[PATCH 0/7] powerpc/ftrace: Patch out -mprofile-kernel instructions

Changes since RFC:
- Patches 1 and 2: No changes
- Patch 3: rename function to ftrace_replace_code_rec() to signify that 
  it acts on a ftrace record.
- Patch 4: numerous small changes, including those suggested by Nick 
  Piggin.
- Patch 4: additionally handle scenario where __ftrace_make_call() can 
  be invoked without having called __ftrace_make_call_prep(), when a 
  kernel module is loaded while function tracing is active.
- Patch 5 to 7: new, to have ftrace claim ownership of two instructions, 
  and to have kprobes work properly with this.


--
On powerpc64, -mprofile-kernel results in two instructions being 
emitted: 'mflr r0' and 'bl _mcount'. So far, we were only nop'ing out 
the branch to _mcount(). This series implements an approach to also nop 
out the preceding mflr.


- Naveen


Naveen N. Rao (7):
  ftrace: Expose flags used for ftrace_replace_code()
  x86/ftrace: Fix use of flags in ftrace_replace_code()
  ftrace: Expose __ftrace_replace_code()
  powerpc/ftrace: Additionally nop out the preceding mflr with
    -mprofile-kernel
  powerpc/ftrace: Update ftrace_location() for powerpc -mprofile-kernel
  kprobes/ftrace: Use ftrace_location() when [dis]arming probes
  powerpc/kprobes: Allow probing on any ftrace address

 arch/powerpc/kernel/kprobes-ftrace.c |  30 +++
 arch/powerpc/kernel/trace/ftrace.c   | 272 ++++++++++++++++++++++++---
 arch/x86/kernel/ftrace.c             |   3 +-
 include/linux/ftrace.h               |   7 +
 kernel/kprobes.c                     |  10 +-
 kernel/trace/ftrace.c                |  17 +-
 6 files changed, 300 insertions(+), 39 deletions(-)

-- 
2.22.0

Since ftrace_replace_code() is a __weak function and can be overridden,
we need to expose the flags that can be set. So, move the flags enum to
the header file.

Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Naveen N. Rao <redacted>
---
 include/linux/ftrace.h | 5 +++++
 kernel/trace/ftrace.c  | 5 -----
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index 25e2995d4a4c..e97789c95c4e 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h

@@ -162,6 +162,11 @@ enum {
 	FTRACE_OPS_FL_TRACE_ARRAY		= 1 << 15,
 };
 
+enum {
+	FTRACE_MODIFY_ENABLE_FL		= (1 << 0),
+	FTRACE_MODIFY_MAY_SLEEP_FL	= (1 << 1),
+};
+
 #ifdef CONFIG_DYNAMIC_FTRACE
 /* The hash used to know what functions callbacks trace */
 struct ftrace_ops_hash {

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 38277af44f5c..5710a6b3edc1 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c

@@ -75,11 +75,6 @@
 #define INIT_OPS_HASH(opsname)
 #endif
 
-enum {
-	FTRACE_MODIFY_ENABLE_FL		= (1 << 0),
-	FTRACE_MODIFY_MAY_SLEEP_FL	= (1 << 1),
-};
-
 struct ftrace_ops ftrace_list_end __read_mostly = {
 	.func		= ftrace_stub,
 	.flags		= FTRACE_OPS_FL_RECURSION_SAFE | FTRACE_OPS_FL_STUB,

-- 
2.22.0

While over-riding ftrace_replace_code(), we still want to reuse the
existing __ftrace_replace_code() function. Rename the function and
make it available for other kernel code.

Signed-off-by: Naveen N. Rao <redacted>
---
 include/linux/ftrace.h | 1 +
 kernel/trace/ftrace.c  | 8 ++++----
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index e97789c95c4e..fa653a561da5 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h

@@ -456,6 +456,7 @@ ftrace_set_early_filter(struct ftrace_ops *ops, char *buf, int enable);
 /* defined in arch */
 extern int ftrace_ip_converted(unsigned long ip);
 extern int ftrace_dyn_arch_init(void);
+extern int ftrace_replace_code_rec(struct dyn_ftrace *rec, int enable);
 extern void ftrace_replace_code(int enable);
 extern int ftrace_update_ftrace_func(ftrace_func_t func);
 extern void ftrace_caller(void);

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 5710a6b3edc1..21d8e201ee80 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c

@@ -2351,8 +2351,8 @@ unsigned long ftrace_get_addr_curr(struct dyn_ftrace *rec)
 		return (unsigned long)FTRACE_ADDR;
 }
 
-static int
-__ftrace_replace_code(struct dyn_ftrace *rec, int enable)
+int
+ftrace_replace_code_rec(struct dyn_ftrace *rec, int enable)
 {
 	unsigned long ftrace_old_addr;
 	unsigned long ftrace_addr;

@@ -2403,7 +2403,7 @@ void __weak ftrace_replace_code(int mod_flags)
 		if (rec->flags & FTRACE_FL_DISABLED)
 			continue;
 
-		failed = __ftrace_replace_code(rec, enable);
+		failed = ftrace_replace_code_rec(rec, enable);
 		if (failed) {
 			ftrace_bug(failed, rec);
 			/* Stop processing */

@@ -5827,7 +5827,7 @@ void ftrace_module_enable(struct module *mod)
 		rec->flags = cnt;
 
 		if (ftrace_start_up && cnt) {
-			int failed = __ftrace_replace_code(rec, 1);
+			int failed = ftrace_replace_code_rec(rec, 1);
 			if (failed) {
 				ftrace_bug(failed, rec);
 				goto out_loop;

-- 
2.22.0

In commit a0572f687fb3c ("ftrace: Allow ftrace_replace_code() to be
schedulable), the generic ftrace_replace_code() function was modified to
accept a flags argument in place of a single 'enable' flag. However, the
x86 version of this function was not updated. Fix the same.

