| Message ID | 20231204143101.64163-2-quic_mnaresh@quicinc.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Add CPU latency QoS support for ufs driver | expand |
On 12/4/23 06:30, Maramaina Naresh wrote: > + u32 (*config_qos_vote)(struct ufs_hba *hba); Please remove the above callback since this patch series does not introduce any instances of this callback. > + > + /* This capability allows the host controller driver to use the PM QoS > + * feature. > + */ > + UFSHCD_CAP_PM_QOS = 1 << 13, > }; Why does it depend on the host driver whether or not PM QoS is enabled? Why isn't it enabled unconditionally? > + * @pm_qos_req: PM QoS request handle > + * @pm_qos_init: flag to check if pm qos init completed > */ Documentation for pm_qos_init is missing. > struct ufs_hba { > void __iomem *mmio_base; > @@ -1076,6 +1089,9 @@ struct ufs_hba { > struct ufs_hw_queue *uhq; > struct ufs_hw_queue *dev_cmd_queue; > struct ufshcd_mcq_opr_info_t mcq_opr[OPR_MAX]; > + struct pm_qos_request pm_qos_req; > + bool pm_qos_init; > + u32 qos_vote; Please rename "pm_qos_init" into "pm_qos_initialized". Thanks, Bart.
On 12/5/2023 12:30 AM, Bart Van Assche wrote: > On 12/4/23 06:30, Maramaina Naresh wrote: >> + u32 (*config_qos_vote)(struct ufs_hba *hba); > > Please remove the above callback since this patch series does not > introduce any instances of this callback. > Sure Bart, i will take care of this comment in next patch set. If some SoC vendor have a different qos vote value then this callback can be added in future. >> + >> + /* This capability allows the host controller driver to use the >> PM QoS >> + * feature. >> + */ >> + UFSHCD_CAP_PM_QOS = 1 << 13, >> }; > > Why does it depend on the host driver whether or not PM QoS is > enabled? Why isn't it enabled unconditionally? For some platform vendors power KPI might be more important than random io KPI. Hence this flag is disabled by default and can be enabled based on platform requirement. > >> + * @pm_qos_req: PM QoS request handle >> + * @pm_qos_init: flag to check if pm qos init completed >> */ > > Documentation for pm_qos_init is missing. > Sorry, i didn't get your comment, i have already added documentation for @pm_qos_init, @pm_qos_req variable as above. Do you want me to add this information some where else as well? >> struct ufs_hba { >> void __iomem *mmio_base; >> @@ -1076,6 +1089,9 @@ struct ufs_hba { >> struct ufs_hw_queue *uhq; >> struct ufs_hw_queue *dev_cmd_queue; >> struct ufshcd_mcq_opr_info_t mcq_opr[OPR_MAX]; >> + struct pm_qos_request pm_qos_req; >> + bool pm_qos_init; >> + u32 qos_vote; > > Please rename "pm_qos_init" into "pm_qos_initialized". > Sure Bart, i will take care of this comment in next patch set. > Thanks, > > Bart. > Thanks, Naresh.
On 12/4/23 21:58, Naresh Maramaina wrote: > On 12/5/2023 12:30 AM, Bart Van Assche wrote: >> On 12/4/23 06:30, Maramaina Naresh wrote: >>> + /* This capability allows the host controller driver to use the >>> PM QoS >>> + * feature. >>> + */ >>> + UFSHCD_CAP_PM_QOS = 1 << 13, >>> }; >> >> Why does it depend on the host driver whether or not PM QoS is >> enabled? Why isn't it enabled unconditionally? > > For some platform vendors power KPI might be more important than random > io KPI. Hence this flag is disabled by default and can be enabled based > on platform requirement. How about leaving this flag out unless if a host vendor asks explicitly for this flag? >> >>> + * @pm_qos_req: PM QoS request handle >>> + * @pm_qos_init: flag to check if pm qos init completed >>> */ >> >> Documentation for pm_qos_init is missing. >> > Sorry, i didn't get your comment, i have already added documentation for > @pm_qos_init, @pm_qos_req variable as above. Do you want me to add this > information some where else as well? Oops, I meant 'qos_vote'. Thanks, Bart.
On 12/5/2023 10:41 PM, Bart Van Assche wrote: > On 12/4/23 21:58, Naresh Maramaina wrote: >> On 12/5/2023 12:30 AM, Bart Van Assche wrote: >>> On 12/4/23 06:30, Maramaina Naresh wrote: >>>> + /* This capability allows the host controller driver to use the >>>> PM QoS >>>> + * feature. >>>> + */ >>>> + UFSHCD_CAP_PM_QOS = 1 << 13, >>>> }; >>> >>> Why does it depend on the host driver whether or not PM QoS is >>> enabled? Why isn't it enabled unconditionally? >> >> For some platform vendors power KPI might be more important than >> random io KPI. Hence this flag is disabled by default and can be >> enabled based on platform requirement. > > How about leaving this flag out unless if a host vendor asks explicitly > for this flag? IMHO, instead of completely removing this flag, how about having flag like "UFSHCD_CAP_DISABLE_PM_QOS" which will make PMQOS enable by default and if some host vendor wants to disable it explicitly, they can enable that flag. Please let me know your opinion. >>> >>>> + * @pm_qos_req: PM QoS request handle >>>> + * @pm_qos_init: flag to check if pm qos init completed >>>> */ >>> >>> Documentation for pm_qos_init is missing. >>> >> Sorry, i didn't get your comment, i have already added documentation >> for @pm_qos_init, @pm_qos_req variable as above. Do you want me to add >> this information some where else as well? > > Oops, I meant 'qos_vote'. Sure. I'll take of this in next patchset. > > Thanks, > > Bart. > Thanks, Naresh
