From testing on sc7180-trogdor devices, reading the GMU registers
needs the GMU clocks to be enabled. Those clocks get turned on in
a6xx_gmu_resume(). Confusingly enough, that function is called as a
result of the runtime_pm of the GPU "struct device", not the GMU
"struct device". Unfortunately the current a6xx_gpu_busy() grabs a
reference to the GMU's "struct device".
The fact that we were grabbing the wrong reference was easily seen to
cause crashes that happen if we change the GPU's pm_runtime usage to
not use autosuspend. It's also believed to cause some long tail GPU
crashes even with autosuspend.
We could look at changing it so that we do pm_runtime_get_if_in_use()
on the GPU's "struct device", but then we run into a different
problem. pm_runtime_get_if_in_use() will return 0 for the GPU's
"struct device" the whole time when we're in the "autosuspend
delay". That is, when we drop the last reference to the GPU but we're
waiting a period before actually suspending then we'll think the GPU
is off. One reason that's bad is that if the GPU didn't actually turn
off then the cycle counter doesn't lose state and that throws off all
of our calculations.
Let's change the code to keep track of the suspend state of
devfreq. msm_devfreq_suspend() is always called before we actually
suspend the GPU and msm_devfreq_resume() after we resume it. This
means we can use the suspended state to know if we're powered or not.
NOTE: one might wonder when exactly our status function is called when
devfreq is supposed to be disabled. The stack crawl I captured was:
msm_devfreq_get_dev_status
devfreq_simple_ondemand_func
devfreq_update_target
qos_notifier_call
qos_max_notifier_call
blocking_notifier_call_chain
pm_qos_update_target
freq_qos_apply
apply_constraint
__dev_pm_qos_update_request
dev_pm_qos_update_request
msm_devfreq_idle_work
Fixes: eadf79286a4b ("drm/msm: Check for powered down HW in the devfreq callbacks")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Patchwork: https://patchwork.freedesktop.org/patch/489124/
Link: https://lore.kernel.org/r/20220610124639.v4.1.Ie846c5352bc307ee4248d7cab998ab3016b85d06@changeid
Signed-off-by: Rob Clark <robdclark@chromium.org>
{
u64 busy_cycles;
- /* Only read the gpu busy if the hardware is already active */
- if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) {
- *out_sample_rate = 1;
- return 0;
- }
-
busy_cycles = gpu_read64(gpu, REG_A5XX_RBBM_PERFCTR_RBBM_0_LO,
REG_A5XX_RBBM_PERFCTR_RBBM_0_HI);
*out_sample_rate = clk_get_rate(gpu->core_clk);
- pm_runtime_put(&gpu->pdev->dev);
-
return busy_cycles;
}
A6XX_GMU_SPTPRAC_PWR_CLK_STATUS_GX_HM_CLK_OFF));
}
-void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp)
+void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp,
+ bool suspended)
{
struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu);
struct a6xx_gpu *a6xx_gpu = to_a6xx_gpu(adreno_gpu);
/*
* This can get called from devfreq while the hardware is idle. Don't
- * bring up the power if it isn't already active
+ * bring up the power if it isn't already active. All we're doing here
+ * is updating the frequency so that when we come back online we're at
+ * the right rate.
*/
- if (pm_runtime_get_if_in_use(gmu->dev) == 0)
+ if (suspended)
return;
if (!gmu->legacy) {
a6xx_hfi_set_freq(gmu, perf_index);
dev_pm_opp_set_opp(&gpu->pdev->dev, opp);
- pm_runtime_put(gmu->dev);
return;
}
dev_err(gmu->dev, "GMU set GPU frequency error: %d\n", ret);
dev_pm_opp_set_opp(&gpu->pdev->dev, opp);
- pm_runtime_put(gmu->dev);
}
unsigned long a6xx_gmu_get_freq(struct msm_gpu *gpu)
return;
gmu->freq = 0; /* so a6xx_gmu_set_freq() doesn't exit early */
- a6xx_gmu_set_freq(gpu, gpu_opp);
+ a6xx_gmu_set_freq(gpu, gpu_opp, false);
dev_pm_opp_put(gpu_opp);
}
/* 19.2MHz */
*out_sample_rate = 19200000;
- /* Only read the gpu busy if the hardware is already active */
- if (pm_runtime_get_if_in_use(a6xx_gpu->gmu.dev) == 0)
- return 0;