Fixes: a0572f687fb3c ("ftrace: Allow ftrace_replace_code() to be schedulable")
Signed-off-by: Naveen N. Rao <redacted>
---
 arch/x86/kernel/ftrace.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c
index 0927bb158ffc..f34005a17051 100644
--- a/arch/x86/kernel/ftrace.c
+++ b/arch/x86/kernel/ftrace.c

@@ -573,8 +573,9 @@ static void run_sync(void)
 		local_irq_disable();
 }
 
-void ftrace_replace_code(int enable)
+void ftrace_replace_code(int mod_flags)
 {
+	int enable = mod_flags & FTRACE_MODIFY_ENABLE_FL;
 	struct ftrace_rec_iter *iter;
 	struct dyn_ftrace *rec;
 	const char *report = "adding breakpoints";

-- 
2.22.0

With -mprofile-kernel, gcc emits 'mflr r0', followed by 'bl _mcount' to
enable function tracing and profiling. So far, with dynamic ftrace, we
used to only patch out the branch to _mcount(). However, mflr is
executed by the branch unit that can only execute one per cycle on
POWER9 and shared with branches, so it would be nice to avoid it where
possible.

We cannot simply nop out the mflr either. When enabling function
tracing, there can be a race if tracing is enabled when some thread was
interrupted after executing a nop'ed out mflr. In this case, the thread
would execute the now-patched-in branch to _mcount() without having
executed the preceding mflr.

To solve this, we now enable function tracing in 2 steps: patch in the
mflr instruction, use synchronize_rcu_tasks() to ensure all existing
threads make progress, and then patch in the branch to _mcount(). We
override ftrace_replace_code() with a powerpc64 variant for this
purpose.

Suggested-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Naveen N. Rao <redacted>
---
 arch/powerpc/kernel/trace/ftrace.c | 241 ++++++++++++++++++++++++++---
 1 file changed, 219 insertions(+), 22 deletions(-)

diff --git a/arch/powerpc/kernel/trace/ftrace.c b/arch/powerpc/kernel/trace/ftrace.c
index 517662a56bdc..5e2b29808af1 100644
--- a/arch/powerpc/kernel/trace/ftrace.c
+++ b/arch/powerpc/kernel/trace/ftrace.c

@@ -125,7 +125,7 @@ __ftrace_make_nop(struct module *mod,
 {
 	unsigned long entry, ptr, tramp;
 	unsigned long ip = rec->ip;
-	unsigned int op, pop;
+	unsigned int op;
 
 	/* read where this goes */
 	if (probe_kernel_read(&op, (void *)ip, sizeof(int))) {

@@ -160,8 +160,6 @@ __ftrace_make_nop(struct module *mod,
 
 #ifdef CONFIG_MPROFILE_KERNEL
 	/* When using -mkernel_profile there is no load to jump over */
-	pop = PPC_INST_NOP;
-
 	if (probe_kernel_read(&op, (void *)(ip - 4), 4)) {
 		pr_err("Fetching instruction at %lx failed.\n", ip - 4);
 		return -EFAULT;

@@ -169,26 +167,23 @@ __ftrace_make_nop(struct module *mod,
 
 	/* We expect either a mflr r0, or a std r0, LRSAVE(r1) */
 	if (op != PPC_INST_MFLR && op != PPC_INST_STD_LR) {
-		pr_err("Unexpected instruction %08x around bl _mcount\n", op);
+		pr_err("Unexpected instruction %08x before bl _mcount\n", op);
 		return -EINVAL;
 	}
-#else
-	/*
-	 * Our original call site looks like:
-	 *
-	 * bl <tramp>
-	 * ld r2,XX(r1)
-	 *
-	 * Milton Miller pointed out that we can not simply nop the branch.
-	 * If a task was preempted when calling a trace function, the nops
-	 * will remove the way to restore the TOC in r2 and the r2 TOC will
-	 * get corrupted.
-	 *
-	 * Use a b +8 to jump over the load.
-	 */
 
-	pop = PPC_INST_BRANCH | 8;	/* b +8 */
+	/* We should patch out the bl to _mcount first */
+	if (patch_instruction((unsigned int *)ip, PPC_INST_NOP)) {
+		pr_err("Patching NOP failed.\n");
+		return -EPERM;
+	}
 
+	/* then, nop out the preceding 'mflr r0' as an optimization */
+	if (op == PPC_INST_MFLR &&
+		patch_instruction((unsigned int *)(ip - 4), PPC_INST_NOP)) {
+		pr_err("Patching NOP failed.\n");
+		return -EPERM;
+	}
+#else
 	/*
 	 * Check what is in the next instruction. We can see ld r2,40(r1), but
 	 * on first pass after boot we will see mflr r0.

