Skip to contentSkip to conversation

Add uprobes support

PATCH Josh Stone 4 participants 5 messages Jan 14, 2013 – Jan 15, 2013 4 messages not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Josh Stone

author
#1 lore raw

Review

Quoted 7 lines · Yannick Brosseau
(quoted diff) +6 -0
+	ret = uprobe_register(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);
 
+	uprobe_unregister(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);

I'm guessing that you're using a Fedora kernel? :)

These functions are not yet exported upstream, but coincidentally I just posted a patch for that a few days ago:

https://patchwork.kernel.org/patch/1962891/
https://lkml.org/lkml/2013/1/13/118

If there's any more complaint about this, maybe you can chime in to advocate for lttng's needs too.

Thanks,
Josh

A message that is not in the archive (sent off-list or not yet received): [email protected]

#3 lore raw
Quoted 8 lines · Josh Stone

Review

Quoted 6 lines · Yannick Brosseau
(quoted diff) +6 -0
+	ret = uprobe_register(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);
+	uprobe_unregister(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);

I'm guessing that you're using a Fedora kernel? :)

Oh, you're right... I did not realize that the vanilla kernel did not contains those export.

Quoted 10 lines

These functions are not yet exported upstream, but coincidentally I just posted a patch for that a few days ago:

https://patchwork.kernel.org/patch/1962891/
https://lkml.org/lkml/2013/1/13/118

If there's any more complaint about this, maybe you can chime in to advocate for lttng's needs too.

Ok, I'll weight in if necessary.

Yannick

A message that is not in the archive (sent off-list or not yet received): [email protected]

#5 lore raw
Quoted 15 lines · * Yannick Brosseau ([email protected]) wrote:
Quoted 8 lines · Josh Stone

Review

Quoted 6 lines · Yannick Brosseau
(quoted diff) +6 -0
+	ret = uprobe_register(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);
+	uprobe_unregister(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);

I'm guessing that you're using a Fedora kernel? :)

Oh, you're right... I did not realize that the vanilla kernel did not contains those export.

Quoted 4 lines

These functions are not yet exported upstream, but coincidentally I just posted a patch for that a few days ago:

Quoted 7 lines
Quoted 6 lines

https://patchwork.kernel.org/patch/1962891/
https://lkml.org/lkml/2013/1/13/118

If there's any more complaint about this, maybe you can chime in to advocate for lttng's needs too.

Ok, I'll weight in if necessary.

Yes, we should be able to help out if needed.

Meanwhile, we could use kallsyms to lookup these symbols and use them, as we already do in a couple of header under wrappers/ in lttng-modules.

Thanks,

Mathieu

Signature · 3 lines
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com

A message that is not in the archive (sent off-list or not yet received): 20130114195104.GA11211@Krystal

#7 lore raw

+1

Especially because Uprobe is in mainline since 3.5, would be nice to support it before it gets in 3.8+ ...

Cheers!
David

Mathieu Desnoyers:

Quoted 33 lines · * Yannick Brosseau ([email protected]) wrote:
Quoted 15 lines · Josh Stone
Quoted 8 lines · Yannick Brosseau

Review

Quoted 6 lines
(quoted diff) +6 -0
+	ret = uprobe_register(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);
+	uprobe_unregister(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);

I'm guessing that you're using a Fedora kernel? :)

Oh, you're right... I did not realize that the vanilla kernel did not contains those export.

Quoted 4 lines

These functions are not yet exported upstream, but coincidentally I just posted a patch for that a few days ago:

Quoted 7 lines
Quoted 6 lines

https://patchwork.kernel.org/patch/1962891/
https://lkml.org/lkml/2013/1/13/118

If there's any more complaint about this, maybe you can chime in to advocate for lttng's needs too.

Ok, I'll weight in if necessary.

Yes, we should be able to help out if needed.

Meanwhile, we could use kallsyms to lookup these symbols and use them, as we already do in a couple of header under wrappers/ in lttng-modules.

Thanks,

Mathieu

#8 lore raw
In reply to a message not in the archive
Quoted 4 lines · * Yannick Brosseau ([email protected]) wrote:

The added support is basic. It create an event with no data associated to the file path + offset specified.
As per the structures currently used we cannot pass a file path bigger than 256 chars.