-
busy_cycles = gmu_read64(&a6xx_gpu->gmu,
REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_L,
REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_H);
-
- pm_runtime_put(a6xx_gpu->gmu.dev);
-
return busy_cycles;
}
-static void a6xx_gpu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp)
+static void a6xx_gpu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp,
+ bool suspended)
{
struct adreno_gpu *adreno_gpu = to_adreno_gpu(gpu);
struct a6xx_gpu *a6xx_gpu = to_a6xx_gpu(adreno_gpu);
mutex_lock(&a6xx_gpu->gmu.lock);
- a6xx_gmu_set_freq(gpu, opp);
+ a6xx_gmu_set_freq(gpu, opp, suspended);
mutex_unlock(&a6xx_gpu->gmu.lock);
}
int a6xx_gmu_init(struct a6xx_gpu *a6xx_gpu, struct device_node *node);
void a6xx_gmu_remove(struct a6xx_gpu *a6xx_gpu);
-void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp);
+void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp,
+ bool suspended);
unsigned long a6xx_gmu_get_freq(struct msm_gpu *gpu);
void a6xx_show(struct msm_gpu *gpu, struct msm_gpu_state *state,
/* for generation specific debugfs: */
void (*debugfs_init)(struct msm_gpu *gpu, struct drm_minor *minor);
#endif
+ /* note: gpu_busy() can assume that we have been pm_resumed */
u64 (*gpu_busy)(struct msm_gpu *gpu, unsigned long *out_sample_rate);
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);
- void (*gpu_set_freq)(struct msm_gpu *gpu, struct dev_pm_opp *opp);
+ /* note: gpu_set_freq() can assume that we have been pm_resumed */
+ void (*gpu_set_freq)(struct msm_gpu *gpu, struct dev_pm_opp *opp,
+ bool suspended);
struct msm_gem_address_space *(*create_address_space)
(struct msm_gpu *gpu, struct platform_device *pdev);
struct msm_gem_address_space *(*create_private_address_space)
/** devfreq: devfreq instance */
struct devfreq *devfreq;
+ /** lock: lock for "suspended", "busy_cycles", and "time" */
+ struct mutex lock;
+
/**
* idle_constraint:
*
* elapsed
*/
struct msm_hrtimer_work boost_work;
+
+ /** suspended: tracks if we're suspended */
+ bool suspended;
};
struct msm_gpu {
u32 flags)
{
struct msm_gpu *gpu = dev_to_gpu(dev);
+ struct msm_gpu_devfreq *df = &gpu->devfreq;
struct dev_pm_opp *opp;
/*
trace_msm_gpu_freq_change(dev_pm_opp_get_freq(opp));
- if (gpu->funcs->gpu_set_freq)
- gpu->funcs->gpu_set_freq(gpu, opp);
- else
+ if (gpu->funcs->gpu_set_freq) {
+ mutex_lock(&df->lock);
+ gpu->funcs->gpu_set_freq(gpu, opp, df->suspended);
+ mutex_unlock(&df->lock);
+ } else {
clk_set_rate(gpu->core_clk, *freq);
+ }
dev_pm_opp_put(opp);
unsigned long sample_rate;
ktime_t time;
+ mutex_lock(&df->lock);
+
status->current_frequency = get_freq(gpu);
- busy_cycles = gpu->funcs->gpu_busy(gpu, &sample_rate);
time = ktime_get();
-
- busy_time = busy_cycles - df->busy_cycles;
status->total_time = ktime_us_delta(time, df->time);
+ df->time = time;
+ if (df->suspended) {
+ mutex_unlock(&df->lock);
+ status->busy_time = 0;
+ return;
+ }
+
+ busy_cycles = gpu->funcs->gpu_busy(gpu, &sample_rate);
+ busy_time = busy_cycles - df->busy_cycles;
df->busy_cycles = busy_cycles;
- df->time = time;
+
+ mutex_unlock(&df->lock);
busy_time *= USEC_PER_SEC;
busy_time = div64_ul(busy_time, sample_rate);
if (!gpu->funcs->gpu_busy)
return;
+ mutex_init(&df->lock);
+
dev_pm_qos_add_request(&gpu->pdev->dev, &df->idle_freq,
DEV_PM_QOS_MAX_FREQUENCY,
PM_QOS_MAX_FREQUENCY_DEFAULT_VALUE);
void msm_devfreq_resume(struct msm_gpu *gpu)
{
struct msm_gpu_devfreq *df = &gpu->devfreq;
+ unsigned long sample_rate;
if (!has_devfreq(gpu))
return;
- df->busy_cycles = 0;
+ mutex_lock(&df->lock);
+ df->busy_cycles = gpu->funcs->gpu_busy(gpu, &sample_rate);
df->time = ktime_get();
+ df->suspended = false;
+ mutex_unlock(&df->lock);
devfreq_resume_device(df->devfreq);
}
if (!has_devfreq(gpu))
return;
+ mutex_lock(&df->lock);
+ df->suspended = true;
+ mutex_unlock(&df->lock);
+
devfreq_suspend_device(df->devfreq);
cancel_idle_work(df);
projects / mirror_ubuntu-kernels.git / commitdiff