On Mon, Dec 04, 2023 at 08:00:59PM +0530, Maramaina Naresh wrote: > Register ufs driver to CPU latency PM QoS framework can improves > ufs device random io performance. > > PM QoS initialization will insert new QoS request into the CPU > latency QoS list with the maximum latency PM_QOS_DEFAULT_VALUE > value. > > UFS driver will vote for performance mode on scale up and power > save mode for scale down. > > If clock scaling feature is not enabled then voting will be based > on clock on or off condition. > > tiotest benchmark tool io performance results on sm8550 platform: > > 1. Without PM QoS support > Type (Speed in) | Average of 18 iterations > Random Write(IPOS) | 41065.13 > Random Read(IPOS) | 37101.3 > > 2. With PM QoS support > Type (Speed in) | Average of 18 iterations > Random Write(IPOS) | 46784.9 > Random Read(IPOS) | 42943.4 > (Improvement % with PM QoS = ~15%). > > Co-developed-by: Nitin Rawat <quic_nitirawa@quicinc.com> > Signed-off-by: Nitin Rawat <quic_nitirawa@quicinc.com> > Signed-off-by: Naveen Kumar Goud Arepalli <quic_narepall@quicinc.com> > Signed-off-by: Maramaina Naresh <quic_mnaresh@quicinc.com> > --- > drivers/ufs/core/ufshcd-priv.h | 8 +++++ > drivers/ufs/core/ufshcd.c | 62 ++++++++++++++++++++++++++++++++++ > include/ufs/ufshcd.h | 16 +++++++++ > 3 files changed, 86 insertions(+) > > diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h > index f42d99ce5bf1..536805f6c4e1 100644 > --- a/drivers/ufs/core/ufshcd-priv.h > +++ b/drivers/ufs/core/ufshcd-priv.h > @@ -241,6 +241,14 @@ static inline void ufshcd_vops_config_scaling_param(struct ufs_hba *hba, > hba->vops->config_scaling_param(hba, p, data); > } > > +static inline u32 ufshcd_vops_config_qos_vote(struct ufs_hba *hba) > +{ > + if (hba->vops && hba->vops->config_qos_vote) > + return hba->vops->config_qos_vote(hba); Please remove this callback as Bart noted. > + > + return UFSHCD_QOS_DEFAULT_VOTE; > +} > + > static inline void ufshcd_vops_reinit_notify(struct ufs_hba *hba) > { > if (hba->vops && hba->vops->reinit_notify) > diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c > index ae9936fc6ffb..13370febd2b5 100644 > --- a/drivers/ufs/core/ufshcd.c > +++ b/drivers/ufs/core/ufshcd.c > @@ -1001,6 +1001,20 @@ static bool ufshcd_is_unipro_pa_params_tuning_req(struct ufs_hba *hba) > return ufshcd_get_local_unipro_ver(hba) < UFS_UNIPRO_VER_1_6; > } > > +/** > + * ufshcd_pm_qos_perf - vote for PM QoS performance or power save mode ufshcd_pm_qos_update() - Update PM QoS request > + * @hba: per adapter instance > + * @on: If True, vote for perf PM QoS mode otherwise power save mode > + */ > +static void ufshcd_pm_qos_perf(struct ufs_hba *hba, bool on) > +{ > + if (!hba->pm_qos_init) > + return; > + > + cpu_latency_qos_update_request(&hba->pm_qos_req, on ? hba->qos_vote > + : PM_QOS_DEFAULT_VALUE); > +} > + > /** > * ufshcd_set_clk_freq - set UFS controller clock frequencies > * @hba: per adapter instance > @@ -1153,6 +1167,10 @@ static int ufshcd_scale_clks(struct ufs_hba *hba, unsigned long freq, > trace_ufshcd_profile_clk_scaling(dev_name(hba->dev), > (scale_up ? "up" : "down"), > ktime_to_us(ktime_sub(ktime_get(), start)), ret); > + > + if (!ret) > + ufshcd_pm_qos_perf(hba, scale_up); Can't you just move this before trace_ufshcd_profile_clk_scaling()? This also avoids checking for !ret. > + > return ret; > } > > @@ -9204,6 +9222,8 @@ static int ufshcd_setup_clocks(struct ufs_hba *hba, bool on) > if (ret) > return ret; > > + if (!ufshcd_is_clkscaling_supported(hba)) > + ufshcd_pm_qos_perf(hba, on); > out: > if (ret) { > list_for_each_entry(clki, head, list) { > @@ -9296,6 +9316,45 @@ static int ufshcd_init_clocks(struct ufs_hba *hba) > return ret; > } > > +/** > + * ufshcd_pm_qos_init - initialize PM QoS instance "Initialize PM QoS request" > + * @hba: per adapter instance > + */ > +static void ufshcd_pm_qos_init(struct ufs_hba *hba) > +{ > + if (!(hba->caps & UFSHCD_CAP_PM_QOS)) > + return; > + > + /* > + * called to configure PM QoS vote value for UFS host, > + * expecting qos vote return value from caller else > + * default vote value will be return. > + */ > + hba->qos_vote = ufshcd_vops_config_qos_vote(hba); No need of this variable too if you get rid of the callback. > + cpu_latency_qos_add_request(&hba->pm_qos_req, > + PM_QOS_DEFAULT_VALUE); > + > + if (cpu_latency_qos_request_active(&hba->pm_qos_req)) > + hba->pm_qos_init = true; Why do you need this flag? > + > + dev_dbg(hba->dev, "%s: QoS %s, qos_vote: %u\n", __func__, > + hba->pm_qos_init ? "initialized" : "uninitialized", > + hba->qos_vote); > +} > + > +/** > + * ufshcd_pm_qos_exit - remove instance from PM QoS > + * @hba: per adapter instance > + */ > +static void ufshcd_pm_qos_exit(struct ufs_hba *hba) > +{ > + if (!hba->pm_qos_init) > + return; > + > + cpu_latency_qos_remove_request(&hba->pm_qos_req); > + hba->pm_qos_init = false; > +} > + [...] > /** > * struct ufs_hba - per adapter private structure > * @mmio_base: UFSHCI base register address > @@ -912,6 +923,8 @@ enum ufshcd_mcq_opr { > * @mcq_base: Multi circular queue registers base address > * @uhq: array of supported hardware queues > * @dev_cmd_queue: Queue for issuing device management commands > + * @pm_qos_req: PM QoS request handle > + * @pm_qos_init: flag to check if pm qos init completed > */ > struct ufs_hba { > void __iomem *mmio_base; > @@ -1076,6 +1089,9 @@ struct ufs_hba { > struct ufs_hw_queue *uhq; > struct ufs_hw_queue *dev_cmd_queue; > struct ufshcd_mcq_opr_info_t mcq_opr[OPR_MAX]; > + struct pm_qos_request pm_qos_req; > + bool pm_qos_init; > + u32 qos_vote; Order doesn't match Kdoc. - Mani