Using a pointer to a user-space string will allow you to overcome this limitation. e.g. passing through the kernel ABI:

char name[LTTNG_KERNEL_SYM_NAME_LEN]
char __user *path;
uint64_t offset;

Quoted 27 lines
---
 README                 |    5 +-
 lttng-abi.c            |    3 +
 lttng-abi.h            |    7 +++
 lttng-events.c         |   18 ++++++
 lttng-events.h         |   38 ++++++++++++
 probes/Makefile        |    5 ++
 probes/lttng-uprobes.c |  156 ++++++++++++++++++++++++++++++++++++++++++++++++
 7 files changed, 231 insertions(+), 1 deletion(-)
 create mode 100644 probes/lttng-uprobes.c
README +1 -1
index 1bcd5b2..7afdaa3 100644
@@ -9,7 +9,7 @@ need for additional patches. Other features:
 - Produces CTF (Common Trace Format) natively,
   (http://www.efficios.com/ctf)
 - Tracepoints, Function tracer, CPU Performance Monitoring Unit (PMU)
-  counters, kprobes, and kretprobes support,
+  counters, kprobes, kretprobes and uprobes support,
 - Integrated interface for both kernel and userspace tracing,
 - Have the ability to attach "context" information to events in the
   trace (e.g. any PMU counter, pid, ppid, tid, comm name, etc).

README CONFIG_KPROBES: · line 91

Quoted 6 lines
README +3 -0
@@ -86,6 +86,9 @@ CONFIG_KPROBES:
 CONFIG_KRETPROBES:
             Dynamic function entry/return probe.
                lttng enable-event -k --function ...
+CONFIG_UPROBES:
+            Dynamic userspace probe.
+               lttng enable-event -k --uprobe ...

It would be great to implement this feature with integrated lookup of file/line number (using dwarf). I think perf already has some code that does it. Adding Masami in the loop would be very relevant. In terms of usability it will make all different in the world. Of course, this won't change the ABI nor lttng-modules implementation, only the lttng-tools side.

Quoted 97 lines
README +0 -0
 * Note about Perf PMU counters support
lttng-abi.c +3 -0
index 25a350a..fe428b2 100644
@@ -610,6 +610,9 @@ int lttng_abi_create_event(struct file *channel_file,
 	case LTTNG_KERNEL_FUNCTION:
 		event_param->u.ftrace.symbol_name[LTTNG_KERNEL_SYM_NAME_LEN - 1] = '\0';
 		break;
+	case LTTNG_KERNEL_UPROBE:
+		event_param->u.uprobe.path[LTTNG_KERNEL_SYM_NAME_LEN - 1] = '\0';
+		break;
 	default:
 		break;
 	}
lttng-abi.h +7 -0
index 8d3ecdd..5abdb39 100644
@@ -34,6 +34,7 @@ enum lttng_kernel_instrumentation {
 	LTTNG_KERNEL_KRETPROBE	= 3,
 	LTTNG_KERNEL_NOOP	= 4,	/* not hooked */
 	LTTNG_KERNEL_SYSCALL	= 5,
+	LTTNG_KERNEL_UPROBE	= 6,
 };
 
 /*
@@ -79,6 +80,11 @@ struct lttng_kernel_function_tracer {
 	char symbol_name[LTTNG_KERNEL_SYM_NAME_LEN];
 }__attribute__((packed));
 
+struct lttng_kernel_uprobe {
+	char path[LTTNG_KERNEL_SYM_NAME_LEN];
+	uint64_t offset;
+}__attribute__((packed));
+
 /*
  * For syscall tracing, name = '\0' means "enable all".
  */
@@ -94,6 +100,7 @@ struct lttng_kernel_event {
 		struct lttng_kernel_kretprobe kretprobe;
 		struct lttng_kernel_kprobe kprobe;
 		struct lttng_kernel_function_tracer ftrace;
+		struct lttng_kernel_uprobe uprobe;
 		char padding[LTTNG_KERNEL_EVENT_PADDING2];
 	} u;
 }__attribute__((packed));
lttng-events.c +18 -0
index 4f30904..38c901f 100644
@@ -392,6 +392,16 @@ struct lttng_event *lttng_event_create(struct lttng_channel *chan,
 		if (!event->desc)
 			goto register_error;
 		break;
+	case LTTNG_KERNEL_UPROBE:
+		ret = lttng_uprobes_register(event_param->name,
+				event_param->u.uprobe.path,
+				event_param->u.uprobe.offset,
+				event);
+		if (ret)
+			goto register_error;
+		ret = try_module_get(event->desc->owner);
+		WARN_ON_ONCE(!ret);
+		break;
 	default:
 		WARN_ON_ONCE(1);
 	}
@@ -443,6 +453,10 @@ int _lttng_event_unregister(struct lttng_event *event)
 	case LTTNG_KERNEL_NOOP:
 		ret = 0;
 		break;
+	case LTTNG_KERNEL_UPROBE:
+		lttng_uprobes_unregister(event);
+		ret = 0;
+		break;
 	default:
 		WARN_ON_ONCE(1);
 	}
@@ -473,6 +487,10 @@ void _lttng_event_destroy(struct lttng_event *event)
 		break;
 	case LTTNG_KERNEL_NOOP:
 		break;
+	case LTTNG_KERNEL_UPROBE:
+		module_put(event->desc->owner);
+		lttng_uprobes_destroy_private(event);
+		break;
 	default:
 		WARN_ON_ONCE(1);
 	}
lttng-events.h +0 -0
index 09d5618..e630db0 100644

