Skip to contentSkip to conversation

x86: kprobes: Show correct blaclkist in debugfs

PATCH In review Masami Hiramatsu 3 participants 13 messages Dec 17, 2018 – Jan 1, 2019

Subsystems (2)
lore
Share & tools
lore

Masami Hiramatsu

author maintainer
#0 lore raw

v2 cover x86: kprobes: Show correct blaclkist in debugfs

This is v2 series for showing correct kprobe blacklist in debugfs.

v1 is here:

https://lkml.org/lkml/2018/12/7/517

I splitted the RFC v1 patch into x86 and generic parts, also added a patch to remove unneeded arch-specific blacklist check function (because those have been added to the generic blacklist.)

If this style is good, I will make another series for the archs which have own arch_within_kprobe_blacklist(), and eventually replace that with arch_populate_kprobe_blacklist() so that user can get the correct kprobe blacklist in debugfs.

Thank you,

---

Masami Hiramatsu (3):
kprobes: Blacklist symbols in arch-defined prohibited area x86/kprobes: Show x86-64 specific blacklisted symbols correctly x86/kprobes: Remove unneeded arch_within_kprobe_blacklist from x86

 arch/x86/kernel/kprobes/core.c |    8 ++---
 include/linux/kprobes.h        |    3 ++
 kernel/kprobes.c               |   67 ++++++++++++++++++++++++++++++++--------
 3 files changed, 59 insertions(+), 19 deletions(-)

--
Masami Hiramatsu (Linaro) <[email protected]>

author maintainer
#1 lore raw

PATCH v2 1/3 kprobes: Blacklist symbols in arch-defined prohibited area

Blacklist symbols in arch-defined probe-prohibited areas. With this change, user can see all symbols which are prohibited to probe in debugfs.

All archtectures which have custom prohibit areas should define its own arch_populate_kprobe_blacklist() function, but unless that, all symbols marked __kprobes are blacklisted.

---
Changes in v2:
- Add area based blacklist adding helper function
- Blacklist __kprobes_text symbols by default
- Add __init to default function

---
 include/linux/kprobes.h |    3 ++
 kernel/kprobes.c        |   67 +++++++++++++++++++++++++++++++++++++----------
 2 files changed, 56 insertions(+), 14 deletions(-)
include/linux/kprobes.h +3 -0
index e64b26c81c2f..e07e91daaacc 100644
@@ -242,10 +242,13 @@ extern int arch_init_kprobes(void);
 extern void show_registers(struct pt_regs *regs);
 extern void kprobes_inc_nmissed_count(struct kprobe *p);
 extern bool arch_within_kprobe_blacklist(unsigned long addr);
+extern int arch_populate_kprobe_blacklist(void);
 extern bool arch_kprobe_on_func_entry(unsigned long offset);
 extern bool kprobe_on_func_entry(kprobe_opcode_t *addr, const char *sym, unsigned long offset);
 
 extern bool within_kprobe_blacklist(unsigned long addr);
+extern int kprobe_add_ksym_blacklist(unsigned long entry);
+extern int kprobe_add_area_blacklist(unsigned long start, unsigned long end);
 
 struct kprobe_insn_cache {
 	struct mutex mutex;
kernel/kprobes.c +53 -14

85 lines not on this page · raw message

#2 lore raw

[tip:perf/core] kprobes: Blacklist symbols in arch-defined prohibited area

kprobes: Blacklist symbols in arch-defined prohibited area
Commit-ID:  fb1a59fae8baa3f3c69b72a87ff94fc4fa5683ec
Gitweb:     https://git.kernel.org/tip/fb1a59fae8baa3f3c69b72a87ff94fc4fa5683ec
Author:     Masami Hiramatsu <[email protected]>
AuthorDate: Mon, 17 Dec 2018 17:20:55 +0900
Committer:  Ingo Molnar <[email protected]>
CommitDate: Mon, 17 Dec 2018 17:48:38 +0100

kprobes: Blacklist symbols in arch-defined prohibited area

Blacklist symbols in arch-defined probe-prohibited areas. With this change, user can see all symbols which are prohibited to probe in debugfs.

All archtectures which have custom prohibit areas should define its own arch_populate_kprobe_blacklist() function, but unless that, all symbols marked __kprobes are blacklisted.

---
 include/linux/kprobes.h |  3 +++
 kernel/kprobes.c        | 67 ++++++++++++++++++++++++++++++++++++++-----------
 2 files changed, 56 insertions(+), 14 deletions(-)
include/linux/kprobes.h +3 -0
index e909413e4e38..5da8a1de2187 100644
@@ -242,10 +242,13 @@ extern int arch_init_kprobes(void);
 extern void show_registers(struct pt_regs *regs);
 extern void kprobes_inc_nmissed_count(struct kprobe *p);
 extern bool arch_within_kprobe_blacklist(unsigned long addr);
+extern int arch_populate_kprobe_blacklist(void);
 extern bool arch_kprobe_on_func_entry(unsigned long offset);
 extern bool kprobe_on_func_entry(kprobe_opcode_t *addr, const char *sym, unsigned long offset);
 
 extern bool within_kprobe_blacklist(unsigned long addr);
+extern int kprobe_add_ksym_blacklist(unsigned long entry);
+extern int kprobe_add_area_blacklist(unsigned long start, unsigned long end);
 
 struct kprobe_insn_cache {
 	struct mutex mutex;
kernel/kprobes.c +53 -14

85 lines not on this page · raw message

author maintainer
#3 lore raw

PATCH v2 2/3 x86/kprobes: Show x86-64 specific blacklisted symbols correctly

In reply to Masami Hiramatsu

Show x86-64 specific blacklisted symbols in debugfs. Since x86-64 prohibits probing on symbols which are in entry text, those should be shown.

---
 arch/x86/kernel/kprobes/core.c |    6 ++++++
 1 file changed, 6 insertions(+)
arch/x86/kernel/kprobes/core.c +6 -0
index 308bf103cc73..2a291fcb6443 100644
@@ -1034,6 +1034,12 @@ bool arch_within_kprobe_blacklist(unsigned long addr)
 		 addr < (unsigned long)__entry_text_end);
 }
 
+int __init arch_populate_kprobe_blacklist(void)
+{
+	return kprobe_add_area_blacklist((unsigned long)__entry_text_start,
+					 (unsigned long)__entry_text_end);
+}
+
 int __init arch_init_kprobes(void)
 {
 	return 0;
#4 lore raw

[tip:perf/core] kprobes/x86: Show x86-64 specific blacklisted symbols correctly

kprobes/x86: Show x86-64 specific blacklisted symbols correctly
Commit-ID:  fe6e65615415987629a2dda583b4495677d8c388
Gitweb:     https://git.kernel.org/tip/fe6e65615415987629a2dda583b4495677d8c388
Author:     Masami Hiramatsu <[email protected]>
AuthorDate: Mon, 17 Dec 2018 17:21:24 +0900
Committer:  Ingo Molnar <[email protected]>
CommitDate: Mon, 17 Dec 2018 17:48:39 +0100

kprobes/x86: Show x86-64 specific blacklisted symbols correctly

Show x86-64 specific blacklisted symbols in debugfs.

Since x86-64 prohibits probing on symbols which are in entry text, those should be shown.

---
 arch/x86/kernel/kprobes/core.c | 6 ++++++
 1 file changed, 6 insertions(+)
arch/x86/kernel/kprobes/core.c +6 -0
index c33b06f5faa4..6011a4a90f0a 100644
@@ -1034,6 +1034,12 @@ bool arch_within_kprobe_blacklist(unsigned long addr)
 		 addr < (unsigned long)__entry_text_end);
 }
 
+int __init arch_populate_kprobe_blacklist(void)
+{
+	return kprobe_add_area_blacklist((unsigned long)__entry_text_start,
+					 (unsigned long)__entry_text_end);
+}
+
 int __init arch_init_kprobes(void)
 {
 	return 0;
author maintainer
#5 lore raw

PATCH v2 3/3 x86/kprobes: Remove unneeded arch_within_kprobe_blacklist from x86

In reply to Masami Hiramatsu

Remove x86 specific arch_within_kprobe_blacklist().

