[Intel-gfx] [PATCH v3 7/8] drm/i915/mtl: Add support for PM DEMAND

Govindapillai, Vinod vinod.govindapillai at intel.com
Thu May 11 23:24:55 UTC 2023


Hello

Thanks for the comments. Pls see some inline replies..

On Thu, 2023-04-27 at 17:24 -0300, Gustavo Sousa wrote:
> Quoting Vinod Govindapillai (2023-04-27 12:00:15)
> > From: Mika Kahola <mika.kahola at intel.com>
> > 
> > Display14 introduces a new way to instruct the PUnit with
> > power and bandwidth requirements of DE. Add the functionality
> > to program the registers and handle waits using interrupts.
> > The current wait time for timeouts is programmed for 10 msecs to
> > factor in the worst case scenarios. Changes made to use REG_BIT
> > for a register that we touched(GEN8_DE_MISC_IER _MMIO).
> > 
> > Wa_14016740474 is added which applies to Xe_LPD+ display
> > 
> > v2: checkpatch warning fixes, simplify program pmdemand part
> > 
> > v3: update to dbufs and pipes values to pmdemand register(stan)
> >    Removed the macro usage in update_pmdemand_values()
> > 
> > Bspec: 66451, 64636, 64602, 64603
> > Cc: Matt Atwood <matthew.s.atwood at intel.com>
> > Cc: Matt Roper <matthew.d.roper at intel.com>
> > Cc: Lucas De Marchi <lucas.demarchi at intel.com>
> > Cc: Gustavo Sousa <gustavo.sousa at intel.com>
> > Signed-off-by: José Roberto de Souza <jose.souza at intel.com>
> > Signed-off-by: Radhakrishna Sripada <radhakrishna.sripada at intel.com>
> > Signed-off-by: Gustavo Sousa <gustavo.sousa at intel.com>
> > Signed-off-by: Mika Kahola <mika.kahola at intel.com>
> > Signed-off-by: Vinod Govindapillai <vinod.govindapillai at intel.com>
> > ---
> > drivers/gpu/drm/i915/Makefile                 |   3 +-
> > drivers/gpu/drm/i915/display/intel_display.c  |   7 +
> > .../gpu/drm/i915/display/intel_display_core.h |   6 +
> > .../drm/i915/display/intel_display_driver.c   |   7 +
> > .../drm/i915/display/intel_display_power.c    |   8 +
> > drivers/gpu/drm/i915/display/intel_pmdemand.c | 455 ++++++++++++++++++
> > drivers/gpu/drm/i915/display/intel_pmdemand.h |  24 +
> > drivers/gpu/drm/i915/i915_irq.c               |  21 +-
> > drivers/gpu/drm/i915/i915_reg.h               |  36 +-
> > 9 files changed, 562 insertions(+), 5 deletions(-)
> > create mode 100644 drivers/gpu/drm/i915/display/intel_pmdemand.c
> > create mode 100644 drivers/gpu/drm/i915/display/intel_pmdemand.h
> > 
> > diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile
> > index 9af76e376ca9..eb899fa86e51 100644
> > --- a/drivers/gpu/drm/i915/Makefile
> > +++ b/drivers/gpu/drm/i915/Makefile
> > @@ -281,7 +281,8 @@ i915-y += \
> >        display/i9xx_wm.o \
> >        display/skl_scaler.o \
> >        display/skl_universal_plane.o \
> > -  display/skl_watermark.o
> > +  display/skl_watermark.o \
> > +  display/intel_pmdemand.o
> > i915-$(CONFIG_ACPI) += \
> >        display/intel_acpi.o \
> >        display/intel_opregion.o
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c
> > b/drivers/gpu/drm/i915/display/intel_display.c
> > index bf391a6cd8d6..f98e235fadc6 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -99,6 +99,7 @@
> > #include "intel_pcode.h"
> > #include "intel_pipe_crc.h"
> > #include "intel_plane_initial.h"
> > +#include "intel_pmdemand.h"
> > #include "intel_pps.h"
> > #include "intel_psr.h"
> > #include "intel_sdvo.h"
> > @@ -6306,6 +6307,10 @@ int intel_atomic_check(struct drm_device *dev,
> >                        return ret;
> >        }
> > 
> > +  ret = intel_pmdemand_atomic_check(state);
> > +  if (ret)
> > +          goto fail;
> > +
> >        ret = intel_atomic_check_crtcs(state);
> >        if (ret)
> >                goto fail;
> > @@ -6960,6 +6965,7 @@ static void intel_atomic_commit_tail(struct intel_atomic_state *state)
> >        }
> > 
> >        intel_sagv_pre_plane_update(state);
> > +  intel_pmdemand_pre_plane_update(state);
> > 
> >        /* Complete the events for pipes that have now been disabled */
> >        for_each_new_intel_crtc_in_state(state, crtc, new_crtc_state, i) {
> > @@ -7070,6 +7076,7 @@ static void intel_atomic_commit_tail(struct intel_atomic_state *state)
> >                intel_verify_planes(state);
> > 
> >        intel_sagv_post_plane_update(state);
> > +  intel_pmdemand_post_plane_update(state);
> > 
> >        drm_atomic_helper_commit_hw_done(&state->base);
> > 
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_core.h
> > b/drivers/gpu/drm/i915/display/intel_display_core.h
> > index 9f66d734edf6..9471a052aa57 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_core.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display_core.h
> > @@ -345,6 +345,12 @@ struct intel_display {
> >                struct intel_global_obj obj;
> >        } dbuf;
> > 
> > +  struct {
> > +          wait_queue_head_t waitqueue;
> > +          struct mutex lock;
> > +          struct intel_global_obj obj;
> > +  } pmdemand;
> > +
> >        struct {
> >                /*
> >                 * dkl.phy_lock protects against concurrent access of the
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_driver.c
> > b/drivers/gpu/drm/i915/display/intel_display_driver.c
> > index 60ce10fc7205..79853d8c3240 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_driver.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display_driver.c
> > @@ -47,6 +47,7 @@
> > #include "intel_opregion.h"
> > #include "intel_overlay.h"
> > #include "intel_plane_initial.h"
> > +#include "intel_pmdemand.h"
> > #include "intel_pps.h"
> > #include "intel_quirks.h"
> > #include "intel_vga.h"