On Wed, Dec 06, 2023 at 07:32:54PM +0530, Naresh Maramaina wrote: > > > On 12/5/2023 10:41 PM, Bart Van Assche wrote: > > On 12/4/23 21:58, Naresh Maramaina wrote: > > > On 12/5/2023 12:30 AM, Bart Van Assche wrote: > > > > On 12/4/23 06:30, Maramaina Naresh wrote: > > > > > + /* This capability allows the host controller driver to > > > > > use the PM QoS > > > > > + * feature. > > > > > + */ > > > > > + UFSHCD_CAP_PM_QOS = 1 << 13, > > > > > }; > > > > > > > > Why does it depend on the host driver whether or not PM QoS is > > > > enabled? Why isn't it enabled unconditionally? > > > > > > For some platform vendors power KPI might be more important than > > > random io KPI. Hence this flag is disabled by default and can be > > > enabled based on platform requirement. > > > > How about leaving this flag out unless if a host vendor asks explicitly > > for this flag? > > IMHO, instead of completely removing this flag, how about having > flag like "UFSHCD_CAP_DISABLE_PM_QOS" which will make PMQOS enable > by default and if some host vendor wants to disable it explicitly, > they can enable that flag. > Please let me know your opinion. > If a vendor wants to disable this feature, then the driver has to be modified. That won't be very convenient. So either this has to be configured through sysfs or Kconfig if flexibility matters. - Mani > > > > > > > > > + * @pm_qos_req: PM QoS request handle > > > > > + * @pm_qos_init: flag to check if pm qos init completed > > > > > */ > > > > > > > > Documentation for pm_qos_init is missing. > > > > > > > Sorry, i didn't get your comment, i have already added documentation > > > for @pm_qos_init, @pm_qos_req variable as above. Do you want me to > > > add this information some where else as well? > > > > Oops, I meant 'qos_vote'. > > Sure. I'll take of this in next patchset. > > > > > Thanks, > > > > Bart. > > > > Thanks, > Naresh >
On 12/6/23 05:32, Manivannan Sadhasivam wrote: > On Wed, Dec 06, 2023 at 07:32:54PM +0530, Naresh Maramaina wrote: >> On 12/5/2023 10:41 PM, Bart Van Assche wrote: >>> On 12/4/23 21:58, Naresh Maramaina wrote: >>>> On 12/5/2023 12:30 AM, Bart Van Assche wrote: >>>>> On 12/4/23 06:30, Maramaina Naresh wrote: >>>>>> + /* This capability allows the host controller driver to >>>>>> use the PM QoS >>>>>> + * feature. >>>>>> + */ >>>>>> + UFSHCD_CAP_PM_QOS = 1 << 13, >>>>>> }; >>>>> >>>>> Why does it depend on the host driver whether or not PM QoS is >>>>> enabled? Why isn't it enabled unconditionally? >>>> >>>> For some platform vendors power KPI might be more important than >>>> random io KPI. Hence this flag is disabled by default and can be >>>> enabled based on platform requirement. >>> >>> How about leaving this flag out unless if a host vendor asks explicitly >>> for this flag? >> >> IMHO, instead of completely removing this flag, how about having >> flag like "UFSHCD_CAP_DISABLE_PM_QOS" which will make PMQOS enable >> by default and if some host vendor wants to disable it explicitly, >> they can enable that flag. >> Please let me know your opinion. That would result in a flag that is tested but that is never set by upstream code. I'm not sure that's acceptable. > If a vendor wants to disable this feature, then the driver has to be modified. > That won't be very convenient. So either this has to be configured through sysfs > or Kconfig if flexibility matters. Kconfig sounds worse to me because changing any Kconfig flag requires a modification of the Android GKI kernel. Thanks, Bart.
On Wed, Dec 06, 2023 at 03:02:04PM -1000, Bart Van Assche wrote: > On 12/6/23 05:32, Manivannan Sadhasivam wrote: > > On Wed, Dec 06, 2023 at 07:32:54PM +0530, Naresh Maramaina wrote: > > > On 12/5/2023 10:41 PM, Bart Van Assche wrote: > > > > On 12/4/23 21:58, Naresh Maramaina wrote: > > > > > On 12/5/2023 12:30 AM, Bart Van Assche wrote: > > > > > > On 12/4/23 06:30, Maramaina Naresh wrote: > > > > > > > + /* This capability allows the host controller driver to > > > > > > > use the PM QoS > > > > > > > + * feature. > > > > > > > + */ > > > > > > > + UFSHCD_CAP_PM_QOS = 1 << 13, > > > > > > > }; > > > > > > > > > > > > Why does it depend on the host driver whether or not PM QoS is > > > > > > enabled? Why isn't it enabled unconditionally? > > > > > > > > > > For some platform vendors power KPI might be more important than > > > > > random io KPI. Hence this flag is disabled by default and can be > > > > > enabled based on platform requirement. > > > > > > > > How about leaving this flag out unless if a host vendor asks explicitly > > > > for this flag? > > > > > > IMHO, instead of completely removing this flag, how about having > > > flag like "UFSHCD_CAP_DISABLE_PM_QOS" which will make PMQOS enable > > > by default and if some host vendor wants to disable it explicitly, > > > they can enable that flag. > > > Please let me know your opinion. > > That would result in a flag that is tested but that is never set by > upstream code. I'm not sure that's acceptable. > Agree. The flag shouldn't be introduced if there are no users. > > If a vendor wants to disable this feature, then the driver has to be modified. > > That won't be very convenient. So either this has to be configured through sysfs > > or Kconfig if flexibility matters. > > Kconfig sounds worse to me because changing any Kconfig flag requires a > modification of the Android GKI kernel. > Hmm, ok. Then I think we can have a sysfs hook to toggle the enable switch. - Mani > Thanks, > > Bart.
On 12/7/2023 3:13 PM, Manivannan Sadhasivam wrote: > On Wed, Dec 06, 2023 at 03:02:04PM -1000, Bart Van Assche wrote: >> On 12/6/23 05:32, Manivannan Sadhasivam wrote: >>> On Wed, Dec 06, 2023 at 07:32:54PM +0530, Naresh Maramaina wrote: >>>> On 12/5/2023 10:41 PM, Bart Van Assche wrote: >>>>> On 12/4/23 21:58, Naresh Maramaina wrote: >>>>>> On 12/5/2023 12:30 AM, Bart Van Assche wrote: >>>>>>> On 12/4/23 06:30, Maramaina Naresh wrote: >>>>>>>> + /* This capability allows the host controller driver to >>>>>>>> use the PM QoS >>>>>>>> + * feature. >>>>>>>> + */ >>>>>>>> + UFSHCD_CAP_PM_QOS = 1 << 13, >>>>>>>> }; >>>>>>> >>>>>>> Why does it depend on the host driver whether or not PM QoS is >>>>>>> enabled? Why isn't it enabled unconditionally? >>>>>> >>>>>> For some platform vendors power KPI might be more important than >>>>>> random io KPI. Hence this flag is disabled by default and can be >>>>>> enabled based on platform requirement. >>>>> >>>>> How about leaving this flag out unless if a host vendor asks explicitly >>>>> for this flag? >>>> >>>> IMHO, instead of completely removing this flag, how about having >>>> flag like "UFSHCD_CAP_DISABLE_PM_QOS" which will make PMQOS enable >>>> by default and if some host vendor wants to disable it explicitly, >>>> they can enable that flag. >>>> Please let me know your opinion. >> >> That would result in a flag that is tested but that is never set by >> upstream code. I'm not sure that's acceptable. >> > > Agree. The flag shouldn't be introduced if there are no users. > >>> If a vendor wants to disable this feature, then the driver has to be modified. >>> That won't be very convenient. So either this has to be configured through sysfs >>> or Kconfig if flexibility matters. >> >> Kconfig sounds worse to me because changing any Kconfig flag requires a >> modification of the Android GKI kernel. >> > > Hmm, ok. Then I think we can have a sysfs hook to toggle the enable switch. Hi Bart, Mani How about keeping this feature enabled by default and having a module parameter to disable pmqos feature if required ? Regards, Nitin > > - Mani > >> Thanks, >> >> Bart. >
On Thu, Dec 07, 2023 at 03:56:43PM +0530, Nitin Rawat wrote: > > > On 12/7/2023 3:13 PM, Manivannan Sadhasivam wrote: > > On Wed, Dec 06, 2023 at 03:02:04PM -1000, Bart Van Assche wrote: > > > On 12/6/23 05:32, Manivannan Sadhasivam wrote: > > > > On Wed, Dec 06, 2023 at 07:32:54PM +0530, Naresh Maramaina wrote: > > > > > On 12/5/2023 10:41 PM, Bart Van Assche wrote: > > > > > > On 12/4/23 21:58, Naresh Maramaina wrote: > > > > > > > On 12/5/2023 12:30 AM, Bart Van Assche wrote: > > > > > > > > On 12/4/23 06:30, Maramaina Naresh wrote: > > > > > > > > > + /* This capability allows the host controller driver to > > > > > > > > > use the PM QoS > > > > > > > > > + * feature. > > > > > > > > > + */ > > > > > > > > > + UFSHCD_CAP_PM_QOS = 1 << 13, > > > > > > > > > }; > > > > > > > > > > > > > > > > Why does it depend on the host driver whether or not PM QoS is > > > > > > > > enabled? Why isn't it enabled unconditionally? > > > > > > > > > > > > > > For some platform vendors power KPI might be more important than > > > > > > > random io KPI. Hence this flag is disabled by default and can be > > > > > > > enabled based on platform requirement. > > > > > > > > > > > > How about leaving this flag out unless if a host vendor asks explicitly > > > > > > for this flag? > > > > > > > > > > IMHO, instead of completely removing this flag, how about having > > > > > flag like "UFSHCD_CAP_DISABLE_PM_QOS" which will make PMQOS enable > > > > > by default and if some host vendor wants to disable it explicitly, > > > > > they can enable that flag. > > > > > Please let me know your opinion. > > > > > > That would result in a flag that is tested but that is never set by > > > upstream code. I'm not sure that's acceptable. > > > > > > > Agree. The flag shouldn't be introduced if there are no users. > > > > > > If a vendor wants to disable this feature, then the driver has to be modified. > > > > That won't be very convenient. So either this has to be configured through sysfs > > > > or Kconfig if flexibility matters. > > > > > > Kconfig sounds worse to me because changing any Kconfig flag requires a > > > modification of the Android GKI kernel. > > > > > > > Hmm, ok. Then I think we can have a sysfs hook to toggle the enable switch. > > Hi Bart, Mani > > How about keeping this feature enabled by default and having a module > parameter to disable pmqos feature if required ? > Module params not encouraged these days unless there are no other feasible options available. - Mani > Regards, > Nitin > > > > > - Mani > > > > > Thanks, > > > > > > Bart. > > >