Since we have already added all blacklisted symbols to kprobe blacklist by arch_populate_kprobe_blacklist(), we don't need arch_within_kprobe_blacklist() on x86 anymore.

---
 arch/x86/kernel/kprobes/core.c |    8 --------
 1 file changed, 8 deletions(-)
arch/x86/kernel/kprobes/core.c +0 -8
index 2a291fcb6443..4ba75afba527 100644
@@ -1026,14 +1026,6 @@ int kprobe_fault_handler(struct pt_regs *regs, int trapnr)
 }
 NOKPROBE_SYMBOL(kprobe_fault_handler);
 
-bool arch_within_kprobe_blacklist(unsigned long addr)
-{
-	return  (addr >= (unsigned long)__kprobes_text_start &&
-		 addr < (unsigned long)__kprobes_text_end) ||
-		(addr >= (unsigned long)__entry_text_start &&
-		 addr < (unsigned long)__entry_text_end);
-}
-
 int __init arch_populate_kprobe_blacklist(void)
 {
 	return kprobe_add_area_blacklist((unsigned long)__entry_text_start,
#6 lore raw

[tip:perf/core] kprobes/x86: Remove unneeded arch_within_kprobe_blacklist from x86

kprobes/x86: Remove unneeded arch_within_kprobe_blacklist from x86
Commit-ID:  8162b3d1a728cf63abf54be4167dd9beec5d9d37
Gitweb:     https://git.kernel.org/tip/8162b3d1a728cf63abf54be4167dd9beec5d9d37
Author:     Masami Hiramatsu <[email protected]>
AuthorDate: Mon, 17 Dec 2018 17:21:53 +0900
Committer:  Ingo Molnar <[email protected]>
CommitDate: Mon, 17 Dec 2018 17:48:40 +0100

kprobes/x86: Remove unneeded arch_within_kprobe_blacklist from x86

Remove x86 specific arch_within_kprobe_blacklist().

Since we have already added all blacklisted symbols to the kprobe blacklist by arch_populate_kprobe_blacklist(), we don't need arch_within_kprobe_blacklist() on x86 anymore.

---
 arch/x86/kernel/kprobes/core.c | 8 --------
 1 file changed, 8 deletions(-)
arch/x86/kernel/kprobes/core.c +0 -8
index 6011a4a90f0a..d5f88fe57064 100644
@@ -1026,14 +1026,6 @@ int kprobe_fault_handler(struct pt_regs *regs, int trapnr)
 }
 NOKPROBE_SYMBOL(kprobe_fault_handler);
 
-bool arch_within_kprobe_blacklist(unsigned long addr)
-{
-	return  (addr >= (unsigned long)__kprobes_text_start &&
-		 addr < (unsigned long)__kprobes_text_end) ||
-		(addr >= (unsigned long)__entry_text_start &&
-		 addr < (unsigned long)__entry_text_end);
-}
-
 int __init arch_populate_kprobe_blacklist(void)
 {
 	return kprobe_add_area_blacklist((unsigned long)__entry_text_start,
Tested-by on the v2 series
#7 lore raw
In reply to Masami Hiramatsu
Quoted 12 lines · Masami Hiramatsu

This is v2 series for showing correct kprobe blacklist in debugfs.

v1 is here:

https://lkml.org/lkml/2018/12/7/517

I splitted the RFC v1 patch into x86 and generic parts, also added a patch to remove unneeded arch-specific blacklist check function (because those have been added to the generic blacklist.)

Quoted 6 lines

If this style is good, I will make another series for the archs which have own arch_within_kprobe_blacklist(), and eventually replace that with arch_populate_kprobe_blacklist() so that user can get the correct kprobe blacklist in debugfs.

