| Message ID | 1539781441-13076-2-git-send-email-smasetty@codeaurora.org (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Headers | show |
| Series | drm/msm: Revive GPU workload profiling | expand |
On Wed, Oct 17, 2018 at 06:33:59PM +0530, Sharat Masetty wrote: > The gpu_busy() function turns out is useful for performance profiling as > well. So we change the input params to pass in the pointer to the previous > busy cycles, this makes the function more generic and removes the dependency > on the gpu devfreq. This patch is good in theory. > Signed-off-by: Sharat Masetty <smasetty@codeaurora.org> > --- > drivers/gpu/drm/msm/adreno/a5xx_gpu.c | 6 +++--- > drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 6 +++--- > drivers/gpu/drm/msm/msm_gpu.c | 3 ++- > drivers/gpu/drm/msm/msm_gpu.h | 2 +- > 4 files changed, 9 insertions(+), 8 deletions(-) > > diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c > index 236a6ed..b5df80c 100644 > --- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c > +++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c > @@ -1428,7 +1428,7 @@ static struct msm_ringbuffer *a5xx_active_ring(struct msm_gpu *gpu) > return a5xx_gpu->cur_ring; > } > > -static unsigned long a5xx_gpu_busy(struct msm_gpu *gpu) > +static unsigned long a5xx_gpu_busy(struct msm_gpu *gpu, u64 *prev_cycles) > { > u64 busy_cycles; > unsigned long busy_time; > @@ -1436,10 +1436,10 @@ static unsigned long a5xx_gpu_busy(struct msm_gpu *gpu) > busy_cycles = gpu_read64(gpu, REG_A5XX_RBBM_PERFCTR_RBBM_0_LO, > REG_A5XX_RBBM_PERFCTR_RBBM_0_HI); > > - busy_time = (busy_cycles - gpu->devfreq.busy_cycles) / > + busy_time = (busy_cycles - *prev_cycles) / > (clk_get_rate(gpu->core_clk) / 1000000); > > - gpu->devfreq.busy_cycles = busy_cycles; > + *prev_cycles = busy_cycles; > > return busy_time; > } > diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > index 07c3c3c..c0cd3ac 100644 > --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > @@ -722,7 +722,7 @@ static void a6xx_destroy(struct msm_gpu *gpu) > kfree(a6xx_gpu); > } > > -static unsigned long a6xx_gpu_busy(struct msm_gpu *gpu) > +static unsigned long a6xx_gpu_busy(struct msm_gpu *gpu, u64 *prev_cycles) > { > struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); > struct a6xx_gpu *a6xx_gpu = to_a6xx_gpu(adreno_gpu); > @@ -733,9 +733,9 @@ static unsigned long a6xx_gpu_busy(struct msm_gpu *gpu) > REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_L, > REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_H); > > - busy_time = ((busy_cycles - gpu->devfreq.busy_cycles) * 10) / 192; > + busy_time = ((busy_cycles - *prev_cycles) * 10) / 192; You must be using older code - this is missing the fixes from Sean Paul. > - gpu->devfreq.busy_cycles = busy_cycles; > + *prev_cycles = busy_cycles; > > return busy_time; > } > diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c > index ca573f6..e9b5426 100644 > --- a/drivers/gpu/drm/msm/msm_gpu.c > +++ b/drivers/gpu/drm/msm/msm_gpu.c > @@ -62,7 +62,8 @@ static int msm_devfreq_get_dev_status(struct device *dev, > else > status->current_frequency = clk_get_rate(gpu->core_clk); > > - status->busy_time = gpu->funcs->gpu_busy(gpu); > + status->busy_time = gpu->funcs->gpu_busy(gpu, > + &gpu->devfreq.busy_cycles); > > time = ktime_get(); > status->total_time = ktime_us_delta(time, gpu->devfreq.time); > diff --git a/drivers/gpu/drm/msm/msm_gpu.h b/drivers/gpu/drm/msm/msm_gpu.h > index 9df48e3..0ff23ca 100644 > --- a/drivers/gpu/drm/msm/msm_gpu.h > +++ b/drivers/gpu/drm/msm/msm_gpu.h > @@ -71,7 +71,7 @@ struct msm_gpu_funcs { > /* for generation specific debugfs: */ > int (*debugfs_init)(struct msm_gpu *gpu, struct drm_minor *minor); > #endif > - unsigned long (*gpu_busy)(struct msm_gpu *gpu); > + unsigned long (*gpu_busy)(struct msm_gpu *gpu, u64 *busy_cycles); > struct msm_gpu_state *(*gpu_state_get)(struct msm_gpu *gpu); > int (*gpu_state_put)(struct msm_gpu_state *state); > unsigned long (*gpu_get_freq)(struct msm_gpu *gpu); > -- > 1.9.1 >
diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c index 236a6ed..b5df80c 100644 --- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c @@ -1428,7 +1428,7 @@ static struct msm_ringbuffer *a5xx_active_ring(struct msm_gpu *gpu) return a5xx_gpu->cur_ring; } -static unsigned long a5xx_gpu_busy(struct msm_gpu *gpu) +static unsigned long a5xx_gpu_busy(struct msm_gpu *gpu, u64 *prev_cycles) { u64 busy_cycles; unsigned long busy_time; @@ -1436,10 +1436,10 @@ static unsigned long a5xx_gpu_busy(struct msm_gpu *gpu) busy_cycles = gpu_read64(gpu, REG_A5XX_RBBM_PERFCTR_RBBM_0_LO, REG_A5XX_RBBM_PERFCTR_RBBM_0_HI); - busy_time = (busy_cycles - gpu->devfreq.busy_cycles) / + busy_time = (busy_cycles - *prev_cycles) / (clk_get_rate(gpu->core_clk) / 1000000); - gpu->devfreq.busy_cycles = busy_cycles; + *prev_cycles = busy_cycles; return busy_time; } diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c index 07c3c3c..c0cd3ac 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c @@ -722,7 +722,7 @@ static void a6xx_destroy(struct msm_gpu *gpu) kfree(a6xx_gpu); } -static unsigned long a6xx_gpu_busy(struct msm_gpu *gpu) +static unsigned long a6xx_gpu_busy(struct msm_gpu *gpu, u64 *prev_cycles) { struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu); struct a6xx_gpu *a6xx_gpu = to_a6xx_gpu(adreno_gpu); @@ -733,9 +733,9 @@ static unsigned long a6xx_gpu_busy(struct msm_gpu *gpu) REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_L, REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_H); - busy_time = ((busy_cycles - gpu->devfreq.busy_cycles) * 10) / 192; + busy_time = ((busy_cycles - *prev_cycles) * 10) / 192; - gpu->devfreq.busy_cycles = busy_cycles; + *prev_cycles = busy_cycles; return busy_time; } diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c index ca573f6..e9b5426 100644 --- a/drivers/gpu/drm/msm/msm_gpu.c +++ b/drivers/gpu/drm/msm/msm_gpu.c @@ -62,7 +62,8 @@ static int msm_devfreq_get_dev_status(struct device *dev, else status->current_frequency = clk_get_rate(gpu->core_clk); - status->busy_time = gpu->funcs->gpu_busy(gpu); + status->busy_time = gpu->funcs->gpu_busy(gpu, + &gpu->devfreq.busy_cycles); time = ktime_get(); status->total_time = ktime_us_delta(time, gpu->devfreq.time); diff --git a/drivers/gpu/drm/msm/msm_gpu.h b/drivers/gpu/drm/msm/msm_gpu.h index 9df48e3..0ff23ca 100644 --- a/drivers/gpu/drm/msm/msm_gpu.h +++ b/drivers/gpu/drm/msm/msm_gpu.h @@ -71,7 +71,7 @@ struct msm_gpu_funcs { /* for generation specific debugfs: */ int (*debugfs_init)(struct msm_gpu *gpu, struct drm_minor *minor); #endif - unsigned long (*gpu_busy)(struct msm_gpu *gpu); + unsigned long (*gpu_busy)(struct msm_gpu *gpu, u64 *busy_cycles); struct msm_gpu_state *(*gpu_state_get)(struct msm_gpu *gpu); int (*gpu_state_put)(struct msm_gpu_state *state); unsigned long (*gpu_get_freq)(struct msm_gpu *gpu);
The gpu_busy() function turns out is useful for performance profiling as well. So we change the input params to pass in the pointer to the previous busy cycles, this makes the function more generic and removes the dependency on the gpu devfreq. Signed-off-by: Sharat Masetty <smasetty@codeaurora.org> --- drivers/gpu/drm/msm/adreno/a5xx_gpu.c | 6 +++--- drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 6 +++--- drivers/gpu/drm/msm/msm_gpu.c | 3 ++- drivers/gpu/drm/msm/msm_gpu.h | 2 +- 4 files changed, 9 insertions(+), 8 deletions(-)