@@ -202,12 +197,25 @@ __ftrace_make_nop(struct module *mod,
 		pr_err("Expected %08x found %08x\n", PPC_INST_LD_TOC, op);
 		return -EINVAL;
 	}
-#endif /* CONFIG_MPROFILE_KERNEL */
 
-	if (patch_instruction((unsigned int *)ip, pop)) {
+	/*
+	 * Our original call site looks like:
+	 *
+	 * bl <tramp>
+	 * ld r2,XX(r1)
+	 *
+	 * Milton Miller pointed out that we can not simply nop the branch.
+	 * If a task was preempted when calling a trace function, the nops
+	 * will remove the way to restore the TOC in r2 and the r2 TOC will
+	 * get corrupted.
+	 *
+	 * Use a b +8 to jump over the load.
+	 */
+	if (patch_instruction((unsigned int *)ip, PPC_INST_BRANCH | 8)) {
 		pr_err("Patching NOP failed.\n");
 		return -EPERM;
 	}
+#endif /* CONFIG_MPROFILE_KERNEL */
 
 	return 0;
 }

@@ -421,6 +429,26 @@ static int __ftrace_make_nop_kernel(struct dyn_ftrace *rec, unsigned long addr)
 		return -EPERM;
 	}
 
+#ifdef CONFIG_MPROFILE_KERNEL
+	/* Nop out the preceding 'mflr r0' as an optimization */
+	if (probe_kernel_read(&op, (void *)(ip - 4), 4)) {
+		pr_err("Fetching instruction at %lx failed.\n", ip - 4);
+		return -EFAULT;
+	}
+
+	/* We expect either a mflr r0, or a std r0, LRSAVE(r1) */
+	if (op != PPC_INST_MFLR && op != PPC_INST_STD_LR) {
+		pr_err("Unexpected instruction %08x before bl _mcount\n", op);
+		return -EINVAL;
+	}
+
+	if (op == PPC_INST_MFLR &&
+		patch_instruction((unsigned int *)(ip - 4), PPC_INST_NOP)) {
+		pr_err("Patching NOP failed.\n");
+		return -EPERM;
+	}
+#endif
+
 	return 0;
 }
 

@@ -429,6 +457,7 @@ int ftrace_make_nop(struct module *mod,
 {
 	unsigned long ip = rec->ip;
 	unsigned int old, new;
+	int rc;
 
 	/*
 	 * If the calling address is more that 24 bits away,

@@ -439,7 +468,34 @@ int ftrace_make_nop(struct module *mod,
 		/* within range */
 		old = ftrace_call_replace(ip, addr, 1);
 		new = PPC_INST_NOP;
-		return ftrace_modify_code(ip, old, new);
+		rc = ftrace_modify_code(ip, old, new);
+#ifdef CONFIG_MPROFILE_KERNEL
+		if (rc)
+			return rc;
+
+		/*
+		 * For -mprofile-kernel, we patch out the preceding 'mflr r0'
+		 * instruction, as an optimization. It is important to nop out
+		 * the branch to _mcount() first, as a lone 'mflr r0' is
+		 * harmless.
+		 */
+		if (probe_kernel_read(&old, (void *)(ip - 4), 4)) {
+			pr_err("Fetching instruction at %lx failed.\n", ip - 4);
+			return -EFAULT;
+		}
+
+		/* We expect either a mflr r0, or a std r0, LRSAVE(r1) */
+		if (old != PPC_INST_MFLR && old != PPC_INST_STD_LR) {
+			pr_err("Unexpected instruction %08x before bl _mcount\n",
+					old);
+			return -EINVAL;
+		}
+
+		if (old == PPC_INST_MFLR)
+			rc = patch_instruction((unsigned int *)(ip - 4),
+					PPC_INST_NOP);
+#endif
+		return rc;
 	} else if (core_kernel_text(ip))
 		return __ftrace_make_nop_kernel(rec, addr);
 

@@ -567,6 +623,37 @@ __ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
 		return -EINVAL;
 	}
 
+#ifdef CONFIG_MPROFILE_KERNEL
+	/*
+	 * We could end up here without having called __ftrace_make_call_prep()
+	 * if function tracing is enabled before a module is loaded.
+	 *
+	 * ftrace_module_enable() --> ftrace_replace_code_rec() -->
+	 *	ftrace_make_call() --> __ftrace_make_call()
+	 *
+	 * In this scenario, the previous instruction will be a NOP. It is
+	 * safe to patch it with a 'mflr r0' since we know for a fact that
+	 * this code is not yet being run.
+	 */
+	ip -= MCOUNT_INSN_SIZE;
+
+	/* read where this goes */
+	if (probe_kernel_read(op, ip, MCOUNT_INSN_SIZE))
+		return -EFAULT;
+
+	/*
+	 * nothing to do if this is using the older -mprofile-kernel
+	 * instruction sequence
+	 */
+	if (op[0] != PPC_INST_NOP)
+		return 0;
+
+	if (patch_instruction((unsigned int *)ip, PPC_INST_MFLR)) {
+		pr_err("Patching MFLR failed.\n");
+		return -EPERM;
+	}
+#endif
+
 	return 0;
 }
 

@@ -863,6 +950,116 @@ void arch_ftrace_update_code(int command)
 	ftrace_modify_all_code(command);
 }
 
