Skip to contentSkip to conversation

x86, ftrace, hw-branch-tracer: reset trace on close

PATCH In review Markus Metzger 2 participants 3 messages Jan 16, 2009 Tracing

lore
Share & tools
lore
#0 lore raw

PATCH 4/5 x86, ftrace, hw-branch-tracer: reset trace on close

Reset the ftrace buffer on close. Since we use cyclic buffers to store the trace, the trace is not contiguous, anyway. This behaviour should be more natural.

---

kernel/trace/trace_hw_branches.c +7 -1
Index: ftrace/kernel/trace/trace_hw_branches.c
===================================================================
@@ -272,6 +272,11 @@
 	mutex_unlock(&bts_tracer_mutex);
 }
 
+static void trace_bts_close(struct trace_iterator *iter)
+{
+	tracing_reset_online_cpus(iter->tr);
+}
+
 void trace_hw_branch_oops(void)
 {
 	mutex_lock(&bts_tracer_mutex);
@@ -290,7 +295,8 @@
 	.print_line	= bts_trace_print_line,
 	.start		= bts_trace_start,
 	.stop		= bts_trace_stop,
-	.open		= trace_bts_prepare
+	.open		= trace_bts_prepare,
+	.close		= trace_bts_close
 };
 
 __init static int init_bts_trace(void)

---------------------------------------------------------------------
Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen Germany
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer
Registergericht: Muenchen HRB 47456 Ust.-IdNr.
VAT Registration No.: DE129385895
Citibank Frankfurt (BLZ 502 109 00) 600119052

This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.

#1 lore raw
Quoted 3 lines · Markus Metzger

Reset the ftrace buffer on close. Since we use cyclic buffers to store the trace, the trace is not contiguous, anyway. This behaviour should be more natural.

So you want multiple reads of /debug/tracing/trace to return different output? Kind of like the trace_pipe? But trace_pipe is a real consumer.

kernel/trace/trace_hw_branches.c

Quoted 21 lines

---

Index: ftrace/kernel/trace/trace_hw_branches.c
===================================================================

kernel/trace/trace_hw_branches.c +5 -0
@@ -272,6 +272,11 @@
 	mutex_unlock(&bts_tracer_mutex);
 }
 
+static void trace_bts_close(struct trace_iterator *iter)
+{
+	tracing_reset_online_cpus(iter->tr);
+}
+
 void trace_hw_branch_oops(void)
 {
 	mutex_lock(&bts_tracer_mutex);
@@ -290,7 +295,8 @@

BTW, can you run your diff with the -p option. This will help in knowing which function the change is in. There's some hunks where it is ambiguous. If you use quilt, you can add:

QUILT_DIFF_OPTS='-p'

to you .quiltrc file.

-- Steve

Quoted 24 lines
kernel/trace/trace_hw_branches.c +2 -2
 	.print_line	= bts_trace_print_line,
 	.start		= bts_trace_start,
 	.stop		= bts_trace_stop,
-	.open		= trace_bts_prepare
+	.open		= trace_bts_prepare,
+	.close		= trace_bts_close
 };
 
 __init static int init_bts_trace(void)
---------------------------------------------------------------------

Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen Germany
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer
Registergericht: Muenchen HRB 47456 Ust.-IdNr.
VAT Registration No.: DE129385895
Citibank Frankfurt (BLZ 502 109 00) 600119052

This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.

#2 lore raw
Quoted 5 lines

-----Original Message-----
From: Steven Rostedt [mailto:[email protected]]
Sent: Friday, January 16, 2009 4:02 PM
To: Metzger, Markus T

Quoted 7 lines
Quoted 3 lines · Markus Metzger

Reset the ftrace buffer on close. Since we use cyclic buffers to store the trace, the trace is not contiguous, anyway. This behaviour should be more natural.

So you want multiple reads of /debug/tracing/trace to return different output? Kind of like the trace_pipe? But trace_pipe is a real consumer.

Yes.
The trace chunks are not contiguous (in this version), so it does not make sense to append them. I don't expect trace_pipe to work with the hw-branch-tracer.

Quoted 7 lines

BTW, can you run your diff with the -p option. This will help in knowing which function the change is in. There's some hunks where it is ambiguous. If you use quilt, you can add:

QUILT_DIFF_OPTS='-p'

to you .quiltrc file.

Thanks.

markus.
---------------------------------------------------------------------
Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen Germany
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Douglas Lusk, Peter Gleissner, Hannes Schwaderer
Registergericht: Muenchen HRB 47456 Ust.-IdNr.
VAT Registration No.: DE129385895
Citibank Frankfurt (BLZ 502 109 00) 600119052

This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.