Welcome! Log In Create A New Profile

Advanced

[PATCH 1/2 v4] perf sched: fix wrong conversion of task state

Posted by Tomoki Sekiyama 
Tomoki Sekiyama
[PATCH 1/2 v4] perf sched: fix wrong conversion of task state
December 13, 2016 05:57AM
Currently sched_out_state() converts the prev_state u64 bitmask to a char
using the bitmask as an index, which may cause invalid memory access.
This fixes the issue by using the __ffs() returned value as an index.

Signed-off-by: Tomoki Sekiyama <[email protected]>
Fixes: cdce9d738b91e ("perf sched: Add sched latency profiling")
Cc: Jiri Olsa <[email protected]>
Cc: David Ahern <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Masami Hiramatsu <[email protected]>
---
Changes from v3:
Added Fixes:
Split the patch into 2 pieces.

tools/perf/builtin-sched.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index 0dfe8df..ce9bef6 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -897,9 +897,10 @@ static int thread_atoms_insert(struct perf_sched *sched, struct thread *thread)

static char sched_out_state(u64 prev_state)
{
- const char *str = TASK_STATE_TO_CHAR_STR;
+ const char str[] = TASK_STATE_TO_CHAR_STR;
+ unsigned int bit = prev_state ? __ffs(prev_state) + 1 : 0;

- return str[prev_state];
+ return bit < ARRAY_SIZE(str) - 1 ? str[bit] : '?';
}

static int
--
2.7.4
Update TASK_STATE_TO_CHAR_STR macro to one from sched.h in the latest
kernel, where 'N' and 'n' are introduced, 'X' and 'Z' are swapped.

Signed-off-by: Tomoki Sekiyama <[email protected]>
Fixes: cdce9d738b91e ("perf sched: Add sched latency profiling")
Cc: Jiri Olsa <[email protected]>
Cc: David Ahern <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Masami Hiramatsu <[email protected]>
---
Changes from v3:
Added Fixes:
Split patch into 2 pieces.

tools/perf/builtin-sched.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index ce9bef6..5776263 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -70,7 +70,7 @@ struct sched_atom {
struct task_desc *wakee;
};

-#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
+#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"

enum thread_state {
THREAD_SLEEPING = 0,
--
2.7.4
On 7/31/16 10:30 PM, Tomoki Sekiyama wrote:
> Currently sched_out_state() converts the prev_state u64 bitmask to a char
> using the bitmask as an index, which may cause invalid memory access.
> This fixes the issue by using the __ffs() returned value as an index.
>
> Signed-off-by: Tomoki Sekiyama <[email protected]>
> Fixes: cdce9d738b91e ("perf sched: Add sched latency profiling")
> Cc: Jiri Olsa <[email protected]>
> Cc: David Ahern <[email protected]>
> Cc: Namhyung Kim <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Masami Hiramatsu <[email protected]>
> ---
> Changes from v3:
> Added Fixes:
> Split the patch into 2 pieces.
>
> tools/perf/builtin-sched.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
> index 0dfe8df..ce9bef6 100644
> --- a/tools/perf/builtin-sched.c
> +++ b/tools/perf/builtin-sched.c
> @@ -897,9 +897,10 @@ static int thread_atoms_insert(struct perf_sched *sched, struct thread *thread)
>
> static char sched_out_state(u64 prev_state)
> {
> - const char *str = TASK_STATE_TO_CHAR_STR;
> + const char str[] = TASK_STATE_TO_CHAR_STR;
> + unsigned int bit = prev_state ? __ffs(prev_state) + 1 : 0;
>
> - return str[prev_state];
> + return bit < ARRAY_SIZE(str) - 1 ? str[bit] : '?';
> }
>
> static int
>


Acked-by: David Ahern <[email protected]>
On 7/31/16 10:30 PM, Tomoki Sekiyama wrote:
> Update TASK_STATE_TO_CHAR_STR macro to one from sched.h in the latest
> kernel, where 'N' and 'n' are introduced, 'X' and 'Z' are swapped.
>
> Signed-off-by: Tomoki Sekiyama <[email protected]>
> Fixes: cdce9d738b91e ("perf sched: Add sched latency profiling")

Wrong Fixes tag for this change.

This is really an update based on kernel changes, the last one being
7dc603c9028ea.


