Skip to contentSkip to conversation

ftrace: fix ftrace misleading comments for arch using it

PATCH In review Li Bin 4 participants 11 messages Dec 6, 2015 – Dec 29, 2015

Subsystems (4)
lore
Share & tools
lore

Li Bin

author
#0 lore raw

v2 cover ftrace: fix ftrace misleading comments for arch using it

Fix the following similar misleading comments of ftrace for arch ia64/metag/powerpc/sh/x86:

Note: Due to modules and __init, code can
disappear and change, we need to protect against faulting as well as code changing. We do this by using the probe_kernel_* functions.

No real locking needed, this code is run through
kstop_machine, or before SMP starts.

Cc: Tony Luck <[email protected]>
Cc: Fenghua Yu <[email protected]>
Cc: [email protected]
Cc: James Hogan <[email protected]>
Cc: [email protected]
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Michael Ellerman <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: Thomas Gleixner <[email protected]>
"Cc: H. Peter Anvin" <[email protected]>
Cc: [email protected]

Li Bin (5):
ia64: ftrace: fix the comments for ftrace_modify_code
metag: ftrace: fix the comments for ftrace_modify_code
powerpc: ftrace: fix the comments for ftrace_modify_code
sh: ftrace: fix the comments for ftrace_modify_code
x86: ftrace: fix the comments for ftrace_modify_code_direct

 arch/ia64/kernel/ftrace.c    |   12 +++++-------
 arch/metag/kernel/ftrace.c   |   11 +++++------
 arch/powerpc/kernel/ftrace.c |   12 +++++-------
 arch/sh/kernel/ftrace.c      |   12 +++++-------
 arch/x86/kernel/ftrace.c     |   12 +++++-------
 5 files changed, 25 insertions(+), 34 deletions(-)
author
#1 lore raw

PATCH v2 1/5 ia64: ftrace: fix the comments for ftrace_modify_code

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

---
 arch/ia64/kernel/ftrace.c |   12 +++++-------
 1 files changed, 5 insertions(+), 7 deletions(-)
arch/ia64/kernel/ftrace.c +5 -7
index 3b0c2aa..cee411e 100644
@@ -97,13 +97,11 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code,
 	unsigned char replaced[MCOUNT_INSN_SIZE];
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing. We do this by using the
-	 *  probe_kernel_* functions.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine, or before SMP starts.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */
 
 	if (!do_check)
Signature · 1 line
1.7.1
author
#2 lore raw

PATCH v2 2/5 metag: ftrace: fix the comments for ftrace_modify_code

In reply to Li Bin

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

---
 arch/metag/kernel/ftrace.c |   11 +++++------
 1 files changed, 5 insertions(+), 6 deletions(-)
arch/metag/kernel/ftrace.c +5 -6
index ed1d685..ac8c039 100644
@@ -54,12 +54,11 @@ static int ftrace_modify_code(unsigned long pc, unsigned char *old_code,
 	unsigned char replaced[MCOUNT_INSN_SIZE];
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */
 
 	/* read the text we want to modify */
Signature · 1 line
1.7.1
maintainer
#3 lore raw

Re: [PATCH v2 2/5] metag: ftrace: fix the comments for ftrace_modify_code

Quoted 11 lines · On Sun, 6 Dec 2015 10:02:55 +0800 Li Bin <[email protected]> wrote:

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

Quoted 4 lines

Hi, can I get a metag maintainer to ack this. Or you can take the patch too.

-- Steve

Quoted 26 lines
---
 arch/metag/kernel/ftrace.c |   11 +++++------
 1 files changed, 5 insertions(+), 6 deletions(-)
arch/metag/kernel/ftrace.c +5 -6
index ed1d685..ac8c039 100644
@@ -54,12 +54,11 @@ static int ftrace_modify_code(unsigned long pc, unsigned char *old_code,
 	unsigned char replaced[MCOUNT_INSN_SIZE];
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */
 
 	/* read the text we want to modify */
Acked-by
#4 lore raw

Re: [PATCH v2 2/5] metag: ftrace: fix the comments for ftrace_modify_code

Hi Steven,

Quoted 14 lines · Steven Rostedt
Quoted 11 lines · On Sun, 6 Dec 2015 10:02:55 +0800 Li Bin <[email protected]> wrote:

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

Quoted 7 lines
Quoted 4 lines

Hi, can I get a metag maintainer to ack this. Or you can take the patch too.

Yep, sorry for the delay,

Cheers
James

Quoted 30 lines

-- Steve

Quoted 26 lines
---
 arch/metag/kernel/ftrace.c |   11 +++++------
 1 files changed, 5 insertions(+), 6 deletions(-)
arch/metag/kernel/ftrace.c +5 -6
index ed1d685..ac8c039 100644
@@ -54,12 +54,11 @@ static int ftrace_modify_code(unsigned long pc, unsigned char *old_code,
 	unsigned char replaced[MCOUNT_INSN_SIZE];
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */
 
 	/* read the text we want to modify */
author
#5 lore raw

PATCH v2 3/5 powerpc: ftrace: fix the comments for ftrace_modify_code

In reply to Li Bin

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Michael Ellerman <[email protected]>
---
 arch/powerpc/kernel/ftrace.c |   12 +++++-------
 1 files changed, 5 insertions(+), 7 deletions(-)
arch/powerpc/kernel/ftrace.c +5 -7
index 44d4d8e..83e3c88 100644
@@ -47,13 +47,11 @@ ftrace_modify_code(unsigned long ip, unsigned int old, unsigned int new)
 	unsigned int replaced;
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing. We do this by using the
-	 *  probe_kernel_* functions.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine, or before SMP starts.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */
 
 	/* read the text we want to modify */
Signature · 1 line
1.7.1
maintainer
#6 lore raw

Re: [PATCH v2 3/5] powerpc: ftrace: fix the comments for ftrace_modify_code

Quoted 11 lines · On Sun, 6 Dec 2015 10:02:56 +0800 Li Bin <[email protected]> wrote:

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

Quoted 6 lines
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Michael Ellerman <[email protected]>

Hi, can I get a POWERPC maintainer to ack this, or you can take it too.

-- Steve

Quoted 27 lines
---
 arch/powerpc/kernel/ftrace.c |   12 +++++-------
 1 files changed, 5 insertions(+), 7 deletions(-)
arch/powerpc/kernel/ftrace.c +5 -7
index 44d4d8e..83e3c88 100644
@@ -47,13 +47,11 @@ ftrace_modify_code(unsigned long ip, unsigned int old, unsigned int new)
 	unsigned int replaced;
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing. We do this by using the
-	 *  probe_kernel_* functions.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine, or before SMP starts.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */
 
 	/* read the text we want to modify */
author
#7 lore raw

PATCH v2 4/5 sh: ftrace: fix the comments for ftrace_modify_code

In reply to Li Bin

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

---
 arch/sh/kernel/ftrace.c |   12 +++++-------
 1 files changed, 5 insertions(+), 7 deletions(-)
arch/sh/kernel/ftrace.c +5 -7
index 079d70e..38993e0 100644
@@ -212,13 +212,11 @@ static int ftrace_modify_code(unsigned long ip, unsigned char *old_code,
 	unsigned char replaced[MCOUNT_INSN_SIZE];
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing. We do this by using the
-	 *  probe_kernel_* functions.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine, or before SMP starts.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */
 
 	/* read the text we want to modify */
Signature · 1 line
1.7.1
author
#8 lore raw

PATCH v2 5/5 x86: ftrace: fix the comments for ftrace_modify_code_direct

In reply to Li Bin

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

---
 arch/x86/kernel/ftrace.c |   12 +++++-------
 1 files changed, 5 insertions(+), 7 deletions(-)
arch/x86/kernel/ftrace.c +5 -7
index 311bcf3..db0f6b1 100644
@@ -106,13 +106,11 @@ ftrace_modify_code_direct(unsigned long ip, unsigned const char *old_code,
 	unsigned char replaced[MCOUNT_INSN_SIZE];
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing. We do this by using the
-	 *  probe_kernel_* functions.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine, or before SMP starts.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */
 
 	/* read the text we want to modify */
Signature · 1 line
1.7.1
maintainer
#9 lore raw

Re: [PATCH v2 5/5] x86: ftrace: fix the comments for ftrace_modify_code_direct

Quoted 31 lines · On Sun, 6 Dec 2015 10:02:58 +0800 Li Bin <[email protected]> wrote:

There is no need to worry about module and __init text disappearing case, because that ftrace has a module notifier that is called when a module is being unloaded and before the text goes away and this code grabs the ftrace_lock mutex and removes the module functions from the ftrace list, such that it will no longer do any modifications to that module's text, the update to make functions be traced or not is done under the ftrace_lock mutex as well. And by now, __init section codes should not been modified by ftrace, because it is black listed in recordmcount.c and ignored by ftrace.

---
 arch/x86/kernel/ftrace.c |   12 +++++-------
 1 files changed, 5 insertions(+), 7 deletions(-)
arch/x86/kernel/ftrace.c +0 -3
index 311bcf3..db0f6b1 100644
@@ -106,13 +106,11 @@ ftrace_modify_code_direct(unsigned long ip, unsigned const char *old_code,
 	unsigned char replaced[MCOUNT_INSN_SIZE];
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing. We do this by using the

arch/x86/kernel/ftrace.c ftrace_modify_code_direct(unsigned long ip, unsigned const char *old_code, · line 114

Quoted 10 lines
arch/x86/kernel/ftrace.c +5 -4
-	 *  probe_kernel_* functions.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine, or before SMP starts.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */

Thomas,

This has been now fixed. You want to ack it and I take it, or can you take it. Or is there something else that should be fixed?

-- Steve

Quoted 2 lines
arch/x86/kernel/ftrace.c +0 -0
 	/* read the text we want to modify */
Reviewed-by
#10 lore raw

Re: [PATCH v2 5/5] x86: ftrace: fix the comments for ftrace_modify_code_direct

Quoted 26 lines · Steven Rostedt
Quoted 11 lines · On Sun, 6 Dec 2015 10:02:58 +0800 Li Bin <[email protected]> wrote:
arch/x86/kernel/ftrace.c +0 -3
index 311bcf3..db0f6b1 100644
@@ -106,13 +106,11 @@ ftrace_modify_code_direct(unsigned long ip, unsigned const char *old_code,
 	unsigned char replaced[MCOUNT_INSN_SIZE];
 
 	/*
-	 * Note: Due to modules and __init, code can
-	 *  disappear and change, we need to protect against faulting
-	 *  as well as code changing. We do this by using the

arch/x86/kernel/ftrace.c ftrace_modify_code_direct(unsigned long ip, unsigned const char *old_code, · line 114

Quoted 10 lines
arch/x86/kernel/ftrace.c +5 -4
-	 *  probe_kernel_* functions.
-	 *
-	 * No real locking needed, this code is run through
-	 * kstop_machine, or before SMP starts.
+	 * Note:
+	 * We are paranoid about modifying text, as if a bug was to happen, it
+	 * could cause us to read or write to someplace that could cause harm.
+	 * Carefully read and modify the code with probe_kernel_*(), and make
+	 * sure what we read is what we expected it to be before modifying it.
 	 */

Thomas,

Quoted 2 lines

This has been now fixed. You want to ack it and I take it, or can you take it. Or is there something else that should be fixed?