Thank you,

Looks good to me. Thanks!

Side question: there are certain symbols in arch/x86/xen that should be blacklisted explicitly, because they're non-attachable.

More exactly, all functions defined in arch/x86/xen/spinlock.c, arch/x86/xen/time.c and arch/x86/xen/irq.c.

The reason is that these files are compiled without -pg to allow the usage of ftrace within a Xen domain apparently (from arch/x86/xen/Makefile):

ifdef CONFIG_FUNCTION_TRACER
# Do not profile debug and lowlevel utilities
CFLAGS_REMOVE_spinlock.o = -pg
CFLAGS_REMOVE_time.o = -pg
CFLAGS_REMOVE_irq.o = -pg
endif

Do you see a nice and clean way to blacklist all these functions (something like arch_populate_kprobe_blacklist()), or should we just flag all of them explicitly with NOKPROBE_SYMBOL()?

Thanks,
-Andrea

author maintainer
#8 lore raw
Quoted 18 lines · On Mon, 17 Dec 2018 16:47:13 +0100 Andrea Righi <[email protected]> wrote:
Quoted 17 lines · Masami Hiramatsu

This is v2 series for showing correct kprobe blacklist in debugfs.

v1 is here:

https://lkml.org/lkml/2018/12/7/517

I splitted the RFC v1 patch into x86 and generic parts, also added a patch to remove unneeded arch-specific blacklist check function (because those have been added to the generic blacklist.)

If this style is good, I will make another series for the archs which have own arch_within_kprobe_blacklist(), and eventually replace that with arch_populate_kprobe_blacklist() so that user can get the correct kprobe blacklist in debugfs.

Quoted 5 lines
Quoted 1 line

Thank you,

Looks good to me. Thanks!

Thank you for testing!

Quoted 11 lines

Side question: there are certain symbols in arch/x86/xen that should be blacklisted explicitly, because they're non-attachable.

More exactly, all functions defined in arch/x86/xen/spinlock.c, arch/x86/xen/time.c and arch/x86/xen/irq.c.

The reason is that these files are compiled without -pg to allow the usage of ftrace within a Xen domain apparently (from arch/x86/xen/Makefile):

Quoted 6 lines

ifdef CONFIG_FUNCTION_TRACER
# Do not profile debug and lowlevel utilities
CFLAGS_REMOVE_spinlock.o = -pg
CFLAGS_REMOVE_time.o = -pg
CFLAGS_REMOVE_irq.o = -pg
endif

Actually, the reason why you can not probe those functions via tracing/kprobe_events is just a side effect. You can probe it if you write a kprobe module. Since the kprobe_events depends on some ftrace tracing functions, it sometimes cause a recursive call problem. To avoid this issue, I have introduced a CONFIG_KPROBE_EVENTS_ON_NOTRACE, see commit 45408c4f9250 ("tracing: kprobes: Prohibit probing on notrace function").

If you set CONFIG_KPROBE_EVENTS_ON_NOTRACE=n, you can continue putting probes on Xen spinlock functions too.

Quoted 3 lines

Do you see a nice and clean way to blacklist all these functions (something like arch_populate_kprobe_blacklist()), or should we just flag all of them explicitly with NOKPROBE_SYMBOL()?

As I pointed, you can probe it via your own kprobe module. Like systemtap, you still can probe it. The blacklist is for "kprobes", not for "kprobe_events". (Those are used to same, but since the above commit, those are different now)

I think the most sane solution is, identifying which (combination of) functions in ftrace (kernel/trace/*) causes a problem, marking those NOKPROBE_SYMBOL() and removing CONFIG_KPROBE_EVENTS_ON_NOTRACE.

Thank you,

Signature · 1 line
Masami Hiramatsu <[email protected]>
#9 lore raw

On Tue, Dec 18, 2018 at 01:50:26PM +0900, Masami Hiramatsu wrote:
...

Quoted 25 lines
Quoted 10 lines

Side question: there are certain symbols in arch/x86/xen that should be blacklisted explicitly, because they're non-attachable.

