Skip to contentSkip to conversation

ring-buffer: Add rb_check_bpage in __rb_allocate_pages

PATCH Superseded Qiujun Huang 2 participants 5 messages Oct 14, 2020

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ring-buffer: Add rb_check_bpage in __rb_allocate_pages

It may be better to check each page is aligned by 4 bytes. The 2 least significant bits of the address will be used as flags.

---
 kernel/trace/ring_buffer.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
kernel/trace/ring_buffer.c +7 -4
index 93ef0ab6ea20..9dec7d58b177 100644
@@ -1420,7 +1420,8 @@ static int rb_check_pages(struct ring_buffer_per_cpu *cpu_buffer)
 	return 0;
 }
 
-static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
+static int __rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
+		long nr_pages, struct list_head *pages, int cpu)
 {
 	struct buffer_page *bpage, *tmp;
 	bool user_thread = current->mm != NULL;
@@ -1464,6 +1465,8 @@ static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
 		if (!bpage)
 			goto free_pages;
 
+		rb_check_bpage(cpu_buffer, bpage);
+
 		list_add(&bpage->list, pages);
 
 		page = alloc_pages_node(cpu_to_node(cpu), mflags, 0);
@@ -1498,7 +1501,7 @@ static int rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
 
 	WARN_ON(!nr_pages);
 
-	if (__rb_allocate_pages(nr_pages, &pages, cpu_buffer->cpu))
+	if (__rb_allocate_pages(cpu_buffer, nr_pages, &pages, cpu_buffer->cpu))
 		return -ENOMEM;
 
 	/*
@@ -2007,7 +2010,7 @@ int ring_buffer_resize(struct trace_buffer *buffer, unsigned long size,
 			 * allocated without receiving ENOMEM
 			 */
 			INIT_LIST_HEAD(&cpu_buffer->new_pages);
-			if (__rb_allocate_pages(cpu_buffer->nr_pages_to_update,
+			if (__rb_allocate_pages(cpu_buffer, cpu_buffer->nr_pages_to_update,
 						&cpu_buffer->new_pages, cpu)) {
 				/* not enough memory for new pages */
 				err = -ENOMEM;
@@ -2073,7 +2076,7 @@ int ring_buffer_resize(struct trace_buffer *buffer, unsigned long size,
 
 		INIT_LIST_HEAD(&cpu_buffer->new_pages);
 		if (cpu_buffer->nr_pages_to_update > 0 &&
-			__rb_allocate_pages(cpu_buffer->nr_pages_to_update,
+			__rb_allocate_pages(cpu_buffer, cpu_buffer->nr_pages_to_update,
 					    &cpu_buffer->new_pages, cpu_id)) {
 			err = -ENOMEM;
 			goto out_err;
Signature · 1 line
2.17.1
maintainer
#1 lore raw
Quoted 23 lines · On Wed, 14 Oct 2020 23:16:14 +0800 Qiujun Huang <[email protected]> wrote:

It may be better to check each page is aligned by 4 bytes. The 2 least significant bits of the address will be used as flags.

---
 kernel/trace/ring_buffer.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
kernel/trace/ring_buffer.c +2 -1
index 93ef0ab6ea20..9dec7d58b177 100644
@@ -1420,7 +1420,8 @@ static int rb_check_pages(struct ring_buffer_per_cpu *cpu_buffer)
 	return 0;
 }
 
-static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
+static int __rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
+		long nr_pages, struct list_head *pages, int cpu)
 {
 	struct buffer_page *bpage, *tmp;
 	bool user_thread = current->mm != NULL;

kernel/trace/ring_buffer.c static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu) · line 1470

Quoted 6 lines
kernel/trace/ring_buffer.c +2 -0
@@ -1464,6 +1465,8 @@ static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
 		if (!bpage)
 			goto free_pages;
 
+		rb_check_bpage(cpu_buffer, bpage);
+
 

Why add it here, and not just add this check to the scan in rb_check_pages()?

-- Steve

#2 lore raw
Quoted 30 lines · Steven Rostedt
Quoted 26 lines · On Wed, 14 Oct 2020 23:16:14 +0800 Qiujun Huang <[email protected]> wrote:

It may be better to check each page is aligned by 4 bytes. The 2 least significant bits of the address will be used as flags.