On 12/6/2023 8:56 PM, Manivannan Sadhasivam wrote: > On Mon, Dec 04, 2023 at 08:00:59PM +0530, Maramaina Naresh wrote: >> Register ufs driver to CPU latency PM QoS framework can improves >> ufs device random io performance. >> >> PM QoS initialization will insert new QoS request into the CPU >> latency QoS list with the maximum latency PM_QOS_DEFAULT_VALUE >> value. >> >> UFS driver will vote for performance mode on scale up and power >> save mode for scale down. >> >> If clock scaling feature is not enabled then voting will be based >> on clock on or off condition. >> >> tiotest benchmark tool io performance results on sm8550 platform: >> >> 1. Without PM QoS support >> Type (Speed in) | Average of 18 iterations >> Random Write(IPOS) | 41065.13 >> Random Read(IPOS) | 37101.3 >> >> 2. With PM QoS support >> Type (Speed in) | Average of 18 iterations >> Random Write(IPOS) | 46784.9 >> Random Read(IPOS) | 42943.4 >> (Improvement % with PM QoS = ~15%). >> >> Co-developed-by: Nitin Rawat <quic_nitirawa@quicinc.com> >> Signed-off-by: Nitin Rawat <quic_nitirawa@quicinc.com> >> Signed-off-by: Naveen Kumar Goud Arepalli <quic_narepall@quicinc.com> >> Signed-off-by: Maramaina Naresh <quic_mnaresh@quicinc.com> >> --- >> drivers/ufs/core/ufshcd-priv.h | 8 +++++ >> drivers/ufs/core/ufshcd.c | 62 ++++++++++++++++++++++++++++++++++ >> include/ufs/ufshcd.h | 16 +++++++++ >> 3 files changed, 86 insertions(+) >> >> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h >> index f42d99ce5bf1..536805f6c4e1 100644 >> --- a/drivers/ufs/core/ufshcd-priv.h >> +++ b/drivers/ufs/core/ufshcd-priv.h >> @@ -241,6 +241,14 @@ static inline void ufshcd_vops_config_scaling_param(struct ufs_hba *hba, >> hba->vops->config_scaling_param(hba, p, data); >> } >> >> +static inline u32 ufshcd_vops_config_qos_vote(struct ufs_hba *hba) >> +{ >> + if (hba->vops && hba->vops->config_qos_vote) >> + return hba->vops->config_qos_vote(hba); > > Please remove this callback as Bart noted. > Sure Mani, will takecare of this comment. >> + >> + return UFSHCD_QOS_DEFAULT_VOTE; >> +} >> + >> static inline void ufshcd_vops_reinit_notify(struct ufs_hba *hba) >> { >> if (hba->vops && hba->vops->reinit_notify) >> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c >> index ae9936fc6ffb..13370febd2b5 100644 >> --- a/drivers/ufs/core/ufshcd.c >> +++ b/drivers/ufs/core/ufshcd.c >> @@ -1001,6 +1001,20 @@ static bool ufshcd_is_unipro_pa_params_tuning_req(struct ufs_hba *hba) >> return ufshcd_get_local_unipro_ver(hba) < UFS_UNIPRO_VER_1_6; >> } >> >> +/** >> + * ufshcd_pm_qos_perf - vote for PM QoS performance or power save mode > > ufshcd_pm_qos_update() - Update PM QoS request > Sure Mani, will takecare of this comment. >> + * @hba: per adapter instance >> + * @on: If True, vote for perf PM QoS mode otherwise power save mode >> + */ >> +static void ufshcd_pm_qos_perf(struct ufs_hba *hba, bool on) >> +{ >> + if (!hba->pm_qos_init) >> + return; >> + >> + cpu_latency_qos_update_request(&hba->pm_qos_req, on ? hba->qos_vote >> + : PM_QOS_DEFAULT_VALUE); >> +} >> + >> /** >> * ufshcd_set_clk_freq - set UFS controller clock frequencies >> * @hba: per adapter instance >> @@ -1153,6 +1167,10 @@ static int ufshcd_scale_clks(struct ufs_hba *hba, unsigned long freq, >> trace_ufshcd_profile_clk_scaling(dev_name(hba->dev), >> (scale_up ? "up" : "down"), >> ktime_to_us(ktime_sub(ktime_get(), start)), ret); >> + >> + if (!ret) >> + ufshcd_pm_qos_perf(hba, scale_up); > > Can't you just move this before trace_ufshcd_profile_clk_scaling()? This also > avoids checking for !ret. > In this case, we need to use goto out; inside if(ret) of ufshcd_vops_clk_scale_notify. will do the above change, to enable ufshcd_pm_qos_perf before the out flag. >> + >> return ret; >> } >> >> @@ -9204,6 +9222,8 @@ static int ufshcd_setup_clocks(struct ufs_hba *hba, bool on) >> if (ret) >> return ret; >> >> + if (!ufshcd_is_clkscaling_supported(hba)) >> + ufshcd_pm_qos_perf(hba, on); >> out: >> if (ret) { >> list_for_each_entry(clki, head, list) { >> @@ -9296,6 +9316,45 @@ static int ufshcd_init_clocks(struct ufs_hba *hba) >> return ret; >> } >> >> +/** >> + * ufshcd_pm_qos_init - initialize PM QoS instance > > "Initialize PM QoS request" > Sure Mani, will takecare of this comment. >> + * @hba: per adapter instance >> + */ >> +static void ufshcd_pm_qos_init(struct ufs_hba *hba) >> +{ >> + if (!(hba->caps & UFSHCD_CAP_PM_QOS)) >> + return; >> + >> + /* >> + * called to configure PM QoS vote value for UFS host, >> + * expecting qos vote return value from caller else >> + * default vote value will be return. >> + */ >> + hba->qos_vote = ufshcd_vops_config_qos_vote(hba); > > No need of this variable too if you get rid of the callback. > >> + cpu_latency_qos_add_request(&hba->pm_qos_req, >> + PM_QOS_DEFAULT_VALUE); >> + >> + if (cpu_latency_qos_request_active(&hba->pm_qos_req)) >> + hba->pm_qos_init = true; > > Why do you need this flag? this flag ensure UFS qos request got added into the Global PM QoS list. > >> + >> + dev_dbg(hba->dev, "%s: QoS %s, qos_vote: %u\n", __func__, >> + hba->pm_qos_init ? "initialized" : "uninitialized", >> + hba->qos_vote); >> +} >> + >> +/** >> + * ufshcd_pm_qos_exit - remove instance from PM QoS >> + * @hba: per adapter instance >> + */ >> +static void ufshcd_pm_qos_exit(struct ufs_hba *hba) >> +{ >> + if (!hba->pm_qos_init) >> + return; >> + >> + cpu_latency_qos_remove_request(&hba->pm_qos_req); >> + hba->pm_qos_init = false; >> +} >> + > > [...] > >> /** >> * struct ufs_hba - per adapter private structure >> * @mmio_base: UFSHCI base register address >> @@ -912,6 +923,8 @@ enum ufshcd_mcq_opr { >> * @mcq_base: Multi circular queue registers base address >> * @uhq: array of supported hardware queues >> * @dev_cmd_queue: Queue for issuing device management commands >> + * @pm_qos_req: PM QoS request handle >> + * @pm_qos_init: flag to check if pm qos init completed >> */ >> struct ufs_hba { >> void __iomem *mmio_base; >> @@ -1076,6 +1089,9 @@ struct ufs_hba { >> struct ufs_hw_queue *uhq; >> struct ufs_hw_queue *dev_cmd_queue; >> struct ufshcd_mcq_opr_info_t mcq_opr[OPR_MAX]; >> + struct pm_qos_request pm_qos_req; >> + bool pm_qos_init; >> + u32 qos_vote; > > Order doesn't match Kdoc. > we are removing qos_vote variable in next patch series. > - Mani > Thanks, Naresh.