More exactly, all functions defined in arch/x86/xen/spinlock.c, arch/x86/xen/time.c and arch/x86/xen/irq.c.

The reason is that these files are compiled without -pg to allow the usage of ftrace within a Xen domain apparently (from arch/x86/xen/Makefile):

Quoted 6 lines

ifdef CONFIG_FUNCTION_TRACER
# Do not profile debug and lowlevel utilities
CFLAGS_REMOVE_spinlock.o = -pg
CFLAGS_REMOVE_time.o = -pg
CFLAGS_REMOVE_irq.o = -pg
endif

Actually, the reason why you can not probe those functions via tracing/kprobe_events is just a side effect. You can probe it if you write a kprobe module. Since the kprobe_events depends on some ftrace tracing functions, it sometimes cause a recursive call problem. To avoid this issue, I have introduced a CONFIG_KPROBE_EVENTS_ON_NOTRACE, see commit 45408c4f9250 ("tracing: kprobes: Prohibit probing on notrace function").

Quoted 2 lines

If you set CONFIG_KPROBE_EVENTS_ON_NOTRACE=n, you can continue putting probes on Xen spinlock functions too.

OK.

Quoted 5 lines
Quoted 4 lines

Do you see a nice and clean way to blacklist all these functions (something like arch_populate_kprobe_blacklist()), or should we just flag all of them explicitly with NOKPROBE_SYMBOL()?

Quoted 7 lines

As I pointed, you can probe it via your own kprobe module. Like systemtap, you still can probe it. The blacklist is for "kprobes", not for "kprobe_events". (Those are used to same, but since the above commit, those are different now)

I think the most sane solution is, identifying which (combination of) functions in ftrace (kernel/trace/*) causes a problem, marking those NOKPROBE_SYMBOL() and removing CONFIG_KPROBE_EVENTS_ON_NOTRACE.

OK. Thanks for the clarification!

-Andrea

#10 lore raw
Quoted 27 lines · Andrea Righi

On Tue, Dec 18, 2018 at 01:50:26PM +0900, Masami Hiramatsu wrote:
...

Quoted 25 lines
Quoted 10 lines

Side question: there are certain symbols in arch/x86/xen that should be blacklisted explicitly, because they're non-attachable.

More exactly, all functions defined in arch/x86/xen/spinlock.c, arch/x86/xen/time.c and arch/x86/xen/irq.c.

The reason is that these files are compiled without -pg to allow the usage of ftrace within a Xen domain apparently (from arch/x86/xen/Makefile):

Quoted 6 lines

ifdef CONFIG_FUNCTION_TRACER
# Do not profile debug and lowlevel utilities
CFLAGS_REMOVE_spinlock.o = -pg
CFLAGS_REMOVE_time.o = -pg
CFLAGS_REMOVE_irq.o = -pg
endif

Actually, the reason why you can not probe those functions via tracing/kprobe_events is just a side effect. You can probe it if you write a kprobe module. Since the kprobe_events depends on some ftrace tracing functions, it sometimes cause a recursive call problem. To avoid this issue, I have introduced a CONFIG_KPROBE_EVENTS_ON_NOTRACE, see commit 45408c4f9250 ("tracing: kprobes: Prohibit probing on notrace function").

Quoted 17 lines
Quoted 2 lines

If you set CONFIG_KPROBE_EVENTS_ON_NOTRACE=n, you can continue putting probes on Xen spinlock functions too.

OK.

Quoted 12 lines
Quoted 4 lines

Do you see a nice and clean way to blacklist all these functions (something like arch_populate_kprobe_blacklist()), or should we just flag all of them explicitly with NOKPROBE_SYMBOL()?

As I pointed, you can probe it via your own kprobe module. Like systemtap, you still can probe it. The blacklist is for "kprobes", not for "kprobe_events". (Those are used to same, but since the above commit, those are different now)

I think the most sane solution is, identifying which (combination of) functions in ftrace (kernel/trace/*) causes a problem, marking those NOKPROBE_SYMBOL() and removing CONFIG_KPROBE_EVENTS_ON_NOTRACE.