+#ifdef CONFIG_MPROFILE_KERNEL
+/* Returns 1 if we patched in the mflr */
+static int __ftrace_make_call_prep(struct dyn_ftrace *rec)
+{
+	void *ip = (void *)rec->ip - MCOUNT_INSN_SIZE;
+	unsigned int op[2];
+
+	/* read where this goes */
+	if (probe_kernel_read(op, ip, sizeof(op)))
+		return -EFAULT;
+
+	if (op[1] != PPC_INST_NOP) {
+		pr_err("Unexpected call sequence at %p: %x %x\n",
+							ip, op[0], op[1]);
+		return -EINVAL;
+	}
+
+	/*
+	 * nothing to do if this is using the older -mprofile-kernel
+	 * instruction sequence
+	 */
+	if (op[0] != PPC_INST_NOP)
+		return 0;
+
+	if (patch_instruction((unsigned int *)ip, PPC_INST_MFLR)) {
+		pr_err("Patching MFLR failed.\n");
+		return -EPERM;
+	}
+
+	return 1;
+}
+
+/*
+ * When enabling function tracing for -mprofile-kernel that uses a
+ * 2-instruction sequence of 'mflr r0, bl _mcount()', we use a 2 step process:
+ * 1. Patch in the 'mflr r0' instruction
+ * 1a. synchronize_rcu_tasks() to ensure that any threads that had executed
+ *     the earlier nop there make progress (and hence do not branch into
+ *     ftrace without executing the preceding mflr)
+ * 2. Patch in the branch to ftrace
+ */
+void ftrace_replace_code(int mod_flags)
+{
+	int enable = mod_flags & FTRACE_MODIFY_ENABLE_FL;
+	int schedulable = mod_flags & FTRACE_MODIFY_MAY_SLEEP_FL;
+	int ret, failed, make_call = 0;
+	struct ftrace_rec_iter *iter;
+	struct dyn_ftrace *rec;
+
+	if (unlikely(!ftrace_enabled))
+		return;
+
+	for_ftrace_rec_iter(iter) {
+		rec = ftrace_rec_iter_record(iter);
+
+		if (rec->flags & FTRACE_FL_DISABLED)
+			continue;
+
+		failed = 0;
+		ret = ftrace_test_record(rec, enable);
+		if (ret == FTRACE_UPDATE_MAKE_CALL) {
+			failed = __ftrace_make_call_prep(rec);
+			if (failed < 0) {
+				ftrace_bug(failed, rec);
+				return;
+			} else if (failed == 1)
+				make_call++;
+		}
+
+		if (!failed) {
+			/* We can patch the record directly */
+			failed = ftrace_replace_code_rec(rec, enable);
+			if (failed) {
+				ftrace_bug(failed, rec);
+				return;
+			}
+		}
+
+		if (schedulable)
+			cond_resched();
+	}
+
+	if (!make_call)
+		/* No records needed patching a preceding mflr */
+		return;
+
+	synchronize_rcu_tasks();
+
+	for_ftrace_rec_iter(iter) {
+		rec = ftrace_rec_iter_record(iter);
+
+		if (rec->flags & FTRACE_FL_DISABLED)
+			continue;
+
+		ret = ftrace_test_record(rec, enable);
+		if (ret == FTRACE_UPDATE_MAKE_CALL)
+			failed = ftrace_replace_code_rec(rec, enable);
+
+		if (failed) {
+			ftrace_bug(failed, rec);
+			return;
+		}
+
+		if (schedulable)
+			cond_resched();
+	}
+
+}
+#endif
+
 #ifdef CONFIG_PPC64
 #define PACATOC offsetof(struct paca_struct, kernel_toc)
 

-- 
2.22.0

Hi Naveen,

Sorry I meant to reply to this earlier .. :/

"Naveen N. Rao" [off-list ref] writes:

With -mprofile-kernel, gcc emits 'mflr r0', followed by 'bl _mcount' to
enable function tracing and profiling. So far, with dynamic ftrace, we
used to only patch out the branch to _mcount(). However, mflr is
executed by the branch unit that can only execute one per cycle on
POWER9 and shared with branches, so it would be nice to avoid it where
possible.

We cannot simply nop out the mflr either. When enabling function
tracing, there can be a race if tracing is enabled when some thread was
interrupted after executing a nop'ed out mflr. In this case, the thread
would execute the now-patched-in branch to _mcount() without having
executed the preceding mflr.

To solve this, we now enable function tracing in 2 steps: patch in the
mflr instruction, use synchronize_rcu_tasks() to ensure all existing
threads make progress, and then patch in the branch to _mcount(). We
override ftrace_replace_code() with a powerpc64 variant for this
purpose.

According to the ISA we're not allowed to patch mflr at runtime. See the
section on "CMODX".

I'm also not convinced the ordering between the two patches is
guaranteed by the ISA, given that there's possibly no isync on the other
CPU.

But I haven't had time to dig into it sorry, hopefully later in the
week?

cheers

Michael Ellerman's on June 19, 2019 3:14 pm:

Hi Naveen,

Sorry I meant to reply to this earlier .. :/

"Naveen N. Rao" [off-list ref] writes:

quoted

With -mprofile-kernel, gcc emits 'mflr r0', followed by 'bl _mcount' to
enable function tracing and profiling. So far, with dynamic ftrace, we
used to only patch out the branch to _mcount(). However, mflr is
executed by the branch unit that can only execute one per cycle on
POWER9 and shared with branches, so it would be nice to avoid it where
possible.

