[v2] i2c: core-smbus: don't trace smbus_reply data on errors

If an smbus transfer fails, there's no guarantee that the output
buffer was written.  So, avoid trying to show the output buffer when
tracing after an error.  This was 'mostly harmless', but would trip
up kasan checking if left-over cruft in byte 0 is a large length,
causing us to read from unwritten memory.

Signed-off-by: John Sperbeck <jsperbeck@google.com>
---
 drivers/i2c/i2c-core-smbus.c | 2 +-
 include/trace/events/smbus.h | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)

On Sun, 10 Feb 2019 19:40:21 -0800
John Sperbeck <jsperbeck@google.com> wrote:

> If an smbus transfer fails, there's no guarantee that the output
> buffer was written.  So, avoid trying to show the output buffer when
> tracing after an error.  This was 'mostly harmless', but would trip
> up kasan checking if left-over cruft in byte 0 is a large length,
> causing us to read from unwritten memory.

This looks fine to me, but I'm not sure how the i2c maintainers feel,
but I always require that a new patch version starts a new thread, and
not be part of a older thread release (causes these patches to be
hidden from those that read patches in threading mode).

-- Steve

> 
> Signed-off-by: John Sperbeck <jsperbeck@google.com>
> ---
>  drivers/i2c/i2c-core-smbus.c | 2 +-
>  include/trace/events/smbus.h | 6 +++---
>  2 files changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/i2c/i2c-core-smbus.c b/drivers/i2c/i2c-core-smbus.c
> index 9cd66cabb84f..132119112596 100644
> --- a/drivers/i2c/i2c-core-smbus.c
> +++ b/drivers/i2c/i2c-core-smbus.c
> @@ -585,7 +585,7 @@ s32 __i2c_smbus_xfer(struct i2c_adapter *adapter, u16 addr,
>  trace:
>  	/* If enabled, the reply tracepoint is conditional on read_write. */
>  	trace_smbus_reply(adapter, addr, flags, read_write,
> -			  command, protocol, data);
> +			  command, protocol, data, res);
>  	trace_smbus_result(adapter, addr, flags, read_write,
>  			   command, protocol, res);
>  
> diff --git a/include/trace/events/smbus.h b/include/trace/events/smbus.h
> index d2fb6e1d3e10..a4892a187842 100644
> --- a/include/trace/events/smbus.h
> +++ b/include/trace/events/smbus.h
> @@ -138,9 +138,9 @@ TRACE_EVENT_CONDITION(smbus_reply,
>  	TP_PROTO(const struct i2c_adapter *adap,
>  		 u16 addr, unsigned short flags,
>  		 char read_write, u8 command, int protocol,
> -		 const union i2c_smbus_data *data),
> -	TP_ARGS(adap, addr, flags, read_write, command, protocol, data),
> -	TP_CONDITION(read_write == I2C_SMBUS_READ),
> +		 const union i2c_smbus_data *data, int res),
> +	TP_ARGS(adap, addr, flags, read_write, command, protocol, data, res),
> +	TP_CONDITION(res >= 0 && read_write == I2C_SMBUS_READ),
>  	TP_STRUCT__entry(
>  		__field(int,	adapter_nr		)
>  		__field(__u16,	addr			)

On Mon, Feb 11, 2019 at 6:24 AM Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Sun, 10 Feb 2019 19:40:21 -0800
> John Sperbeck <jsperbeck@google.com> wrote:
>
> > If an smbus transfer fails, there's no guarantee that the output
> > buffer was written.  So, avoid trying to show the output buffer when
> > tracing after an error.  This was 'mostly harmless', but would trip
> > up kasan checking if left-over cruft in byte 0 is a large length,
> > causing us to read from unwritten memory.
>
> This looks fine to me, but I'm not sure how the i2c maintainers feel,
> but I always require that a new patch version starts a new thread, and
> not be part of a older thread release (causes these patches to be
> hidden from those that read patches in threading mode).

That sounds reasonable to me.  Unless I hear otherwise, I'll create a
v3 as a new thread.  Thanks.