lttng-events.h line 31

Quoted 6 lines
lttng-events.h +3 -0
@@ -26,6 +26,9 @@
 #include <linux/version.h>
 #include <linux/list.h>
 #include <linux/kprobes.h>
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,5,0))
+#include <linux/uprobes.h>
+#endif

please just include a wrapper/uprobes.h , so we move the ifdef to wrapper/.

lttng-events.h struct lttng_event { · line 212

Quoted 11 lines
lttng-events.h +4 -0
 #include "wrapper/uuid.h"
 #include "lttng-abi.h"
 #include "lttng-abi-old.h"
@@ -203,6 +206,13 @@ struct lttng_event {
 		struct {
 			char *symbol_name;
 		} ftrace;
+		struct {
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(3,5,0))
+			struct uprobe_consumer up_consumer;
+#endif

an ifdef in a wrapper that defines struct uprobe_consumer locally if necessary would be better. We need to shove all this compatibility code under wrappers/ whenever possible.

Quoted 46 lines
lttng-events.h +31 -0
+			struct inode *inode;
+			loff_t offset;
+		} uprobe;
 	} u;
 	struct list_head list;		/* Event list */
 	unsigned int metadata_dumped:1;
@@ -406,6 +416,34 @@ void lttng_kprobes_destroy_private(struct lttng_event *event)
 }
 #endif
 
+#ifdef CONFIG_UPROBES
+int lttng_uprobes_register(const char *name,
+		const char *path,
+		uint64_t offset,
+		struct lttng_event *event);
+void lttng_uprobes_unregister(struct lttng_event *event);
+void lttng_uprobes_destroy_private(struct lttng_event *event);
+#else
+static inline
+int lttng_uprobes_register(const char *name,
+		const char *path,
+		uint64_t offset,
+		struct lttng_event *event)
+{
+	return -ENOSYS;
+}
+
+static inline
+void lttng_uprobes_unregister(struct lttng_event *event)
+{
+}
+
+static inline
+void lttng_uprobes_destroy_private(struct lttng_event *event)
+{
+}
+#endif
+
 #ifdef CONFIG_KRETPROBES
 int lttng_kretprobes_register(const char *name,
 		const char *symbol_name,
probes/Makefile +0 -0
index 088cd5f..de0f1a7 100644

probes/Makefile ifneq ($(CONFIG_DYNAMIC_FTRACE),) · line 215

Quoted 8 lines
probes/Makefile +5 -0
@@ -208,6 +208,11 @@ ifneq ($(CONFIG_DYNAMIC_FTRACE),)
 obj-m += lttng-ftrace.o
 endif
 
+ifneq ($(CONFIG_UPROBES),)
+obj-m += lttng-uprobes.o
+endif
+
+

Extra whiteline to remove.