> Cc: Jiri Olsa <[email protected]>
> Cc: David Ahern <[email protected]>
> Cc: Namhyung Kim <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Masami Hiramatsu <[email protected]>
> ---
> Changes from v3:
> Added Fixes:
> Split patch into 2 pieces.
>
> tools/perf/builtin-sched.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
> index ce9bef6..5776263 100644
> --- a/tools/perf/builtin-sched.c
> +++ b/tools/perf/builtin-sched.c
> @@ -70,7 +70,7 @@ struct sched_atom {
> struct task_desc *wakee;
> };
>
> -#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
> +#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
>
> enum thread_state {
> THREAD_SLEEPING = 0,
>

Acked-by: David Ahern <[email protected]>
Masami Hiramatsu
Re: [PATCH 1/2 v4] perf sched: fix wrong conversion of task state
December 13, 2016 05:59AM
On Mon, 1 Aug 2016 13:30:25 +0900
Tomoki Sekiyama <[email protected]> wrote:

> Currently sched_out_state() converts the prev_state u64 bitmask to a char
> using the bitmask as an index, which may cause invalid memory access.
> This fixes the issue by using the __ffs() returned value as an index.

OK, looks good to me:)

Acked-by: Masami Hiramatsu <[email protected]>

Thanks!

>
> Signed-off-by: Tomoki Sekiyama <[email protected]>
> Fixes: cdce9d738b91e ("perf sched: Add sched latency profiling")
> Cc: Jiri Olsa <[email protected]>
> Cc: David Ahern <[email protected]>
> Cc: Namhyung Kim <[email protected]>
> Cc: Peter Zijlstra <[email protected]>
> Cc: Masami Hiramatsu <[email protected]>
> ---
> Changes from v3:
> Added Fixes:
> Split the patch into 2 pieces.
>
> tools/perf/builtin-sched.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
> index 0dfe8df..ce9bef6 100644
> --- a/tools/perf/builtin-sched.c
> +++ b/tools/perf/builtin-sched.c
> @@ -897,9 +897,10 @@ static int thread_atoms_insert(struct perf_sched *sched, struct thread *thread)
>
> static char sched_out_state(u64 prev_state)
> {
> - const char *str = TASK_STATE_TO_CHAR_STR;
> + const char str[] = TASK_STATE_TO_CHAR_STR;
> + unsigned int bit = prev_state ? __ffs(prev_state) + 1 : 0;
>
> - return str[prev_state];
> + return bit < ARRAY_SIZE(str) - 1 ? str[bit] : '?';
> }
>
> static int
> --
> 2.7.4
>


--
Masami Hiramatsu <[email protected]>
On Mon, 1 Aug 2016 07:58:19 -0600
David Ahern <[email protected]> wrote:

> On 7/31/16 10:30 PM, Tomoki Sekiyama wrote:
> > Update TASK_STATE_TO_CHAR_STR macro to one from sched.h in the latest
> > kernel, where 'N' and 'n' are introduced, 'X' and 'Z' are swapped.
> >
> > Signed-off-by: Tomoki Sekiyama <[email protected]>
> > Fixes: cdce9d738b91e ("perf sched: Add sched latency profiling")
>
> Wrong Fixes tag for this change.
>
> This is really an update based on kernel changes, the last one being
> 7dc603c9028ea.

Except for the above mistake, I'm also OK for the code.

Thanks,

>
>
> > Cc: Jiri Olsa <[email protected]>
> > Cc: David Ahern <[email protected]>
> > Cc: Namhyung Kim <[email protected]>
> > Cc: Peter Zijlstra <[email protected]>
> > Cc: Masami Hiramatsu <[email protected]>
> > ---
> > Changes from v3:
> > Added Fixes:
> > Split patch into 2 pieces.
> >
> > tools/perf/builtin-sched.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
> > index ce9bef6..5776263 100644
> > --- a/tools/perf/builtin-sched.c
> > +++ b/tools/perf/builtin-sched.c
> > @@ -70,7 +70,7 @@ struct sched_atom {
> > struct task_desc *wakee;
> > };
> >
> > -#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
> > +#define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
> >
> > enum thread_state {
> > THREAD_SLEEPING = 0,
> >
>
> Acked-by: David Ahern <[email protected]>


--
Masami Hiramatsu <[email protected]>
Sorry, only registered users may post in this forum.

Click here to login