Messages in this thread |  | | Date | Wed, 22 Oct 2014 19:25:40 +0200 | Subject | Re: [PATCH v2 05/12] perf/x86: add cross-HT counter exclusion infrastructure | From | Stephane Eranian <> |
| |
On Wed, Oct 22, 2014 at 5:07 PM, Jiri Olsa <jolsa@redhat.com> wrote: > On Thu, Oct 09, 2014 at 06:34:39PM +0200, Stephane Eranian wrote: >> From: Maria Dimakopoulou <maria.n.dimakopoulou@gmail.com> >> > > SNIP > >> +struct intel_excl_cntrs *allocate_excl_cntrs(int cpu) >> +{ >> + struct intel_excl_cntrs *c; >> + int i; >> + >> + c = kzalloc_node(sizeof(struct intel_excl_cntrs), >> + GFP_KERNEL, cpu_to_node(cpu)); >> + if (c) { >> + spin_lock_init(&c->lock); >> + for (i = 0; i < X86_PMC_IDX_MAX; i++) { >> + c->states[0].state[i] = INTEL_EXCL_UNUSED; >> + c->states[0].init_state[i] = INTEL_EXCL_UNUSED; >> + >> + c->states[1].state[i] = INTEL_EXCL_UNUSED; >> + c->states[1].init_state[i] = INTEL_EXCL_UNUSED; > > 'c' is allocated with kzalloc_node, seems there's no need > to re-initialize states it to zero again with INTEL_EXCL_UNUSED > You are assuming that INTEL_EXCL_UNUSED=0 always. That means it needs to be clear there is a dependency here.
I am happy removing that if someone confirms the assumption is common in the kernel for this kind of fields.
> jirka
|  |