drm/i915: Make sandybridge_pcode_read() deal with the second data register
Subsystems (2)
PATCH v3 1/2 drm/i915: Make sandybridge_pcode_read() deal with the second data register
From: Ville Syrjälä <[email protected]>
The pcode mailbox has two data registers. So far we've only ever used the one, but that's about to change. Expose the second data register to the callers of sandybridge_pcode_read().
--- drivers/gpu/drm/i915/i915_debugfs.c | 4 ++-- drivers/gpu/drm/i915/intel_pm.c | 12 +++++++----- drivers/gpu/drm/i915/intel_sideband.c | 15 +++++++++------ drivers/gpu/drm/i915/intel_sideband.h | 3 ++- 4 files changed, 20 insertions(+), 14 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +2 -2
@@ -1494,7 +1494,7 @@ static int gen6_drpc_info(struct seq_file *m)if (INTEL_GEN (dev_priv) <=7 )sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , - &rc6vids); + &rc6vids,NULL );seq_printf (m,"RC1e Enabled: %s\n" ,yesno (rcctl1 & GEN6_RC_CTL_RC1e_ENABLE));
@@ -1777,7 +1777,7 @@ static int i915_ring_freq_table(struct seq_file *m, void *unused) ia_freq = gpu_freq;sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_MIN_FREQ_TABLE , - &ia_freq); + &ia_freq,NULL ); seq_printf(m, "%d\t\t%d\t\t\t\t%d\n", intel_gpu_freq(dev_priv, (gpu_freq * (IS_GEN9_BC(dev_priv) ||
drivers/gpu/drm/i915/intel_pm.c +7 -5
@@ -2822,7 +2822,7 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv, val =0 ;/* data0 to be programmed to 0 for first set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENCY , - &val); + &val,NULL );if (ret) {DRM_ERROR ("SKL Mailbox read error = %d\n" , ret);
@@ -2841,7 +2841,7 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv, val =1 ;/* data0 to be programmed to 1 for second set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENCY , - &val); + &val,NULL );if (ret) {DRM_ERROR ("SKL Mailbox read error = %d\n" , ret);return ;
@@ -7061,7 +7061,7 @@ static void gen6_init_rps_frequencies(struct drm_i915_private *dev_priv)if (sandybridge_pcode_read (dev_priv,HSW_PCODE_DYNAMIC_DUTY_CYCLE_CONTROL , - &ddcc_status) ==0 ) + &ddcc_status,NULL ) ==0 ) rps->efficient_freq = clamp_t(u8, ((ddcc_status >> 8) & 0xff),
@@ -7408,7 +7408,8 @@ static void gen6_enable_rc6(struct drm_i915_private *dev_priv) GEN6_RC_CTL_HW_ENABLE); rc6vids =0 ; - ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , &rc6vids); + ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , + &rc6vids,NULL );if (IS_GEN (dev_priv,6 ) && ret) {DRM_DEBUG_DRIVER ("Couldn't check for BIOS workaround\n" ); }else if (IS_GEN (dev_priv,6 ) && (GEN6_DECODE_RC6_VID (rc6vids &0xff ) <450 )) {
@@ -8555,7 +8556,8 @@ void intel_init_gt_powersave(struct drm_i915_private *dev_priv) IS_IVYBRIDGE(dev_priv) || IS_HASWELL(dev_priv)) {u32 params =0 ; -sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , ¶ms); +sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , + ¶ms,NULL );if (params &BIT (31 )) {/* OC supported */ DRM_DEBUG_DRIVER("Overclocking supported, max: %dMHz, overclock: %dMHz\n", (rps->max_freq & 0xff) * 50,
drivers/gpu/drm/i915/intel_sideband.c +9 -6
@@ -374,7 +374,7 @@ static inline int gen7_check_mailbox_status(u32 mbox) }static int __sandybridge_pcode_rw (struct drm_i915_private *i915, -u32 mbox,u32 *val, +u32 mbox,u32 *val,u32 *val1,int fast_timeout_us,int slow_timeout_ms,bool is_read)
@@ -393,7 +393,7 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,return -EAGAIN ;intel_uncore_write_fw (uncore,GEN6_PCODE_DATA , *val); -intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 ,0 ); +intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 , val1 ? *val1 :0 );intel_uncore_write_fw (uncore,GEN6_PCODE_MAILBOX ,GEN6_PCODE_READY | mbox);
@@ -407,6 +407,8 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,if (is_read) *val =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA ); +if (is_read && val1) + *val1 =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA1 );if (INTEL_GEN (i915) >6 )return gen7_check_mailbox_status (mbox);
@@ -414,12 +416,13 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,return gen6_check_mailbox_status (mbox); } -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox,u32 *val) +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1) {int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, val, + err =__sandybridge_pcode_rw (i915, mbox, val, val1,500 ,0 , true);mutex_unlock (&i915->sb_lock);
@@ -440,7 +443,7 @@ int sandybridge_pcode_write_timeout(struct drm_i915_private *i915,int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, &val, + err =__sandybridge_pcode_rw (i915, mbox, &val,NULL , fast_timeout_us, slow_timeout_ms, false);mutex_unlock (&i915->sb_lock);
@@ -457,7 +460,7 @@ static bool skl_pcode_try_request(struct drm_i915_private *i915, u32 mbox, u32 request, u32 reply_mask, u32 reply, u32 *status) { - *status =__sandybridge_pcode_rw (i915, mbox, &request, + *status =__sandybridge_pcode_rw (i915, mbox, &request,NULL ,500 ,0 , true);
drivers/gpu/drm/i915/intel_sideband.h +2 -1
@@ -127,7 +127,8 @@ u32 intel_sbi_read(struct drm_i915_private *i915, u16 reg,void intel_sbi_write (struct drm_i915_private *i915,u16 reg,u32 value,enum intel_sbi_destination destination); -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox,u32 *val); +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1);int sandybridge_pcode_write_timeout (struct drm_i915_private *i915,u32 mbox,u32 val,int fast_timeout_us,int slow_timeout_ms);
Signature · 6 lines
2.21.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 2/2 drm/i915: Make sure we have enough memory bandwidth on ICL
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbers
--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/intel_bw.c create mode 100644 drivers/gpu/drm/i915/intel_bw.h
drivers/gpu/drm/i915/Makefile +1 -0
@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \
drivers/gpu/drm/i915/i915_drv.c +229 -0
247 lines not on this page · raw message
drivers/gpu/drm/i915/i915_drv.h +10 -0
@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h>
@@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {
@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select (struct drm_i915_private *dev_priv);
drivers/gpu/drm/i915/i915_reg.h +3 -0
@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((point) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define HSW_PCODE_DE_WRITE_FREQ_REQ 0x17
drivers/gpu/drm/i915/intel_atomic_plane.c +20 -0
@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); } +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state) +{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i]; + +return cpp * crtc_state->pixel_rate; +} + int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_state, struct intel_crtc_state *new_crtc_state, const struct intel_plane_state *old_plane_state,
@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_ new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)
@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_if (new_plane_state->base.visible || old_plane_state->base.visible) new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,
drivers/gpu/drm/i915/intel_atomic_plane.h +2 -0
@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state);void intel_update_plane (struct intel_plane *plane,const struct intel_crtc_state *crtc_state,const struct intel_plane_state *plane_state);
drivers/gpu/drm/i915/intel_bw.c new file +181 -0
181 lines not on this page · raw message
drivers/gpu/drm/i915/intel_bw.h new file +46 -0
46 lines not on this page · raw message
drivers/gpu/drm/i915/intel_display.c +39 -1
@@ -50,6 +50,7 @@#include "intel_acpi.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" +#include "intel_bw.h" #include "intel_color.h" #include "intel_cdclk.h" #include "intel_crt.h"
@@ -2863,6 +2864,7 @@ static void intel_plane_disable_noatomic(struct intel_crtc *crtc,intel_set_plane_visible (crtc_state, plane_state, false);fixup_active_planes (crtc_state); + crtc_state->data_rate[plane->id] =0 ;if (plane->id ==PLANE_PRIMARY )intel_pre_disable_primary_noatomic (&crtc->base);
@@ -6590,6 +6592,8 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc,struct intel_encoder *encoder;struct intel_crtc *intel_crtc =to_intel_crtc (crtc);struct drm_i915_private *dev_priv =to_i915 (crtc->dev); +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);enum intel_display_power_domain domain;struct intel_plane *plane;u64 domains;
@@ -6652,6 +6656,9 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc, dev_priv->active_crtcs &= ~(1 << intel_crtc->pipe); dev_priv->min_cdclk[intel_crtc->pipe] =0 ; dev_priv->min_voltage_level[intel_crtc->pipe] =0 ; + + bw_state->data_rate[intel_crtc->pipe] =0 ; + bw_state->num_active_planes[intel_crtc->pipe] =0 ; }/*
@@ -11176,6 +11183,7 @@ int intel_plane_atomic_calc_changes(const struct intel_crtc_state *old_crtc_statif (!is_crtc_enabled) { plane_state->visible = visible = false;to_intel_crtc_state (crtc_state)->active_planes &= ~BIT (plane->id); +to_intel_crtc_state (crtc_state)->data_rate[plane->id] =0 ; }if (!was_visible && !visible)
@@ -13296,7 +13304,15 @@ static int intel_atomic_check(struct drm_device *dev,return ret;intel_fbc_choose_crtc (dev_priv, intel_state); -return calc_watermark_data (intel_state); + ret =calc_watermark_data (intel_state); +if (ret) +return ret; + + ret =intel_bw_atomic_check (intel_state); +if (ret) +return ret; + +return 0 ; } static int intel_atomic_prepare_commit(struct drm_device *dev,
@@ -15696,6 +15712,10 @@ int intel_modeset_init(struct drm_device *dev)drm_mode_config_init (dev); + ret =intel_bw_init (dev_priv); +if (ret) +return ret; + dev->mode_config.min_width =0 ; dev->mode_config.min_height =0 ;
@@ -16318,8 +16338,11 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)drm_connector_list_iter_end (&conn_iter);for_each_intel_crtc (dev, crtc) { +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);struct intel_crtc_state *crtc_state =to_intel_crtc_state (crtc->base.state); +struct intel_plane *plane;int min_cdclk =0 ;memset (&crtc->base.mode,0 ,sizeof (crtc->base.mode));
@@ -16358,6 +16381,21 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) dev_priv->min_voltage_level[crtc->pipe] = crtc_state->min_voltage_level; +for_each_intel_plane_on_crtc (&dev_priv->drm, crtc, plane) { +const struct intel_plane_state *plane_state = +to_intel_plane_state (plane->base.state); + +/* +* FIXME don't have the fb yet, so can't +* use intel_plane_data_rate() :( +*/ +if (plane_state->base.visible) + crtc_state->data_rate[plane->id] = +4 * crtc_state->pixel_rate; + } + +intel_bw_crtc_update (bw_state, crtc_state); +intel_pipe_config_sanity_check (dev_priv, crtc_state); } }
drivers/gpu/drm/i915/intel_drv.h +2 -0
@@ -885,6 +885,8 @@ struct intel_crtc_state {struct intel_crtc_wm_state wm; +u32 data_rate[I915_MAX_PLANES ]; +/* Gamma mode programmed on the pipe */ u32 gamma_mode;
Signature · 6 lines
2.21.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
Quoted 23 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
Quoted 1 line
TODO: make it robust and deal with all the open questions
TODO: Add comments detailing structures
Quoted 166 lines
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-)create mode 100644
drivers/gpu/drm/i915/intel_bw.ccreate mode 100644drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +115
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ; + + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point) +{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point), + &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + +DRM_DEBUG_KMS ("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%d tRCD=%d tRC=%d\n" , + i, sp->dclk, sp->t_rp, sp->t_rdpre, sp->t_ras, + sp->t_rcd, sp->t_rc); + } + +return 0 ; +} + +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i;
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1562
Quoted 10 lines
drivers/gpu/drm/i915/i915_drv.c +10
-0+ +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +};
intel_sa_info? Doesn't seem very descriptive. also very close to intel_sagv_info
Quoted 508 lines
drivers/gpu/drm/i915/i915_drv.c +104
-0+ +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave / num_channels) << i; + bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1 ; + +for (j =0 ; j < si.num_points; j++) { +const struct intel_qgv_point *sp = &si.points[j]; +int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the +* assumed burst length? +*/ + ct =max_t (int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup -1 ) * si.t_bl + sp->t_rdpre); + bw =icl_calc_bw (sp->dclk, clpchgroup *32 * num_channels, ct); + + bi->deratedbw[j] =min (maxdebw, + bw *9 /10 );/* 90% */ + +DRM_DEBUG_KMS ("BW%d / QGV %d: num_planes=%d deratedbw=%d\n" , + i, j, bi->num_planes, bi->deratedbw[j]); + } + +if (bi->num_planes ==1 ) +break ; + } + +return 0 ; +} + +static unsigned int icl_max_bw (struct drm_i915_private *dev_priv, +int num_planes,int qgv_point) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +const struct intel_bw_info *bi = + &dev_priv->max_bw[i]; + +if (num_planes >= bi->num_planes) +return bi->deratedbw[qgv_point]; + } + +return 0 ; +} + +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes) +{ +if (IS_ICELAKE (dev_priv)) +/* +* FIXME with SAGV disabled maybe we can assume +* point 1 will always be used? Seems to match +* the behaviour observed in the wild. +*/ +return min3 (icl_max_bw (dev_priv, num_planes,0 ), +icl_max_bw (dev_priv, num_planes,1 ), +icl_max_bw (dev_priv, num_planes,2 )); +else +return UINT_MAX ; +} +static void intel_get_dram_info (struct drm_i915_private *dev_priv) {@@ -1655,6 +1882,8 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)*/ intel_get_dram_info (dev_priv); +if (INTEL_GEN (dev_priv) >=11 ) +icl_get_bw_info (dev_priv);return 0 ;drivers/gpu/drm/i915/i915_drv.h +10
-0@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h> @@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select (struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((point) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define HSW_PCODE_DE_WRITE_FREQ_REQ 0x17drivers/gpu/drm/i915/intel_atomic_plane.c +20
-0@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); } +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state) +{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i]; + +return cpp * crtc_state->pixel_rate; +} + int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_state, struct intel_crtc_state *new_crtc_state, const struct intel_plane_state *old_plane_state,@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_ new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_if (new_plane_state->base.visible || old_plane_state->base.visible) new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.h +2
-0@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state);void intel_update_plane (struct intel_plane *plane,const struct intel_crtc_state *crtc_state,const struct intel_plane_state *plane_state);drivers/gpu/drm/i915/intel_bw.c new file +181
-0@@ -0,0 +1,181 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright © 2019 Intel Corporation +*/ + +#include <drm/drm_atomic_state_helper.h> + +#include "intel_bw.h" +#include "intel_drv.h" + +static unsigned int intel_bw_crtc_num_active_planes (const struct intel_crtc_state *crtc_state) +{ +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +return hweight8 (crtc_state->active_planes & ~BIT (PLANE_CURSOR )); +} + +static unsigned int intel_bw_crtc_data_rate (const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); +unsigned int data_rate =0 ; +enum plane_id plane_id; + +for_each_plane_id_on_crtc (crtc, plane_id) { +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +if (plane_id ==PLANE_CURSOR ) +continue ; + + data_rate += crtc_state->data_rate[plane_id]; + } + +return data_rate; +} + +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); + + bw_state->data_rate[crtc->pipe] = +intel_bw_crtc_data_rate (crtc_state); + bw_state->num_active_planes[crtc->pipe] = +intel_bw_crtc_num_active_planes (crtc_state); + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); +} + +static unsigned int intel_bw_num_active_planes (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int num_active_planes =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + num_active_planes += bw_state->num_active_planes[pipe]; + +return num_active_planes; +} + +static unsigned int intel_bw_data_rate (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int data_rate =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + data_rate += bw_state->data_rate[pipe]; + +return data_rate; +} + +int intel_bw_atomic_check (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct intel_crtc_state *new_crtc_state, *old_crtc_state; +struct intel_bw_state *bw_state =NULL ; +unsigned int data_rate, max_data_rate; +unsigned int num_active_planes; +struct intel_crtc *crtc; +int i; + +/* FIXME earlier gens need some checks too */ +if (INTEL_GEN (dev_priv) <11 ) +return 0 ; + +for_each_oldnew_intel_crtc_in_state (state, crtc, old_crtc_state, + new_crtc_state, i) { +unsigned int old_data_rate = +intel_bw_crtc_data_rate (old_crtc_state); +unsigned int new_data_rate = +intel_bw_crtc_data_rate (new_crtc_state); +unsigned int old_active_planes = +intel_bw_crtc_num_active_planes (old_crtc_state); +unsigned int new_active_planes = +intel_bw_crtc_num_active_planes (new_crtc_state); + +/* +* Avoid locking the bw state when +* nothing significant has changed. +*/ +if (old_data_rate == new_data_rate && + old_active_planes == new_active_planes) +continue ; + + bw_state =intel_atomic_get_bw_state (state); +if (IS_ERR (bw_state)) +return PTR_ERR (bw_state); + + bw_state->data_rate[crtc->pipe] = new_data_rate; + bw_state->num_active_planes[crtc->pipe] = new_active_planes; + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); + } + +if (!bw_state) +return 0 ; + + data_rate =intel_bw_data_rate (dev_priv, bw_state); + num_active_planes =intel_bw_num_active_planes (dev_priv, bw_state); + + max_data_rate =intel_max_data_rate (dev_priv, num_active_planes); + + data_rate =DIV_ROUND_UP (data_rate,1000 ); + +if (data_rate > max_data_rate) { +DRM_DEBUG_KMS ("Bandwidth %u MB/s exceeds max available %d MB/s (%d active planes)\n" , + data_rate, max_data_rate, num_active_planes); +return -EINVAL ; + } + +return 0 ; +} + +static struct drm_private_state *intel_bw_duplicate_state (struct drm_private_obj *obj) +{ +struct intel_bw_state *state; + + state =kmemdup (obj->state,sizeof (*state),GFP_KERNEL ); +if (!state) +return NULL ; + +__drm_atomic_helper_private_obj_duplicate_state (obj, &state->base); + +return &state->base; +} + +static void intel_bw_destroy_state (struct drm_private_obj *obj, +struct drm_private_state *state) +{ +kfree (state); +} + +static const struct drm_private_state_funcs intel_bw_funcs = { + .atomic_duplicate_state = intel_bw_duplicate_state, + .atomic_destroy_state = intel_bw_destroy_state, +}; + +int intel_bw_init (struct drm_i915_private *dev_priv) +{ +struct intel_bw_state *state; + + state =kzalloc (sizeof (*state),GFP_KERNEL ); +if (!state) +return -ENOMEM ; + +drm_atomic_private_obj_init (&dev_priv->drm, &dev_priv->bw_obj, + &state->base, &intel_bw_funcs); + +return 0 ; +}drivers/gpu/drm/i915/intel_bw.h new file +46
-0@@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2019 Intel Corporation +*/ + +#ifndef __INTEL_BW_H__ +#define __INTEL_BW_H__ + +#include <drm/drm_atomic.h> + +#include "i915_drv.h" +#include "intel_display.h" + +struct drm_i915_private ; +struct intel_atomic_state ; +struct intel_crtc_state ; + +struct intel_bw_state { +struct drm_private_state base; + +unsigned int data_rate[I915_MAX_PIPES ]; +u8 num_active_planes[I915_MAX_PIPES ]; +}; + +#define to_intel_bw_state (x) container_of((x), struct intel_bw_state, base) + +static inline struct intel_bw_state * +intel_atomic_get_bw_state (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct drm_private_state *bw_state; + + bw_state =drm_atomic_get_private_obj_state (&state->base, + &dev_priv->bw_obj); +if (IS_ERR (bw_state)) +return ERR_CAST (bw_state); + +return to_intel_bw_state (bw_state); +} + +int intel_bw_init (struct drm_i915_private *dev_priv); +int intel_bw_atomic_check (struct intel_atomic_state *state); +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state); + +#endif /* __INTEL_BW_H__ */ drivers/gpu/drm/i915/intel_display.c +8
-0@@ -50,6 +50,7 @@#include "intel_acpi.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" +#include "intel_bw.h" #include "intel_color.h" #include "intel_cdclk.h" #include "intel_crt.h" @@ -2863,6 +2864,7 @@ static void intel_plane_disable_noatomic(struct intel_crtc *crtc,intel_set_plane_visible (crtc_state, plane_state, false);fixup_active_planes (crtc_state); + crtc_state->data_rate[plane->id] =0 ;if (plane->id ==PLANE_PRIMARY )intel_pre_disable_primary_noatomic (&crtc->base);@@ -6590,6 +6592,8 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc,struct intel_encoder *encoder;struct intel_crtc *intel_crtc =to_intel_crtc (crtc);struct drm_i915_private *dev_priv =to_i915 (crtc->dev); +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);enum intel_display_power_domain domain;struct intel_plane *plane;u64 domains;@@ -6652,6 +6656,9 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc, dev_priv->active_crtcs &= ~(1 << intel_crtc->pipe); dev_priv->min_cdclk[intel_crtc->pipe] =0 ; dev_priv->min_voltage_level[intel_crtc->pipe] =0 ; + + bw_state->data_rate[intel_crtc->pipe] =0 ; + bw_state->num_active_planes[intel_crtc->pipe] =0 ; }/* @@ -11176,6 +11183,7 @@ int intel_plane_atomic_calc_changes(const struct intel_crtc_state *old_crtc_statif (!is_crtc_enabled) { plane_state->visible = visible = false;to_intel_crtc_state (crtc_state)->active_planes &= ~BIT (plane->id); +to_intel_crtc_state (crtc_state)->data_rate[plane->id] =0 ; }if (!was_visible && !visible)
drivers/gpu/drm/i915/intel_display.c static int intel_atomic_check(struct drm_device *dev, · line 13311
Quoted 9 lines
drivers/gpu/drm/i915/intel_display.c +5
-1@@ -13296,7 +13304,15 @@ static int intel_atomic_check(struct drm_device *dev,return ret;intel_fbc_choose_crtc (dev_priv, intel_state); -return calc_watermark_data (intel_state); + ret =calc_watermark_data (intel_state); +if (ret) +return ret; + + ret =intel_bw_atomic_check (intel_state);
This should be before calc_watermark_data. If we don't have the bandwidth why calculate the WMs.
