Messages in this thread | | | Date | Wed, 27 Jun 2012 03:05:13 +0200 | Subject | Re: [PATCH V5 0/13] perf: Intel uncore pmu counting support | From | Stephane Eranian <> |
| |
Hi,
If you compile the uncore support in 32-bit mode, you will get a warning on the hrtimer_start_range_ns() functions because the interval is passed as ktime_t whereas the function expects unsigned long. With 64-bit, no problem ktime_t is a union with s64. But in 32-bit mode, there is a possible truncation of the delta. This needs to be fixed.
On Thu, Jun 14, 2012 at 7:41 AM, Stephane Eranian <eranian@google.com> wrote: > On Thu, Jun 14, 2012 at 4:18 AM, Yan, Zheng <zheng.z.yan@intel.com> wrote: >> On 06/13/2012 05:02 PM, Peter Zijlstra wrote: >>> On Wed, 2012-06-13 at 09:41 +0800, Yan, Zheng wrote: >>>> Peter suggests keeping the uncore names as they're listed in the intel >>>> doc. For Sandybirdge-EP, uncore names are something like: Cbo, iMC, >>>> QPI. >>> >>> No they're not, they're C-Box etc.. but I'm fine with doing a tolower on >>> all of it. >>> >> >> The reason I choose CBox instead of C-Box is that '-' is a separate symbol >> in the flex rules. '-' is used for matching events such as LLC-load-misses. >> I don't know how to allow letter '-' in the pmu name, but without leading >> to ambiguity. >> > I would drop the -, just call it cbox.
| |