Messages in this thread | | | Date | Tue, 23 Nov 2010 09:13:05 -0500 | From | Don Zickus <> | Subject | Re: [PATCH] x86, perf, nmi: Disable perf if counters are not accessable |
| |
On Tue, Nov 23, 2010 at 11:46:06AM +0100, Peter Zijlstra wrote: > On Mon, 2010-11-22 at 16:55 -0500, Don Zickus wrote: > > +static bool check_hw_exists(void) > > +{ > > + u64 val, val_new; > > + > > + val = 0xabcdUL; > > + (void) checking_wrmsrl(x86_pmu.perfctr, val); > > + rdmsrl_safe(x86_pmu.perfctr, &val_new); > > + if (val != val_new) > > + return false; > > + > > + return true; > > +} > > If I can make any sense of the implementation of native_read_msr_safe() > then I think it doesn't actually sets val_new in case it faults, it just > returns -EIO. > > So I changed it to: > > static bool check_hw_exists(void) > { > u64 val, val_new = 0; > int ret = 0; > > val = 0xabcdUL; > ret |= checking_wrmsrl(x86_pmu.perfctr, val); > ret |= rdmsrl_safe(x86_pmu.perfctr, &val_new); > if (ret || val != val_new) > return false; > > return true; > } > > And have applied the patch,
Looks good to me. Thanks!
Cheers, Don
| |