Quoted 57 lines
drivers/gpu/drm/i915/intel_display.c +26
-0+if (ret) +return ret; + +return 0 ; } static int intel_atomic_prepare_commit(struct drm_device *dev,@@ -15696,6 +15712,10 @@ int intel_modeset_init(struct drm_device *dev)drm_mode_config_init (dev); + ret =intel_bw_init (dev_priv); +if (ret) +return ret; + dev->mode_config.min_width =0 ; dev->mode_config.min_height =0 ;@@ -16318,8 +16338,11 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)drm_connector_list_iter_end (&conn_iter);for_each_intel_crtc (dev, crtc) { +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);struct intel_crtc_state *crtc_state =to_intel_crtc_state (crtc->base.state); +struct intel_plane *plane;int min_cdclk =0 ;memset (&crtc->base.mode,0 ,sizeof (crtc->base.mode));@@ -16358,6 +16381,21 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) dev_priv->min_voltage_level[crtc->pipe] = crtc_state->min_voltage_level; +for_each_intel_plane_on_crtc (&dev_priv->drm, crtc, plane) { +const struct intel_plane_state *plane_state = +to_intel_plane_state (plane->base.state); + +/* +* FIXME don't have the fb yet, so can't +* use intel_plane_data_rate() :( +*/ +if (plane_state->base.visible) + crtc_state->data_rate[plane->id] = +4 * crtc_state->pixel_rate; + } + +intel_bw_crtc_update (bw_state, crtc_state); +intel_pipe_config_sanity_check (dev_priv, crtc_state); } }drivers/gpu/drm/i915/intel_drv.h +0
-0
drivers/gpu/drm/i915/intel_drv.h struct intel_crtc_state { · line 892
Quoted 8 lines
drivers/gpu/drm/i915/intel_drv.h +2
-0@@ -885,6 +885,8 @@ struct intel_crtc_state {struct intel_crtc_wm_state wm; +u32 data_rate[I915_MAX_PLANES ]; +/* Gamma mode programmed on the pipe */ u32 gamma_mode;
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
Quoted 199 lines · Clinton Taylor
Quoted 23 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
Quoted 1 line
TODO: make it robust and deal with all the open questions
TODO: Add comments detailing structures
Quoted 170 lines
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-)create mode 100644
drivers/gpu/drm/i915/intel_bw.ccreate mode 100644drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +119
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ; + + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point) +{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point), + &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + +DRM_DEBUG_KMS ("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%d tRCD=%d tRC=%d\n" , + i, sp->dclk, sp->t_rp, sp->t_rdpre, sp->t_ras, + sp->t_rcd, sp->t_rc); + } + +return 0 ; +} + +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); +
Quoted 8 lines
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1562
Quoted 6 lines
drivers/gpu/drm/i915/i915_drv.c +6
-0+return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +};
intel_sa_info?
sa == system agent.
Quoted 2 lines
Doesn't seem very descriptive. also very close to
intel_sagv_info
I was meaning to rename intel_sagv_info to intel_qgv_info but apparently forgot. Oh, I should really add a comment explaining what qgv stands for.
Quoted 511 lines
Quoted 511 lines
drivers/gpu/drm/i915/i915_drv.c +104
-0+ +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave / num_channels) << i; + bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1 ; + +for (j =0 ; j < si.num_points; j++) { +const struct intel_qgv_point *sp = &si.points[j]; +int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the +* assumed burst length? +*/ + ct =max_t (int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup -1 ) * si.t_bl + sp->t_rdpre); + bw =icl_calc_bw (sp->dclk, clpchgroup *32 * num_channels, ct); + + bi->deratedbw[j] =min (maxdebw, + bw *9 /10 );/* 90% */ + +DRM_DEBUG_KMS ("BW%d / QGV %d: num_planes=%d deratedbw=%d\n" , + i, j, bi->num_planes, bi->deratedbw[j]); + } + +if (bi->num_planes ==1 ) +break ; + } + +return 0 ; +} + +static unsigned int icl_max_bw (struct drm_i915_private *dev_priv, +int num_planes,int qgv_point) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +const struct intel_bw_info *bi = + &dev_priv->max_bw[i]; + +if (num_planes >= bi->num_planes) +return bi->deratedbw[qgv_point]; + } + +return 0 ; +} + +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes) +{ +if (IS_ICELAKE (dev_priv)) +/* +* FIXME with SAGV disabled maybe we can assume +* point 1 will always be used? Seems to match +* the behaviour observed in the wild. +*/ +return min3 (icl_max_bw (dev_priv, num_planes,0 ), +icl_max_bw (dev_priv, num_planes,1 ), +icl_max_bw (dev_priv, num_planes,2 )); +else +return UINT_MAX ; +} +static void intel_get_dram_info (struct drm_i915_private *dev_priv) {@@ -1655,6 +1882,8 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)*/ intel_get_dram_info (dev_priv); +if (INTEL_GEN (dev_priv) >=11 ) +icl_get_bw_info (dev_priv);return 0 ;drivers/gpu/drm/i915/i915_drv.h +10
-0@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h> @@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select (struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((point) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define HSW_PCODE_DE_WRITE_FREQ_REQ 0x17drivers/gpu/drm/i915/intel_atomic_plane.c +20
-0@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); } +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state) +{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i]; + +return cpp * crtc_state->pixel_rate; +} + int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_state, struct intel_crtc_state *new_crtc_state, const struct intel_plane_state *old_plane_state,@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_ new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_if (new_plane_state->base.visible || old_plane_state->base.visible) new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.h +2
-0@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state);void intel_update_plane (struct intel_plane *plane,const struct intel_crtc_state *crtc_state,const struct intel_plane_state *plane_state);drivers/gpu/drm/i915/intel_bw.c new file +181
-0@@ -0,0 +1,181 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright © 2019 Intel Corporation +*/ + +#include <drm/drm_atomic_state_helper.h> + +#include "intel_bw.h" +#include "intel_drv.h" + +static unsigned int intel_bw_crtc_num_active_planes (const struct intel_crtc_state *crtc_state) +{ +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +return hweight8 (crtc_state->active_planes & ~BIT (PLANE_CURSOR )); +} + +static unsigned int intel_bw_crtc_data_rate (const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); +unsigned int data_rate =0 ; +enum plane_id plane_id; + +for_each_plane_id_on_crtc (crtc, plane_id) { +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +if (plane_id ==PLANE_CURSOR ) +continue ; + + data_rate += crtc_state->data_rate[plane_id]; + } + +return data_rate; +} + +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); + + bw_state->data_rate[crtc->pipe] = +intel_bw_crtc_data_rate (crtc_state); + bw_state->num_active_planes[crtc->pipe] = +intel_bw_crtc_num_active_planes (crtc_state); + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); +} + +static unsigned int intel_bw_num_active_planes (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int num_active_planes =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + num_active_planes += bw_state->num_active_planes[pipe]; + +return num_active_planes; +} + +static unsigned int intel_bw_data_rate (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int data_rate =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + data_rate += bw_state->data_rate[pipe]; + +return data_rate; +} + +int intel_bw_atomic_check (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct intel_crtc_state *new_crtc_state, *old_crtc_state; +struct intel_bw_state *bw_state =NULL ; +unsigned int data_rate, max_data_rate; +unsigned int num_active_planes; +struct intel_crtc *crtc; +int i; + +/* FIXME earlier gens need some checks too */ +if (INTEL_GEN (dev_priv) <11 ) +return 0 ; + +for_each_oldnew_intel_crtc_in_state (state, crtc, old_crtc_state, + new_crtc_state, i) { +unsigned int old_data_rate = +intel_bw_crtc_data_rate (old_crtc_state); +unsigned int new_data_rate = +intel_bw_crtc_data_rate (new_crtc_state); +unsigned int old_active_planes = +intel_bw_crtc_num_active_planes (old_crtc_state); +unsigned int new_active_planes = +intel_bw_crtc_num_active_planes (new_crtc_state); + +/* +* Avoid locking the bw state when +* nothing significant has changed. +*/ +if (old_data_rate == new_data_rate && + old_active_planes == new_active_planes) +continue ; + + bw_state =intel_atomic_get_bw_state (state); +if (IS_ERR (bw_state)) +return PTR_ERR (bw_state); + + bw_state->data_rate[crtc->pipe] = new_data_rate; + bw_state->num_active_planes[crtc->pipe] = new_active_planes; + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); + } + +if (!bw_state) +return 0 ; + + data_rate =intel_bw_data_rate (dev_priv, bw_state); + num_active_planes =intel_bw_num_active_planes (dev_priv, bw_state); + + max_data_rate =intel_max_data_rate (dev_priv, num_active_planes); + + data_rate =DIV_ROUND_UP (data_rate,1000 ); + +if (data_rate > max_data_rate) { +DRM_DEBUG_KMS ("Bandwidth %u MB/s exceeds max available %d MB/s (%d active planes)\n" , + data_rate, max_data_rate, num_active_planes); +return -EINVAL ; + } + +return 0 ; +} + +static struct drm_private_state *intel_bw_duplicate_state (struct drm_private_obj *obj) +{ +struct intel_bw_state *state; + + state =kmemdup (obj->state,sizeof (*state),GFP_KERNEL ); +if (!state) +return NULL ; + +__drm_atomic_helper_private_obj_duplicate_state (obj, &state->base); + +return &state->base; +} + +static void intel_bw_destroy_state (struct drm_private_obj *obj, +struct drm_private_state *state) +{ +kfree (state); +} + +static const struct drm_private_state_funcs intel_bw_funcs = { + .atomic_duplicate_state = intel_bw_duplicate_state, + .atomic_destroy_state = intel_bw_destroy_state, +}; + +int intel_bw_init (struct drm_i915_private *dev_priv) +{ +struct intel_bw_state *state; + + state =kzalloc (sizeof (*state),GFP_KERNEL ); +if (!state) +return -ENOMEM ; + +drm_atomic_private_obj_init (&dev_priv->drm, &dev_priv->bw_obj, + &state->base, &intel_bw_funcs); + +return 0 ; +}drivers/gpu/drm/i915/intel_bw.h new file +46
-0@@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2019 Intel Corporation +*/ + +#ifndef __INTEL_BW_H__ +#define __INTEL_BW_H__ + +#include <drm/drm_atomic.h> + +#include "i915_drv.h" +#include "intel_display.h" + +struct drm_i915_private ; +struct intel_atomic_state ; +struct intel_crtc_state ; + +struct intel_bw_state { +struct drm_private_state base; + +unsigned int data_rate[I915_MAX_PIPES ]; +u8 num_active_planes[I915_MAX_PIPES ]; +}; + +#define to_intel_bw_state (x) container_of((x), struct intel_bw_state, base) + +static inline struct intel_bw_state * +intel_atomic_get_bw_state (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct drm_private_state *bw_state; + + bw_state =drm_atomic_get_private_obj_state (&state->base, + &dev_priv->bw_obj); +if (IS_ERR (bw_state)) +return ERR_CAST (bw_state); + +return to_intel_bw_state (bw_state); +} + +int intel_bw_init (struct drm_i915_private *dev_priv); +int intel_bw_atomic_check (struct intel_atomic_state *state); +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state); + +#endif /* __INTEL_BW_H__ */ drivers/gpu/drm/i915/intel_display.c +8
-0@@ -50,6 +50,7 @@#include "intel_acpi.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" +#include "intel_bw.h" #include "intel_color.h" #include "intel_cdclk.h" #include "intel_crt.h" @@ -2863,6 +2864,7 @@ static void intel_plane_disable_noatomic(struct intel_crtc *crtc,intel_set_plane_visible (crtc_state, plane_state, false);fixup_active_planes (crtc_state); + crtc_state->data_rate[plane->id] =0 ;if (plane->id ==PLANE_PRIMARY )intel_pre_disable_primary_noatomic (&crtc->base);@@ -6590,6 +6592,8 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc,struct intel_encoder *encoder;struct intel_crtc *intel_crtc =to_intel_crtc (crtc);struct drm_i915_private *dev_priv =to_i915 (crtc->dev); +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);enum intel_display_power_domain domain;struct intel_plane *plane;u64 domains;@@ -6652,6 +6656,9 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc, dev_priv->active_crtcs &= ~(1 << intel_crtc->pipe); dev_priv->min_cdclk[intel_crtc->pipe] =0 ; dev_priv->min_voltage_level[intel_crtc->pipe] =0 ; + + bw_state->data_rate[intel_crtc->pipe] =0 ; + bw_state->num_active_planes[intel_crtc->pipe] =0 ; }/* @@ -11176,6 +11183,7 @@ int intel_plane_atomic_calc_changes(const struct intel_crtc_state *old_crtc_statif (!is_crtc_enabled) { plane_state->visible = visible = false;to_intel_crtc_state (crtc_state)->active_planes &= ~BIT (plane->id); +to_intel_crtc_state (crtc_state)->data_rate[plane->id] =0 ; }if (!was_visible && !visible)@@ -13296,7 +13304,15 @@ static int intel_atomic_check(struct drm_device *dev,return ret;
Quoted 9 lines
drivers/gpu/drm/i915/intel_display.c static int intel_atomic_check(struct drm_device *dev, · line 13311
Quoted 7 lines
drivers/gpu/drm/i915/intel_display.c +5
-1intel_fbc_choose_crtc (dev_priv, intel_state); -return calc_watermark_data (intel_state); + ret =calc_watermark_data (intel_state); +if (ret) +return ret; + + ret =intel_bw_atomic_check (intel_state);This should be before
calc_watermark_data. If we don't have the bandwidth why calculate the WMs.
I wanted to keep it last to minimize the time we hold the global lock. Although maybe we might want to consider the bw when calculating the watermarks. There are some bandwidth related workarounds in the wm code for earlier platforms that might be relaxed if we had an actual idea on the bw utilization. I'll have to think about it.
Quoted 65 lines
Quoted 65 lines
drivers/gpu/drm/i915/intel_display.c +26
-0+if (ret) +return ret; + +return 0 ; } static int intel_atomic_prepare_commit(struct drm_device *dev,@@ -15696,6 +15712,10 @@ int intel_modeset_init(struct drm_device *dev)drm_mode_config_init (dev); + ret =intel_bw_init (dev_priv); +if (ret) +return ret; + dev->mode_config.min_width =0 ; dev->mode_config.min_height =0 ;@@ -16318,8 +16338,11 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)drm_connector_list_iter_end (&conn_iter);for_each_intel_crtc (dev, crtc) { +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);struct intel_crtc_state *crtc_state =to_intel_crtc_state (crtc->base.state); +struct intel_plane *plane;int min_cdclk =0 ;memset (&crtc->base.mode,0 ,sizeof (crtc->base.mode));@@ -16358,6 +16381,21 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) dev_priv->min_voltage_level[crtc->pipe] = crtc_state->min_voltage_level; +for_each_intel_plane_on_crtc (&dev_priv->drm, crtc, plane) { +const struct intel_plane_state *plane_state = +to_intel_plane_state (plane->base.state); + +/* +* FIXME don't have the fb yet, so can't +* use intel_plane_data_rate() :( +*/ +if (plane_state->base.visible) + crtc_state->data_rate[plane->id] = +4 * crtc_state->pixel_rate; + } + +intel_bw_crtc_update (bw_state, crtc_state); +intel_pipe_config_sanity_check (dev_priv, crtc_state); } }drivers/gpu/drm/i915/intel_drv.h +2
-0@@ -885,6 +885,8 @@ struct intel_crtc_state {struct intel_crtc_wm_state wm; +u32 data_rate[I915_MAX_PLANES ]; +/* Gamma mode programmed on the pipe */ u32 gamma_mode;
Signature · 6 lines
Ville Syrjälä Intel _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
In reply to Ville SyrjalaQuoted 126 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/intel_bw.c create mode 100644 drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +46
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ;
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private · line 1495
Quoted 10 lines
drivers/gpu/drm/i915/i915_drv.c +9
-0+ + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point)
Are we trying to retrieve the dram timing parameters to calculate the latency? If so can that be seperated as latency calculation instead of using it under bw info below?
Quoted 131 lines
drivers/gpu/drm/i915/i915_drv.c +7
-0+{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | + ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO(point),
drivers/gpu/drm/i915/i915_drv.c +36
-0+ &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + + DRM_DEBUG_KMS("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%dtRCD=%d tRC=%d\n",
drivers/gpu/drm/i915/i915_drv.c +1
-0+ i, sp->dclk, sp->t_rp, sp->t_rdpre, sp-Quoted 1 line
t_ras,drivers/gpu/drm/i915/i915_drv.c +66
-0+ sp->t_rcd, sp->t_rc); + } + +return 0 ; +} + +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +}; + +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave /
num_channels) << i;drivers/gpu/drm/i915/i915_drv.c +1
-0+ bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1;
drivers/gpu/drm/i915/i915_drv.c +3
-0+ +for (j =0 ; j < si.num_points; j++) { + const struct intel_qgv_point *sp =&si.points[j];
drivers/gpu/drm/i915/i915_drv.c +11
-0+int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the +* assumed burst length? +*/ + ct = max_t(int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup - 1) * si.t_bl + sp-
Quoted 1 line
Quoted 1 line
t_rdpre);
For logical flow can we move the above timing related calculations to a seperate function along with fixme to delink bandwidth and latency calculations?
- Radhakrishna(RK) Sripada
Quoted 209 lines
drivers/gpu/drm/i915/i915_drv.c +1
-0+ bw = icl_calc_bw(sp->dclk, clpchgroup * 32 *
num_channels, ct);drivers/gpu/drm/i915/i915_drv.c +5
-0+ + bi->deratedbw[j] =min (maxdebw, + bw *9 /10 );/* 90% */ + + DRM_DEBUG_KMS("BW%d / QGV %d: num_planes=%dderatedbw=%d\n",
drivers/gpu/drm/i915/i915_drv.c +1
-0+ i, j, bi->num_planes, bi-Quoted 1 line
deratedbw[j]);
drivers/gpu/drm/i915/i915_drv.c +43
-0+ } + +if (bi->num_planes ==1 ) +break ; + } + +return 0 ; +} + +static unsigned int icl_max_bw (struct drm_i915_private *dev_priv, +int num_planes,int qgv_point) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +const struct intel_bw_info *bi = + &dev_priv->max_bw[i]; + +if (num_planes >= bi->num_planes) +return bi->deratedbw[qgv_point]; + } + +return 0 ; +} + +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes) +{ +if (IS_ICELAKE (dev_priv)) +/* +* FIXME with SAGV disabled maybe we can assume +* point 1 will always be used? Seems to match +* the behaviour observed in the wild. +*/ +return min3 (icl_max_bw (dev_priv, num_planes,0 ), +icl_max_bw (dev_priv, num_planes,1 ), +icl_max_bw (dev_priv, num_planes,2 )); +else +return UINT_MAX ; +} +static void intel_get_dram_info (struct drm_i915_private *dev_priv) {@@ -1655,6 +1882,8 @@ static int i915_driver_init_hw(structdrm_i915_private *dev_priv) */ intel_get_dram_info (dev_priv); +if (INTEL_GEN (dev_priv) >=11 ) +icl_get_bw_info (dev_priv);return 0 ;drivers/gpu/drm/i915/i915_drv.h +10
-0@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h> @@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select(struct
drm_i915_private*dev_priv);drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((poin t) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define
HSW_PCODE_DE_WRITE_FREQ_REQ0x17drivers/gpu/drm/i915/intel_atomic_plane.c +16
-0@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); } +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state) +{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i]; + +return cpp * crtc_state->pixel_rate; +} +int intel_plane_atomic_check_with_state(const struct
intel_crtc_state*old_crtc_state,
structintel_crtc_state*new_crtc_state,
const structintel_plane_state*old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.c +0
-0@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(conststruct
intel_crtc_state*old_crtc_drivers/gpu/drm/i915/intel_atomic_plane.c +1
-0new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(conststruct
intel_crtc_state*old_crtc_
if (new_plane_state->base.visible|| old_plane_state-Quoted 1 line
base.visible)
drivers/gpu/drm/i915/intel_atomic_plane.c +3
-0new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.h +2
-0@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state); void intel_update_plane(struct intel_plane *plane,drivers/gpu/drm/i915/intel_atomic_plane.h +0
-0const struct intel_crtc_state *crtc_state, const struct intel_plane_state *plane_state);drivers/gpu/drm/i915/intel_bw.c new file +11
-0@@ -0,0 +1,181 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright © 2019 Intel Corporation +*/ + +#include <drm/drm_atomic_state_helper.h> + +#include "intel_bw.h" +#include "intel_drv.h" + +static unsigned int intel_bw_crtc_num_active_planes(const struct
intel_crtc_state*crtc_state)
drivers/gpu/drm/i915/intel_bw.c line 16
Quoted 5 lines
drivers/gpu/drm/i915/intel_bw.c new file +5
-0+{ +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/
Can this assumption be made when the cursor sizes are 256x256(chrome)?