---
 kernel/trace/ring_buffer.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
kernel/trace/ring_buffer.c +2 -1
index 93ef0ab6ea20..9dec7d58b177 100644
@@ -1420,7 +1420,8 @@ static int rb_check_pages(struct ring_buffer_per_cpu *cpu_buffer)
      return 0;
 }
 
-static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
+static int __rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
+             long nr_pages, struct list_head *pages, int cpu)
 {
      struct buffer_page *bpage, *tmp;
      bool user_thread = current->mm != NULL;
@@ -1464,6 +1465,8 @@ static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
              if (!bpage)
                      goto free_pages;
 
Quoted 6 lines

kernel/trace/ring_buffer.c static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu) · line 1471

Quoted 3 lines
kernel/trace/ring_buffer.c +2 -0
+             rb_check_bpage(cpu_buffer, bpage);
+
 

Why add it here, and not just add this check to the scan in rb_check_pages()?

rb_head_page_deactivate() in rb_check_pages() will clear the 2 LSB first.

Quoted 2 lines

-- Steve

maintainer
#3 lore raw
Quoted 31 lines · On Wed, 14 Oct 2020 23:48:05 +0800 Qiujun Huang <[email protected]> wrote:
Quoted 30 lines · Steven Rostedt
Quoted 26 lines · On Wed, 14 Oct 2020 23:16:14 +0800 Qiujun Huang <[email protected]> wrote:

It may be better to check each page is aligned by 4 bytes. The 2 least significant bits of the address will be used as flags.

---
 kernel/trace/ring_buffer.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
kernel/trace/ring_buffer.c +2 -1
index 93ef0ab6ea20..9dec7d58b177 100644
@@ -1420,7 +1420,8 @@ static int rb_check_pages(struct ring_buffer_per_cpu *cpu_buffer)
      return 0;
 }
 
-static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
+static int __rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
+             long nr_pages, struct list_head *pages, int cpu)
 {
      struct buffer_page *bpage, *tmp;
      bool user_thread = current->mm != NULL;
@@ -1464,6 +1465,8 @@ static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
              if (!bpage)
                      goto free_pages;
 
Quoted 9 lines
Quoted 6 lines

kernel/trace/ring_buffer.c static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu) · line 1471

Quoted 3 lines
kernel/trace/ring_buffer.c +2 -0
+             rb_check_bpage(cpu_buffer, bpage);
+
 

Why add it here, and not just add this check to the scan in rb_check_pages()?

rb_head_page_deactivate() in rb_check_pages() will clear the 2 LSB first.

Well, you could just add another scan there, but if you want to do it this way, then remove passing the int cpu to these functions, and use the cpu_buffer->cpu, as keeping the cpu is just redundant.

Also, did you see an issue? This check is more of me being paranoid to make sure we don't crash later. I've honestly never seen it trigger.

-- Steve

#4 lore raw
Quoted 42 lines · Steven Rostedt
Quoted 38 lines · On Wed, 14 Oct 2020 23:48:05 +0800 Qiujun Huang <[email protected]> wrote:
Quoted 36 lines · Steven Rostedt
Quoted 23 lines · On Wed, 14 Oct 2020 23:16:14 +0800 Qiujun Huang <[email protected]> wrote:

It may be better to check each page is aligned by 4 bytes. The 2 least significant bits of the address will be used as flags.

---
 kernel/trace/ring_buffer.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
kernel/trace/ring_buffer.c +2 -1
index 93ef0ab6ea20..9dec7d58b177 100644
@@ -1420,7 +1420,8 @@ static int rb_check_pages(struct ring_buffer_per_cpu *cpu_buffer)
      return 0;
 }
 
-static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
+static int __rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
+             long nr_pages, struct list_head *pages, int cpu)
 {
      struct buffer_page *bpage, *tmp;
      bool user_thread = current->mm != NULL;

kernel/trace/ring_buffer.c static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu) · line 1470

Quoted 6 lines
kernel/trace/ring_buffer.c +2 -0
@@ -1464,6 +1465,8 @@ static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu)
              if (!bpage)
                      goto free_pages;
 
+             rb_check_bpage(cpu_buffer, bpage);
+
 

Why add it here, and not just add this check to the scan in rb_check_pages()?

Quoted 6 lines
Quoted 2 lines

rb_head_page_deactivate() in rb_check_pages() will clear the 2 LSB first.

Well, you could just add another scan there, but if you want to do it this way, then remove passing the int cpu to these functions, and use the cpu_buffer->cpu, as keeping the cpu is just redundant.

Get it.

Quoted 2 lines

Also, did you see an issue? This check is more of me being paranoid to

No, I'm a little paranoid too following the code :-)

Quoted 3 lines

make sure we don't crash later. I've honestly never seen it trigger.

-- Steve