> > @@ -211,6 +212,8 @@ int intel_display_driver_probe_noirq(struct drm_i915_private *i915)
> >        if (ret < 0)
> >                goto cleanup_vga;
> > 
> > +  intel_pmdemand_init(i915);
> > +
> >        intel_power_domains_init_hw(i915, false);
> > 
> >        if (!HAS_DISPLAY(i915))
> > @@ -240,6 +243,10 @@ int intel_display_driver_probe_noirq(struct drm_i915_private *i915)
> >        if (ret)
> >                goto cleanup_vga_client_pw_domain_dmc;
> > 
> > +  ret = intel_pmdemand_state_init(i915);
> > +  if (ret)
> > +          goto cleanup_vga_client_pw_domain_dmc;
> > +
> >        init_llist_head(&i915->display.atomic_helper.free_list);
> >        INIT_WORK(&i915->display.atomic_helper.free_work,
> >                  intel_atomic_helper_free_state_worker);
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_power.c
> > b/drivers/gpu/drm/i915/display/intel_display_power.c
> > index 5150069f3f82..f5c5a486efbc 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_power.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display_power.c
> > @@ -20,6 +20,7 @@
> > #include "intel_mchbar_regs.h"
> > #include "intel_pch_refclk.h"
> > #include "intel_pcode.h"
> > +#include "intel_pmdemand.h"
> > #include "intel_pps_regs.h"
> > #include "intel_snps_phy.h"
> > #include "skl_watermark.h"
> > @@ -1085,6 +1086,10 @@ static void gen9_dbuf_enable(struct drm_i915_private *dev_priv)
> >        dev_priv->display.dbuf.enabled_slices =
> >                intel_enabled_dbuf_slices_mask(dev_priv);
> > 
> > +  if (DISPLAY_VER(dev_priv) >= 14)
> > +          intel_program_dbuf_pmdemand(dev_priv, BIT(DBUF_S1) |
> > +                                      dev_priv->display.dbuf.enabled_slices);
> > +
> >        /*
> >         * Just power up at least 1 slice, we will
> >         * figure out later which slices we have and what we need.
> > @@ -1096,6 +1101,9 @@ static void gen9_dbuf_enable(struct drm_i915_private *dev_priv)
> > static void gen9_dbuf_disable(struct drm_i915_private *dev_priv)
> > {
> >        gen9_dbuf_slices_update(dev_priv, 0);
> > +
> > +  if (DISPLAY_VER(dev_priv) >= 14)
> > +          intel_program_dbuf_pmdemand(dev_priv, 0);
> > }
> > 
> > static void gen12_dbuf_slices_config(struct drm_i915_private *dev_priv)
> > diff --git a/drivers/gpu/drm/i915/display/intel_pmdemand.c
> > b/drivers/gpu/drm/i915/display/intel_pmdemand.c
> > new file mode 100644
> > index 000000000000..df6429e7059d
> > --- /dev/null
> > +++ b/drivers/gpu/drm/i915/display/intel_pmdemand.c
> > @@ -0,0 +1,455 @@
> > +// SPDX-License-Identifier: MIT
> > +/*
> > + * Copyright © 2024 Intel Corporation
> > + */
> > +
> > +#include <linux/bitops.h>
> > +
> > +#include "i915_drv.h"
> > +#include "i915_reg.h"
> > +#include "intel_bw.h"
> > +#include "intel_cdclk.h"
> > +#include "intel_cx0_phy.h"
> > +#include "intel_de.h"
> > +#include "intel_display.h"
> > +#include "intel_display_trace.h"
> > +#include "intel_pmdemand.h"
> > +#include "skl_watermark.h"
> > +
> > +struct intel_pmdemand_state {
> > +  struct intel_global_state base;
> > +
> > +  u16 qclk_gv_bw;
> > +  u8 voltage_index;
> > +  u8 qclk_gv_index;
> > +  u8 active_pipes;
> > +  u8 dbufs;
> > +  u8 active_phys_plls_mask;
> 
> Is u8 enough for the mask? The enum phy shows 9 possible PHY_* members.
> Also, I think having BUILD_BUG_ON() somewhere in this file to make sure
> we have enough bits would be nice.
Thanks. updated.
> 
> > +  u16 cdclk_freq_mhz;
> > +  u16 ddiclk_freq_mhz;
> > +  u8 scalers;
> > +};
> > +
> > +#define to_intel_pmdemand_state(x) container_of((x), struct intel_pmdemand_state, base)
> > +
> > +static struct intel_global_state *
> > +intel_pmdemand_duplicate_state(struct intel_global_obj *obj)
> > +{
> > +  struct intel_pmdemand_state *pmdmnd_state;
> > +
> > +  pmdmnd_state = kmemdup(obj->state, sizeof(*pmdmnd_state), GFP_KERNEL);
> > +  if (!pmdmnd_state)
> > +          return NULL;
> > +
> > +  return &pmdmnd_state->base;
> > +}
> > +
> > +static void intel_pmdemand_destroy_state(struct intel_global_obj *obj,
> > +                                   struct intel_global_state *state)
> > +{
> > +  kfree(state);
> > +}
> > +
> > +static const struct intel_global_state_funcs intel_pmdemand_funcs = {
> > +  .atomic_duplicate_state = intel_pmdemand_duplicate_state,
> > +  .atomic_destroy_state = intel_pmdemand_destroy_state,
> > +};
> > +
> > +static struct intel_pmdemand_state *
> > +intel_atomic_get_pmdemand_state(struct intel_atomic_state *state)
> > +{
> > +  struct drm_i915_private *i915 = to_i915(state->base.dev);
> > +  struct intel_global_state *pmdemand_state;
> > +
> > +  pmdemand_state =
> > +          intel_atomic_get_global_obj_state(state,
> > +                                            &i915->display.pmdemand.obj);
> > +  if (IS_ERR(pmdemand_state))
> > +          return ERR_CAST(pmdemand_state);
> > +
> > +  return to_intel_pmdemand_state(pmdemand_state);
> > +}
> > +
> > +static struct intel_pmdemand_state *
> > +intel_atomic_get_old_pmdemand_state(struct intel_atomic_state *state)
> > +{
> > +  struct drm_i915_private *i915 = to_i915(state->base.dev);
> > +  struct intel_global_state *pmdemand_state;
> > +
> > +  pmdemand_state = intel_atomic_get_old_global_obj_state(state, &i915->display.pmdemand.obj);
> 
> Wouldn't it be safer if we returned early here when pmdemand_state is
> NULL?
> 
> I think to_intel_pmdemand_state(NULL) pmdemand_state just happens to
> work (i.e. still returns NULL) because the "base" member is at the
> beginning of the struct. However, we shouldn't rely on that IMO.