Quoted 364 lines
drivers/gpu/drm/i915/intel_bw.c new file +1
-0+ return hweight8(crtc_state->active_planes &~BIT(PLANE_CURSOR));
drivers/gpu/drm/i915/intel_bw.c new file +3
-0+} + +static unsigned int intel_bw_crtc_data_rate(const struct
intel_crtc_state*crtc_state)drivers/gpu/drm/i915/intel_bw.c new file +36
-0+{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); +unsigned int data_rate =0 ; +enum plane_id plane_id; + +for_each_plane_id_on_crtc (crtc, plane_id) { +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +if (plane_id ==PLANE_CURSOR ) +continue ; + + data_rate += crtc_state->data_rate[plane_id]; + } + +return data_rate; +} + +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); + + bw_state->data_rate[crtc->pipe] = +intel_bw_crtc_data_rate (crtc_state); + bw_state->num_active_planes[crtc->pipe] = +intel_bw_crtc_num_active_planes (crtc_state); + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); +} + +static unsigned int intel_bw_num_active_planes(struct
drm_i915_private*dev_priv,drivers/gpu/drm/i915/intel_bw.c new file +1
-0+ const struct
intel_bw_state*bw_state)drivers/gpu/drm/i915/intel_bw.c new file +11
-0+{ +unsigned int num_active_planes =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + num_active_planes += bw_state->num_active_planes[pipe]; + +return num_active_planes; +} + +static unsigned int intel_bw_data_rate(struct drm_i915_private
*dev_priv,drivers/gpu/drm/i915/intel_bw.c new file +1
-0+ const struct intel_bw_state
*bw_state)drivers/gpu/drm/i915/intel_bw.c new file +25
-0+{ +unsigned int data_rate =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + data_rate += bw_state->data_rate[pipe]; + +return data_rate; +} + +int intel_bw_atomic_check (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct intel_crtc_state *new_crtc_state, *old_crtc_state; +struct intel_bw_state *bw_state =NULL ; +unsigned int data_rate, max_data_rate; +unsigned int num_active_planes; +struct intel_crtc *crtc; +int i; + +/* FIXME earlier gens need some checks too */ +if (INTEL_GEN (dev_priv) <11 ) +return 0 ; + + for_each_oldnew_intel_crtc_in_state(state, crtc,
old_crtc_state,drivers/gpu/drm/i915/intel_bw.c new file +7
-0+ new_crtc_state, i) { +unsigned int old_data_rate = +intel_bw_crtc_data_rate (old_crtc_state); +unsigned int new_data_rate = +intel_bw_crtc_data_rate (new_crtc_state); +unsigned int old_active_planes = +intel_bw_crtc_num_active_planes (old_crtc_state);
drivers/gpu/drm/i915/intel_bw.c new file +2
-0+unsigned int new_active_planes = +intel_bw_crtc_num_active_planes (new_crtc_state);
drivers/gpu/drm/i915/intel_bw.c new file +15
-0+ +/* +* Avoid locking the bw state when +* nothing significant has changed. +*/ +if (old_data_rate == new_data_rate && + old_active_planes == new_active_planes) +continue ; + + bw_state =intel_atomic_get_bw_state (state); +if (IS_ERR (bw_state)) +return PTR_ERR (bw_state); + + bw_state->data_rate[crtc->pipe] = new_data_rate; + bw_state->num_active_planes[crtc->pipe] =
new_active_planes;drivers/gpu/drm/i915/intel_bw.c new file +2
-0+ + DRM_DEBUG_KMS("pipe %c data rate %u num active planes%u\n",
drivers/gpu/drm/i915/intel_bw.c new file +10
-0+pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); + } + +if (!bw_state) +return 0 ; + + data_rate =intel_bw_data_rate (dev_priv, bw_state); + num_active_planes = intel_bw_num_active_planes(dev_priv,
bw_state);drivers/gpu/drm/i915/intel_bw.c new file +2
-0+ + max_data_rate = intel_max_data_rate(dev_priv,
num_active_planes);drivers/gpu/drm/i915/intel_bw.c new file +5
-0+ + data_rate =DIV_ROUND_UP (data_rate,1000 ); + +if (data_rate > max_data_rate) { + DRM_DEBUG_KMS("Bandwidth %u MB/s exceeds max available%d MB/s (%d active planes)\n",
drivers/gpu/drm/i915/intel_bw.c new file +1
-0+ data_rate, max_data_rate,
num_active_planes);drivers/gpu/drm/i915/intel_bw.c new file +7
-0+return -EINVAL ; + } + +return 0 ; +} + +static struct drm_private_state *intel_bw_duplicate_state(struct
drm_private_obj*obj)drivers/gpu/drm/i915/intel_bw.c new file +8
-0+{ +struct intel_bw_state *state; + + state =kmemdup (obj->state,sizeof (*state),GFP_KERNEL ); +if (!state) +return NULL ; + + __drm_atomic_helper_private_obj_duplicate_state(obj, &state-Quoted 1 line
base);
drivers/gpu/drm/i915/intel_bw.c new file +28
-0+ +return &state->base; +} + +static void intel_bw_destroy_state (struct drm_private_obj *obj, +struct drm_private_state *state) +{ +kfree (state); +} + +static const struct drm_private_state_funcs intel_bw_funcs = { + .atomic_duplicate_state = intel_bw_duplicate_state, + .atomic_destroy_state = intel_bw_destroy_state, +}; + +int intel_bw_init (struct drm_i915_private *dev_priv) +{ +struct intel_bw_state *state; + + state =kzalloc (sizeof (*state),GFP_KERNEL ); +if (!state) +return -ENOMEM ; + +drm_atomic_private_obj_init (&dev_priv->drm, &dev_priv->bw_obj, + &state->base, &intel_bw_funcs); + +return 0 ; +}drivers/gpu/drm/i915/intel_bw.h new file +25
-0@@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2019 Intel Corporation +*/ + +#ifndef __INTEL_BW_H__ +#define __INTEL_BW_H__ + +#include <drm/drm_atomic.h> + +#include "i915_drv.h" +#include "intel_display.h" + +struct drm_i915_private ; +struct intel_atomic_state ; +struct intel_crtc_state ; + +struct intel_bw_state { +struct drm_private_state base; + +unsigned int data_rate[I915_MAX_PIPES ]; +u8 num_active_planes[I915_MAX_PIPES ]; +}; + +#define to_intel_bw_state (x) container_of((x), struct
intel_bw_state, base)drivers/gpu/drm/i915/intel_bw.h new file +21
-0+ +static inline struct intel_bw_state * +intel_atomic_get_bw_state (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct drm_private_state *bw_state; + + bw_state =drm_atomic_get_private_obj_state (&state->base, + &dev_priv->bw_obj); +if (IS_ERR (bw_state)) +return ERR_CAST (bw_state); + +return to_intel_bw_state (bw_state); +} + +int intel_bw_init (struct drm_i915_private *dev_priv); +int intel_bw_atomic_check (struct intel_atomic_state *state); +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state); + +#endif /* __INTEL_BW_H__ */ drivers/gpu/drm/i915/intel_display.c +2
-0@@ -50,6 +50,7 @@#include "intel_acpi.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" +#include "intel_bw.h" #include "intel_color.h" #include "intel_cdclk.h" #include "intel_crt.h" @@ -2863,6 +2864,7 @@ static void intel_plane_disable_noatomic(structintel_crtc *crtc,intel_set_plane_visible (crtc_state ,plane_state ,false );fixup_active_planes (crtc_state); + crtc_state->data_rate[plane->id] =0 ;if (plane->id ==PLANE_PRIMARY )drivers/gpu/drm/i915/intel_display.c +2
-0intel_pre_disable_primary_noatomic (&crtc->base);@@ -6590,6 +6592,8 @@ static void intel_crtc_disable_noatomic(structdrm_crtc *crtc, struct intel_encoder *encoder;struct intel_crtc *intel_crtc =to_intel_crtc (crtc);struct drm_i915_private *dev_priv =to_i915 (crtc->dev); +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);enum intel_display_power_domain domain;struct intel_plane *plane;drivers/gpu/drm/i915/intel_display.c +3
-0u64 domains;@@ -6652,6 +6656,9 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc, dev_priv->active_crtcs &= ~(1 << intel_crtc->pipe); dev_priv->min_cdclk[intel_crtc->pipe] =0 ; dev_priv->min_voltage_level[intel_crtc->pipe] =0 ; + + bw_state->data_rate[intel_crtc->pipe] =0 ; + bw_state->num_active_planes[intel_crtc->pipe] =0 ; }/*
drivers/gpu/drm/i915/intel_display.c +0
-0@@ -11176,6 +11183,7 @@ int intel_plane_atomic_calc_changes(conststruct
intel_crtc_state*old_crtc_statdrivers/gpu/drm/i915/intel_display.c +0
-0if (!is_crtc_enabled) { plane_state->visible = visible = false; to_intel_crtc_state(crtc_state)->active_planes &=~BIT(plane->id);
drivers/gpu/drm/i915/intel_display.c +1
-0+ to_intel_crtc_state(crtc_state)->data_rate[plane->id] =0;
drivers/gpu/drm/i915/intel_display.c +9
-1}if (!was_visible && !visible)@@ -13296,7 +13304,15 @@ static int intel_atomic_check(structdrm_device *dev, return ret;intel_fbc_choose_crtc (dev_priv, intel_state); -return calc_watermark_data (intel_state); + ret =calc_watermark_data (intel_state); +if (ret) +return ret; + + ret =intel_bw_atomic_check (intel_state); +if (ret) +return ret; + +return 0 ; }static int intel_atomic_prepare_commit(struct
drm_device*dev,drivers/gpu/drm/i915/intel_display.c +7
-0@@ -15696,6 +15712,10 @@ int intel_modeset_init(struct drm_device *dev)drm_mode_config_init (dev); + ret =intel_bw_init (dev_priv); +if (ret) +return ret; + dev->mode_config.min_width =0 ; dev->mode_config.min_height =0 ;@@ -16318,8 +16338,11 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) drm_connector_list_iter_end(&conn_iter);for_each_intel_crtc (dev, crtc) { +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);struct intel_crtc_state *crtc_state =to_intel_crtc_state (crtc->base.state); +struct intel_plane *plane;int min_cdclk =0 ;drivers/gpu/drm/i915/intel_display.c +15
-0memset (&crtc->base.mode,0 ,sizeof (crtc->base.mode));@@ -16358,6 +16381,21 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) dev_priv->min_voltage_level[crtc->pipe] = crtc_state->min_voltage_level; +for_each_intel_plane_on_crtc (&dev_priv->drm, crtc, plane) { +const struct intel_plane_state *plane_state = +to_intel_plane_state (plane- >base.state); + +/* +* FIXME don't have the fb yet, so can't +* use intel_plane_data_rate() :( +*/ +if (plane_state->base.visible) + crtc_state->data_rate[plane->id] = +4 * crtc_state->pixel_rate; + } + +intel_bw_crtc_update (bw_state, crtc_state); +drivers/gpu/drm/i915/intel_display.c +0
-0intel_pipe_config_sanity_check (dev_priv, crtc_state); } }drivers/gpu/drm/i915/intel_drv.h +0
-0
drivers/gpu/drm/i915/intel_drv.h struct intel_crtc_state { · line 892
Quoted 8 lines
drivers/gpu/drm/i915/intel_drv.h +2
-0@@ -885,6 +885,8 @@ struct intel_crtc_state {struct intel_crtc_wm_state wm; +u32 data_rate[I915_MAX_PLANES ]; +/* Gamma mode programmed on the pipe */ u32 gamma_mode;
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
Quoted 271 lines · Radhakrishna
Quoted 126 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/intel_bw.c create mode 100644 drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +46
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ;drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private · line 1495
Quoted 10 lines
drivers/gpu/drm/i915/i915_drv.c +9
-0+ + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point)Are we trying to retrieve the dram timing parameters to calculate the latency? If so can that be seperated as latency calculation instead of using it under bw info below?
Quoted 131 lines
drivers/gpu/drm/i915/i915_drv.c +7
-0+{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | + ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO(point),
drivers/gpu/drm/i915/i915_drv.c +36
-0+ &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + + DRM_DEBUG_KMS("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%dtRCD=%d tRC=%d\n",
drivers/gpu/drm/i915/i915_drv.c +1
-0+ i, sp->dclk, sp->t_rp, sp->t_rdpre, sp-Quoted 1 line
t_ras,drivers/gpu/drm/i915/i915_drv.c +66
-0+ sp->t_rcd, sp->t_rc); + } + +return 0 ; +} + +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +}; + +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave /
num_channels) << i;drivers/gpu/drm/i915/i915_drv.c +1
-0+ bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1;
drivers/gpu/drm/i915/i915_drv.c +3
-0+ +for (j =0 ; j < si.num_points; j++) { + const struct intel_qgv_point *sp =&si.points[j];
drivers/gpu/drm/i915/i915_drv.c +11
-0+int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the +* assumed burst length? +*/ + ct = max_t(int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup - 1) * si.t_bl + sp-
Quoted 4 lines
Quoted 1 line
Quoted 1 line
t_rdpre);For logical flow can we move the above timing related calculations to a seperate function along with fixme to delink bandwidth and latency calculations?
I don't see what you would want to delink. This is all just calculating the effective memory bandwidth limit.
Signature · 6 lines
Ville Syrjälä Intel _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
Quoted 277 lines · Ville Syrjälä
Quoted 275 lines · Radhakrishna
Quoted 136 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/intel_bw.c create mode 100644 drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +14
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info(struct
drm_i915_private*dev_priv,drivers/gpu/drm/i915/i915_drv.c +39
-0+ struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ; + + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info(struct drm_i915_private
*dev_priv,drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1492
Quoted 2 lines
drivers/gpu/drm/i915/i915_drv.c +2
-0+struct intel_qgv_point *sp, + int point)Are we trying to retrieve the dram timing parameters to calculate the
latency? If so can that be seperated as latency calculation instead of
using it under bw info below?Quoted 131 lines
drivers/gpu/drm/i915/i915_drv.c +7
-0+{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | + ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO(point),
drivers/gpu/drm/i915/i915_drv.c +36
-0+ &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + + DRM_DEBUG_KMS("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%dtRCD=%d tRC=%d\n",
drivers/gpu/drm/i915/i915_drv.c +1
-0+ i, sp->dclk, sp->t_rp, sp->t_rdpre, sp-Quoted 1 line
t_ras,drivers/gpu/drm/i915/i915_drv.c +66
-0+ sp->t_rcd, sp->t_rc); + } + +return 0 ; +} + +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +}; + +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave /
num_channels) << i;drivers/gpu/drm/i915/i915_drv.c +1
-0+ bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1;
drivers/gpu/drm/i915/i915_drv.c +3
-0+ +for (j =0 ; j < si.num_points; j++) { + const struct intel_qgv_point *sp =&si.points[j];
drivers/gpu/drm/i915/i915_drv.c +11
-0+int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the +* assumed burst length? +*/ + ct = max_t(int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup - 1) * si.t_bl + sp-
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
Quoted 1 line
t_rdpre);For logical flow can we move the above timing related calculations to a
seperate function along with fixme to delink bandwidth and latency calculations?I don't see what you would want to delink. This is all just calculating the effective memory bandwidth limit.
I am assuming the variables t_rc, t_rp, t_rcd are related to dram latency/timing information. I was wondering if this latency calculation is moved to a seperate inline function for readability purposes.
- Radhakrishna(RK) Sripada
Quoted 1 line
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
In reply to Ville SyrjalaQuoted 169 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/intel_bw.c create mode 100644 drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +94
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ; + + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point) +{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point), + &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i);
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1541
Quoted 10 lines
drivers/gpu/drm/i915/i915_drv.c +10
-0+if (ret) +return ret; + +DRM_DEBUG_KMS ("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%d tRCD=%d tRC=%d\n" , + i, sp->dclk, sp->t_rp, sp->t_rdpre, sp->t_ras, + sp->t_rcd, sp->t_rc); + } + +return 0 ; +}
It might make sense to separate the pcode readout stuff into a patch of its own. Aside from the si->t_bl assignment, the functions above are straightforward details from the pcode HAS. If we wind up needing to drop the complicated algorithm below and replace it with a different one, the pcode readout part won't need to change.
Quoted 66 lines
drivers/gpu/drm/i915/i915_drv.c +66
-0+ +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +}; + +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave / num_channels) << i; + bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1 ; + +for (j =0 ; j < si.num_points; j++) { +const struct intel_qgv_point *sp = &si.points[j]; +int ct, bw;
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1617
Quoted 10 lines
drivers/gpu/drm/i915/i915_drv.c +10
-0+ +/* +* Max row cycle time +* +* FIXME what is the logic behind the +* assumed burst length? +*/ + ct =max_t (int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup -1 ) * si.t_bl + sp->t_rdpre); + bw =icl_calc_bw (sp->dclk, clpchgroup *32 * num_channels, ct);
The HAS document uses *64 instead of *32 for the BW calculation here. Are we doubling the value somewhere else that I'm overlooking?
Quoted 36 lines
drivers/gpu/drm/i915/i915_drv.c +36
-0+ + bi->deratedbw[j] =min (maxdebw, + bw *9 /10 );/* 90% */ + +DRM_DEBUG_KMS ("BW%d / QGV %d: num_planes=%d deratedbw=%d\n" , + i, j, bi->num_planes, bi->deratedbw[j]); + } + +if (bi->num_planes ==1 ) +break ; + } + +return 0 ; +} + +static unsigned int icl_max_bw (struct drm_i915_private *dev_priv, +int num_planes,int qgv_point) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +const struct intel_bw_info *bi = + &dev_priv->max_bw[i]; + +if (num_planes >= bi->num_planes) +return bi->deratedbw[qgv_point]; + } + +return 0 ; +} + +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes) +{ +if (IS_ICELAKE (dev_priv)) +/*
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1663
Quoted 10 lines
drivers/gpu/drm/i915/i915_drv.c +10
-0+* FIXME with SAGV disabled maybe we can assume +* point 1 will always be used? Seems to match +* the behaviour observed in the wild. +*/ +return min3 (icl_max_bw (dev_priv, num_planes,0 ), +icl_max_bw (dev_priv, num_planes,1 ), +icl_max_bw (dev_priv, num_planes,2 )); + else + return UINT_MAX; +}
Any specific reason reason some of these functions are in i915_drv.c? Seems like they could just go in the new intel_bw.c.
Quoted 72 lines
drivers/gpu/drm/i915/i915_drv.c +3
-0+static void intel_get_dram_info (struct drm_i915_private *dev_priv) {@@ -1655,6 +1882,8 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)*/ intel_get_dram_info (dev_priv); +if (INTEL_GEN (dev_priv) >=11 ) +icl_get_bw_info (dev_priv);return 0 ;drivers/gpu/drm/i915/i915_drv.h +10
-0@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h> @@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select (struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((point) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define HSW_PCODE_DE_WRITE_FREQ_REQ 0x17drivers/gpu/drm/i915/intel_atomic_plane.c +2
-0@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); } +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state)
drivers/gpu/drm/i915/intel_atomic_plane.c intel_plane_destroy_state(struct drm_plane *plane, · line 128
Quoted 10 lines
drivers/gpu/drm/i915/intel_atomic_plane.c +10
-0+{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i];
Will this handle NV12 properly? This will give us 1+2, but the algorithm document indicates "NV12 should be considered as 4 bytes per pixel."
Quoted 51 lines
drivers/gpu/drm/i915/intel_atomic_plane.c +8
-0+ +return cpp * crtc_state->pixel_rate; +} + int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_state, struct intel_crtc_state *new_crtc_state, const struct intel_plane_state *old_plane_state,@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_ new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_if (new_plane_state->base.visible || old_plane_state->base.visible) new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.h +2
-0@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state);void intel_update_plane (struct intel_plane *plane,const struct intel_crtc_state *crtc_state,const struct intel_plane_state *plane_state);drivers/gpu/drm/i915/intel_bw.c new file +7
-0@@ -0,0 +1,181 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright © 2019 Intel Corporation +*/ + +#include <drm/drm_atomic_state_helper.h> +
drivers/gpu/drm/i915/intel_bw.c line 17
Quoted 10 lines
drivers/gpu/drm/i915/intel_bw.c new file +10
-0+#include "intel_bw.h" +#include "intel_drv.h" + +static unsigned int intel_bw_crtc_num_active_planes (const struct intel_crtc_state *crtc_state) +{ +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +return hweight8 (crtc_state->active_planes & ~BIT (PLANE_CURSOR ));
Do we also need to account for NV12 slave planes? IIRC, they're not set in the active_planes bitmask, but they're still reading from DRAM, so I imagine they'd count toward the calculations here?
Quoted 192 lines
drivers/gpu/drm/i915/intel_bw.c new file +164
-0+} + +static unsigned int intel_bw_crtc_data_rate (const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); +unsigned int data_rate =0 ; +enum plane_id plane_id; + +for_each_plane_id_on_crtc (crtc, plane_id) { +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +if (plane_id ==PLANE_CURSOR ) +continue ; + + data_rate += crtc_state->data_rate[plane_id]; + } + +return data_rate; +} + +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); + + bw_state->data_rate[crtc->pipe] = +intel_bw_crtc_data_rate (crtc_state); + bw_state->num_active_planes[crtc->pipe] = +intel_bw_crtc_num_active_planes (crtc_state); + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); +} + +static unsigned int intel_bw_num_active_planes (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int num_active_planes =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + num_active_planes += bw_state->num_active_planes[pipe]; + +return num_active_planes; +} + +static unsigned int intel_bw_data_rate (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int data_rate =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + data_rate += bw_state->data_rate[pipe]; + +return data_rate; +} + +int intel_bw_atomic_check (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct intel_crtc_state *new_crtc_state, *old_crtc_state; +struct intel_bw_state *bw_state =NULL ; +unsigned int data_rate, max_data_rate; +unsigned int num_active_planes; +struct intel_crtc *crtc; +int i; + +/* FIXME earlier gens need some checks too */ +if (INTEL_GEN (dev_priv) <11 ) +return 0 ; + +for_each_oldnew_intel_crtc_in_state (state, crtc, old_crtc_state, + new_crtc_state, i) { +unsigned int old_data_rate = +intel_bw_crtc_data_rate (old_crtc_state); +unsigned int new_data_rate = +intel_bw_crtc_data_rate (new_crtc_state); +unsigned int old_active_planes = +intel_bw_crtc_num_active_planes (old_crtc_state); +unsigned int new_active_planes = +intel_bw_crtc_num_active_planes (new_crtc_state); + +/* +* Avoid locking the bw state when +* nothing significant has changed. +*/ +if (old_data_rate == new_data_rate && + old_active_planes == new_active_planes) +continue ; + + bw_state =intel_atomic_get_bw_state (state); +if (IS_ERR (bw_state)) +return PTR_ERR (bw_state); + + bw_state->data_rate[crtc->pipe] = new_data_rate; + bw_state->num_active_planes[crtc->pipe] = new_active_planes; + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); + } + +if (!bw_state) +return 0 ; + + data_rate =intel_bw_data_rate (dev_priv, bw_state); + num_active_planes =intel_bw_num_active_planes (dev_priv, bw_state); + + max_data_rate =intel_max_data_rate (dev_priv, num_active_planes); + + data_rate =DIV_ROUND_UP (data_rate,1000 ); + +if (data_rate > max_data_rate) { +DRM_DEBUG_KMS ("Bandwidth %u MB/s exceeds max available %d MB/s (%d active planes)\n" , + data_rate, max_data_rate, num_active_planes); +return -EINVAL ; + } + +return 0 ; +} + +static struct drm_private_state *intel_bw_duplicate_state (struct drm_private_obj *obj) +{ +struct intel_bw_state *state; + + state =kmemdup (obj->state,sizeof (*state),GFP_KERNEL ); +if (!state) +return NULL ; + +__drm_atomic_helper_private_obj_duplicate_state (obj, &state->base); + +return &state->base; +} + +static void intel_bw_destroy_state (struct drm_private_obj *obj, +struct drm_private_state *state) +{ +kfree (state); +} + +static const struct drm_private_state_funcs intel_bw_funcs = { + .atomic_duplicate_state = intel_bw_duplicate_state, + .atomic_destroy_state = intel_bw_destroy_state, +}; + +int intel_bw_init (struct drm_i915_private *dev_priv) +{ +struct intel_bw_state *state; + + state =kzalloc (sizeof (*state),GFP_KERNEL ); +if (!state) +return -ENOMEM ; + +drm_atomic_private_obj_init (&dev_priv->drm, &dev_priv->bw_obj, + &state->base, &intel_bw_funcs); + +return 0 ; +}drivers/gpu/drm/i915/intel_bw.h new file +22
-0@@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2019 Intel Corporation +*/ + +#ifndef __INTEL_BW_H__ +#define __INTEL_BW_H__ + +#include <drm/drm_atomic.h> + +#include "i915_drv.h" +#include "intel_display.h" + +struct drm_i915_private; +struct intel_atomic_state; +struct intel_crtc_state; + +struct intel_bw_state { + struct drm_private_state base; + + unsigned int data_rate[I915_MAX_PIPES]; + u8 num_active_planes[I915_MAX_PIPES];
drivers/gpu/drm/i915/intel_bw.h line 32
Quoted 10 lines
drivers/gpu/drm/i915/intel_bw.h new file +10
-0+}; + +#define to_intel_bw_state (x) container_of((x), struct intel_bw_state, base) + +static inline struct intel_bw_state * +intel_atomic_get_bw_state (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct drm_private_state *bw_state; +
Do we need to grab a lock here? Otherwise I don't see anything preventing parallel commits that update disjoint sets of CRTCs from exceeding the combined memory bandwidth.