I'm planning to spend a little bit more time on this and see if I can identify the problematic ftrace functions and eventually drop CONFIG_KPROBE_EVENTS_ON_NOTRACE, following the sane solution.

However, in the meantime, with the following patch I've been able to get a more reliable kprobes blacklist and show also the notrace functions in debugfs when CONFIG_KPROBE_EVENTS_ON_NOTRACE is off.

It's probably ugly and inefficient, because it's iterating over all symbols in x86's arch_populate_kprobe_blacklist(), but it seems to work for my specific use case, so I thought it shouldn't be bad to share it, just in case (maybe someone else is also interested).

Thanks,

From: Andrea Righi <[email protected]>
Subject: [PATCH] x86: kprobes: automatically blacklist all non-traceable functions

Iterate over all symbols to detect those that are non-traceable and blacklist them.

---
 arch/x86/kernel/kprobes/core.c | 11 +++++++++--
 kernel/kprobes.c               | 22 ++++++++++++++++++++--
 2 files changed, 29 insertions(+), 4 deletions(-)
arch/x86/kernel/kprobes/core.c +9 -2
index 4ba75afba527..8cc7191ba3f9 100644
@@ -1026,10 +1026,17 @@ int kprobe_fault_handler(struct pt_regs *regs, int trapnr)
 }
 NOKPROBE_SYMBOL(kprobe_fault_handler);
 
+static int do_kprobes_arch_blacklist(void *data, const char *name,
+				     struct module *mod, unsigned long addr)
+{
+	if (arch_within_kprobe_blacklist(addr))
+		kprobe_add_ksym_blacklist(addr);
+	return 0;
+}
+
 int __init arch_populate_kprobe_blacklist(void)
 {
-	return kprobe_add_area_blacklist((unsigned long)__entry_text_start,
-					 (unsigned long)__entry_text_end);
+	return kallsyms_on_each_symbol(do_kprobes_arch_blacklist, NULL);
 }
 
 int __init arch_init_kprobes(void)
kernel/kprobes.c +20 -2
index f4ddfdd2d07e..2e824cd536ba 100644
@@ -1389,11 +1389,29 @@ static int register_aggr_kprobe(struct kprobe *orig_p, struct kprobe *p)
 	return ret;
 }
 
+#if defined(CONFIG_KPROBES_ON_FTRACE) && \
+	!defined(CONFIG_KPROBE_EVENTS_ON_NOTRACE)
+static bool within_notrace(unsigned long addr)
+{
+	unsigned long offset, size;
+
+	if (!kallsyms_lookup_size_offset(addr, &size, &offset))
+		return true;
+	return !ftrace_location_range(addr - offset, addr - offset + size);
+}
+#else
+static bool within_notrace(unsigned long addr)
+{
+	return false;
+}
+#endif
+
 bool __weak arch_within_kprobe_blacklist(unsigned long addr)
 {
 	/* The __kprobes marked functions and entry code must not be probed */
-	return addr >= (unsigned long)__kprobes_text_start &&
-	       addr < (unsigned long)__kprobes_text_end;
+	return (addr >= (unsigned long)__kprobes_text_start &&
+	       addr < (unsigned long)__kprobes_text_end) ||
+	       within_notrace(addr);
 }
 
 bool within_kprobe_blacklist(unsigned long addr)
Signature · 1 line
2.17.1
author maintainer
#11 lore raw

Hi Andrea,

Sorry for late reply,

Quoted 46 lines · On Thu, 27 Dec 2018 18:09:34 +0100 Andrea Righi <[email protected]> wrote:

On Tue, Dec 18, 2018 at 06:24:35PM +0100, Andrea Righi wrote:,

Quoted 27 lines · Masami Hiramatsu

...

Quoted 25 lines
Quoted 10 lines

Side question: there are certain symbols in arch/x86/xen that should be blacklisted explicitly, because they're non-attachable.