Quoted 113 lines
probes/Makefile +0 -0
 endif
 
 else
probes/lttng-uprobes.c new file +104 -0
new file mode 100644
index 0000000..1d68273
@@ -0,0 +1,156 @@
+/*
+ * probes/lttng-uprobes.c
+ *
+ * LTTng uprobes integration module.
+ *
+ * Copyright (C) 2013 Yannick Brosseau <[email protected]>
+ * Copyright (C) 2009-2012 Mathieu Desnoyers <[email protected]>
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; only
+ * version 2.1 of the License.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <linux/module.h>
+#include <linux/uprobes.h>
+#include <linux/slab.h>
+#include <linux/namei.h>
+#include <linux/fs.h>
+#include "../lttng-events.h"
+#include "../wrapper/ringbuffer/frontend_types.h"
+#include "../wrapper/vmalloc.h"
+#include "../lttng-tracer.h"
+
+static
+int lttng_uprobes_handler_pre(struct uprobe_consumer *uc, struct pt_regs *regs)
+{
+	struct lttng_event *event =
+		container_of(uc, struct lttng_event, u.uprobe.up_consumer);
+	struct lttng_channel *chan = event->chan;
+	struct lib_ring_buffer_ctx ctx;
+	int ret;
+
+	if (unlikely(!ACCESS_ONCE(chan->session->active)))
+		return 0;
+	if (unlikely(!ACCESS_ONCE(chan->enabled)))
+		return 0;
+	if (unlikely(!ACCESS_ONCE(event->enabled)))
+		return 0;
+
+	lib_ring_buffer_ctx_init(&ctx, chan->chan, event, 0,
+				 0, -1);
+	ret = chan->ops->event_reserve(&ctx, event->id);
+	if (ret < 0)
+		return 0;
+	lib_ring_buffer_align_ctx(&ctx, 0);
+	chan->ops->event_write(&ctx, 0, 0);
+	chan->ops->event_commit(&ctx);
+	return 0;
+}
+
+/*
+ * Create event description
+ */
+static
+int lttng_create_uprobe_event(const char *name, struct lttng_event *event)
+{
+	struct lttng_event_desc *desc;
+	int ret;
+
+	desc = kzalloc(sizeof(*event->desc), GFP_KERNEL);
+	if (!desc)
+		return -ENOMEM;
+	desc->name = kstrdup(name, GFP_KERNEL);
+	if (!desc->name) {
+		ret = -ENOMEM;
+		goto error_str;
+	}
+	desc->owner = THIS_MODULE;
+	event->desc = desc;
+
+	return 0;
+
+error_str:
+	kfree(desc);
+	return ret;
+}
+
+int lttng_uprobes_register(const char *name,
+			   const char *path_name,
+			   uint64_t offset,
+			   struct lttng_event *event)
+{
+	int ret;
+
+	if (path_name[0] == '\0')
+		path_name = NULL;
+
+	ret = lttng_create_uprobe_event(name, event);
+	if (ret)
+		goto error;
+	memset(&event->u.uprobe.up_consumer, 0,
+		sizeof(event->u.uprobe.up_consumer));
+	event->u.uprobe.up_consumer.handler = lttng_uprobes_handler_pre;
+	if (path_name) {

probes/lttng-uprobes.c line 114

Quoted 10 lines
probes/lttng-uprobes.c new file +10 -0
+		struct path path;
+		ret = kern_path(path_name, LOOKUP_FOLLOW, &path);
+		if (ret)
+			goto path_error;
+
+		event->u.uprobe.inode = igrab(path.dentry->d_inode);
+	}
+	event->u.uprobe.offset = offset;
+
+	/* TODO Is this also true for uprobes?

The memory we just zmalloc'd could theoretically trigger a minor page fault (if the memory segment is ever very very large and the kernel needs to fall back on vmalloc).

Quoted 8 lines
probes/lttng-uprobes.c new file +8 -0
+	 * Ensure the memory we just allocated don't trigger page faults.
+	 * Well.. kprobes itself puts the page fault handler on the blacklist,
+	 * but we can never be too careful.
+	 */
+	wrapper_vmalloc_sync_all();
+
+	ret = uprobe_register(event->u.uprobe.inode,
+			event->u.uprobe.offset,

probes/lttng-uprobes.c line 132

Quoted 10 lines
probes/lttng-uprobes.c new file +10 -0
+			&event->u.uprobe.up_consumer);
+	if (ret)
+		goto register_error;
+	return 0;
+
+register_error:
+	iput(event->u.uprobe.inode);
+path_error:
+	kfree(event->desc->name);
+	kfree(event->desc);

not sure why kfree these, which are not allocated within this function ? Should be cleaned up.

probes/lttng-uprobes.c line 142

Quoted 10 lines
probes/lttng-uprobes.c new file +10 -0
+error:
+	return ret;
+}
+EXPORT_SYMBOL_GPL(lttng_uprobes_register);
+
+void lttng_uprobes_unregister(struct lttng_event *event)
+{
+	uprobe_unregister(event->u.uprobe.inode,
+			event->u.uprobe.offset,
+			&event->u.uprobe.up_consumer);

Hrm, we should probably report to uprobes developers that uprobe_unregister() should return errors instead of silently returning.

Thanks !

Mathieu

Quoted 22 lines
probes/lttng-uprobes.c new file +14 -0
+}
+EXPORT_SYMBOL_GPL(lttng_uprobes_unregister);
+
+void lttng_uprobes_destroy_private(struct lttng_event *event)
+{
+	iput(event->u.uprobe.inode);
+	kfree(event->desc->name);
+	kfree(event->desc);
+}
+EXPORT_SYMBOL_GPL(lttng_uprobes_destroy_private);
+
+MODULE_LICENSE("GPL and additional rights");
+MODULE_AUTHOR("Yannick Brosseau");
+MODULE_DESCRIPTION("Linux Trace Toolkit Uprobes Support");
Signature · 7 lines
1.7.10.4


_______________________________________________
lttng-dev mailing list
[email protected]
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Signature · 3 lines
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com