Granted, grabbing a central lock seems really painful since then plane updates on one CRTC may wind up blocking plane updates on another CRTC. But I don't know if there's an easy way around that with the algorithm we've been given either. Was the mention of potentially chopping up the bandwidth in the commit message how you plan to address this? I.e., treat it like we do with DDB space on modesets --- since we globally lock everything on any modeset we can divide it up at that point and then intra-CRTC updates will only have to worry about their own allocation thereafter? That could potentially limit plane usage more than is actually necessary in some cases; not sure how to weigh the impact of that vs the downside of stuff like adding/removing/resizing planes blocking block updates to unrelated CRTC's.
Matt
Quoted 143 lines
drivers/gpu/drm/i915/intel_bw.h new file +14
-0+ bw_state =drm_atomic_get_private_obj_state (&state->base, + &dev_priv->bw_obj); +if (IS_ERR (bw_state)) +return ERR_CAST (bw_state); + + return to_intel_bw_state(bw_state); +} + +int intel_bw_init (struct drm_i915_private *dev_priv); +int intel_bw_atomic_check (struct intel_atomic_state *state); +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state); + +#endif /* __INTEL_BW_H__ */ drivers/gpu/drm/i915/intel_display.c +39
-1@@ -50,6 +50,7 @@#include "intel_acpi.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" +#include "intel_bw.h" #include "intel_color.h" #include "intel_cdclk.h" #include "intel_crt.h" @@ -2863,6 +2864,7 @@ static void intel_plane_disable_noatomic(struct intel_crtc *crtc,intel_set_plane_visible (crtc_state, plane_state, false);fixup_active_planes (crtc_state); + crtc_state->data_rate[plane->id] =0 ;if (plane->id ==PLANE_PRIMARY )intel_pre_disable_primary_noatomic (&crtc->base);@@ -6590,6 +6592,8 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc,struct intel_encoder *encoder;struct intel_crtc *intel_crtc =to_intel_crtc (crtc);struct drm_i915_private *dev_priv =to_i915 (crtc->dev); +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);enum intel_display_power_domain domain;struct intel_plane *plane;u64 domains;@@ -6652,6 +6656,9 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc, dev_priv->active_crtcs &= ~(1 << intel_crtc->pipe); dev_priv->min_cdclk[intel_crtc->pipe] =0 ; dev_priv->min_voltage_level[intel_crtc->pipe] =0 ; + + bw_state->data_rate[intel_crtc->pipe] =0 ; + bw_state->num_active_planes[intel_crtc->pipe] =0 ; }/* @@ -11176,6 +11183,7 @@ int intel_plane_atomic_calc_changes(const struct intel_crtc_state *old_crtc_statif (!is_crtc_enabled) { plane_state->visible = visible = false;to_intel_crtc_state (crtc_state)->active_planes &= ~BIT (plane->id); +to_intel_crtc_state (crtc_state)->data_rate[plane->id] =0 ; }if (!was_visible && !visible)@@ -13296,7 +13304,15 @@ static int intel_atomic_check(struct drm_device *dev,return ret;intel_fbc_choose_crtc (dev_priv, intel_state); -return calc_watermark_data (intel_state); + ret =calc_watermark_data (intel_state); +if (ret) +return ret; + + ret =intel_bw_atomic_check (intel_state); +if (ret) +return ret; + +return 0 ; } static int intel_atomic_prepare_commit(struct drm_device *dev,@@ -15696,6 +15712,10 @@ int intel_modeset_init(struct drm_device *dev)drm_mode_config_init (dev); + ret =intel_bw_init (dev_priv); +if (ret) +return ret; + dev->mode_config.min_width =0 ; dev->mode_config.min_height =0 ;@@ -16318,8 +16338,11 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)drm_connector_list_iter_end (&conn_iter);for_each_intel_crtc (dev, crtc) { +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);struct intel_crtc_state *crtc_state =to_intel_crtc_state (crtc->base.state); +struct intel_plane *plane;int min_cdclk =0 ;memset (&crtc->base.mode,0 ,sizeof (crtc->base.mode));@@ -16358,6 +16381,21 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) dev_priv->min_voltage_level[crtc->pipe] = crtc_state->min_voltage_level; +for_each_intel_plane_on_crtc (&dev_priv->drm, crtc, plane) { +const struct intel_plane_state *plane_state = +to_intel_plane_state (plane->base.state); + +/* +* FIXME don't have the fb yet, so can't +* use intel_plane_data_rate() :( +*/ +if (plane_state->base.visible) + crtc_state->data_rate[plane->id] = +4 * crtc_state->pixel_rate; + } + +intel_bw_crtc_update (bw_state, crtc_state); +intel_pipe_config_sanity_check (dev_priv, crtc_state); } }drivers/gpu/drm/i915/intel_drv.h +2
-0@@ -885,6 +885,8 @@ struct intel_crtc_state {struct intel_crtc_wm_state wm; +u32 data_rate[I915_MAX_PLANES ]; +/* Gamma mode programmed on the pipe */ u32 gamma_mode;Signature · 6 lines
2.21.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 9 lines
Matt Roper Graphics Software Engineer IoTG Platform Enabling & Development Intel Corporation (916) 356-2795 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
Quoted 178 lines · Matt Roper
Quoted 177 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/intel_bw.c create mode 100644 drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +102
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ; + + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point) +{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point), + &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + +DRM_DEBUG_KMS ("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%d tRCD=%d tRC=%d\n" , + i, sp->dclk, sp->t_rp, sp->t_rdpre, sp->t_ras, + sp->t_rcd, sp->t_rc); + } +
Quoted 8 lines
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1541
Quoted 2 lines
drivers/gpu/drm/i915/i915_drv.c +2
-0+return 0 ; +}It might make sense to separate the pcode readout stuff into a patch of its own. Aside from the
si->t_blassignment, the functions above are straightforward details from the pcode HAS. If we wind up needing to drop the complicated algorithm below and replace it with a different one, the pcode readout part won't need to change.
Perhaps. OTOH it does mean having a patch that adds a bunch of unused code.
Quoted 72 lines
Quoted 72 lines
drivers/gpu/drm/i915/i915_drv.c +71
-0+ +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +}; + +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave / num_channels) << i; + bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1 ; + +for (j =0 ; j < si.num_points; j++) { +const struct intel_qgv_point *sp = &si.points[j]; +int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the
Quoted 8 lines
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1617
Quoted 5 lines
drivers/gpu/drm/i915/i915_drv.c +5
-0+* assumed burst length? +*/ + ct =max_t (int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup -1 ) * si.t_bl + sp->t_rdpre); + bw =icl_calc_bw (sp->dclk, clpchgroup *32 * num_channels, ct);The HAS document uses *64 instead of *32 for the BW calculation here. Are we doubling the value somewhere else that I'm overlooking?
It was 32 originally, then was updated to 64, and the dropped back to 32 because 64 caused everything to get doubled from what it was supposed to be. The reason was that pcode is giving doubled values to what the 64 version of the algorithm was expecting.
Quoted 43 lines
Quoted 43 lines
drivers/gpu/drm/i915/i915_drv.c +41
-0+ + bi->deratedbw[j] =min (maxdebw, + bw *9 /10 );/* 90% */ + +DRM_DEBUG_KMS ("BW%d / QGV %d: num_planes=%d deratedbw=%d\n" , + i, j, bi->num_planes, bi->deratedbw[j]); + } + +if (bi->num_planes ==1 ) +break ; + } + +return 0 ; +} + +static unsigned int icl_max_bw (struct drm_i915_private *dev_priv, +int num_planes,int qgv_point) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +const struct intel_bw_info *bi = + &dev_priv->max_bw[i]; + +if (num_planes >= bi->num_planes) +return bi->deratedbw[qgv_point]; + } + +return 0 ; +} + +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes) +{ +if (IS_ICELAKE (dev_priv)) +/* +* FIXME with SAGV disabled maybe we can assume +* point 1 will always be used? Seems to match +* the behaviour observed in the wild. +*/ + return min3(icl_max_bw(dev_priv, num_planes, 0),
Quoted 8 lines
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1663
Quoted 5 lines
drivers/gpu/drm/i915/i915_drv.c +5
-0+icl_max_bw (dev_priv, num_planes,1 ), + icl_max_bw(dev_priv, num_planes, 2)); + else + return UINT_MAX; +}Any specific reason reason some of these functions are in
i915_drv.c? Seems like they could just go in the newintel_bw.c.
They're here mostly because the DRAM readout was here. But yeah, should probably move it all to intel_bw.c.
Quoted 71 lines
Quoted 71 lines
drivers/gpu/drm/i915/i915_drv.c +3
-0+static void intel_get_dram_info (struct drm_i915_private *dev_priv) {@@ -1655,6 +1882,8 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)*/ intel_get_dram_info (dev_priv); +if (INTEL_GEN (dev_priv) >=11 ) +icl_get_bw_info (dev_priv);return 0 ;drivers/gpu/drm/i915/i915_drv.h +10
-0@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h> @@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select (struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((point) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define HSW_PCODE_DE_WRITE_FREQ_REQ 0x17drivers/gpu/drm/i915/intel_atomic_plane.c +0
-0@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); }
Quoted 16 lines
drivers/gpu/drm/i915/intel_atomic_plane.c intel_plane_destroy_state(struct drm_plane *plane, · line 128
Quoted 12 lines
drivers/gpu/drm/i915/intel_atomic_plane.c +12
-0+unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state) +{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i];Will this handle NV12 properly? This will give us 1+2, but the algorithm document indicates "NV12 should be considered as 4 bytes per pixel."
Hmm. That is a bit of a strange way to handle NV12. Probably need to ask why that is the recommendation.
Quoted 58 lines
Quoted 58 lines
drivers/gpu/drm/i915/intel_atomic_plane.c +8
-0+ +return cpp * crtc_state->pixel_rate; +} + int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_state, struct intel_crtc_state *new_crtc_state, const struct intel_plane_state *old_plane_state,@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_ new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_if (new_plane_state->base.visible || old_plane_state->base.visible) new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.h +2
-0@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state);void intel_update_plane (struct intel_plane *plane,const struct intel_crtc_state *crtc_state,const struct intel_plane_state *plane_state);drivers/gpu/drm/i915/intel_bw.c new file +13
-0@@ -0,0 +1,181 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright © 2019 Intel Corporation +*/ + +#include <drm/drm_atomic_state_helper.h> + +#include "intel_bw.h" +#include "intel_drv.h" + +static unsigned int intel_bw_crtc_num_active_planes (const struct intel_crtc_state *crtc_state) +{ +/*
Quoted 8 lines
drivers/gpu/drm/i915/intel_bw.c line 17
Quoted 4 lines
drivers/gpu/drm/i915/intel_bw.c new file +4
-0+* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +return hweight8 (crtc_state->active_planes & ~BIT (PLANE_CURSOR ));Do we also need to account for NV12 slave planes? IIRC, they're not set in the
active_planesbitmask, but they're still reading from DRAM, so I imagine they'd count toward the calculations here?
They should be part of active_planes.
Quoted 199 lines
Quoted 199 lines
drivers/gpu/drm/i915/intel_bw.c new file +164
-0+} + +static unsigned int intel_bw_crtc_data_rate (const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); +unsigned int data_rate =0 ; +enum plane_id plane_id; + +for_each_plane_id_on_crtc (crtc, plane_id) { +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +if (plane_id ==PLANE_CURSOR ) +continue ; + + data_rate += crtc_state->data_rate[plane_id]; + } + +return data_rate; +} + +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); + + bw_state->data_rate[crtc->pipe] = +intel_bw_crtc_data_rate (crtc_state); + bw_state->num_active_planes[crtc->pipe] = +intel_bw_crtc_num_active_planes (crtc_state); + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); +} + +static unsigned int intel_bw_num_active_planes (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int num_active_planes =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + num_active_planes += bw_state->num_active_planes[pipe]; + +return num_active_planes; +} + +static unsigned int intel_bw_data_rate (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int data_rate =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + data_rate += bw_state->data_rate[pipe]; + +return data_rate; +} + +int intel_bw_atomic_check (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct intel_crtc_state *new_crtc_state, *old_crtc_state; +struct intel_bw_state *bw_state =NULL ; +unsigned int data_rate, max_data_rate; +unsigned int num_active_planes; +struct intel_crtc *crtc; +int i; + +/* FIXME earlier gens need some checks too */ +if (INTEL_GEN (dev_priv) <11 ) +return 0 ; + +for_each_oldnew_intel_crtc_in_state (state, crtc, old_crtc_state, + new_crtc_state, i) { +unsigned int old_data_rate = +intel_bw_crtc_data_rate (old_crtc_state); +unsigned int new_data_rate = +intel_bw_crtc_data_rate (new_crtc_state); +unsigned int old_active_planes = +intel_bw_crtc_num_active_planes (old_crtc_state); +unsigned int new_active_planes = +intel_bw_crtc_num_active_planes (new_crtc_state); + +/* +* Avoid locking the bw state when +* nothing significant has changed. +*/ +if (old_data_rate == new_data_rate && + old_active_planes == new_active_planes) +continue ; + + bw_state =intel_atomic_get_bw_state (state); +if (IS_ERR (bw_state)) +return PTR_ERR (bw_state); + + bw_state->data_rate[crtc->pipe] = new_data_rate; + bw_state->num_active_planes[crtc->pipe] = new_active_planes; + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); + } + +if (!bw_state) +return 0 ; + + data_rate =intel_bw_data_rate (dev_priv, bw_state); + num_active_planes =intel_bw_num_active_planes (dev_priv, bw_state); + + max_data_rate =intel_max_data_rate (dev_priv, num_active_planes); + + data_rate =DIV_ROUND_UP (data_rate,1000 ); + +if (data_rate > max_data_rate) { +DRM_DEBUG_KMS ("Bandwidth %u MB/s exceeds max available %d MB/s (%d active planes)\n" , + data_rate, max_data_rate, num_active_planes); +return -EINVAL ; + } + +return 0 ; +} + +static struct drm_private_state *intel_bw_duplicate_state (struct drm_private_obj *obj) +{ +struct intel_bw_state *state; + + state =kmemdup (obj->state,sizeof (*state),GFP_KERNEL ); +if (!state) +return NULL ; + +__drm_atomic_helper_private_obj_duplicate_state (obj, &state->base); + +return &state->base; +} + +static void intel_bw_destroy_state (struct drm_private_obj *obj, +struct drm_private_state *state) +{ +kfree (state); +} + +static const struct drm_private_state_funcs intel_bw_funcs = { + .atomic_duplicate_state = intel_bw_duplicate_state, + .atomic_destroy_state = intel_bw_destroy_state, +}; + +int intel_bw_init (struct drm_i915_private *dev_priv) +{ +struct intel_bw_state *state; + + state =kzalloc (sizeof (*state),GFP_KERNEL ); +if (!state) +return -ENOMEM ; + +drm_atomic_private_obj_init (&dev_priv->drm, &dev_priv->bw_obj, + &state->base, &intel_bw_funcs); + +return 0 ; +}drivers/gpu/drm/i915/intel_bw.h new file +28
-0@@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2019 Intel Corporation +*/ + +#ifndef __INTEL_BW_H__ +#define __INTEL_BW_H__ + +#include <drm/drm_atomic.h> + +#include "i915_drv.h" +#include "intel_display.h" + +struct drm_i915_private ; +struct intel_atomic_state ; +struct intel_crtc_state ; + +struct intel_bw_state { +struct drm_private_state base; + +unsigned int data_rate[I915_MAX_PIPES ]; +u8 num_active_planes[I915_MAX_PIPES ]; +}; + +#define to_intel_bw_state (x) container_of((x), struct intel_bw_state, base) + +static inline struct intel_bw_state * +intel_atomic_get_bw_state (struct intel_atomic_state *state)
Quoted 8 lines
drivers/gpu/drm/i915/intel_bw.h line 32
Quoted 4 lines
drivers/gpu/drm/i915/intel_bw.h new file +4
-0+{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct drm_private_state *bw_state; +Do we need to grab a lock here? Otherwise I don't see anything preventing parallel commits that update disjoint sets of CRTCs from exceeding the combined memory bandwidth.
Private objs have a lock embedded in them now. But even with that it is not quite race free since we don't serialize the actual commits. So there is still a chance of temporary excursion above the limit.
Quoted 13 lines
Granted, grabbing a central lock seems really painful since then plane updates on one CRTC may wind up blocking plane updates on another CRTC. But I don't know if there's an easy way around that with the algorithm we've been given either. Was the mention of potentially chopping up the bandwidth in the commit message how you plan to address this? I.e., treat it like we do with DDB space on modesets --- since we globally lock everything on any modeset we can divide it up at that point and then intra-CRTC updates will only have to worry about their own allocation thereafter? That could potentially limit plane usage more than is actually necessary in some cases; not sure how to weigh the impact of that vs the downside of stuff like adding/removing/resizing planes blocking block updates to unrelated CRTC's.
Yeah, that was the idea. This should avoid the aforementioned problem of temporarily exceeding the bw limit due to arbitrary commit order. I even started implementing this but ran into too much code that needs rework, so I put it on hold for now.
And as you note it would likely leave some bandwidth unutilized. I was also pondering about some kind of hybrid approach where each pipe would still have its own allocation, but we could somehow reallocate that dynamically without resorting to a full modeset. So each pipe would be independent until one reaches its current bw limit, at which point we try to steal unused bw from the other pipes for the pipe that ran out. Ie. the global serialization would only happen when someone is actually pushing the limits.
Signature · 6 lines
Ville Syrjälä Intel _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Matt Roper
Reviewed-byRe: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
Quoted 420 lines · Ville Syrjälä
Quoted 178 lines · Matt Roper
Quoted 177 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/intel_bw.c create mode 100644 drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +102
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ; + + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point) +{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point), + &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + +DRM_DEBUG_KMS ("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%d tRCD=%d tRC=%d\n" , + i, sp->dclk, sp->t_rp, sp->t_rdpre, sp->t_ras, + sp->t_rcd, sp->t_rc); + } +Quoted 8 lines
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1541
Quoted 2 lines
drivers/gpu/drm/i915/i915_drv.c +2
-0+return 0 ; +}It might make sense to separate the pcode readout stuff into a patch of its own. Aside from the
si->t_blassignment, the functions above are straightforward details from the pcode HAS. If we wind up needing to drop the complicated algorithm below and replace it with a different one, the pcode readout part won't need to change.Perhaps. OTOH it does mean having a patch that adds a bunch of unused code.
Quoted 72 lines
Quoted 72 lines
drivers/gpu/drm/i915/i915_drv.c +71
-0+ +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +}; + +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave / num_channels) << i; + bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1 ; + +for (j =0 ; j < si.num_points; j++) { +const struct intel_qgv_point *sp = &si.points[j]; +int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the Quoted 8 lines
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1617
Quoted 5 lines
drivers/gpu/drm/i915/i915_drv.c +5
-0+* assumed burst length? +*/ + ct =max_t (int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup -1 ) * si.t_bl + sp->t_rdpre); + bw =icl_calc_bw (sp->dclk, clpchgroup *32 * num_channels, ct);The HAS document uses *64 instead of *32 for the BW calculation here. Are we doubling the value somewhere else that I'm overlooking?
It was 32 originally, then was updated to 64, and the dropped back to 32 because 64 caused everything to get doubled from what it was supposed to be. The reason was that pcode is giving doubled values to what the 64 version of the algorithm was expecting.
Quoted 43 lines
Quoted 43 lines
drivers/gpu/drm/i915/i915_drv.c +41
-0+ + bi->deratedbw[j] =min (maxdebw, + bw *9 /10 );/* 90% */ + +DRM_DEBUG_KMS ("BW%d / QGV %d: num_planes=%d deratedbw=%d\n" , + i, j, bi->num_planes, bi->deratedbw[j]); + } + +if (bi->num_planes ==1 ) +break ; + } + +return 0 ; +} + +static unsigned int icl_max_bw (struct drm_i915_private *dev_priv, +int num_planes,int qgv_point) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +const struct intel_bw_info *bi = + &dev_priv->max_bw[i]; + +if (num_planes >= bi->num_planes) +return bi->deratedbw[qgv_point]; + } + +return 0 ; +} + +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes) +{ +if (IS_ICELAKE (dev_priv)) +/* +* FIXME with SAGV disabled maybe we can assume +* point 1 will always be used? Seems to match +* the behaviour observed in the wild. +*/ + return min3(icl_max_bw(dev_priv, num_planes, 0),Quoted 8 lines
drivers/gpu/drm/i915/i915_drv.c bxt_get_dram_info(struct drm_i915_private *dev_priv) · line 1663
Quoted 5 lines
drivers/gpu/drm/i915/i915_drv.c +5
-0+icl_max_bw (dev_priv, num_planes,1 ), + icl_max_bw(dev_priv, num_planes, 2)); + else + return UINT_MAX; +}Any specific reason reason some of these functions are in
i915_drv.c? Seems like they could just go in the newintel_bw.c.They're here mostly because the DRAM readout was here. But yeah, should probably move it all to
intel_bw.c.Quoted 71 lines
Quoted 71 lines
drivers/gpu/drm/i915/i915_drv.c +3
-0+static void intel_get_dram_info (struct drm_i915_private *dev_priv) {@@ -1655,6 +1882,8 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)*/ intel_get_dram_info (dev_priv); +if (INTEL_GEN (dev_priv) >=11 ) +icl_get_bw_info (dev_priv);return 0 ;drivers/gpu/drm/i915/i915_drv.h +10
-0@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h> @@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select (struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((point) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define HSW_PCODE_DE_WRITE_FREQ_REQ 0x17drivers/gpu/drm/i915/intel_atomic_plane.c +0
-0@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); }Quoted 16 lines
drivers/gpu/drm/i915/intel_atomic_plane.c intel_plane_destroy_state(struct drm_plane *plane, · line 128
Quoted 12 lines
drivers/gpu/drm/i915/intel_atomic_plane.c +12
-0+unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state) +{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i];Will this handle NV12 properly? This will give us 1+2, but the algorithm document indicates "NV12 should be considered as 4 bytes per pixel."
