Messages in this thread |  | | Date | Tue, 4 Mar 2014 21:27:14 -0600 (CST) | From | Christoph Lameter <> | Subject | Re: [PATCH 05/48] percpu: Add preemption checks to __this_cpu ops |
| |
On Tue, 4 Mar 2014, Andrew Morton wrote:
> > print_symbol("caller is %s\n", (long)__builtin_return_address(0)); > > dump_stack(); > > I wonder if there's any point in printing __builtin_return_address. > Doesn't dump_stack() tell us the same thing?
Yes it does. However, it was there before and software may scan the logs for it.
> > +notrace void __this_cpu_preempt_check(const char *op) > > +{ > > + char text[40]; > > + > > + snprintf(text, sizeof(text), "__this_cpu_%s()", op); > > + check_preemption_disabled(text); > > +} > > I'd like to see a comment here telling scared readers why this can > never overflow text[].
Ok. I can also add VM_BUG_ON(strlen(op) >= sizeof(text)) ?
|  |