Well. It is a valid point. But the base pointer is the first member for exaclty the reason you
pointed out. So  to prevent someone from accidentally move that "base" from that position, I added a
BUILD_BUG_ON() check. There are few other state objects which lack such a check. I will address that
as a separate patch.

> 
> > +
> > +  return to_intel_pmdemand_state(pmdemand_state);
> > +}
> > +
> > +static struct intel_pmdemand_state *
> > +intel_atomic_get_new_pmdemand_state(struct intel_atomic_state *state)
> > +{
> > +  struct drm_i915_private *i915 = to_i915(state->base.dev);
> > +  struct intel_global_state *pmdemand_state;
> > +
> > +  pmdemand_state = intel_atomic_get_new_global_obj_state(state, &i915->display.pmdemand.obj);
> 
> Just as with intel_atomic_get_old_pmdemand_state(), shouldn't we return
> early if pmdemand_state is NULL here?
> 
> > +
> > +  return to_intel_pmdemand_state(pmdemand_state);
> > +}
> > +
> > +int intel_pmdemand_state_init(struct drm_i915_private *i915)
> > +{
> > +  struct intel_pmdemand_state *pmdemand_state;
> > +
> > +  pmdemand_state = kzalloc(sizeof(*pmdemand_state), GFP_KERNEL);
> > +  if (!pmdemand_state)
> > +          return -ENOMEM;
> > +
> > +  intel_atomic_global_obj_init(i915, &i915->display.pmdemand.obj,
> > +                               &pmdemand_state->base,
> > +                               &intel_pmdemand_funcs);
> > +
> > +
> > +  if (IS_MTL_DISPLAY_STEP(i915, STEP_A0, STEP_C0))
> > +          /* Wa_14016740474 */
> > +          intel_de_rmw(i915, XELPD_CHICKEN_DCPR_3, 0, DMD_RSP_TIMEOUT_DISABLE);
> > +
> > +  return 0;
> > +}
> > +
> > +void intel_pmdemand_init(struct drm_i915_private *i915)
> > +{
> > +  mutex_init(&i915->display.pmdemand.lock);
> > +  init_waitqueue_head(&i915->display.pmdemand.waitqueue);
> > +}
> 
> The functions intel_pmdemand_state_init() and intel_pmdemand_init() are
> both called from the same place. Furthermore,
> intel_pmdemand_state_init() isn't only initializing the state, as the
> Wa_14016740474 workaround is programmed there. Could we have only the
> function intel_pmdemand_init() and incorporate what
> intel_pmdemand_state_init() does in it?

