| Message ID | 20220609073317.1.Ie846c5352bc307ee4248d7cab998ab3016b85d06@changeid (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | drm/msm: Grab the GPU runtime in a6xx routines, not the GMU one | expand |
On 6/9/2022 8:03 PM, Douglas Anderson wrote: > >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". > > Let's grab a reference to the correct device. Incidentally, this makes > us match the a5xx routine more closely. > > This is easily shown to fix crashes that happen if we change the GPU's > pm_runtime usage to not use autosuspend. It's also believed to fix > some long tail GPU crashes even with autosuspend. > > NOTE: the crashes I've seen were fixed by _only_ fixing > a6xx_gpu_busy(). However, I believe that the same arguments should be > made to a6xx_gmu_set_freq() so I've changed that function too. > > Fixes: eadf79286a4b ("drm/msm: Check for powered down HW in the devfreq callbacks") > Signed-off-by: Douglas Anderson <dianders@chromium.org> > --- > > drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 6 +++--- > drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 4 ++-- > 2 files changed, 5 insertions(+), 5 deletions(-) > > diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c > index 9f76f5b15759..b79ad2e0649c 100644 > --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c > +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c > @@ -129,13 +129,13 @@ void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp) > * This can get called from devfreq while the hardware is idle. Don't > * bring up the power if it isn't already active > */ > - if (pm_runtime_get_if_in_use(gmu->dev) == 0) > + if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) Wouldn't we return early here when this fn is called from a6xx_gmu_set_initial_freq()? -Akhil. > 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); > + pm_runtime_put(&gpu->pdev->dev); > return; > } > > @@ -159,7 +159,7 @@ void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp) > 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); > + pm_runtime_put(&gpu->pdev->dev); > } > > unsigned long a6xx_gmu_get_freq(struct msm_gpu *gpu) > diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > index 841e47a0b06b..87568d0b6ef8 100644 > --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > @@ -1659,7 +1659,7 @@ static u64 a6xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) > *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) > + if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) > return 0; > > busy_cycles = gmu_read64(&a6xx_gpu->gmu, > @@ -1667,7 +1667,7 @@ static u64 a6xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) > REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_H); > > > - pm_runtime_put(a6xx_gpu->gmu.dev); > + pm_runtime_put(&gpu->pdev->dev); > > return busy_cycles; > }
On Thu, Jun 9, 2022 at 7:34 AM Douglas Anderson <dianders@chromium.org> wrote: > > 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". > > Let's grab a reference to the correct device. Incidentally, this makes > us match the a5xx routine more closely. > > This is easily shown to fix crashes that happen if we change the GPU's > pm_runtime usage to not use autosuspend. It's also believed to fix > some long tail GPU crashes even with autosuspend. > > NOTE: the crashes I've seen were fixed by _only_ fixing > a6xx_gpu_busy(). However, I believe that the same arguments should be > made to a6xx_gmu_set_freq() so I've changed that function too. > > Fixes: eadf79286a4b ("drm/msm: Check for powered down HW in the devfreq callbacks") > Signed-off-by: Douglas Anderson <dianders@chromium.org> > --- > > drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 6 +++--- > drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 4 ++-- > 2 files changed, 5 insertions(+), 5 deletions(-) > > diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c > index 9f76f5b15759..b79ad2e0649c 100644 > --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c > +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c > @@ -129,13 +129,13 @@ void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp) > * This can get called from devfreq while the hardware is idle. Don't > * bring up the power if it isn't already active > */ > - if (pm_runtime_get_if_in_use(gmu->dev) == 0) > + if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) IMHO, if we do end up using the GPU's runpm instead of the GMU's, we should probably just move this _get_if_in_use() into msm_gpu_devfreq, etc. (And probably also this should be "<= 0".. I have that change locally but haven't sent a patch yet BR, -R > 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); > + pm_runtime_put(&gpu->pdev->dev); > return; > } > > @@ -159,7 +159,7 @@ void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp) > 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); > + pm_runtime_put(&gpu->pdev->dev); > } > > unsigned long a6xx_gmu_get_freq(struct msm_gpu *gpu) > diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > index 841e47a0b06b..87568d0b6ef8 100644 > --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c > @@ -1659,7 +1659,7 @@ static u64 a6xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) > *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) > + if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) > return 0; > > busy_cycles = gmu_read64(&a6xx_gpu->gmu, > @@ -1667,7 +1667,7 @@ static u64 a6xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) > REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_H); > > > - pm_runtime_put(a6xx_gpu->gmu.dev); > + pm_runtime_put(&gpu->pdev->dev); > > return busy_cycles; > } > -- > 2.36.1.255.ge46751e96f-goog >
On 6/9/2022 9:24 PM, Rob Clark wrote: > On Thu, Jun 9, 2022 at 7:34 AM Douglas Anderson <dianders@chromium.org> wrote: >> 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". >> >> Let's grab a reference to the correct device. Incidentally, this makes >> us match the a5xx routine more closely. >> >> This is easily shown to fix crashes that happen if we change the GPU's >> pm_runtime usage to not use autosuspend. It's also believed to fix >> some long tail GPU crashes even with autosuspend. >> >> NOTE: the crashes I've seen were fixed by _only_ fixing >> a6xx_gpu_busy(). However, I believe that the same arguments should be >> made to a6xx_gmu_set_freq() so I've changed that function too. >> >> Fixes: eadf79286a4b ("drm/msm: Check for powered down HW in the devfreq callbacks") >> Signed-off-by: Douglas Anderson <dianders@chromium.org> >> --- >> >> drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 6 +++--- >> drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 4 ++-- >> 2 files changed, 5 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c >> index 9f76f5b15759..b79ad2e0649c 100644 >> --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c >> +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c >> @@ -129,13 +129,13 @@ void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp) >> * This can get called from devfreq while the hardware is idle. Don't >> * bring up the power if it isn't already active >> */ >> - if (pm_runtime_get_if_in_use(gmu->dev) == 0) >> + if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) > IMHO, if we do end up using the GPU's runpm instead of the GMU's, we > should probably just move this _get_if_in_use() into msm_gpu_devfreq, + 1 this. > etc. (And probably also this should be "<= 0".. I have that change > locally but haven't sent a patch yet .. and skip return here when CONFIG_PM is disabled. -Akhil. > > BR, > -R > >> 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); >> + pm_runtime_put(&gpu->pdev->dev); >> return; >> } >> >> @@ -159,7 +159,7 @@ void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp) >> 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); >> + pm_runtime_put(&gpu->pdev->dev); >> } >> >> unsigned long a6xx_gmu_get_freq(struct msm_gpu *gpu) >> diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c >> index 841e47a0b06b..87568d0b6ef8 100644 >> --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c >> +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c >> @@ -1659,7 +1659,7 @@ static u64 a6xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) >> *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) >> + if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) >> return 0; >> >> busy_cycles = gmu_read64(&a6xx_gpu->gmu, >> @@ -1667,7 +1667,7 @@ static u64 a6xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) >> REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_H); >> >> >> - pm_runtime_put(a6xx_gpu->gmu.dev); >> + pm_runtime_put(&gpu->pdev->dev); >> >> return busy_cycles; >> } >> -- >> 2.36.1.255.ge46751e96f-goog >>
diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c index 9f76f5b15759..b79ad2e0649c 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c @@ -129,13 +129,13 @@ void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp) * This can get called from devfreq while the hardware is idle. Don't * bring up the power if it isn't already active */ - if (pm_runtime_get_if_in_use(gmu->dev) == 0) + if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) 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); + pm_runtime_put(&gpu->pdev->dev); return; } @@ -159,7 +159,7 @@ void a6xx_gmu_set_freq(struct msm_gpu *gpu, struct dev_pm_opp *opp) 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); + pm_runtime_put(&gpu->pdev->dev); } unsigned long a6xx_gmu_get_freq(struct msm_gpu *gpu) diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c index 841e47a0b06b..87568d0b6ef8 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c @@ -1659,7 +1659,7 @@ static u64 a6xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) *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) + if (pm_runtime_get_if_in_use(&gpu->pdev->dev) == 0) return 0; busy_cycles = gmu_read64(&a6xx_gpu->gmu, @@ -1667,7 +1667,7 @@ static u64 a6xx_gpu_busy(struct msm_gpu *gpu, unsigned long *out_sample_rate) REG_A6XX_GMU_CX_GMU_POWER_COUNTER_XOCLK_0_H); - pm_runtime_put(a6xx_gpu->gmu.dev); + pm_runtime_put(&gpu->pdev->dev); return busy_cycles; }
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". Let's grab a reference to the correct device. Incidentally, this makes us match the a5xx routine more closely. This is easily shown to fix crashes that happen if we change the GPU's pm_runtime usage to not use autosuspend. It's also believed to fix some long tail GPU crashes even with autosuspend. NOTE: the crashes I've seen were fixed by _only_ fixing a6xx_gpu_busy(). However, I believe that the same arguments should be made to a6xx_gmu_set_freq() so I've changed that function too. Fixes: eadf79286a4b ("drm/msm: Check for powered down HW in the devfreq callbacks") Signed-off-by: Douglas Anderson <dianders@chromium.org> --- drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 6 +++--- drivers/gpu/drm/msm/adreno/a6xx_gpu.c | 4 ++-- 2 files changed, 5 insertions(+), 5 deletions(-)