On 12/6/2023 8:56 PM, Manivannan Sadhasivam wrote: > On Mon, Dec 04, 2023 at 08:00:59PM +0530, Maramaina Naresh wrote: >> Register ufs driver to CPU latency PM QoS framework can improves >> ufs device random io performance. >> >> PM QoS initialization will insert new QoS request into the CPU >> latency QoS list with the maximum latency PM_QOS_DEFAULT_VALUE >> value. >> >> UFS driver will vote for performance mode on scale up and power >> save mode for scale down. >> >> If clock scaling feature is not enabled then voting will be based >> on clock on or off condition. >> >> tiotest benchmark tool io performance results on sm8550 platform: >> >> 1. Without PM QoS support >> Type (Speed in) | Average of 18 iterations >> Random Write(IPOS) | 41065.13 >> Random Read(IPOS) | 37101.3 >> >> 2. With PM QoS support >> Type (Speed in) | Average of 18 iterations >> Random Write(IPOS) | 46784.9 >> Random Read(IPOS) | 42943.4 >> (Improvement % with PM QoS = ~15%). >> >> Co-developed-by: Nitin Rawat <quic_nitirawa@quicinc.com> >> Signed-off-by: Nitin Rawat <quic_nitirawa@quicinc.com> >> Signed-off-by: Naveen Kumar Goud Arepalli <quic_narepall@quicinc.com> >> Signed-off-by: Maramaina Naresh <quic_mnaresh@quicinc.com> >> --- >> drivers/ufs/core/ufshcd-priv.h | 8 +++++ >> drivers/ufs/core/ufshcd.c | 62 ++++++++++++++++++++++++++++++++++ >> include/ufs/ufshcd.h | 16 +++++++++ >> 3 files changed, 86 insertions(+) >> >> diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h >> index f42d99ce5bf1..536805f6c4e1 100644 >> --- a/drivers/ufs/core/ufshcd-priv.h >> +++ b/drivers/ufs/core/ufshcd-priv.h >> @@ -241,6 +241,14 @@ static inline void ufshcd_vops_config_scaling_param(struct ufs_hba *hba, >> hba->vops->config_scaling_param(hba, p, data); >> } >> >> +static inline u32 ufshcd_vops_config_qos_vote(struct ufs_hba *hba) >> +{ >> + if (hba->vops && hba->vops->config_qos_vote) >> + return hba->vops->config_qos_vote(hba); > > Please remove this callback as Bart noted. > Sure Mani, will takecare of this comment. >> + >> + return UFSHCD_QOS_DEFAULT_VOTE; >> +} >> + >> static inline void ufshcd_vops_reinit_notify(struct ufs_hba *hba) >> { >> if (hba->vops && hba->vops->reinit_notify) >> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c >> index ae9936fc6ffb..13370febd2b5 100644 >> --- a/drivers/ufs/core/ufshcd.c >> +++ b/drivers/ufs/core/ufshcd.c >> @@ -1001,6 +1001,20 @@ static bool ufshcd_is_unipro_pa_params_tuning_req(struct ufs_hba *hba) >> return ufshcd_get_local_unipro_ver(hba) < UFS_UNIPRO_VER_1_6; >> } >> >> +/** >> + * ufshcd_pm_qos_perf - vote for PM QoS performance or power save mode > > ufshcd_pm_qos_update() - Update PM QoS request > Sure Mani, will takecare of this comment. >> + * @hba: per adapter instance >> + * @on: If True, vote for perf PM QoS mode otherwise power save mode >> + */ >> +static void ufshcd_pm_qos_perf(struct ufs_hba *hba, bool on) >> +{ >> + if (!hba->pm_qos_init) >> + return; >> + >> + cpu_latency_qos_update_request(&hba->pm_qos_req, on ? hba->qos_vote >> + : PM_QOS_DEFAULT_VALUE); >> +} >> + >> /** >> * ufshcd_set_clk_freq - set UFS controller clock frequencies >> * @hba: per adapter instance >> @@ -1153,6 +1167,10 @@ static int ufshcd_scale_clks(struct ufs_hba *hba, unsigned long freq, >> trace_ufshcd_profile_clk_scaling(dev_name(hba->dev), >> (scale_up ? "up" : "down"), >> ktime_to_us(ktime_sub(ktime_get(), start)), ret); >> + >> + if (!ret) >> + ufshcd_pm_qos_perf(hba, scale_up); > > Can't you just move this before trace_ufshcd_profile_clk_scaling()? This also > avoids checking for !ret. > In this case, we need to use goto out; inside if condition of ufshcd_vops_clk_scale_notify. we can enable ufshcd_pm_qos_perf only when ufshcd_vops_clk_scale_notify is successful. Will add goto out; in next patch set. >> + >> return ret; >> } >> >> @@ -9204,6 +9222,8 @@ static int ufshcd_setup_clocks(struct ufs_hba *hba, bool on) >> if (ret) >> return ret; >> >> + if (!ufshcd_is_clkscaling_supported(hba)) >> + ufshcd_pm_qos_perf(hba, on); >> out: >> if (ret) { >> list_for_each_entry(clki, head, list) { >> @@ -9296,6 +9316,45 @@ static int ufshcd_init_clocks(struct ufs_hba *hba) >> return ret; >> } >> >> +/** >> + * ufshcd_pm_qos_init - initialize PM QoS instance > > "Initialize PM QoS request" > Sure Mani, will takecare of this comment. >> + * @hba: per adapter instance >> + */ >> +static void ufshcd_pm_qos_init(struct ufs_hba *hba) >> +{ >> + if (!(hba->caps & UFSHCD_CAP_PM_QOS)) >> + return; >> + >> + /* >> + * called to configure PM QoS vote value for UFS host, >> + * expecting qos vote return value from caller else >> + * default vote value will be return. >> + */ >> + hba->qos_vote = ufshcd_vops_config_qos_vote(hba); > > No need of this variable too if you get rid of the callback. > >> + cpu_latency_qos_add_request(&hba->pm_qos_req, >> + PM_QOS_DEFAULT_VALUE); >> + >> + if (cpu_latency_qos_request_active(&hba->pm_qos_req)) >> + hba->pm_qos_init = true; > > Why do you need this flag? this flag ensure UFS qos request got added into the list. > >> + >> + dev_dbg(hba->dev, "%s: QoS %s, qos_vote: %u\n", __func__, >> + hba->pm_qos_init ? "initialized" : "uninitialized", >> + hba->qos_vote); >> +} >> + >> +/** >> + * ufshcd_pm_qos_exit - remove instance from PM QoS >> + * @hba: per adapter instance >> + */ >> +static void ufshcd_pm_qos_exit(struct ufs_hba *hba) >> +{ >> + if (!hba->pm_qos_init) >> + return; >> + >> + cpu_latency_qos_remove_request(&hba->pm_qos_req); >> + hba->pm_qos_init = false; >> +} >> + > > [...] > >> /** >> * struct ufs_hba - per adapter private structure >> * @mmio_base: UFSHCI base register address >> @@ -912,6 +923,8 @@ enum ufshcd_mcq_opr { >> * @mcq_base: Multi circular queue registers base address >> * @uhq: array of supported hardware queues >> * @dev_cmd_queue: Queue for issuing device management commands >> + * @pm_qos_req: PM QoS request handle >> + * @pm_qos_init: flag to check if pm qos init completed >> */ >> struct ufs_hba { >> void __iomem *mmio_base; >> @@ -1076,6 +1089,9 @@ struct ufs_hba { >> struct ufs_hw_queue *uhq; >> struct ufs_hw_queue *dev_cmd_queue; >> struct ufshcd_mcq_opr_info_t mcq_opr[OPR_MAX]; >> + struct pm_qos_request pm_qos_req; >> + bool pm_qos_init; >> + u32 qos_vote; > > Order doesn't match Kdoc. > qos_vote variable will be remove as per latest comment. > - Mani > Thanks, Naresh.