Quoted 2 lines
Hmm. That is a bit of a strange way to handle NV12. Probably need to ask why that is the recommendation.
Any update on this? I agree that the recommendation to consider it 4 bytes seems odd from a software point of view, but I'd like to get confirmation from whoever came up with the algorithm as to whether it's truly important or whether they just specified that because they thought it would be simpler somehow.
If you've already got confirmation from someone on the hardware side that this is okay, then I think this patch is good enough for now:
and we can figure out something more sophisicated to do about the global locking situation in followup patches. Please also push on the hardware guys to update the algorithm documentation with the outcome here and also the 32 vs 64 thing that you clarified above.
Matt
Quoted 312 lines
Quoted 58 lines
Quoted 58 lines
drivers/gpu/drm/i915/intel_atomic_plane.c +8
-0+ +return cpp * crtc_state->pixel_rate; +} + int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_state, struct intel_crtc_state *new_crtc_state, const struct intel_plane_state *old_plane_state,@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_ new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_if (new_plane_state->base.visible || old_plane_state->base.visible) new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.h +2
-0@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state);void intel_update_plane (struct intel_plane *plane,const struct intel_crtc_state *crtc_state,const struct intel_plane_state *plane_state);drivers/gpu/drm/i915/intel_bw.c new file +13
-0@@ -0,0 +1,181 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright © 2019 Intel Corporation +*/ + +#include <drm/drm_atomic_state_helper.h> + +#include "intel_bw.h" +#include "intel_drv.h" + +static unsigned int intel_bw_crtc_num_active_planes (const struct intel_crtc_state *crtc_state) +{ +/* Quoted 8 lines
drivers/gpu/drm/i915/intel_bw.c line 17
Quoted 4 lines
drivers/gpu/drm/i915/intel_bw.c new file +4
-0+* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +return hweight8 (crtc_state->active_planes & ~BIT (PLANE_CURSOR ));Do we also need to account for NV12 slave planes? IIRC, they're not set in the
active_planesbitmask, but they're still reading from DRAM, so I imagine they'd count toward the calculations here?They should be part of
active_planes.Quoted 199 lines
Quoted 199 lines
drivers/gpu/drm/i915/intel_bw.c new file +164
-0+} + +static unsigned int intel_bw_crtc_data_rate (const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); +unsigned int data_rate =0 ; +enum plane_id plane_id; + +for_each_plane_id_on_crtc (crtc, plane_id) { +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +if (plane_id ==PLANE_CURSOR ) +continue ; + + data_rate += crtc_state->data_rate[plane_id]; + } + +return data_rate; +} + +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); + + bw_state->data_rate[crtc->pipe] = +intel_bw_crtc_data_rate (crtc_state); + bw_state->num_active_planes[crtc->pipe] = +intel_bw_crtc_num_active_planes (crtc_state); + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); +} + +static unsigned int intel_bw_num_active_planes (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int num_active_planes =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + num_active_planes += bw_state->num_active_planes[pipe]; + +return num_active_planes; +} + +static unsigned int intel_bw_data_rate (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int data_rate =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + data_rate += bw_state->data_rate[pipe]; + +return data_rate; +} + +int intel_bw_atomic_check (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct intel_crtc_state *new_crtc_state, *old_crtc_state; +struct intel_bw_state *bw_state =NULL ; +unsigned int data_rate, max_data_rate; +unsigned int num_active_planes; +struct intel_crtc *crtc; +int i; + +/* FIXME earlier gens need some checks too */ +if (INTEL_GEN (dev_priv) <11 ) +return 0 ; + +for_each_oldnew_intel_crtc_in_state (state, crtc, old_crtc_state, + new_crtc_state, i) { +unsigned int old_data_rate = +intel_bw_crtc_data_rate (old_crtc_state); +unsigned int new_data_rate = +intel_bw_crtc_data_rate (new_crtc_state); +unsigned int old_active_planes = +intel_bw_crtc_num_active_planes (old_crtc_state); +unsigned int new_active_planes = +intel_bw_crtc_num_active_planes (new_crtc_state); + +/* +* Avoid locking the bw state when +* nothing significant has changed. +*/ +if (old_data_rate == new_data_rate && + old_active_planes == new_active_planes) +continue ; + + bw_state =intel_atomic_get_bw_state (state); +if (IS_ERR (bw_state)) +return PTR_ERR (bw_state); + + bw_state->data_rate[crtc->pipe] = new_data_rate; + bw_state->num_active_planes[crtc->pipe] = new_active_planes; + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); + } + +if (!bw_state) +return 0 ; + + data_rate =intel_bw_data_rate (dev_priv, bw_state); + num_active_planes =intel_bw_num_active_planes (dev_priv, bw_state); + + max_data_rate =intel_max_data_rate (dev_priv, num_active_planes); + + data_rate =DIV_ROUND_UP (data_rate,1000 ); + +if (data_rate > max_data_rate) { +DRM_DEBUG_KMS ("Bandwidth %u MB/s exceeds max available %d MB/s (%d active planes)\n" , + data_rate, max_data_rate, num_active_planes); +return -EINVAL ; + } + +return 0 ; +} + +static struct drm_private_state *intel_bw_duplicate_state (struct drm_private_obj *obj) +{ +struct intel_bw_state *state; + + state =kmemdup (obj->state,sizeof (*state),GFP_KERNEL ); +if (!state) +return NULL ; + +__drm_atomic_helper_private_obj_duplicate_state (obj, &state->base); + +return &state->base; +} + +static void intel_bw_destroy_state (struct drm_private_obj *obj, +struct drm_private_state *state) +{ +kfree (state); +} + +static const struct drm_private_state_funcs intel_bw_funcs = { + .atomic_duplicate_state = intel_bw_duplicate_state, + .atomic_destroy_state = intel_bw_destroy_state, +}; + +int intel_bw_init (struct drm_i915_private *dev_priv) +{ +struct intel_bw_state *state; + + state =kzalloc (sizeof (*state),GFP_KERNEL ); +if (!state) +return -ENOMEM ; + +drm_atomic_private_obj_init (&dev_priv->drm, &dev_priv->bw_obj, + &state->base, &intel_bw_funcs); + +return 0 ; +}drivers/gpu/drm/i915/intel_bw.h new file +28
-0@@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2019 Intel Corporation +*/ + +#ifndef __INTEL_BW_H__ +#define __INTEL_BW_H__ + +#include <drm/drm_atomic.h> + +#include "i915_drv.h" +#include "intel_display.h" + +struct drm_i915_private ; +struct intel_atomic_state ; +struct intel_crtc_state ; + +struct intel_bw_state { +struct drm_private_state base; + +unsigned int data_rate[I915_MAX_PIPES ]; +u8 num_active_planes[I915_MAX_PIPES ]; +}; + +#define to_intel_bw_state (x) container_of((x), struct intel_bw_state, base) + +static inline struct intel_bw_state * +intel_atomic_get_bw_state (struct intel_atomic_state *state)Quoted 8 lines
drivers/gpu/drm/i915/intel_bw.h line 32
Quoted 4 lines
drivers/gpu/drm/i915/intel_bw.h new file +4
-0+{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct drm_private_state *bw_state; +Do we need to grab a lock here? Otherwise I don't see anything preventing parallel commits that update disjoint sets of CRTCs from exceeding the combined memory bandwidth.
Private objs have a lock embedded in them now. But even with that it is not quite race free since we don't serialize the actual commits. So there is still a chance of temporary excursion above the limit.
Quoted 13 lines
Granted, grabbing a central lock seems really painful since then plane updates on one CRTC may wind up blocking plane updates on another CRTC. But I don't know if there's an easy way around that with the algorithm we've been given either. Was the mention of potentially chopping up the bandwidth in the commit message how you plan to address this? I.e., treat it like we do with DDB space on modesets --- since we globally lock everything on any modeset we can divide it up at that point and then intra-CRTC updates will only have to worry about their own allocation thereafter? That could potentially limit plane usage more than is actually necessary in some cases; not sure how to weigh the impact of that vs the downside of stuff like adding/removing/resizing planes blocking block updates to unrelated CRTC's.
Yeah, that was the idea. This should avoid the aforementioned problem of temporarily exceeding the bw limit due to arbitrary commit order. I even started implementing this but ran into too much code that needs rework, so I put it on hold for now.
And as you note it would likely leave some bandwidth unutilized. I was also pondering about some kind of hybrid approach where each pipe would still have its own allocation, but we could somehow reallocate that dynamically without resorting to a full modeset. So each pipe would be independent until one reaches its current bw limit, at which point we try to steal unused bw from the other pipes for the pipe that ran out. Ie. the global serialization would only happen when someone is actually pushing the limits.
Signature · 2 lines
Ville Syrjälä Intel
Signature · 9 lines
Matt Roper Graphics Software Engineer IoTG Platform Enabling & Development Intel Corporation (916) 356-2795 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
In reply to Ville SyrjalaOp 03-05-2019 om 21:08 schreef Ville Syrjala:
Quoted 32 lines
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbers
Quoted 2 lines
Signed-off-by Ville Syrjälä <[email protected]>---
Splitting out the HW readout would be nice, it would make it easier to review the separate parts that this patch tries to accomplish. :)
Quoted 740 lines
drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/intel_bw.c create mode 100644 drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +229
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ; + + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point) +{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point), + &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + +DRM_DEBUG_KMS ("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%d tRCD=%d tRC=%d\n" , + i, sp->dclk, sp->t_rp, sp->t_rdpre, sp->t_ras, + sp->t_rcd, sp->t_rc); + } + +return 0 ; +} + +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +}; + +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave / num_channels) << i; + bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1 ; + +for (j =0 ; j < si.num_points; j++) { +const struct intel_qgv_point *sp = &si.points[j]; +int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the +* assumed burst length? +*/ + ct =max_t (int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup -1 ) * si.t_bl + sp->t_rdpre); + bw =icl_calc_bw (sp->dclk, clpchgroup *32 * num_channels, ct); + + bi->deratedbw[j] =min (maxdebw, + bw *9 /10 );/* 90% */ + +DRM_DEBUG_KMS ("BW%d / QGV %d: num_planes=%d deratedbw=%d\n" , + i, j, bi->num_planes, bi->deratedbw[j]); + } + +if (bi->num_planes ==1 ) +break ; + } + +return 0 ; +} + +static unsigned int icl_max_bw (struct drm_i915_private *dev_priv, +int num_planes,int qgv_point) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +const struct intel_bw_info *bi = + &dev_priv->max_bw[i]; + +if (num_planes >= bi->num_planes) +return bi->deratedbw[qgv_point]; + } + +return 0 ; +} + +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes) +{ +if (IS_ICELAKE (dev_priv)) +/* +* FIXME with SAGV disabled maybe we can assume +* point 1 will always be used? Seems to match +* the behaviour observed in the wild. +*/ +return min3 (icl_max_bw (dev_priv, num_planes,0 ), +icl_max_bw (dev_priv, num_planes,1 ), +icl_max_bw (dev_priv, num_planes,2 )); +else +return UINT_MAX ; +} +static void intel_get_dram_info (struct drm_i915_private *dev_priv) {@@ -1655,6 +1882,8 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)*/ intel_get_dram_info (dev_priv); +if (INTEL_GEN (dev_priv) >=11 ) +icl_get_bw_info (dev_priv);return 0 ;drivers/gpu/drm/i915/i915_drv.h +10
-0@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h> @@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select (struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((point) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define HSW_PCODE_DE_WRITE_FREQ_REQ 0x17drivers/gpu/drm/i915/intel_atomic_plane.c +20
-0@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); } +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state) +{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i]; + +return cpp * crtc_state->pixel_rate; +} + int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_state, struct intel_crtc_state *new_crtc_state, const struct intel_plane_state *old_plane_state,@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_ new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_if (new_plane_state->base.visible || old_plane_state->base.visible) new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.h +2
-0@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state);void intel_update_plane (struct intel_plane *plane,const struct intel_crtc_state *crtc_state,const struct intel_plane_state *plane_state);drivers/gpu/drm/i915/intel_bw.c new file +181
-0@@ -0,0 +1,181 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright © 2019 Intel Corporation +*/ + +#include <drm/drm_atomic_state_helper.h> + +#include "intel_bw.h" +#include "intel_drv.h" + +static unsigned int intel_bw_crtc_num_active_planes (const struct intel_crtc_state *crtc_state) +{ +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +return hweight8 (crtc_state->active_planes & ~BIT (PLANE_CURSOR )); +} + +static unsigned int intel_bw_crtc_data_rate (const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); +unsigned int data_rate =0 ; +enum plane_id plane_id; + +for_each_plane_id_on_crtc (crtc, plane_id) { +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +if (plane_id ==PLANE_CURSOR ) +continue ; + + data_rate += crtc_state->data_rate[plane_id]; + } + +return data_rate; +} + +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); + + bw_state->data_rate[crtc->pipe] = +intel_bw_crtc_data_rate (crtc_state); + bw_state->num_active_planes[crtc->pipe] = +intel_bw_crtc_num_active_planes (crtc_state); + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); +} + +static unsigned int intel_bw_num_active_planes (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int num_active_planes =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + num_active_planes += bw_state->num_active_planes[pipe]; + +return num_active_planes; +} + +static unsigned int intel_bw_data_rate (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int data_rate =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + data_rate += bw_state->data_rate[pipe]; + +return data_rate; +} + +int intel_bw_atomic_check (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct intel_crtc_state *new_crtc_state, *old_crtc_state; +struct intel_bw_state *bw_state =NULL ; +unsigned int data_rate, max_data_rate; +unsigned int num_active_planes; +struct intel_crtc *crtc; +int i; + +/* FIXME earlier gens need some checks too */ +if (INTEL_GEN (dev_priv) <11 ) +return 0 ; + +for_each_oldnew_intel_crtc_in_state (state, crtc, old_crtc_state, + new_crtc_state, i) { +unsigned int old_data_rate = +intel_bw_crtc_data_rate (old_crtc_state); +unsigned int new_data_rate = +intel_bw_crtc_data_rate (new_crtc_state); +unsigned int old_active_planes = +intel_bw_crtc_num_active_planes (old_crtc_state); +unsigned int new_active_planes = +intel_bw_crtc_num_active_planes (new_crtc_state); + +/* +* Avoid locking the bw state when +* nothing significant has changed. +*/ +if (old_data_rate == new_data_rate && + old_active_planes == new_active_planes) +continue ; + + bw_state =intel_atomic_get_bw_state (state); +if (IS_ERR (bw_state)) +return PTR_ERR (bw_state); + + bw_state->data_rate[crtc->pipe] = new_data_rate; + bw_state->num_active_planes[crtc->pipe] = new_active_planes; + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); + } + +if (!bw_state) +return 0 ; + + data_rate =intel_bw_data_rate (dev_priv, bw_state); + num_active_planes =intel_bw_num_active_planes (dev_priv, bw_state); + + max_data_rate =intel_max_data_rate (dev_priv, num_active_planes); + + data_rate =DIV_ROUND_UP (data_rate,1000 ); + +if (data_rate > max_data_rate) { +DRM_DEBUG_KMS ("Bandwidth %u MB/s exceeds max available %d MB/s (%d active planes)\n" , + data_rate, max_data_rate, num_active_planes); +return -EINVAL ; + } + +return 0 ; +} + +static struct drm_private_state *intel_bw_duplicate_state (struct drm_private_obj *obj) +{ +struct intel_bw_state *state; + + state =kmemdup (obj->state,sizeof (*state),GFP_KERNEL ); +if (!state) +return NULL ; + +__drm_atomic_helper_private_obj_duplicate_state (obj, &state->base); + +return &state->base; +} + +static void intel_bw_destroy_state (struct drm_private_obj *obj, +struct drm_private_state *state) +{ +kfree (state); +} + +static const struct drm_private_state_funcs intel_bw_funcs = { + .atomic_duplicate_state = intel_bw_duplicate_state, + .atomic_destroy_state = intel_bw_destroy_state, +}; + +int intel_bw_init (struct drm_i915_private *dev_priv) +{ +struct intel_bw_state *state; + + state =kzalloc (sizeof (*state),GFP_KERNEL ); +if (!state) +return -ENOMEM ; + +drm_atomic_private_obj_init (&dev_priv->drm, &dev_priv->bw_obj, + &state->base, &intel_bw_funcs); + +return 0 ; +}drivers/gpu/drm/i915/intel_bw.h new file +46
-0@@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2019 Intel Corporation +*/ + +#ifndef __INTEL_BW_H__ +#define __INTEL_BW_H__ + +#include <drm/drm_atomic.h> + +#include "i915_drv.h" +#include "intel_display.h" + +struct drm_i915_private ; +struct intel_atomic_state ; +struct intel_crtc_state ; + +struct intel_bw_state { +struct drm_private_state base; + +unsigned int data_rate[I915_MAX_PIPES ]; +u8 num_active_planes[I915_MAX_PIPES ]; +}; + +#define to_intel_bw_state (x) container_of((x), struct intel_bw_state, base) + +static inline struct intel_bw_state * +intel_atomic_get_bw_state (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct drm_private_state *bw_state; + + bw_state =drm_atomic_get_private_obj_state (&state->base, + &dev_priv->bw_obj); +if (IS_ERR (bw_state)) +return ERR_CAST (bw_state); + +return to_intel_bw_state (bw_state); +} + +int intel_bw_init (struct drm_i915_private *dev_priv); +int intel_bw_atomic_check (struct intel_atomic_state *state); +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state); + +#endif /* __INTEL_BW_H__ */ drivers/gpu/drm/i915/intel_display.c +39
-1@@ -50,6 +50,7 @@#include "intel_acpi.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" +#include "intel_bw.h" #include "intel_color.h" #include "intel_cdclk.h" #include "intel_crt.h" @@ -2863,6 +2864,7 @@ static void intel_plane_disable_noatomic(struct intel_crtc *crtc,intel_set_plane_visible (crtc_state, plane_state, false);fixup_active_planes (crtc_state); + crtc_state->data_rate[plane->id] =0 ;if (plane->id ==PLANE_PRIMARY )intel_pre_disable_primary_noatomic (&crtc->base);@@ -6590,6 +6592,8 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc,struct intel_encoder *encoder;struct intel_crtc *intel_crtc =to_intel_crtc (crtc);struct drm_i915_private *dev_priv =to_i915 (crtc->dev); +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);enum intel_display_power_domain domain;struct intel_plane *plane;u64 domains;@@ -6652,6 +6656,9 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc, dev_priv->active_crtcs &= ~(1 << intel_crtc->pipe); dev_priv->min_cdclk[intel_crtc->pipe] =0 ; dev_priv->min_voltage_level[intel_crtc->pipe] =0 ; + + bw_state->data_rate[intel_crtc->pipe] =0 ; + bw_state->num_active_planes[intel_crtc->pipe] =0 ; }/* @@ -11176,6 +11183,7 @@ int intel_plane_atomic_calc_changes(const struct intel_crtc_state *old_crtc_statif (!is_crtc_enabled) { plane_state->visible = visible = false;to_intel_crtc_state (crtc_state)->active_planes &= ~BIT (plane->id); +to_intel_crtc_state (crtc_state)->data_rate[plane->id] =0 ; }if (!was_visible && !visible)@@ -13296,7 +13304,15 @@ static int intel_atomic_check(struct drm_device *dev,return ret;intel_fbc_choose_crtc (dev_priv, intel_state); -return calc_watermark_data (intel_state); + ret =calc_watermark_data (intel_state); +if (ret) +return ret; + + ret =intel_bw_atomic_check (intel_state); +if (ret) +return ret; + +return 0 ; } static int intel_atomic_prepare_commit(struct drm_device *dev,@@ -15696,6 +15712,10 @@ int intel_modeset_init(struct drm_device *dev)drm_mode_config_init (dev); + ret =intel_bw_init (dev_priv); +if (ret) +return ret; + dev->mode_config.min_width =0 ; dev->mode_config.min_height =0 ;@@ -16318,8 +16338,11 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)drm_connector_list_iter_end (&conn_iter);for_each_intel_crtc (dev, crtc) { +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);struct intel_crtc_state *crtc_state =to_intel_crtc_state (crtc->base.state); +struct intel_plane *plane;int min_cdclk =0 ;memset (&crtc->base.mode,0 ,sizeof (crtc->base.mode));@@ -16358,6 +16381,21 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) dev_priv->min_voltage_level[crtc->pipe] = crtc_state->min_voltage_level; +for_each_intel_plane_on_crtc (&dev_priv->drm, crtc, plane) { +const struct intel_plane_state *plane_state = +to_intel_plane_state (plane->base.state); + +/* +* FIXME don't have the fb yet, so can't +* use intel_plane_data_rate() :( +*/ +if (plane_state->base.visible) + crtc_state->data_rate[plane->id] = +4 * crtc_state->pixel_rate; + } + +intel_bw_crtc_update (bw_state, crtc_state); +intel_pipe_config_sanity_check (dev_priv, crtc_state); } }drivers/gpu/drm/i915/intel_drv.h +0
-0
drivers/gpu/drm/i915/intel_drv.h struct intel_crtc_state { · line 892
Quoted 8 lines
drivers/gpu/drm/i915/intel_drv.h +2
-0@@ -885,6 +885,8 @@ struct intel_crtc_state {struct intel_crtc_wm_state wm; +u32 data_rate[I915_MAX_PLANES ]; +/* Gamma mode programmed on the pipe */ u32 gamma_mode;
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 2/2] drm/i915: Make sure we have enough memory bandwidth on ICL
In reply to Ville SyrjalaNit: Commit message V4 and Patch Subject V3
-Clint
Quoted 774 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
ICL has so many planes that it can easily exceed the maximum effective memory bandwidth of the system. We must therefore check that we don't exceed that limit.