We cannot simply nop out the mflr either. When enabling function
tracing, there can be a race if tracing is enabled when some thread was
interrupted after executing a nop'ed out mflr. In this case, the thread
would execute the now-patched-in branch to _mcount() without having
executed the preceding mflr.

To solve this, we now enable function tracing in 2 steps: patch in the
mflr instruction, use synchronize_rcu_tasks() to ensure all existing
threads make progress, and then patch in the branch to _mcount(). We
override ftrace_replace_code() with a powerpc64 variant for this
purpose.

According to the ISA we're not allowed to patch mflr at runtime. See the
section on "CMODX".

According to "quasi patch class" engineering note, we can patch
anything with a preferred nop. But that's written as an optional
facility, which we don't have a feature to test for.

I'm also not convinced the ordering between the two patches is
guaranteed by the ISA, given that there's possibly no isync on the other
CPU.

Will they go through a context synchronizing event?

synchronize_rcu_tasks() should ensure a thread is scheduled away, but
I'm not actually sure it guarantees CSI if it's kernel->kernel. Could
do a smp_call_function to do the isync on each CPU to be sure.

Thanks,
Nick

Nicholas Piggin wrote:

Michael Ellerman's on June 19, 2019 3:14 pm:

quoted

Hi Naveen,

Sorry I meant to reply to this earlier .. :/

No problem. Thanks for the questions.

quoted

"Naveen N. Rao" [off-list ref] writes:

quoted

With -mprofile-kernel, gcc emits 'mflr r0', followed by 'bl _mcount' to
enable function tracing and profiling. So far, with dynamic ftrace, we
used to only patch out the branch to _mcount(). However, mflr is
executed by the branch unit that can only execute one per cycle on
POWER9 and shared with branches, so it would be nice to avoid it where
possible.

We cannot simply nop out the mflr either. When enabling function
tracing, there can be a race if tracing is enabled when some thread was
interrupted after executing a nop'ed out mflr. In this case, the thread
would execute the now-patched-in branch to _mcount() without having
executed the preceding mflr.

To solve this, we now enable function tracing in 2 steps: patch in the
mflr instruction, use synchronize_rcu_tasks() to ensure all existing
threads make progress, and then patch in the branch to _mcount(). We
override ftrace_replace_code() with a powerpc64 variant for this
purpose.

According to the ISA we're not allowed to patch mflr at runtime. See the
section on "CMODX".

According to "quasi patch class" engineering note, we can patch
anything with a preferred nop. But that's written as an optional
facility, which we don't have a feature to test for.

Hmm... I wonder what the implications are. We've been patching in a 
'trap' for kprobes for a long time now, along with having to patch back 
the original instruction (which can be anything), when the probe is 
removed.

quoted

I'm also not convinced the ordering between the two patches is
guaranteed by the ISA, given that there's possibly no isync on the other
CPU.

Will they go through a context synchronizing event?

synchronize_rcu_tasks() should ensure a thread is scheduled away, but
I'm not actually sure it guarantees CSI if it's kernel->kernel. Could
do a smp_call_function to do the isync on each CPU to be sure.

Good point. Per 
Documentation/RCU/Design/Requirements/Requirements.html#Tasks RCU:
"The solution, in the form of Tasks RCU, is to have implicit read-side 
critical sections that are delimited by voluntary context switches, that 
is, calls to schedule(), cond_resched(), and synchronize_rcu_tasks(). In 
addition, transitions to and from userspace execution also delimit 
tasks-RCU read-side critical sections."

I suppose transitions to/from userspace, as well as calls to schedule() 
result in context synchronizing instruction being executed. But, if some 
tasks call cond_resched() and synchronize_rcu_tasks(), we probably won't 
have a CSI executed.

Also:
"In CONFIG_PREEMPT=n kernels, trampolines cannot be preempted, so these 
APIs map to call_rcu(), synchronize_rcu(), and rcu_barrier(), 
respectively."

In this scenario as well, I think we won't have a CSI executed in case 
of cond_resched().

Should we enhance patch_instruction() to handle that?


- Naveen

Naveen N. Rao's on June 19, 2019 7:53 pm:

Nicholas Piggin wrote:

quoted

Michael Ellerman's on June 19, 2019 3:14 pm:

quoted

Hi Naveen,

Sorry I meant to reply to this earlier .. :/

No problem. Thanks for the questions.

quoted

quoted

"Naveen N. Rao" [off-list ref] writes:

quoted

With -mprofile-kernel, gcc emits 'mflr r0', followed by 'bl _mcount' to
enable function tracing and profiling. So far, with dynamic ftrace, we
used to only patch out the branch to _mcount(). However, mflr is
executed by the branch unit that can only execute one per cycle on
POWER9 and shared with branches, so it would be nice to avoid it where
possible.

We cannot simply nop out the mflr either. When enabling function
tracing, there can be a race if tracing is enabled when some thread was
interrupted after executing a nop'ed out mflr. In this case, the thread
would execute the now-patched-in branch to _mcount() without having
executed the preceding mflr.