More exactly, all functions defined in arch/x86/xen/spinlock.c, arch/x86/xen/time.c and arch/x86/xen/irq.c.

The reason is that these files are compiled without -pg to allow the usage of ftrace within a Xen domain apparently (from arch/x86/xen/Makefile):

Quoted 6 lines

ifdef CONFIG_FUNCTION_TRACER
# Do not profile debug and lowlevel utilities
CFLAGS_REMOVE_spinlock.o = -pg
CFLAGS_REMOVE_time.o = -pg
CFLAGS_REMOVE_irq.o = -pg
endif

Actually, the reason why you can not probe those functions via tracing/kprobe_events is just a side effect. You can probe it if you write a kprobe module. Since the kprobe_events depends on some ftrace tracing functions, it sometimes cause a recursive call problem. To avoid this issue, I have introduced a CONFIG_KPROBE_EVENTS_ON_NOTRACE, see commit 45408c4f9250 ("tracing: kprobes: Prohibit probing on notrace function").

Quoted 17 lines
Quoted 2 lines

If you set CONFIG_KPROBE_EVENTS_ON_NOTRACE=n, you can continue putting probes on Xen spinlock functions too.

OK.

Quoted 12 lines
Quoted 4 lines

Do you see a nice and clean way to blacklist all these functions (something like arch_populate_kprobe_blacklist()), or should we just flag all of them explicitly with NOKPROBE_SYMBOL()?

As I pointed, you can probe it via your own kprobe module. Like systemtap, you still can probe it. The blacklist is for "kprobes", not for "kprobe_events". (Those are used to same, but since the above commit, those are different now)

I think the most sane solution is, identifying which (combination of) functions in ftrace (kernel/trace/*) causes a problem, marking those NOKPROBE_SYMBOL() and removing CONFIG_KPROBE_EVENTS_ON_NOTRACE.

Quoted 7 lines

I'm planning to spend a little bit more time on this and see if I can identify the problematic ftrace functions and eventually drop CONFIG_KPROBE_EVENTS_ON_NOTRACE, following the sane solution.

However, in the meantime, with the following patch I've been able to get a more reliable kprobes blacklist and show also the notrace functions in debugfs when CONFIG_KPROBE_EVENTS_ON_NOTRACE is off.

Hmm, if CONFIG_KPROBE_EVENTS_ON_NOTRACE=n, we already have a whitelist of functions in /sys/kernel/debug/tracing/available_filter_functions, so I don't think we need a blacklist.

Quoted 4 lines

It's probably ugly and inefficient, because it's iterating over all symbols in x86's arch_populate_kprobe_blacklist(), but it seems to work for my specific use case, so I thought it shouldn't be bad to share it, just in case (maybe someone else is also interested).

Hmm, but in that case, it limits other native kprobes users like systemtap to disable probing on notrace functions with no reasons. That may not be acceptable.

OK, I'll retry to find which notrace function combination tracing with kprobes are problematic. Let me do it...

Thank you,

Quoted 78 lines

Thanks,

From: Andrea Righi <[email protected]>
Subject: [PATCH] x86: kprobes: automatically blacklist all non-traceable functions

Iterate over all symbols to detect those that are non-traceable and blacklist them.

---
 arch/x86/kernel/kprobes/core.c | 11 +++++++++--
 kernel/kprobes.c               | 22 ++++++++++++++++++++--
 2 files changed, 29 insertions(+), 4 deletions(-)
arch/x86/kernel/kprobes/core.c +9 -2
index 4ba75afba527..8cc7191ba3f9 100644
@@ -1026,10 +1026,17 @@ int kprobe_fault_handler(struct pt_regs *regs, int trapnr)
 }
 NOKPROBE_SYMBOL(kprobe_fault_handler);
 
