Messages in this thread | | | Date | Fri, 1 Jun 2012 11:35:48 +0200 | Subject | Re: [PATCH] perf: Fix intel shared extra msr allocation | From | Stephane Eranian <> |
| |
On Fri, Jun 1, 2012 at 5:20 AM, Yan, Zheng <zheng.z.yan@intel.com> wrote: > From: "Yan, Zheng" <zheng.z.yan@intel.com> > > intel_shared_reg_get/put_constraints() can be indirectly called > by validate_group(). In that case, they should avoid modifying > the perf_event date structure because the event can be already > in active state. Otherwise the shared extra msr's reference > count will be left in inconsistent state. > I understand the problem but I am wondering if you actually saw it in real life. The reason I am asking is because of the way validate_group() collects the events and how they are added to sibling_list. The new event is added at the tail. Thus it will come last, and will get to __intel_shared_reg_get_constraints() last, thus I am wondering if it can really modify the programming on the existing events.
See more comments inline.
> Signed-off-by: Zheng Yan <zheng.z.yan@intel.com> > --- > arch/x86/kernel/cpu/perf_event_intel.c | 31 +++++++++++++++++++++++-------- > 1 files changed, 23 insertions(+), 8 deletions(-) > > diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c > index 166546e..10840d0 100644 > --- a/arch/x86/kernel/cpu/perf_event_intel.c > +++ b/arch/x86/kernel/cpu/perf_event_intel.c > @@ -1119,11 +1119,21 @@ intel_bts_constraints(struct perf_event *event) > return NULL; > } > > -static bool intel_try_alt_er(struct perf_event *event, int orig_idx) > +static bool intel_try_alt_er(struct perf_event *event, int *idx, > + int orig_idx, bool fake_cpuc) > { > - if (!(x86_pmu.er_flags & ERF_HAS_RSP_1)) > + if (!(x86_pmu.er_flags & ERF_HAS_RSP_1) || *idx != orig_idx) > return false; > > + /* don't modify the event structure if the cpuc is faked */ > + if (fake_cpuc) { > + if (*idx == EXTRA_REG_RSP_0) > + *idx = EXTRA_REG_RSP_1; > + else if (*idx == EXTRA_REG_RSP_1) > + *idx = EXTRA_REG_RSP_0; > + return (*idx != orig_idx); > + } > + I understand that.
> if (event->hw.extra_reg.idx == EXTRA_REG_RSP_0) { > event->hw.config &= ~INTEL_ARCH_EVENT_MASK; > event->hw.config |= 0x01bb; > @@ -1139,6 +1149,7 @@ static bool intel_try_alt_er(struct perf_event *event, int orig_idx) > if (event->hw.extra_reg.idx == orig_idx) > return false; > > + *idx = event->hw.extra_reg.idx; > return true; > } > > @@ -1155,16 +1166,18 @@ __intel_shared_reg_get_constraints(struct cpu_hw_events *cpuc, > struct hw_perf_event_extra *reg) > { > struct event_constraint *c = &emptyconstraint; > + struct intel_shared_regs *shared_regs = cpuc->shared_regs; > struct er_account *era; > unsigned long flags; > int orig_idx = reg->idx; > + int idx = orig_idx; > > - /* already allocated shared msr */ > - if (reg->alloc) > + /* shared msr is already allocated and cpuc is not faked */ > + if (reg->alloc && shared_regs->core_id != -1) > return NULL; /* call x86_get_event_constraint() */ > I don't understand what you need this stuff. Shared_regs is faked as well.
> again: > - era = &cpuc->shared_regs->regs[reg->idx]; > + era = &shared_regs->regs[idx]; > /* > * we use spin_lock_irqsave() to avoid lockdep issues when > * passing a fake cpuc > @@ -1181,14 +1194,16 @@ __intel_shared_reg_get_constraints(struct cpu_hw_events *cpuc, > atomic_inc(&era->ref); > > /* no need to reallocate during incremental event scheduling */ > - reg->alloc = 1; > + if (shared_regs->core_id != -1) > + reg->alloc = 1; > > /* > * need to call x86_get_event_constraint() > * to check if associated event has constraints > */ > c = NULL; > - } else if (intel_try_alt_er(event, orig_idx)) { > + } else if (intel_try_alt_er(event, &idx, orig_idx, > + shared_regs->core_id == -1)) { > raw_spin_unlock_irqrestore(&era->lock, flags); > goto again; > } > @@ -1208,7 +1223,7 @@ __intel_shared_reg_put_constraints(struct cpu_hw_events *cpuc, > * allocated. Also takes care of event which do > * not use an extra shared reg > */ > - if (!reg->alloc) > + if (!reg->alloc || cpuc->shared_regs->core_id == -1) > return; > > era = &cpuc->shared_regs->regs[reg->idx]; > -- > 1.7.7.6 > -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |