[Beignet] [PATCH] Fix timestamp on HASWELL

[He Junyan]

OK, This patch LGTM

On Mon, 2014-05-26 at 19:25 +0800, Li Peng wrote:
> The GPU timestamp should be lower 36 bit on HASWELL
> 
> Signed-off-by: Li Peng <peng.li at intel.com>
> ---
>  src/cl_driver.h         |  2 +-
>  src/cl_event.c          |  4 ++--
>  src/intel/intel_gpgpu.c | 26 +++++++++++++++++---------
>  3 files changed, 20 insertions(+), 12 deletions(-)
> 
> diff --git a/src/cl_driver.h b/src/cl_driver.h
> index 9dc2330..3e01c92 100644
> --- a/src/cl_driver.h
> +++ b/src/cl_driver.h
> @@ -193,7 +193,7 @@ typedef void (cl_gpgpu_event_delete_cb)(cl_gpgpu_event);
>  extern cl_gpgpu_event_delete_cb *cl_gpgpu_event_delete;
>  
>  /* Get a event time stamp */
> -typedef void (cl_gpgpu_event_get_exec_timestamp_cb)(cl_gpgpu_event, int, uint64_t*);
> +typedef void (cl_gpgpu_event_get_exec_timestamp_cb)(cl_gpgpu, cl_gpgpu_event, int, uint64_t*);
>  extern cl_gpgpu_event_get_exec_timestamp_cb *cl_gpgpu_event_get_exec_timestamp;
>  
>  /* Get current GPU time stamp */
> diff --git a/src/cl_event.c b/src/cl_event.c
> index 727ee1f..30e0e06 100644
> --- a/src/cl_event.c
> +++ b/src/cl_event.c
> @@ -514,11 +514,11 @@ cl_int cl_event_get_timestamp(cl_event event, cl_profiling_info param_name)
>      event->timestamp[param_name - CL_PROFILING_COMMAND_QUEUED] = ret_val;
>      return CL_SUCCESS;
>    } else if(param_name == CL_PROFILING_COMMAND_START) {
> -    cl_gpgpu_event_get_exec_timestamp(event->gpgpu_event, 0, &ret_val);
> +    cl_gpgpu_event_get_exec_timestamp(gpgpu, event->gpgpu_event, 0, &ret_val);
>      event->timestamp[param_name - CL_PROFILING_COMMAND_QUEUED] = ret_val;
>      return CL_SUCCESS;
>    } else if (param_name == CL_PROFILING_COMMAND_END) {
> -    cl_gpgpu_event_get_exec_timestamp(event->gpgpu_event, 1, &ret_val);
> +    cl_gpgpu_event_get_exec_timestamp(gpgpu, event->gpgpu_event, 1, &ret_val);
>      event->timestamp[param_name - CL_PROFILING_COMMAND_QUEUED] = ret_val;
>      return CL_SUCCESS;
>    }
> diff --git a/src/intel/intel_gpgpu.c b/src/intel/intel_gpgpu.c
> index b7b712f..2ab2bb7 100644
> --- a/src/intel/intel_gpgpu.c
> +++ b/src/intel/intel_gpgpu.c
> @@ -1110,8 +1110,12 @@ intel_gpgpu_event_get_gpu_cur_timestamp(intel_gpgpu_t* gpgpu, uint64_t* ret_ts)
>    drm_intel_bufmgr *bufmgr = gpgpu->drv->bufmgr;
>  
>    drm_intel_reg_read(bufmgr, TIMESTAMP_ADDR, &result);
> -  result = result & 0xFFFFFFFFF0000000;
> -  result = result >> 28;
> +  if (IS_HASWELL(gpgpu->drv->device_id)) {
> +    result = result & 0x0000000FFFFFFFFF;
> +  } else {
> +    result = result & 0xFFFFFFFFF0000000;
> +    result = result >> 28;
> +  }
>    result *= 80;
>  
>    *ret_ts = result;
> @@ -1120,8 +1124,8 @@ intel_gpgpu_event_get_gpu_cur_timestamp(intel_gpgpu_t* gpgpu, uint64_t* ret_ts)
>  
>  /* Get the GPU execute time. */
>  static void
> -intel_gpgpu_event_get_exec_timestamp(intel_event_t *event,
> -                                int index, uint64_t* ret_ts)
> +intel_gpgpu_event_get_exec_timestamp(intel_gpgpu_t* gpgpu, intel_event_t *event,
> +				     int index, uint64_t* ret_ts)
>  {
>    uint64_t result = 0;
>  
> @@ -1131,11 +1135,15 @@ intel_gpgpu_event_get_exec_timestamp(intel_event_t *event,
>    uint64_t* ptr = event->ts_buf->virtual;
>    result = ptr[index];
>  
> -  /* According to BSpec, the timestamp counter should be 36 bits,
> -     but comparing to the timestamp counter from IO control reading,
> -     we find the first 4 bits seems to be fake. In order to keep the
> -     timestamp counter conformable, we just skip the first 4 bits. */
> -  result = ((result & 0x0FFFFFFFF) << 4) * 80; //convert to nanoseconds
> +  if (IS_HASWELL(gpgpu->drv->device_id))
> +    result = (result & 0xFFFFFFFFF) * 80; //convert to nanoseconds
> +  else
> +    /* According to BSpec, the timestamp counter should be 36 bits,
> +       but comparing to the timestamp counter from IO control reading,
> +       we find the first 4 bits seems to be fake. In order to keep the
> +       timestamp counter conformable, we just skip the first 4 bits.
> +     */
> +    result = ((result & 0x0FFFFFFFF) << 4) * 80; //convert to nanoseconds
>    *ret_ts = result;
>  
>    drm_intel_gem_bo_unmap_gtt(event->ts_buf);