c
===================================================================
@@ -3243,22 +3243,6 @@ void ftrace_dump(void)
 	(sizeof(struct trace_field) - offsetof(struct trace_field, print.buf))
 #define TRACE_CONT_BUF_SIZE sizeof(struct trace_field)
 
-/**
- * ftrace_printk - printf formatting in the ftrace buffer
- * @fmt - the printf format for printing.
- *
- * Note: __ftrace_printk is an internal function for ftrace_printk and
- *       the @ip is passed in via the ftrace_printk macro.
- *
- * This function allows a kernel developer to debug fast path sections
- * that printk is not appropriate for. By scattering in various
- * printk like tracing in the code, a developer can quickly see
- * where problems are occurring.
- *
- * This is intended as a debugging tool for the developer only.
- * Please reframe from leaving ftrace_printks scattered around in
- * your code.
- */
 int __ftrace_printk(unsigned long ip, const char *fmt, ...)
 {
 	struct trace_array *tr = &global_trace;
#7 lore raw

Re: [PATCH v2 3/2] ftrace: ftrace_printk doc moved

Quoted 16 lines · On Fri, 1 Aug 2008 14:17:42 -0400 (EDT) Steven Rostedt wrote:

Based on Randy Dunlap's suggestion, the ftrace_printk kernel-doc belongs with the ftrace_printk macro that should be used. Not with the __ftrace_printk internal function.

---
 include/linux/ftrace.h |   19 ++++++++++++++++++-
 kernel/trace/trace.c   |   16 ----------------
 2 files changed, 18 insertions(+), 17 deletions(-)

Index: linux-tip.git/include/linux/ftrace.h
===================================================================

include/linux/ftrace.h static inline void tracer_disable(void) · line 143

Quoted 8 lines
include/linux/ftrace.h +4 -1
@@ -137,7 +137,24 @@ static inline void tracer_disable(void)
 extern void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3);
 extern void ftrace_dump(void);
-# define ftrace_printk(x...) __ftrace_printk(_THIS_IP_, x)
+
+/**
+ * ftrace_printk - printf formatting in the ftrace buffer
+ * @fmt - the printf format for printing.

* @fmt: the printf format for printing

include/linux/ftrace.h static inline void tracer_disable(void) · line 154

Quoted 11 lines
include/linux/ftrace.h +11 -0
+ *
+ * Note: __ftrace_printk is an internal function for ftrace_printk and
+ *       the @ip is passed in via the ftrace_printk macro.
+ *
+ * This function allows a kernel developer to debug fast path sections
+ * that printk is not appropriate for. By scattering in various
+ * printk like tracing in the code, a developer can quickly see
+ * where problems are occurring.
+ *
+ * This is intended as a debugging tool for the developer only.
+ * Please reframe from leaving ftrace_printks scattered around in

refrain

Quoted 23 lines
include/linux/ftrace.h +3 -0
+ * your code.
+ */
+# define ftrace_printk(fmt...) __ftrace_printk(_THIS_IP_, fmt)
 extern int
 __ftrace_printk(unsigned long ip, const char *fmt, ...)
 	__attribute__ ((format (printf, 2, 3)));

Index: linux-tip.git/kernel/trace/trace.c
===================================================================

kernel/trace/trace.c +0 -9
@@ -3243,22 +3243,6 @@ void ftrace_dump(void)
 	(sizeof(struct trace_field) - offsetof(struct trace_field, print.buf))
 #define TRACE_CONT_BUF_SIZE sizeof(struct trace_field)
 
-/**
- * ftrace_printk - printf formatting in the ftrace buffer
- * @fmt - the printf format for printing.
- *
- * Note: __ftrace_printk is an internal function for ftrace_printk and
- *       the @ip is passed in via the ftrace_printk macro.
- *
- * This function allows a kernel developer to debug fast path sections
- * that printk is not appropriate for. By scattering in various

kernel/trace/trace.c void ftrace_dump(void) · line 3248

Quoted 10 lines
kernel/trace/trace.c +0 -7
- * printk like tracing in the code, a developer can quickly see
- * where problems are occurring.
- *
- * This is intended as a debugging tool for the developer only.
- * Please reframe from leaving ftrace_printks scattered around in
- * your code.
- */
 int __ftrace_printk(unsigned long ip, const char *fmt, ...)
 {
 	struct trace_array *tr = &global_trace;

---
~Randy
Linux Plumbers Conference, 17-19 September 2008, Portland, Oregon USA
http://linuxplumbersconf.org/

author maintainer
#8 lore raw

Re: [PATCH v2 3/2] ftrace: ftrace_printk doc moved

Quoted 13 lines · Randy Dunlap
Quoted 11 lines · On Fri, 1 Aug 2008 14:17:42 -0400 (EDT) Steven Rostedt wrote:

Based on Randy Dunlap's suggestion, the ftrace_printk kernel-doc belongs with the ftrace_printk macro that should be used. Not with the __ftrace_printk internal function.

---
 include/linux/ftrace.h |   19 ++++++++++++++++++-
 kernel/trace/trace.c   |   16 ----------------
 2 files changed, 18 insertions(+), 17 deletions(-)
Quoted 15 lines

include/linux/ftrace.h static inline void tracer_disable(void) · line 143

Quoted 13 lines

Index: linux-tip.git/include/linux/ftrace.h
===================================================================

include/linux/ftrace.h +4 -1
@@ -137,7 +137,24 @@ static inline void tracer_disable(void)
 extern void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3);
 extern void ftrace_dump(void);
-# define ftrace_printk(x...) __ftrace_printk(_THIS_IP_, x)
+
+/**
+ * ftrace_printk - printf formatting in the ftrace buffer
+ * @fmt - the printf format for printing.

* @fmt: the printf format for printing

OK, thanks.

Quoted 13 lines

include/linux/ftrace.h static inline void tracer_disable(void) · line 154

Quoted 12 lines
include/linux/ftrace.h +11 -0
+ *
+ * Note: __ftrace_printk is an internal function for ftrace_printk and
+ *       the @ip is passed in via the ftrace_printk macro.
+ *
+ * This function allows a kernel developer to debug fast path sections
+ * that printk is not appropriate for. By scattering in various
+ * printk like tracing in the code, a developer can quickly see
+ * where problems are occurring.
+ *
+ * This is intended as a debugging tool for the developer only.
+ * Please reframe from leaving ftrace_printks scattered around in
Quoted 1 line

refrain

Heh, I manually fixed that before sending out the original. I never fixed it in the quilt queue. Darn!

-- Steve

Quoted 5 lines
Quoted 5 lines
include/linux/ftrace.h +3 -0
+ * your code.
+ */
+# define ftrace_printk(fmt...) __ftrace_printk(_THIS_IP_, fmt)
 extern int
author maintainer
#9 lore raw

[PATCH v3 3/2] ftrace: ftrace_printk doc moved

In reply to Randy Dunlap

Based on Randy Dunlap's suggestion, the ftrace_printk kernel-doc belongs with the ftrace_printk macro that should be used. Not with the __ftrace_printk internal function.

---
 include/linux/ftrace.h |   19 ++++++++++++++++++-
 kernel/trace/trace.c   |   16 ----------------
 2 files changed, 18 insertions(+), 17 deletions(-)
include/linux/ftrace.h +18 -1
Index: linux-tip.git/include/linux/ftrace.h
===================================================================
@@ -137,7 +137,24 @@ static inline void tracer_disable(void)
 extern void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3);
 extern void ftrace_dump(void);
-# define ftrace_printk(x...) __ftrace_printk(_THIS_IP_, x)
+
+/**
+ * ftrace_printk - printf formatting in the ftrace buffer
+ * @fmt: the printf format for printing
+ *
+ * Note: __ftrace_printk is an internal function for ftrace_printk and
+ *       the @ip is passed in via the ftrace_printk macro.
+ *
+ * This function allows a kernel developer to debug fast path sections
+ * that printk is not appropriate for. By scattering in various
+ * printk like tracing in the code, a developer can quickly see
+ * where problems are occurring.
+ *
+ * This is intended as a debugging tool for the developer only.
+ * Please refrain from leaving ftrace_printks scattered around in
+ * your code.
+ */
+# define ftrace_printk(fmt...) __ftrace_printk(_THIS_IP_, fmt)
 extern int
 __ftrace_printk(unsigned long ip, const char *fmt, ...)
 	__attribute__ ((format (printf, 2, 3)));
kernel/trace/trace.c +0 -16
Index: linux-tip.git/kernel/trace/trace.c
===================================================================
@@ -3243,22 +3243,6 @@ void ftrace_dump(void)
 	(sizeof(struct trace_field) - offsetof(struct trace_field, print.buf))
 #define TRACE_CONT_BUF_SIZE sizeof(struct trace_field)
 
-/**
- * ftrace_printk - printf formatting in the ftrace buffer
- * @fmt - the printf format for printing.
- *
- * Note: __ftrace_printk is an internal function for ftrace_printk and
- *       the @ip is passed in via the ftrace_printk macro.
- *
- * This function allows a kernel developer to debug fast path sections
- * that printk is not appropriate for. By scattering in various
- * printk like tracing in the code, a developer can quickly see
- * where problems are occurring.
- *
- * This is intended as a debugging tool for the developer only.
- * Please reframe from leaving ftrace_printks scattered around in
- * your code.
- */
 int __ftrace_printk(unsigned long ip, const char *fmt, ...)
 {
 	struct trace_array *tr = &global_trace;
Acked-by
#10 lore raw

Re: [PATCH v3 3/2] ftrace: ftrace_printk doc moved

Quoted 6 lines · On Fri, 1 Aug 2008 16:45:49 -0400 (EDT) Steven Rostedt wrote:

Based on Randy Dunlap's suggestion, the ftrace_printk kernel-doc belongs with the ftrace_printk macro that should be used. Not with the __ftrace_printk internal function.

Quoted 52 lines
---
 include/linux/ftrace.h |   19 ++++++++++++++++++-
 kernel/trace/trace.c   |   16 ----------------
 2 files changed, 18 insertions(+), 17 deletions(-)

Index: linux-tip.git/include/linux/ftrace.h
===================================================================

include/linux/ftrace.h +18 -1
@@ -137,7 +137,24 @@ static inline void tracer_disable(void)
 extern void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3);
 extern void ftrace_dump(void);
-# define ftrace_printk(x...) __ftrace_printk(_THIS_IP_, x)
+
+/**
+ * ftrace_printk - printf formatting in the ftrace buffer
+ * @fmt: the printf format for printing
+ *
+ * Note: __ftrace_printk is an internal function for ftrace_printk and
+ *       the @ip is passed in via the ftrace_printk macro.
+ *
+ * This function allows a kernel developer to debug fast path sections
+ * that printk is not appropriate for. By scattering in various
+ * printk like tracing in the code, a developer can quickly see
+ * where problems are occurring.
+ *
+ * This is intended as a debugging tool for the developer only.
+ * Please refrain from leaving ftrace_printks scattered around in
+ * your code.
+ */
+# define ftrace_printk(fmt...) __ftrace_printk(_THIS_IP_, fmt)
 extern int
 __ftrace_printk(unsigned long ip, const char *fmt, ...)
 	__attribute__ ((format (printf, 2, 3)));

Index: linux-tip.git/kernel/trace/trace.c
===================================================================

kernel/trace/trace.c +0 -9
@@ -3243,22 +3243,6 @@ void ftrace_dump(void)
 	(sizeof(struct trace_field) - offsetof(struct trace_field, print.buf))
 #define TRACE_CONT_BUF_SIZE sizeof(struct trace_field)
 
-/**
- * ftrace_printk - printf formatting in the ftrace buffer
- * @fmt - the printf format for printing.
- *
- * Note: __ftrace_printk is an internal function for ftrace_printk and
- *       the @ip is passed in via the ftrace_printk macro.
- *
- * This function allows a kernel developer to debug fast path sections
- * that printk is not appropriate for. By scattering in various

kernel/trace/trace.c void ftrace_dump(void) · line 3248

Quoted 10 lines
kernel/trace/trace.c +0 -7
- * printk like tracing in the code, a developer can quickly see
- * where problems are occurring.
- *
- * This is intended as a debugging tool for the developer only.
- * Please reframe from leaving ftrace_printks scattered around in
- * your code.
- */
 int __ftrace_printk(unsigned long ip, const char *fmt, ...)
 {
 	struct trace_array *tr = &global_trace;

---
~Randy
Linux Plumbers Conference, 17-19 September 2008, Portland, Oregon USA
http://linuxplumbersconf.org/

Applied
#11 lore raw
In reply to Steven Rostedt
Quoted 5 lines · Steven Rostedt

I previously sent out a patch to add printk formatted recording to the ftrace buffers. Andrew Morton suggested to get rid of my hacked trace_cont structure and do it properly with one structure at the top and two other structures to handle the difference between the fields and the continuation entry.

applied all 3 patches to tip/tracing/ftrace - thanks Steve.

Ingo

#12 lore raw

PATCH 1/1 ftrace: build fix

Quoted 2 lines

From 00968bd78e3346f35dc809f7a7c811a3d88cf07c Mon Sep 17 00:00:00 2001

From: Ingo Molnar <[email protected]>
Date: Fri, 15 Aug 2008 17:48:02 +0200
Subject: [PATCH] ftrace: build fix
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit

fix:

In file included from init/main.c:65:
include/linux/ftrace.h:166: error: expected ‘,' or ‘;' before ‘{' token make[1]: *** [init/main.o] Error 1
make: *** [init/main.o] Error 2

---
 include/linux/ftrace.h |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)
include/linux/ftrace.h +4 -1
index f0d8e48..0d19aab 100644
@@ -162,7 +162,10 @@ extern void ftrace_dump(void);
 static inline void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) { }
 static inline int
-ftrace_printk(const char *fmt, ...) __attribute__ ((format (printf, 1, 0)))
+ftrace_printk(const char *fmt, ...) __attribute__ ((format (printf, 1, 0)));
+
+static inline int
+ftrace_printk(const char *fmt, ...)
 {
 	return 0;
 }