----- "Jeff Moyer" <jmoyer@xxxxxxxxxx> wrote:
> Bernhard Walle <bwalle@xxxxxxx> writes:
>
> > That should also go mainline.
> >
> >
> > Signed-off-by : Sachin Sant <sachinp@xxxxxxxxxx>
> > Acked-by: Bernhard Walle <bwalle@xxxxxxx>
> >
> >
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> > task.c | 2 +-
> >
> >
> > That should also go mainline.
> >
> >
> > Signed-off-by : Sachin Sant <sachinp@xxxxxxxxxx>
> > Acked-by: Bernhard Walle <bwalle@xxxxxxx>
> >
> > diff --git a/task.c b/task.c
> > --- a/task.c
> > +++ b/task.c
> > @@ -2902,7 +2902,7 @@
> > {
> > int i, c;
> > struct task_context *tcp;
> > - char format[10];
> > + char format[15];
>
>
> c = strlen(buf);
> sprintf(format, "[%c%dll%c] ", '%', c,
> pc->output_radix == 10 ? 'u' : 'x');
>
> Looks like it should be 11, no? You have 6 characters + '\0' + the %d,
> which can be up to 1500 (BUFSIZE). Of course, it wouldn't kill us to
> use snprintf, either.
You're right, it can only be a maximum of 11 bytes, i.e., either "[%20llu] "
(decimal) or "[%16llx] " (hex), plus their trailing NULLs. It's apparently
worked OK by dumb luck by putting the NULL either in the stack variable location
of the preceding "tcp" pointer or the "buf[BUFSIZE]" array (I guess???), which
as it turns out, is safe to do.
Anyway I'll bump it up...
Thanks,
Dave
--
Crash-utility mailing list
Crash-utility@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/crash-utility