+static int do_kprobes_arch_blacklist(void *data, const char *name,
+				     struct module *mod, unsigned long addr)
+{
+	if (arch_within_kprobe_blacklist(addr))
+		kprobe_add_ksym_blacklist(addr);
+	return 0;
+}
+
 int __init arch_populate_kprobe_blacklist(void)
 {
-	return kprobe_add_area_blacklist((unsigned long)__entry_text_start,
-					 (unsigned long)__entry_text_end);
+	return kallsyms_on_each_symbol(do_kprobes_arch_blacklist, NULL);
 }
 
 int __init arch_init_kprobes(void)
kernel/kprobes.c +20 -2
index f4ddfdd2d07e..2e824cd536ba 100644
@@ -1389,11 +1389,29 @@ static int register_aggr_kprobe(struct kprobe *orig_p, struct kprobe *p)
 	return ret;
 }
 
+#if defined(CONFIG_KPROBES_ON_FTRACE) && \
+	!defined(CONFIG_KPROBE_EVENTS_ON_NOTRACE)
+static bool within_notrace(unsigned long addr)
+{
+	unsigned long offset, size;
+
+	if (!kallsyms_lookup_size_offset(addr, &size, &offset))
+		return true;
+	return !ftrace_location_range(addr - offset, addr - offset + size);
+}
+#else
+static bool within_notrace(unsigned long addr)
+{
+	return false;
+}
+#endif
+
 bool __weak arch_within_kprobe_blacklist(unsigned long addr)
 {
 	/* The __kprobes marked functions and entry code must not be probed */
-	return addr >= (unsigned long)__kprobes_text_start &&
-	       addr < (unsigned long)__kprobes_text_end;
+	return (addr >= (unsigned long)__kprobes_text_start &&
+	       addr < (unsigned long)__kprobes_text_end) ||
+	       within_notrace(addr);
 }
 
 bool within_kprobe_blacklist(unsigned long addr)
Signature · 1 line
2.17.1
Signature · 1 line
Masami Hiramatsu <[email protected]>
#12 lore raw

On Tue, Jan 01, 2019 at 10:16:54PM +0900, Masami Hiramatsu wrote:
...

Quoted 16 lines
Quoted 16 lines
Quoted 11 lines
Quoted 11 lines
Quoted 3 lines

Do you see a nice and clean way to blacklist all these functions (something like arch_populate_kprobe_blacklist()), or should we just flag all of them explicitly with NOKPROBE_SYMBOL()?

As I pointed, you can probe it via your own kprobe module. Like systemtap, you still can probe it. The blacklist is for "kprobes", not for "kprobe_events". (Those are used to same, but since the above commit, those are different now)

I think the most sane solution is, identifying which (combination of) functions in ftrace (kernel/trace/*) causes a problem, marking those NOKPROBE_SYMBOL() and removing CONFIG_KPROBE_EVENTS_ON_NOTRACE.

I'm planning to spend a little bit more time on this and see if I can identify the problematic ftrace functions and eventually drop CONFIG_KPROBE_EVENTS_ON_NOTRACE, following the sane solution.

Quoted 7 lines
Quoted 3 lines

However, in the meantime, with the following patch I've been able to get a more reliable kprobes blacklist and show also the notrace functions in debugfs when CONFIG_KPROBE_EVENTS_ON_NOTRACE is off.

Hmm, if CONFIG_KPROBE_EVENTS_ON_NOTRACE=n, we already have a whitelist of functions in /sys/kernel/debug/tracing/available_filter_functions, so I don't think we need a blacklist.

OK.

Quoted 8 lines
Quoted 5 lines

It's probably ugly and inefficient, because it's iterating over all symbols in x86's arch_populate_kprobe_blacklist(), but it seems to work for my specific use case, so I thought it shouldn't be bad to share it, just in case (maybe someone else is also interested).

Hmm, but in that case, it limits other native kprobes users like systemtap to disable probing on notrace functions with no reasons. That may not be acceptable.

True...

Quoted 3 lines

OK, I'll retry to find which notrace function combination tracing with kprobes are problematic. Let me do it...

OK. Thanks tons for looking into this!

-Andrea