I tried that earlier and wasn't possible. Because the intel_power_domains_init_hw() will call the
pmdemand debuf config. And I can't recall the other issue. But I will confirm this again and update
as I am not able to try this locally.
> 
> > +
> > +static bool pmdemand_needs_update(struct intel_atomic_state *state)
> > +{
> > +  bool changed = false;
> > +  struct intel_crtc *crtc;
> > +  int i;
> > +  const struct intel_bw_state *new_bw_state, *old_bw_state;
> > +  const struct intel_cdclk_state *new_cdclk_state;
> > +  const struct intel_crtc_state *new_crtc_state, *old_crtc_state;
> > +  const struct intel_dbuf_state *new_dbuf_state, *old_dbuf_state;
> > +
> > +  for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
> > +                                      new_crtc_state, i) {
> > +          new_bw_state = intel_atomic_get_new_bw_state(state);
> > +          old_bw_state = intel_atomic_get_old_bw_state(state);
> > +
> > +          new_dbuf_state = intel_atomic_get_new_dbuf_state(state);
> > +          old_dbuf_state = intel_atomic_get_old_dbuf_state(state);
> > +
> > +          new_cdclk_state = intel_atomic_get_new_cdclk_state(state);
> > +
> > +          if ((new_bw_state && new_bw_state->qgv_point_peakbw !=
> > +               old_bw_state->qgv_point_peakbw) ||
> > +              (new_dbuf_state && new_dbuf_state->active_pipes !=
> > +               old_dbuf_state->active_pipes) || new_cdclk_state)
> > +                  changed = true;
> > +
> > +          /*
> > +           * break needs to be removed, if some crtc_state dependent
> > +           * parameters are added here
> > +           */
> > +          break;
> > +  }
> > +
> > +  return changed;
> > +}
> > +
> > +int intel_pmdemand_atomic_check(struct intel_atomic_state *state)
> > +{
> > +  struct drm_i915_private *i915 = to_i915(state->base.dev);
> > +  int port_clock = 0;
> > +  struct intel_crtc *crtc;
> > +  struct intel_encoder *encoder;
> > +  const struct intel_bw_state *new_bw_state;
> > +  const struct intel_cdclk_state *new_cdclk_state;
> > +  const struct intel_crtc_state *new_crtc_state, *old_crtc_state;
> > +  const struct intel_dbuf_state *new_dbuf_state;
> > +  struct intel_pmdemand_state *new_pmdemand_state;
> > +  enum phy phy;
> > +  int i, ret;
> > +
> > +  if (DISPLAY_VER(i915) < 14)
> > +          return 0;
> > +
> > +  if (!pmdemand_needs_update(state))
> > +          return 0;
> > +
> > +  new_pmdemand_state = intel_atomic_get_pmdemand_state(state);
> > +  if (IS_ERR(new_pmdemand_state))
> > +          return PTR_ERR(new_pmdemand_state);
> > +
> > +  ret = intel_atomic_lock_global_state(&new_pmdemand_state->base);
> > +  if (ret)
> > +          return ret;
> > +
> > +  /* Punit figures out the voltage index based on bandwidth*/
> > +  new_bw_state = intel_atomic_get_bw_state(state);
> > +  if (IS_ERR(new_bw_state))
> > +          return PTR_ERR(new_bw_state);
> > +
> > +  /* firmware will calculate the qclck_gc_index, requirement is set to 0 */
> > +  new_pmdemand_state->qclk_gv_index = 0;
> > +  new_pmdemand_state->qclk_gv_bw =
> > +          min_t(u16, new_bw_state->qgv_point_peakbw, 0xffff);
> > +
> > +  new_dbuf_state = intel_atomic_get_dbuf_state(state);
> > +  if (IS_ERR(new_dbuf_state))
> > +          return PTR_ERR(new_dbuf_state);
> > +
> > +  i = hweight8(new_dbuf_state->active_pipes);
> > +  new_pmdemand_state->active_pipes = min(i, 3);
> > +
> > +  new_cdclk_state = intel_atomic_get_cdclk_state(state);
> > +  if (IS_ERR(new_cdclk_state))
> > +          return PTR_ERR(new_cdclk_state);
> > +
> > +  new_pmdemand_state->voltage_index =
> > +          new_cdclk_state->logical.voltage_level;
> > +  /* KHz to MHz */
> > +  new_pmdemand_state->cdclk_freq_mhz =
> > +          DIV_ROUND_UP(new_cdclk_state->logical.cdclk, 1000);
> > +
> > +  new_pmdemand_state->active_phys_plls_mask = 0;
> > +
> > +  for_each_oldnew_intel_crtc_in_state(state, crtc, old_crtc_state,
> > +                                      new_crtc_state, i) {
> > +          if (!new_crtc_state->hw.active)
> > +                  continue;
> > +
> > +          encoder = intel_get_crtc_new_encoder(state, new_crtc_state);
> > +          if (!encoder)
> > +                  continue;
> > +
> > +          phy = intel_port_to_phy(i915, encoder->port);
> > +
> > +          if (intel_is_c10phy(i915, phy))
> > +                  new_pmdemand_state->active_phys_plls_mask |= BIT(phy);
> > +
> > +          port_clock = max(port_clock, new_crtc_state->port_clock);
> > +  }
> 
> As previously noted in https://patchwork.freedesktop.org/patch/530495 ,
> I'm under the impression that this loop would not let us account for al
> active crtcs, only those currently being touched by this atomic
> transaction. Am I wrong to assume that
> for_each_oldnew_intel_crtc_in_state() would only iterate over crtcs
> touched by the atomic update?

I checked the intel_bw_check_data_rate() which should be doing something similar to find out about
the datarate from eachi pipe and figure out the total data rate. So I thought this should be
sufficient. 

