2015-04-27 12:17:26

by Alex Bennée

Subject: [PATCH] tracing: make ftrace_print_array_seq compute buf_len

The only caller to this function was getting it wrong. I favoured
pushing the calculation to as close to the need as possible rather than
fixing the one caller.

Signed-off-by: Alex Bennée <[email protected]>
---
include/linux/ftrace_event.h | 2 +-
kernel/trace/trace_output.c | 3 ++-
2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h
index c674ee8..e6b0262 100644
--- a/include/linux/ftrace_event.h
+++ b/include/linux/ftrace_event.h
@@ -45,7 +45,7 @@ const char *ftrace_print_hex_seq(struct trace_seq *p,
const unsigned char *buf, int len);

const char *ftrace_print_array_seq(struct trace_seq *p,
- const void *buf, int buf_len,
+ const void *buf, int len,
size_t el_size);

struct trace_iterator;
diff --git a/kernel/trace/trace_output.c b/kernel/trace/trace_output.c
index 692bf71..a42703d 100644
--- a/kernel/trace/trace_output.c
+++ b/kernel/trace/trace_output.c
@@ -178,12 +178,13 @@ ftrace_print_hex_seq(struct trace_seq *p, const unsigned char *buf, int buf_len)
EXPORT_SYMBOL(ftrace_print_hex_seq);

const char *
-ftrace_print_array_seq(struct trace_seq *p, const void *buf, int buf_len,
+ftrace_print_array_seq(struct trace_seq *p, const void *buf, int len,
size_t el_size)
{
const char *ret = trace_seq_buffer_ptr(p);
const char *prefix = "";
void *ptr = (void *)buf;
+ size_t buf_len = len * el_size;

trace_seq_putc(p, '{');

--
2.3.5

2015-04-27 14:22:11

by Dave Martin

Subject: Re: [PATCH] tracing: make ftrace_print_array_seq compute buf_len

On Mon, Apr 27, 2015 at 01:17:48PM +0100, Alex Benn?e wrote:
> The only caller to this function was getting it wrong. I favoured

What caller?

Wrong in what way?

> pushing the calculation to as close to the need as possible rather than
> fixing the one caller.

This seems reasonable, but...

>
> Signed-off-by: Alex Benn?e <[email protected]>
> ---
> include/linux/ftrace_event.h | 2 +-
> kernel/trace/trace_output.c | 3 ++-
> 2 files changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h
> index c674ee8..e6b0262 100644
> --- a/include/linux/ftrace_event.h
> +++ b/include/linux/ftrace_event.h
> @@ -45,7 +45,7 @@ const char *ftrace_print_hex_seq(struct trace_seq *p,
> const unsigned char *buf, int len);
>
> const char *ftrace_print_array_seq(struct trace_seq *p,
> - const void *buf, int buf_len,
> + const void *buf, int len,

How is the name "len" less confusing than "buf_len"?

I suggest matching the name to the equivalent argument of the
__print_array macro -- i.e., "count".

Cheers
---Dave

[...]

2015-04-27 18:00:29

by Steven Rostedt

Subject: Re: [PATCH] tracing: make ftrace_print_array_seq compute buf_len

On Mon, 27 Apr 2015 15:22:02 +0100
Dave P Martin <[email protected]> wrote:

> On Mon, Apr 27, 2015 at 01:17:48PM +0100, Alex Bennée wrote:
> > The only caller to this function was getting it wrong. I favoured
>
> What caller?
>
> Wrong in what way?

Yes, please add that info to the change log.

>
> > pushing the calculation to as close to the need as possible rather than
> > fixing the one caller.
>
> This seems reasonable, but...
>
> >
> > Signed-off-by: Alex Bennée <[email protected]>
> > ---
> > include/linux/ftrace_event.h | 2 +-
> > kernel/trace/trace_output.c | 3 ++-
> > 2 files changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h
> > index c674ee8..e6b0262 100644
> > --- a/include/linux/ftrace_event.h
> > +++ b/include/linux/ftrace_event.h
> > @@ -45,7 +45,7 @@ const char *ftrace_print_hex_seq(struct trace_seq *p,
> > const unsigned char *buf, int len);
> >
> > const char *ftrace_print_array_seq(struct trace_seq *p,
> > - const void *buf, int buf_len,
> > + const void *buf, int len,
>
> How is the name "len" less confusing than "buf_len"?
>
> I suggest matching the name to the equivalent argument of the
> __print_array macro -- i.e., "count".

I agree, please change the variable name to "count", that will make more
sense.

Thanks,

-- Steve

2015-04-28 14:40:30

by Alex Bennée

Subject: Re: [PATCH] tracing: make ftrace_print_array_seq compute buf_len

Steven Rostedt <[email protected]> writes:

> On Mon, 27 Apr 2015 15:22:02 +0100
> Dave P Martin <[email protected]> wrote:
>
>> On Mon, Apr 27, 2015 at 01:17:48PM +0100, Alex Bennée wrote:
>> > The only caller to this function was getting it wrong. I favoured
>>
>> What caller?
>>
>> Wrong in what way?
>
> Yes, please add that info to the change log.

It was ftrace's own __print_array, I'll add the info to v2.

>
>>
>> > pushing the calculation to as close to the need as possible rather than
>> > fixing the one caller.
>>
>> This seems reasonable, but...
>>
>> >
>> > Signed-off-by: Alex Bennée <[email protected]>
>> > ---
>> > include/linux/ftrace_event.h | 2 +-
>> > kernel/trace/trace_output.c | 3 ++-
>> > 2 files changed, 3 insertions(+), 2 deletions(-)
>> >
>> > diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h
>> > index c674ee8..e6b0262 100644
>> > --- a/include/linux/ftrace_event.h
>> > +++ b/include/linux/ftrace_event.h
>> > @@ -45,7 +45,7 @@ const char *ftrace_print_hex_seq(struct trace_seq *p,
>> > const unsigned char *buf, int len);
>> >
>> > const char *ftrace_print_array_seq(struct trace_seq *p,
>> > - const void *buf, int buf_len,
>> > + const void *buf, int len,
>>
>> How is the name "len" less confusing than "buf_len"?
>>
>> I suggest matching the name to the equivalent argument of the
>> __print_array macro -- i.e., "count".
>
> I agree, please change the variable name to "count", that will make more
> sense.

OK I'll re-spin today.

>
> Thanks,
>
> -- Steve

--
Alex Bennée