To solve this, we now enable function tracing in 2 steps: patch in the
mflr instruction, use synchronize_rcu_tasks() to ensure all existing
threads make progress, and then patch in the branch to _mcount(). We
override ftrace_replace_code() with a powerpc64 variant for this
purpose.

According to the ISA we're not allowed to patch mflr at runtime. See the
section on "CMODX".

According to "quasi patch class" engineering note, we can patch
anything with a preferred nop. But that's written as an optional
facility, which we don't have a feature to test for.

Hmm... I wonder what the implications are. We've been patching in a 
'trap' for kprobes for a long time now, along with having to patch back 
the original instruction (which can be anything), when the probe is 
removed.

Will have to check what implementations support "quasi patch class"
instructions. IIRC recent POWER processors are okay. May have to add
a feature test though.

quoted

quoted

I'm also not convinced the ordering between the two patches is
guaranteed by the ISA, given that there's possibly no isync on the other
CPU.

Will they go through a context synchronizing event?

synchronize_rcu_tasks() should ensure a thread is scheduled away, but
I'm not actually sure it guarantees CSI if it's kernel->kernel. Could
do a smp_call_function to do the isync on each CPU to be sure.

Good point. Per 
Documentation/RCU/Design/Requirements/Requirements.html#Tasks RCU:
"The solution, in the form of Tasks RCU, is to have implicit read-side 
critical sections that are delimited by voluntary context switches, that 
is, calls to schedule(), cond_resched(), and synchronize_rcu_tasks(). In 
addition, transitions to and from userspace execution also delimit 
tasks-RCU read-side critical sections."

I suppose transitions to/from userspace, as well as calls to schedule() 
result in context synchronizing instruction being executed. But, if some 
tasks call cond_resched() and synchronize_rcu_tasks(), we probably won't 
have a CSI executed.

Also:
"In CONFIG_PREEMPT=n kernels, trampolines cannot be preempted, so these 
APIs map to call_rcu(), synchronize_rcu(), and rcu_barrier(), 
respectively."

In this scenario as well, I think we won't have a CSI executed in case 
of cond_resched().

Should we enhance patch_instruction() to handle that?

Well, not sure. Do we have many post-boot callers of it? Should
they take care of their own synchronization requirements?

Thanks,
Nick

Nicholas Piggin wrote:

Naveen N. Rao's on June 19, 2019 7:53 pm:

quoted

Nicholas Piggin wrote:

quoted

Michael Ellerman's on June 19, 2019 3:14 pm:

quoted

I'm also not convinced the ordering between the two patches is
guaranteed by the ISA, given that there's possibly no isync on the other
CPU.

Will they go through a context synchronizing event?

synchronize_rcu_tasks() should ensure a thread is scheduled away, but
I'm not actually sure it guarantees CSI if it's kernel->kernel. Could
do a smp_call_function to do the isync on each CPU to be sure.

Good point. Per 
Documentation/RCU/Design/Requirements/Requirements.html#Tasks RCU:
"The solution, in the form of Tasks RCU, is to have implicit read-side 
critical sections that are delimited by voluntary context switches, that 
is, calls to schedule(), cond_resched(), and synchronize_rcu_tasks(). In 
addition, transitions to and from userspace execution also delimit 
tasks-RCU read-side critical sections."

I suppose transitions to/from userspace, as well as calls to schedule() 
result in context synchronizing instruction being executed. But, if some 
tasks call cond_resched() and synchronize_rcu_tasks(), we probably won't 
have a CSI executed.

Also:
"In CONFIG_PREEMPT=n kernels, trampolines cannot be preempted, so these 
APIs map to call_rcu(), synchronize_rcu(), and rcu_barrier(), 
respectively."

In this scenario as well, I think we won't have a CSI executed in case 
of cond_resched().

Should we enhance patch_instruction() to handle that?

Well, not sure. Do we have many post-boot callers of it? Should
they take care of their own synchronization requirements?

Kprobes and ftrace are the two users (along with anything else that may 
use jump labels).

Looking at this from the CMODX perspective: the main example quoted of 
an erratic behavior is when any variant of the patched instruction 
causes an exception.

With ftrace, I think we are ok since we only ever patch a 'nop' or a 
'bl' (and the 'mflr' now), none of which should cause an exception. As 
such, the existing patch_instruction() should suffice.

However, with kprobes, we patch a 'trap' (or a branch in case of 
optprobes) on most instructions. I wonder if we should be issuing an 
'isync' on all cpus in this case. Or, even if that is sufficient or 
necessary.


Thanks,
Naveen

With KPROBES_ON_FTRACE, kprobe is allowed to be inserted on instructions
that branch to _mcount (referred to as ftrace location). With
-mprofile-kernel, we now include the preceding 'mflr r0' as being part
of the ftrace location.

However, by default, probing on an instruction that is not actually the
branch to _mcount() is prohibited, as that is considered to not be at an
instruction boundary. This is not the case on powerpc, so allow the same
by overriding arch_check_ftrace_location()

In addition, we update kprobe_ftrace_handler() to detect this scenarios
and to pass the proper nip to the pre and post probe handlers.

Signed-off-by: Naveen N. Rao <redacted>
---
 arch/powerpc/kernel/kprobes-ftrace.c | 30 ++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/arch/powerpc/kernel/kprobes-ftrace.c b/arch/powerpc/kernel/kprobes-ftrace.c
index 972cb28174b2..6a0bd3c16cb6 100644
--- a/arch/powerpc/kernel/kprobes-ftrace.c
+++ b/arch/powerpc/kernel/kprobes-ftrace.c

@@ -12,14 +12,34 @@
 #include <linux/preempt.h>
 #include <linux/ftrace.h>
 
+/*
+ * With -mprofile-kernel, we patch two instructions -- the branch to _mcount
+ * as well as the preceding 'mflr r0'. Both these instructions are claimed
+ * by ftrace and we should allow probing on either instruction.
+ */
+int arch_check_ftrace_location(struct kprobe *p)
+{
+	if (ftrace_location((unsigned long)p->addr))
+		p->flags |= KPROBE_FLAG_FTRACE;
+	return 0;
+}
+
 /* Ftrace callback handler for kprobes */
 void kprobe_ftrace_handler(unsigned long nip, unsigned long parent_nip,
 			   struct ftrace_ops *ops, struct pt_regs *regs)
 {
 	struct kprobe *p;
+	int mflr_kprobe = 0;
 	struct kprobe_ctlblk *kcb;
 
 	p = get_kprobe((kprobe_opcode_t *)nip);
+	if (unlikely(!p)) {
+		p = get_kprobe((kprobe_opcode_t *)(nip - MCOUNT_INSN_SIZE));
+		if (!p)
+			return;
+		mflr_kprobe = 1;
+	}
+
 	if (unlikely(!p) || kprobe_disabled(p))
 		return;
 

@@ -33,6 +53,9 @@ void kprobe_ftrace_handler(unsigned long nip, unsigned long parent_nip,
 		 */
 		regs->nip -= MCOUNT_INSN_SIZE;
 
+		if (mflr_kprobe)
+			regs->nip -= MCOUNT_INSN_SIZE;
+
 		__this_cpu_write(current_kprobe, p);
 		kcb->kprobe_status = KPROBE_HIT_ACTIVE;
 		if (!p->pre_handler || !p->pre_handler(p, regs)) {

@@ -45,6 +68,8 @@ void kprobe_ftrace_handler(unsigned long nip, unsigned long parent_nip,
 				kcb->kprobe_status = KPROBE_HIT_SSDONE;
 				p->post_handler(p, regs, 0);
 			}
+			if (mflr_kprobe)
+				regs->nip += MCOUNT_INSN_SIZE;
 		}
 		/*
 		 * If pre_handler returns !0, it changes regs->nip. We have to

@@ -57,6 +82,11 @@ NOKPROBE_SYMBOL(kprobe_ftrace_handler);
 
 int arch_prepare_kprobe_ftrace(struct kprobe *p)
 {
+	if ((unsigned long)p->addr & 0x03) {
+		printk("Attempt to register kprobe at an unaligned address\n");
+		return -EILSEQ;
+	}
+
 	p->ainsn.insn = NULL;
 	p->ainsn.boostable = -1;
 	return 0;

-- 
2.22.0

On Tue, 18 Jun 2019 20:17:06 +0530
"Naveen N. Rao" [off-list ref] wrote:

Hmm, is this really unlikely? If we put a kprobe on the second instruction address,
we will see p == NULL always.

+		p = get_kprobe((kprobe_opcode_t *)(nip - MCOUNT_INSN_SIZE));
+		if (!p)

Here will be unlikely, because we can not find kprobe at both of nip and
nip - MCOUNT_INSN_SIZE.

+			return;
+		mflr_kprobe = 1;
+	}
+
 	if (unlikely(!p) || kprobe_disabled(p))

"unlikely(!p)" is not needed here.

Thank you,

-- 
Masami Hiramatsu [off-list ref]

On Fri, 2019-06-21 at 23:50 +0900, Masami Hiramatsu wrote:

On Tue, 18 Jun 2019 20:17:06 +0530
"Naveen N. Rao" [off-list ref] wrote:

trivia:

quoted

diff --git a/arch/powerpc/kernel/kprobes-ftrace.c b/arch/powerpc/kernel/kprobes-ftrace.c

[]

quoted

@@ -57,6 +82,11 @@ NOKPROBE_SYMBOL(kprobe_ftrace_handler);
 
 int arch_prepare_kprobe_ftrace(struct kprobe *p)
 {
+	if ((unsigned long)p->addr & 0x03) {
+		printk("Attempt to register kprobe at an unaligned address\n");

Please use the appropriate KERN_<LEVEL> or pr_<level>

Masami Hiramatsu wrote:

On Tue, 18 Jun 2019 20:17:06 +0530
"Naveen N. Rao" [off-list ref] wrote:

quoted

With KPROBES_ON_FTRACE, kprobe is allowed to be inserted on instructions
that branch to _mcount (referred to as ftrace location). With
-mprofile-kernel, we now include the preceding 'mflr r0' as being part
of the ftrace location.

However, by default, probing on an instruction that is not actually the
branch to _mcount() is prohibited, as that is considered to not be at an
instruction boundary. This is not the case on powerpc, so allow the same
by overriding arch_check_ftrace_location()

In addition, we update kprobe_ftrace_handler() to detect this scenarios
and to pass the proper nip to the pre and post probe handlers.

Signed-off-by: Naveen N. Rao <redacted>
---
 arch/powerpc/kernel/kprobes-ftrace.c | 30 ++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/arch/powerpc/kernel/kprobes-ftrace.c b/arch/powerpc/kernel/kprobes-ftrace.c
index 972cb28174b2..6a0bd3c16cb6 100644
--- a/arch/powerpc/kernel/kprobes-ftrace.c
+++ b/arch/powerpc/kernel/kprobes-ftrace.c

@@ -12,14 +12,34 @@
 #include <linux/preempt.h>
 #include <linux/ftrace.h>
 
+/*
+ * With -mprofile-kernel, we patch two instructions -- the branch to _mcount
+ * as well as the preceding 'mflr r0'. Both these instructions are claimed
+ * by ftrace and we should allow probing on either instruction.
+ */
+int arch_check_ftrace_location(struct kprobe *p)
+{
+	if (ftrace_location((unsigned long)p->addr))
+		p->flags |= KPROBE_FLAG_FTRACE;
+	return 0;
+}
+
 /* Ftrace callback handler for kprobes */
 void kprobe_ftrace_handler(unsigned long nip, unsigned long parent_nip,
 			   struct ftrace_ops *ops, struct pt_regs *regs)
 {
 	struct kprobe *p;
+	int mflr_kprobe = 0;
 	struct kprobe_ctlblk *kcb;
 
 	p = get_kprobe((kprobe_opcode_t *)nip);
+	if (unlikely(!p)) {

Hmm, is this really unlikely? If we put a kprobe on the second instruction address,
we will see p == NULL always.

quoted

+		p = get_kprobe((kprobe_opcode_t *)(nip - MCOUNT_INSN_SIZE));
+		if (!p)

Here will be unlikely, because we can not find kprobe at both of nip and
nip - MCOUNT_INSN_SIZE.

quoted

+			return;
+		mflr_kprobe = 1;
+	}
+
 	if (unlikely(!p) || kprobe_disabled(p))

"unlikely(!p)" is not needed here.

...

Joe Perches wrote:

On Fri, 2019-06-21 at 23:50 +0900, Masami Hiramatsu wrote:

quoted

On Tue, 18 Jun 2019 20:17:06 +0530
"Naveen N. Rao" [off-list ref] wrote:

trivia:

quoted

quoted

diff --git a/arch/powerpc/kernel/kprobes-ftrace.c b/arch/powerpc/kernel/kprobes-ftrace.c

[]

quoted

quoted

@@ -57,6 +82,11 @@ NOKPROBE_SYMBOL(kprobe_ftrace_handler);
 
 int arch_prepare_kprobe_ftrace(struct kprobe *p)
 {
+	if ((unsigned long)p->addr & 0x03) {
+		printk("Attempt to register kprobe at an unaligned address\n");

Please use the appropriate KERN_<LEVEL> or pr_<level>

All good points. Thanks for the review.


- Naveen

Ftrace location could include more than a single instruction in case of
some architectures (powerpc64, for now). In this case, kprobe is
permitted on any of those instructions, and uses ftrace infrastructure
for functioning.

However, [dis]arm_kprobe_ftrace() uses the kprobe address when setting
up ftrace filter IP. This won't work if the address points to any
instruction apart from the one that has a branch to _mcount(). To
resolve this, have [dis]arm_kprobe_ftrace() use ftrace_function() to
identify the filter IP.

Signed-off-by: Naveen N. Rao <redacted>
---
 kernel/kprobes.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/kernel/kprobes.c b/kernel/kprobes.c
index 445337c107e0..282ee704e2d8 100644
--- a/kernel/kprobes.c
+++ b/kernel/kprobes.c

@@ -978,10 +978,10 @@ static int prepare_kprobe(struct kprobe *p)
 /* Caller must lock kprobe_mutex */
 static int arm_kprobe_ftrace(struct kprobe *p)
 {
+	unsigned long ftrace_ip = ftrace_location((unsigned long)p->addr);
 	int ret = 0;
 
-	ret = ftrace_set_filter_ip(&kprobe_ftrace_ops,
-				   (unsigned long)p->addr, 0, 0);
+	ret = ftrace_set_filter_ip(&kprobe_ftrace_ops, ftrace_ip, 0, 0);
 	if (ret) {
 		pr_debug("Failed to arm kprobe-ftrace at %pS (%d)\n",
 			 p->addr, ret);

@@ -1005,13 +1005,14 @@ static int arm_kprobe_ftrace(struct kprobe *p)
 	 * non-empty filter_hash for IPMODIFY ops, we're safe from an accidental
 	 * empty filter_hash which would undesirably trace all functions.
 	 */
-	ftrace_set_filter_ip(&kprobe_ftrace_ops, (unsigned long)p->addr, 1, 0);
+	ftrace_set_filter_ip(&kprobe_ftrace_ops, ftrace_ip, 1, 0);
 	return ret;
 }
 
 /* Caller must lock kprobe_mutex */
 static int disarm_kprobe_ftrace(struct kprobe *p)
 {
+	unsigned long ftrace_ip = ftrace_location((unsigned long)p->addr);
 	int ret = 0;
 
 	if (kprobe_ftrace_enabled == 1) {