The algorithm is very magic number heavy and lacks sufficient explanation for now. We also have no sane way to query the memory clock and timings, so we must rely on a combination of raw readout from the memory controller and hardcoded assumptions. The memory controller values obviously change as the system jumps between the different SAGV points, so we try to stabilize it first by disabling SAGV for the duration of the readout.
The utilized bandwidth is tracked via a device wide atomic private object. That is actually not robust because we can't afford to enforce strict global ordering between the pipes. Thus I think I'll need to change this to simply chop up the available bandwidth between all the active pipes. Each pipe can then do whatever it wants as long as it doesn't exceed its budget. That scheme will also require that we assume that any number of planes could be active at any time.
TODO: make it robust and deal with all the open questions
v2: Sleep longer after disabling SAGV
v3: Poll for the dclk to get raised (seen it take 250ms!)
If the system has 2133MT/s memory then we pointlessly wait one full second :(
v4: Use the new pcode interface to get the qgv points rather
that using hardcoded numbersSigned-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 229 ++++++++++++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 10 + drivers/gpu/drm/i915/i915_reg.h | 3 + drivers/gpu/drm/i915/intel_atomic_plane.c | 20 ++ drivers/gpu/drm/i915/intel_atomic_plane.h | 2 + drivers/gpu/drm/i915/intel_bw.c | 181 +++++++++++++++++ drivers/gpu/drm/i915/intel_bw.h | 46 +++++ drivers/gpu/drm/i915/intel_display.c | 40 +++- drivers/gpu/drm/i915/intel_drv.h | 2 + 10 files changed, 533 insertions(+), 1 deletion(-)create mode 100644
drivers/gpu/drm/i915/intel_bw.ccreate mode 100644drivers/gpu/drm/i915/intel_bw.hdrivers/gpu/drm/i915/Makefile +1
-0@@ -138,6 +138,7 @@ i915-y += intel_audio.o \ intel_atomic.o \ intel_atomic_plane.o \ intel_bios.o \ + intel_bw.o \ intel_cdclk.o \ intel_color.o \ intel_combo_phy.o \drivers/gpu/drm/i915/i915_drv.c +229
-0@@ -70,6 +70,7 @@#include "intel_overlay.h" #include "intel_pipe_crc.h" #include "intel_pm.h" +#include "intel_sideband.h" #include "intel_sprite.h" #include "intel_uc.h" @@ -1435,6 +1436,232 @@ bxt_get_dram_info(struct drm_i915_private *dev_priv)return 0 ; } +struct intel_qgv_point { +u16 dclk, t_rp, t_rdpre, t_rc, t_ras, t_rcd; +}; + +struct intel_sagv_info { +struct intel_qgv_point points[3 ]; +u8 num_points; +u8 num_channels; +u8 t_bl; +enum intel_dram_type dram_type; +}; + +static int icl_pcode_read_mem_global_info (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +u32 val =0 ; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_GLOBAL_INFO , + &val,NULL ); +if (ret) +return ret; + +switch (val &0xf ) { +case 0 : + si->dram_type =INTEL_DRAM_DDR4 ; +break ; +case 1 : + si->dram_type =INTEL_DRAM_DDR3 ; +break ; +case 2 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +case 3 : + si->dram_type =INTEL_DRAM_LPDDR3 ; +break ; +default : +MISSING_CASE (val &0xf ); +break ; + } + + si->num_channels = (val &0xf0 ) >>4 ; + si->num_points = (val &0xf00 ) >>8 ; + + si->t_bl = si->dram_type ==INTEL_DRAM_DDR4 ?4 :8 ; + +return 0 ; +} + +static int icl_pcode_read_qgv_point_info (struct drm_i915_private *dev_priv, +struct intel_qgv_point *sp, +int point) +{ +u32 val =0 , val2; +int ret; + + ret =sandybridge_pcode_read (dev_priv, +ICL_PCODE_MEM_SUBSYSYSTEM_INFO | +ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point), + &val, &val2); +if (ret) +return ret; + + sp->dclk = val &0xffff ; + sp->t_rp = (val &0xff0000 ) >>16 ; + sp->t_rcd = (val &0xff000000 ) >>24 ; + + sp->t_rdpre = val2 &0xff ; + sp->t_ras = (val2 &0xff00 ) >>8 ; + + sp->t_rc = sp->t_rp + sp->t_ras; + +return 0 ; +} + +static int icl_get_qgv_points (struct drm_i915_private *dev_priv, +struct intel_sagv_info *si) +{ +int i, ret; + + ret =icl_pcode_read_mem_global_info (dev_priv, si); +if (ret) +return ret; + +if (WARN_ON (si->num_points >ARRAY_SIZE (si->points))) + si->num_points =ARRAY_SIZE (si->points); + +for (i =0 ; i < si->num_points; i++) { +struct intel_qgv_point *sp = &si->points[i]; + + ret =icl_pcode_read_qgv_point_info (dev_priv, sp, i); +if (ret) +return ret; + +DRM_DEBUG_KMS ("QGV %d: DCLK=%d tRP=%d tRDPRE=%d tRAS=%d tRCD=%d tRC=%d\n" , + i, sp->dclk, sp->t_rp, sp->t_rdpre, sp->t_ras, + sp->t_rcd, sp->t_rc); + } + +return 0 ; +} + +static int icl_calc_bw (int dclk,int num,int den) +{ +/* multiples of 16.666MHz (100/6) */ +return DIV_ROUND_CLOSEST (num * dclk *100 , den *6 ); +} + +static int icl_sagv_max_dclk (const struct intel_sagv_info *si) +{ +u16 dclk =0 ; +int i; + +for (i =0 ; i < si->num_points; i++) + dclk =max (dclk, si->points[i].dclk); + +return dclk; +} + +struct intel_sa_info { +u8 deburst, mpagesize, deprogbwlimit, displayrtids; +}; + +static const struct intel_sa_info icl_sa_info = { + .deburst =8 , + .mpagesize =16 , + .deprogbwlimit =25 ,/* GB/s */ + .displayrtids =128 , +}; + +static int icl_get_bw_info (struct drm_i915_private *dev_priv) +{ +struct intel_sagv_info si = {}; +const struct intel_sa_info *sa = &icl_sa_info; +bool is_y_tile = true;/* assume y tile may be used */ +int num_channels; +int deinterleave; +int ipqdepth, ipqdepthpch; +int dclk_max; +int maxdebw; +int i, ret; + + ret =icl_get_qgv_points (dev_priv, &si); +if (ret) +return ret; + num_channels = si.num_channels; + + deinterleave =DIV_ROUND_UP (num_channels, is_y_tile ?4 :2 ); + dclk_max =icl_sagv_max_dclk (&si); + + ipqdepthpch =16 ; + + maxdebw =min (sa->deprogbwlimit *1000 , +icl_calc_bw (dclk_max,16 ,1 ) *6 /10 );/* 60% */ + ipqdepth =min (ipqdepthpch, sa->displayrtids / num_channels); + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +struct intel_bw_info *bi = &dev_priv->max_bw[i]; +int clpchgroup; +int j; + + clpchgroup = (sa->deburst * deinterleave / num_channels) << i; + bi->num_planes = (ipqdepth - clpchgroup) / clpchgroup +1 ; + +for (j =0 ; j < si.num_points; j++) { +const struct intel_qgv_point *sp = &si.points[j]; +int ct, bw; + +/* +* Max row cycle time +* +* FIXME what is the logic behind the +* assumed burst length? +*/ + ct =max_t (int, sp->t_rc, sp->t_rp + sp->t_rcd + + (clpchgroup -1 ) * si.t_bl + sp->t_rdpre); + bw =icl_calc_bw (sp->dclk, clpchgroup *32 * num_channels, ct); + + bi->deratedbw[j] =min (maxdebw, + bw *9 /10 );/* 90% */ + +DRM_DEBUG_KMS ("BW%d / QGV %d: num_planes=%d deratedbw=%d\n" , + i, j, bi->num_planes, bi->deratedbw[j]); + } + +if (bi->num_planes ==1 ) +break ; + } + +return 0 ; +} + +static unsigned int icl_max_bw (struct drm_i915_private *dev_priv, +int num_planes,int qgv_point) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (dev_priv->max_bw); i++) { +const struct intel_bw_info *bi = + &dev_priv->max_bw[i]; + +if (num_planes >= bi->num_planes) +return bi->deratedbw[qgv_point]; + } + +return 0 ; +} + +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes) +{ +if (IS_ICELAKE (dev_priv)) +/* +* FIXME with SAGV disabled maybe we can assume +* point 1 will always be used? Seems to match +* the behaviour observed in the wild. +*/ +return min3 (icl_max_bw (dev_priv, num_planes,0 ), +icl_max_bw (dev_priv, num_planes,1 ), +icl_max_bw (dev_priv, num_planes,2 )); +else +return UINT_MAX ; +} +static void intel_get_dram_info (struct drm_i915_private *dev_priv) {@@ -1655,6 +1882,8 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)*/ intel_get_dram_info (dev_priv); +if (INTEL_GEN (dev_priv) >=11 ) +icl_get_bw_info (dev_priv);return 0 ;drivers/gpu/drm/i915/i915_drv.h +10
-0@@ -54,6 +54,7 @@#include <drm/drm_cache.h> #include <drm/drm_util.h> #include <drm/drm_dsc.h> +#include <drm/drm_atomic.h> #include <drm/drm_connector.h> #include <drm/i915_mei_hdcp_interface.h> @@ -1837,6 +1838,13 @@ struct drm_i915_private { } type; } dram_info; +struct intel_bw_info { +int num_planes; +int deratedbw[3 ]; + } max_bw[6 ]; + +struct drm_private_obj bw_obj; +struct i915_runtime_pm runtime_pm;struct {@@ -2706,6 +2714,8 @@ extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);extern unsigned long i915_gfx_val (struct drm_i915_private *dev_priv);extern void i915_update_gfx_val (struct drm_i915_private *dev_priv);int vlv_force_gfx_clock (struct drm_i915_private *dev_priv,bool on); +unsigned int intel_max_data_rate (struct drm_i915_private *dev_priv, +int num_planes);u32 intel_calculate_mcr_s_ss_select (struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/i915_reg.h +3
-0@@ -8774,6 +8774,9 @@ enum {#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9#define GEN6_READ_OC_PARAMS 0xc +#define ICL_PCODE_MEM_SUBSYSYSTEM_INFO 0xd +#define ICL_PCODE_MEM_SS_READ_GLOBAL_INFO (0x0 << 8) +#define ICL_PCODE_MEM_SS_READ_QGV_POINT_INFO (point) (((point) << 16) | (0x1 << 8))#define GEN6_PCODE_READ_D_COMP 0x10#define GEN6_PCODE_WRITE_D_COMP 0x11#define HSW_PCODE_DE_WRITE_FREQ_REQ 0x17drivers/gpu/drm/i915/intel_atomic_plane.c +20
-0@@ -114,6 +114,22 @@ intel_plane_destroy_state(struct drm_plane *plane,drm_atomic_helper_plane_destroy_state (plane, state); } +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state) +{ +const struct drm_framebuffer *fb = plane_state->base.fb; +unsigned int cpp =0 ; +int i; + +if (!plane_state->base.visible) +return 0 ; + +for (i =0 ; i < fb->format->num_planes; i++) + cpp += fb->format->cpp[i]; + +return cpp * crtc_state->pixel_rate; +} + int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_state, struct intel_crtc_state *new_crtc_state, const struct intel_plane_state *old_plane_state,@@ -125,6 +141,7 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_ new_crtc_state->active_planes &= ~BIT (plane->id); new_crtc_state->nv12_planes &= ~BIT (plane->id); new_crtc_state->c8_planes &= ~BIT (plane->id); + new_crtc_state->data_rate[plane->id] =0 ; new_plane_state->base.visible = false;if (!new_plane_state->base.crtc && !old_plane_state->base.crtc)@@ -149,6 +166,9 @@ int intel_plane_atomic_check_with_state(const struct intel_crtc_state *old_crtc_if (new_plane_state->base.visible || old_plane_state->base.visible) new_crtc_state->update_planes |=BIT (plane->id); + new_crtc_state->data_rate[plane->id] = +intel_plane_data_rate (new_crtc_state, new_plane_state); + return intel_plane_atomic_calc_changes(old_crtc_state, &new_crtc_state->base, old_plane_state,drivers/gpu/drm/i915/intel_atomic_plane.h +2
-0@@ -15,6 +15,8 @@ struct intel_plane_state;extern const struct drm_plane_helper_funcs intel_plane_helper_funcs; +unsigned int intel_plane_data_rate (const struct intel_crtc_state *crtc_state, +const struct intel_plane_state *plane_state);void intel_update_plane (struct intel_plane *plane,const struct intel_crtc_state *crtc_state,const struct intel_plane_state *plane_state);drivers/gpu/drm/i915/intel_bw.c new file +181
-0@@ -0,0 +1,181 @@ +// SPDX-License-Identifier: MIT +/* +* Copyright © 2019 Intel Corporation +*/ + +#include <drm/drm_atomic_state_helper.h> + +#include "intel_bw.h" +#include "intel_drv.h" + +static unsigned int intel_bw_crtc_num_active_planes (const struct intel_crtc_state *crtc_state) +{ +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +return hweight8 (crtc_state->active_planes & ~BIT (PLANE_CURSOR )); +} + +static unsigned int intel_bw_crtc_data_rate (const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); +unsigned int data_rate =0 ; +enum plane_id plane_id; + +for_each_plane_id_on_crtc (crtc, plane_id) { +/* +* We assume cursors are small enough +* to not not cause bandwidth problems. +*/ +if (plane_id ==PLANE_CURSOR ) +continue ; + + data_rate += crtc_state->data_rate[plane_id]; + } + +return data_rate; +} + +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state) +{ +struct intel_crtc *crtc =to_intel_crtc (crtc_state->base.crtc); + + bw_state->data_rate[crtc->pipe] = +intel_bw_crtc_data_rate (crtc_state); + bw_state->num_active_planes[crtc->pipe] = +intel_bw_crtc_num_active_planes (crtc_state); + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); +} + +static unsigned int intel_bw_num_active_planes (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int num_active_planes =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + num_active_planes += bw_state->num_active_planes[pipe]; + +return num_active_planes; +} + +static unsigned int intel_bw_data_rate (struct drm_i915_private *dev_priv, +const struct intel_bw_state *bw_state) +{ +unsigned int data_rate =0 ; +enum pipe pipe; + +for_each_pipe (dev_priv, pipe) + data_rate += bw_state->data_rate[pipe]; + +return data_rate; +} + +int intel_bw_atomic_check (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct intel_crtc_state *new_crtc_state, *old_crtc_state; +struct intel_bw_state *bw_state =NULL ; +unsigned int data_rate, max_data_rate; +unsigned int num_active_planes; +struct intel_crtc *crtc; +int i; + +/* FIXME earlier gens need some checks too */ +if (INTEL_GEN (dev_priv) <11 ) +return 0 ; + +for_each_oldnew_intel_crtc_in_state (state, crtc, old_crtc_state, + new_crtc_state, i) { +unsigned int old_data_rate = +intel_bw_crtc_data_rate (old_crtc_state); +unsigned int new_data_rate = +intel_bw_crtc_data_rate (new_crtc_state); +unsigned int old_active_planes = +intel_bw_crtc_num_active_planes (old_crtc_state); +unsigned int new_active_planes = +intel_bw_crtc_num_active_planes (new_crtc_state); + +/* +* Avoid locking the bw state when +* nothing significant has changed. +*/ +if (old_data_rate == new_data_rate && + old_active_planes == new_active_planes) +continue ; + + bw_state =intel_atomic_get_bw_state (state); +if (IS_ERR (bw_state)) +return PTR_ERR (bw_state); + + bw_state->data_rate[crtc->pipe] = new_data_rate; + bw_state->num_active_planes[crtc->pipe] = new_active_planes; + +DRM_DEBUG_KMS ("pipe %c data rate %u num active planes %u\n" , +pipe_name (crtc->pipe), + bw_state->data_rate[crtc->pipe], + bw_state->num_active_planes[crtc->pipe]); + } + +if (!bw_state) +return 0 ; + + data_rate =intel_bw_data_rate (dev_priv, bw_state); + num_active_planes =intel_bw_num_active_planes (dev_priv, bw_state); + + max_data_rate =intel_max_data_rate (dev_priv, num_active_planes); + + data_rate =DIV_ROUND_UP (data_rate,1000 ); + +if (data_rate > max_data_rate) { +DRM_DEBUG_KMS ("Bandwidth %u MB/s exceeds max available %d MB/s (%d active planes)\n" , + data_rate, max_data_rate, num_active_planes); +return -EINVAL ; + } + +return 0 ; +} + +static struct drm_private_state *intel_bw_duplicate_state (struct drm_private_obj *obj) +{ +struct intel_bw_state *state; + + state =kmemdup (obj->state,sizeof (*state),GFP_KERNEL ); +if (!state) +return NULL ; + +__drm_atomic_helper_private_obj_duplicate_state (obj, &state->base); + +return &state->base; +} + +static void intel_bw_destroy_state (struct drm_private_obj *obj, +struct drm_private_state *state) +{ +kfree (state); +} + +static const struct drm_private_state_funcs intel_bw_funcs = { + .atomic_duplicate_state = intel_bw_duplicate_state, + .atomic_destroy_state = intel_bw_destroy_state, +}; + +int intel_bw_init (struct drm_i915_private *dev_priv) +{ +struct intel_bw_state *state; + + state =kzalloc (sizeof (*state),GFP_KERNEL ); +if (!state) +return -ENOMEM ; + +drm_atomic_private_obj_init (&dev_priv->drm, &dev_priv->bw_obj, + &state->base, &intel_bw_funcs); + +return 0 ; +}drivers/gpu/drm/i915/intel_bw.h new file +46
-0@@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2019 Intel Corporation +*/ + +#ifndef __INTEL_BW_H__ +#define __INTEL_BW_H__ + +#include <drm/drm_atomic.h> + +#include "i915_drv.h" +#include "intel_display.h" + +struct drm_i915_private ; +struct intel_atomic_state ; +struct intel_crtc_state ; + +struct intel_bw_state { +struct drm_private_state base; + +unsigned int data_rate[I915_MAX_PIPES ]; +u8 num_active_planes[I915_MAX_PIPES ]; +}; + +#define to_intel_bw_state (x) container_of((x), struct intel_bw_state, base) + +static inline struct intel_bw_state * +intel_atomic_get_bw_state (struct intel_atomic_state *state) +{ +struct drm_i915_private *dev_priv =to_i915 (state->base.dev); +struct drm_private_state *bw_state; + + bw_state =drm_atomic_get_private_obj_state (&state->base, + &dev_priv->bw_obj); +if (IS_ERR (bw_state)) +return ERR_CAST (bw_state); + +return to_intel_bw_state (bw_state); +} + +int intel_bw_init (struct drm_i915_private *dev_priv); +int intel_bw_atomic_check (struct intel_atomic_state *state); +void intel_bw_crtc_update (struct intel_bw_state *bw_state, +const struct intel_crtc_state *crtc_state); + +#endif /* __INTEL_BW_H__ */ drivers/gpu/drm/i915/intel_display.c +39
-1@@ -50,6 +50,7 @@#include "intel_acpi.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" +#include "intel_bw.h" #include "intel_color.h" #include "intel_cdclk.h" #include "intel_crt.h" @@ -2863,6 +2864,7 @@ static void intel_plane_disable_noatomic(struct intel_crtc *crtc,intel_set_plane_visible (crtc_state, plane_state, false);fixup_active_planes (crtc_state); + crtc_state->data_rate[plane->id] =0 ;if (plane->id ==PLANE_PRIMARY )intel_pre_disable_primary_noatomic (&crtc->base);@@ -6590,6 +6592,8 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc,struct intel_encoder *encoder;struct intel_crtc *intel_crtc =to_intel_crtc (crtc);struct drm_i915_private *dev_priv =to_i915 (crtc->dev); +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);enum intel_display_power_domain domain;struct intel_plane *plane;u64 domains;@@ -6652,6 +6656,9 @@ static void intel_crtc_disable_noatomic(struct drm_crtc *crtc, dev_priv->active_crtcs &= ~(1 << intel_crtc->pipe); dev_priv->min_cdclk[intel_crtc->pipe] =0 ; dev_priv->min_voltage_level[intel_crtc->pipe] =0 ; + + bw_state->data_rate[intel_crtc->pipe] =0 ; + bw_state->num_active_planes[intel_crtc->pipe] =0 ; }/* @@ -11176,6 +11183,7 @@ int intel_plane_atomic_calc_changes(const struct intel_crtc_state *old_crtc_statif (!is_crtc_enabled) { plane_state->visible = visible = false;to_intel_crtc_state (crtc_state)->active_planes &= ~BIT (plane->id); +to_intel_crtc_state (crtc_state)->data_rate[plane->id] =0 ; }if (!was_visible && !visible)@@ -13296,7 +13304,15 @@ static int intel_atomic_check(struct drm_device *dev,return ret;intel_fbc_choose_crtc (dev_priv, intel_state); -return calc_watermark_data (intel_state); + ret =calc_watermark_data (intel_state); +if (ret) +return ret; + + ret =intel_bw_atomic_check (intel_state); +if (ret) +return ret; + +return 0 ; } static int intel_atomic_prepare_commit(struct drm_device *dev,@@ -15696,6 +15712,10 @@ int intel_modeset_init(struct drm_device *dev)drm_mode_config_init (dev); + ret =intel_bw_init (dev_priv); +if (ret) +return ret; + dev->mode_config.min_width =0 ; dev->mode_config.min_height =0 ;@@ -16318,8 +16338,11 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev)drm_connector_list_iter_end (&conn_iter);for_each_intel_crtc (dev, crtc) { +struct intel_bw_state *bw_state = +to_intel_bw_state (dev_priv->bw_obj.state);struct intel_crtc_state *crtc_state =to_intel_crtc_state (crtc->base.state); +struct intel_plane *plane;int min_cdclk =0 ;memset (&crtc->base.mode,0 ,sizeof (crtc->base.mode));@@ -16358,6 +16381,21 @@ static void intel_modeset_readout_hw_state(struct drm_device *dev) dev_priv->min_voltage_level[crtc->pipe] = crtc_state->min_voltage_level; +for_each_intel_plane_on_crtc (&dev_priv->drm, crtc, plane) { +const struct intel_plane_state *plane_state = +to_intel_plane_state (plane->base.state); + +/* +* FIXME don't have the fb yet, so can't +* use intel_plane_data_rate() :( +*/ +if (plane_state->base.visible) + crtc_state->data_rate[plane->id] = +4 * crtc_state->pixel_rate; + } + +intel_bw_crtc_update (bw_state, crtc_state); +intel_pipe_config_sanity_check (dev_priv, crtc_state); } }drivers/gpu/drm/i915/intel_drv.h +0
-0
drivers/gpu/drm/i915/intel_drv.h struct intel_crtc_state { · line 892
Quoted 8 lines
drivers/gpu/drm/i915/intel_drv.h +2
-0@@ -885,6 +885,8 @@ struct intel_crtc_state {struct intel_crtc_wm_state wm; +u32 data_rate[I915_MAX_PLANES ]; +/* Gamma mode programmed on the pipe */ u32 gamma_mode;
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.CHECKPATCH: warning for series starting with [v3,1/2] drm/i915: Make sandybridge_pcode_read() deal with the second data register
In reply to Ville Syrjala== Series Details ==
== Series Details ==
Series: series starting with [v3,1/2] drm/i915: Make sandybridge_pcode_read() deal with the second data register
URL : https://patchwork.freedesktop.org/series/60271/
State : warning
== Summary ==
$ dim checkpatch origin/drm-tip f4b65571ca4a drm/i915: Make sandybridge_pcode_read() deal with the second data register 404a2b3b0603 drm/i915: Make sure we have enough memory bandwidth on ICL -:415: WARNING:FILE_PATH_CHANGES: added, moved or deleted file(s), does MAINTAINERS need updating? #415: new file mode 100644
total: 0 errors, 1 warnings, 0 checks, 668 lines checked
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.SPARSE: warning for series starting with [v3,1/2] drm/i915: Make sandybridge_pcode_read() deal with the second data register
In reply to Ville Syrjala== Series Details ==
== Series Details ==
Series: series starting with [v3,1/2] drm/i915: Make sandybridge_pcode_read() deal with the second data register
URL : https://patchwork.freedesktop.org/series/60271/
State : warning
== Summary ==
$ dim sparse origin/drm-tip Sparse version: v0.5.2 Commit: drm/i915: Make sandybridge_pcode_read() deal with the second data register Okay!
