2013-04-11 06:55:10

by Namhyung Kim

Subject: [PATCH 1/2] tracing: Fix possible NULL pointer dereferences

From: Namhyung Kim <[email protected]>

Currently set_ftrace_pid and set_graph_function files use seq_lseek
for their fops. However seq_open() is called only for FMODE_READ in
the fops->open() so that if an user tries to seek one of those file
when she open it for writing, it sees NULL seq_file and then panic.

It can be easily reproduced with following command:

$ cd /sys/kernel/debug/tracing
$ echo 1234 | sudo tee -a set_ftrace_pid

In this example, GNU coreutils' tee opens the file with fopen(, "a")
and then the fopen() internally calls lseek().

Signed-off-by: Namhyung Kim <[email protected]>
---
include/linux/ftrace.h | 2 +-
kernel/trace/ftrace.c | 10 +++++-----
kernel/trace/trace_stack.c | 2 +-
3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index 832422d706f4..c7687d8c0b30 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h
@@ -396,7 +396,7 @@ ssize_t ftrace_filter_write(struct file *file, const char __user *ubuf,
size_t cnt, loff_t *ppos);
ssize_t ftrace_notrace_write(struct file *file, const char __user *ubuf,
size_t cnt, loff_t *ppos);
-loff_t ftrace_regex_lseek(struct file *file, loff_t offset, int whence);
+loff_t ftrace_filter_lseek(struct file *file, loff_t offset, int whence);
int ftrace_regex_release(struct inode *inode, struct file *file);

void __init
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 03953fbd1b2a..be2bcb7ab2f7 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -2688,7 +2688,7 @@ ftrace_notrace_open(struct inode *inode, struct file *file)
}

loff_t
-ftrace_regex_lseek(struct file *file, loff_t offset, int whence)
+ftrace_filter_lseek(struct file *file, loff_t offset, int whence)
{
loff_t ret;

@@ -3613,7 +3613,7 @@ static const struct file_operations ftrace_filter_fops = {
.open = ftrace_filter_open,
.read = seq_read,
.write = ftrace_filter_write,
- .llseek = ftrace_regex_lseek,
+ .llseek = ftrace_filter_lseek,
.release = ftrace_regex_release,
};

@@ -3621,7 +3621,7 @@ static const struct file_operations ftrace_notrace_fops = {
.open = ftrace_notrace_open,
.read = seq_read,
.write = ftrace_notrace_write,
- .llseek = ftrace_regex_lseek,
+ .llseek = ftrace_filter_lseek,
.release = ftrace_regex_release,
};

@@ -3826,8 +3826,8 @@ static const struct file_operations ftrace_graph_fops = {
.open = ftrace_graph_open,
.read = seq_read,
.write = ftrace_graph_write,
+ .llseek = ftrace_filter_lseek,
.release = ftrace_graph_release,
- .llseek = seq_lseek,
};
#endif /* CONFIG_FUNCTION_GRAPH_TRACER */

@@ -4481,7 +4481,7 @@ static const struct file_operations ftrace_pid_fops = {
.open = ftrace_pid_open,
.write = ftrace_pid_write,
.read = seq_read,
- .llseek = seq_lseek,
+ .llseek = ftrace_filter_lseek,
.release = ftrace_pid_release,
};

diff --git a/kernel/trace/trace_stack.c b/kernel/trace/trace_stack.c
index 8c3f37e2dc43..b20428c5efe2 100644
--- a/kernel/trace/trace_stack.c
+++ b/kernel/trace/trace_stack.c
@@ -382,7 +382,7 @@ static const struct file_operations stack_trace_filter_fops = {
.open = stack_trace_filter_open,
.read = seq_read,
.write = ftrace_filter_write,
- .llseek = ftrace_regex_lseek,
+ .llseek = ftrace_filter_lseek,
.release = ftrace_regex_release,
};

--
1.7.11.7

2013-04-11 06:55:17

by Namhyung Kim

Subject: [PATCH 2/2] tracing: Reset ftrace_graph_filter_enabled if count is zero

From: Namhyung Kim <[email protected]>

The ftrace_graph_count can be decreased with a "!" pattern, so that
the enabled flag should be updated too.

Signed-off-by: Namhyung Kim <[email protected]>
---
kernel/trace/ftrace.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index be2bcb7ab2f7..3b5ef16692c5 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -3779,7 +3779,11 @@ out:
if (fail)
return -EINVAL;

- ftrace_graph_filter_enabled = 1;
+ if (*idx)
+ ftrace_graph_filter_enabled = 1;
+ else
+ ftrace_graph_filter_enabled = 0;
+
return 0;
}

--
1.7.11.7

2013-04-11 07:02:12

by Namhyung Kim

Subject: [PATCH v2 2/2] tracing: Reset ftrace_graph_filter_enabled if count is zero

From: Namhyung Kim <[email protected]>

The ftrace_graph_count can be decreased with a "!" pattern, so that
the enabled flag should be updated too.

Signed-off-by: Namhyung Kim <[email protected]>
---
kernel/trace/ftrace.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index be2bcb7ab2f7..daf8df20b764 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -3779,7 +3779,8 @@ out:
if (fail)
return -EINVAL;

- ftrace_graph_filter_enabled = 1;
+ ftrace_graph_filter_enabled = !!(*idx);
+
return 0;
}

--
1.7.11.7

2013-04-12 18:13:18

by Steven Rostedt

Subject: Re: [PATCH v2 2/2] tracing: Reset ftrace_graph_filter_enabled if count is zero

On Thu, 2013-04-11 at 16:01 +0900, Namhyung Kim wrote:
> From: Namhyung Kim <[email protected]>
>
> The ftrace_graph_count can be decreased with a "!" pattern, so that
> the enabled flag should be updated too.

Thanks!

Once again I'll push patch 1 to 3.9 and stable as that's a nasty bug.

But this patch isn't that big of a deal, and as we are in the later
-rc's I'll label it with stable and add it to the 3.10 queue.

Thanks,

-- Steve

>
> Signed-off-by: Namhyung Kim <[email protected]>
> ---
> kernel/trace/ftrace.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
> index be2bcb7ab2f7..daf8df20b764 100644
> --- a/kernel/trace/ftrace.c
> +++ b/kernel/trace/ftrace.c
> @@ -3779,7 +3779,8 @@ out:
> if (fail)
> return -EINVAL;
>
> - ftrace_graph_filter_enabled = 1;
> + ftrace_graph_filter_enabled = !!(*idx);
> +
> return 0;
> }
>