> 
> > +
> > +  /* To MHz */
> > +  new_pmdemand_state->ddiclk_freq_mhz = DIV_ROUND_UP(port_clock, 1000);
> > +
> > +  /*
> > +   * Setting scalers to max as it can not be calculated during flips and
> > +   * fastsets without taking global states locks.
> > +   */
> > +  new_pmdemand_state->scalers = 7;
> > +
> > +  ret = intel_atomic_serialize_global_state(&new_pmdemand_state->base);
> > +  if (ret)
> > +          return ret;
> > +
> > +  return 0;
> > +}
> > +
> > +static bool intel_pmdemand_check_prev_transaction(struct drm_i915_private *i915)
> > +{
> > +  return !((intel_de_read(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(1)) &
> > +            XELPDP_PMDEMAND_REQ_ENABLE) ||
> > +          (intel_de_read(i915, GEN12_DCPR_STATUS_1) &
> > +           XELPDP_PMDEMAND_INFLIGHT_STATUS));
> > +}
> > +
> > +static bool intel_pmdemand_req_complete(struct drm_i915_private *i915)
> > +{
> > +  return !(intel_de_read(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(1)) &
> > +           XELPDP_PMDEMAND_REQ_ENABLE);
> > +}
> > +
> > +static int intel_pmdemand_wait(struct drm_i915_private *i915)
> > +{
> > +  DEFINE_WAIT(wait);
> > +  int ret;
> > +  const unsigned int timeout_ms = 10;
> > +
> > +  ret = wait_event_timeout(i915->display.pmdemand.waitqueue,
> > +                           intel_pmdemand_req_complete(i915),
> > +                           msecs_to_jiffies_timeout(timeout_ms));
> > +  if (ret == 0)
> > +          drm_err(&i915->drm,
> > +                  "timed out waiting for Punit PM Demand Response\n");
> > +
> > +  return ret;
> > +}
> > +
> > +/* Required to be programmed during Display Init Sequences. */
> > +void intel_program_dbuf_pmdemand(struct drm_i915_private *i915,
> > +                           u8 dbuf_slices)
> > +{
> > +  u32 dbufs = min_t(u32, hweight8(dbuf_slices), 3);
> > +
> > +  mutex_lock(&i915->display.pmdemand.lock);
> > +  if (drm_WARN_ON(&i915->drm,
> > +                  !intel_pmdemand_check_prev_transaction(i915)))
> > +          goto unlock;
> > +
> > +  intel_de_rmw(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(0),
> > +               XELPDP_PMDEMAND_DBUFS_MASK, XELPDP_PMDEMAND_DBUFS(dbufs));
> > +  intel_de_rmw(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(1), 0,
> > +               XELPDP_PMDEMAND_REQ_ENABLE);
> > +
> > +  intel_pmdemand_wait(i915);
> > +
> > +unlock:
> > +  mutex_unlock(&i915->display.pmdemand.lock);
> > +}
> > +
> > +static void update_pmdemand_values(const struct intel_pmdemand_state *new,
> > +                             const struct intel_pmdemand_state *old,
> > +                             u32 *reg1, u32 *reg2)
> > +{
> > +  u32 plls, tmp;
> > +
> > +  /*
> > +   * The pmdemand parameter updates happens in two steps. Pre plane and
> > +   * post plane updates. During the pre plane, as DE might still be
> > +   * handling with some old operations, to avoid unwanted performance
> > +   * issues, program the pmdemand parameters with higher of old and new
> > +   * values. And then after once settled, use the new parameter values
> > +   * as part of the post plane update.
> > +   */
> > +
> > +  /* Set 1*/
> > +  *reg1 &= ~XELPDP_PMDEMAND_QCLK_GV_BW_MASK;
> > +  tmp = old ? max(old->qclk_gv_bw, new->qclk_gv_bw) : new->qclk_gv_bw;
> > +  *reg1 |= XELPDP_PMDEMAND_QCLK_GV_BW(tmp);
> > +
> > +  *reg1 &= ~XELPDP_PMDEMAND_VOLTAGE_INDEX_MASK;
> > +  tmp = old ? max(old->voltage_index, new->voltage_index) :
> > +              new->voltage_index;
> > +  *reg1 |= XELPDP_PMDEMAND_VOLTAGE_INDEX(tmp);
> > +
> > +  *reg1 &= ~XELPDP_PMDEMAND_QCLK_GV_INDEX_MASK;
> > +  tmp = old ? max(old->qclk_gv_index, new->qclk_gv_index) :
> > +              new->qclk_gv_index;
> > +  *reg1 |= XELPDP_PMDEMAND_QCLK_GV_INDEX(tmp);
> > +
> > +  *reg1 &= ~XELPDP_PMDEMAND_PIPES_MASK;
> > +  tmp = old ? max(old->active_pipes, new->active_pipes) :
> > +              new->active_pipes;
> > +  *reg1 |= XELPDP_PMDEMAND_PIPES(tmp);
> > +
> > +  *reg1 &= ~XELPDP_PMDEMAND_PHYS_MASK;
> > +  plls = hweight32(new->active_phys_plls_mask);
> > +  if (old)
> > +          plls = max(plls, hweight32(old->active_phys_plls_mask));
> > +  *reg1 |= XELPDP_PMDEMAND_PHYS(plls);
> 
> If plls > 7, we would be potentially programming this wrong (e.g. for
> plls=8, we would setting the field to 0).

Thanks for pointing that out. Fixed.

> 
> > +
> > +  /* Set 2*/
> > +  *reg2 &= ~XELPDP_PMDEMAND_CDCLK_FREQ_MASK;
> > +  tmp = old ? max(old->cdclk_freq_mhz, new->cdclk_freq_mhz) :
> > +              new->cdclk_freq_mhz;
> > +  *reg2 |= XELPDP_PMDEMAND_CDCLK_FREQ(tmp);
> > +
> > +  *reg2 &= ~XELPDP_PMDEMAND_DDICLK_FREQ_MASK;
> > +  tmp = old ? max(old->ddiclk_freq_mhz, new->ddiclk_freq_mhz) :
> > +              new->ddiclk_freq_mhz;
> > +  *reg2 |= XELPDP_PMDEMAND_DDICLK_FREQ(tmp);
> > +
> > +  /* Hard code scalers to 7*/
> 
> I think this comment can be dropped: the hardcoding happens in
> intel_pmdemand_atomic_check().

