Re: [PATCH]tracing: Fix inaccurate function definition in trace_events_filter.c

[Li Zefan <lizf@cn.fujitsu.com>]

Wenji Huang wrote:
> Use strcmp instead of strncmp to do full match.
> 

It's intended to use strncmp instead of strcmp..

field(char, name, size) will be treated as a STATIC_STRING by
filter code. But this field isn't necessarily a string, it
can be an array of chars, which may not be ended with '\0'.
So in this case, we can't use strcmp.

So do you really see incorrect filter results?

I think STATIC_STRING and DYN_STRING should be fine, because
the "len" parameter is not the length of the string, but
the length of the array, which including the tailing '\0'.

For FILTER_PTR_STRING, the filter is not working correctly:

  # echo 'func == vt' > events/bkl/lock_kernel/filter
  # echo 1 > events/bkl/lock_kernel/enable
  ...
  # cat trace
   Xorg-1484  [000]  1973.392586: lock_kernel: ... func=vt_ioctl()
    gpm-1402  [001]  1974.027740: lock_kernel: ... func=vt_ioctl()

Actually I already made 2 patches to fix this, and fix another
bug in regex_match_end(). I'll send them out soon.

> Signed-off-by: Wenji Huang <wenji.huang@oracle.com>
> ---
>  kernel/trace/trace_events_filter.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c
> index 50504cb..06dc376 100644
> --- a/kernel/trace/trace_events_filter.c
> +++ b/kernel/trace/trace_events_filter.c
> @@ -254,7 +254,7 @@ static int filter_pred_none(struct filter_pred *pred, void *event,
>  /* Basic regex callbacks */
>  static int regex_match_full(char *str, struct regex *r, int len)
>  {
> -	if (strncmp(str, r->pattern, len) == 0)
> +	if (strcmp(str, r->pattern) == 0)
>  		return 1;
>  	return 0;
>  }