Commit: drm/i915: Make sure we have enough memory bandwidth on ICL +drivers/gpu/drm/i915/i915_drv.c:1555:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1555:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1593:19: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1593:19: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1595:20: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1595:20: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1615:30: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1615:30: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1619:44: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1619:44: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +drivers/gpu/drm/i915/i915_drv.c:1658:24: warning: expression using sizeof(void) +./include/uapi/linux/perf_event.h:147:56: warning: cast truncates bits from constant value (8000000000000000 becomes 0)
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✓ Fi.CI.BAT: success for series starting with [v3,1/2] drm/i915: Make sandybridge_pcode_read() deal with the second data register
In reply to Ville Syrjala== Series Details ==
== Series Details ==
Series: series starting with [v3,1/2] drm/i915: Make sandybridge_pcode_read() deal with the second data register
URL : https://patchwork.freedesktop.org/series/60271/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6041 -> Patchwork_12962
====================================================
Summary
-------
**SUCCESS**
No regressions found.
External URL: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/
Known issues
------------
Here are the changes found in Patchwork_12962 that come from known issues:
### IGT changes ###
#### Issues hit ####
* igt@gem_exec_suspend@basic-s3:
- fi-blb-e6850: [PASS][1] -> [INCOMPLETE][2] ([fdo#107718] / [fdo#110581])
[1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/fi-blb-e6850/igt@[email protected]
[2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/fi-blb-e6850/igt@[email protected]
* igt@i915_pm_rpm@module-reload:
- fi-skl-6770hq: [PASS][3] -> [FAIL][4] ([fdo#108511])
[3]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/fi-skl-6770hq/igt@[email protected]
[4]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/fi-skl-6770hq/igt@[email protected]
* igt@i915_selftest@live_hangcheck:
- fi-icl-u3: [PASS][5] -> [INCOMPLETE][6] ([fdo#107713] / [fdo#108569] / [fdo#110581])
[5]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/fi-icl-u3/igt@i915_selftest@live_hangcheck.html
[6]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/fi-icl-u3/igt@i915_selftest@live_hangcheck.html
#### Possible fixes ####
* igt@i915_selftest@live_hangcheck:
- fi-skl-iommu: [INCOMPLETE][7] ([fdo#108602] / [fdo#108744] / [fdo#110581]) -> [PASS][8]
[7]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/fi-skl-iommu/igt@i915_selftest@live_hangcheck.html
[8]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/fi-skl-iommu/igt@i915_selftest@live_hangcheck.html
* igt@kms_chamelium@hdmi-hpd-fast:
- fi-kbl-7500u: [FAIL][9] ([fdo#109485]) -> [PASS][10]
[9]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/fi-kbl-7500u/igt@[email protected]
[10]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/fi-kbl-7500u/igt@[email protected]
* igt@kms_cursor_legacy@basic-busy-flip-before-cursor-legacy:
- {fi-icl-u2}: [DMESG-WARN][11] -> [PASS][12]
[11]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/fi-icl-u2/igt@[email protected]
[12]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/fi-icl-u2/igt@[email protected]
{name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).
[fdo#102505]: https://bugs.freedesktop.org/show_bug.cgi?id=102505
[fdo#107713]: https://bugs.freedesktop.org/show_bug.cgi?id=107713
[fdo#107718]: https://bugs.freedesktop.org/show_bug.cgi?id=107718
[fdo#108511]: https://bugs.freedesktop.org/show_bug.cgi?id=108511
[fdo#108569]: https://bugs.freedesktop.org/show_bug.cgi?id=108569
[fdo#108602]: https://bugs.freedesktop.org/show_bug.cgi?id=108602
[fdo#108744]: https://bugs.freedesktop.org/show_bug.cgi?id=108744
[fdo#109485]: https://bugs.freedesktop.org/show_bug.cgi?id=109485
[fdo#110581]: https://bugs.freedesktop.org/show_bug.cgi?id=110581
[fdo#110595]: https://bugs.freedesktop.org/show_bug.cgi?id=110595
Participating hosts (51 -> 45)
------------------------------
Additional (2): fi-bsw-n3050 fi-pnv-d510
Missing (8): fi-kbl-soraka fi-ilk-m540 fi-hsw-4200u fi-byt-squawks fi-bsw-cyan fi-icl-y fi-byt-clapper fi-bdw-samus
Build changes
-------------
* Linux: CI_DRM_6041 -> Patchwork_12962
CI_DRM_6041: 014903e8b7de5d69a17de628345ed31db1600b73 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4972: f052e49a43cc9704ea5f240df15dd9d3dfed68ab @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_12962: 404a2b3b06037095b181fcf288f6c56328d3a174 @ git://anongit.freedesktop.org/gfx-ci/linux
== Linux commits ==
404a2b3b0603 drm/i915: Make sure we have enough memory bandwidth on ICL f4b65571ca4a drm/i915: Make sandybridge_pcode_read() deal with the second data register
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✓ Fi.CI.IGT: success for series starting with [v3,1/2] drm/i915: Make sandybridge_pcode_read() deal with the second data register
In reply to Ville Syrjala== Series Details ==
== Series Details ==
Series: series starting with [v3,1/2] drm/i915: Make sandybridge_pcode_read() deal with the second data register
URL : https://patchwork.freedesktop.org/series/60271/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6041_full -> Patchwork_12962_full
====================================================
Summary
-------
**SUCCESS**
No regressions found.
Known issues
------------
Here are the changes found in Patchwork_12962_full that come from known issues:
### IGT changes ###
#### Issues hit ####
* igt@gem_eio@reset-stress:
- shard-skl: [PASS][1] -> [FAIL][2] ([fdo#105957])
[1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl1/igt@[email protected]
[2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl1/igt@[email protected]
* igt@gem_workarounds@suspend-resume:
- shard-apl: [PASS][3] -> [DMESG-WARN][4] ([fdo#108566])
[3]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-apl2/igt@[email protected]
[4]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-apl1/igt@[email protected]
* igt@i915_pm_rpm@drm-resources-equal:
- shard-skl: [PASS][5] -> [INCOMPLETE][6] ([fdo#107807] / [fdo#110581])
[5]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl10/igt@[email protected]
[6]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl8/igt@[email protected]
* igt@kms_cursor_crc@cursor-256x256-suspend:
- shard-kbl: [PASS][7] -> [DMESG-WARN][8] ([fdo#103313])
[7]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-kbl2/igt@[email protected]
[8]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-kbl5/igt@[email protected]
* igt@kms_draw_crc@draw-method-xrgb8888-render-xtiled:
- shard-skl: [PASS][9] -> [FAIL][10] ([fdo#103184] / [fdo#103232])
[9]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl8/igt@[email protected]
[10]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl10/igt@[email protected]
* igt@kms_frontbuffer_tracking@fbc-1p-primscrn-cur-indfb-draw-blt:
- shard-iclb: [PASS][11] -> [FAIL][12] ([fdo#103167]) +3 similar issues
[11]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb6/igt@[email protected]
[12]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb4/igt@[email protected]
* igt@kms_pipe_crc_basic@suspend-read-crc-pipe-b:
- shard-snb: [PASS][13] -> [DMESG-WARN][14] ([fdo#102365])
[13]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-snb5/igt@[email protected]
[14]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-snb1/igt@[email protected]
* igt@kms_plane_multiple@atomic-pipe-c-tiling-yf:
- shard-apl: [PASS][15] -> [DMESG-WARN][16] ([fdo#103558] / [fdo#105602]) +19 similar issues
[15]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-apl2/igt@[email protected]
[16]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-apl1/igt@[email protected]
* igt@kms_plane_scaling@pipe-a-scaler-with-pixel-format:
- shard-glk: [PASS][17] -> [SKIP][18] ([fdo#109271] / [fdo#109278]) +1 similar issue
[17]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-glk9/igt@[email protected]
[18]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-glk1/igt@[email protected]
* igt@kms_psr2_su@page_flip:
- shard-iclb: [PASS][19] -> [SKIP][20] ([fdo#109642])
[19]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb2/igt@kms_psr2_su@page_flip.html
[20]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb5/igt@kms_psr2_su@page_flip.html
* igt@kms_psr@psr2_primary_page_flip:
- shard-iclb: [PASS][21] -> [SKIP][22] ([fdo#109441]) +1 similar issue
[21]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb2/igt@kms_psr@psr2_primary_page_flip.html
[22]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb6/igt@kms_psr@psr2_primary_page_flip.html
* igt@kms_rotation_crc@multiplane-rotation:
- shard-glk: [PASS][23] -> [DMESG-FAIL][24] ([fdo#105763] / [fdo#106538])
[23]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-glk8/igt@[email protected]
[24]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-glk1/igt@[email protected]
* igt@kms_setmode@basic:
- shard-kbl: [PASS][25] -> [FAIL][26] ([fdo#99912])
[25]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-kbl2/igt@[email protected]
[26]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-kbl5/igt@[email protected]
* igt@kms_sysfs_edid_timing:
- shard-iclb: [PASS][27] -> [FAIL][28] ([fdo#100047])
[27]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb1/igt@kms_sysfs_edid_timing.html
[28]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb3/igt@kms_sysfs_edid_timing.html
#### Possible fixes ####
* igt@debugfs_test@read_all_entries_display_off:
- shard-skl: [INCOMPLETE][29] ([fdo#104108] / [fdo#110581]) -> [PASS][30]
[29]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl5/igt@debugfs_test@read_all_entries_display_off.html
[30]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl5/igt@debugfs_test@read_all_entries_display_off.html
* igt@gem_tiled_swapping@non-threaded:
- shard-snb: [DMESG-WARN][31] ([fdo#108686]) -> [PASS][32]
[31]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-snb6/igt@[email protected]
[32]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-snb4/igt@[email protected]
* igt@i915_pm_rpm@i2c:
- shard-iclb: [FAIL][33] ([fdo#104097]) -> [PASS][34]
[33]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb6/igt@[email protected]
[34]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb4/igt@[email protected]
* igt@i915_pm_rpm@pm-tiling:
- shard-skl: [INCOMPLETE][35] ([fdo#107807] / [fdo#110581]) -> [PASS][36]
[35]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl6/igt@[email protected]
[36]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl7/igt@[email protected]
* igt@kms_dp_dsc@basic-dsc-enable-edp:
- shard-iclb: [SKIP][37] ([fdo#109349]) -> [PASS][38]
[37]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb8/igt@[email protected]
[38]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb2/igt@[email protected]
* igt@kms_flip@flip-vs-expired-vblank-interruptible:
- shard-skl: [FAIL][39] ([fdo#105363]) -> [PASS][40]
[39]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl2/igt@[email protected]
[40]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl9/igt@[email protected]
* igt@kms_flip@flip-vs-suspend:
- shard-skl: [INCOMPLETE][41] ([fdo#107773] / [fdo#109507] / [fdo#110581]) -> [PASS][42]
[41]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl5/igt@[email protected]
[42]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl7/igt@[email protected]
* igt@kms_frontbuffer_tracking@fbc-suspend:
- shard-apl: [DMESG-WARN][43] ([fdo#108566]) -> [PASS][44] +3 similar issues
[43]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-apl3/igt@[email protected]
[44]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-apl4/igt@[email protected]
* igt@kms_frontbuffer_tracking@fbcpsr-1p-primscrn-pri-indfb-draw-blt:
- shard-iclb: [FAIL][45] ([fdo#103167]) -> [PASS][46] +4 similar issues
[45]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb4/igt@kms_frontbuffer_tracking@fbcpsr-1p-primscrn-pri-indfb-draw-blt.html
[46]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb1/igt@kms_frontbuffer_tracking@fbcpsr-1p-primscrn-pri-indfb-draw-blt.html
* igt@kms_plane_alpha_blend@pipe-b-constant-alpha-min:
- shard-skl: [FAIL][47] ([fdo#108145]) -> [PASS][48]
[47]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl9/igt@[email protected]
[48]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl2/igt@[email protected]
* igt@kms_plane_lowres@pipe-a-tiling-x:
- shard-iclb: [FAIL][49] ([fdo#103166]) -> [PASS][50]
[49]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb6/igt@[email protected]
[50]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb4/igt@[email protected]
* igt@kms_psr@psr2_cursor_mmap_cpu:
- shard-iclb: [SKIP][51] ([fdo#109441]) -> [PASS][52] +1 similar issue
[51]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-iclb1/igt@kms_psr@psr2_cursor_mmap_cpu.html
[52]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-iclb2/igt@kms_psr@psr2_cursor_mmap_cpu.html
#### Warnings ####
* igt@kms_concurrent@pipe-d:
- shard-apl: [SKIP][53] ([fdo#109271] / [fdo#109278]) -> [SKIP][54] ([fdo#105602] / [fdo#109271] / [fdo#109278]) +2 similar issues
[53]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-apl2/igt@[email protected]
[54]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-apl1/igt@[email protected]
* igt@kms_frontbuffer_tracking@fbcpsr-1p-primscrn-pri-indfb-draw-mmap-gtt:
- shard-skl: [FAIL][55] ([fdo#103167]) -> [FAIL][56] ([fdo#108040])
[55]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-skl8/igt@kms_frontbuffer_tracking@fbcpsr-1p-primscrn-pri-indfb-draw-mmap-gtt.html
[56]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-skl10/igt@kms_frontbuffer_tracking@fbcpsr-1p-primscrn-pri-indfb-draw-mmap-gtt.html
* igt@kms_frontbuffer_tracking@psr-slowdraw:
- shard-apl: [SKIP][57] ([fdo#109271]) -> [SKIP][58] ([fdo#105602] / [fdo#109271]) +15 similar issues
[57]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-apl2/igt@[email protected]
[58]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-apl1/igt@[email protected]
* igt@kms_plane_alpha_blend@pipe-b-alpha-basic:
- shard-apl: [FAIL][59] ([fdo#108145]) -> [DMESG-FAIL][60] ([fdo#103558] / [fdo#105602] / [fdo#108145])
[59]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6041/shard-apl2/igt@[email protected]
[60]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/shard-apl1/igt@[email protected]
[fdo#100047]: https://bugs.freedesktop.org/show_bug.cgi?id=100047
[fdo#102365]: https://bugs.freedesktop.org/show_bug.cgi?id=102365
[fdo#103166]: https://bugs.freedesktop.org/show_bug.cgi?id=103166
[fdo#103167]: https://bugs.freedesktop.org/show_bug.cgi?id=103167
[fdo#103184]: https://bugs.freedesktop.org/show_bug.cgi?id=103184
[fdo#103232]: https://bugs.freedesktop.org/show_bug.cgi?id=103232
[fdo#103313]: https://bugs.freedesktop.org/show_bug.cgi?id=103313
[fdo#103558]: https://bugs.freedesktop.org/show_bug.cgi?id=103558
[fdo#104097]: https://bugs.freedesktop.org/show_bug.cgi?id=104097
[fdo#104108]: https://bugs.freedesktop.org/show_bug.cgi?id=104108
[fdo#105363]: https://bugs.freedesktop.org/show_bug.cgi?id=105363
[fdo#105602]: https://bugs.freedesktop.org/show_bug.cgi?id=105602
[fdo#105763]: https://bugs.freedesktop.org/show_bug.cgi?id=105763
[fdo#105957]: https://bugs.freedesktop.org/show_bug.cgi?id=105957
[fdo#106538]: https://bugs.freedesktop.org/show_bug.cgi?id=106538
[fdo#107773]: https://bugs.freedesktop.org/show_bug.cgi?id=107773
[fdo#107807]: https://bugs.freedesktop.org/show_bug.cgi?id=107807
[fdo#108040]: https://bugs.freedesktop.org/show_bug.cgi?id=108040
[fdo#108145]: https://bugs.freedesktop.org/show_bug.cgi?id=108145
[fdo#108566]: https://bugs.freedesktop.org/show_bug.cgi?id=108566
[fdo#108686]: https://bugs.freedesktop.org/show_bug.cgi?id=108686
[fdo#109271]: https://bugs.freedesktop.org/show_bug.cgi?id=109271
[fdo#109278]: https://bugs.freedesktop.org/show_bug.cgi?id=109278
[fdo#109349]: https://bugs.freedesktop.org/show_bug.cgi?id=109349
[fdo#109441]: https://bugs.freedesktop.org/show_bug.cgi?id=109441
[fdo#109507]: https://bugs.freedesktop.org/show_bug.cgi?id=109507
[fdo#109642]: https://bugs.freedesktop.org/show_bug.cgi?id=109642
[fdo#110581]: https://bugs.freedesktop.org/show_bug.cgi?id=110581
[fdo#99912]: https://bugs.freedesktop.org/show_bug.cgi?id=99912
Participating hosts (10 -> 10)
------------------------------
No changes in participating hosts
Build changes
-------------
* Linux: CI_DRM_6041 -> Patchwork_12962
CI_DRM_6041: 014903e8b7de5d69a17de628345ed31db1600b73 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4972: f052e49a43cc9704ea5f240df15dd9d3dfed68ab @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_12962: 404a2b3b06037095b181fcf288f6c56328d3a174 @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4509: fdc5a4ca11124ab8413c7988896eec4c97336694 @ git://anongit.freedesktop.org/piglit
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_12962/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Very straight forward. Nit variable names val and val1, maybe val0 and val1.
-Clint
Quoted 157 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
The pcode mailbox has two data registers. So far we've only ever used the one, but that's about to change. Expose the second data register to the callers of
sandybridge_pcode_read().Signed-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/i915_debugfs.c | 4 ++-- drivers/gpu/drm/i915/intel_pm.c | 12 +++++++----- drivers/gpu/drm/i915/intel_sideband.c | 15 +++++++++------ drivers/gpu/drm/i915/intel_sideband.h | 3 ++- 4 files changed, 20 insertions(+), 14 deletions(-)drivers/gpu/drm/i915/i915_debugfs.c +2
-2@@ -1494,7 +1494,7 @@ static int gen6_drpc_info(struct seq_file *m)if (INTEL_GEN (dev_priv) <=7 )sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , - &rc6vids); + &rc6vids,NULL );seq_printf (m,"RC1e Enabled: %s\n" ,yesno (rcctl1 & GEN6_RC_CTL_RC1e_ENABLE));@@ -1777,7 +1777,7 @@ static int i915_ring_freq_table(struct seq_file *m, void *unused) ia_freq = gpu_freq;sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_MIN_FREQ_TABLE , - &ia_freq); + &ia_freq,NULL ); seq_printf(m, "%d\t\t%d\t\t\t\t%d\n", intel_gpu_freq(dev_priv, (gpu_freq * (IS_GEN9_BC(dev_priv) ||drivers/gpu/drm/i915/intel_pm.c +7
-5@@ -2822,7 +2822,7 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv, val =0 ;/* data0 to be programmed to 0 for first set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENCY , - &val); + &val,NULL );if (ret) {DRM_ERROR ("SKL Mailbox read error = %d\n" , ret);@@ -2841,7 +2841,7 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv, val =1 ;/* data0 to be programmed to 1 for second set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENCY , - &val); + &val,NULL );if (ret) {DRM_ERROR ("SKL Mailbox read error = %d\n" , ret);return ;@@ -7061,7 +7061,7 @@ static void gen6_init_rps_frequencies(struct drm_i915_private *dev_priv)if (sandybridge_pcode_read (dev_priv,HSW_PCODE_DYNAMIC_DUTY_CYCLE_CONTROL , - &ddcc_status) ==0 ) + &ddcc_status,NULL ) ==0 ) rps->efficient_freq = clamp_t(u8, ((ddcc_status >> 8) & 0xff),@@ -7408,7 +7408,8 @@ static void gen6_enable_rc6(struct drm_i915_private *dev_priv) GEN6_RC_CTL_HW_ENABLE); rc6vids =0 ; - ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , &rc6vids); + ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , + &rc6vids,NULL );if (IS_GEN (dev_priv,6 ) && ret) {DRM_DEBUG_DRIVER ("Couldn't check for BIOS workaround\n" ); }else if (IS_GEN (dev_priv,6 ) && (GEN6_DECODE_RC6_VID (rc6vids &0xff ) <450 )) {@@ -8555,7 +8556,8 @@ void intel_init_gt_powersave(struct drm_i915_private *dev_priv) IS_IVYBRIDGE(dev_priv) || IS_HASWELL(dev_priv)) {u32 params =0 ; -sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , ¶ms); +sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , + ¶ms,NULL );if (params &BIT (31 )) {/* OC supported */ DRM_DEBUG_DRIVER("Overclocking supported, max: %dMHz, overclock: %dMHz\n", (rps->max_freq & 0xff) * 50,drivers/gpu/drm/i915/intel_sideband.c +9
-6@@ -374,7 +374,7 @@ static inline int gen7_check_mailbox_status(u32 mbox) }static int __sandybridge_pcode_rw (struct drm_i915_private *i915, -u32 mbox,u32 *val, +u32 mbox,u32 *val,u32 *val1,int fast_timeout_us,int slow_timeout_ms,bool is_read)@@ -393,7 +393,7 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,return -EAGAIN ;intel_uncore_write_fw (uncore,GEN6_PCODE_DATA , *val); -intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 ,0 ); +intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 , val1 ? *val1 :0 );intel_uncore_write_fw (uncore,GEN6_PCODE_MAILBOX ,GEN6_PCODE_READY | mbox);@@ -407,6 +407,8 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,if (is_read) *val =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA ); +if (is_read && val1) + *val1 =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA1 );if (INTEL_GEN (i915) >6 )return gen7_check_mailbox_status (mbox);@@ -414,12 +416,13 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,return gen6_check_mailbox_status (mbox); } -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox,u32 *val) +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1) {int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, val, + err =__sandybridge_pcode_rw (i915, mbox, val, val1,500 ,0 , true);mutex_unlock (&i915->sb_lock);@@ -440,7 +443,7 @@ int sandybridge_pcode_write_timeout(struct drm_i915_private *i915,int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, &val, + err =__sandybridge_pcode_rw (i915, mbox, &val,NULL , fast_timeout_us, slow_timeout_ms, false);mutex_unlock (&i915->sb_lock);@@ -457,7 +460,7 @@ static bool skl_pcode_try_request(struct drm_i915_private *i915, u32 mbox, u32 request, u32 reply_mask, u32 reply, u32 *status) { - *status =__sandybridge_pcode_rw (i915, mbox, &request, + *status =__sandybridge_pcode_rw (i915, mbox, &request,NULL ,500 ,0 , true);drivers/gpu/drm/i915/intel_sideband.h +0
-0
drivers/gpu/drm/i915/intel_sideband.h u32 intel_sbi_read(struct drm_i915_private *i915, u16 reg, · line 134
Quoted 9 lines
drivers/gpu/drm/i915/intel_sideband.h +2
-1@@ -127,7 +127,8 @@ u32 intel_sbi_read(struct drm_i915_private *i915, u16 reg,void intel_sbi_write (struct drm_i915_private *i915,u16 reg,u32 value,enum intel_sbi_destination destination); -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox,u32 *val); +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1);int sandybridge_pcode_write_timeout (struct drm_i915_private *i915,u32 mbox,u32 val,int fast_timeout_us,int slow_timeout_ms);
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 1 line · Clinton Taylor
Very straight forward. Nit variable names val and val1, maybe val0 and val1.