Done
> 
> > +  *reg2 &= ~XELPDP_PMDEMAND_SCALERS_MASK;
> > +  tmp = old ? max(old->scalers, new->scalers) : new->scalers;
> > +  *reg2 |= XELPDP_PMDEMAND_SCALERS(tmp);
> > +
> > +  /*
> > +   * Active_PLLs starts with 1 because of CDCLK PLL.
> > +   * TODO: Missing to account genlock filter when it gets used.
> > +   */
> > +  *reg2 &= ~XELPDP_PMDEMAND_PLLS_MASK;
> 
> I think we are missing the ternary operator here to select the maximum
> value for the pre-plane case.
As pet the above comments, it is just the plls from step earlier + 1

> 
> > +  *reg2 |= XELPDP_PMDEMAND_PLLS(plls + 1);
> > +}
> > +
> > +static void intel_program_pmdemand(struct drm_i915_private *i915,
> > +                             const struct intel_pmdemand_state *new,
> > +                             const struct intel_pmdemand_state *old)
> > +{
> > +  bool changed = false;
> > +  u32 reg1, mod_reg1;
> > +  u32 reg2, mod_reg2;
> > +
> > +  mutex_lock(&i915->display.pmdemand.lock);
> > +  if (drm_WARN_ON(&i915->drm,
> > +                  !intel_pmdemand_check_prev_transaction(i915)))
> > +          goto unlock;
> 
> According to the spec, we should wait and timeout after 10ms here.
> 
> > +
> > +  reg1 = intel_de_read(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(0));
> > +  mod_reg1 = reg1;
> > +
> > +  reg2 = intel_de_read(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(1));
> > +  mod_reg2 = reg2;
> > +
> > +  update_pmdemand_values(new, old, &mod_reg1, &mod_reg2);
> > +
> > +  if (reg1 != mod_reg1) {
> > +          intel_de_write(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(0),
> > +                         mod_reg1);
> > +          changed = true;
> > +  }
> > +
> > +  if (reg2 != mod_reg2) {
> > +          intel_de_write(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(1),
> > +                         mod_reg2);
> > +          changed = true;
> > +  }
> > +
> > +  /* Initiate pm demand request only if register values are changed */
> > +  if (changed) {
> 
> Nitpick: we could have
> 
>     if (!changed)
>             goto unlock;
> 
> and dedent the block below.