On Thu, Dec 07, 2023 at 04:51:01PM +0530, Manivannan Sadhasivam wrote: > On Thu, Dec 07, 2023 at 03:56:43PM +0530, Nitin Rawat wrote: > > > > > > On 12/7/2023 3:13 PM, Manivannan Sadhasivam wrote: > > > On Wed, Dec 06, 2023 at 03:02:04PM -1000, Bart Van Assche wrote: > > > > On 12/6/23 05:32, Manivannan Sadhasivam wrote: > > > > > On Wed, Dec 06, 2023 at 07:32:54PM +0530, Naresh Maramaina wrote: > > > > > > On 12/5/2023 10:41 PM, Bart Van Assche wrote: > > > > > > > On 12/4/23 21:58, Naresh Maramaina wrote: > > > > > > > > On 12/5/2023 12:30 AM, Bart Van Assche wrote: > > > > > > > > > On 12/4/23 06:30, Maramaina Naresh wrote: > > > > > > > > > > + /* This capability allows the host controller driver to > > > > > > > > > > use the PM QoS > > > > > > > > > > + * feature. > > > > > > > > > > + */ > > > > > > > > > > + UFSHCD_CAP_PM_QOS = 1 << 13, > > > > > > > > > > }; > > > > > > > > > > > > > > > > > > Why does it depend on the host driver whether or not PM QoS is > > > > > > > > > enabled? Why isn't it enabled unconditionally? > > > > > > > > > > > > > > > > For some platform vendors power KPI might be more important than > > > > > > > > random io KPI. Hence this flag is disabled by default and can be > > > > > > > > enabled based on platform requirement. > > > > > > > > > > > > > > How about leaving this flag out unless if a host vendor asks explicitly > > > > > > > for this flag? > > > > > > > > > > > > IMHO, instead of completely removing this flag, how about having > > > > > > flag like "UFSHCD_CAP_DISABLE_PM_QOS" which will make PMQOS enable > > > > > > by default and if some host vendor wants to disable it explicitly, > > > > > > they can enable that flag. > > > > > > Please let me know your opinion. > > > > > > > > That would result in a flag that is tested but that is never set by > > > > upstream code. I'm not sure that's acceptable. > > > > > > > > > > Agree. The flag shouldn't be introduced if there are no users. > > > > > > > > If a vendor wants to disable this feature, then the driver has to be modified. > > > > > That won't be very convenient. So either this has to be configured through sysfs > > > > > or Kconfig if flexibility matters. > > > > > > > > Kconfig sounds worse to me because changing any Kconfig flag requires a > > > > modification of the Android GKI kernel. > > > > > > > > > > Hmm, ok. Then I think we can have a sysfs hook to toggle the enable switch. > > > > Hi Bart, Mani > > > > How about keeping this feature enabled by default and having a module > > parameter to disable pmqos feature if required ? > > > > Module params not encouraged these days unless there are no other feasible > options available. Yeah, one of the problem with module param is that we can't have different settings for two two UFS controllers. Anyways, this setting need not be a module param, there is nothing special about this setting that is tied to module loading (driver probe) time, AFAICT. Thanks, Pavan
diff --git a/drivers/ufs/core/ufshcd-priv.h b/drivers/ufs/core/ufshcd-priv.h index f42d99ce5bf1..536805f6c4e1 100644 --- a/drivers/ufs/core/ufshcd-priv.h +++ b/drivers/ufs/core/ufshcd-priv.h @@ -241,6 +241,14 @@ static inline void ufshcd_vops_config_scaling_param(struct ufs_hba *hba, hba->vops->config_scaling_param(hba, p, data); } +static inline u32 ufshcd_vops_config_qos_vote(struct ufs_hba *hba) +{ + if (hba->vops && hba->vops->config_qos_vote) + return hba->vops->config_qos_vote(hba); + + return UFSHCD_QOS_DEFAULT_VOTE; +} + static inline void ufshcd_vops_reinit_notify(struct ufs_hba *hba) { if (hba->vops && hba->vops->reinit_notify) diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c index ae9936fc6ffb..13370febd2b5 100644 --- a/drivers/ufs/core/ufshcd.c +++ b/drivers/ufs/core/ufshcd.c @@ -1001,6 +1001,20 @@ static bool ufshcd_is_unipro_pa_params_tuning_req(struct ufs_hba *hba) return ufshcd_get_local_unipro_ver(hba) < UFS_UNIPRO_VER_1_6; } +/** + * ufshcd_pm_qos_perf - vote for PM QoS performance or power save mode + * @hba: per adapter instance + * @on: If True, vote for perf PM QoS mode otherwise power save mode + */ +static void ufshcd_pm_qos_perf(struct ufs_hba *hba, bool on) +{ + if (!hba->pm_qos_init) + return; + + cpu_latency_qos_update_request(&hba->pm_qos_req, on ? hba->qos_vote + : PM_QOS_DEFAULT_VALUE); +} + /** * ufshcd_set_clk_freq - set UFS controller clock frequencies * @hba: per adapter instance @@ -1153,6 +1167,10 @@ static int ufshcd_scale_clks(struct ufs_hba *hba, unsigned long freq, trace_ufshcd_profile_clk_scaling(dev_name(hba->dev), (scale_up ? "up" : "down"), ktime_to_us(ktime_sub(ktime_get(), start)), ret); + + if (!ret) + ufshcd_pm_qos_perf(hba, scale_up); + return ret; } @@ -9204,6 +9222,8 @@ static int ufshcd_setup_clocks(struct ufs_hba *hba, bool on) if (ret) return ret; + if (!ufshcd_is_clkscaling_supported(hba)) + ufshcd_pm_qos_perf(hba, on); out: if (ret) { list_for_each_entry(clki, head, list) { @@ -9296,6 +9316,45 @@ static int ufshcd_init_clocks(struct ufs_hba *hba) return ret; } +/** + * ufshcd_pm_qos_init - initialize PM QoS instance + * @hba: per adapter instance + */ +static void ufshcd_pm_qos_init(struct ufs_hba *hba) +{ + if (!(hba->caps & UFSHCD_CAP_PM_QOS)) + return; + + /* + * called to configure PM QoS vote value for UFS host, + * expecting qos vote return value from caller else + * default vote value will be return. + */ + hba->qos_vote = ufshcd_vops_config_qos_vote(hba); + cpu_latency_qos_add_request(&hba->pm_qos_req, + PM_QOS_DEFAULT_VALUE); + + if (cpu_latency_qos_request_active(&hba->pm_qos_req)) + hba->pm_qos_init = true; + + dev_dbg(hba->dev, "%s: QoS %s, qos_vote: %u\n", __func__, + hba->pm_qos_init ? "initialized" : "uninitialized", + hba->qos_vote); +} + +/** + * ufshcd_pm_qos_exit - remove instance from PM QoS + * @hba: per adapter instance + */ +static void ufshcd_pm_qos_exit(struct ufs_hba *hba) +{ + if (!hba->pm_qos_init) + return; + + cpu_latency_qos_remove_request(&hba->pm_qos_req); + hba->pm_qos_init = false; +} + static int ufshcd_variant_hba_init(struct ufs_hba *hba) { int err = 0; @@ -9381,6 +9440,7 @@ static int ufshcd_hba_init(struct ufs_hba *hba) static void ufshcd_hba_exit(struct ufs_hba *hba) { if (hba->is_powered) { + ufshcd_pm_qos_exit(hba); ufshcd_exit_clk_scaling(hba); ufshcd_exit_clk_gating(hba); if (hba->eh_wq) @@ -10030,6 +10090,7 @@ static int ufshcd_suspend(struct ufs_hba *hba) ufshcd_vreg_set_lpm(hba); /* Put the host controller in low power mode if possible */ ufshcd_hba_vreg_set_lpm(hba); + ufshcd_pm_qos_perf(hba, false); return ret; } @@ -10576,6 +10637,7 @@ int ufshcd_init(struct ufs_hba *hba, void __iomem *mmio_base, unsigned int irq) ufs_sysfs_add_nodes(hba->dev); device_enable_async_suspend(dev); + ufshcd_pm_qos_init(hba); return 0; free_tmf_queue: diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h index d862c8ddce03..e9f2bad8934e 100644 --- a/include/ufs/ufshcd.h +++ b/include/ufs/ufshcd.h @@ -320,6 +320,9 @@ struct ufs_pwr_mode_info { * @phy_initialization: used to initialize phys * @device_reset: called to issue a reset pulse on the UFS device * @config_scaling_param: called to configure clock scaling parameters + * @config_qos_vote: called to configure PM QoS vote value for UFS host, + * expecting qos vote return value from caller else + default vote value will be return * @program_key: program or evict an inline encryption key * @event_notify: called to notify important events * @reinit_notify: called to notify reinit of UFSHCD during max gear switch @@ -362,6 +365,7 @@ struct ufs_hba_variant_ops { void (*config_scaling_param)(struct ufs_hba *hba, struct devfreq_dev_profile *profile, struct devfreq_simple_ondemand_data *data); + u32 (*config_qos_vote)(struct ufs_hba *hba); int (*program_key)(struct ufs_hba *hba, const union ufs_crypto_cfg_entry *cfg, int slot); void (*event_notify)(struct ufs_hba *hba, @@ -720,6 +724,11 @@ enum ufshcd_caps { * WriteBooster when scaling the clock down. */ UFSHCD_CAP_WB_WITH_CLK_SCALING = 1 << 12, + + /* This capability allows the host controller driver to use the PM QoS + * feature. + */ + UFSHCD_CAP_PM_QOS = 1 << 13, }; struct ufs_hba_variant_params { @@ -793,6 +802,8 @@ enum ufshcd_mcq_opr { OPR_MAX, }; +#define UFSHCD_QOS_DEFAULT_VOTE 0 + /** * struct ufs_hba - per adapter private structure * @mmio_base: UFSHCI base register address @@ -912,6 +923,8 @@ enum ufshcd_mcq_opr { * @mcq_base: Multi circular queue registers base address * @uhq: array of supported hardware queues * @dev_cmd_queue: Queue for issuing device management commands + * @pm_qos_req: PM QoS request handle + * @pm_qos_init: flag to check if pm qos init completed */ struct ufs_hba { void __iomem *mmio_base; @@ -1076,6 +1089,9 @@ struct ufs_hba { struct ufs_hw_queue *uhq; struct ufs_hw_queue *dev_cmd_queue; struct ufshcd_mcq_opr_info_t mcq_opr[OPR_MAX]; + struct pm_qos_request pm_qos_req; + bool pm_qos_init; + u32 qos_vote; }; /**