On Mon, Mar 26, 2018 at 07:32:26AM -0700, Tejun Heo wrote:
> On Fri, Mar 23, 2018 at 11:46:12AM +0300, Dan Carpenter wrote:
> > Hello Tejun Heo,
> >
> > The patch 38b78eb85540: "tracing: Factorize filter creation" from Dec
> > 15, 2011, leads to the following static checker warning:
> >
> > kernel/trace/trace_events_filter.c:1794 apply_event_filter()
> > error: uninitialized symbol 'filter'.
> >
> > kernel/trace/trace_events_filter.c
> > 1764 int apply_event_filter(struct trace_event_file *file, char *filter_string)
> ...
> > 1785
> > 1786 err = create_filter(call, filter_string, true, &filter);
> > ^^^^^^^
> > 1787
> > 1788 /*
> > 1789 * Always swap the call filter with the new filter
> > 1790 * even if there was an error. If there was an error
> > 1791 * in the filter, we disable the filter and show the error
> > 1792 * string
> > 1793 */
> > 1794 if (filter) {
> > ^^^^^^
> > I guess the fix is probably to set filter to NULL in create_filter()?
>
> @filterp is the outparam and create_filter() always sets it, so the
> code doesn't look wrong to me.
It's not set if create_filter_start() fails at the start of
create_filter(). I'm looking at today's linux-next.
1686 /**
1687 * create_filter - create a filter for a trace_event_call
1688 * @call: trace_event_call to create a filter for
1689 * @filter_str: filter string
1690 * @set_str: remember @filter_str and enable detailed error in filter
1691 * @filterp: out param for created filter (always updated on return)
^^^^^^^^^^^^^^^^^^^^^^^^
Ah... You're right about the comment.
1692 *
1693 * Creates a filter for @call with @filter_str. If @set_str is %true,
1694 * @filter_str is copied and recorded in the new filter.
1695 *
1696 * On success, returns 0 and *@filterp points to the new filter. On
1697 * failure, returns -errno and *@filterp may point to %NULL or to a new
1698 * filter. In the latter case, the returned filter contains error
1699 * information if @set_str is %true and the caller is responsible for
1700 * freeing it.
1701 */
1702 static int create_filter(struct trace_event_call *call,
1703 char *filter_string, bool set_str,
1704 struct event_filter **filterp)
1705 {
1706 struct filter_parse_error *pe = NULL;
1707 struct event_filter *filter = NULL;
1708 int err;
1709
1710 err = create_filter_start(filter_string, set_str, &pe, &filter);
1711 if (err)
1712 return err;
^^^^^^^^^^
But it's not set here.
1713
1714 err = process_preds(call, filter_string, filter, pe);
1715 if (err && set_str)
1716 append_filter_err(pe, filter);
1717
1718 *filterp = filter;
1719 return err;
1720 }
regards,
dan carpenter
--
To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