Done.
> 
> > +          drm_dbg_kms(&i915->drm,
> > +                      "initate pmdemand request values: (0x%x 0x%x)\n",
> > +                      mod_reg1, mod_reg2);
> > +
> > +          intel_de_rmw(i915, XELPDP_INITIATE_PMDEMAND_REQUEST(1), 0,
> > +                       XELPDP_PMDEMAND_REQ_ENABLE);
> > +
> > +          intel_pmdemand_wait(i915);
> > +  }
> > +
> > +unlock:
> > +  mutex_unlock(&i915->display.pmdemand.lock);
> > +}
> > +
> > +static bool
> > +intel_pmdemand_state_changed(const struct intel_pmdemand_state *new,
> > +                       const struct intel_pmdemand_state *old)
> > +{
> > +  return memcmp(&new->qclk_gv_bw, &old->qclk_gv_bw,
> > +                sizeof(*new) - offsetof(typeof(*new), qclk_gv_bw)) != 0;
> > +}
> > +
> > +void intel_pmdemand_pre_plane_update(struct intel_atomic_state *state)
> > +{
> > +  struct drm_i915_private *i915 = to_i915(state->base.dev);
> > +  const struct intel_pmdemand_state *new_pmdmnd_state =
> > +          intel_atomic_get_new_pmdemand_state(state);
> > +  const struct intel_pmdemand_state *old_pmdmnd_state =
> > +          intel_atomic_get_old_pmdemand_state(state);
> > +
> > +  if (DISPLAY_VER(i915) < 14)
> > +          return;
> > +
> > +  if (!new_pmdmnd_state ||
> > +      !intel_pmdemand_state_changed(new_pmdmnd_state, old_pmdmnd_state))
> > +          return;
> > +
> > +  intel_program_pmdemand(i915, new_pmdmnd_state, old_pmdmnd_state);
> > +}
> > +
> > +void intel_pmdemand_post_plane_update(struct intel_atomic_state *state)
> > +{
> > +  struct drm_i915_private *i915 = to_i915(state->base.dev);
> > +  const struct intel_pmdemand_state *new_pmdmnd_state =
> > +          intel_atomic_get_new_pmdemand_state(state);
> > +  const struct intel_pmdemand_state *old_pmdmnd_state =
> > +          intel_atomic_get_old_pmdemand_state(state);
> > +
> > +  if (DISPLAY_VER(i915) < 14)
> > +          return;
> > +
> > +  if (!new_pmdmnd_state ||
> > +      !intel_pmdemand_state_changed(new_pmdmnd_state, old_pmdmnd_state))
> > +          return;
> > +
> > +  intel_program_pmdemand(i915, new_pmdmnd_state, NULL);
> > +}
> > diff --git a/drivers/gpu/drm/i915/display/intel_pmdemand.h
> > b/drivers/gpu/drm/i915/display/intel_pmdemand.h
> > new file mode 100644
> > index 000000000000..0114f4e0225a
> > --- /dev/null
> > +++ b/drivers/gpu/drm/i915/display/intel_pmdemand.h
> > @@ -0,0 +1,24 @@
> > +/* SPDX-License-Identifier: MIT */
> > +/*
> > + * Copyright © 2023 Intel Corporation
> > + */
> > +
> > +#ifndef __INTEL_PMDEMAND_H__
> > +#define __INTEL_PMDEMAND_H__
> > +
> > +#include <linux/types.h>
> > +
> > +struct drm_i915_private;
> > +struct intel_atomic_state;
> > +struct intel_crtc_state;
> > +struct intel_plane_state;
> > +
> > +void intel_pmdemand_init(struct drm_i915_private *i915);
> > +int intel_pmdemand_state_init(struct drm_i915_private *i915);
> > +void intel_program_dbuf_pmdemand(struct drm_i915_private *i915,
> > +                           u8 dbuf_slices);
> > +void intel_pmdemand_pre_plane_update(struct intel_atomic_state *state);
> > +void intel_pmdemand_post_plane_update(struct intel_atomic_state *state);
> > +int intel_pmdemand_atomic_check(struct intel_atomic_state *state);
> > +
> > +#endif /* __INTEL_PMDEMAND_H__ */
> > diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
> > index 2b94b8ca8ec9..907fa3aee179 100644
> > --- a/drivers/gpu/drm/i915/i915_irq.c
> > +++ b/drivers/gpu/drm/i915/i915_irq.c
> > @@ -41,6 +41,7 @@
> > #include "display/intel_fifo_underrun.h"
> > #include "display/intel_hotplug.h"
> > #include "display/intel_lpe_audio.h"
> > +#include "display/intel_pmdemand.h"
> > #include "display/intel_psr.h"
> > #include "display/intel_psr_regs.h"
> > 
> > @@ -1986,12 +1987,25 @@ static u32 gen8_de_pipe_fault_mask(struct drm_i915_private *dev_priv)
> >                return GEN8_DE_PIPE_IRQ_FAULT_ERRORS;
> > }
> > 
> > +static void intel_pmdemand_irq_handler(struct drm_i915_private *dev_priv)
> > +{
> > +  wake_up_all(&dev_priv->display.pmdemand.waitqueue);
> > +}
> > +
> > static void
> > gen8_de_misc_irq_handler(struct drm_i915_private *dev_priv, u32 iir)
> > {
> >        bool found = false;
> > 
> > -  if (iir & GEN8_DE_MISC_GSE) {
> > +  if (DISPLAY_VER(dev_priv) >= 14 &&
> > +      (iir & (XELPDP_PMDEMAND_RSP | XELPDP_PMDEMAND_RSPTOUT_ERR))) {
> > +          if (iir & XELPDP_PMDEMAND_RSPTOUT_ERR)
> 
> I think we should have the (iir & (XELPDP_PMDEMAND_RSP |
> XELPDP_PMDEMAND_RSPTOUT_ERR)) part as nested if statement here.
> Otherwise, when the interrupt did not happen, we could endup checking
> for the GEN8_DE_MISC_GSE even when DISPLAY_VER(dev_priv) >= 14.
> 
> Even though we know that iir & GEN8_DE_MISC_GSE would be false in this
> situation (because both XELPDP_PMDEMAND_RSPTOUT_ERR and GEN8_DE_MISC_GSE
> map to the same bit), I think having that one checked only for previous
> display engines would sound more correct semantically speaking.

Thanks. updated.

> 
> > +                  drm_dbg(&dev_priv->drm,
> > +                          "Error waiting for Punit PM Demand Response\n");
> > +
> > +          intel_pmdemand_irq_handler(dev_priv);
> > +          found = true;
> > +  } else if (iir & GEN8_DE_MISC_GSE) {
> >                intel_opregion_asle_intr(dev_priv);
> >                found = true;
> >        }
> > @@ -3742,7 +3756,10 @@ static void gen8_de_irq_postinstall(struct drm_i915_private *dev_priv)
> >        if (IS_GEMINILAKE(dev_priv) || IS_BROXTON(dev_priv))
> >                de_port_masked |= BXT_DE_PORT_GMBUS;
> > 
> > -  if (DISPLAY_VER(dev_priv) >= 11) {
> > +  if (DISPLAY_VER(dev_priv) >= 14)
> > +          de_misc_masked |= XELPDP_PMDEMAND_RSPTOUT_ERR |
> > +                            XELPDP_PMDEMAND_RSP;
> > +  else if (DISPLAY_VER(dev_priv) >= 11) {
> >                enum port port;
> > 
> >                if (intel_bios_is_dsi_present(dev_priv, &port))
> > diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
> > index dde6e91055bd..60c007aea1ce 100644
> > --- a/drivers/gpu/drm/i915/i915_reg.h
> > +++ b/drivers/gpu/drm/i915/i915_reg.h
> > @@ -4426,8 +4426,10 @@
> > #define GEN8_DE_MISC_IMR _MMIO(0x44464)
> > #define GEN8_DE_MISC_IIR _MMIO(0x44468)
> > #define GEN8_DE_MISC_IER _MMIO(0x4446c)
> > -#define  GEN8_DE_MISC_GSE         (1 << 27)
> > -#define  GEN8_DE_EDP_PSR          (1 << 19)
> > +#define  XELPDP_PMDEMAND_RSPTOUT_ERR      REG_BIT(27)
> > +#define  GEN8_DE_MISC_GSE         REG_BIT(27)
> > +#define  GEN8_DE_EDP_PSR          REG_BIT(19)
> > +#define  XELPDP_PMDEMAND_RSP              REG_BIT(3)
> > 
> > #define GEN8_PCU_ISR _MMIO(0x444e0)
> > #define GEN8_PCU_IMR _MMIO(0x444e4)
> > @@ -4512,6 +4514,33 @@
> > #define  XELPDP_DP_ALT_HPD_LONG_DETECT         REG_BIT(1)
> > #define  XELPDP_DP_ALT_HPD_SHORT_DETECT                REG_BIT(0)
> > 
> > +#define XELPDP_INITIATE_PMDEMAND_REQUEST(dword)           _MMIO(0x45230 + 4 * (dword))
> > +#define  XELPDP_PMDEMAND_QCLK_GV_BW_MASK          REG_GENMASK(31, 16)
> > +#define  XELPDP_PMDEMAND_QCLK_GV_BW(x)                   
> > REG_FIELD_PREP(XELPDP_PMDEMAND_QCLK_GV_BW_MASK, x)
> > +#define  XELPDP_PMDEMAND_VOLTAGE_INDEX_MASK               REG_GENMASK(14, 12)
> > +#define  XELPDP_PMDEMAND_VOLTAGE_INDEX(x)        
> > REG_FIELD_PREP(XELPDP_PMDEMAND_VOLTAGE_INDEX_MASK, x)
> > +#define  XELPDP_PMDEMAND_QCLK_GV_INDEX_MASK               REG_GENMASK(11, 8)
> > +#define  XELPDP_PMDEMAND_QCLK_GV_INDEX(x)        
> > REG_FIELD_PREP(XELPDP_PMDEMAND_QCLK_GV_INDEX_MASK, x)
> > +#define  XELPDP_PMDEMAND_PIPES_MASK                       REG_GENMASK(7, 6)
> > +#define  XELPDP_PMDEMAND_PIPES(x)                 REG_FIELD_PREP(XELPDP_PMDEMAND_PIPES_MASK, x)
> > +#define  XELPDP_PMDEMAND_DBUFS_MASK                       REG_GENMASK(5, 4)
> > +#define  XELPDP_PMDEMAND_DBUFS(x)                 REG_FIELD_PREP(XELPDP_PMDEMAND_DBUFS_MASK, x)
> > +#define  XELPDP_PMDEMAND_PHYS_MASK                        REG_GENMASK(2, 0)
> > +#define  XELPDP_PMDEMAND_PHYS(x)                  REG_FIELD_PREP(XELPDP_PMDEMAND_PHYS_MASK, x)
> > +
> > +#define  XELPDP_PMDEMAND_REQ_ENABLE                       REG_BIT(31)
> > +#define  XELPDP_PMDEMAND_CDCLK_FREQ_MASK          REG_GENMASK(30, 20)
> > +#define  XELPDP_PMDEMAND_CDCLK_FREQ(x)                   
> > REG_FIELD_PREP(XELPDP_PMDEMAND_CDCLK_FREQ_MASK, x)
> > +#define  XELPDP_PMDEMAND_DDICLK_FREQ_MASK         REG_GENMASK(18, 8)
> > +#define  XELPDP_PMDEMAND_DDICLK_FREQ(x)                  
> > REG_FIELD_PREP(XELPDP_PMDEMAND_DDICLK_FREQ_MASK, x)
> > +#define  XELPDP_PMDEMAND_SCALERS_MASK                     REG_GENMASK(6, 4)
> > +#define  XELPDP_PMDEMAND_SCALERS(x)                      
> > REG_FIELD_PREP(XELPDP_PMDEMAND_SCALERS_MASK, x)
> > +#define  XELPDP_PMDEMAND_PLLS_MASK                        REG_GENMASK(2, 0)
> > +#define  XELPDP_PMDEMAND_PLLS(x)                  REG_FIELD_PREP(XELPDP_PMDEMAND_PLLS_MASK, x)
> > +
> > +#define GEN12_DCPR_STATUS_1                               _MMIO(0x46440)
> > +#define  XELPDP_PMDEMAND_INFLIGHT_STATUS          REG_BIT(26)
> > +
> > #define ILK_DISPLAY_CHICKEN2   _MMIO(0x42004)
> > /* Required on all Ironlake and Sandybridge according to the B-Spec. */
> > #define   ILK_ELPIN_409_SELECT REG_BIT(25)
> > @@ -4671,6 +4700,9 @@
> > #define   DCPR_SEND_RESP_IMM                   REG_BIT(25)
> > #define   DCPR_CLEAR_MEMSTAT_DIS               REG_BIT(24)
> > 
> > +#define XELPD_CHICKEN_DCPR_3                      _MMIO(0x46438)
> > +#define   DMD_RSP_TIMEOUT_DISABLE         REG_BIT(19)
> > +
> > #define SKL_DFSM                       _MMIO(0x51000)
> > #define   SKL_DFSM_DISPLAY_PM_DISABLE  (1 << 27)
> > #define   SKL_DFSM_DISPLAY_HDCP_DISABLE        (1 << 25)
> > -- 
> > 2.34.1
> > 



More information about the Intel-gfx mailing list