The registers are named DATA and DATA1, so I called the variables val and val1. I guess I could have renamed them to data and data1 to make the relationship even more explicit.
Quoted 173 lines
Reviewed-by Clint Taylor <[email protected]>-Clint
Quoted 166 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
The pcode mailbox has two data registers. So far we've only ever used the one, but that's about to change. Expose the second data register to the callers of
sandybridge_pcode_read().Signed-off-by Ville Syrjälä <[email protected]>--- drivers/gpu/drm/i915/i915_debugfs.c | 4 ++-- drivers/gpu/drm/i915/intel_pm.c | 12 +++++++----- drivers/gpu/drm/i915/intel_sideband.c | 15 +++++++++------ drivers/gpu/drm/i915/intel_sideband.h | 3 ++- 4 files changed, 20 insertions(+), 14 deletions(-)drivers/gpu/drm/i915/i915_debugfs.c +2
-2@@ -1494,7 +1494,7 @@ static int gen6_drpc_info(struct seq_file *m)if (INTEL_GEN (dev_priv) <=7 )sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , - &rc6vids); + &rc6vids,NULL );seq_printf (m,"RC1e Enabled: %s\n" ,yesno (rcctl1 & GEN6_RC_CTL_RC1e_ENABLE));@@ -1777,7 +1777,7 @@ static int i915_ring_freq_table(struct seq_file *m, void *unused) ia_freq = gpu_freq;sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_MIN_FREQ_TABLE , - &ia_freq); + &ia_freq,NULL ); seq_printf(m, "%d\t\t%d\t\t\t\t%d\n", intel_gpu_freq(dev_priv, (gpu_freq * (IS_GEN9_BC(dev_priv) ||drivers/gpu/drm/i915/intel_pm.c +7
-5@@ -2822,7 +2822,7 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv, val =0 ;/* data0 to be programmed to 0 for first set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENCY , - &val); + &val,NULL );if (ret) {DRM_ERROR ("SKL Mailbox read error = %d\n" , ret);@@ -2841,7 +2841,7 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv, val =1 ;/* data0 to be programmed to 1 for second set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENCY , - &val); + &val,NULL );if (ret) {DRM_ERROR ("SKL Mailbox read error = %d\n" , ret);return ;@@ -7061,7 +7061,7 @@ static void gen6_init_rps_frequencies(struct drm_i915_private *dev_priv)if (sandybridge_pcode_read (dev_priv,HSW_PCODE_DYNAMIC_DUTY_CYCLE_CONTROL , - &ddcc_status) ==0 ) + &ddcc_status,NULL ) ==0 ) rps->efficient_freq = clamp_t(u8, ((ddcc_status >> 8) & 0xff),@@ -7408,7 +7408,8 @@ static void gen6_enable_rc6(struct drm_i915_private *dev_priv) GEN6_RC_CTL_HW_ENABLE); rc6vids =0 ; - ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , &rc6vids); + ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , + &rc6vids,NULL );if (IS_GEN (dev_priv,6 ) && ret) {DRM_DEBUG_DRIVER ("Couldn't check for BIOS workaround\n" ); }else if (IS_GEN (dev_priv,6 ) && (GEN6_DECODE_RC6_VID (rc6vids &0xff ) <450 )) {@@ -8555,7 +8556,8 @@ void intel_init_gt_powersave(struct drm_i915_private *dev_priv) IS_IVYBRIDGE(dev_priv) || IS_HASWELL(dev_priv)) {u32 params =0 ; -sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , ¶ms); +sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , + ¶ms,NULL );if (params &BIT (31 )) {/* OC supported */ DRM_DEBUG_DRIVER("Overclocking supported, max: %dMHz, overclock: %dMHz\n", (rps->max_freq & 0xff) * 50,drivers/gpu/drm/i915/intel_sideband.c +9
-6@@ -374,7 +374,7 @@ static inline int gen7_check_mailbox_status(u32 mbox) }static int __sandybridge_pcode_rw (struct drm_i915_private *i915, -u32 mbox,u32 *val, +u32 mbox,u32 *val,u32 *val1,int fast_timeout_us,int slow_timeout_ms,bool is_read)@@ -393,7 +393,7 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,return -EAGAIN ;intel_uncore_write_fw (uncore,GEN6_PCODE_DATA , *val); -intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 ,0 ); +intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 , val1 ? *val1 :0 );intel_uncore_write_fw (uncore,GEN6_PCODE_MAILBOX ,GEN6_PCODE_READY | mbox);@@ -407,6 +407,8 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,if (is_read) *val =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA ); +if (is_read && val1) + *val1 =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA1 );if (INTEL_GEN (i915) >6 )return gen7_check_mailbox_status (mbox);@@ -414,12 +416,13 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,return gen6_check_mailbox_status (mbox); } -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox,u32 *val) +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1) {int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, val, + err =__sandybridge_pcode_rw (i915, mbox, val, val1,500 ,0 , true);mutex_unlock (&i915->sb_lock);@@ -440,7 +443,7 @@ int sandybridge_pcode_write_timeout(struct drm_i915_private *i915,int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, &val, + err =__sandybridge_pcode_rw (i915, mbox, &val,NULL , fast_timeout_us, slow_timeout_ms, false);mutex_unlock (&i915->sb_lock);@@ -457,7 +460,7 @@ static bool skl_pcode_try_request(struct drm_i915_private *i915, u32 mbox, u32 request, u32 reply_mask, u32 reply, u32 *status) { - *status =__sandybridge_pcode_rw (i915, mbox, &request, + *status =__sandybridge_pcode_rw (i915, mbox, &request,NULL ,500 ,0 , true);drivers/gpu/drm/i915/intel_sideband.h +2
-1@@ -127,7 +127,8 @@ u32 intel_sbi_read(struct drm_i915_private *i915, u16 reg,void intel_sbi_write (struct drm_i915_private *i915,u16 reg,u32 value,enum intel_sbi_destination destination); -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox,u32 *val); +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1);int sandybridge_pcode_write_timeout (struct drm_i915_private *i915,u32 mbox,u32 val,int fast_timeout_us,int slow_timeout_ms);
Signature · 6 lines
Ville Syrjälä Intel _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 8 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
The pcode mailbox has two data registers. So far we've only ever used the one, but that's about to change. Expose the second data register to
the callers ofsandybridge_pcode_read().Signed-off-by Ville Syrjälä <[email protected]>
LGTM
Quoted 195 lines
--- drivers/gpu/drm/i915/i915_debugfs.c | 4 ++-- drivers/gpu/drm/i915/intel_pm.c | 12 +++++++----- drivers/gpu/drm/i915/intel_sideband.c | 15 +++++++++------ drivers/gpu/drm/i915/intel_sideband.h | 3 ++- 4 files changed, 20 insertions(+), 14 deletions(-)drivers/gpu/drm/i915/i915_debugfs.c +1
-1@@ -1494,7 +1494,7 @@ static int gen6_drpc_info(struct seq_file *m)if (INTEL_GEN (dev_priv) <=7 )sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , - &rc6vids); + &rc6vids,NULL ); seq_printf(m, "RC1e Enabled: %s\n",drivers/gpu/drm/i915/i915_debugfs.c +1
-1yesno(rcctl1 & GEN6_RC_CTL_RC1e_ENABLE));@@ -1777,7 +1777,7 @@ static int i915_ring_freq_table(struct seq_file *m, void *unused) ia_freq = gpu_freq;sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_MIN_FREQ_TABLE , - &ia_freq); + &ia_freq,NULL ); seq_printf(m, "%d\t\t%d\t\t\t\t%d\n", intel_gpu_freq(dev_priv, (gpu_freq *(IS_GEN9_BC(dev_pr
iv) ||drivers/gpu/drm/i915/intel_pm.c +1
-1@@ -2822,7 +2822,7 @@ static void intel_read_wm_latency(structdrm_i915_private *dev_priv, val =0 ;/* data0 to be programmed to 0 for first set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENC Y, - &val); + &val,NULL );drivers/gpu/drm/i915/intel_pm.c +0
-0if (ret) { DRM_ERROR("SKL Mailbox read error = %d\n",ret);
drivers/gpu/drm/i915/intel_pm.c +1
-1@@ -2841,7 +2841,7 @@ static void intel_read_wm_latency(structdrm_i915_private *dev_priv, val =1 ;/* data0 to be programmed to 1 for second set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENC Y, - &val); + &val,NULL );drivers/gpu/drm/i915/intel_pm.c +0
-0if (ret) { DRM_ERROR("SKL Mailbox read error = %d\n",ret);
drivers/gpu/drm/i915/intel_pm.c +1
-1return ;@@ -7061,7 +7061,7 @@ static void gen6_init_rps_frequencies(struct drm_i915_private *dev_priv) if (sandybridge_pcode_read(dev_priv,HSW_PCODE_DYNAMIC_DUTY_CYCLE _CONTROL, - &ddcc_status) == 0) + &ddcc_status, NULL) == 0) rps->efficient_freq =drivers/gpu/drm/i915/intel_pm.c +2
-1clamp_t(u8, ((ddcc_status >> 8) & 0xff),@@ -7408,7 +7408,8 @@ static void gen6_enable_rc6(struct drm_i915_private *dev_priv) GEN6_RC_CTL_HW_ENABLE); rc6vids =0 ; - ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , &rc6vids); + ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , + &rc6vids,NULL );if (IS_GEN (dev_priv,6 ) && ret) {DRM_DEBUG_DRIVER("Couldn't check for BIOS
workaround\n");
} else if (IS_GEN(dev_priv, 6)&& (GEN6_DECODE_RC6_VID(rc6vids
&0xff) < 450)) {drivers/gpu/drm/i915/intel_pm.c +2
-1@@ -8555,7 +8556,8 @@ void intel_init_gt_powersave(struct drm_i915_private *dev_priv) IS_IVYBRIDGE(dev_priv) || IS_HASWELL(dev_priv)) {u32 params =0 ; -sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , ¶ms); +sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , + ¶ms,NULL );if (params &BIT (31 )) {/* OC supported */ DRM_DEBUG_DRIVER("Overclocking supported, max:
%dMHz, overclock: %dMHz\n",drivers/gpu/drm/i915/intel_pm.c +0
-0(rps->max_freq & 0xff) * 50,drivers/gpu/drm/i915/intel_sideband.c +1
-1@@ -374,7 +374,7 @@ static inline int gen7_check_mailbox_status(u32 mbox) } static int __sandybridge_pcode_rw(struct drm_i915_private *i915, - u32 mbox, u32 *val, + u32 mbox, u32 *val, u32 *val1, int fast_timeout_us, int slow_timeout_ms,drivers/gpu/drm/i915/intel_sideband.c +1
-1bool is_read)@@ -393,7 +393,7 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915, return -EAGAIN ;intel_uncore_write_fw (uncore,GEN6_PCODE_DATA , *val); -intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 ,0 ); +intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 , val1 ? *val1 :0 ); intel_uncore_write_fw(uncore,
GEN6_PCODE_MAILBOX,GEN6_PCODE_READY|
mbox);drivers/gpu/drm/i915/intel_sideband.c +2
-0@@ -407,6 +407,8 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915, if (is_read) *val =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA ); +if (is_read && val1) + *val1 =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA1 );if (INTEL_GEN (i915) >6 )drivers/gpu/drm/i915/intel_sideband.c +3
-2return gen7_check_mailbox_status (mbox);@@ -414,12 +416,13 @@ static int __sandybridge_pcode_rw(structdrm_i915_private *i915, return gen6_check_mailbox_status(mbox); } -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, u32 *val) +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1) {int err;mutex_lock (&i915->sb_lock); - err = __sandybridge_pcode_rw(i915, mbox, val, + err = __sandybridge_pcode_rw(i915, mbox, val, val1, 500, 0,drivers/gpu/drm/i915/intel_sideband.c +1
-1true);mutex_unlock (&i915->sb_lock);@@ -440,7 +443,7 @@ int sandybridge_pcode_write_timeout(structdrm_i915_private *i915, int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, &val, + err =__sandybridge_pcode_rw (i915, mbox, &val,NULL , fast_timeout_us, slow_timeout_ms, false);drivers/gpu/drm/i915/intel_sideband.c +1
-1mutex_unlock (&i915->sb_lock);@@ -457,7 +460,7 @@ static bool skl_pcode_try_request(struct drm_i915_private *i915, u32 mbox, u32 request, u32 reply_mask, u32 reply, u32 *status) { - *status = __sandybridge_pcode_rw(i915, mbox, &request, + *status = __sandybridge_pcode_rw(i915, mbox, &request, NULL, 500, 0,drivers/gpu/drm/i915/intel_sideband.c +0
-0true);drivers/gpu/drm/i915/intel_sideband.h +2
-1@@ -127,7 +127,8 @@ u32 intel_sbi_read(struct drm_i915_private *i915,u16 reg, void intel_sbi_write(struct drm_i915_private *i915, u16 reg, u32 value,enum intel_sbi_destination destination); -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, u32 *val); +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1);int sandybridge_pcode_write_timeout(struct
drm_i915_private*i915,
u32 mbox,
drivers/gpu/drm/i915/intel_sideband.h u32 intel_sbi_read(struct drm_i915_private *i915, u16 reg, · line 134
Quoted 2 lines
drivers/gpu/drm/i915/intel_sideband.h +0
-0u32 val, int fast_timeout_us, int slow_timeout_ms);
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 7 lines · Ville Syrjala
From: Ville Syrjälä <[email protected]>
The pcode mailbox has two data registers. So far we've only ever used the one, but that's about to change. Expose the second data register to the callers of
sandybridge_pcode_read().Signed-off-by Ville Syrjälä <[email protected]>
Quoted 166 lines
--- drivers/gpu/drm/i915/i915_debugfs.c | 4 ++-- drivers/gpu/drm/i915/intel_pm.c | 12 +++++++----- drivers/gpu/drm/i915/intel_sideband.c | 15 +++++++++------ drivers/gpu/drm/i915/intel_sideband.h | 3 ++- 4 files changed, 20 insertions(+), 14 deletions(-)drivers/gpu/drm/i915/i915_debugfs.c +2
-2@@ -1494,7 +1494,7 @@ static int gen6_drpc_info(struct seq_file *m)if (INTEL_GEN (dev_priv) <=7 )sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , - &rc6vids); + &rc6vids,NULL );seq_printf (m,"RC1e Enabled: %s\n" ,yesno (rcctl1 & GEN6_RC_CTL_RC1e_ENABLE));@@ -1777,7 +1777,7 @@ static int i915_ring_freq_table(struct seq_file *m, void *unused) ia_freq = gpu_freq;sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_MIN_FREQ_TABLE , - &ia_freq); + &ia_freq,NULL ); seq_printf(m, "%d\t\t%d\t\t\t\t%d\n", intel_gpu_freq(dev_priv, (gpu_freq * (IS_GEN9_BC(dev_priv) ||drivers/gpu/drm/i915/intel_pm.c +7
-5@@ -2822,7 +2822,7 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv, val =0 ;/* data0 to be programmed to 0 for first set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENCY , - &val); + &val,NULL );if (ret) {DRM_ERROR ("SKL Mailbox read error = %d\n" , ret);@@ -2841,7 +2841,7 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv, val =1 ;/* data0 to be programmed to 1 for second set */ ret =sandybridge_pcode_read (dev_priv,GEN9_PCODE_READ_MEM_LATENCY , - &val); + &val,NULL );if (ret) {DRM_ERROR ("SKL Mailbox read error = %d\n" , ret);return ;@@ -7061,7 +7061,7 @@ static void gen6_init_rps_frequencies(struct drm_i915_private *dev_priv)if (sandybridge_pcode_read (dev_priv,HSW_PCODE_DYNAMIC_DUTY_CYCLE_CONTROL , - &ddcc_status) ==0 ) + &ddcc_status,NULL ) ==0 ) rps->efficient_freq = clamp_t(u8, ((ddcc_status >> 8) & 0xff),@@ -7408,7 +7408,8 @@ static void gen6_enable_rc6(struct drm_i915_private *dev_priv) GEN6_RC_CTL_HW_ENABLE); rc6vids =0 ; - ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , &rc6vids); + ret =sandybridge_pcode_read (dev_priv,GEN6_PCODE_READ_RC6VIDS , + &rc6vids,NULL );if (IS_GEN (dev_priv,6 ) && ret) {DRM_DEBUG_DRIVER ("Couldn't check for BIOS workaround\n" ); }else if (IS_GEN (dev_priv,6 ) && (GEN6_DECODE_RC6_VID (rc6vids &0xff ) <450 )) {@@ -8555,7 +8556,8 @@ void intel_init_gt_powersave(struct drm_i915_private *dev_priv) IS_IVYBRIDGE(dev_priv) || IS_HASWELL(dev_priv)) {u32 params =0 ; -sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , ¶ms); +sandybridge_pcode_read (dev_priv,GEN6_READ_OC_PARAMS , + ¶ms,NULL );if (params &BIT (31 )) {/* OC supported */ DRM_DEBUG_DRIVER("Overclocking supported, max: %dMHz, overclock: %dMHz\n", (rps->max_freq & 0xff) * 50,drivers/gpu/drm/i915/intel_sideband.c +9
-6@@ -374,7 +374,7 @@ static inline int gen7_check_mailbox_status(u32 mbox) }static int __sandybridge_pcode_rw (struct drm_i915_private *i915, -u32 mbox,u32 *val, +u32 mbox,u32 *val,u32 *val1,int fast_timeout_us,int slow_timeout_ms,bool is_read)@@ -393,7 +393,7 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,return -EAGAIN ;intel_uncore_write_fw (uncore,GEN6_PCODE_DATA , *val); -intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 ,0 ); +intel_uncore_write_fw (uncore,GEN6_PCODE_DATA1 , val1 ? *val1 :0 );intel_uncore_write_fw (uncore,GEN6_PCODE_MAILBOX ,GEN6_PCODE_READY | mbox);@@ -407,6 +407,8 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,if (is_read) *val =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA ); +if (is_read && val1) + *val1 =intel_uncore_read_fw (uncore,GEN6_PCODE_DATA1 );if (INTEL_GEN (i915) >6 )return gen7_check_mailbox_status (mbox);@@ -414,12 +416,13 @@ static int __sandybridge_pcode_rw(struct drm_i915_private *i915,return gen6_check_mailbox_status (mbox); } -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox,u32 *val) +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1) {int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, val, + err =__sandybridge_pcode_rw (i915, mbox, val, val1,500 ,0 , true);mutex_unlock (&i915->sb_lock);@@ -440,7 +443,7 @@ int sandybridge_pcode_write_timeout(struct drm_i915_private *i915,int err;mutex_lock (&i915->sb_lock); - err =__sandybridge_pcode_rw (i915, mbox, &val, + err =__sandybridge_pcode_rw (i915, mbox, &val,NULL , fast_timeout_us, slow_timeout_ms, false);mutex_unlock (&i915->sb_lock);@@ -457,7 +460,7 @@ static bool skl_pcode_try_request(struct drm_i915_private *i915, u32 mbox, u32 request, u32 reply_mask, u32 reply, u32 *status) { - *status =__sandybridge_pcode_rw (i915, mbox, &request, + *status =__sandybridge_pcode_rw (i915, mbox, &request,NULL ,500 ,0 , true);drivers/gpu/drm/i915/intel_sideband.h +2
-1@@ -127,7 +127,8 @@ u32 intel_sbi_read(struct drm_i915_private *i915, u16 reg,void intel_sbi_write (struct drm_i915_private *i915,u16 reg,u32 value,enum intel_sbi_destination destination); -int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox,u32 *val); +int sandybridge_pcode_read (struct drm_i915_private *i915,u32 mbox, +u32 *val,u32 *val1);int sandybridge_pcode_write_timeout (struct drm_i915_private *i915,u32 mbox,u32 val,int fast_timeout_us,int slow_timeout_ms);Signature · 6 lines
2.21.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 9 lines
Matt Roper Graphics Software Engineer IoTG Platform Enabling & Development Intel Corporation (916) 356-2795 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx