diff options
| author | Linus Torvalds <torvalds@linux-foundation.org> | 2014-12-15 15:52:01 -0800 |
|---|---|---|
| committer | Linus Torvalds <torvalds@linux-foundation.org> | 2014-12-15 15:52:01 -0800 |
| commit | 988adfdffdd43cfd841df734664727993076d7cb (patch) | |
| tree | 6794f7bba8f595500c2b7d33376ad6614adcfaf2 /drivers/gpu/drm/radeon | |
| parent | 26178ec11ef3c6c814bf16a0a2b9c2f7242e3c64 (diff) | |
| parent | 4e0cd68115620bc3236ff4e58e4c073948629b41 (diff) | |
| download | op-kernel-dev-988adfdffdd43cfd841df734664727993076d7cb.zip op-kernel-dev-988adfdffdd43cfd841df734664727993076d7cb.tar.gz | |
Merge branch 'drm-next' of git://people.freedesktop.org/~airlied/linux
Pull drm updates from Dave Airlie:
"Highlights:
- AMD KFD driver merge
This is the AMD HSA interface for exposing a lowlevel interface for
GPGPU use. They have an open source userspace built on top of this
interface, and the code looks as good as it was going to get out of
tree.
- Initial atomic modesetting work
The need for an atomic modesetting interface to allow userspace to
try and send a complete set of modesetting state to the driver has
arisen, and been suffering from neglect this past year. No more,
the start of the common code and changes for msm driver to use it
are in this tree. Ongoing work to get the userspace ioctl finished
and the code clean will probably wait until next kernel.
- DisplayID 1.3 and tiled monitor exposed to userspace.
Tiled monitor property is now exposed for userspace to make use of.
- Rockchip drm driver merged.
- imx gpu driver moved out of staging
Other stuff:
- core:
panel - MIPI DSI + new panels.
expose suggested x/y properties for virtual GPUs
- i915:
Initial Skylake (SKL) support
gen3/4 reset work
start of dri1/ums removal
infoframe tracking
fixes for lots of things.
- nouveau:
tegra k1 voltage support
GM204 modesetting support
GT21x memory reclocking work
- radeon:
CI dpm fixes
GPUVM improvements
Initial DPM fan control
- rcar-du:
HDMI support added
removed some support for old boards
slave encoder driver for Analog Devices adv7511
- exynos:
Exynos4415 SoC support
- msm:
a4xx gpu support
atomic helper conversion
- tegra:
iommu support
universal plane support
ganged-mode DSI support
- sti:
HDMI i2c improvements
- vmwgfx:
some late fixes.
- qxl:
use suggested x/y properties"
* 'drm-next' of git://people.freedesktop.org/~airlied/linux: (969 commits)
drm: sti: fix module compilation issue
drm/i915: save/restore GMBUS freq across suspend/resume on gen4
drm: sti: correctly cleanup CRTC and planes
drm: sti: add HQVDP plane
drm: sti: add cursor plane
drm: sti: enable auxiliary CRTC
drm: sti: fix delay in VTG programming
drm: sti: prepare sti_tvout to support auxiliary crtc
drm: sti: use drm_crtc_vblank_{on/off} instead of drm_vblank_{on/off}
drm: sti: fix hdmi avi infoframe
drm: sti: remove event lock while disabling vblank
drm: sti: simplify gdp code
drm: sti: clear all mixer control
drm: sti: remove gpio for HDMI hot plug detection
drm: sti: allow to change hdmi ddc i2c adapter
drm/doc: Document drm_add_modes_noedid() usage
drm/i915: Remove '& 0xffff' from the mask given to WA_REG()
drm/i915: Invert the mask and val arguments in wa_add() and WA_REG()
drm: Zero out DRM object memory upon cleanup
drm/i915/bdw: Fix the write setting up the WIZ hashing mode
...
Diffstat (limited to 'drivers/gpu/drm/radeon')
58 files changed, 3236 insertions, 922 deletions
diff --git a/drivers/gpu/drm/radeon/Makefile b/drivers/gpu/drm/radeon/Makefile index d01b879..12bc212 100644 --- a/drivers/gpu/drm/radeon/Makefile +++ b/drivers/gpu/drm/radeon/Makefile @@ -80,7 +80,8 @@ radeon-y += radeon_device.o radeon_asic.o radeon_kms.o \ r600_dpm.o rs780_dpm.o rv6xx_dpm.o rv770_dpm.o rv730_dpm.o rv740_dpm.o \ rv770_smc.o cypress_dpm.o btc_dpm.o sumo_dpm.o sumo_smc.o trinity_dpm.o \ trinity_smc.o ni_dpm.o si_smc.o si_dpm.o kv_smc.o kv_dpm.o ci_smc.o \ - ci_dpm.o dce6_afmt.o radeon_vm.o radeon_ucode.o radeon_ib.o radeon_mn.o + ci_dpm.o dce6_afmt.o radeon_vm.o radeon_ucode.o radeon_ib.o radeon_mn.o \ + radeon_sync.o # add async DMA block radeon-y += \ @@ -104,6 +105,7 @@ radeon-y += \ radeon_vce.o \ vce_v1_0.o \ vce_v2_0.o \ + radeon_kfd.o radeon-$(CONFIG_COMPAT) += radeon_ioc32.o radeon-$(CONFIG_VGA_SWITCHEROO) += radeon_atpx_handler.o diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c index 30d242b..d59ec49 100644 --- a/drivers/gpu/drm/radeon/atombios_crtc.c +++ b/drivers/gpu/drm/radeon/atombios_crtc.c @@ -2039,6 +2039,7 @@ int atombios_crtc_mode_set(struct drm_crtc *crtc, atombios_crtc_set_base(crtc, x, y, old_fb); atombios_overscan_setup(crtc, mode, adjusted_mode); atombios_scaler_setup(crtc); + radeon_cursor_reset(crtc); /* update the hw version fpr dpm */ radeon_crtc->hw_mode = *adjusted_mode; diff --git a/drivers/gpu/drm/radeon/ci_dpm.c b/drivers/gpu/drm/radeon/ci_dpm.c index 11a55e9..f373a81 100644 --- a/drivers/gpu/drm/radeon/ci_dpm.c +++ b/drivers/gpu/drm/radeon/ci_dpm.c @@ -46,15 +46,15 @@ static const struct ci_pt_defaults defaults_hawaii_xt = { 1, 0xF, 0xFD, 0x19, 5, 0x14, 0, 0xB0000, - { 0x84, 0x0, 0x0, 0x7F, 0x0, 0x0, 0x5A, 0x60, 0x51, 0x8E, 0x79, 0x6B, 0x5F, 0x90, 0x79 }, - { 0x1EA, 0x1EA, 0x1EA, 0x224, 0x224, 0x224, 0x24F, 0x24F, 0x24F, 0x28E, 0x28E, 0x28E, 0x2BC, 0x2BC, 0x2BC } + { 0x2E, 0x00, 0x00, 0x88, 0x00, 0x00, 0x72, 0x60, 0x51, 0xA7, 0x79, 0x6B, 0x90, 0xBD, 0x79 }, + { 0x217, 0x217, 0x217, 0x242, 0x242, 0x242, 0x269, 0x269, 0x269, 0x2A1, 0x2A1, 0x2A1, 0x2C9, 0x2C9, 0x2C9 } }; static const struct ci_pt_defaults defaults_hawaii_pro = { 1, 0xF, 0xFD, 0x19, 5, 0x14, 0, 0x65062, - { 0x93, 0x0, 0x0, 0x97, 0x0, 0x0, 0x6B, 0x60, 0x51, 0x95, 0x79, 0x6B, 0x5F, 0x90, 0x79 }, - { 0x1EA, 0x1EA, 0x1EA, 0x224, 0x224, 0x224, 0x24F, 0x24F, 0x24F, 0x28E, 0x28E, 0x28E, 0x2BC, 0x2BC, 0x2BC } + { 0x2E, 0x00, 0x00, 0x88, 0x00, 0x00, 0x72, 0x60, 0x51, 0xA7, 0x79, 0x6B, 0x90, 0xBD, 0x79 }, + { 0x217, 0x217, 0x217, 0x242, 0x242, 0x242, 0x269, 0x269, 0x269, 0x2A1, 0x2A1, 0x2A1, 0x2C9, 0x2C9, 0x2C9 } }; static const struct ci_pt_defaults defaults_bonaire_xt = @@ -184,6 +184,9 @@ static int ci_set_overdrive_target_tdp(struct radeon_device *rdev, u32 target_tdp); static int ci_update_uvd_dpm(struct radeon_device *rdev, bool gate); +static PPSMC_Result ci_send_msg_to_smc_with_parameter(struct radeon_device *rdev, + PPSMC_Msg msg, u32 parameter); + static struct ci_power_info *ci_get_pi(struct radeon_device *rdev) { struct ci_power_info *pi = rdev->pm.dpm.priv; @@ -249,7 +252,10 @@ static void ci_initialize_powertune_defaults(struct radeon_device *rdev) if (pi->caps_power_containment) { pi->caps_cac = true; - pi->enable_bapm_feature = true; + if (rdev->family == CHIP_HAWAII) + pi->enable_bapm_feature = false; + else + pi->enable_bapm_feature = true; pi->enable_tdc_limit_feature = true; pi->enable_pkg_pwr_tracking_feature = true; } @@ -352,6 +358,21 @@ static int ci_populate_dw8(struct radeon_device *rdev) return 0; } +static int ci_populate_fuzzy_fan(struct radeon_device *rdev) +{ + struct ci_power_info *pi = ci_get_pi(rdev); + + if ((rdev->pm.dpm.fan.fan_output_sensitivity & (1 << 15)) || + (rdev->pm.dpm.fan.fan_output_sensitivity == 0)) + rdev->pm.dpm.fan.fan_output_sensitivity = + rdev->pm.dpm.fan.default_fan_output_sensitivity; + + pi->smc_powertune_table.FuzzyFan_PwmSetDelta = + cpu_to_be16(rdev->pm.dpm.fan.fan_output_sensitivity); + + return 0; +} + static int ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(struct radeon_device *rdev) { struct ci_power_info *pi = ci_get_pi(rdev); @@ -477,6 +498,9 @@ static int ci_populate_pm_base(struct radeon_device *rdev) ret = ci_populate_dw8(rdev); if (ret) return ret; + ret = ci_populate_fuzzy_fan(rdev); + if (ret) + return ret; ret = ci_min_max_v_gnbl_pm_lid_from_bapm_vddc(rdev); if (ret) return ret; @@ -690,6 +714,25 @@ static int ci_enable_smc_cac(struct radeon_device *rdev, bool enable) return ret; } +static int ci_enable_thermal_based_sclk_dpm(struct radeon_device *rdev, + bool enable) +{ + struct ci_power_info *pi = ci_get_pi(rdev); + PPSMC_Result smc_result = PPSMC_Result_OK; + + if (pi->thermal_sclk_dpm_enabled) { + if (enable) + smc_result = ci_send_msg_to_smc(rdev, PPSMC_MSG_ENABLE_THERMAL_DPM); + else + smc_result = ci_send_msg_to_smc(rdev, PPSMC_MSG_DISABLE_THERMAL_DPM); + } + + if (smc_result == PPSMC_Result_OK) + return 0; + else + return -EINVAL; +} + static int ci_power_control_set_level(struct radeon_device *rdev) { struct ci_power_info *pi = ci_get_pi(rdev); @@ -700,13 +743,11 @@ static int ci_power_control_set_level(struct radeon_device *rdev) int ret = 0; bool adjust_polarity = false; /* ??? */ - if (pi->caps_power_containment && - (pi->power_containment_features & POWERCONTAINMENT_FEATURE_BAPM)) { + if (pi->caps_power_containment) { adjust_percent = adjust_polarity ? rdev->pm.dpm.tdp_adjustment : (-1 * rdev->pm.dpm.tdp_adjustment); target_tdp = ((100 + adjust_percent) * (s32)cac_tdp_table->configurable_tdp) / 100; - target_tdp *= 256; ret = ci_set_overdrive_target_tdp(rdev, (u32)target_tdp); } @@ -814,7 +855,7 @@ static void ci_apply_state_adjust_rules(struct radeon_device *rdev, } } -static int ci_set_thermal_temperature_range(struct radeon_device *rdev, +static int ci_thermal_set_temperature_range(struct radeon_device *rdev, int min_temp, int max_temp) { int low_temp = 0 * 1000; @@ -850,6 +891,350 @@ static int ci_set_thermal_temperature_range(struct radeon_device *rdev, return 0; } +static int ci_thermal_enable_alert(struct radeon_device *rdev, + bool enable) +{ + u32 thermal_int = RREG32_SMC(CG_THERMAL_INT); + PPSMC_Result result; + + if (enable) { + thermal_int &= ~(THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW); + WREG32_SMC(CG_THERMAL_INT, thermal_int); + rdev->irq.dpm_thermal = false; + result = ci_send_msg_to_smc(rdev, PPSMC_MSG_Thermal_Cntl_Enable); + if (result != PPSMC_Result_OK) { + DRM_DEBUG_KMS("Could not enable thermal interrupts.\n"); + return -EINVAL; + } + } else { + thermal_int |= THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW; + WREG32_SMC(CG_THERMAL_INT, thermal_int); + rdev->irq.dpm_thermal = true; + result = ci_send_msg_to_smc(rdev, PPSMC_MSG_Thermal_Cntl_Disable); + if (result != PPSMC_Result_OK) { + DRM_DEBUG_KMS("Could not disable thermal interrupts.\n"); + return -EINVAL; + } + } + + return 0; +} + +static void ci_fan_ctrl_set_static_mode(struct radeon_device *rdev, u32 mode) +{ + struct ci_power_info *pi = ci_get_pi(rdev); + u32 tmp; + + if (pi->fan_ctrl_is_in_default_mode) { + tmp = (RREG32_SMC(CG_FDO_CTRL2) & FDO_PWM_MODE_MASK) >> FDO_PWM_MODE_SHIFT; + pi->fan_ctrl_default_mode = tmp; + tmp = (RREG32_SMC(CG_FDO_CTRL2) & TMIN_MASK) >> TMIN_SHIFT; + pi->t_min = tmp; + pi->fan_ctrl_is_in_default_mode = false; + } + + tmp = RREG32_SMC(CG_FDO_CTRL2) & ~TMIN_MASK; + tmp |= TMIN(0); + WREG32_SMC(CG_FDO_CTRL2, tmp); + + tmp = RREG32_SMC(CG_FDO_CTRL2) & ~FDO_PWM_MODE_MASK; + tmp |= FDO_PWM_MODE(mode); + WREG32_SMC(CG_FDO_CTRL2, tmp); +} + +static int ci_thermal_setup_fan_table(struct radeon_device *rdev) +{ + struct ci_power_info *pi = ci_get_pi(rdev); + SMU7_Discrete_FanTable fan_table = { FDO_MODE_HARDWARE }; + u32 duty100; + u32 t_diff1, t_diff2, pwm_diff1, pwm_diff2; + u16 fdo_min, slope1, slope2; + u32 reference_clock, tmp; + int ret; + u64 tmp64; + + if (!pi->fan_table_start) { + rdev->pm.dpm.fan.ucode_fan_control = false; + return 0; + } + + duty100 = (RREG32_SMC(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT; + + if (duty100 == 0) { + rdev->pm.dpm.fan.ucode_fan_control = false; + return 0; + } + + tmp64 = (u64)rdev->pm.dpm.fan.pwm_min * duty100; + do_div(tmp64, 10000); + fdo_min = (u16)tmp64; + + t_diff1 = rdev->pm.dpm.fan.t_med - rdev->pm.dpm.fan.t_min; + t_diff2 = rdev->pm.dpm.fan.t_high - rdev->pm.dpm.fan.t_med; + + pwm_diff1 = rdev->pm.dpm.fan.pwm_med - rdev->pm.dpm.fan.pwm_min; + pwm_diff2 = rdev->pm.dpm.fan.pwm_high - rdev->pm.dpm.fan.pwm_med; + + slope1 = (u16)((50 + ((16 * duty100 * pwm_diff1) / t_diff1)) / 100); + slope2 = (u16)((50 + ((16 * duty100 * pwm_diff2) / t_diff2)) / 100); + + fan_table.TempMin = cpu_to_be16((50 + rdev->pm.dpm.fan.t_min) / 100); + fan_table.TempMed = cpu_to_be16((50 + rdev->pm.dpm.fan.t_med) / 100); + fan_table.TempMax = cpu_to_be16((50 + rdev->pm.dpm.fan.t_max) / 100); + + fan_table.Slope1 = cpu_to_be16(slope1); + fan_table.Slope2 = cpu_to_be16(slope2); + + fan_table.FdoMin = cpu_to_be16(fdo_min); + + fan_table.HystDown = cpu_to_be16(rdev->pm.dpm.fan.t_hyst); + + fan_table.HystUp = cpu_to_be16(1); + + fan_table.HystSlope = cpu_to_be16(1); + + fan_table.TempRespLim = cpu_to_be16(5); + + reference_clock = radeon_get_xclk(rdev); + + fan_table.RefreshPeriod = cpu_to_be32((rdev->pm.dpm.fan.cycle_delay * + reference_clock) / 1600); + + fan_table.FdoMax = cpu_to_be16((u16)duty100); + + tmp = (RREG32_SMC(CG_MULT_THERMAL_CTRL) & TEMP_SEL_MASK) >> TEMP_SEL_SHIFT; + fan_table.TempSrc = (uint8_t)tmp; + + ret = ci_copy_bytes_to_smc(rdev, + pi->fan_table_start, + (u8 *)(&fan_table), + sizeof(fan_table), + pi->sram_end); + + if (ret) { + DRM_ERROR("Failed to load fan table to the SMC."); + rdev->pm.dpm.fan.ucode_fan_control = false; + } + + return 0; +} + +static int ci_fan_ctrl_start_smc_fan_control(struct radeon_device *rdev) +{ + struct ci_power_info *pi = ci_get_pi(rdev); + PPSMC_Result ret; + + if (pi->caps_od_fuzzy_fan_control_support) { + ret = ci_send_msg_to_smc_with_parameter(rdev, + PPSMC_StartFanControl, + FAN_CONTROL_FUZZY); + if (ret != PPSMC_Result_OK) + return -EINVAL; + ret = ci_send_msg_to_smc_with_parameter(rdev, + PPSMC_MSG_SetFanPwmMax, + rdev->pm.dpm.fan.default_max_fan_pwm); + if (ret != PPSMC_Result_OK) + return -EINVAL; + } else { + ret = ci_send_msg_to_smc_with_parameter(rdev, + PPSMC_StartFanControl, + FAN_CONTROL_TABLE); + if (ret != PPSMC_Result_OK) + return -EINVAL; + } + + return 0; +} + +#if 0 +static int ci_fan_ctrl_stop_smc_fan_control(struct radeon_device *rdev) +{ + PPSMC_Result ret; + + ret = ci_send_msg_to_smc(rdev, PPSMC_StopFanControl); + if (ret == PPSMC_Result_OK) + return 0; + else + return -EINVAL; +} + +static int ci_fan_ctrl_get_fan_speed_percent(struct radeon_device *rdev, + u32 *speed) +{ + u32 duty, duty100; + u64 tmp64; + + if (rdev->pm.no_fan) + return -ENOENT; + + duty100 = (RREG32_SMC(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT; + duty = (RREG32_SMC(CG_THERMAL_STATUS) & FDO_PWM_DUTY_MASK) >> FDO_PWM_DUTY_SHIFT; + + if (duty100 == 0) + return -EINVAL; + + tmp64 = (u64)duty * 100; + do_div(tmp64, duty100); + *speed = (u32)tmp64; + + if (*speed > 100) + *speed = 100; + + return 0; +} + +static int ci_fan_ctrl_set_fan_speed_percent(struct radeon_device *rdev, + u32 speed) +{ + u32 tmp; + u32 duty, duty100; + u64 tmp64; + + if (rdev->pm.no_fan) + return -ENOENT; + + if (speed > 100) + return -EINVAL; + + if (rdev->pm.dpm.fan.ucode_fan_control) + ci_fan_ctrl_stop_smc_fan_control(rdev); + + duty100 = (RREG32_SMC(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT; + + if (duty100 == 0) + return -EINVAL; + + tmp64 = (u64)speed * duty100; + do_div(tmp64, 100); + duty = (u32)tmp64; + + tmp = RREG32_SMC(CG_FDO_CTRL0) & ~FDO_STATIC_DUTY_MASK; + tmp |= FDO_STATIC_DUTY(duty); + WREG32_SMC(CG_FDO_CTRL0, tmp); + + ci_fan_ctrl_set_static_mode(rdev, FDO_PWM_MODE_STATIC); + + return 0; +} + +static int ci_fan_ctrl_get_fan_speed_rpm(struct radeon_device *rdev, + u32 *speed) +{ + u32 tach_period; + u32 xclk = radeon_get_xclk(rdev); + + if (rdev->pm.no_fan) + return -ENOENT; + + if (rdev->pm.fan_pulses_per_revolution == 0) + return -ENOENT; + + tach_period = (RREG32_SMC(CG_TACH_STATUS) & TACH_PERIOD_MASK) >> TACH_PERIOD_SHIFT; + if (tach_period == 0) + return -ENOENT; + + *speed = 60 * xclk * 10000 / tach_period; + + return 0; +} + +static int ci_fan_ctrl_set_fan_speed_rpm(struct radeon_device *rdev, + u32 speed) +{ + u32 tach_period, tmp; + u32 xclk = radeon_get_xclk(rdev); + + if (rdev->pm.no_fan) + return -ENOENT; + + if (rdev->pm.fan_pulses_per_revolution == 0) + return -ENOENT; + + if ((speed < rdev->pm.fan_min_rpm) || + (speed > rdev->pm.fan_max_rpm)) + return -EINVAL; + + if (rdev->pm.dpm.fan.ucode_fan_control) + ci_fan_ctrl_stop_smc_fan_control(rdev); + + tach_period = 60 * xclk * 10000 / (8 * speed); + tmp = RREG32_SMC(CG_TACH_CTRL) & ~TARGET_PERIOD_MASK; + tmp |= TARGET_PERIOD(tach_period); + WREG32_SMC(CG_TACH_CTRL, tmp); + + ci_fan_ctrl_set_static_mode(rdev, FDO_PWM_MODE_STATIC_RPM); + + return 0; +} +#endif + +static void ci_fan_ctrl_set_default_mode(struct radeon_device *rdev) +{ + struct ci_power_info *pi = ci_get_pi(rdev); + u32 tmp; + + if (!pi->fan_ctrl_is_in_default_mode) { + tmp = RREG32_SMC(CG_FDO_CTRL2) & ~FDO_PWM_MODE_MASK; + tmp |= FDO_PWM_MODE(pi->fan_ctrl_default_mode); + WREG32_SMC(CG_FDO_CTRL2, tmp); + + tmp = RREG32_SMC(CG_FDO_CTRL2) & ~TMIN_MASK; + tmp |= TMIN(pi->t_min); + WREG32_SMC(CG_FDO_CTRL2, tmp); + pi->fan_ctrl_is_in_default_mode = true; + } +} + +static void ci_thermal_start_smc_fan_control(struct radeon_device *rdev) +{ + if (rdev->pm.dpm.fan.ucode_fan_control) { + ci_fan_ctrl_start_smc_fan_control(rdev); + ci_fan_ctrl_set_static_mode(rdev, FDO_PWM_MODE_STATIC); + } +} + +static void ci_thermal_initialize(struct radeon_device *rdev) +{ + u32 tmp; + + if (rdev->pm.fan_pulses_per_revolution) { + tmp = RREG32_SMC(CG_TACH_CTRL) & ~EDGE_PER_REV_MASK; + tmp |= EDGE_PER_REV(rdev->pm.fan_pulses_per_revolution -1); + WREG32_SMC(CG_TACH_CTRL, tmp); + } + + tmp = RREG32_SMC(CG_FDO_CTRL2) & ~TACH_PWM_RESP_RATE_MASK; + tmp |= TACH_PWM_RESP_RATE(0x28); + WREG32_SMC(CG_FDO_CTRL2, tmp); +} + +static int ci_thermal_start_thermal_controller(struct radeon_device *rdev) +{ + int ret; + + ci_thermal_initialize(rdev); + ret = ci_thermal_set_temperature_range(rdev, R600_TEMP_RANGE_MIN, R600_TEMP_RANGE_MAX); + if (ret) + return ret; + ret = ci_thermal_enable_alert(rdev, true); + if (ret) + return ret; + if (rdev->pm.dpm.fan.ucode_fan_control) { + ret = ci_thermal_setup_fan_table(rdev); + if (ret) + return ret; + ci_thermal_start_smc_fan_control(rdev); + } + + return 0; +} + +static void ci_thermal_stop_thermal_controller(struct radeon_device *rdev) +{ + if (!rdev->pm.no_fan) + ci_fan_ctrl_set_default_mode(rdev); +} + #if 0 static int ci_read_smc_soft_register(struct radeon_device *rdev, u16 reg_offset, u32 *value) @@ -1253,7 +1638,7 @@ static int ci_dpm_force_state_sclk(struct radeon_device *rdev, u32 n) if (!pi->sclk_dpm_key_disabled) { PPSMC_Result smc_result = - ci_send_msg_to_smc_with_parameter(rdev, PPSMC_MSG_DPM_ForceState, n); + ci_send_msg_to_smc_with_parameter(rdev, PPSMC_MSG_SCLKDPM_SetEnabledMask, 1 << n); if (smc_result != PPSMC_Result_OK) return -EINVAL; } @@ -1267,7 +1652,7 @@ static int ci_dpm_force_state_mclk(struct radeon_device *rdev, u32 n) if (!pi->mclk_dpm_key_disabled) { PPSMC_Result smc_result = - ci_send_msg_to_smc_with_parameter(rdev, PPSMC_MSG_MCLKDPM_ForceState, n); + ci_send_msg_to_smc_with_parameter(rdev, PPSMC_MSG_MCLKDPM_SetEnabledMask, 1 << n); if (smc_result != PPSMC_Result_OK) return -EINVAL; } @@ -2042,6 +2427,33 @@ static int ci_force_switch_to_arb_f0(struct radeon_device *rdev) return ni_copy_and_switch_arb_sets(rdev, tmp, MC_CG_ARB_FREQ_F0); } +static void ci_register_patching_mc_arb(struct radeon_device *rdev, + const u32 engine_clock, + const u32 memory_clock, + u32 *dram_timimg2) +{ + bool patch; + u32 tmp, tmp2; + + tmp = RREG32(MC_SEQ_MISC0); + patch = ((tmp & 0x0000f00) == 0x300) ? true : false; + + if (patch && + ((rdev->pdev->device == 0x67B0) || + (rdev->pdev->device == 0x67B1))) { + if ((memory_clock > 100000) && (memory_clock <= 125000)) { + tmp2 = (((0x31 * engine_clock) / 125000) - 1) & 0xff; + *dram_timimg2 &= ~0x00ff0000; + *dram_timimg2 |= tmp2 << 16; + } else if ((memory_clock > 125000) && (memory_clock <= 137500)) { + tmp2 = (((0x36 * engine_clock) / 137500) - 1) & 0xff; + *dram_timimg2 &= ~0x00ff0000; + *dram_timimg2 |= tmp2 << 16; + } + } +} + + static int ci_populate_memory_timing_parameters(struct radeon_device *rdev, u32 sclk, u32 mclk, @@ -2057,6 +2469,8 @@ static int ci_populate_memory_timing_parameters(struct radeon_device *rdev, dram_timing2 = RREG32(MC_ARB_DRAM_TIMING2); burst_time = RREG32(MC_ARB_BURST_TIME) & STATE0_MASK; + ci_register_patching_mc_arb(rdev, sclk, mclk, &dram_timing2); + arb_regs->McArbDramTiming = cpu_to_be32(dram_timing); arb_regs->McArbDramTiming2 = cpu_to_be32(dram_timing2); arb_regs->McArbBurstTime = (u8)burst_time; @@ -2351,10 +2765,10 @@ static int ci_calculate_mclk_params(struct radeon_device *rdev, u32 tmp; u32 reference_clock = rdev->clock.mpll.reference_freq; - if (pi->mem_gddr5) - freq_nom = memory_clock * 4; + if (mpll_param.qdr == 1) + freq_nom = memory_clock * 4 * (1 << mpll_param.post_div); else - freq_nom = memory_clock * 2; + freq_nom = memory_clock * 2 * (1 << mpll_param.post_div); tmp = (freq_nom / reference_clock); tmp = tmp * tmp; @@ -2434,7 +2848,6 @@ static int ci_populate_single_memory_level(struct radeon_device *rdev, &memory_level->MinVddcPhases); memory_level->EnabledForThrottle = 1; - memory_level->EnabledForActivity = 1; memory_level->UpH = 0; memory_level->DownH = 100; memory_level->VoltageDownH = 0; @@ -2767,7 +3180,6 @@ static int ci_populate_single_graphic_level(struct radeon_device *rdev, graphic_level->CcPwrDynRm = 0; graphic_level->CcPwrDynRm1 = 0; - graphic_level->EnabledForActivity = 1; graphic_level->EnabledForThrottle = 1; graphic_level->UpH = 0; graphic_level->DownH = 0; @@ -2816,10 +3228,13 @@ static int ci_populate_all_graphic_levels(struct radeon_device *rdev) &pi->smc_state_table.GraphicsLevel[i]); if (ret) return ret; + if (i > 1) + pi->smc_state_table.GraphicsLevel[i].DeepSleepDivId = 0; if (i == (dpm_table->sclk_table.count - 1)) pi->smc_state_table.GraphicsLevel[i].DisplayWatermark = PPSMC_DISPLAY_WATERMARK_HIGH; } + pi->smc_state_table.GraphicsLevel[0].EnabledForActivity = 1; pi->smc_state_table.GraphicsDpmLevelCount = (u8)dpm_table->sclk_table.count; pi->dpm_level_enable_mask.sclk_dpm_enable_mask = @@ -2863,6 +3278,16 @@ static int ci_populate_all_memory_levels(struct radeon_device *rdev) return ret; } + pi->smc_state_table.MemoryLevel[0].EnabledForActivity = 1; + + if ((dpm_table->mclk_table.count >= 2) && + ((rdev->pdev->device == 0x67B0) || (rdev->pdev->device == 0x67B1))) { + pi->smc_state_table.MemoryLevel[1].MinVddc = + pi->smc_state_table.MemoryLevel[0].MinVddc; + pi->smc_state_table.MemoryLevel[1].MinVddcPhases = + pi->smc_state_table.MemoryLevel[0].MinVddcPhases; + } + pi->smc_state_table.MemoryLevel[0].ActivityLevel = cpu_to_be16(0x1F); pi->smc_state_table.MemoryDpmLevelCount = (u8)dpm_table->mclk_table.count; @@ -2919,9 +3344,14 @@ static int ci_setup_default_pcie_tables(struct radeon_device *rdev) &pi->dpm_table.pcie_speed_table, SMU7_MAX_LEVELS_LINK); - ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0, - pi->pcie_gen_powersaving.min, - pi->pcie_lane_powersaving.min); + if (rdev->family == CHIP_BONAIRE) + ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0, + pi->pcie_gen_powersaving.min, + pi->pcie_lane_powersaving.max); + else + ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 0, + pi->pcie_gen_powersaving.min, + pi->pcie_lane_powersaving.min); ci_setup_pcie_table_entry(&pi->dpm_table.pcie_speed_table, 1, pi->pcie_gen_performance.min, pi->pcie_lane_performance.min); @@ -2988,19 +3418,21 @@ static int ci_setup_default_dpm_tables(struct radeon_device *rdev) allowed_sclk_vddc_table->entries[i].clk)) { pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].value = allowed_sclk_vddc_table->entries[i].clk; - pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].enabled = true; + pi->dpm_table.sclk_table.dpm_levels[pi->dpm_table.sclk_table.count].enabled = + (i == 0) ? true : false; pi->dpm_table.sclk_table.count++; } } pi->dpm_table.mclk_table.count = 0; for (i = 0; i < allowed_mclk_table->count; i++) { - if ((i==0) || + if ((i == 0) || (pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count-1].value != allowed_mclk_table->entries[i].clk)) { pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].value = allowed_mclk_table->entries[i].clk; - pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].enabled = true; + pi->dpm_table.mclk_table.dpm_levels[pi->dpm_table.mclk_table.count].enabled = + (i == 0) ? true : false; pi->dpm_table.mclk_table.count++; } } @@ -3166,7 +3598,7 @@ static int ci_init_smc_table(struct radeon_device *rdev) table->VddcVddciDelta = 4000; table->PhaseResponseTime = 0; table->MemoryThermThrottleEnable = 1; - table->PCIeBootLinkLevel = 0; + table->PCIeBootLinkLevel = pi->dpm_table.pcie_speed_table.count - 1; table->PCIeGenInterval = 1; if (pi->voltage_control == CISLANDS_VOLTAGE_CONTROL_BY_SVID2) table->SVI2Enable = 1; @@ -3320,6 +3752,8 @@ static int ci_upload_dpm_level_enable_mask(struct radeon_device *rdev) struct ci_power_info *pi = ci_get_pi(rdev); PPSMC_Result result; + ci_apply_disp_minimum_voltage_request(rdev); + if (!pi->sclk_dpm_key_disabled) { if (pi->dpm_level_enable_mask.sclk_dpm_enable_mask) { result = ci_send_msg_to_smc_with_parameter(rdev, @@ -3339,7 +3773,7 @@ static int ci_upload_dpm_level_enable_mask(struct radeon_device *rdev) return -EINVAL; } } - +#if 0 if (!pi->pcie_dpm_key_disabled) { if (pi->dpm_level_enable_mask.pcie_dpm_enable_mask) { result = ci_send_msg_to_smc_with_parameter(rdev, @@ -3349,9 +3783,7 @@ static int ci_upload_dpm_level_enable_mask(struct radeon_device *rdev) return -EINVAL; } } - - ci_apply_disp_minimum_voltage_request(rdev); - +#endif return 0; } @@ -3377,7 +3809,7 @@ static void ci_find_dpm_states_clocks_in_dpm_table(struct radeon_device *rdev, pi->need_update_smu7_dpm_table |= DPMTABLE_OD_UPDATE_SCLK; } else { /* XXX check display min clock requirements */ - if (0 != CISLAND_MINIMUM_ENGINE_CLOCK) + if (CISLAND_MINIMUM_ENGINE_CLOCK != CISLAND_MINIMUM_ENGINE_CLOCK) pi->need_update_smu7_dpm_table |= DPMTABLE_UPDATE_SCLK; } @@ -3707,62 +4139,61 @@ int ci_dpm_force_performance_level(struct radeon_device *rdev, enum radeon_dpm_forced_level level) { struct ci_power_info *pi = ci_get_pi(rdev); - PPSMC_Result smc_result; u32 tmp, levels, i; int ret; if (level == RADEON_DPM_FORCED_LEVEL_HIGH) { - if ((!pi->sclk_dpm_key_disabled) && - pi->dpm_level_enable_mask.sclk_dpm_enable_mask) { + if ((!pi->pcie_dpm_key_disabled) && + pi->dpm_level_enable_mask.pcie_dpm_enable_mask) { levels = 0; - tmp = pi->dpm_level_enable_mask.sclk_dpm_enable_mask; + tmp = pi->dpm_level_enable_mask.pcie_dpm_enable_mask; while (tmp >>= 1) levels++; if (levels) { - ret = ci_dpm_force_state_sclk(rdev, levels); + ret = ci_dpm_force_state_pcie(rdev, level); if (ret) return ret; for (i = 0; i < rdev->usec_timeout; i++) { - tmp = (RREG32_SMC(TARGET_AND_CURRENT_PROFILE_INDEX) & - CURR_SCLK_INDEX_MASK) >> CURR_SCLK_INDEX_SHIFT; + tmp = (RREG32_SMC(TARGET_AND_CURRENT_PROFILE_INDEX_1) & + CURR_PCIE_INDEX_MASK) >> CURR_PCIE_INDEX_SHIFT; if (tmp == levels) break; udelay(1); } } } - if ((!pi->mclk_dpm_key_disabled) && - pi->dpm_level_enable_mask.mclk_dpm_enable_mask) { + if ((!pi->sclk_dpm_key_disabled) && + pi->dpm_level_enable_mask.sclk_dpm_enable_mask) { levels = 0; - tmp = pi->dpm_level_enable_mask.mclk_dpm_enable_mask; + tmp = pi->dpm_level_enable_mask.sclk_dpm_enable_mask; while (tmp >>= 1) levels++; if (levels) { - ret = ci_dpm_force_state_mclk(rdev, levels); + ret = ci_dpm_force_state_sclk(rdev, levels); if (ret) return ret; for (i = 0; i < rdev->usec_timeout; i++) { tmp = (RREG32_SMC(TARGET_AND_CURRENT_PROFILE_INDEX) & - CURR_MCLK_INDEX_MASK) >> CURR_MCLK_INDEX_SHIFT; + CURR_SCLK_INDEX_MASK) >> CURR_SCLK_INDEX_SHIFT; if (tmp == levels) break; udelay(1); } } } - if ((!pi->pcie_dpm_key_disabled) && - pi->dpm_level_enable_mask.pcie_dpm_enable_mask) { + if ((!pi->mclk_dpm_key_disabled) && + pi->dpm_level_enable_mask.mclk_dpm_enable_mask) { levels = 0; - tmp = pi->dpm_level_enable_mask.pcie_dpm_enable_mask; + tmp = pi->dpm_level_enable_mask.mclk_dpm_enable_mask; while (tmp >>= 1) levels++; if (levels) { - ret = ci_dpm_force_state_pcie(rdev, level); + ret = ci_dpm_force_state_mclk(rdev, levels); if (ret) return ret; for (i = 0; i < rdev->usec_timeout; i++) { - tmp = (RREG32_SMC(TARGET_AND_CURRENT_PROFILE_INDEX_1) & - CURR_PCIE_INDEX_MASK) >> CURR_PCIE_INDEX_SHIFT; + tmp = (RREG32_SMC(TARGET_AND_CURRENT_PROFILE_INDEX) & + CURR_MCLK_INDEX_MASK) >> CURR_MCLK_INDEX_SHIFT; if (tmp == levels) break; udelay(1); @@ -3816,21 +4247,17 @@ int ci_dpm_force_performance_level(struct radeon_device *rdev, } } } else if (level == RADEON_DPM_FORCED_LEVEL_AUTO) { - if (!pi->sclk_dpm_key_disabled) { - smc_result = ci_send_msg_to_smc(rdev, PPSMC_MSG_NoForcedLevel); - if (smc_result != PPSMC_Result_OK) - return -EINVAL; - } - if (!pi->mclk_dpm_key_disabled) { - smc_result = ci_send_msg_to_smc(rdev, PPSMC_MSG_MCLKDPM_NoForcedLevel); - if (smc_result != PPSMC_Result_OK) - return -EINVAL; - } if (!pi->pcie_dpm_key_disabled) { - smc_result = ci_send_msg_to_smc(rdev, PPSMC_MSG_PCIeDPM_UnForceLevel); + PPSMC_Result smc_result; + + smc_result = ci_send_msg_to_smc(rdev, + PPSMC_MSG_PCIeDPM_UnForceLevel); if (smc_result != PPSMC_Result_OK) return -EINVAL; } + ret = ci_upload_dpm_level_enable_mask(rdev); + if (ret) + return ret; } rdev->pm.dpm.forced_level = level; @@ -4036,6 +4463,96 @@ static int ci_copy_vbios_mc_reg_table(const struct atom_mc_reg_table *table, return 0; } +static int ci_register_patching_mc_seq(struct radeon_device *rdev, + struct ci_mc_reg_table *table) +{ + u8 i, k; + u32 tmp; + bool patch; + + tmp = RREG32(MC_SEQ_MISC0); + patch = ((tmp & 0x0000f00) == 0x300) ? true : false; + + if (patch && + ((rdev->pdev->device == 0x67B0) || + (rdev->pdev->device == 0x67B1))) { + for (i = 0; i < table->last; i++) { + if (table->last >= SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE) + return -EINVAL; + switch(table->mc_reg_address[i].s1 >> 2) { + case MC_SEQ_MISC1: + for (k = 0; k < table->num_entries; k++) { + if ((table->mc_reg_table_entry[k].mclk_max == 125000) || + (table->mc_reg_table_entry[k].mclk_max == 137500)) + table->mc_reg_table_entry[k].mc_data[i] = + (table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFF8) | + 0x00000007; + } + break; + case MC_SEQ_WR_CTL_D0: + for (k = 0; k < table->num_entries; k++) { + if ((table->mc_reg_table_entry[k].mclk_max == 125000) || + (table->mc_reg_table_entry[k].mclk_max == 137500)) + table->mc_reg_table_entry[k].mc_data[i] = + (table->mc_reg_table_entry[k].mc_data[i] & 0xFFFF0F00) | + 0x0000D0DD; + } + break; + case MC_SEQ_WR_CTL_D1: + for (k = 0; k < table->num_entries; k++) { + if ((table->mc_reg_table_entry[k].mclk_max == 125000) || + (table->mc_reg_table_entry[k].mclk_max == 137500)) + table->mc_reg_table_entry[k].mc_data[i] = + (table->mc_reg_table_entry[k].mc_data[i] & 0xFFFF0F00) | + 0x0000D0DD; + } + break; + case MC_SEQ_WR_CTL_2: + for (k = 0; k < table->num_entries; k++) { + if ((table->mc_reg_table_entry[k].mclk_max == 125000) || + (table->mc_reg_table_entry[k].mclk_max == 137500)) + table->mc_reg_table_entry[k].mc_data[i] = 0; + } + break; + case MC_SEQ_CAS_TIMING: + for (k = 0; k < table->num_entries; k++) { + if (table->mc_reg_table_entry[k].mclk_max == 125000) + table->mc_reg_table_entry[k].mc_data[i] = + (table->mc_reg_table_entry[k].mc_data[i] & 0xFFE0FE0F) | + 0x000C0140; + else if (table->mc_reg_table_entry[k].mclk_max == 137500) + table->mc_reg_table_entry[k].mc_data[i] = + (table->mc_reg_table_entry[k].mc_data[i] & 0xFFE0FE0F) | + 0x000C0150; + } + break; + case MC_SEQ_MISC_TIMING: + for (k = 0; k < table->num_entries; k++) { + if (table->mc_reg_table_entry[k].mclk_max == 125000) + table->mc_reg_table_entry[k].mc_data[i] = + (table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFE0) | + 0x00000030; + else if (table->mc_reg_table_entry[k].mclk_max == 137500) + table->mc_reg_table_entry[k].mc_data[i] = + (table->mc_reg_table_entry[k].mc_data[i] & 0xFFFFFFE0) | + 0x00000035; + } + break; + default: + break; + } + } + + WREG32(MC_SEQ_IO_DEBUG_INDEX, 3); + tmp = RREG32(MC_SEQ_IO_DEBUG_DATA); + tmp = (tmp & 0xFFF8FFFF) | (1 << 16); + WREG32(MC_SEQ_IO_DEBUG_INDEX, 3); + WREG32(MC_SEQ_IO_DEBUG_DATA, tmp); + } + + return 0; +} + static int ci_initialize_mc_reg_table(struct radeon_device *rdev) { struct ci_power_info *pi = ci_get_pi(rdev); @@ -4079,6 +4596,10 @@ static int ci_initialize_mc_reg_table(struct radeon_device *rdev) ci_set_s0_mc_reg_index(ci_table); + ret = ci_register_patching_mc_seq(rdev, ci_table); + if (ret) + goto init_mc_done; + ret = ci_set_mc_special_registers(rdev, ci_table); if (ret) goto init_mc_done; @@ -4675,36 +5196,51 @@ int ci_dpm_enable(struct radeon_device *rdev) return ret; } + ret = ci_power_control_set_level(rdev); + if (ret) { + DRM_ERROR("ci_power_control_set_level failed\n"); + return ret; + } + ci_enable_auto_throttle_source(rdev, RADEON_DPM_AUTO_THROTTLE_SRC_THERMAL, true); + ret = ci_enable_thermal_based_sclk_dpm(rdev, true); + if (ret) { + DRM_ERROR("ci_enable_thermal_based_sclk_dpm failed\n"); + return ret; + } + + ci_thermal_start_thermal_controller(rdev); + ci_update_current_ps(rdev, boot_ps); return 0; } -int ci_dpm_late_enable(struct radeon_device *rdev) +static int ci_set_temperature_range(struct radeon_device *rdev) { int ret; - if (rdev->irq.installed && - r600_is_internal_thermal_sensor(rdev->pm.int_thermal_type)) { -#if 0 - PPSMC_Result result; -#endif - ret = ci_set_thermal_temperature_range(rdev, R600_TEMP_RANGE_MIN, R600_TEMP_RANGE_MAX); - if (ret) { - DRM_ERROR("ci_set_thermal_temperature_range failed\n"); - return ret; - } - rdev->irq.dpm_thermal = true; - radeon_irq_set(rdev); -#if 0 - result = ci_send_msg_to_smc(rdev, PPSMC_MSG_EnableThermalInterrupt); + ret = ci_thermal_enable_alert(rdev, false); + if (ret) + return ret; + ret = ci_thermal_set_temperature_range(rdev, R600_TEMP_RANGE_MIN, R600_TEMP_RANGE_MAX); + if (ret) + return ret; + ret = ci_thermal_enable_alert(rdev, true); + if (ret) + return ret; - if (result != PPSMC_Result_OK) - DRM_DEBUG_KMS("Could not enable thermal interrupts.\n"); -#endif - } + return ret; +} + +int ci_dpm_late_enable(struct radeon_device *rdev) +{ + int ret; + + ret = ci_set_temperature_range(rdev); + if (ret) + return ret; ci_dpm_powergate_uvd(rdev, true); @@ -4721,6 +5257,8 @@ void ci_dpm_disable(struct radeon_device *rdev) if (!ci_is_smc_running(rdev)) return; + ci_thermal_stop_thermal_controller(rdev); + if (pi->thermal_protection) ci_enable_thermal_protection(rdev, false); ci_enable_power_containment(rdev, false); @@ -4729,12 +5267,13 @@ void ci_dpm_disable(struct radeon_device *rdev) ci_enable_spread_spectrum(rdev, false); ci_enable_auto_throttle_source(rdev, RADEON_DPM_AUTO_THROTTLE_SRC_THERMAL, false); ci_stop_dpm(rdev); - ci_enable_ds_master_switch(rdev, true); + ci_enable_ds_master_switch(rdev, false); ci_enable_ulv(rdev, false); ci_clear_vc(rdev); ci_reset_to_default(rdev); ci_dpm_stop_smc(rdev); ci_force_switch_to_arb_f0(rdev); + ci_enable_thermal_based_sclk_dpm(rdev, false); ci_update_current_ps(rdev, boot_ps); } @@ -4804,11 +5343,6 @@ int ci_dpm_set_power_state(struct radeon_device *rdev) return 0; } -int ci_dpm_power_control_set_level(struct radeon_device *rdev) -{ - return ci_power_control_set_level(rdev); -} - void ci_dpm_reset_asic(struct radeon_device *rdev) { ci_set_boot_state(rdev); @@ -5068,6 +5602,8 @@ void ci_dpm_fini(struct radeon_device *rdev) int ci_dpm_init(struct radeon_device *rdev) { int index = GetIndexIntoMasterTable(DATA, ASIC_InternalSS_Info); + SMU7_Discrete_DpmTable *dpm_table; + struct radeon_gpio_rec gpio; u16 data_offset, size; u8 frev, crev; struct ci_power_info *pi; @@ -5137,6 +5673,7 @@ int ci_dpm_init(struct radeon_device *rdev) pi->sclk_dpm_key_disabled = 0; pi->mclk_dpm_key_disabled = 0; pi->pcie_dpm_key_disabled = 0; + pi->thermal_sclk_dpm_enabled = 0; /* mclk dpm is unstable on some R7 260X cards with the old mc ucode */ if ((rdev->pdev->device == 0x6658) && @@ -5201,6 +5738,55 @@ int ci_dpm_init(struct radeon_device *rdev) pi->uvd_enabled = false; + dpm_table = &pi->smc_state_table; + + gpio = radeon_atombios_lookup_gpio(rdev, VDDC_VRHOT_GPIO_PINID); + if (gpio.valid) { + dpm_table->VRHotGpio = gpio.shift; + rdev->pm.dpm.platform_caps |= ATOM_PP_PLATFORM_CAP_REGULATOR_HOT; + } else { + dpm_table->VRHotGpio = CISLANDS_UNUSED_GPIO_PIN; + rdev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_REGULATOR_HOT; + } + + gpio = radeon_atombios_lookup_gpio(rdev, PP_AC_DC_SWITCH_GPIO_PINID); + if (gpio.valid) { + dpm_table->AcDcGpio = gpio.shift; + rdev->pm.dpm.platform_caps |= ATOM_PP_PLATFORM_CAP_HARDWAREDC; + } else { + dpm_table->AcDcGpio = CISLANDS_UNUSED_GPIO_PIN; + rdev->pm.dpm.platform_caps &= ~ATOM_PP_PLATFORM_CAP_HARDWAREDC; + } + + gpio = radeon_atombios_lookup_gpio(rdev, VDDC_PCC_GPIO_PINID); + if (gpio.valid) { + u32 tmp = RREG32_SMC(CNB_PWRMGT_CNTL); + + switch (gpio.shift) { + case 0: + tmp &= ~GNB_SLOW_MODE_MASK; + tmp |= GNB_SLOW_MODE(1); + break; + case 1: + tmp &= ~GNB_SLOW_MODE_MASK; + tmp |= GNB_SLOW_MODE(2); + break; + case 2: + tmp |= GNB_SLOW; + break; + case 3: + tmp |= FORCE_NB_PS1; + break; + case 4: + tmp |= DPM_ENABLED; + break; + default: + DRM_ERROR("Invalid PCC GPIO: %u!\n", gpio.shift); + break; + } + WREG32_SMC(CNB_PWRMGT_CNTL, tmp); + } + pi->voltage_control = CISLANDS_VOLTAGE_CONTROL_NONE; pi->vddci_control = CISLANDS_VOLTAGE_CONTROL_NONE; pi->mvdd_control = CISLANDS_VOLTAGE_CONTROL_NONE; @@ -5262,6 +5848,8 @@ int ci_dpm_init(struct radeon_device *rdev) rdev->pm.dpm.dyn_state.max_clock_voltage_on_dc = rdev->pm.dpm.dyn_state.max_clock_voltage_on_ac; + pi->fan_ctrl_is_in_default_mode = true; + return 0; } diff --git a/drivers/gpu/drm/radeon/ci_dpm.h b/drivers/gpu/drm/radeon/ci_dpm.h index 93bbed9..84e3d3b 100644 --- a/drivers/gpu/drm/radeon/ci_dpm.h +++ b/drivers/gpu/drm/radeon/ci_dpm.h @@ -33,6 +33,8 @@ #define CISLANDS_MAX_HARDWARE_POWERLEVELS 2 +#define CISLANDS_UNUSED_GPIO_PIN 0x7F + struct ci_pl { u32 mclk; u32 sclk; @@ -237,6 +239,7 @@ struct ci_power_info { u32 sclk_dpm_key_disabled; u32 mclk_dpm_key_disabled; u32 pcie_dpm_key_disabled; + u32 thermal_sclk_dpm_enabled; struct ci_pcie_perf_range pcie_gen_performance; struct ci_pcie_perf_range pcie_lane_performance; struct ci_pcie_perf_range pcie_gen_powersaving; @@ -264,6 +267,7 @@ struct ci_power_info { bool caps_automatic_dc_transition; bool caps_sclk_throttle_low_notification; bool caps_dynamic_ac_timing; + bool caps_od_fuzzy_fan_control_support; /* flags */ bool thermal_protection; bool pcie_performance_request; @@ -285,6 +289,10 @@ struct ci_power_info { struct ci_ps current_ps; struct radeon_ps requested_rps; struct ci_ps requested_ps; + /* fan control */ + bool fan_ctrl_is_in_default_mode; + u32 t_min; + u32 fan_ctrl_default_mode; }; #define CISLANDS_VOLTAGE_CONTROL_NONE 0x0 diff --git a/drivers/gpu/drm/radeon/ci_smc.c b/drivers/gpu/drm/radeon/ci_smc.c index b630edc..e78bcad 100644 --- a/drivers/gpu/drm/radeon/ci_smc.c +++ b/drivers/gpu/drm/radeon/ci_smc.c @@ -129,7 +129,7 @@ void ci_reset_smc(struct radeon_device *rdev) int ci_program_jump_on_start(struct radeon_device *rdev) { - static u8 data[] = { 0xE0, 0x00, 0x80, 0x40 }; + static const u8 data[] = { 0xE0, 0x00, 0x80, 0x40 }; return ci_copy_bytes_to_smc(rdev, 0x0, data, 4, sizeof(data)+1); } diff --git a/drivers/gpu/drm/radeon/cik.c b/drivers/gpu/drm/radeon/cik.c index 89c01fa..6dcde37 100644 --- a/drivers/gpu/drm/radeon/cik.c +++ b/drivers/gpu/drm/radeon/cik.c @@ -32,6 +32,7 @@ #include "cik_blit_shaders.h" #include "radeon_ucode.h" #include "clearstate_ci.h" +#include "radeon_kfd.h" MODULE_FIRMWARE("radeon/BONAIRE_pfp.bin"); MODULE_FIRMWARE("radeon/BONAIRE_me.bin"); @@ -1563,6 +1564,8 @@ static const u32 godavari_golden_registers[] = static void cik_init_golden_registers(struct radeon_device *rdev) { + /* Some of the registers might be dependent on GRBM_GFX_INDEX */ + mutex_lock(&rdev->grbm_idx_mutex); switch (rdev->family) { case CHIP_BONAIRE: radeon_program_register_sequence(rdev, @@ -1637,6 +1640,7 @@ static void cik_init_golden_registers(struct radeon_device *rdev) default: break; } + mutex_unlock(&rdev->grbm_idx_mutex); } /** @@ -1806,7 +1810,7 @@ int ci_mc_load_microcode(struct radeon_device *rdev) { const __be32 *fw_data = NULL; const __le32 *new_fw_data = NULL; - u32 running, blackout = 0; + u32 running, blackout = 0, tmp; u32 *io_mc_regs = NULL; const __le32 *new_io_mc_regs = NULL; int i, regs_size, ucode_size; @@ -1866,6 +1870,15 @@ int ci_mc_load_microcode(struct radeon_device *rdev) WREG32(MC_SEQ_IO_DEBUG_DATA, io_mc_regs[(i << 1) + 1]); } } + + tmp = RREG32(MC_SEQ_MISC0); + if ((rdev->pdev->device == 0x6649) && ((tmp & 0xff00) == 0x5600)) { + WREG32(MC_SEQ_IO_DEBUG_INDEX, 5); + WREG32(MC_SEQ_IO_DEBUG_DATA, 0x00000023); + WREG32(MC_SEQ_IO_DEBUG_INDEX, 9); + WREG32(MC_SEQ_IO_DEBUG_DATA, 0x000001f0); + } + /* load the MC ucode */ for (i = 0; i < ucode_size; i++) { if (rdev->new_fw) @@ -3419,6 +3432,7 @@ static void cik_setup_rb(struct radeon_device *rdev, u32 disabled_rbs = 0; u32 enabled_rbs = 0; + mutex_lock(&rdev->grbm_idx_mutex); for (i = 0; i < se_num; i++) { for (j = 0; j < sh_per_se; j++) { cik_select_se_sh(rdev, i, j); @@ -3430,6 +3444,7 @@ static void cik_setup_rb(struct radeon_device *rdev, } } cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); + mutex_unlock(&rdev->grbm_idx_mutex); mask = 1; for (i = 0; i < max_rb_num_per_se * se_num; i++) { @@ -3440,6 +3455,7 @@ static void cik_setup_rb(struct radeon_device *rdev, rdev->config.cik.backend_enable_mask = enabled_rbs; + mutex_lock(&rdev->grbm_idx_mutex); for (i = 0; i < se_num; i++) { cik_select_se_sh(rdev, i, 0xffffffff); data = 0; @@ -3467,6 +3483,7 @@ static void cik_setup_rb(struct radeon_device *rdev, WREG32(PA_SC_RASTER_CONFIG, data); } cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); + mutex_unlock(&rdev->grbm_idx_mutex); } /** @@ -3684,6 +3701,12 @@ static void cik_gpu_init(struct radeon_device *rdev) /* set HW defaults for 3D engine */ WREG32(CP_MEQ_THRESHOLDS, MEQ1_START(0x30) | MEQ2_START(0x60)); + mutex_lock(&rdev->grbm_idx_mutex); + /* + * making sure that the following register writes will be broadcasted + * to all the shaders + */ + cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); WREG32(SX_DEBUG_1, 0x20); WREG32(TA_CNTL_AUX, 0x00010000); @@ -3739,6 +3762,7 @@ static void cik_gpu_init(struct radeon_device *rdev) WREG32(PA_CL_ENHANCE, CLIP_VTX_REORDER_ENA | NUM_CLIP_SEQ(3)); WREG32(PA_SC_ENHANCE, ENABLE_PA_SC_OUT_OF_ORDER); + mutex_unlock(&rdev->grbm_idx_mutex); udelay(50); } @@ -3970,31 +3994,27 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev, unsigned num_gpu_pages, struct reservation_object *resv) { - struct radeon_semaphore *sem = NULL; struct radeon_fence *fence; + struct radeon_sync sync; int ring_index = rdev->asic->copy.blit_ring_index; struct radeon_ring *ring = &rdev->ring[ring_index]; u32 size_in_bytes, cur_size_in_bytes, control; int i, num_loops; int r = 0; - r = radeon_semaphore_create(rdev, &sem); - if (r) { - DRM_ERROR("radeon: moving bo (%d).\n", r); - return ERR_PTR(r); - } + radeon_sync_create(&sync); size_in_bytes = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT); num_loops = DIV_ROUND_UP(size_in_bytes, 0x1fffff); r = radeon_ring_lock(rdev, ring, num_loops * 7 + 18); if (r) { DRM_ERROR("radeon: moving bo (%d).\n", r); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } - radeon_semaphore_sync_resv(rdev, sem, resv, false); - radeon_semaphore_sync_rings(rdev, sem, ring->idx); + radeon_sync_resv(rdev, &sync, resv, false); + radeon_sync_rings(rdev, &sync, ring->idx); for (i = 0; i < num_loops; i++) { cur_size_in_bytes = size_in_bytes; @@ -4018,12 +4038,12 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev, r = radeon_fence_emit(rdev, &fence, ring->idx); if (r) { radeon_ring_unlock_undo(rdev, ring); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } radeon_ring_unlock_commit(rdev, ring, false); - radeon_semaphore_free(rdev, &sem, fence); + radeon_sync_free(rdev, &sync, fence); return fence; } @@ -4046,6 +4066,7 @@ struct radeon_fence *cik_copy_cpdma(struct radeon_device *rdev, void cik_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) { struct radeon_ring *ring = &rdev->ring[ib->ring]; + unsigned vm_id = ib->vm ? ib->vm->ids[ib->ring].id : 0; u32 header, control = INDIRECT_BUFFER_VALID; if (ib->is_const_ib) { @@ -4074,8 +4095,7 @@ void cik_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) header = PACKET3(PACKET3_INDIRECT_BUFFER, 2); } - control |= ib->length_dw | - (ib->vm ? (ib->vm->id << 24) : 0); + control |= ib->length_dw | (vm_id << 24); radeon_ring_write(ring, header); radeon_ring_write(ring, @@ -4675,12 +4695,11 @@ static int cik_mec_init(struct radeon_device *rdev) /* * KV: 2 MEC, 4 Pipes/MEC, 8 Queues/Pipe - 64 Queues total * CI/KB: 1 MEC, 4 Pipes/MEC, 8 Queues/Pipe - 32 Queues total + * Nonetheless, we assign only 1 pipe because all other pipes will + * be handled by KFD */ - if (rdev->family == CHIP_KAVERI) - rdev->mec.num_mec = 2; - else - rdev->mec.num_mec = 1; - rdev->mec.num_pipe = 4; + rdev->mec.num_mec = 1; + rdev->mec.num_pipe = 1; rdev->mec.num_queue = rdev->mec.num_mec * rdev->mec.num_pipe * 8; if (rdev->mec.hpd_eop_obj == NULL) { @@ -4822,28 +4841,24 @@ static int cik_cp_compute_resume(struct radeon_device *rdev) /* init the pipes */ mutex_lock(&rdev->srbm_mutex); - for (i = 0; i < (rdev->mec.num_pipe * rdev->mec.num_mec); i++) { - int me = (i < 4) ? 1 : 2; - int pipe = (i < 4) ? i : (i - 4); - eop_gpu_addr = rdev->mec.hpd_eop_gpu_addr + (i * MEC_HPD_SIZE * 2); + eop_gpu_addr = rdev->mec.hpd_eop_gpu_addr; - cik_srbm_select(rdev, me, pipe, 0, 0); + cik_srbm_select(rdev, 0, 0, 0, 0); - /* write the EOP addr */ - WREG32(CP_HPD_EOP_BASE_ADDR, eop_gpu_addr >> 8); - WREG32(CP_HPD_EOP_BASE_ADDR_HI, upper_32_bits(eop_gpu_addr) >> 8); + /* write the EOP addr */ + WREG32(CP_HPD_EOP_BASE_ADDR, eop_gpu_addr >> 8); + WREG32(CP_HPD_EOP_BASE_ADDR_HI, upper_32_bits(eop_gpu_addr) >> 8); - /* set the VMID assigned */ - WREG32(CP_HPD_EOP_VMID, 0); + /* set the VMID assigned */ + WREG32(CP_HPD_EOP_VMID, 0); + + /* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */ + tmp = RREG32(CP_HPD_EOP_CONTROL); + tmp &= ~EOP_SIZE_MASK; + tmp |= order_base_2(MEC_HPD_SIZE / 8); + WREG32(CP_HPD_EOP_CONTROL, tmp); - /* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */ - tmp = RREG32(CP_HPD_EOP_CONTROL); - tmp &= ~EOP_SIZE_MASK; - tmp |= order_base_2(MEC_HPD_SIZE / 8); - WREG32(CP_HPD_EOP_CONTROL, tmp); - } - cik_srbm_select(rdev, 0, 0, 0, 0); mutex_unlock(&rdev->srbm_mutex); /* init the queues. Just two for now. */ @@ -5897,8 +5912,13 @@ int cik_ib_parse(struct radeon_device *rdev, struct radeon_ib *ib) */ int cik_vm_init(struct radeon_device *rdev) { - /* number of VMs */ - rdev->vm_manager.nvm = 16; + /* + * number of VMs + * VMID 0 is reserved for System + * radeon graphics/compute will use VMIDs 1-7 + * amdkfd will use VMIDs 8-15 + */ + rdev->vm_manager.nvm = RADEON_NUM_OF_VMIDS; /* base offset of vram pages */ if (rdev->flags & RADEON_IS_IGP) { u64 tmp = RREG32(MC_VM_FB_OFFSET); @@ -5958,26 +5978,23 @@ static void cik_vm_decode_fault(struct radeon_device *rdev, * Update the page table base and flush the VM TLB * using the CP (CIK). */ -void cik_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) +void cik_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr) { - struct radeon_ring *ring = &rdev->ring[ridx]; - int usepfp = (ridx == RADEON_RING_TYPE_GFX_INDEX); - - if (vm == NULL) - return; + int usepfp = (ring->idx == RADEON_RING_TYPE_GFX_INDEX); radeon_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3)); radeon_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) | WRITE_DATA_DST_SEL(0))); - if (vm->id < 8) { + if (vm_id < 8) { radeon_ring_write(ring, - (VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm->id << 2)) >> 2); + (VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm_id << 2)) >> 2); } else { radeon_ring_write(ring, - (VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + ((vm->id - 8) << 2)) >> 2); + (VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + ((vm_id - 8) << 2)) >> 2); } radeon_ring_write(ring, 0); - radeon_ring_write(ring, vm->pd_gpu_addr >> 12); + radeon_ring_write(ring, pd_addr >> 12); /* update SH_MEM_* regs */ radeon_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3)); @@ -5985,7 +6002,7 @@ void cik_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) WRITE_DATA_DST_SEL(0))); radeon_ring_write(ring, SRBM_GFX_CNTL >> 2); radeon_ring_write(ring, 0); - radeon_ring_write(ring, VMID(vm->id)); + radeon_ring_write(ring, VMID(vm_id)); radeon_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 6)); radeon_ring_write(ring, (WRITE_DATA_ENGINE_SEL(usepfp) | @@ -6006,7 +6023,7 @@ void cik_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) radeon_ring_write(ring, VMID(0)); /* HDP flush */ - cik_hdp_flush_cp_ring_emit(rdev, ridx); + cik_hdp_flush_cp_ring_emit(rdev, ring->idx); /* bits 0-15 are the VM contexts0-15 */ radeon_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3)); @@ -6014,7 +6031,7 @@ void cik_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) WRITE_DATA_DST_SEL(0))); radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2); radeon_ring_write(ring, 0); - radeon_ring_write(ring, 1 << vm->id); + radeon_ring_write(ring, 1 << vm_id); /* compute doesn't have PFP */ if (usepfp) { @@ -6059,6 +6076,7 @@ static void cik_wait_for_rlc_serdes(struct radeon_device *rdev) u32 i, j, k; u32 mask; + mutex_lock(&rdev->grbm_idx_mutex); for (i = 0; i < rdev->config.cik.max_shader_engines; i++) { for (j = 0; j < rdev->config.cik.max_sh_per_se; j++) { cik_select_se_sh(rdev, i, j); @@ -6070,6 +6088,7 @@ static void cik_wait_for_rlc_serdes(struct radeon_device *rdev) } } cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); + mutex_unlock(&rdev->grbm_idx_mutex); mask = SE_MASTER_BUSY_MASK | GC_MASTER_BUSY | TC0_MASTER_BUSY | TC1_MASTER_BUSY; for (k = 0; k < rdev->usec_timeout; k++) { @@ -6204,10 +6223,12 @@ static int cik_rlc_resume(struct radeon_device *rdev) WREG32(RLC_LB_CNTR_INIT, 0); WREG32(RLC_LB_CNTR_MAX, 0x00008000); + mutex_lock(&rdev->grbm_idx_mutex); cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); WREG32(RLC_LB_INIT_CU_MASK, 0xffffffff); WREG32(RLC_LB_PARAMS, 0x00600408); WREG32(RLC_LB_CNTL, 0x80000004); + mutex_unlock(&rdev->grbm_idx_mutex); WREG32(RLC_MC_CNTL, 0); WREG32(RLC_UCODE_CNTL, 0); @@ -6274,11 +6295,13 @@ static void cik_enable_cgcg(struct radeon_device *rdev, bool enable) tmp = cik_halt_rlc(rdev); + mutex_lock(&rdev->grbm_idx_mutex); cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); WREG32(RLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff); WREG32(RLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff); tmp2 = BPM_ADDR_MASK | CGCG_OVERRIDE_0 | CGLS_ENABLE; WREG32(RLC_SERDES_WR_CTRL, tmp2); + mutex_unlock(&rdev->grbm_idx_mutex); cik_update_rlc(rdev, tmp); @@ -6314,17 +6337,20 @@ static void cik_enable_mgcg(struct radeon_device *rdev, bool enable) } orig = data = RREG32(RLC_CGTT_MGCG_OVERRIDE); + data |= 0x00000001; data &= 0xfffffffd; if (orig != data) WREG32(RLC_CGTT_MGCG_OVERRIDE, data); tmp = cik_halt_rlc(rdev); + mutex_lock(&rdev->grbm_idx_mutex); cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); WREG32(RLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff); WREG32(RLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff); data = BPM_ADDR_MASK | MGCG_OVERRIDE_0; WREG32(RLC_SERDES_WR_CTRL, data); + mutex_unlock(&rdev->grbm_idx_mutex); cik_update_rlc(rdev, tmp); @@ -6345,7 +6371,7 @@ static void cik_enable_mgcg(struct radeon_device *rdev, bool enable) } } else { orig = data = RREG32(RLC_CGTT_MGCG_OVERRIDE); - data |= 0x00000002; + data |= 0x00000003; if (orig != data) WREG32(RLC_CGTT_MGCG_OVERRIDE, data); @@ -6368,11 +6394,13 @@ static void cik_enable_mgcg(struct radeon_device *rdev, bool enable) tmp = cik_halt_rlc(rdev); + mutex_lock(&rdev->grbm_idx_mutex); cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); WREG32(RLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff); WREG32(RLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff); data = BPM_ADDR_MASK | MGCG_OVERRIDE_1; WREG32(RLC_SERDES_WR_CTRL, data); + mutex_unlock(&rdev->grbm_idx_mutex); cik_update_rlc(rdev, tmp); } @@ -6801,10 +6829,12 @@ static u32 cik_get_cu_active_bitmap(struct radeon_device *rdev, u32 se, u32 sh) u32 mask = 0, tmp, tmp1; int i; + mutex_lock(&rdev->grbm_idx_mutex); cik_select_se_sh(rdev, se, sh); tmp = RREG32(CC_GC_SHADER_ARRAY_CONFIG); tmp1 = RREG32(GC_USER_SHADER_ARRAY_CONFIG); cik_select_se_sh(rdev, 0xffffffff, 0xffffffff); + mutex_unlock(&rdev->grbm_idx_mutex); tmp &= 0xffff0000; @@ -7288,8 +7318,7 @@ static int cik_irq_init(struct radeon_device *rdev) int cik_irq_set(struct radeon_device *rdev) { u32 cp_int_cntl; - u32 cp_m1p0, cp_m1p1, cp_m1p2, cp_m1p3; - u32 cp_m2p0, cp_m2p1, cp_m2p2, cp_m2p3; + u32 cp_m1p0; u32 crtc1 = 0, crtc2 = 0, crtc3 = 0, crtc4 = 0, crtc5 = 0, crtc6 = 0; u32 hpd1, hpd2, hpd3, hpd4, hpd5, hpd6; u32 grbm_int_cntl = 0; @@ -7323,13 +7352,6 @@ int cik_irq_set(struct radeon_device *rdev) dma_cntl1 = RREG32(SDMA0_CNTL + SDMA1_REGISTER_OFFSET) & ~TRAP_ENABLE; cp_m1p0 = RREG32(CP_ME1_PIPE0_INT_CNTL) & ~TIME_STAMP_INT_ENABLE; - cp_m1p1 = RREG32(CP_ME1_PIPE1_INT_CNTL) & ~TIME_STAMP_INT_ENABLE; - cp_m1p2 = RREG32(CP_ME1_PIPE2_INT_CNTL) & ~TIME_STAMP_INT_ENABLE; - cp_m1p3 = RREG32(CP_ME1_PIPE3_INT_CNTL) & ~TIME_STAMP_INT_ENABLE; - cp_m2p0 = RREG32(CP_ME2_PIPE0_INT_CNTL) & ~TIME_STAMP_INT_ENABLE; - cp_m2p1 = RREG32(CP_ME2_PIPE1_INT_CNTL) & ~TIME_STAMP_INT_ENABLE; - cp_m2p2 = RREG32(CP_ME2_PIPE2_INT_CNTL) & ~TIME_STAMP_INT_ENABLE; - cp_m2p3 = RREG32(CP_ME2_PIPE3_INT_CNTL) & ~TIME_STAMP_INT_ENABLE; if (rdev->flags & RADEON_IS_IGP) thermal_int = RREG32_SMC(CG_THERMAL_INT_CTRL) & @@ -7351,33 +7373,6 @@ int cik_irq_set(struct radeon_device *rdev) case 0: cp_m1p0 |= TIME_STAMP_INT_ENABLE; break; - case 1: - cp_m1p1 |= TIME_STAMP_INT_ENABLE; - break; - case 2: - cp_m1p2 |= TIME_STAMP_INT_ENABLE; - break; - case 3: - cp_m1p2 |= TIME_STAMP_INT_ENABLE; - break; - default: - DRM_DEBUG("si_irq_set: sw int cp1 invalid pipe %d\n", ring->pipe); - break; - } - } else if (ring->me == 2) { - switch (ring->pipe) { - case 0: - cp_m2p0 |= TIME_STAMP_INT_ENABLE; - break; - case 1: - cp_m2p1 |= TIME_STAMP_INT_ENABLE; - break; - case 2: - cp_m2p2 |= TIME_STAMP_INT_ENABLE; - break; - case 3: - cp_m2p2 |= TIME_STAMP_INT_ENABLE; - break; default: DRM_DEBUG("si_irq_set: sw int cp1 invalid pipe %d\n", ring->pipe); break; @@ -7394,33 +7389,6 @@ int cik_irq_set(struct radeon_device *rdev) case 0: cp_m1p0 |= TIME_STAMP_INT_ENABLE; break; - case 1: - cp_m1p1 |= TIME_STAMP_INT_ENABLE; - break; - case 2: - cp_m1p2 |= TIME_STAMP_INT_ENABLE; - break; - case 3: - cp_m1p2 |= TIME_STAMP_INT_ENABLE; - break; - default: - DRM_DEBUG("si_irq_set: sw int cp2 invalid pipe %d\n", ring->pipe); - break; - } - } else if (ring->me == 2) { - switch (ring->pipe) { - case 0: - cp_m2p0 |= TIME_STAMP_INT_ENABLE; - break; - case 1: - cp_m2p1 |= TIME_STAMP_INT_ENABLE; - break; - case 2: - cp_m2p2 |= TIME_STAMP_INT_ENABLE; - break; - case 3: - cp_m2p2 |= TIME_STAMP_INT_ENABLE; - break; default: DRM_DEBUG("si_irq_set: sw int cp2 invalid pipe %d\n", ring->pipe); break; @@ -7509,13 +7477,6 @@ int cik_irq_set(struct radeon_device *rdev) WREG32(SDMA0_CNTL + SDMA1_REGISTER_OFFSET, dma_cntl1); WREG32(CP_ME1_PIPE0_INT_CNTL, cp_m1p0); - WREG32(CP_ME1_PIPE1_INT_CNTL, cp_m1p1); - WREG32(CP_ME1_PIPE2_INT_CNTL, cp_m1p2); - WREG32(CP_ME1_PIPE3_INT_CNTL, cp_m1p3); - WREG32(CP_ME2_PIPE0_INT_CNTL, cp_m2p0); - WREG32(CP_ME2_PIPE1_INT_CNTL, cp_m2p1); - WREG32(CP_ME2_PIPE2_INT_CNTL, cp_m2p2); - WREG32(CP_ME2_PIPE3_INT_CNTL, cp_m2p3); WREG32(GRBM_INT_CNTL, grbm_int_cntl); @@ -7832,6 +7793,10 @@ restart_ih: while (rptr != wptr) { /* wptr/rptr are in bytes! */ ring_index = rptr / 4; + + radeon_kfd_interrupt(rdev, + (const void *) &rdev->ih.ring[ring_index]); + src_id = le32_to_cpu(rdev->ih.ring[ring_index]) & 0xff; src_data = le32_to_cpu(rdev->ih.ring[ring_index + 1]) & 0xfffffff; ring_id = le32_to_cpu(rdev->ih.ring[ring_index + 2]) & 0xff; @@ -8521,6 +8486,10 @@ static int cik_startup(struct radeon_device *rdev) if (r) return r; + r = radeon_kfd_resume(rdev); + if (r) + return r; + return 0; } @@ -8569,6 +8538,7 @@ int cik_resume(struct radeon_device *rdev) */ int cik_suspend(struct radeon_device *rdev) { + radeon_kfd_suspend(rdev); radeon_pm_suspend(rdev); dce6_audio_fini(rdev); radeon_vm_manager_fini(rdev); diff --git a/drivers/gpu/drm/radeon/cik_reg.h b/drivers/gpu/drm/radeon/cik_reg.h index ca1bb61..79c45e8 100644 --- a/drivers/gpu/drm/radeon/cik_reg.h +++ b/drivers/gpu/drm/radeon/cik_reg.h @@ -147,4 +147,140 @@ #define CIK_LB_DESKTOP_HEIGHT 0x6b0c +#define CP_HQD_IQ_RPTR 0xC970u +#define AQL_ENABLE (1U << 0) + +#define IDLE (1 << 2) + +struct cik_mqd { + uint32_t header; + uint32_t compute_dispatch_initiator; + uint32_t compute_dim_x; + uint32_t compute_dim_y; + uint32_t compute_dim_z; + uint32_t compute_start_x; + uint32_t compute_start_y; + uint32_t compute_start_z; + uint32_t compute_num_thread_x; + uint32_t compute_num_thread_y; + uint32_t compute_num_thread_z; + uint32_t compute_pipelinestat_enable; + uint32_t compute_perfcount_enable; + uint32_t compute_pgm_lo; + uint32_t compute_pgm_hi; + uint32_t compute_tba_lo; + uint32_t compute_tba_hi; + uint32_t compute_tma_lo; + uint32_t compute_tma_hi; + uint32_t compute_pgm_rsrc1; + uint32_t compute_pgm_rsrc2; + uint32_t compute_vmid; + uint32_t compute_resource_limits; + uint32_t compute_static_thread_mgmt_se0; + uint32_t compute_static_thread_mgmt_se1; + uint32_t compute_tmpring_size; + uint32_t compute_static_thread_mgmt_se2; + uint32_t compute_static_thread_mgmt_se3; + uint32_t compute_restart_x; + uint32_t compute_restart_y; + uint32_t compute_restart_z; + uint32_t compute_thread_trace_enable; + uint32_t compute_misc_reserved; + uint32_t compute_user_data_0; + uint32_t compute_user_data_1; + uint32_t compute_user_data_2; + uint32_t compute_user_data_3; + uint32_t compute_user_data_4; + uint32_t compute_user_data_5; + uint32_t compute_user_data_6; + uint32_t compute_user_data_7; + uint32_t compute_user_data_8; + uint32_t compute_user_data_9; + uint32_t compute_user_data_10; + uint32_t compute_user_data_11; + uint32_t compute_user_data_12; + uint32_t compute_user_data_13; + uint32_t compute_user_data_14; + uint32_t compute_user_data_15; + uint32_t cp_compute_csinvoc_count_lo; + uint32_t cp_compute_csinvoc_count_hi; + uint32_t cp_mqd_base_addr_lo; + uint32_t cp_mqd_base_addr_hi; + uint32_t cp_hqd_active; + uint32_t cp_hqd_vmid; + uint32_t cp_hqd_persistent_state; + uint32_t cp_hqd_pipe_priority; + uint32_t cp_hqd_queue_priority; + uint32_t cp_hqd_quantum; + uint32_t cp_hqd_pq_base_lo; + uint32_t cp_hqd_pq_base_hi; + uint32_t cp_hqd_pq_rptr; + uint32_t cp_hqd_pq_rptr_report_addr_lo; + uint32_t cp_hqd_pq_rptr_report_addr_hi; + uint32_t cp_hqd_pq_wptr_poll_addr_lo; + uint32_t cp_hqd_pq_wptr_poll_addr_hi; + uint32_t cp_hqd_pq_doorbell_control; + uint32_t cp_hqd_pq_wptr; + uint32_t cp_hqd_pq_control; + uint32_t cp_hqd_ib_base_addr_lo; + uint32_t cp_hqd_ib_base_addr_hi; + uint32_t cp_hqd_ib_rptr; + uint32_t cp_hqd_ib_control; + uint32_t cp_hqd_iq_timer; + uint32_t cp_hqd_iq_rptr; + uint32_t cp_hqd_dequeue_request; + uint32_t cp_hqd_dma_offload; + uint32_t cp_hqd_sema_cmd; + uint32_t cp_hqd_msg_type; + uint32_t cp_hqd_atomic0_preop_lo; + uint32_t cp_hqd_atomic0_preop_hi; + uint32_t cp_hqd_atomic1_preop_lo; + uint32_t cp_hqd_atomic1_preop_hi; + uint32_t cp_hqd_hq_status0; + uint32_t cp_hqd_hq_control0; + uint32_t cp_mqd_control; + uint32_t cp_mqd_query_time_lo; + uint32_t cp_mqd_query_time_hi; + uint32_t cp_mqd_connect_start_time_lo; + uint32_t cp_mqd_connect_start_time_hi; + uint32_t cp_mqd_connect_end_time_lo; + uint32_t cp_mqd_connect_end_time_hi; + uint32_t cp_mqd_connect_end_wf_count; + uint32_t cp_mqd_connect_end_pq_rptr; + uint32_t cp_mqd_connect_end_pq_wptr; + uint32_t cp_mqd_connect_end_ib_rptr; + uint32_t reserved_96; + uint32_t reserved_97; + uint32_t reserved_98; + uint32_t reserved_99; + uint32_t iqtimer_pkt_header; + uint32_t iqtimer_pkt_dw0; + uint32_t iqtimer_pkt_dw1; + uint32_t iqtimer_pkt_dw2; + uint32_t iqtimer_pkt_dw3; + uint32_t iqtimer_pkt_dw4; + uint32_t iqtimer_pkt_dw5; + uint32_t iqtimer_pkt_dw6; + uint32_t reserved_108; + uint32_t reserved_109; + uint32_t reserved_110; + uint32_t reserved_111; + uint32_t queue_doorbell_id0; + uint32_t queue_doorbell_id1; + uint32_t queue_doorbell_id2; + uint32_t queue_doorbell_id3; + uint32_t queue_doorbell_id4; + uint32_t queue_doorbell_id5; + uint32_t queue_doorbell_id6; + uint32_t queue_doorbell_id7; + uint32_t queue_doorbell_id8; + uint32_t queue_doorbell_id9; + uint32_t queue_doorbell_id10; + uint32_t queue_doorbell_id11; + uint32_t queue_doorbell_id12; + uint32_t queue_doorbell_id13; + uint32_t queue_doorbell_id14; + uint32_t queue_doorbell_id15; +}; + #endif diff --git a/drivers/gpu/drm/radeon/cik_sdma.c b/drivers/gpu/drm/radeon/cik_sdma.c index d748963..dde5c7e 100644 --- a/drivers/gpu/drm/radeon/cik_sdma.c +++ b/drivers/gpu/drm/radeon/cik_sdma.c @@ -134,7 +134,7 @@ void cik_sdma_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) { struct radeon_ring *ring = &rdev->ring[ib->ring]; - u32 extra_bits = (ib->vm ? ib->vm->id : 0) & 0xf; + u32 extra_bits = (ib->vm ? ib->vm->ids[ib->ring].id : 0) & 0xf; if (rdev->wb.enabled) { u32 next_rptr = ring->wptr + 5; @@ -541,31 +541,27 @@ struct radeon_fence *cik_copy_dma(struct radeon_device *rdev, unsigned num_gpu_pages, struct reservation_object *resv) { - struct radeon_semaphore *sem = NULL; struct radeon_fence *fence; + struct radeon_sync sync; int ring_index = rdev->asic->copy.dma_ring_index; struct radeon_ring *ring = &rdev->ring[ring_index]; u32 size_in_bytes, cur_size_in_bytes; int i, num_loops; int r = 0; - r = radeon_semaphore_create(rdev, &sem); - if (r) { - DRM_ERROR("radeon: moving bo (%d).\n", r); - return ERR_PTR(r); - } + radeon_sync_create(&sync); size_in_bytes = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT); num_loops = DIV_ROUND_UP(size_in_bytes, 0x1fffff); r = radeon_ring_lock(rdev, ring, num_loops * 7 + 14); if (r) { DRM_ERROR("radeon: moving bo (%d).\n", r); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } - radeon_semaphore_sync_resv(rdev, sem, resv, false); - radeon_semaphore_sync_rings(rdev, sem, ring->idx); + radeon_sync_resv(rdev, &sync, resv, false); + radeon_sync_rings(rdev, &sync, ring->idx); for (i = 0; i < num_loops; i++) { cur_size_in_bytes = size_in_bytes; @@ -586,12 +582,12 @@ struct radeon_fence *cik_copy_dma(struct radeon_device *rdev, r = radeon_fence_emit(rdev, &fence, ring->idx); if (r) { radeon_ring_unlock_undo(rdev, ring); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } radeon_ring_unlock_commit(rdev, ring, false); - radeon_semaphore_free(rdev, &sem, fence); + radeon_sync_free(rdev, &sync, fence); return fence; } @@ -904,25 +900,21 @@ void cik_sdma_vm_pad_ib(struct radeon_ib *ib) * Update the page table base and flush the VM TLB * using sDMA (CIK). */ -void cik_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) +void cik_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr) { - struct radeon_ring *ring = &rdev->ring[ridx]; - - if (vm == NULL) - return; - radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000)); - if (vm->id < 8) { - radeon_ring_write(ring, (VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm->id << 2)) >> 2); + if (vm_id < 8) { + radeon_ring_write(ring, (VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm_id << 2)) >> 2); } else { - radeon_ring_write(ring, (VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + ((vm->id - 8) << 2)) >> 2); + radeon_ring_write(ring, (VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + ((vm_id - 8) << 2)) >> 2); } - radeon_ring_write(ring, vm->pd_gpu_addr >> 12); + radeon_ring_write(ring, pd_addr >> 12); /* update SH_MEM_* regs */ radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000)); radeon_ring_write(ring, SRBM_GFX_CNTL >> 2); - radeon_ring_write(ring, VMID(vm->id)); + radeon_ring_write(ring, VMID(vm_id)); radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000)); radeon_ring_write(ring, SH_MEM_BASES >> 2); @@ -945,11 +937,11 @@ void cik_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm radeon_ring_write(ring, VMID(0)); /* flush HDP */ - cik_sdma_hdp_flush_ring_emit(rdev, ridx); + cik_sdma_hdp_flush_ring_emit(rdev, ring->idx); /* flush TLB */ radeon_ring_write(ring, SDMA_PACKET(SDMA_OPCODE_SRBM_WRITE, 0, 0xf000)); radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2); - radeon_ring_write(ring, 1 << vm->id); + radeon_ring_write(ring, 1 << vm_id); } diff --git a/drivers/gpu/drm/radeon/cikd.h b/drivers/gpu/drm/radeon/cikd.h index 0c6e1b5..ba85986 100644 --- a/drivers/gpu/drm/radeon/cikd.h +++ b/drivers/gpu/drm/radeon/cikd.h @@ -30,6 +30,8 @@ #define CIK_RB_BITMAP_WIDTH_PER_SH 2 #define HAWAII_RB_BITMAP_WIDTH_PER_SH 4 +#define RADEON_NUM_OF_VMIDS 8 + /* DIDT IND registers */ #define DIDT_SQ_CTRL0 0x0 # define DIDT_CTRL_EN (1 << 0) @@ -184,7 +186,10 @@ #define DIG_THERM_DPM(x) ((x) << 14) #define DIG_THERM_DPM_MASK 0x003FC000 #define DIG_THERM_DPM_SHIFT 14 - +#define CG_THERMAL_STATUS 0xC0300008 +#define FDO_PWM_DUTY(x) ((x) << 9) +#define FDO_PWM_DUTY_MASK (0xff << 9) +#define FDO_PWM_DUTY_SHIFT 9 #define CG_THERMAL_INT 0xC030000C #define CI_DIG_THERM_INTH(x) ((x) << 8) #define CI_DIG_THERM_INTH_MASK 0x0000FF00 @@ -194,7 +199,10 @@ #define CI_DIG_THERM_INTL_SHIFT 16 #define THERM_INT_MASK_HIGH (1 << 24) #define THERM_INT_MASK_LOW (1 << 25) - +#define CG_MULT_THERMAL_CTRL 0xC0300010 +#define TEMP_SEL(x) ((x) << 20) +#define TEMP_SEL_MASK (0xff << 20) +#define TEMP_SEL_SHIFT 20 #define CG_MULT_THERMAL_STATUS 0xC0300014 #define ASIC_MAX_TEMP(x) ((x) << 0) #define ASIC_MAX_TEMP_MASK 0x000001ff @@ -203,6 +211,36 @@ #define CTF_TEMP_MASK 0x0003fe00 #define CTF_TEMP_SHIFT 9 +#define CG_FDO_CTRL0 0xC0300064 +#define FDO_STATIC_DUTY(x) ((x) << 0) +#define FDO_STATIC_DUTY_MASK 0x000000FF +#define FDO_STATIC_DUTY_SHIFT 0 +#define CG_FDO_CTRL1 0xC0300068 +#define FMAX_DUTY100(x) ((x) << 0) +#define FMAX_DUTY100_MASK 0x000000FF +#define FMAX_DUTY100_SHIFT 0 +#define CG_FDO_CTRL2 0xC030006C +#define TMIN(x) ((x) << 0) +#define TMIN_MASK 0x000000FF +#define TMIN_SHIFT 0 +#define FDO_PWM_MODE(x) ((x) << 11) +#define FDO_PWM_MODE_MASK (7 << 11) +#define FDO_PWM_MODE_SHIFT 11 +#define TACH_PWM_RESP_RATE(x) ((x) << 25) +#define TACH_PWM_RESP_RATE_MASK (0x7f << 25) +#define TACH_PWM_RESP_RATE_SHIFT 25 +#define CG_TACH_CTRL 0xC0300070 +# define EDGE_PER_REV(x) ((x) << 0) +# define EDGE_PER_REV_MASK (0x7 << 0) +# define EDGE_PER_REV_SHIFT 0 +# define TARGET_PERIOD(x) ((x) << 3) +# define TARGET_PERIOD_MASK 0xfffffff8 +# define TARGET_PERIOD_SHIFT 3 +#define CG_TACH_STATUS 0xC0300074 +# define TACH_PERIOD(x) ((x) << 0) +# define TACH_PERIOD_MASK 0xffffffff +# define TACH_PERIOD_SHIFT 0 + #define CG_ECLK_CNTL 0xC05000AC # define ECLK_DIVIDER_MASK 0x7f # define ECLK_DIR_CNTL_EN (1 << 8) @@ -1137,6 +1175,9 @@ #define SH_MEM_ALIGNMENT_MODE_UNALIGNED 3 #define DEFAULT_MTYPE(x) ((x) << 4) #define APE1_MTYPE(x) ((x) << 7) +/* valid for both DEFAULT_MTYPE and APE1_MTYPE */ +#define MTYPE_CACHED 0 +#define MTYPE_NONCACHED 3 #define SX_DEBUG_1 0x9060 @@ -1447,6 +1488,16 @@ #define CP_HQD_ACTIVE 0xC91C #define CP_HQD_VMID 0xC920 +#define CP_HQD_PERSISTENT_STATE 0xC924u +#define DEFAULT_CP_HQD_PERSISTENT_STATE (0x33U << 8) + +#define CP_HQD_PIPE_PRIORITY 0xC928u +#define CP_HQD_QUEUE_PRIORITY 0xC92Cu +#define CP_HQD_QUANTUM 0xC930u +#define QUANTUM_EN 1U +#define QUANTUM_SCALE_1MS (1U << 4) +#define QUANTUM_DURATION(x) ((x) << 8) + #define CP_HQD_PQ_BASE 0xC934 #define CP_HQD_PQ_BASE_HI 0xC938 #define CP_HQD_PQ_RPTR 0xC93C @@ -1474,12 +1525,32 @@ #define PRIV_STATE (1 << 30) #define KMD_QUEUE (1 << 31) -#define CP_HQD_DEQUEUE_REQUEST 0xC974 +#define CP_HQD_IB_BASE_ADDR 0xC95Cu +#define CP_HQD_IB_BASE_ADDR_HI 0xC960u +#define CP_HQD_IB_RPTR 0xC964u +#define CP_HQD_IB_CONTROL 0xC968u +#define IB_ATC_EN (1U << 23) +#define DEFAULT_MIN_IB_AVAIL_SIZE (3U << 20) + +#define CP_HQD_DEQUEUE_REQUEST 0xC974 +#define DEQUEUE_REQUEST_DRAIN 1 +#define DEQUEUE_REQUEST_RESET 2 #define CP_MQD_CONTROL 0xC99C #define MQD_VMID(x) ((x) << 0) #define MQD_VMID_MASK (0xf << 0) +#define CP_HQD_SEMA_CMD 0xC97Cu +#define CP_HQD_MSG_TYPE 0xC980u +#define CP_HQD_ATOMIC0_PREOP_LO 0xC984u +#define CP_HQD_ATOMIC0_PREOP_HI 0xC988u +#define CP_HQD_ATOMIC1_PREOP_LO 0xC98Cu +#define CP_HQD_ATOMIC1_PREOP_HI 0xC990u +#define CP_HQD_HQ_SCHEDULER0 0xC994u +#define CP_HQD_HQ_SCHEDULER1 0xC998u + +#define SH_STATIC_MEM_CONFIG 0x9604u + #define DB_RENDER_CONTROL 0x28000 #define PA_SC_RASTER_CONFIG 0x28350 @@ -2069,4 +2140,20 @@ #define VCE_CMD_IB_AUTO 0x00000005 #define VCE_CMD_SEMAPHORE 0x00000006 +#define ATC_VMID0_PASID_MAPPING 0x339Cu +#define ATC_VMID_PASID_MAPPING_UPDATE_STATUS 0x3398u +#define ATC_VMID_PASID_MAPPING_VALID (1U << 31) + +#define ATC_VM_APERTURE0_CNTL 0x3310u +#define ATS_ACCESS_MODE_NEVER 0 +#define ATS_ACCESS_MODE_ALWAYS 1 + +#define ATC_VM_APERTURE0_CNTL2 0x3318u +#define ATC_VM_APERTURE0_HIGH_ADDR 0x3308u +#define ATC_VM_APERTURE0_LOW_ADDR 0x3300u +#define ATC_VM_APERTURE1_CNTL 0x3314u +#define ATC_VM_APERTURE1_CNTL2 0x331Cu +#define ATC_VM_APERTURE1_HIGH_ADDR 0x330Cu +#define ATC_VM_APERTURE1_LOW_ADDR 0x3304u + #endif diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c index 5c8b358..924b1b7 100644 --- a/drivers/gpu/drm/radeon/evergreen_cs.c +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -35,7 +35,7 @@ #define MIN(a,b) (((a)<(b))?(a):(b)) int r600_dma_cs_next_reloc(struct radeon_cs_parser *p, - struct radeon_cs_reloc **cs_reloc); + struct radeon_bo_list **cs_reloc); struct evergreen_cs_track { u32 group_size; u32 nbanks; @@ -1094,7 +1094,7 @@ static int evergreen_cs_parse_packet0(struct radeon_cs_parser *p, static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) { struct evergreen_cs_track *track = (struct evergreen_cs_track *)p->track; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; u32 last_reg; u32 m, i, tmp, *ib; int r; @@ -1792,7 +1792,7 @@ static bool evergreen_is_safe_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) static int evergreen_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct evergreen_cs_track *track; volatile u32 *ib; unsigned idx; @@ -2661,7 +2661,7 @@ int evergreen_cs_parse(struct radeon_cs_parser *p) p->track = NULL; return r; } - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->chunk_ib->length_dw); #if 0 for (r = 0; r < p->ib.length_dw; r++) { printk(KERN_INFO "%05d 0x%08X\n", r, p->ib.ptr[r]); @@ -2684,8 +2684,8 @@ int evergreen_cs_parse(struct radeon_cs_parser *p) **/ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) { - struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; - struct radeon_cs_reloc *src_reloc, *dst_reloc, *dst2_reloc; + struct radeon_cs_chunk *ib_chunk = p->chunk_ib; + struct radeon_bo_list *src_reloc, *dst_reloc, *dst2_reloc; u32 header, cmd, count, sub_cmd; volatile u32 *ib = p->ib.ptr; u32 idx; @@ -3100,7 +3100,7 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) DRM_ERROR("Unknown packet type %d at %d !\n", cmd, idx); return -EINVAL; } - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->chunk_ib->length_dw); #if 0 for (r = 0; r < p->ib->length_dw; r++) { printk(KERN_INFO "%05d 0x%08X\n", r, p->ib.ptr[r]); diff --git a/drivers/gpu/drm/radeon/evergreen_dma.c b/drivers/gpu/drm/radeon/evergreen_dma.c index 66bcfad..96535aa 100644 --- a/drivers/gpu/drm/radeon/evergreen_dma.c +++ b/drivers/gpu/drm/radeon/evergreen_dma.c @@ -110,31 +110,27 @@ struct radeon_fence *evergreen_copy_dma(struct radeon_device *rdev, unsigned num_gpu_pages, struct reservation_object *resv) { - struct radeon_semaphore *sem = NULL; struct radeon_fence *fence; + struct radeon_sync sync; int ring_index = rdev->asic->copy.dma_ring_index; struct radeon_ring *ring = &rdev->ring[ring_index]; u32 size_in_dw, cur_size_in_dw; int i, num_loops; int r = 0; - r = radeon_semaphore_create(rdev, &sem); - if (r) { - DRM_ERROR("radeon: moving bo (%d).\n", r); - return ERR_PTR(r); - } + radeon_sync_create(&sync); size_in_dw = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT) / 4; num_loops = DIV_ROUND_UP(size_in_dw, 0xfffff); r = radeon_ring_lock(rdev, ring, num_loops * 5 + 11); if (r) { DRM_ERROR("radeon: moving bo (%d).\n", r); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } - radeon_semaphore_sync_resv(rdev, sem, resv, false); - radeon_semaphore_sync_rings(rdev, sem, ring->idx); + radeon_sync_resv(rdev, &sync, resv, false); + radeon_sync_rings(rdev, &sync, ring->idx); for (i = 0; i < num_loops; i++) { cur_size_in_dw = size_in_dw; @@ -153,12 +149,12 @@ struct radeon_fence *evergreen_copy_dma(struct radeon_device *rdev, r = radeon_fence_emit(rdev, &fence, ring->idx); if (r) { radeon_ring_unlock_undo(rdev, ring); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } radeon_ring_unlock_commit(rdev, ring, false); - radeon_semaphore_free(rdev, &sem, fence); + radeon_sync_free(rdev, &sync, fence); return fence; } diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c index 3faee58..360de9f 100644 --- a/drivers/gpu/drm/radeon/ni.c +++ b/drivers/gpu/drm/radeon/ni.c @@ -1373,6 +1373,7 @@ void cayman_fence_ring_emit(struct radeon_device *rdev, void cayman_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) { struct radeon_ring *ring = &rdev->ring[ib->ring]; + unsigned vm_id = ib->vm ? ib->vm->ids[ib->ring].id : 0; u32 cp_coher_cntl = PACKET3_FULL_CACHE_ENA | PACKET3_TC_ACTION_ENA | PACKET3_SH_ACTION_ENA; @@ -1395,15 +1396,14 @@ void cayman_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) #endif (ib->gpu_addr & 0xFFFFFFFC)); radeon_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFF); - radeon_ring_write(ring, ib->length_dw | - (ib->vm ? (ib->vm->id << 24) : 0)); + radeon_ring_write(ring, ib->length_dw | (vm_id << 24)); /* flush read cache over gart for this vmid */ radeon_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3)); radeon_ring_write(ring, PACKET3_ENGINE_ME | cp_coher_cntl); radeon_ring_write(ring, 0xFFFFFFFF); radeon_ring_write(ring, 0); - radeon_ring_write(ring, ((ib->vm ? ib->vm->id : 0) << 24) | 10); /* poll interval */ + radeon_ring_write(ring, (vm_id << 24) | 10); /* poll interval */ } static void cayman_cp_enable(struct radeon_device *rdev, bool enable) @@ -2502,15 +2502,11 @@ void cayman_vm_decode_fault(struct radeon_device *rdev, * Update the page table base and flush the VM TLB * using the CP (cayman-si). */ -void cayman_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) +void cayman_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr) { - struct radeon_ring *ring = &rdev->ring[ridx]; - - if (vm == NULL) - return; - - radeon_ring_write(ring, PACKET0(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm->id << 2), 0)); - radeon_ring_write(ring, vm->pd_gpu_addr >> 12); + radeon_ring_write(ring, PACKET0(VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm_id << 2), 0)); + radeon_ring_write(ring, pd_addr >> 12); /* flush hdp cache */ radeon_ring_write(ring, PACKET0(HDP_MEM_COHERENCY_FLUSH_CNTL, 0)); @@ -2518,7 +2514,7 @@ void cayman_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) /* bits 0-7 are the VM contexts0-7 */ radeon_ring_write(ring, PACKET0(VM_INVALIDATE_REQUEST, 0)); - radeon_ring_write(ring, 1 << vm->id); + radeon_ring_write(ring, 1 << vm_id); /* sync PFP to ME, otherwise we might get invalid PFP reads */ radeon_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0)); diff --git a/drivers/gpu/drm/radeon/ni_dma.c b/drivers/gpu/drm/radeon/ni_dma.c index f26f0a9..50f8861 100644 --- a/drivers/gpu/drm/radeon/ni_dma.c +++ b/drivers/gpu/drm/radeon/ni_dma.c @@ -123,6 +123,7 @@ void cayman_dma_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) { struct radeon_ring *ring = &rdev->ring[ib->ring]; + unsigned vm_id = ib->vm ? ib->vm->ids[ib->ring].id : 0; if (rdev->wb.enabled) { u32 next_rptr = ring->wptr + 4; @@ -140,7 +141,7 @@ void cayman_dma_ring_ib_execute(struct radeon_device *rdev, */ while ((ring->wptr & 7) != 5) radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0)); - radeon_ring_write(ring, DMA_IB_PACKET(DMA_PACKET_INDIRECT_BUFFER, ib->vm ? ib->vm->id : 0, 0)); + radeon_ring_write(ring, DMA_IB_PACKET(DMA_PACKET_INDIRECT_BUFFER, vm_id, 0)); radeon_ring_write(ring, (ib->gpu_addr & 0xFFFFFFE0)); radeon_ring_write(ring, (ib->length_dw << 12) | (upper_32_bits(ib->gpu_addr) & 0xFF)); @@ -446,16 +447,12 @@ void cayman_dma_vm_pad_ib(struct radeon_ib *ib) ib->ptr[ib->length_dw++] = DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0); } -void cayman_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) +void cayman_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr) { - struct radeon_ring *ring = &rdev->ring[ridx]; - - if (vm == NULL) - return; - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0)); - radeon_ring_write(ring, (0xf << 16) | ((VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm->id << 2)) >> 2)); - radeon_ring_write(ring, vm->pd_gpu_addr >> 12); + radeon_ring_write(ring, (0xf << 16) | ((VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm_id << 2)) >> 2)); + radeon_ring_write(ring, pd_addr >> 12); /* flush hdp cache */ radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0)); @@ -465,6 +462,6 @@ void cayman_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm /* bits 0-7 are the VM contexts0-7 */ radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0)); radeon_ring_write(ring, (0xf << 16) | (VM_INVALIDATE_REQUEST >> 2)); - radeon_ring_write(ring, 1 << vm->id); + radeon_ring_write(ring, 1 << vm_id); } diff --git a/drivers/gpu/drm/radeon/ppsmc.h b/drivers/gpu/drm/radeon/ppsmc.h index 5670b82..7e5724a 100644 --- a/drivers/gpu/drm/radeon/ppsmc.h +++ b/drivers/gpu/drm/radeon/ppsmc.h @@ -56,6 +56,14 @@ #define PPSMC_STATEFLAG_DEEPSLEEP_THROTTLE 0x20 #define PPSMC_STATEFLAG_DEEPSLEEP_BYPASS 0x40 +#define FDO_MODE_HARDWARE 0 +#define FDO_MODE_PIECE_WISE_LINEAR 1 + +enum FAN_CONTROL { + FAN_CONTROL_FUZZY, + FAN_CONTROL_TABLE +}; + #define PPSMC_Result_OK ((uint8_t)0x01) #define PPSMC_Result_Failed ((uint8_t)0xFF) @@ -79,6 +87,8 @@ typedef uint8_t PPSMC_Result; #define PPSMC_MSG_DisableCac ((uint8_t)0x54) #define PPSMC_TDPClampingActive ((uint8_t)0x59) #define PPSMC_TDPClampingInactive ((uint8_t)0x5A) +#define PPSMC_StartFanControl ((uint8_t)0x5B) +#define PPSMC_StopFanControl ((uint8_t)0x5C) #define PPSMC_MSG_NoDisplay ((uint8_t)0x5D) #define PPSMC_MSG_HasDisplay ((uint8_t)0x5E) #define PPSMC_MSG_UVDPowerOFF ((uint8_t)0x60) @@ -106,6 +116,7 @@ typedef uint8_t PPSMC_Result; #define PPSMC_MSG_SAMUDPM_SetEnabledMask ((uint16_t) 0x130) #define PPSMC_MSG_MCLKDPM_ForceState ((uint16_t) 0x131) #define PPSMC_MSG_MCLKDPM_NoForcedLevel ((uint16_t) 0x132) +#define PPSMC_MSG_Thermal_Cntl_Disable ((uint16_t) 0x133) #define PPSMC_MSG_Voltage_Cntl_Disable ((uint16_t) 0x135) #define PPSMC_MSG_PCIeDPM_Enable ((uint16_t) 0x136) #define PPSMC_MSG_PCIeDPM_Disable ((uint16_t) 0x13d) @@ -149,6 +160,10 @@ typedef uint8_t PPSMC_Result; #define PPSMC_MSG_MASTER_DeepSleep_ON ((uint16_t) 0x18F) #define PPSMC_MSG_MASTER_DeepSleep_OFF ((uint16_t) 0x190) #define PPSMC_MSG_Remove_DC_Clamp ((uint16_t) 0x191) +#define PPSMC_MSG_SetFanPwmMax ((uint16_t) 0x19A) + +#define PPSMC_MSG_ENABLE_THERMAL_DPM ((uint16_t) 0x19C) +#define PPSMC_MSG_DISABLE_THERMAL_DPM ((uint16_t) 0x19D) #define PPSMC_MSG_API_GetSclkFrequency ((uint16_t) 0x200) #define PPSMC_MSG_API_GetMclkFrequency ((uint16_t) 0x201) @@ -157,10 +172,11 @@ typedef uint8_t PPSMC_Result; #define PPSMC_MSG_DPM_Config ((uint32_t) 0x102) #define PPSMC_MSG_DPM_ForceState ((uint32_t) 0x104) #define PPSMC_MSG_PG_SIMD_Config ((uint32_t) 0x108) -#define PPSMC_MSG_DPM_N_LevelsDisabled ((uint32_t) 0x112) +#define PPSMC_MSG_Thermal_Cntl_Enable ((uint32_t) 0x10a) #define PPSMC_MSG_Voltage_Cntl_Enable ((uint32_t) 0x109) #define PPSMC_MSG_VCEPowerOFF ((uint32_t) 0x10e) #define PPSMC_MSG_VCEPowerON ((uint32_t) 0x10f) +#define PPSMC_MSG_DPM_N_LevelsDisabled ((uint32_t) 0x112) #define PPSMC_MSG_DCE_RemoveVoltageAdjustment ((uint32_t) 0x11d) #define PPSMC_MSG_DCE_AllowVoltageAdjustment ((uint32_t) 0x11e) #define PPSMC_MSG_EnableBAPM ((uint32_t) 0x120) diff --git a/drivers/gpu/drm/radeon/pptable.h b/drivers/gpu/drm/radeon/pptable.h index 2d53299..4c2eec4 100644 --- a/drivers/gpu/drm/radeon/pptable.h +++ b/drivers/gpu/drm/radeon/pptable.h @@ -96,6 +96,14 @@ typedef struct _ATOM_PPLIB_FANTABLE2 USHORT usTMax; // The max temperature } ATOM_PPLIB_FANTABLE2; +typedef struct _ATOM_PPLIB_FANTABLE3 +{ + ATOM_PPLIB_FANTABLE2 basicTable2; + UCHAR ucFanControlMode; + USHORT usFanPWMMax; + USHORT usFanOutputSensitivity; +} ATOM_PPLIB_FANTABLE3; + typedef struct _ATOM_PPLIB_EXTENDEDHEADER { USHORT usSize; diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c index b53b31a..74f06d5 100644 --- a/drivers/gpu/drm/radeon/r100.c +++ b/drivers/gpu/drm/radeon/r100.c @@ -1254,7 +1254,7 @@ int r100_reloc_pitch_offset(struct radeon_cs_parser *p, int r; u32 tile_flags = 0; u32 tmp; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; u32 value; r = radeon_cs_packet_next_reloc(p, &reloc, 0); @@ -1293,7 +1293,7 @@ int r100_packet3_load_vbpntr(struct radeon_cs_parser *p, int idx) { unsigned c, i; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; int r = 0; volatile uint32_t *ib; @@ -1542,7 +1542,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; volatile uint32_t *ib; uint32_t tmp; @@ -1901,7 +1901,7 @@ int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, static int r100_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; unsigned idx; volatile uint32_t *ib; @@ -2061,7 +2061,7 @@ int r100_cs_parse(struct radeon_cs_parser *p) } if (r) return r; - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->chunk_ib->length_dw); return 0; } diff --git a/drivers/gpu/drm/radeon/r200.c b/drivers/gpu/drm/radeon/r200.c index 732d493..c70e6d5 100644 --- a/drivers/gpu/drm/radeon/r200.c +++ b/drivers/gpu/drm/radeon/r200.c @@ -146,7 +146,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; volatile uint32_t *ib; uint32_t tmp; diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c index 1bc4704..064ad55 100644 --- a/drivers/gpu/drm/radeon/r300.c +++ b/drivers/gpu/drm/radeon/r300.c @@ -598,7 +598,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx, unsigned reg) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; volatile uint32_t *ib; uint32_t tmp, tile_flags = 0; @@ -1142,7 +1142,7 @@ fail: static int r300_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r100_cs_track *track; volatile uint32_t *ib; unsigned idx; @@ -1283,7 +1283,7 @@ int r300_cs_parse(struct radeon_cs_parser *p) if (r) { return r; } - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->chunk_ib->length_dw); return 0; } diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index 56b0292..ef5d606 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -2889,31 +2889,27 @@ struct radeon_fence *r600_copy_cpdma(struct radeon_device *rdev, unsigned num_gpu_pages, struct reservation_object *resv) { - struct radeon_semaphore *sem = NULL; struct radeon_fence *fence; + struct radeon_sync sync; int ring_index = rdev->asic->copy.blit_ring_index; struct radeon_ring *ring = &rdev->ring[ring_index]; u32 size_in_bytes, cur_size_in_bytes, tmp; int i, num_loops; int r = 0; - r = radeon_semaphore_create(rdev, &sem); - if (r) { - DRM_ERROR("radeon: moving bo (%d).\n", r); - return ERR_PTR(r); - } + radeon_sync_create(&sync); size_in_bytes = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT); num_loops = DIV_ROUND_UP(size_in_bytes, 0x1fffff); r = radeon_ring_lock(rdev, ring, num_loops * 6 + 24); if (r) { DRM_ERROR("radeon: moving bo (%d).\n", r); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } - radeon_semaphore_sync_resv(rdev, sem, resv, false); - radeon_semaphore_sync_rings(rdev, sem, ring->idx); + radeon_sync_resv(rdev, &sync, resv, false); + radeon_sync_rings(rdev, &sync, ring->idx); radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1)); radeon_ring_write(ring, (WAIT_UNTIL - PACKET3_SET_CONFIG_REG_OFFSET) >> 2); @@ -2942,12 +2938,12 @@ struct radeon_fence *r600_copy_cpdma(struct radeon_device *rdev, r = radeon_fence_emit(rdev, &fence, ring->idx); if (r) { radeon_ring_unlock_undo(rdev, ring); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } radeon_ring_unlock_commit(rdev, ring, false); - radeon_semaphore_free(rdev, &sem, fence); + radeon_sync_free(rdev, &sync, fence); return fence; } diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c index c47537a..acc1f99 100644 --- a/drivers/gpu/drm/radeon/r600_cs.c +++ b/drivers/gpu/drm/radeon/r600_cs.c @@ -969,7 +969,7 @@ static int r600_cs_parse_packet0(struct radeon_cs_parser *p, static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) { struct r600_cs_track *track = (struct r600_cs_track *)p->track; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; u32 m, i, tmp, *ib; int r; @@ -1626,7 +1626,7 @@ static bool r600_is_safe_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) static int r600_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; struct r600_cs_track *track; volatile u32 *ib; unsigned idx; @@ -2316,7 +2316,7 @@ int r600_cs_parse(struct radeon_cs_parser *p) p->track = NULL; return r; } - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->chunk_ib->length_dw); #if 0 for (r = 0; r < p->ib.length_dw; r++) { printk(KERN_INFO "%05d 0x%08X\n", r, p->ib.ptr[r]); @@ -2351,10 +2351,10 @@ static void r600_cs_parser_fini(struct radeon_cs_parser *parser, int error) static int r600_cs_parser_relocs_legacy(struct radeon_cs_parser *p) { - if (p->chunk_relocs_idx == -1) { + if (p->chunk_relocs == NULL) { return 0; } - p->relocs = kzalloc(sizeof(struct radeon_cs_reloc), GFP_KERNEL); + p->relocs = kzalloc(sizeof(struct radeon_bo_list), GFP_KERNEL); if (p->relocs == NULL) { return -ENOMEM; } @@ -2398,7 +2398,7 @@ int r600_cs_legacy(struct drm_device *dev, void *data, struct drm_file *filp, /* Copy the packet into the IB, the parser will read from the * input memory (cached) and write to the IB (which can be * uncached). */ - ib_chunk = &parser.chunks[parser.chunk_ib_idx]; + ib_chunk = parser.chunk_ib; parser.ib.length_dw = ib_chunk->length_dw; *l = parser.ib.length_dw; if (copy_from_user(ib, ib_chunk->user_ptr, ib_chunk->length_dw * 4)) { @@ -2435,24 +2435,24 @@ void r600_cs_legacy_init(void) * GPU offset using the provided start. **/ int r600_dma_cs_next_reloc(struct radeon_cs_parser *p, - struct radeon_cs_reloc **cs_reloc) + struct radeon_bo_list **cs_reloc) { struct radeon_cs_chunk *relocs_chunk; unsigned idx; *cs_reloc = NULL; - if (p->chunk_relocs_idx == -1) { + if (p->chunk_relocs == NULL) { DRM_ERROR("No relocation chunk !\n"); return -EINVAL; } - relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + relocs_chunk = p->chunk_relocs; idx = p->dma_reloc_idx; if (idx >= p->nrelocs) { DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", idx, p->nrelocs); return -EINVAL; } - *cs_reloc = p->relocs_ptr[idx]; + *cs_reloc = &p->relocs[idx]; p->dma_reloc_idx++; return 0; } @@ -2472,8 +2472,8 @@ int r600_dma_cs_next_reloc(struct radeon_cs_parser *p, **/ int r600_dma_cs_parse(struct radeon_cs_parser *p) { - struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; - struct radeon_cs_reloc *src_reloc, *dst_reloc; + struct radeon_cs_chunk *ib_chunk = p->chunk_ib; + struct radeon_bo_list *src_reloc, *dst_reloc; u32 header, cmd, count, tiled; volatile u32 *ib = p->ib.ptr; u32 idx, idx_value; @@ -2619,7 +2619,7 @@ int r600_dma_cs_parse(struct radeon_cs_parser *p) DRM_ERROR("Unknown packet type %d at %d !\n", cmd, idx); return -EINVAL; } - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->chunk_ib->length_dw); #if 0 for (r = 0; r < p->ib->length_dw; r++) { printk(KERN_INFO "%05d 0x%08X\n", r, p->ib.ptr[r]); diff --git a/drivers/gpu/drm/radeon/r600_dma.c b/drivers/gpu/drm/radeon/r600_dma.c index cf0df45..d2dd29a 100644 --- a/drivers/gpu/drm/radeon/r600_dma.c +++ b/drivers/gpu/drm/radeon/r600_dma.c @@ -441,31 +441,27 @@ struct radeon_fence *r600_copy_dma(struct radeon_device *rdev, unsigned num_gpu_pages, struct reservation_object *resv) { - struct radeon_semaphore *sem = NULL; struct radeon_fence *fence; + struct radeon_sync sync; int ring_index = rdev->asic->copy.dma_ring_index; struct radeon_ring *ring = &rdev->ring[ring_index]; u32 size_in_dw, cur_size_in_dw; int i, num_loops; int r = 0; - r = radeon_semaphore_create(rdev, &sem); - if (r) { - DRM_ERROR("radeon: moving bo (%d).\n", r); - return ERR_PTR(r); - } + radeon_sync_create(&sync); size_in_dw = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT) / 4; num_loops = DIV_ROUND_UP(size_in_dw, 0xFFFE); r = radeon_ring_lock(rdev, ring, num_loops * 4 + 8); if (r) { DRM_ERROR("radeon: moving bo (%d).\n", r); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } - radeon_semaphore_sync_resv(rdev, sem, resv, false); - radeon_semaphore_sync_rings(rdev, sem, ring->idx); + radeon_sync_resv(rdev, &sync, resv, false); + radeon_sync_rings(rdev, &sync, ring->idx); for (i = 0; i < num_loops; i++) { cur_size_in_dw = size_in_dw; @@ -484,12 +480,12 @@ struct radeon_fence *r600_copy_dma(struct radeon_device *rdev, r = radeon_fence_emit(rdev, &fence, ring->idx); if (r) { radeon_ring_unlock_undo(rdev, ring); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } radeon_ring_unlock_commit(rdev, ring, false); - radeon_semaphore_free(rdev, &sem, fence); + radeon_sync_free(rdev, &sync, fence); return fence; } diff --git a/drivers/gpu/drm/radeon/r600_dpm.c b/drivers/gpu/drm/radeon/r600_dpm.c index b5c73df..843b65f 100644 --- a/drivers/gpu/drm/radeon/r600_dpm.c +++ b/drivers/gpu/drm/radeon/r600_dpm.c @@ -811,6 +811,7 @@ union power_info { union fan_info { struct _ATOM_PPLIB_FANTABLE fan; struct _ATOM_PPLIB_FANTABLE2 fan2; + struct _ATOM_PPLIB_FANTABLE3 fan3; }; static int r600_parse_clk_voltage_dep_table(struct radeon_clock_voltage_dependency_table *radeon_table, @@ -900,6 +901,14 @@ int r600_parse_extended_power_table(struct radeon_device *rdev) else rdev->pm.dpm.fan.t_max = 10900; rdev->pm.dpm.fan.cycle_delay = 100000; + if (fan_info->fan.ucFanTableFormat >= 3) { + rdev->pm.dpm.fan.control_mode = fan_info->fan3.ucFanControlMode; + rdev->pm.dpm.fan.default_max_fan_pwm = + le16_to_cpu(fan_info->fan3.usFanPWMMax); + rdev->pm.dpm.fan.default_fan_output_sensitivity = 4836; + rdev->pm.dpm.fan.fan_output_sensitivity = + le16_to_cpu(fan_info->fan3.usFanOutputSensitivity); + } rdev->pm.dpm.fan.ucode_fan_control = true; } } diff --git a/drivers/gpu/drm/radeon/r600_dpm.h b/drivers/gpu/drm/radeon/r600_dpm.h index 46b9d2a..bd499d7 100644 --- a/drivers/gpu/drm/radeon/r600_dpm.h +++ b/drivers/gpu/drm/radeon/r600_dpm.h @@ -96,6 +96,9 @@ #define R600_TEMP_RANGE_MIN (90 * 1000) #define R600_TEMP_RANGE_MAX (120 * 1000) +#define FDO_PWM_MODE_STATIC 1 +#define FDO_PWM_MODE_STATIC_RPM 5 + enum r600_power_level { R600_POWER_LEVEL_LOW = 0, R600_POWER_LEVEL_MEDIUM = 1, diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index a9717b3..54529b8 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -150,9 +150,6 @@ extern int radeon_backlight; /* number of hw syncs before falling back on blocking */ #define RADEON_NUM_SYNCS 4 -/* number of hw syncs before falling back on blocking */ -#define RADEON_NUM_SYNCS 4 - /* hardcode those limit for now */ #define RADEON_VA_IB_OFFSET (1 << 20) #define RADEON_VA_RESERVED_SIZE (8 << 20) @@ -363,14 +360,15 @@ struct radeon_fence_driver { }; struct radeon_fence { - struct fence base; + struct fence base; - struct radeon_device *rdev; - uint64_t seq; + struct radeon_device *rdev; + uint64_t seq; /* RB, DMA, etc. */ - unsigned ring; + unsigned ring; + bool is_vm_update; - wait_queue_t fence_wake; + wait_queue_t fence_wake; }; int radeon_fence_driver_start_ring(struct radeon_device *rdev, int ring); @@ -452,12 +450,22 @@ struct radeon_mman { #endif }; +struct radeon_bo_list { + struct radeon_bo *robj; + struct ttm_validate_buffer tv; + uint64_t gpu_offset; + unsigned prefered_domains; + unsigned allowed_domains; + uint32_t tiling_flags; +}; + /* bo virtual address in a specific vm */ struct radeon_bo_va { /* protected by bo being reserved */ struct list_head bo_list; uint32_t flags; uint64_t addr; + struct radeon_fence *last_pt_update; unsigned ref_count; /* protected by vm mutex */ @@ -474,7 +482,7 @@ struct radeon_bo { struct list_head list; /* Protected by tbo.reserved */ u32 initial_domain; - struct ttm_place placements[3]; + struct ttm_place placements[4]; struct ttm_placement placement; struct ttm_buffer_object tbo; struct ttm_bo_kmap_obj kmap; @@ -576,10 +584,9 @@ int radeon_mode_dumb_mmap(struct drm_file *filp, * Semaphores. */ struct radeon_semaphore { - struct radeon_sa_bo *sa_bo; - signed waiters; - uint64_t gpu_addr; - struct radeon_fence *sync_to[RADEON_NUM_RINGS]; + struct radeon_sa_bo *sa_bo; + signed waiters; + uint64_t gpu_addr; }; int radeon_semaphore_create(struct radeon_device *rdev, @@ -588,20 +595,33 @@ bool radeon_semaphore_emit_signal(struct radeon_device *rdev, int ring, struct radeon_semaphore *semaphore); bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ring, struct radeon_semaphore *semaphore); -void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, - struct radeon_fence *fence); -int radeon_semaphore_sync_resv(struct radeon_device *rdev, - struct radeon_semaphore *semaphore, - struct reservation_object *resv, - bool shared); -int radeon_semaphore_sync_rings(struct radeon_device *rdev, - struct radeon_semaphore *semaphore, - int waiting_ring); void radeon_semaphore_free(struct radeon_device *rdev, struct radeon_semaphore **semaphore, struct radeon_fence *fence); /* + * Synchronization + */ +struct radeon_sync { + struct radeon_semaphore *semaphores[RADEON_NUM_SYNCS]; + struct radeon_fence *sync_to[RADEON_NUM_RINGS]; + struct radeon_fence *last_vm_update; +}; + +void radeon_sync_create(struct radeon_sync *sync); +void radeon_sync_fence(struct radeon_sync *sync, + struct radeon_fence *fence); +int radeon_sync_resv(struct radeon_device *rdev, + struct radeon_sync *sync, + struct reservation_object *resv, + bool shared); +int radeon_sync_rings(struct radeon_device *rdev, + struct radeon_sync *sync, + int waiting_ring); +void radeon_sync_free(struct radeon_device *rdev, struct radeon_sync *sync, + struct radeon_fence *fence); + +/* * GART structures, functions & helpers */ struct radeon_mc; @@ -701,6 +721,10 @@ struct radeon_doorbell { int radeon_doorbell_get(struct radeon_device *rdev, u32 *page); void radeon_doorbell_free(struct radeon_device *rdev, u32 doorbell); +void radeon_doorbell_get_kfd_info(struct radeon_device *rdev, + phys_addr_t *aperture_base, + size_t *aperture_size, + size_t *start_offset); /* * IRQS. @@ -814,7 +838,7 @@ struct radeon_ib { struct radeon_fence *fence; struct radeon_vm *vm; bool is_const_ib; - struct radeon_semaphore *semaphore; + struct radeon_sync sync; }; struct radeon_ring { @@ -891,33 +915,40 @@ struct radeon_vm_pt { uint64_t addr; }; +struct radeon_vm_id { + unsigned id; + uint64_t pd_gpu_addr; + /* last flushed PD/PT update */ + struct radeon_fence *flushed_updates; + /* last use of vmid */ + struct radeon_fence *last_id_use; +}; + struct radeon_vm { - struct rb_root va; - unsigned id; + struct mutex mutex; + + struct rb_root va; + + /* protecting invalidated and freed */ + spinlock_t status_lock; /* BOs moved, but not yet updated in the PT */ - struct list_head invalidated; + struct list_head invalidated; /* BOs freed, but not yet updated in the PT */ - struct list_head freed; + struct list_head freed; /* contains the page directory */ - struct radeon_bo *page_directory; - uint64_t pd_gpu_addr; - unsigned max_pde_used; + struct radeon_bo *page_directory; + unsigned max_pde_used; /* array of page tables, one for each page directory entry */ - struct radeon_vm_pt *page_tables; + struct radeon_vm_pt *page_tables; - struct radeon_bo_va *ib_bo_va; + struct radeon_bo_va *ib_bo_va; - struct mutex mutex; - /* last fence for cs using this vm */ - struct radeon_fence *fence; - /* last flush or NULL if we still need to flush */ - struct radeon_fence *last_flush; - /* last use of vmid */ - struct radeon_fence *last_id_use; + /* for id and flush management per ring */ + struct radeon_vm_id ids[RADEON_NUM_RINGS]; }; struct radeon_vm_manager { @@ -1025,19 +1056,7 @@ void cayman_dma_fini(struct radeon_device *rdev); /* * CS. */ -struct radeon_cs_reloc { - struct drm_gem_object *gobj; - struct radeon_bo *robj; - struct ttm_validate_buffer tv; - uint64_t gpu_offset; - unsigned prefered_domains; - unsigned allowed_domains; - uint32_t tiling_flags; - uint32_t handle; -}; - struct radeon_cs_chunk { - uint32_t chunk_id; uint32_t length_dw; uint32_t *kdata; void __user *user_ptr; @@ -1055,16 +1074,15 @@ struct radeon_cs_parser { unsigned idx; /* relocations */ unsigned nrelocs; - struct radeon_cs_reloc *relocs; - struct radeon_cs_reloc **relocs_ptr; - struct radeon_cs_reloc *vm_bos; + struct radeon_bo_list *relocs; + struct radeon_bo_list *vm_bos; struct list_head validated; unsigned dma_reloc_idx; /* indices of various chunks */ - int chunk_ib_idx; - int chunk_relocs_idx; - int chunk_flags_idx; - int chunk_const_ib_idx; + struct radeon_cs_chunk *chunk_ib; + struct radeon_cs_chunk *chunk_relocs; + struct radeon_cs_chunk *chunk_flags; + struct radeon_cs_chunk *chunk_const_ib; struct radeon_ib ib; struct radeon_ib const_ib; void *track; @@ -1078,7 +1096,7 @@ struct radeon_cs_parser { static inline u32 radeon_get_ib_value(struct radeon_cs_parser *p, int idx) { - struct radeon_cs_chunk *ibc = &p->chunks[p->chunk_ib_idx]; + struct radeon_cs_chunk *ibc = p->chunk_ib; if (ibc->kdata) return ibc->kdata[idx]; @@ -1490,6 +1508,10 @@ struct radeon_dpm_fan { u8 t_hyst; u32 cycle_delay; u16 t_max; + u8 control_mode; + u16 default_max_fan_pwm; + u16 default_fan_output_sensitivity; + u16 fan_output_sensitivity; bool ucode_fan_control; }; @@ -1623,6 +1645,11 @@ struct radeon_pm { /* internal thermal controller on rv6xx+ */ enum radeon_int_thermal_type int_thermal_type; struct device *int_hwmon_dev; + /* fan control parameters */ + bool no_fan; + u8 fan_pulses_per_revolution; + u8 fan_min_rpm; + u8 fan_max_rpm; /* dpm */ bool dpm_enabled; struct radeon_dpm dpm; @@ -1785,7 +1812,8 @@ struct radeon_asic_ring { void (*hdp_flush)(struct radeon_device *rdev, struct radeon_ring *ring); bool (*emit_semaphore)(struct radeon_device *rdev, struct radeon_ring *cp, struct radeon_semaphore *semaphore, bool emit_wait); - void (*vm_flush)(struct radeon_device *rdev, int ridx, struct radeon_vm *vm); + void (*vm_flush)(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr); /* testing functions */ int (*ring_test)(struct radeon_device *rdev, struct radeon_ring *cp); @@ -2388,6 +2416,8 @@ struct radeon_device { struct radeon_atcs atcs; /* srbm instance registers */ struct mutex srbm_mutex; + /* GRBM index mutex. Protects concurrents access to GRBM index */ + struct mutex grbm_idx_mutex; /* clock, powergating flags */ u32 cg_flags; u32 pg_flags; @@ -2400,6 +2430,10 @@ struct radeon_device { u64 vram_pin_size; u64 gart_pin_size; + /* amdkfd interface */ + struct kfd_dev *kfd; + struct radeon_sa_manager kfd_bo; + struct mutex mn_lock; DECLARE_HASHTABLE(mn_hash, 7); }; @@ -2831,7 +2865,7 @@ static inline void radeon_ring_write(struct radeon_ring *ring, uint32_t v) #define radeon_ring_ib_execute(rdev, r, ib) (rdev)->asic->ring[(r)]->ib_execute((rdev), (ib)) #define radeon_ring_ib_parse(rdev, r, ib) (rdev)->asic->ring[(r)]->ib_parse((rdev), (ib)) #define radeon_ring_is_lockup(rdev, r, cp) (rdev)->asic->ring[(r)]->is_lockup((rdev), (cp)) -#define radeon_ring_vm_flush(rdev, r, vm) (rdev)->asic->ring[(r)]->vm_flush((rdev), (r), (vm)) +#define radeon_ring_vm_flush(rdev, r, vm_id, pd_addr) (rdev)->asic->ring[(r)->idx]->vm_flush((rdev), (r), (vm_id), (pd_addr)) #define radeon_ring_get_rptr(rdev, r) (rdev)->asic->ring[(r)->idx]->get_rptr((rdev), (r)) #define radeon_ring_get_wptr(rdev, r) (rdev)->asic->ring[(r)->idx]->get_wptr((rdev), (r)) #define radeon_ring_set_wptr(rdev, r) (rdev)->asic->ring[(r)->idx]->set_wptr((rdev), (r)) @@ -2940,14 +2974,14 @@ int radeon_vm_manager_init(struct radeon_device *rdev); void radeon_vm_manager_fini(struct radeon_device *rdev); int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm); void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm); -struct radeon_cs_reloc *radeon_vm_get_bos(struct radeon_device *rdev, +struct radeon_bo_list *radeon_vm_get_bos(struct radeon_device *rdev, struct radeon_vm *vm, struct list_head *head); struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, struct radeon_vm *vm, int ring); void radeon_vm_flush(struct radeon_device *rdev, struct radeon_vm *vm, - int ring); + int ring, struct radeon_fence *fence); void radeon_vm_fence(struct radeon_device *rdev, struct radeon_vm *vm, struct radeon_fence *fence); @@ -3054,7 +3088,7 @@ bool radeon_cs_packet_next_is_pkt3_nop(struct radeon_cs_parser *p); void radeon_cs_dump_packet(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt); int radeon_cs_packet_next_reloc(struct radeon_cs_parser *p, - struct radeon_cs_reloc **cs_reloc, + struct radeon_bo_list **cs_reloc, int nomm); int r600_cs_common_vline_parse(struct radeon_cs_parser *p, uint32_t *vline_start_end, diff --git a/drivers/gpu/drm/radeon/radeon_asic.h b/drivers/gpu/drm/radeon/radeon_asic.h index d8ace5b..2a45d54 100644 --- a/drivers/gpu/drm/radeon/radeon_asic.h +++ b/drivers/gpu/drm/radeon/radeon_asic.h @@ -599,7 +599,8 @@ int cayman_asic_reset(struct radeon_device *rdev); void cayman_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib); int cayman_vm_init(struct radeon_device *rdev); void cayman_vm_fini(struct radeon_device *rdev); -void cayman_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm); +void cayman_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr); uint32_t cayman_vm_page_flags(struct radeon_device *rdev, uint32_t flags); int evergreen_ib_parse(struct radeon_device *rdev, struct radeon_ib *ib); int evergreen_dma_ib_parse(struct radeon_device *rdev, struct radeon_ib *ib); @@ -624,7 +625,8 @@ void cayman_dma_vm_set_pages(struct radeon_device *rdev, uint32_t incr, uint32_t flags); void cayman_dma_vm_pad_ib(struct radeon_ib *ib); -void cayman_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm); +void cayman_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr); u32 cayman_gfx_get_rptr(struct radeon_device *rdev, struct radeon_ring *ring); @@ -699,7 +701,8 @@ int si_irq_set(struct radeon_device *rdev); int si_irq_process(struct radeon_device *rdev); int si_vm_init(struct radeon_device *rdev); void si_vm_fini(struct radeon_device *rdev); -void si_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm); +void si_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr); int si_ib_parse(struct radeon_device *rdev, struct radeon_ib *ib); struct radeon_fence *si_copy_dma(struct radeon_device *rdev, uint64_t src_offset, uint64_t dst_offset, @@ -721,7 +724,8 @@ void si_dma_vm_set_pages(struct radeon_device *rdev, uint64_t addr, unsigned count, uint32_t incr, uint32_t flags); -void si_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm); +void si_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr); u32 si_get_xclk(struct radeon_device *rdev); uint64_t si_get_gpu_clock_counter(struct radeon_device *rdev); int si_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk); @@ -793,7 +797,8 @@ int cik_irq_set(struct radeon_device *rdev); int cik_irq_process(struct radeon_device *rdev); int cik_vm_init(struct radeon_device *rdev); void cik_vm_fini(struct radeon_device *rdev); -void cik_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm); +void cik_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr); void cik_sdma_vm_copy_pages(struct radeon_device *rdev, struct radeon_ib *ib, @@ -811,7 +816,8 @@ void cik_sdma_vm_set_pages(struct radeon_device *rdev, uint32_t incr, uint32_t flags); void cik_sdma_vm_pad_ib(struct radeon_ib *ib); -void cik_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm); +void cik_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr); int cik_ib_parse(struct radeon_device *rdev, struct radeon_ib *ib); u32 cik_gfx_get_rptr(struct radeon_device *rdev, struct radeon_ring *ring); diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c index df69b92..dbc94f3 100644 --- a/drivers/gpu/drm/radeon/radeon_atombios.c +++ b/drivers/gpu/drm/radeon/radeon_atombios.c @@ -196,8 +196,8 @@ void radeon_atombios_i2c_init(struct radeon_device *rdev) } } -static struct radeon_gpio_rec radeon_lookup_gpio(struct radeon_device *rdev, - u8 id) +struct radeon_gpio_rec radeon_atombios_lookup_gpio(struct radeon_device *rdev, + u8 id) { struct atom_context *ctx = rdev->mode_info.atom_context; struct radeon_gpio_rec gpio; @@ -221,6 +221,7 @@ static struct radeon_gpio_rec radeon_lookup_gpio(struct radeon_device *rdev, if (id == pin->ucGPIO_ID) { gpio.id = pin->ucGPIO_ID; gpio.reg = le16_to_cpu(pin->usGpioPin_AIndex) * 4; + gpio.shift = pin->ucGpioPinBitShift; gpio.mask = (1 << pin->ucGpioPinBitShift); gpio.valid = true; break; @@ -801,7 +802,7 @@ bool radeon_get_atom_connector_info_from_object_table(struct drm_device *dev) hpd_record = (ATOM_HPD_INT_RECORD *) record; - gpio = radeon_lookup_gpio(rdev, + gpio = radeon_atombios_lookup_gpio(rdev, hpd_record->ucHPDIntGPIOID); hpd = radeon_atom_get_hpd_info_from_gpio(rdev, &gpio); hpd.plugged_state = hpd_record->ucPlugged_PinState; @@ -2128,7 +2129,7 @@ static int radeon_atombios_parse_power_table_1_3(struct radeon_device *rdev) rdev->pm.power_state[state_index].clock_info[0].voltage.type = VOLTAGE_GPIO; rdev->pm.power_state[state_index].clock_info[0].voltage.gpio = - radeon_lookup_gpio(rdev, + radeon_atombios_lookup_gpio(rdev, power_info->info.asPowerPlayInfo[i].ucVoltageDropIndex); if (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH) rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = @@ -2164,7 +2165,7 @@ static int radeon_atombios_parse_power_table_1_3(struct radeon_device *rdev) rdev->pm.power_state[state_index].clock_info[0].voltage.type = VOLTAGE_GPIO; rdev->pm.power_state[state_index].clock_info[0].voltage.gpio = - radeon_lookup_gpio(rdev, + radeon_atombios_lookup_gpio(rdev, power_info->info_2.asPowerPlayInfo[i].ucVoltageDropIndex); if (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH) rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = @@ -2200,7 +2201,7 @@ static int radeon_atombios_parse_power_table_1_3(struct radeon_device *rdev) rdev->pm.power_state[state_index].clock_info[0].voltage.type = VOLTAGE_GPIO; rdev->pm.power_state[state_index].clock_info[0].voltage.gpio = - radeon_lookup_gpio(rdev, + radeon_atombios_lookup_gpio(rdev, power_info->info_3.asPowerPlayInfo[i].ucVoltageDropIndex); if (misc & ATOM_PM_MISCINFO_VOLTAGE_DROP_ACTIVE_HIGH) rdev->pm.power_state[state_index].clock_info[0].voltage.active_high = @@ -2248,6 +2249,14 @@ static void radeon_atombios_add_pplib_thermal_controller(struct radeon_device *r /* add the i2c bus for thermal/fan chip */ if (controller->ucType > 0) { + if (controller->ucFanParameters & ATOM_PP_FANPARAMETERS_NOFAN) + rdev->pm.no_fan = true; + rdev->pm.fan_pulses_per_revolution = + controller->ucFanParameters & ATOM_PP_FANPARAMETERS_TACHOMETER_PULSES_PER_REVOLUTION_MASK; + if (rdev->pm.fan_pulses_per_revolution) { + rdev->pm.fan_min_rpm = controller->ucFanMinRPM; + rdev->pm.fan_max_rpm = controller->ucFanMaxRPM; + } if (controller->ucType == ATOM_PP_THERMALCONTROLLER_RV6xx) { DRM_INFO("Internal thermal controller %s fan control\n", (controller->ucFanParameters & diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c index 6f377de..c830863 100644 --- a/drivers/gpu/drm/radeon/radeon_cs.c +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -77,22 +77,18 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p) struct drm_device *ddev = p->rdev->ddev; struct radeon_cs_chunk *chunk; struct radeon_cs_buckets buckets; - unsigned i, j; - bool duplicate, need_mmap_lock = false; + unsigned i; + bool need_mmap_lock = false; int r; - if (p->chunk_relocs_idx == -1) { + if (p->chunk_relocs == NULL) { return 0; } - chunk = &p->chunks[p->chunk_relocs_idx]; + chunk = p->chunk_relocs; p->dma_reloc_idx = 0; /* FIXME: we assume that each relocs use 4 dwords */ p->nrelocs = chunk->length_dw / 4; - p->relocs_ptr = kcalloc(p->nrelocs, sizeof(void *), GFP_KERNEL); - if (p->relocs_ptr == NULL) { - return -ENOMEM; - } - p->relocs = kcalloc(p->nrelocs, sizeof(struct radeon_cs_reloc), GFP_KERNEL); + p->relocs = kcalloc(p->nrelocs, sizeof(struct radeon_bo_list), GFP_KERNEL); if (p->relocs == NULL) { return -ENOMEM; } @@ -101,31 +97,17 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p) for (i = 0; i < p->nrelocs; i++) { struct drm_radeon_cs_reloc *r; + struct drm_gem_object *gobj; unsigned priority; - duplicate = false; r = (struct drm_radeon_cs_reloc *)&chunk->kdata[i*4]; - for (j = 0; j < i; j++) { - if (r->handle == p->relocs[j].handle) { - p->relocs_ptr[i] = &p->relocs[j]; - duplicate = true; - break; - } - } - if (duplicate) { - p->relocs[i].handle = 0; - continue; - } - - p->relocs[i].gobj = drm_gem_object_lookup(ddev, p->filp, - r->handle); - if (p->relocs[i].gobj == NULL) { + gobj = drm_gem_object_lookup(ddev, p->filp, r->handle); + if (gobj == NULL) { DRM_ERROR("gem object lookup failed 0x%x\n", r->handle); return -ENOENT; } - p->relocs_ptr[i] = &p->relocs[i]; - p->relocs[i].robj = gem_to_radeon_bo(p->relocs[i].gobj); + p->relocs[i].robj = gem_to_radeon_bo(gobj); /* The userspace buffer priorities are from 0 to 15. A higher * number means the buffer is more important. @@ -184,7 +166,6 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p) p->relocs[i].tv.bo = &p->relocs[i].robj->tbo; p->relocs[i].tv.shared = !r->write_domain; - p->relocs[i].handle = r->handle; radeon_cs_buckets_add(&buckets, &p->relocs[i].tv.head, priority); @@ -251,15 +232,15 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority static int radeon_cs_sync_rings(struct radeon_cs_parser *p) { - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; int r; list_for_each_entry(reloc, &p->validated, tv.head) { struct reservation_object *resv; resv = reloc->robj->tbo.resv; - r = radeon_semaphore_sync_resv(p->rdev, p->ib.semaphore, resv, - reloc->tv.shared); + r = radeon_sync_resv(p->rdev, &p->ib.sync, resv, + reloc->tv.shared); if (r) return r; } @@ -282,13 +263,11 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) INIT_LIST_HEAD(&p->validated); p->idx = 0; p->ib.sa_bo = NULL; - p->ib.semaphore = NULL; p->const_ib.sa_bo = NULL; - p->const_ib.semaphore = NULL; - p->chunk_ib_idx = -1; - p->chunk_relocs_idx = -1; - p->chunk_flags_idx = -1; - p->chunk_const_ib_idx = -1; + p->chunk_ib = NULL; + p->chunk_relocs = NULL; + p->chunk_flags = NULL; + p->chunk_const_ib = NULL; p->chunks_array = kcalloc(cs->num_chunks, sizeof(uint64_t), GFP_KERNEL); if (p->chunks_array == NULL) { return -ENOMEM; @@ -315,24 +294,23 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) return -EFAULT; } p->chunks[i].length_dw = user_chunk.length_dw; - p->chunks[i].chunk_id = user_chunk.chunk_id; - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_RELOCS) { - p->chunk_relocs_idx = i; + if (user_chunk.chunk_id == RADEON_CHUNK_ID_RELOCS) { + p->chunk_relocs = &p->chunks[i]; } - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_IB) { - p->chunk_ib_idx = i; + if (user_chunk.chunk_id == RADEON_CHUNK_ID_IB) { + p->chunk_ib = &p->chunks[i]; /* zero length IB isn't useful */ if (p->chunks[i].length_dw == 0) return -EINVAL; } - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_CONST_IB) { - p->chunk_const_ib_idx = i; + if (user_chunk.chunk_id == RADEON_CHUNK_ID_CONST_IB) { + p->chunk_const_ib = &p->chunks[i]; /* zero length CONST IB isn't useful */ if (p->chunks[i].length_dw == 0) return -EINVAL; } - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_FLAGS) { - p->chunk_flags_idx = i; + if (user_chunk.chunk_id == RADEON_CHUNK_ID_FLAGS) { + p->chunk_flags = &p->chunks[i]; /* zero length flags aren't useful */ if (p->chunks[i].length_dw == 0) return -EINVAL; @@ -341,10 +319,10 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) size = p->chunks[i].length_dw; cdata = (void __user *)(unsigned long)user_chunk.chunk_data; p->chunks[i].user_ptr = cdata; - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_CONST_IB) + if (user_chunk.chunk_id == RADEON_CHUNK_ID_CONST_IB) continue; - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_IB) { + if (user_chunk.chunk_id == RADEON_CHUNK_ID_IB) { if (!p->rdev || !(p->rdev->flags & RADEON_IS_AGP)) continue; } @@ -357,7 +335,7 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) if (copy_from_user(p->chunks[i].kdata, cdata, size)) { return -EFAULT; } - if (p->chunks[i].chunk_id == RADEON_CHUNK_ID_FLAGS) { + if (user_chunk.chunk_id == RADEON_CHUNK_ID_FLAGS) { p->cs_flags = p->chunks[i].kdata[0]; if (p->chunks[i].length_dw > 1) ring = p->chunks[i].kdata[1]; @@ -398,8 +376,8 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) static int cmp_size_smaller_first(void *priv, struct list_head *a, struct list_head *b) { - struct radeon_cs_reloc *la = list_entry(a, struct radeon_cs_reloc, tv.head); - struct radeon_cs_reloc *lb = list_entry(b, struct radeon_cs_reloc, tv.head); + struct radeon_bo_list *la = list_entry(a, struct radeon_bo_list, tv.head); + struct radeon_bo_list *lb = list_entry(b, struct radeon_bo_list, tv.head); /* Sort A before B if A is smaller. */ return (int)la->robj->tbo.num_pages - (int)lb->robj->tbo.num_pages; @@ -440,13 +418,15 @@ static void radeon_cs_parser_fini(struct radeon_cs_parser *parser, int error, bo if (parser->relocs != NULL) { for (i = 0; i < parser->nrelocs; i++) { - if (parser->relocs[i].gobj) - drm_gem_object_unreference_unlocked(parser->relocs[i].gobj); + struct radeon_bo *bo = parser->relocs[i].robj; + if (bo == NULL) + continue; + + drm_gem_object_unreference_unlocked(&bo->gem_base); } } kfree(parser->track); kfree(parser->relocs); - kfree(parser->relocs_ptr); drm_free_large(parser->vm_bos); for (i = 0; i < parser->nchunks; i++) drm_free_large(parser->chunks[i].kdata); @@ -461,7 +441,7 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev, { int r; - if (parser->chunk_ib_idx == -1) + if (parser->chunk_ib == NULL) return 0; if (parser->cs_flags & RADEON_CS_USE_VM) @@ -521,10 +501,6 @@ static int radeon_bo_vm_update_pte(struct radeon_cs_parser *p, for (i = 0; i < p->nrelocs; i++) { struct radeon_bo *bo; - /* ignore duplicates */ - if (p->relocs_ptr[i] != &p->relocs[i]) - continue; - bo = p->relocs[i].robj; bo_va = radeon_vm_bo_find(vm, bo); if (bo_va == NULL) { @@ -535,6 +511,8 @@ static int radeon_bo_vm_update_pte(struct radeon_cs_parser *p, r = radeon_vm_bo_update(rdev, bo_va, &bo->tbo.mem); if (r) return r; + + radeon_sync_fence(&p->ib.sync, bo_va->last_pt_update); } return radeon_vm_clear_invalids(rdev, vm); @@ -547,7 +525,7 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev, struct radeon_vm *vm = &fpriv->vm; int r; - if (parser->chunk_ib_idx == -1) + if (parser->chunk_ib == NULL) return 0; if ((parser->cs_flags & RADEON_CS_USE_VM) == 0) return 0; @@ -579,10 +557,9 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev, DRM_ERROR("Failed to sync rings: %i\n", r); goto out; } - radeon_semaphore_sync_fence(parser->ib.semaphore, vm->fence); if ((rdev->family >= CHIP_TAHITI) && - (parser->chunk_const_ib_idx != -1)) { + (parser->chunk_const_ib != NULL)) { r = radeon_ib_schedule(rdev, &parser->ib, &parser->const_ib, true); } else { r = radeon_ib_schedule(rdev, &parser->ib, NULL, true); @@ -609,7 +586,7 @@ static int radeon_cs_ib_fill(struct radeon_device *rdev, struct radeon_cs_parser struct radeon_vm *vm = NULL; int r; - if (parser->chunk_ib_idx == -1) + if (parser->chunk_ib == NULL) return 0; if (parser->cs_flags & RADEON_CS_USE_VM) { @@ -617,8 +594,8 @@ static int radeon_cs_ib_fill(struct radeon_device *rdev, struct radeon_cs_parser vm = &fpriv->vm; if ((rdev->family >= CHIP_TAHITI) && - (parser->chunk_const_ib_idx != -1)) { - ib_chunk = &parser->chunks[parser->chunk_const_ib_idx]; + (parser->chunk_const_ib != NULL)) { + ib_chunk = parser->chunk_const_ib; if (ib_chunk->length_dw > RADEON_IB_VM_MAX_SIZE) { DRM_ERROR("cs IB CONST too big: %d\n", ib_chunk->length_dw); return -EINVAL; @@ -637,13 +614,13 @@ static int radeon_cs_ib_fill(struct radeon_device *rdev, struct radeon_cs_parser return -EFAULT; } - ib_chunk = &parser->chunks[parser->chunk_ib_idx]; + ib_chunk = parser->chunk_ib; if (ib_chunk->length_dw > RADEON_IB_VM_MAX_SIZE) { DRM_ERROR("cs IB too big: %d\n", ib_chunk->length_dw); return -EINVAL; } } - ib_chunk = &parser->chunks[parser->chunk_ib_idx]; + ib_chunk = parser->chunk_ib; r = radeon_ib_get(rdev, parser->ring, &parser->ib, vm, ib_chunk->length_dw * 4); @@ -735,7 +712,7 @@ int radeon_cs_packet_parse(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, unsigned idx) { - struct radeon_cs_chunk *ib_chunk = &p->chunks[p->chunk_ib_idx]; + struct radeon_cs_chunk *ib_chunk = p->chunk_ib; struct radeon_device *rdev = p->rdev; uint32_t header; @@ -829,7 +806,7 @@ void radeon_cs_dump_packet(struct radeon_cs_parser *p, * GPU offset using the provided start. **/ int radeon_cs_packet_next_reloc(struct radeon_cs_parser *p, - struct radeon_cs_reloc **cs_reloc, + struct radeon_bo_list **cs_reloc, int nomm) { struct radeon_cs_chunk *relocs_chunk; @@ -837,12 +814,12 @@ int radeon_cs_packet_next_reloc(struct radeon_cs_parser *p, unsigned idx; int r; - if (p->chunk_relocs_idx == -1) { + if (p->chunk_relocs == NULL) { DRM_ERROR("No relocation chunk !\n"); return -EINVAL; } *cs_reloc = NULL; - relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + relocs_chunk = p->chunk_relocs; r = radeon_cs_packet_parse(p, &p3reloc, p->idx); if (r) return r; @@ -868,6 +845,6 @@ int radeon_cs_packet_next_reloc(struct radeon_cs_parser *p, (u64)relocs_chunk->kdata[idx + 3] << 32; (*cs_reloc)->gpu_offset |= relocs_chunk->kdata[idx + 0]; } else - *cs_reloc = p->relocs_ptr[(idx / 4)]; + *cs_reloc = &p->relocs[(idx / 4)]; return 0; } diff --git a/drivers/gpu/drm/radeon/radeon_cursor.c b/drivers/gpu/drm/radeon/radeon_cursor.c index 9630e8d..45e5406 100644 --- a/drivers/gpu/drm/radeon/radeon_cursor.c +++ b/drivers/gpu/drm/radeon/radeon_cursor.c @@ -117,106 +117,7 @@ static void radeon_show_cursor(struct drm_crtc *crtc) } } -static void radeon_set_cursor(struct drm_crtc *crtc, struct drm_gem_object *obj, - uint64_t gpu_addr) -{ - struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); - struct radeon_device *rdev = crtc->dev->dev_private; - - if (ASIC_IS_DCE4(rdev)) { - WREG32(EVERGREEN_CUR_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset, - upper_32_bits(gpu_addr)); - WREG32(EVERGREEN_CUR_SURFACE_ADDRESS + radeon_crtc->crtc_offset, - gpu_addr & 0xffffffff); - } else if (ASIC_IS_AVIVO(rdev)) { - if (rdev->family >= CHIP_RV770) { - if (radeon_crtc->crtc_id) - WREG32(R700_D2CUR_SURFACE_ADDRESS_HIGH, upper_32_bits(gpu_addr)); - else - WREG32(R700_D1CUR_SURFACE_ADDRESS_HIGH, upper_32_bits(gpu_addr)); - } - WREG32(AVIVO_D1CUR_SURFACE_ADDRESS + radeon_crtc->crtc_offset, - gpu_addr & 0xffffffff); - } else { - radeon_crtc->legacy_cursor_offset = gpu_addr - radeon_crtc->legacy_display_base_addr; - /* offset is from DISP(2)_BASE_ADDRESS */ - WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, radeon_crtc->legacy_cursor_offset); - } -} - -int radeon_crtc_cursor_set(struct drm_crtc *crtc, - struct drm_file *file_priv, - uint32_t handle, - uint32_t width, - uint32_t height) -{ - struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); - struct radeon_device *rdev = crtc->dev->dev_private; - struct drm_gem_object *obj; - struct radeon_bo *robj; - uint64_t gpu_addr; - int ret; - - if (!handle) { - /* turn off cursor */ - radeon_hide_cursor(crtc); - obj = NULL; - goto unpin; - } - - if ((width > radeon_crtc->max_cursor_width) || - (height > radeon_crtc->max_cursor_height)) { - DRM_ERROR("bad cursor width or height %d x %d\n", width, height); - return -EINVAL; - } - - obj = drm_gem_object_lookup(crtc->dev, file_priv, handle); - if (!obj) { - DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, radeon_crtc->crtc_id); - return -ENOENT; - } - - robj = gem_to_radeon_bo(obj); - ret = radeon_bo_reserve(robj, false); - if (unlikely(ret != 0)) - goto fail; - /* Only 27 bit offset for legacy cursor */ - ret = radeon_bo_pin_restricted(robj, RADEON_GEM_DOMAIN_VRAM, - ASIC_IS_AVIVO(rdev) ? 0 : 1 << 27, - &gpu_addr); - radeon_bo_unreserve(robj); - if (ret) - goto fail; - - radeon_crtc->cursor_width = width; - radeon_crtc->cursor_height = height; - - radeon_lock_cursor(crtc, true); - radeon_set_cursor(crtc, obj, gpu_addr); - radeon_show_cursor(crtc); - radeon_lock_cursor(crtc, false); - -unpin: - if (radeon_crtc->cursor_bo) { - robj = gem_to_radeon_bo(radeon_crtc->cursor_bo); - ret = radeon_bo_reserve(robj, false); - if (likely(ret == 0)) { - radeon_bo_unpin(robj); - radeon_bo_unreserve(robj); - } - drm_gem_object_unreference_unlocked(radeon_crtc->cursor_bo); - } - - radeon_crtc->cursor_bo = obj; - return 0; -fail: - drm_gem_object_unreference_unlocked(obj); - - return ret; -} - -int radeon_crtc_cursor_move(struct drm_crtc *crtc, - int x, int y) +static int radeon_cursor_move_locked(struct drm_crtc *crtc, int x, int y) { struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); struct radeon_device *rdev = crtc->dev->dev_private; @@ -281,7 +182,6 @@ int radeon_crtc_cursor_move(struct drm_crtc *crtc, } } - radeon_lock_cursor(crtc, true); if (ASIC_IS_DCE4(rdev)) { WREG32(EVERGREEN_CUR_POSITION + radeon_crtc->crtc_offset, (x << 16) | y); WREG32(EVERGREEN_CUR_HOT_SPOT + radeon_crtc->crtc_offset, (xorigin << 16) | yorigin); @@ -308,7 +208,173 @@ int radeon_crtc_cursor_move(struct drm_crtc *crtc, WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, (radeon_crtc->legacy_cursor_offset + (yorigin * 256))); } + + radeon_crtc->cursor_x = x; + radeon_crtc->cursor_y = y; + + return 0; +} + +int radeon_crtc_cursor_move(struct drm_crtc *crtc, + int x, int y) +{ + int ret; + + radeon_lock_cursor(crtc, true); + ret = radeon_cursor_move_locked(crtc, x, y); radeon_lock_cursor(crtc, false); + return ret; +} + +static int radeon_set_cursor(struct drm_crtc *crtc, struct drm_gem_object *obj) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_device *rdev = crtc->dev->dev_private; + struct radeon_bo *robj = gem_to_radeon_bo(obj); + uint64_t gpu_addr; + int ret; + + ret = radeon_bo_reserve(robj, false); + if (unlikely(ret != 0)) + goto fail; + /* Only 27 bit offset for legacy cursor */ + ret = radeon_bo_pin_restricted(robj, RADEON_GEM_DOMAIN_VRAM, + ASIC_IS_AVIVO(rdev) ? 0 : 1 << 27, + &gpu_addr); + radeon_bo_unreserve(robj); + if (ret) + goto fail; + + if (ASIC_IS_DCE4(rdev)) { + WREG32(EVERGREEN_CUR_SURFACE_ADDRESS_HIGH + radeon_crtc->crtc_offset, + upper_32_bits(gpu_addr)); + WREG32(EVERGREEN_CUR_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + gpu_addr & 0xffffffff); + } else if (ASIC_IS_AVIVO(rdev)) { + if (rdev->family >= CHIP_RV770) { + if (radeon_crtc->crtc_id) + WREG32(R700_D2CUR_SURFACE_ADDRESS_HIGH, upper_32_bits(gpu_addr)); + else + WREG32(R700_D1CUR_SURFACE_ADDRESS_HIGH, upper_32_bits(gpu_addr)); + } + WREG32(AVIVO_D1CUR_SURFACE_ADDRESS + radeon_crtc->crtc_offset, + gpu_addr & 0xffffffff); + } else { + radeon_crtc->legacy_cursor_offset = gpu_addr - radeon_crtc->legacy_display_base_addr; + /* offset is from DISP(2)_BASE_ADDRESS */ + WREG32(RADEON_CUR_OFFSET + radeon_crtc->crtc_offset, radeon_crtc->legacy_cursor_offset); + } + return 0; + +fail: + drm_gem_object_unreference_unlocked(obj); + + return ret; +} + +int radeon_crtc_cursor_set2(struct drm_crtc *crtc, + struct drm_file *file_priv, + uint32_t handle, + uint32_t width, + uint32_t height, + int32_t hot_x, + int32_t hot_y) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct drm_gem_object *obj; + int ret; + + if (!handle) { + /* turn off cursor */ + radeon_hide_cursor(crtc); + obj = NULL; + goto unpin; + } + + if ((width > radeon_crtc->max_cursor_width) || + (height > radeon_crtc->max_cursor_height)) { + DRM_ERROR("bad cursor width or height %d x %d\n", width, height); + return -EINVAL; + } + + obj = drm_gem_object_lookup(crtc->dev, file_priv, handle); + if (!obj) { + DRM_ERROR("Cannot find cursor object %x for crtc %d\n", handle, radeon_crtc->crtc_id); + return -ENOENT; + } + + radeon_crtc->cursor_width = width; + radeon_crtc->cursor_height = height; + + radeon_lock_cursor(crtc, true); + + if (hot_x != radeon_crtc->cursor_hot_x || + hot_y != radeon_crtc->cursor_hot_y) { + int x, y; + + x = radeon_crtc->cursor_x + radeon_crtc->cursor_hot_x - hot_x; + y = radeon_crtc->cursor_y + radeon_crtc->cursor_hot_y - hot_y; + + radeon_cursor_move_locked(crtc, x, y); + + radeon_crtc->cursor_hot_x = hot_x; + radeon_crtc->cursor_hot_y = hot_y; + } + + ret = radeon_set_cursor(crtc, obj); + + if (ret) + DRM_ERROR("radeon_set_cursor returned %d, not changing cursor\n", + ret); + else + radeon_show_cursor(crtc); + + radeon_lock_cursor(crtc, false); + +unpin: + if (radeon_crtc->cursor_bo) { + struct radeon_bo *robj = gem_to_radeon_bo(radeon_crtc->cursor_bo); + ret = radeon_bo_reserve(robj, false); + if (likely(ret == 0)) { + radeon_bo_unpin(robj); + radeon_bo_unreserve(robj); + } + if (radeon_crtc->cursor_bo != obj) + drm_gem_object_unreference_unlocked(radeon_crtc->cursor_bo); + } + + radeon_crtc->cursor_bo = obj; + return 0; +} + +/** + * radeon_cursor_reset - Re-set the current cursor, if any. + * + * @crtc: drm crtc + * + * If the CRTC passed in currently has a cursor assigned, this function + * makes sure it's visible. + */ +void radeon_cursor_reset(struct drm_crtc *crtc) +{ + struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + int ret; + + if (radeon_crtc->cursor_bo) { + radeon_lock_cursor(crtc, true); + + radeon_cursor_move_locked(crtc, radeon_crtc->cursor_x, + radeon_crtc->cursor_y); + + ret = radeon_set_cursor(crtc, radeon_crtc->cursor_bo); + if (ret) + DRM_ERROR("radeon_set_cursor returned %d, not showing " + "cursor\n", ret); + else + radeon_show_cursor(crtc); + + radeon_lock_cursor(crtc, false); + } } diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c index 995a8b1..0ec6516 100644 --- a/drivers/gpu/drm/radeon/radeon_device.c +++ b/drivers/gpu/drm/radeon/radeon_device.c @@ -377,6 +377,37 @@ void radeon_doorbell_free(struct radeon_device *rdev, u32 doorbell) __clear_bit(doorbell, rdev->doorbell.used); } +/** + * radeon_doorbell_get_kfd_info - Report doorbell configuration required to + * setup KFD + * + * @rdev: radeon_device pointer + * @aperture_base: output returning doorbell aperture base physical address + * @aperture_size: output returning doorbell aperture size in bytes + * @start_offset: output returning # of doorbell bytes reserved for radeon. + * + * Radeon and the KFD share the doorbell aperture. Radeon sets it up, + * takes doorbells required for its own rings and reports the setup to KFD. + * Radeon reserved doorbells are at the start of the doorbell aperture. + */ +void radeon_doorbell_get_kfd_info(struct radeon_device *rdev, + phys_addr_t *aperture_base, + size_t *aperture_size, + size_t *start_offset) +{ + /* The first num_doorbells are used by radeon. + * KFD takes whatever's left in the aperture. */ + if (rdev->doorbell.size > rdev->doorbell.num_doorbells * sizeof(u32)) { + *aperture_base = rdev->doorbell.base; + *aperture_size = rdev->doorbell.size; + *start_offset = rdev->doorbell.num_doorbells * sizeof(u32); + } else { + *aperture_base = 0; + *aperture_size = 0; + *start_offset = 0; + } +} + /* * radeon_wb_*() * Writeback is the the method by which the the GPU updates special pages @@ -1273,6 +1304,7 @@ int radeon_device_init(struct radeon_device *rdev, mutex_init(&rdev->pm.mutex); mutex_init(&rdev->gpu_clock_mutex); mutex_init(&rdev->srbm_mutex); + mutex_init(&rdev->grbm_idx_mutex); init_rwsem(&rdev->pm.mclk_lock); init_rwsem(&rdev->exclusive_lock); init_waitqueue_head(&rdev->irq.vblank_queue); diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c index 00ead8c..1021169 100644 --- a/drivers/gpu/drm/radeon/radeon_display.c +++ b/drivers/gpu/drm/radeon/radeon_display.c @@ -32,6 +32,7 @@ #include <linux/pm_runtime.h> #include <drm/drm_crtc_helper.h> +#include <drm/drm_plane_helper.h> #include <drm/drm_edid.h> #include <linux/gcd.h> @@ -634,7 +635,7 @@ radeon_crtc_set_config(struct drm_mode_set *set) return ret; } static const struct drm_crtc_funcs radeon_crtc_funcs = { - .cursor_set = radeon_crtc_cursor_set, + .cursor_set2 = radeon_crtc_cursor_set2, .cursor_move = radeon_crtc_cursor_move, .gamma_set = radeon_crtc_gamma_set, .set_config = radeon_crtc_set_config, diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index dcffa30..4f50fb0 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -41,6 +41,8 @@ #include <drm/drm_gem.h> #include "drm_crtc_helper.h" +#include "radeon_kfd.h" + /* * KMS wrapper. * - 2.0.0 - initial interface @@ -654,12 +656,15 @@ static int __init radeon_init(void) #endif } + radeon_kfd_init(); + /* let modprobe override vga console setting */ return drm_pci_init(driver, pdriver); } static void __exit radeon_exit(void) { + radeon_kfd_fini(); drm_pci_exit(driver, pdriver); radeon_unregister_atpx_handler(); } diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c index 0ea1db8..29b9220 100644 --- a/drivers/gpu/drm/radeon/radeon_fb.c +++ b/drivers/gpu/drm/radeon/radeon_fb.c @@ -48,10 +48,40 @@ struct radeon_fbdev { struct radeon_device *rdev; }; +/** + * radeon_fb_helper_set_par - Hide cursor on CRTCs used by fbdev. + * + * @info: fbdev info + * + * This function hides the cursor on all CRTCs used by fbdev. + */ +static int radeon_fb_helper_set_par(struct fb_info *info) +{ + int ret; + + ret = drm_fb_helper_set_par(info); + + /* XXX: with universal plane support fbdev will automatically disable + * all non-primary planes (including the cursor) + */ + if (ret == 0) { + struct drm_fb_helper *fb_helper = info->par; + int i; + + for (i = 0; i < fb_helper->crtc_count; i++) { + struct drm_crtc *crtc = fb_helper->crtc_info[i].mode_set.crtc; + + radeon_crtc_cursor_set2(crtc, NULL, 0, 0, 0, 0, 0); + } + } + + return ret; +} + static struct fb_ops radeonfb_ops = { .owner = THIS_MODULE, .fb_check_var = drm_fb_helper_check_var, - .fb_set_par = drm_fb_helper_set_par, + .fb_set_par = radeon_fb_helper_set_par, .fb_fillrect = cfb_fillrect, .fb_copyarea = cfb_copyarea, .fb_imageblit = cfb_imageblit, diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c index 9951670..d13d1b5 100644 --- a/drivers/gpu/drm/radeon/radeon_fence.c +++ b/drivers/gpu/drm/radeon/radeon_fence.c @@ -140,6 +140,7 @@ int radeon_fence_emit(struct radeon_device *rdev, (*fence)->rdev = rdev; (*fence)->seq = seq; (*fence)->ring = ring; + (*fence)->is_vm_update = false; fence_init(&(*fence)->base, &radeon_fence_ops, &rdev->fence_queue.lock, rdev->fence_context + ring, seq); radeon_fence_ring_emit(rdev, ring, *fence); diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index c194497..fe48f22 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -394,9 +394,10 @@ int radeon_gem_set_domain_ioctl(struct drm_device *dev, void *data, return r; } -int radeon_mode_dumb_mmap(struct drm_file *filp, - struct drm_device *dev, - uint32_t handle, uint64_t *offset_p) +static int radeon_mode_mmap(struct drm_file *filp, + struct drm_device *dev, + uint32_t handle, bool dumb, + uint64_t *offset_p) { struct drm_gem_object *gobj; struct radeon_bo *robj; @@ -405,6 +406,14 @@ int radeon_mode_dumb_mmap(struct drm_file *filp, if (gobj == NULL) { return -ENOENT; } + + /* + * We don't allow dumb mmaps on objects created using another + * interface. + */ + WARN_ONCE(dumb && !(gobj->dumb || gobj->import_attach), + "Illegal dumb map of GPU buffer.\n"); + robj = gem_to_radeon_bo(gobj); if (radeon_ttm_tt_has_userptr(robj->tbo.ttm)) { drm_gem_object_unreference_unlocked(gobj); @@ -415,12 +424,20 @@ int radeon_mode_dumb_mmap(struct drm_file *filp, return 0; } +int radeon_mode_dumb_mmap(struct drm_file *filp, + struct drm_device *dev, + uint32_t handle, uint64_t *offset_p) +{ + return radeon_mode_mmap(filp, dev, handle, true, offset_p); +} + int radeon_gem_mmap_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) { struct drm_radeon_gem_mmap *args = data; - return radeon_mode_dumb_mmap(filp, dev, args->handle, &args->addr_ptr); + return radeon_mode_mmap(filp, dev, args->handle, false, + &args->addr_ptr); } int radeon_gem_busy_ioctl(struct drm_device *dev, void *data, @@ -518,6 +535,68 @@ out: return r; } +/** + * radeon_gem_va_update_vm -update the bo_va in its VM + * + * @rdev: radeon_device pointer + * @bo_va: bo_va to update + * + * Update the bo_va directly after setting it's address. Errors are not + * vital here, so they are not reported back to userspace. + */ +static void radeon_gem_va_update_vm(struct radeon_device *rdev, + struct radeon_bo_va *bo_va) +{ + struct ttm_validate_buffer tv, *entry; + struct radeon_bo_list *vm_bos; + struct ww_acquire_ctx ticket; + struct list_head list; + unsigned domain; + int r; + + INIT_LIST_HEAD(&list); + + tv.bo = &bo_va->bo->tbo; + tv.shared = true; + list_add(&tv.head, &list); + + vm_bos = radeon_vm_get_bos(rdev, bo_va->vm, &list); + if (!vm_bos) + return; + + r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL); + if (r) + goto error_free; + + list_for_each_entry(entry, &list, head) { + domain = radeon_mem_type_to_domain(entry->bo->mem.mem_type); + /* if anything is swapped out don't swap it in here, + just abort and wait for the next CS */ + if (domain == RADEON_GEM_DOMAIN_CPU) + goto error_unreserve; + } + + mutex_lock(&bo_va->vm->mutex); + r = radeon_vm_clear_freed(rdev, bo_va->vm); + if (r) + goto error_unlock; + + if (bo_va->it.start) + r = radeon_vm_bo_update(rdev, bo_va, &bo_va->bo->tbo.mem); + +error_unlock: + mutex_unlock(&bo_va->vm->mutex); + +error_unreserve: + ttm_eu_backoff_reservation(&ticket, &list); + +error_free: + drm_free_large(vm_bos); + + if (r) + DRM_ERROR("Couldn't update BO_VA (%d)\n", r); +} + int radeon_gem_va_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) { @@ -601,6 +680,7 @@ int radeon_gem_va_ioctl(struct drm_device *dev, void *data, if (bo_va->it.start) { args->operation = RADEON_VA_RESULT_VA_EXIST; args->offset = bo_va->it.start * RADEON_GPU_PAGE_SIZE; + radeon_bo_unreserve(rbo); goto out; } r = radeon_vm_bo_set_addr(rdev, bo_va, args->offset, args->flags); @@ -611,12 +691,13 @@ int radeon_gem_va_ioctl(struct drm_device *dev, void *data, default: break; } + if (!r) + radeon_gem_va_update_vm(rdev, bo_va); args->operation = RADEON_VA_RESULT_OK; if (r) { args->operation = RADEON_VA_RESULT_ERROR; } out: - radeon_bo_unreserve(rbo); drm_gem_object_unreference_unlocked(gobj); return r; } @@ -682,6 +763,7 @@ int radeon_mode_dumb_create(struct drm_file *file_priv, return -ENOMEM; r = drm_gem_handle_create(file_priv, gobj, &handle); + gobj->dumb = true; /* drop reference from allocate - handle holds it now */ drm_gem_object_unreference_unlocked(gobj); if (r) { diff --git a/drivers/gpu/drm/radeon/radeon_ib.c b/drivers/gpu/drm/radeon/radeon_ib.c index 3f39fcc..c39ce1f 100644 --- a/drivers/gpu/drm/radeon/radeon_ib.c +++ b/drivers/gpu/drm/radeon/radeon_ib.c @@ -64,10 +64,7 @@ int radeon_ib_get(struct radeon_device *rdev, int ring, return r; } - r = radeon_semaphore_create(rdev, &ib->semaphore); - if (r) { - return r; - } + radeon_sync_create(&ib->sync); ib->ring = ring; ib->fence = NULL; @@ -96,7 +93,7 @@ int radeon_ib_get(struct radeon_device *rdev, int ring, */ void radeon_ib_free(struct radeon_device *rdev, struct radeon_ib *ib) { - radeon_semaphore_free(rdev, &ib->semaphore, ib->fence); + radeon_sync_free(rdev, &ib->sync, ib->fence); radeon_sa_bo_free(rdev, &ib->sa_bo, ib->fence); radeon_fence_unref(&ib->fence); } @@ -145,11 +142,11 @@ int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib, if (ib->vm) { struct radeon_fence *vm_id_fence; vm_id_fence = radeon_vm_grab_id(rdev, ib->vm, ib->ring); - radeon_semaphore_sync_fence(ib->semaphore, vm_id_fence); + radeon_sync_fence(&ib->sync, vm_id_fence); } /* sync with other rings */ - r = radeon_semaphore_sync_rings(rdev, ib->semaphore, ib->ring); + r = radeon_sync_rings(rdev, &ib->sync, ib->ring); if (r) { dev_err(rdev->dev, "failed to sync rings (%d)\n", r); radeon_ring_unlock_undo(rdev, ring); @@ -157,11 +154,12 @@ int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib, } if (ib->vm) - radeon_vm_flush(rdev, ib->vm, ib->ring); + radeon_vm_flush(rdev, ib->vm, ib->ring, + ib->sync.last_vm_update); if (const_ib) { radeon_ring_ib_execute(rdev, const_ib->ring, const_ib); - radeon_semaphore_free(rdev, &const_ib->semaphore, NULL); + radeon_sync_free(rdev, &const_ib->sync, NULL); } radeon_ring_ib_execute(rdev, ib->ring, ib); r = radeon_fence_emit(rdev, &ib->fence, ib->ring); diff --git a/drivers/gpu/drm/radeon/radeon_kfd.c b/drivers/gpu/drm/radeon/radeon_kfd.c new file mode 100644 index 0000000..065d020 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_kfd.c @@ -0,0 +1,563 @@ +/* + * Copyright 2014 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#include <linux/module.h> +#include <linux/fdtable.h> +#include <linux/uaccess.h> +#include <drm/drmP.h> +#include "radeon.h" +#include "cikd.h" +#include "cik_reg.h" +#include "radeon_kfd.h" + +#define CIK_PIPE_PER_MEC (4) + +struct kgd_mem { + struct radeon_sa_bo *sa_bo; + uint64_t gpu_addr; + void *ptr; +}; + +static int init_sa_manager(struct kgd_dev *kgd, unsigned int size); +static void fini_sa_manager(struct kgd_dev *kgd); + +static int allocate_mem(struct kgd_dev *kgd, size_t size, size_t alignment, + enum kgd_memory_pool pool, struct kgd_mem **mem); + +static void free_mem(struct kgd_dev *kgd, struct kgd_mem *mem); + +static uint64_t get_vmem_size(struct kgd_dev *kgd); +static uint64_t get_gpu_clock_counter(struct kgd_dev *kgd); + +static uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd); + +/* + * Register access functions + */ + +static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid, + uint32_t sh_mem_config, uint32_t sh_mem_ape1_base, + uint32_t sh_mem_ape1_limit, uint32_t sh_mem_bases); + +static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid, + unsigned int vmid); + +static int kgd_init_memory(struct kgd_dev *kgd); + +static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id, + uint32_t hpd_size, uint64_t hpd_gpu_addr); + +static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id, + uint32_t queue_id, uint32_t __user *wptr); + +static bool kgd_hqd_is_occupies(struct kgd_dev *kgd, uint64_t queue_address, + uint32_t pipe_id, uint32_t queue_id); + +static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type, + unsigned int timeout, uint32_t pipe_id, + uint32_t queue_id); + +static const struct kfd2kgd_calls kfd2kgd = { + .init_sa_manager = init_sa_manager, + .fini_sa_manager = fini_sa_manager, + .allocate_mem = allocate_mem, + .free_mem = free_mem, + .get_vmem_size = get_vmem_size, + .get_gpu_clock_counter = get_gpu_clock_counter, + .get_max_engine_clock_in_mhz = get_max_engine_clock_in_mhz, + .program_sh_mem_settings = kgd_program_sh_mem_settings, + .set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping, + .init_memory = kgd_init_memory, + .init_pipeline = kgd_init_pipeline, + .hqd_load = kgd_hqd_load, + .hqd_is_occupies = kgd_hqd_is_occupies, + .hqd_destroy = kgd_hqd_destroy, +}; + +static const struct kgd2kfd_calls *kgd2kfd; + +bool radeon_kfd_init(void) +{ + bool (*kgd2kfd_init_p)(unsigned, const struct kfd2kgd_calls*, + const struct kgd2kfd_calls**); + + kgd2kfd_init_p = symbol_request(kgd2kfd_init); + + if (kgd2kfd_init_p == NULL) + return false; + + if (!kgd2kfd_init_p(KFD_INTERFACE_VERSION, &kfd2kgd, &kgd2kfd)) { + symbol_put(kgd2kfd_init); + kgd2kfd = NULL; + + return false; + } + + return true; +} + +void radeon_kfd_fini(void) +{ + if (kgd2kfd) { + kgd2kfd->exit(); + symbol_put(kgd2kfd_init); + } +} + +void radeon_kfd_device_probe(struct radeon_device *rdev) +{ + if (kgd2kfd) + rdev->kfd = kgd2kfd->probe((struct kgd_dev *)rdev, rdev->pdev); +} + +void radeon_kfd_device_init(struct radeon_device *rdev) +{ + if (rdev->kfd) { + struct kgd2kfd_shared_resources gpu_resources = { + .compute_vmid_bitmap = 0xFF00, + + .first_compute_pipe = 1, + .compute_pipe_count = 8 - 1, + }; + + radeon_doorbell_get_kfd_info(rdev, + &gpu_resources.doorbell_physical_address, + &gpu_resources.doorbell_aperture_size, + &gpu_resources.doorbell_start_offset); + + kgd2kfd->device_init(rdev->kfd, &gpu_resources); + } +} + +void radeon_kfd_device_fini(struct radeon_device *rdev) +{ + if (rdev->kfd) { + kgd2kfd->device_exit(rdev->kfd); + rdev->kfd = NULL; + } +} + +void radeon_kfd_interrupt(struct radeon_device *rdev, const void *ih_ring_entry) +{ + if (rdev->kfd) + kgd2kfd->interrupt(rdev->kfd, ih_ring_entry); +} + +void radeon_kfd_suspend(struct radeon_device *rdev) +{ + if (rdev->kfd) + kgd2kfd->suspend(rdev->kfd); +} + +int radeon_kfd_resume(struct radeon_device *rdev) +{ + int r = 0; + + if (rdev->kfd) + r = kgd2kfd->resume(rdev->kfd); + + return r; +} + +static u32 pool_to_domain(enum kgd_memory_pool p) +{ + switch (p) { + case KGD_POOL_FRAMEBUFFER: return RADEON_GEM_DOMAIN_VRAM; + default: return RADEON_GEM_DOMAIN_GTT; + } +} + +static int init_sa_manager(struct kgd_dev *kgd, unsigned int size) +{ + struct radeon_device *rdev = (struct radeon_device *)kgd; + int r; + + BUG_ON(kgd == NULL); + + r = radeon_sa_bo_manager_init(rdev, &rdev->kfd_bo, + size, + RADEON_GPU_PAGE_SIZE, + RADEON_GEM_DOMAIN_GTT, + RADEON_GEM_GTT_WC); + + if (r) + return r; + + r = radeon_sa_bo_manager_start(rdev, &rdev->kfd_bo); + if (r) + radeon_sa_bo_manager_fini(rdev, &rdev->kfd_bo); + + return r; +} + +static void fini_sa_manager(struct kgd_dev *kgd) +{ + struct radeon_device *rdev = (struct radeon_device *)kgd; + + BUG_ON(kgd == NULL); + + radeon_sa_bo_manager_suspend(rdev, &rdev->kfd_bo); + radeon_sa_bo_manager_fini(rdev, &rdev->kfd_bo); +} + +static int allocate_mem(struct kgd_dev *kgd, size_t size, size_t alignment, + enum kgd_memory_pool pool, struct kgd_mem **mem) +{ + struct radeon_device *rdev = (struct radeon_device *)kgd; + u32 domain; + int r; + + BUG_ON(kgd == NULL); + + domain = pool_to_domain(pool); + if (domain != RADEON_GEM_DOMAIN_GTT) { + dev_err(rdev->dev, + "Only allowed to allocate gart memory for kfd\n"); + return -EINVAL; + } + + *mem = kmalloc(sizeof(struct kgd_mem), GFP_KERNEL); + if ((*mem) == NULL) + return -ENOMEM; + + r = radeon_sa_bo_new(rdev, &rdev->kfd_bo, &(*mem)->sa_bo, size, + alignment); + if (r) { + dev_err(rdev->dev, "failed to get memory for kfd (%d)\n", r); + return r; + } + + (*mem)->ptr = radeon_sa_bo_cpu_addr((*mem)->sa_bo); + (*mem)->gpu_addr = radeon_sa_bo_gpu_addr((*mem)->sa_bo); + + return 0; +} + +static void free_mem(struct kgd_dev *kgd, struct kgd_mem *mem) +{ + struct radeon_device *rdev = (struct radeon_device *)kgd; + + BUG_ON(kgd == NULL); + + radeon_sa_bo_free(rdev, &mem->sa_bo, NULL); + kfree(mem); +} + +static uint64_t get_vmem_size(struct kgd_dev *kgd) +{ + struct radeon_device *rdev = (struct radeon_device *)kgd; + + BUG_ON(kgd == NULL); + + return rdev->mc.real_vram_size; +} + +static uint64_t get_gpu_clock_counter(struct kgd_dev *kgd) +{ + struct radeon_device *rdev = (struct radeon_device *)kgd; + + return rdev->asic->get_gpu_clock_counter(rdev); +} + +static uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd) +{ + struct radeon_device *rdev = (struct radeon_device *)kgd; + + /* The sclk is in quantas of 10kHz */ + return rdev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk / 100; +} + +static inline struct radeon_device *get_radeon_device(struct kgd_dev *kgd) +{ + return (struct radeon_device *)kgd; +} + +static void write_register(struct kgd_dev *kgd, uint32_t offset, uint32_t value) +{ + struct radeon_device *rdev = get_radeon_device(kgd); + + writel(value, (void __iomem *)(rdev->rmmio + offset)); +} + +static uint32_t read_register(struct kgd_dev *kgd, uint32_t offset) +{ + struct radeon_device *rdev = get_radeon_device(kgd); + + return readl((void __iomem *)(rdev->rmmio + offset)); +} + +static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe, + uint32_t queue, uint32_t vmid) +{ + struct radeon_device *rdev = get_radeon_device(kgd); + uint32_t value = PIPEID(pipe) | MEID(mec) | VMID(vmid) | QUEUEID(queue); + + mutex_lock(&rdev->srbm_mutex); + write_register(kgd, SRBM_GFX_CNTL, value); +} + +static void unlock_srbm(struct kgd_dev *kgd) +{ + struct radeon_device *rdev = get_radeon_device(kgd); + + write_register(kgd, SRBM_GFX_CNTL, 0); + mutex_unlock(&rdev->srbm_mutex); +} + +static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id, + uint32_t queue_id) +{ + uint32_t mec = (++pipe_id / CIK_PIPE_PER_MEC) + 1; + uint32_t pipe = (pipe_id % CIK_PIPE_PER_MEC); + + lock_srbm(kgd, mec, pipe, queue_id, 0); +} + +static void release_queue(struct kgd_dev *kgd) +{ + unlock_srbm(kgd); +} + +static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid, + uint32_t sh_mem_config, + uint32_t sh_mem_ape1_base, + uint32_t sh_mem_ape1_limit, + uint32_t sh_mem_bases) +{ + lock_srbm(kgd, 0, 0, 0, vmid); + + write_register(kgd, SH_MEM_CONFIG, sh_mem_config); + write_register(kgd, SH_MEM_APE1_BASE, sh_mem_ape1_base); + write_register(kgd, SH_MEM_APE1_LIMIT, sh_mem_ape1_limit); + write_register(kgd, SH_MEM_BASES, sh_mem_bases); + + unlock_srbm(kgd); +} + +static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid, + unsigned int vmid) +{ + /* + * We have to assume that there is no outstanding mapping. + * The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 + * because a mapping is in progress or because a mapping finished and + * the SW cleared it. + * So the protocol is to always wait & clear. + */ + uint32_t pasid_mapping = (pasid == 0) ? 0 : + (uint32_t)pasid | ATC_VMID_PASID_MAPPING_VALID; + + write_register(kgd, ATC_VMID0_PASID_MAPPING + vmid*sizeof(uint32_t), + pasid_mapping); + + while (!(read_register(kgd, ATC_VMID_PASID_MAPPING_UPDATE_STATUS) & + (1U << vmid))) + cpu_relax(); + write_register(kgd, ATC_VMID_PASID_MAPPING_UPDATE_STATUS, 1U << vmid); + + return 0; +} + +static int kgd_init_memory(struct kgd_dev *kgd) +{ + /* + * Configure apertures: + * LDS: 0x60000000'00000000 - 0x60000001'00000000 (4GB) + * Scratch: 0x60000001'00000000 - 0x60000002'00000000 (4GB) + * GPUVM: 0x60010000'00000000 - 0x60020000'00000000 (1TB) + */ + int i; + uint32_t sh_mem_bases = PRIVATE_BASE(0x6000) | SHARED_BASE(0x6000); + + for (i = 8; i < 16; i++) { + uint32_t sh_mem_config; + + lock_srbm(kgd, 0, 0, 0, i); + + sh_mem_config = ALIGNMENT_MODE(SH_MEM_ALIGNMENT_MODE_UNALIGNED); + sh_mem_config |= DEFAULT_MTYPE(MTYPE_NONCACHED); + + write_register(kgd, SH_MEM_CONFIG, sh_mem_config); + + write_register(kgd, SH_MEM_BASES, sh_mem_bases); + + /* Scratch aperture is not supported for now. */ + write_register(kgd, SH_STATIC_MEM_CONFIG, 0); + + /* APE1 disabled for now. */ + write_register(kgd, SH_MEM_APE1_BASE, 1); + write_register(kgd, SH_MEM_APE1_LIMIT, 0); + + unlock_srbm(kgd); + } + + return 0; +} + +static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id, + uint32_t hpd_size, uint64_t hpd_gpu_addr) +{ + uint32_t mec = (++pipe_id / CIK_PIPE_PER_MEC) + 1; + uint32_t pipe = (pipe_id % CIK_PIPE_PER_MEC); + + lock_srbm(kgd, mec, pipe, 0, 0); + write_register(kgd, CP_HPD_EOP_BASE_ADDR, + lower_32_bits(hpd_gpu_addr >> 8)); + write_register(kgd, CP_HPD_EOP_BASE_ADDR_HI, + upper_32_bits(hpd_gpu_addr >> 8)); + write_register(kgd, CP_HPD_EOP_VMID, 0); + write_register(kgd, CP_HPD_EOP_CONTROL, hpd_size); + unlock_srbm(kgd); + + return 0; +} + +static inline struct cik_mqd *get_mqd(void *mqd) +{ + return (struct cik_mqd *)mqd; +} + +static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id, + uint32_t queue_id, uint32_t __user *wptr) +{ + uint32_t wptr_shadow, is_wptr_shadow_valid; + struct cik_mqd *m; + + m = get_mqd(mqd); + + is_wptr_shadow_valid = !get_user(wptr_shadow, wptr); + + acquire_queue(kgd, pipe_id, queue_id); + write_register(kgd, CP_MQD_BASE_ADDR, m->cp_mqd_base_addr_lo); + write_register(kgd, CP_MQD_BASE_ADDR_HI, m->cp_mqd_base_addr_hi); + write_register(kgd, CP_MQD_CONTROL, m->cp_mqd_control); + + write_register(kgd, CP_HQD_PQ_BASE, m->cp_hqd_pq_base_lo); + write_register(kgd, CP_HQD_PQ_BASE_HI, m->cp_hqd_pq_base_hi); + write_register(kgd, CP_HQD_PQ_CONTROL, m->cp_hqd_pq_control); + + write_register(kgd, CP_HQD_IB_CONTROL, m->cp_hqd_ib_control); + write_register(kgd, CP_HQD_IB_BASE_ADDR, m->cp_hqd_ib_base_addr_lo); + write_register(kgd, CP_HQD_IB_BASE_ADDR_HI, m->cp_hqd_ib_base_addr_hi); + + write_register(kgd, CP_HQD_IB_RPTR, m->cp_hqd_ib_rptr); + + write_register(kgd, CP_HQD_PERSISTENT_STATE, + m->cp_hqd_persistent_state); + write_register(kgd, CP_HQD_SEMA_CMD, m->cp_hqd_sema_cmd); + write_register(kgd, CP_HQD_MSG_TYPE, m->cp_hqd_msg_type); + + write_register(kgd, CP_HQD_ATOMIC0_PREOP_LO, + m->cp_hqd_atomic0_preop_lo); + + write_register(kgd, CP_HQD_ATOMIC0_PREOP_HI, + m->cp_hqd_atomic0_preop_hi); + + write_register(kgd, CP_HQD_ATOMIC1_PREOP_LO, + m->cp_hqd_atomic1_preop_lo); + + write_register(kgd, CP_HQD_ATOMIC1_PREOP_HI, + m->cp_hqd_atomic1_preop_hi); + + write_register(kgd, CP_HQD_PQ_RPTR_REPORT_ADDR, + m->cp_hqd_pq_rptr_report_addr_lo); + + write_register(kgd, CP_HQD_PQ_RPTR_REPORT_ADDR_HI, + m->cp_hqd_pq_rptr_report_addr_hi); + + write_register(kgd, CP_HQD_PQ_RPTR, m->cp_hqd_pq_rptr); + + write_register(kgd, CP_HQD_PQ_WPTR_POLL_ADDR, + m->cp_hqd_pq_wptr_poll_addr_lo); + + write_register(kgd, CP_HQD_PQ_WPTR_POLL_ADDR_HI, + m->cp_hqd_pq_wptr_poll_addr_hi); + + write_register(kgd, CP_HQD_PQ_DOORBELL_CONTROL, + m->cp_hqd_pq_doorbell_control); + + write_register(kgd, CP_HQD_VMID, m->cp_hqd_vmid); + + write_register(kgd, CP_HQD_QUANTUM, m->cp_hqd_quantum); + + write_register(kgd, CP_HQD_PIPE_PRIORITY, m->cp_hqd_pipe_priority); + write_register(kgd, CP_HQD_QUEUE_PRIORITY, m->cp_hqd_queue_priority); + + write_register(kgd, CP_HQD_IQ_RPTR, m->cp_hqd_iq_rptr); + + if (is_wptr_shadow_valid) + write_register(kgd, CP_HQD_PQ_WPTR, wptr_shadow); + + write_register(kgd, CP_HQD_ACTIVE, m->cp_hqd_active); + release_queue(kgd); + + return 0; +} + +static bool kgd_hqd_is_occupies(struct kgd_dev *kgd, uint64_t queue_address, + uint32_t pipe_id, uint32_t queue_id) +{ + uint32_t act; + bool retval = false; + uint32_t low, high; + + acquire_queue(kgd, pipe_id, queue_id); + act = read_register(kgd, CP_HQD_ACTIVE); + if (act) { + low = lower_32_bits(queue_address >> 8); + high = upper_32_bits(queue_address >> 8); + + if (low == read_register(kgd, CP_HQD_PQ_BASE) && + high == read_register(kgd, CP_HQD_PQ_BASE_HI)) + retval = true; + } + release_queue(kgd); + return retval; +} + +static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type, + unsigned int timeout, uint32_t pipe_id, + uint32_t queue_id) +{ + uint32_t temp; + + acquire_queue(kgd, pipe_id, queue_id); + write_register(kgd, CP_HQD_PQ_DOORBELL_CONTROL, 0); + + write_register(kgd, CP_HQD_DEQUEUE_REQUEST, reset_type); + + while (true) { + temp = read_register(kgd, CP_HQD_ACTIVE); + if (temp & 0x1) + break; + if (timeout == 0) { + pr_err("kfd: cp queue preemption time out (%dms)\n", + temp); + return -ETIME; + } + msleep(20); + timeout -= 20; + } + + release_queue(kgd); + return 0; +} diff --git a/drivers/gpu/drm/radeon/radeon_kfd.h b/drivers/gpu/drm/radeon/radeon_kfd.h new file mode 100644 index 0000000..f90e161 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_kfd.h @@ -0,0 +1,47 @@ +/* + * Copyright 2014 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +/* + * radeon_kfd.h defines the private interface between the + * AMD kernel graphics drivers and the AMD KFD. + */ + +#ifndef RADEON_KFD_H_INCLUDED +#define RADEON_KFD_H_INCLUDED + +#include <linux/types.h> +#include "../amd/include/kgd_kfd_interface.h" + +struct radeon_device; + +bool radeon_kfd_init(void); +void radeon_kfd_fini(void); + +void radeon_kfd_suspend(struct radeon_device *rdev); +int radeon_kfd_resume(struct radeon_device *rdev); +void radeon_kfd_interrupt(struct radeon_device *rdev, + const void *ih_ring_entry); +void radeon_kfd_device_probe(struct radeon_device *rdev); +void radeon_kfd_device_init(struct radeon_device *rdev); +void radeon_kfd_device_fini(struct radeon_device *rdev); + +#endif /* RADEON_KFD_H_INCLUDED */ diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index 0358676..3cf9c1f 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -34,6 +34,8 @@ #include <linux/slab.h> #include <linux/pm_runtime.h> +#include "radeon_kfd.h" + #if defined(CONFIG_VGA_SWITCHEROO) bool radeon_has_atpx(void); #else @@ -63,6 +65,8 @@ int radeon_driver_unload_kms(struct drm_device *dev) pm_runtime_get_sync(dev->dev); + radeon_kfd_device_fini(rdev); + radeon_acpi_fini(rdev); radeon_modeset_fini(rdev); @@ -142,6 +146,9 @@ int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags) "Error during ACPI methods call\n"); } + radeon_kfd_device_probe(rdev); + radeon_kfd_device_init(rdev); + if (radeon_is_px(dev)) { pm_runtime_use_autosuspend(dev->dev); pm_runtime_set_autosuspend_delay(dev->dev, 5000); @@ -621,8 +628,6 @@ int radeon_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv) RADEON_VA_IB_OFFSET, RADEON_VM_PAGE_READABLE | RADEON_VM_PAGE_SNOOPED); - - radeon_bo_unreserve(rdev->ring_tmp_bo.bo); if (r) { radeon_vm_fini(rdev, vm); kfree(fpriv); diff --git a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c index cafb1cc..678b438 100644 --- a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c +++ b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c @@ -1054,6 +1054,7 @@ static int radeon_crtc_mode_set(struct drm_crtc *crtc, DRM_ERROR("Mode need scaling but only first crtc can do that.\n"); } } + radeon_cursor_reset(crtc); return 0; } diff --git a/drivers/gpu/drm/radeon/radeon_mode.h b/drivers/gpu/drm/radeon/radeon_mode.h index 04db2fd..390db89 100644 --- a/drivers/gpu/drm/radeon/radeon_mode.h +++ b/drivers/gpu/drm/radeon/radeon_mode.h @@ -321,6 +321,10 @@ struct radeon_crtc { uint32_t crtc_offset; struct drm_gem_object *cursor_bo; uint64_t cursor_addr; + int cursor_x; + int cursor_y; + int cursor_hot_x; + int cursor_hot_y; int cursor_width; int cursor_height; int max_cursor_width; @@ -462,6 +466,7 @@ struct radeon_gpio_rec { u8 id; u32 reg; u32 mask; + u32 shift; }; struct radeon_hpd { @@ -748,6 +753,8 @@ extern bool radeon_atombios_get_ppll_ss_info(struct radeon_device *rdev, extern bool radeon_atombios_get_asic_ss_info(struct radeon_device *rdev, struct radeon_atom_ss *ss, int id, u32 clock); +extern struct radeon_gpio_rec radeon_atombios_lookup_gpio(struct radeon_device *rdev, + u8 id); extern void radeon_compute_pll_legacy(struct radeon_pll *pll, uint64_t freq, @@ -802,13 +809,16 @@ extern int radeon_crtc_set_base_atomic(struct drm_crtc *crtc, extern int radeon_crtc_do_set_base(struct drm_crtc *crtc, struct drm_framebuffer *fb, int x, int y, int atomic); -extern int radeon_crtc_cursor_set(struct drm_crtc *crtc, - struct drm_file *file_priv, - uint32_t handle, - uint32_t width, - uint32_t height); +extern int radeon_crtc_cursor_set2(struct drm_crtc *crtc, + struct drm_file *file_priv, + uint32_t handle, + uint32_t width, + uint32_t height, + int32_t hot_x, + int32_t hot_y); extern int radeon_crtc_cursor_move(struct drm_crtc *crtc, int x, int y); +extern void radeon_cursor_reset(struct drm_crtc *crtc); extern int radeon_get_crtc_scanoutpos(struct drm_device *dev, int crtc, unsigned int flags, diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 4c0d786..7d68223 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -99,22 +99,39 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) rbo->placement.placement = rbo->placements; rbo->placement.busy_placement = rbo->placements; - if (domain & RADEON_GEM_DOMAIN_VRAM) + if (domain & RADEON_GEM_DOMAIN_VRAM) { + /* Try placing BOs which don't need CPU access outside of the + * CPU accessible part of VRAM + */ + if ((rbo->flags & RADEON_GEM_NO_CPU_ACCESS) && + rbo->rdev->mc.visible_vram_size < rbo->rdev->mc.real_vram_size) { + rbo->placements[c].fpfn = + rbo->rdev->mc.visible_vram_size >> PAGE_SHIFT; + rbo->placements[c++].flags = TTM_PL_FLAG_WC | + TTM_PL_FLAG_UNCACHED | + TTM_PL_FLAG_VRAM; + } + + rbo->placements[c].fpfn = 0; rbo->placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_VRAM; + } if (domain & RADEON_GEM_DOMAIN_GTT) { if (rbo->flags & RADEON_GEM_GTT_UC) { + rbo->placements[c].fpfn = 0; rbo->placements[c++].flags = TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_TT; } else if ((rbo->flags & RADEON_GEM_GTT_WC) || (rbo->rdev->flags & RADEON_IS_AGP)) { + rbo->placements[c].fpfn = 0; rbo->placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_TT; } else { + rbo->placements[c].fpfn = 0; rbo->placements[c++].flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_TT; } @@ -122,30 +139,35 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) if (domain & RADEON_GEM_DOMAIN_CPU) { if (rbo->flags & RADEON_GEM_GTT_UC) { + rbo->placements[c].fpfn = 0; rbo->placements[c++].flags = TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_SYSTEM; } else if ((rbo->flags & RADEON_GEM_GTT_WC) || rbo->rdev->flags & RADEON_IS_AGP) { + rbo->placements[c].fpfn = 0; rbo->placements[c++].flags = TTM_PL_FLAG_WC | TTM_PL_FLAG_UNCACHED | TTM_PL_FLAG_SYSTEM; } else { + rbo->placements[c].fpfn = 0; rbo->placements[c++].flags = TTM_PL_FLAG_CACHED | TTM_PL_FLAG_SYSTEM; } } - if (!c) + if (!c) { + rbo->placements[c].fpfn = 0; rbo->placements[c++].flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM; + } rbo->placement.num_placement = c; rbo->placement.num_busy_placement = c; for (i = 0; i < c; ++i) { - rbo->placements[i].fpfn = 0; if ((rbo->flags & RADEON_GEM_CPU_ACCESS) && - (rbo->placements[i].flags & TTM_PL_FLAG_VRAM)) + (rbo->placements[i].flags & TTM_PL_FLAG_VRAM) && + !rbo->placements[i].fpfn) rbo->placements[i].lpfn = rbo->rdev->mc.visible_vram_size >> PAGE_SHIFT; else @@ -157,9 +179,7 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) * improve fragmentation quality. * 512kb was measured as the most optimal number. */ - if (!((rbo->flags & RADEON_GEM_CPU_ACCESS) && - (rbo->placements[i].flags & TTM_PL_FLAG_VRAM)) && - rbo->tbo.mem.size > 512 * 1024) { + if (rbo->tbo.mem.size > 512 * 1024) { for (i = 0; i < c; i++) { rbo->placements[i].flags |= TTM_PL_FLAG_TOPDOWN; } @@ -489,25 +509,29 @@ int radeon_bo_list_validate(struct radeon_device *rdev, struct ww_acquire_ctx *ticket, struct list_head *head, int ring) { - struct radeon_cs_reloc *lobj; - struct radeon_bo *bo; + struct radeon_bo_list *lobj; + struct list_head duplicates; int r; u64 bytes_moved = 0, initial_bytes_moved; u64 bytes_moved_threshold = radeon_bo_get_threshold_for_moves(rdev); - r = ttm_eu_reserve_buffers(ticket, head, true); + INIT_LIST_HEAD(&duplicates); + r = ttm_eu_reserve_buffers(ticket, head, true, &duplicates); if (unlikely(r != 0)) { return r; } list_for_each_entry(lobj, head, tv.head) { - bo = lobj->robj; + struct radeon_bo *bo = lobj->robj; if (!bo->pin_count) { u32 domain = lobj->prefered_domains; u32 allowed = lobj->allowed_domains; u32 current_domain = radeon_mem_type_to_domain(bo->tbo.mem.mem_type); + WARN_ONCE(bo->gem_base.dumb, + "GPU use of dumb buffer is illegal.\n"); + /* Check if this buffer will be moved and don't move it * if we have moved too many buffers for this IB already. * @@ -546,6 +570,12 @@ int radeon_bo_list_validate(struct radeon_device *rdev, lobj->gpu_offset = radeon_bo_gpu_offset(bo); lobj->tiling_flags = bo->tiling_flags; } + + list_for_each_entry(lobj, &duplicates, tv.head) { + lobj->gpu_offset = radeon_bo_gpu_offset(lobj->robj); + lobj->tiling_flags = lobj->robj->tiling_flags; + } + return 0; } @@ -750,8 +780,8 @@ int radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo) { struct radeon_device *rdev; struct radeon_bo *rbo; - unsigned long offset, size; - int r; + unsigned long offset, size, lpfn; + int i, r; if (!radeon_ttm_bo_is_radeon_bo(bo)) return 0; @@ -768,7 +798,13 @@ int radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo) /* hurrah the memory is not visible ! */ radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_VRAM); - rbo->placements[0].lpfn = rdev->mc.visible_vram_size >> PAGE_SHIFT; + lpfn = rdev->mc.visible_vram_size >> PAGE_SHIFT; + for (i = 0; i < rbo->placement.num_placement; i++) { + /* Force into visible VRAM */ + if ((rbo->placements[i].flags & TTM_PL_FLAG_VRAM) && + (!rbo->placements[i].lpfn || rbo->placements[i].lpfn > lpfn)) + rbo->placements[i].lpfn = lpfn; + } r = ttm_bo_validate(bo, &rbo->placement, false, false); if (unlikely(r == -ENOMEM)) { radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_GTT); @@ -799,3 +835,22 @@ int radeon_bo_wait(struct radeon_bo *bo, u32 *mem_type, bool no_wait) ttm_bo_unreserve(&bo->tbo); return r; } + +/** + * radeon_bo_fence - add fence to buffer object + * + * @bo: buffer object in question + * @fence: fence to add + * @shared: true if fence should be added shared + * + */ +void radeon_bo_fence(struct radeon_bo *bo, struct radeon_fence *fence, + bool shared) +{ + struct reservation_object *resv = bo->tbo.resv; + + if (shared) + reservation_object_add_shared_fence(resv, &fence->base); + else + reservation_object_add_excl_fence(resv, &fence->base); +} diff --git a/drivers/gpu/drm/radeon/radeon_object.h b/drivers/gpu/drm/radeon/radeon_object.h index 1b8ec79..3b0b377 100644 --- a/drivers/gpu/drm/radeon/radeon_object.h +++ b/drivers/gpu/drm/radeon/radeon_object.h @@ -155,6 +155,8 @@ extern void radeon_bo_move_notify(struct ttm_buffer_object *bo, struct ttm_mem_reg *new_mem); extern int radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo); extern int radeon_bo_get_surface_reg(struct radeon_bo *bo); +extern void radeon_bo_fence(struct radeon_bo *bo, struct radeon_fence *fence, + bool shared); /* * sub allocation diff --git a/drivers/gpu/drm/radeon/radeon_semaphore.c b/drivers/gpu/drm/radeon/radeon_semaphore.c index 6deb08f..e6ad54c 100644 --- a/drivers/gpu/drm/radeon/radeon_semaphore.c +++ b/drivers/gpu/drm/radeon/radeon_semaphore.c @@ -34,15 +34,14 @@ int radeon_semaphore_create(struct radeon_device *rdev, struct radeon_semaphore **semaphore) { - uint64_t *cpu_addr; - int i, r; + int r; *semaphore = kmalloc(sizeof(struct radeon_semaphore), GFP_KERNEL); if (*semaphore == NULL) { return -ENOMEM; } - r = radeon_sa_bo_new(rdev, &rdev->ring_tmp_bo, &(*semaphore)->sa_bo, - 8 * RADEON_NUM_SYNCS, 8); + r = radeon_sa_bo_new(rdev, &rdev->ring_tmp_bo, + &(*semaphore)->sa_bo, 8, 8); if (r) { kfree(*semaphore); *semaphore = NULL; @@ -51,12 +50,7 @@ int radeon_semaphore_create(struct radeon_device *rdev, (*semaphore)->waiters = 0; (*semaphore)->gpu_addr = radeon_sa_bo_gpu_addr((*semaphore)->sa_bo); - cpu_addr = radeon_sa_bo_cpu_addr((*semaphore)->sa_bo); - for (i = 0; i < RADEON_NUM_SYNCS; ++i) - cpu_addr[i] = 0; - - for (i = 0; i < RADEON_NUM_RINGS; ++i) - (*semaphore)->sync_to[i] = NULL; + *((uint64_t *)radeon_sa_bo_cpu_addr((*semaphore)->sa_bo)) = 0; return 0; } @@ -95,146 +89,6 @@ bool radeon_semaphore_emit_wait(struct radeon_device *rdev, int ridx, return false; } -/** - * radeon_semaphore_sync_fence - use the semaphore to sync to a fence - * - * @semaphore: semaphore object to add fence to - * @fence: fence to sync to - * - * Sync to the fence using this semaphore object - */ -void radeon_semaphore_sync_fence(struct radeon_semaphore *semaphore, - struct radeon_fence *fence) -{ - struct radeon_fence *other; - - if (!fence) - return; - - other = semaphore->sync_to[fence->ring]; - semaphore->sync_to[fence->ring] = radeon_fence_later(fence, other); -} - -/** - * radeon_semaphore_sync_to - use the semaphore to sync to a reservation object - * - * @sema: semaphore object to add fence from reservation object to - * @resv: reservation object with embedded fence - * @shared: true if we should onyl sync to the exclusive fence - * - * Sync to the fence using this semaphore object - */ -int radeon_semaphore_sync_resv(struct radeon_device *rdev, - struct radeon_semaphore *sema, - struct reservation_object *resv, - bool shared) -{ - struct reservation_object_list *flist; - struct fence *f; - struct radeon_fence *fence; - unsigned i; - int r = 0; - - /* always sync to the exclusive fence */ - f = reservation_object_get_excl(resv); - fence = f ? to_radeon_fence(f) : NULL; - if (fence && fence->rdev == rdev) - radeon_semaphore_sync_fence(sema, fence); - else if (f) - r = fence_wait(f, true); - - flist = reservation_object_get_list(resv); - if (shared || !flist || r) - return r; - - for (i = 0; i < flist->shared_count; ++i) { - f = rcu_dereference_protected(flist->shared[i], - reservation_object_held(resv)); - fence = to_radeon_fence(f); - if (fence && fence->rdev == rdev) - radeon_semaphore_sync_fence(sema, fence); - else - r = fence_wait(f, true); - - if (r) - break; - } - return r; -} - -/** - * radeon_semaphore_sync_rings - sync ring to all registered fences - * - * @rdev: radeon_device pointer - * @semaphore: semaphore object to use for sync - * @ring: ring that needs sync - * - * Ensure that all registered fences are signaled before letting - * the ring continue. The caller must hold the ring lock. - */ -int radeon_semaphore_sync_rings(struct radeon_device *rdev, - struct radeon_semaphore *semaphore, - int ring) -{ - unsigned count = 0; - int i, r; - - for (i = 0; i < RADEON_NUM_RINGS; ++i) { - struct radeon_fence *fence = semaphore->sync_to[i]; - - /* check if we really need to sync */ - if (!radeon_fence_need_sync(fence, ring)) - continue; - - /* prevent GPU deadlocks */ - if (!rdev->ring[i].ready) { - dev_err(rdev->dev, "Syncing to a disabled ring!"); - return -EINVAL; - } - - if (++count > RADEON_NUM_SYNCS) { - /* not enough room, wait manually */ - r = radeon_fence_wait(fence, false); - if (r) - return r; - continue; - } - - /* allocate enough space for sync command */ - r = radeon_ring_alloc(rdev, &rdev->ring[i], 16); - if (r) { - return r; - } - - /* emit the signal semaphore */ - if (!radeon_semaphore_emit_signal(rdev, i, semaphore)) { - /* signaling wasn't successful wait manually */ - radeon_ring_undo(&rdev->ring[i]); - r = radeon_fence_wait(fence, false); - if (r) - return r; - continue; - } - - /* we assume caller has already allocated space on waiters ring */ - if (!radeon_semaphore_emit_wait(rdev, ring, semaphore)) { - /* waiting wasn't successful wait manually */ - radeon_ring_undo(&rdev->ring[i]); - r = radeon_fence_wait(fence, false); - if (r) - return r; - continue; - } - - radeon_ring_commit(rdev, &rdev->ring[i], false); - radeon_fence_note_sync(fence, ring); - - semaphore->gpu_addr += 8; - } - - return 0; -} - void radeon_semaphore_free(struct radeon_device *rdev, struct radeon_semaphore **semaphore, struct radeon_fence *fence) diff --git a/drivers/gpu/drm/radeon/radeon_sync.c b/drivers/gpu/drm/radeon/radeon_sync.c new file mode 100644 index 0000000..02ac8a1 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_sync.c @@ -0,0 +1,220 @@ +/* + * Copyright 2014 Advanced Micro Devices, Inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + */ +/* + * Authors: + * Christian König <christian.koenig@amd.com> + */ + +#include <drm/drmP.h> +#include "radeon.h" +#include "radeon_trace.h" + +/** + * radeon_sync_create - zero init sync object + * + * @sync: sync object to initialize + * + * Just clear the sync object for now. + */ +void radeon_sync_create(struct radeon_sync *sync) +{ + unsigned i; + + for (i = 0; i < RADEON_NUM_SYNCS; ++i) + sync->semaphores[i] = NULL; + + for (i = 0; i < RADEON_NUM_RINGS; ++i) + sync->sync_to[i] = NULL; + + sync->last_vm_update = NULL; +} + +/** + * radeon_sync_fence - use the semaphore to sync to a fence + * + * @sync: sync object to add fence to + * @fence: fence to sync to + * + * Sync to the fence using the semaphore objects + */ +void radeon_sync_fence(struct radeon_sync *sync, + struct radeon_fence *fence) +{ + struct radeon_fence *other; + + if (!fence) + return; + + other = sync->sync_to[fence->ring]; + sync->sync_to[fence->ring] = radeon_fence_later(fence, other); + + if (fence->is_vm_update) { + other = sync->last_vm_update; + sync->last_vm_update = radeon_fence_later(fence, other); + } +} + +/** + * radeon_sync_resv - use the semaphores to sync to a reservation object + * + * @sync: sync object to add fences from reservation object to + * @resv: reservation object with embedded fence + * @shared: true if we should only sync to the exclusive fence + * + * Sync to the fence using the semaphore objects + */ +int radeon_sync_resv(struct radeon_device *rdev, + struct radeon_sync *sync, + struct reservation_object *resv, + bool shared) +{ + struct reservation_object_list *flist; + struct fence *f; + struct radeon_fence *fence; + unsigned i; + int r = 0; + + /* always sync to the exclusive fence */ + f = reservation_object_get_excl(resv); + fence = f ? to_radeon_fence(f) : NULL; + if (fence && fence->rdev == rdev) + radeon_sync_fence(sync, fence); + else if (f) + r = fence_wait(f, true); + + flist = reservation_object_get_list(resv); + if (shared || !flist || r) + return r; + + for (i = 0; i < flist->shared_count; ++i) { + f = rcu_dereference_protected(flist->shared[i], + reservation_object_held(resv)); + fence = to_radeon_fence(f); + if (fence && fence->rdev == rdev) + radeon_sync_fence(sync, fence); + else + r = fence_wait(f, true); + + if (r) + break; + } + return r; +} + +/** + * radeon_sync_rings - sync ring to all registered fences + * + * @rdev: radeon_device pointer + * @sync: sync object to use + * @ring: ring that needs sync + * + * Ensure that all registered fences are signaled before letting + * the ring continue. The caller must hold the ring lock. + */ +int radeon_sync_rings(struct radeon_device *rdev, + struct radeon_sync *sync, + int ring) +{ + unsigned count = 0; + int i, r; + + for (i = 0; i < RADEON_NUM_RINGS; ++i) { + struct radeon_fence *fence = sync->sync_to[i]; + struct radeon_semaphore *semaphore; + + /* check if we really need to sync */ + if (!radeon_fence_need_sync(fence, ring)) + continue; + + /* prevent GPU deadlocks */ + if (!rdev->ring[i].ready) { + dev_err(rdev->dev, "Syncing to a disabled ring!"); + return -EINVAL; + } + + if (count >= RADEON_NUM_SYNCS) { + /* not enough room, wait manually */ + r = radeon_fence_wait(fence, false); + if (r) + return r; + continue; + } + r = radeon_semaphore_create(rdev, &semaphore); + if (r) + return r; + + sync->semaphores[count++] = semaphore; + + /* allocate enough space for sync command */ + r = radeon_ring_alloc(rdev, &rdev->ring[i], 16); + if (r) + return r; + + /* emit the signal semaphore */ + if (!radeon_semaphore_emit_signal(rdev, i, semaphore)) { + /* signaling wasn't successful wait manually */ + radeon_ring_undo(&rdev->ring[i]); + r = radeon_fence_wait(fence, false); + if (r) + return r; + continue; + } + + /* we assume caller has already allocated space on waiters ring */ + if (!radeon_semaphore_emit_wait(rdev, ring, semaphore)) { + /* waiting wasn't successful wait manually */ + radeon_ring_undo(&rdev->ring[i]); + r = radeon_fence_wait(fence, false); + if (r) + return r; + continue; + } + + radeon_ring_commit(rdev, &rdev->ring[i], false); + radeon_fence_note_sync(fence, ring); + } + + return 0; +} + +/** + * radeon_sync_free - free the sync object + * + * @rdev: radeon_device pointer + * @sync: sync object to use + * @fence: fence to use for the free + * + * Free the sync object by freeing all semaphores in it. + */ +void radeon_sync_free(struct radeon_device *rdev, + struct radeon_sync *sync, + struct radeon_fence *fence) +{ + unsigned i; + + for (i = 0; i < RADEON_NUM_SYNCS; ++i) + radeon_semaphore_free(rdev, &sync->semaphores[i], fence); +} diff --git a/drivers/gpu/drm/radeon/radeon_trace.h b/drivers/gpu/drm/radeon/radeon_trace.h index 9db74a9..ce075cb 100644 --- a/drivers/gpu/drm/radeon/radeon_trace.h +++ b/drivers/gpu/drm/radeon/radeon_trace.h @@ -38,7 +38,7 @@ TRACE_EVENT(radeon_cs, TP_fast_assign( __entry->ring = p->ring; - __entry->dw = p->chunks[p->chunk_ib_idx].length_dw; + __entry->dw = p->chunk_ib->length_dw; __entry->fences = radeon_fence_count_emitted( p->rdev, p->ring); ), diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index 8624979..d02aa1d 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -196,9 +196,32 @@ static void radeon_evict_flags(struct ttm_buffer_object *bo, rbo = container_of(bo, struct radeon_bo, tbo); switch (bo->mem.mem_type) { case TTM_PL_VRAM: - if (rbo->rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready == false) + if (rbo->rdev->ring[radeon_copy_ring_index(rbo->rdev)].ready == false) radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_CPU); - else + else if (rbo->rdev->mc.visible_vram_size < rbo->rdev->mc.real_vram_size && + bo->mem.start < (rbo->rdev->mc.visible_vram_size >> PAGE_SHIFT)) { + unsigned fpfn = rbo->rdev->mc.visible_vram_size >> PAGE_SHIFT; + int i; + + /* Try evicting to the CPU inaccessible part of VRAM + * first, but only set GTT as busy placement, so this + * BO will be evicted to GTT rather than causing other + * BOs to be evicted from VRAM + */ + radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_VRAM | + RADEON_GEM_DOMAIN_GTT); + rbo->placement.num_busy_placement = 0; + for (i = 0; i < rbo->placement.num_placement; i++) { + if (rbo->placements[i].flags & TTM_PL_FLAG_VRAM) { + if (rbo->placements[0].fpfn < fpfn) + rbo->placements[0].fpfn = fpfn; + } else { + rbo->placement.busy_placement = + &rbo->placements[i]; + rbo->placement.num_busy_placement = 1; + } + } + } else radeon_ttm_placement_from_domain(rbo, RADEON_GEM_DOMAIN_GTT); break; case TTM_PL_TT: diff --git a/drivers/gpu/drm/radeon/radeon_uvd.c b/drivers/gpu/drm/radeon/radeon_uvd.c index 11b6624..c10b2ae 100644 --- a/drivers/gpu/drm/radeon/radeon_uvd.c +++ b/drivers/gpu/drm/radeon/radeon_uvd.c @@ -488,12 +488,12 @@ static int radeon_uvd_cs_reloc(struct radeon_cs_parser *p, unsigned buf_sizes[], bool *has_msg_cmd) { struct radeon_cs_chunk *relocs_chunk; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; unsigned idx, cmd, offset; uint64_t start, end; int r; - relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + relocs_chunk = p->chunk_relocs; offset = radeon_get_ib_value(p, data0); idx = radeon_get_ib_value(p, data1); if (idx >= relocs_chunk->length_dw) { @@ -502,7 +502,7 @@ static int radeon_uvd_cs_reloc(struct radeon_cs_parser *p, return -EINVAL; } - reloc = p->relocs_ptr[(idx / 4)]; + reloc = &p->relocs[(idx / 4)]; start = reloc->gpu_offset; end = start + radeon_bo_size(reloc->robj); start += offset; @@ -610,13 +610,13 @@ int radeon_uvd_cs_parse(struct radeon_cs_parser *p) [0x00000003] = 2048, }; - if (p->chunks[p->chunk_ib_idx].length_dw % 16) { + if (p->chunk_ib->length_dw % 16) { DRM_ERROR("UVD IB length (%d) not 16 dwords aligned!\n", - p->chunks[p->chunk_ib_idx].length_dw); + p->chunk_ib->length_dw); return -EINVAL; } - if (p->chunk_relocs_idx == -1) { + if (p->chunk_relocs == NULL) { DRM_ERROR("No relocation chunk !\n"); return -EINVAL; } @@ -640,7 +640,7 @@ int radeon_uvd_cs_parse(struct radeon_cs_parser *p) DRM_ERROR("Unknown packet type %d !\n", pkt.type); return -EINVAL; } - } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); + } while (p->idx < p->chunk_ib->length_dw); if (!has_msg_cmd) { DRM_ERROR("UVD-IBs need a msg command!\n"); diff --git a/drivers/gpu/drm/radeon/radeon_vce.c b/drivers/gpu/drm/radeon/radeon_vce.c index 9e85757..976fe43 100644 --- a/drivers/gpu/drm/radeon/radeon_vce.c +++ b/drivers/gpu/drm/radeon/radeon_vce.c @@ -453,11 +453,11 @@ int radeon_vce_cs_reloc(struct radeon_cs_parser *p, int lo, int hi, unsigned size) { struct radeon_cs_chunk *relocs_chunk; - struct radeon_cs_reloc *reloc; + struct radeon_bo_list *reloc; uint64_t start, end, offset; unsigned idx; - relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + relocs_chunk = p->chunk_relocs; offset = radeon_get_ib_value(p, lo); idx = radeon_get_ib_value(p, hi); @@ -467,7 +467,7 @@ int radeon_vce_cs_reloc(struct radeon_cs_parser *p, int lo, int hi, return -EINVAL; } - reloc = p->relocs_ptr[(idx / 4)]; + reloc = &p->relocs[(idx / 4)]; start = reloc->gpu_offset; end = start + radeon_bo_size(reloc->robj); start += offset; @@ -534,7 +534,7 @@ int radeon_vce_cs_parse(struct radeon_cs_parser *p) uint32_t *size = &tmp; int i, r; - while (p->idx < p->chunks[p->chunk_ib_idx].length_dw) { + while (p->idx < p->chunk_ib->length_dw) { uint32_t len = radeon_get_ib_value(p, p->idx); uint32_t cmd = radeon_get_ib_value(p, p->idx + 1); diff --git a/drivers/gpu/drm/radeon/radeon_vm.c b/drivers/gpu/drm/radeon/radeon_vm.c index dfde266..cde48c4 100644 --- a/drivers/gpu/drm/radeon/radeon_vm.c +++ b/drivers/gpu/drm/radeon/radeon_vm.c @@ -125,41 +125,37 @@ void radeon_vm_manager_fini(struct radeon_device *rdev) * Add the page directory to the list of BOs to * validate for command submission (cayman+). */ -struct radeon_cs_reloc *radeon_vm_get_bos(struct radeon_device *rdev, +struct radeon_bo_list *radeon_vm_get_bos(struct radeon_device *rdev, struct radeon_vm *vm, struct list_head *head) { - struct radeon_cs_reloc *list; + struct radeon_bo_list *list; unsigned i, idx; list = drm_malloc_ab(vm->max_pde_used + 2, - sizeof(struct radeon_cs_reloc)); + sizeof(struct radeon_bo_list)); if (!list) return NULL; /* add the vm page table to the list */ - list[0].gobj = NULL; list[0].robj = vm->page_directory; list[0].prefered_domains = RADEON_GEM_DOMAIN_VRAM; list[0].allowed_domains = RADEON_GEM_DOMAIN_VRAM; list[0].tv.bo = &vm->page_directory->tbo; - list[0].tv.shared = false; + list[0].tv.shared = true; list[0].tiling_flags = 0; - list[0].handle = 0; list_add(&list[0].tv.head, head); for (i = 0, idx = 1; i <= vm->max_pde_used; i++) { if (!vm->page_tables[i].bo) continue; - list[idx].gobj = NULL; list[idx].robj = vm->page_tables[i].bo; list[idx].prefered_domains = RADEON_GEM_DOMAIN_VRAM; list[idx].allowed_domains = RADEON_GEM_DOMAIN_VRAM; list[idx].tv.bo = &list[idx].robj->tbo; - list[idx].tv.shared = false; + list[idx].tv.shared = true; list[idx].tiling_flags = 0; - list[idx].handle = 0; list_add(&list[idx++].tv.head, head); } @@ -182,15 +178,18 @@ struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, struct radeon_vm *vm, int ring) { struct radeon_fence *best[RADEON_NUM_RINGS] = {}; + struct radeon_vm_id *vm_id = &vm->ids[ring]; + unsigned choices[2] = {}; unsigned i; /* check if the id is still valid */ - if (vm->last_id_use && vm->last_id_use == rdev->vm_manager.active[vm->id]) + if (vm_id->id && vm_id->last_id_use && + vm_id->last_id_use == rdev->vm_manager.active[vm_id->id]) return NULL; /* we definately need to flush */ - radeon_fence_unref(&vm->last_flush); + vm_id->pd_gpu_addr = ~0ll; /* skip over VMID 0, since it is the system VM */ for (i = 1; i < rdev->vm_manager.nvm; ++i) { @@ -198,8 +197,8 @@ struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, if (fence == NULL) { /* found a free one */ - vm->id = i; - trace_radeon_vm_grab_id(vm->id, ring); + vm_id->id = i; + trace_radeon_vm_grab_id(i, ring); return NULL; } @@ -211,8 +210,8 @@ struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, for (i = 0; i < 2; ++i) { if (choices[i]) { - vm->id = choices[i]; - trace_radeon_vm_grab_id(vm->id, ring); + vm_id->id = choices[i]; + trace_radeon_vm_grab_id(choices[i], ring); return rdev->vm_manager.active[choices[i]]; } } @@ -228,6 +227,7 @@ struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, * @rdev: radeon_device pointer * @vm: vm we want to flush * @ring: ring to use for flush + * @updates: last vm update that is waited for * * Flush the vm (cayman+). * @@ -235,15 +235,21 @@ struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, */ void radeon_vm_flush(struct radeon_device *rdev, struct radeon_vm *vm, - int ring) + int ring, struct radeon_fence *updates) { uint64_t pd_addr = radeon_bo_gpu_offset(vm->page_directory); + struct radeon_vm_id *vm_id = &vm->ids[ring]; + + if (pd_addr != vm_id->pd_gpu_addr || !vm_id->flushed_updates || + radeon_fence_is_earlier(vm_id->flushed_updates, updates)) { + + trace_radeon_vm_flush(pd_addr, ring, vm->ids[ring].id); + radeon_fence_unref(&vm_id->flushed_updates); + vm_id->flushed_updates = radeon_fence_ref(updates); + vm_id->pd_gpu_addr = pd_addr; + radeon_ring_vm_flush(rdev, &rdev->ring[ring], + vm_id->id, vm_id->pd_gpu_addr); - /* if we can't remember our last VM flush then flush now! */ - if (!vm->last_flush || pd_addr != vm->pd_gpu_addr) { - trace_radeon_vm_flush(pd_addr, ring, vm->id); - vm->pd_gpu_addr = pd_addr; - radeon_ring_vm_flush(rdev, ring, vm); } } @@ -263,18 +269,13 @@ void radeon_vm_fence(struct radeon_device *rdev, struct radeon_vm *vm, struct radeon_fence *fence) { - radeon_fence_unref(&vm->fence); - vm->fence = radeon_fence_ref(fence); - - radeon_fence_unref(&rdev->vm_manager.active[vm->id]); - rdev->vm_manager.active[vm->id] = radeon_fence_ref(fence); + unsigned vm_id = vm->ids[fence->ring].id; - radeon_fence_unref(&vm->last_id_use); - vm->last_id_use = radeon_fence_ref(fence); + radeon_fence_unref(&rdev->vm_manager.active[vm_id]); + rdev->vm_manager.active[vm_id] = radeon_fence_ref(fence); - /* we just flushed the VM, remember that */ - if (!vm->last_flush) - vm->last_flush = radeon_fence_ref(fence); + radeon_fence_unref(&vm->ids[fence->ring].last_id_use); + vm->ids[fence->ring].last_id_use = radeon_fence_ref(fence); } /** @@ -387,35 +388,25 @@ static void radeon_vm_set_pages(struct radeon_device *rdev, static int radeon_vm_clear_bo(struct radeon_device *rdev, struct radeon_bo *bo) { - struct ttm_validate_buffer tv; - struct ww_acquire_ctx ticket; - struct list_head head; struct radeon_ib ib; unsigned entries; uint64_t addr; int r; - memset(&tv, 0, sizeof(tv)); - tv.bo = &bo->tbo; - tv.shared = false; - - INIT_LIST_HEAD(&head); - list_add(&tv.head, &head); - - r = ttm_eu_reserve_buffers(&ticket, &head, true); - if (r) + r = radeon_bo_reserve(bo, false); + if (r) return r; - r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false); - if (r) - goto error; + r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false); + if (r) + goto error_unreserve; addr = radeon_bo_gpu_offset(bo); entries = radeon_bo_size(bo) / 8; r = radeon_ib_get(rdev, R600_RING_TYPE_DMA_INDEX, &ib, NULL, 256); if (r) - goto error; + goto error_unreserve; ib.length_dw = 0; @@ -425,15 +416,16 @@ static int radeon_vm_clear_bo(struct radeon_device *rdev, r = radeon_ib_schedule(rdev, &ib, NULL, false); if (r) - goto error; + goto error_free; - ttm_eu_fence_buffer_objects(&ticket, &head, &ib.fence->base); - radeon_ib_free(rdev, &ib); + ib.fence->is_vm_update = true; + radeon_bo_fence(bo, ib.fence, false); - return 0; +error_free: + radeon_ib_free(rdev, &ib); -error: - ttm_eu_backoff_reservation(&ticket, &head); +error_unreserve: + radeon_bo_unreserve(bo); return r; } @@ -449,7 +441,7 @@ error: * Validate and set the offset requested within the vm address space. * Returns 0 for success, error for failure. * - * Object has to be reserved! + * Object has to be reserved and gets unreserved by this function! */ int radeon_vm_bo_set_addr(struct radeon_device *rdev, struct radeon_bo_va *bo_va, @@ -495,7 +487,9 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, tmp->vm = vm; tmp->addr = bo_va->addr; tmp->bo = radeon_bo_ref(bo_va->bo); + spin_lock(&vm->status_lock); list_add(&tmp->vm_status, &vm->freed); + spin_unlock(&vm->status_lock); } interval_tree_remove(&bo_va->it, &vm->va); @@ -575,7 +569,7 @@ int radeon_vm_bo_set_addr(struct radeon_device *rdev, } mutex_unlock(&vm->mutex); - return radeon_bo_reserve(bo_va->bo, false); + return 0; } /** @@ -699,17 +693,15 @@ int radeon_vm_update_page_directory(struct radeon_device *rdev, if (ib.length_dw != 0) { radeon_asic_vm_pad_ib(rdev, &ib); - radeon_semaphore_sync_resv(rdev, ib.semaphore, pd->tbo.resv, false); - radeon_semaphore_sync_fence(ib.semaphore, vm->last_id_use); + radeon_sync_resv(rdev, &ib.sync, pd->tbo.resv, true); WARN_ON(ib.length_dw > ndw); r = radeon_ib_schedule(rdev, &ib, NULL, false); if (r) { radeon_ib_free(rdev, &ib); return r; } - radeon_fence_unref(&vm->fence); - vm->fence = radeon_fence_ref(ib.fence); - radeon_fence_unref(&vm->last_flush); + ib.fence->is_vm_update = true; + radeon_bo_fence(pd, ib.fence, false); } radeon_ib_free(rdev, &ib); @@ -808,11 +800,11 @@ static void radeon_vm_frag_ptes(struct radeon_device *rdev, * * Global and local mutex must be locked! */ -static void radeon_vm_update_ptes(struct radeon_device *rdev, - struct radeon_vm *vm, - struct radeon_ib *ib, - uint64_t start, uint64_t end, - uint64_t dst, uint32_t flags) +static int radeon_vm_update_ptes(struct radeon_device *rdev, + struct radeon_vm *vm, + struct radeon_ib *ib, + uint64_t start, uint64_t end, + uint64_t dst, uint32_t flags) { uint64_t mask = RADEON_VM_PTE_COUNT - 1; uint64_t last_pte = ~0, last_dst = ~0; @@ -825,8 +817,12 @@ static void radeon_vm_update_ptes(struct radeon_device *rdev, struct radeon_bo *pt = vm->page_tables[pt_idx].bo; unsigned nptes; uint64_t pte; + int r; - radeon_semaphore_sync_resv(rdev, ib->semaphore, pt->tbo.resv, false); + radeon_sync_resv(rdev, &ib->sync, pt->tbo.resv, true); + r = reservation_object_reserve_shared(pt->tbo.resv); + if (r) + return r; if ((addr & ~mask) == (end & ~mask)) nptes = end - addr; @@ -860,6 +856,33 @@ static void radeon_vm_update_ptes(struct radeon_device *rdev, last_pte + 8 * count, last_dst, flags); } + + return 0; +} + +/** + * radeon_vm_fence_pts - fence page tables after an update + * + * @vm: requested vm + * @start: start of GPU address range + * @end: end of GPU address range + * @fence: fence to use + * + * Fence the page tables in the range @start - @end (cayman+). + * + * Global and local mutex must be locked! + */ +static void radeon_vm_fence_pts(struct radeon_vm *vm, + uint64_t start, uint64_t end, + struct radeon_fence *fence) +{ + unsigned i; + + start >>= radeon_vm_block_size; + end >>= radeon_vm_block_size; + + for (i = start; i <= end; ++i) + radeon_bo_fence(vm->page_tables[i].bo, fence, true); } /** @@ -892,7 +915,9 @@ int radeon_vm_bo_update(struct radeon_device *rdev, return -EINVAL; } + spin_lock(&vm->status_lock); list_del_init(&bo_va->vm_status); + spin_unlock(&vm->status_lock); bo_va->flags &= ~RADEON_VM_PAGE_VALID; bo_va->flags &= ~RADEON_VM_PAGE_SYSTEM; @@ -961,23 +986,34 @@ int radeon_vm_bo_update(struct radeon_device *rdev, return r; ib.length_dw = 0; - radeon_vm_update_ptes(rdev, vm, &ib, bo_va->it.start, - bo_va->it.last + 1, addr, - radeon_vm_page_flags(bo_va->flags)); + if (!(bo_va->flags & RADEON_VM_PAGE_VALID)) { + unsigned i; + + for (i = 0; i < RADEON_NUM_RINGS; ++i) + radeon_sync_fence(&ib.sync, vm->ids[i].last_id_use); + } + + r = radeon_vm_update_ptes(rdev, vm, &ib, bo_va->it.start, + bo_va->it.last + 1, addr, + radeon_vm_page_flags(bo_va->flags)); + if (r) { + radeon_ib_free(rdev, &ib); + return r; + } radeon_asic_vm_pad_ib(rdev, &ib); WARN_ON(ib.length_dw > ndw); - radeon_semaphore_sync_fence(ib.semaphore, vm->fence); r = radeon_ib_schedule(rdev, &ib, NULL, false); if (r) { radeon_ib_free(rdev, &ib); return r; } - radeon_fence_unref(&vm->fence); - vm->fence = radeon_fence_ref(ib.fence); + ib.fence->is_vm_update = true; + radeon_vm_fence_pts(vm, bo_va->it.start, bo_va->it.last + 1, ib.fence); + radeon_fence_unref(&bo_va->last_pt_update); + bo_va->last_pt_update = radeon_fence_ref(ib.fence); radeon_ib_free(rdev, &ib); - radeon_fence_unref(&vm->last_flush); return 0; } @@ -996,16 +1032,25 @@ int radeon_vm_bo_update(struct radeon_device *rdev, int radeon_vm_clear_freed(struct radeon_device *rdev, struct radeon_vm *vm) { - struct radeon_bo_va *bo_va, *tmp; + struct radeon_bo_va *bo_va; int r; - list_for_each_entry_safe(bo_va, tmp, &vm->freed, vm_status) { + spin_lock(&vm->status_lock); + while (!list_empty(&vm->freed)) { + bo_va = list_first_entry(&vm->freed, + struct radeon_bo_va, vm_status); + spin_unlock(&vm->status_lock); + r = radeon_vm_bo_update(rdev, bo_va, NULL); radeon_bo_unref(&bo_va->bo); + radeon_fence_unref(&bo_va->last_pt_update); kfree(bo_va); if (r) return r; + + spin_lock(&vm->status_lock); } + spin_unlock(&vm->status_lock); return 0; } @@ -1024,14 +1069,23 @@ int radeon_vm_clear_freed(struct radeon_device *rdev, int radeon_vm_clear_invalids(struct radeon_device *rdev, struct radeon_vm *vm) { - struct radeon_bo_va *bo_va, *tmp; + struct radeon_bo_va *bo_va; int r; - list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, vm_status) { + spin_lock(&vm->status_lock); + while (!list_empty(&vm->invalidated)) { + bo_va = list_first_entry(&vm->invalidated, + struct radeon_bo_va, vm_status); + spin_unlock(&vm->status_lock); + r = radeon_vm_bo_update(rdev, bo_va, NULL); if (r) return r; + + spin_lock(&vm->status_lock); } + spin_unlock(&vm->status_lock); + return 0; } @@ -1054,14 +1108,17 @@ void radeon_vm_bo_rmv(struct radeon_device *rdev, mutex_lock(&vm->mutex); interval_tree_remove(&bo_va->it, &vm->va); + spin_lock(&vm->status_lock); list_del(&bo_va->vm_status); if (bo_va->addr) { bo_va->bo = radeon_bo_ref(bo_va->bo); list_add(&bo_va->vm_status, &vm->freed); } else { + radeon_fence_unref(&bo_va->last_pt_update); kfree(bo_va); } + spin_unlock(&vm->status_lock); mutex_unlock(&vm->mutex); } @@ -1082,10 +1139,10 @@ void radeon_vm_bo_invalidate(struct radeon_device *rdev, list_for_each_entry(bo_va, &bo->va, bo_list) { if (bo_va->addr) { - mutex_lock(&bo_va->vm->mutex); + spin_lock(&bo_va->vm->status_lock); list_del(&bo_va->vm_status); list_add(&bo_va->vm_status, &bo_va->vm->invalidated); - mutex_unlock(&bo_va->vm->mutex); + spin_unlock(&bo_va->vm->status_lock); } } } @@ -1103,15 +1160,17 @@ int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) const unsigned align = min(RADEON_VM_PTB_ALIGN_SIZE, RADEON_VM_PTE_COUNT * 8); unsigned pd_size, pd_entries, pts_size; - int r; + int i, r; - vm->id = 0; vm->ib_bo_va = NULL; - vm->fence = NULL; - vm->last_flush = NULL; - vm->last_id_use = NULL; + for (i = 0; i < RADEON_NUM_RINGS; ++i) { + vm->ids[i].id = 0; + vm->ids[i].flushed_updates = NULL; + vm->ids[i].last_id_use = NULL; + } mutex_init(&vm->mutex); vm->va = RB_ROOT; + spin_lock_init(&vm->status_lock); INIT_LIST_HEAD(&vm->invalidated); INIT_LIST_HEAD(&vm->freed); @@ -1165,11 +1224,13 @@ void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm) if (!r) { list_del_init(&bo_va->bo_list); radeon_bo_unreserve(bo_va->bo); + radeon_fence_unref(&bo_va->last_pt_update); kfree(bo_va); } } list_for_each_entry_safe(bo_va, tmp, &vm->freed, vm_status) { radeon_bo_unref(&bo_va->bo); + radeon_fence_unref(&bo_va->last_pt_update); kfree(bo_va); } @@ -1179,9 +1240,10 @@ void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm) radeon_bo_unref(&vm->page_directory); - radeon_fence_unref(&vm->fence); - radeon_fence_unref(&vm->last_flush); - radeon_fence_unref(&vm->last_id_use); + for (i = 0; i < RADEON_NUM_RINGS; ++i) { + radeon_fence_unref(&vm->ids[i].flushed_updates); + radeon_fence_unref(&vm->ids[i].last_id_use); + } mutex_destroy(&vm->mutex); } diff --git a/drivers/gpu/drm/radeon/rv770_dma.c b/drivers/gpu/drm/radeon/rv770_dma.c index 7f34bad..acff6e0 100644 --- a/drivers/gpu/drm/radeon/rv770_dma.c +++ b/drivers/gpu/drm/radeon/rv770_dma.c @@ -44,31 +44,27 @@ struct radeon_fence *rv770_copy_dma(struct radeon_device *rdev, unsigned num_gpu_pages, struct reservation_object *resv) { - struct radeon_semaphore *sem = NULL; struct radeon_fence *fence; + struct radeon_sync sync; int ring_index = rdev->asic->copy.dma_ring_index; struct radeon_ring *ring = &rdev->ring[ring_index]; u32 size_in_dw, cur_size_in_dw; int i, num_loops; int r = 0; - r = radeon_semaphore_create(rdev, &sem); - if (r) { - DRM_ERROR("radeon: moving bo (%d).\n", r); - return ERR_PTR(r); - } + radeon_sync_create(&sync); size_in_dw = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT) / 4; num_loops = DIV_ROUND_UP(size_in_dw, 0xFFFF); r = radeon_ring_lock(rdev, ring, num_loops * 5 + 8); if (r) { DRM_ERROR("radeon: moving bo (%d).\n", r); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } - radeon_semaphore_sync_resv(rdev, sem, resv, false); - radeon_semaphore_sync_rings(rdev, sem, ring->idx); + radeon_sync_resv(rdev, &sync, resv, false); + radeon_sync_rings(rdev, &sync, ring->idx); for (i = 0; i < num_loops; i++) { cur_size_in_dw = size_in_dw; @@ -87,12 +83,12 @@ struct radeon_fence *rv770_copy_dma(struct radeon_device *rdev, r = radeon_fence_emit(rdev, &fence, ring->idx); if (r) { radeon_ring_unlock_undo(rdev, ring); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } radeon_ring_unlock_commit(rdev, ring, false); - radeon_semaphore_free(rdev, &sem, fence); + radeon_sync_free(rdev, &sync, fence); return fence; } diff --git a/drivers/gpu/drm/radeon/si.c b/drivers/gpu/drm/radeon/si.c index 7d5083d..60df444 100644 --- a/drivers/gpu/drm/radeon/si.c +++ b/drivers/gpu/drm/radeon/si.c @@ -3365,6 +3365,7 @@ void si_fence_ring_emit(struct radeon_device *rdev, void si_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) { struct radeon_ring *ring = &rdev->ring[ib->ring]; + unsigned vm_id = ib->vm ? ib->vm->ids[ib->ring].id : 0; u32 header; if (ib->is_const_ib) { @@ -3400,14 +3401,13 @@ void si_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) #endif (ib->gpu_addr & 0xFFFFFFFC)); radeon_ring_write(ring, upper_32_bits(ib->gpu_addr) & 0xFFFF); - radeon_ring_write(ring, ib->length_dw | - (ib->vm ? (ib->vm->id << 24) : 0)); + radeon_ring_write(ring, ib->length_dw | (vm_id << 24)); if (!ib->is_const_ib) { /* flush read cache over gart for this vmid */ radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1)); radeon_ring_write(ring, (CP_COHER_CNTL2 - PACKET3_SET_CONFIG_REG_START) >> 2); - radeon_ring_write(ring, ib->vm ? ib->vm->id : 0); + radeon_ring_write(ring, vm_id); radeon_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3)); radeon_ring_write(ring, PACKET3_TCL1_ACTION_ENA | PACKET3_TC_ACTION_ENA | @@ -5023,27 +5023,23 @@ static void si_vm_decode_fault(struct radeon_device *rdev, block, mc_id); } -void si_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) +void si_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr) { - struct radeon_ring *ring = &rdev->ring[ridx]; - - if (vm == NULL) - return; - /* write new base address */ radeon_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3)); radeon_ring_write(ring, (WRITE_DATA_ENGINE_SEL(1) | WRITE_DATA_DST_SEL(0))); - if (vm->id < 8) { + if (vm_id < 8) { radeon_ring_write(ring, - (VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm->id << 2)) >> 2); + (VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm_id << 2)) >> 2); } else { radeon_ring_write(ring, - (VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + ((vm->id - 8) << 2)) >> 2); + (VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + ((vm_id - 8) << 2)) >> 2); } radeon_ring_write(ring, 0); - radeon_ring_write(ring, vm->pd_gpu_addr >> 12); + radeon_ring_write(ring, pd_addr >> 12); /* flush hdp cache */ radeon_ring_write(ring, PACKET3(PACKET3_WRITE_DATA, 3)); @@ -5059,7 +5055,7 @@ void si_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) WRITE_DATA_DST_SEL(0))); radeon_ring_write(ring, VM_INVALIDATE_REQUEST >> 2); radeon_ring_write(ring, 0); - radeon_ring_write(ring, 1 << vm->id); + radeon_ring_write(ring, 1 << vm_id); /* sync PFP to ME, otherwise we might get invalid PFP reads */ radeon_ring_write(ring, PACKET3(PACKET3_PFP_SYNC_ME, 0)); diff --git a/drivers/gpu/drm/radeon/si_dma.c b/drivers/gpu/drm/radeon/si_dma.c index b58f12b..f5cc777 100644 --- a/drivers/gpu/drm/radeon/si_dma.c +++ b/drivers/gpu/drm/radeon/si_dma.c @@ -185,20 +185,17 @@ void si_dma_vm_set_pages(struct radeon_device *rdev, } } -void si_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) -{ - struct radeon_ring *ring = &rdev->ring[ridx]; - - if (vm == NULL) - return; +void si_dma_vm_flush(struct radeon_device *rdev, struct radeon_ring *ring, + unsigned vm_id, uint64_t pd_addr) +{ radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0)); - if (vm->id < 8) { - radeon_ring_write(ring, (0xf << 16) | ((VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm->id << 2)) >> 2)); + if (vm_id < 8) { + radeon_ring_write(ring, (0xf << 16) | ((VM_CONTEXT0_PAGE_TABLE_BASE_ADDR + (vm_id << 2)) >> 2)); } else { - radeon_ring_write(ring, (0xf << 16) | ((VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + ((vm->id - 8) << 2)) >> 2)); + radeon_ring_write(ring, (0xf << 16) | ((VM_CONTEXT8_PAGE_TABLE_BASE_ADDR + ((vm_id - 8) << 2)) >> 2)); } - radeon_ring_write(ring, vm->pd_gpu_addr >> 12); + radeon_ring_write(ring, pd_addr >> 12); /* flush hdp cache */ radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0)); @@ -208,7 +205,7 @@ void si_dma_vm_flush(struct radeon_device *rdev, int ridx, struct radeon_vm *vm) /* bits 0-7 are the VM contexts0-7 */ radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SRBM_WRITE, 0, 0, 0, 0)); radeon_ring_write(ring, (0xf << 16) | (VM_INVALIDATE_REQUEST >> 2)); - radeon_ring_write(ring, 1 << vm->id); + radeon_ring_write(ring, 1 << vm_id); } /** @@ -229,31 +226,27 @@ struct radeon_fence *si_copy_dma(struct radeon_device *rdev, unsigned num_gpu_pages, struct reservation_object *resv) { - struct radeon_semaphore *sem = NULL; struct radeon_fence *fence; + struct radeon_sync sync; int ring_index = rdev->asic->copy.dma_ring_index; struct radeon_ring *ring = &rdev->ring[ring_index]; u32 size_in_bytes, cur_size_in_bytes; int i, num_loops; int r = 0; - r = radeon_semaphore_create(rdev, &sem); - if (r) { - DRM_ERROR("radeon: moving bo (%d).\n", r); - return ERR_PTR(r); - } + radeon_sync_create(&sync); size_in_bytes = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT); num_loops = DIV_ROUND_UP(size_in_bytes, 0xfffff); r = radeon_ring_lock(rdev, ring, num_loops * 5 + 11); if (r) { DRM_ERROR("radeon: moving bo (%d).\n", r); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } - radeon_semaphore_sync_resv(rdev, sem, resv, false); - radeon_semaphore_sync_rings(rdev, sem, ring->idx); + radeon_sync_resv(rdev, &sync, resv, false); + radeon_sync_rings(rdev, &sync, ring->idx); for (i = 0; i < num_loops; i++) { cur_size_in_bytes = size_in_bytes; @@ -272,12 +265,12 @@ struct radeon_fence *si_copy_dma(struct radeon_device *rdev, r = radeon_fence_emit(rdev, &fence, ring->idx); if (r) { radeon_ring_unlock_undo(rdev, ring); - radeon_semaphore_free(rdev, &sem, NULL); + radeon_sync_free(rdev, &sync, NULL); return ERR_PTR(r); } radeon_ring_unlock_commit(rdev, ring, false); - radeon_semaphore_free(rdev, &sem, fence); + radeon_sync_free(rdev, &sync, fence); return fence; } diff --git a/drivers/gpu/drm/radeon/si_dpm.c b/drivers/gpu/drm/radeon/si_dpm.c index 676e6c2..32e354b 100644 --- a/drivers/gpu/drm/radeon/si_dpm.c +++ b/drivers/gpu/drm/radeon/si_dpm.c @@ -3398,6 +3398,15 @@ static int si_process_firmware_header(struct radeon_device *rdev) ret = si_read_smc_sram_dword(rdev, SISLANDS_SMC_FIRMWARE_HEADER_LOCATION + + SISLANDS_SMC_FIRMWARE_HEADER_fanTable, + &tmp, si_pi->sram_end); + if (ret) + return ret; + + si_pi->fan_table_start = tmp; + + ret = si_read_smc_sram_dword(rdev, + SISLANDS_SMC_FIRMWARE_HEADER_LOCATION + SISLANDS_SMC_FIRMWARE_HEADER_mcArbDramAutoRefreshTable, &tmp, si_pi->sram_end); if (ret) @@ -5817,8 +5826,33 @@ void si_dpm_setup_asic(struct radeon_device *rdev) si_enable_acpi_power_management(rdev); } -static int si_set_thermal_temperature_range(struct radeon_device *rdev, - int min_temp, int max_temp) +static int si_thermal_enable_alert(struct radeon_device *rdev, + bool enable) +{ + u32 thermal_int = RREG32(CG_THERMAL_INT); + + if (enable) { + PPSMC_Result result; + + thermal_int &= ~(THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW); + WREG32(CG_THERMAL_INT, thermal_int); + rdev->irq.dpm_thermal = false; + result = si_send_msg_to_smc(rdev, PPSMC_MSG_EnableThermalInterrupt); + if (result != PPSMC_Result_OK) { + DRM_DEBUG_KMS("Could not enable thermal interrupts.\n"); + return -EINVAL; + } + } else { + thermal_int |= THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW; + WREG32(CG_THERMAL_INT, thermal_int); + rdev->irq.dpm_thermal = true; + } + + return 0; +} + +static int si_thermal_set_temperature_range(struct radeon_device *rdev, + int min_temp, int max_temp) { int low_temp = 0 * 1000; int high_temp = 255 * 1000; @@ -5842,6 +5876,309 @@ static int si_set_thermal_temperature_range(struct radeon_device *rdev, return 0; } +static void si_fan_ctrl_set_static_mode(struct radeon_device *rdev, u32 mode) +{ + struct si_power_info *si_pi = si_get_pi(rdev); + u32 tmp; + + if (si_pi->fan_ctrl_is_in_default_mode) { + tmp = (RREG32(CG_FDO_CTRL2) & FDO_PWM_MODE_MASK) >> FDO_PWM_MODE_SHIFT; + si_pi->fan_ctrl_default_mode = tmp; + tmp = (RREG32(CG_FDO_CTRL2) & TMIN_MASK) >> TMIN_SHIFT; + si_pi->t_min = tmp; + si_pi->fan_ctrl_is_in_default_mode = false; + } + + tmp = RREG32(CG_FDO_CTRL2) & ~TMIN_MASK; + tmp |= TMIN(0); + WREG32(CG_FDO_CTRL2, tmp); + + tmp = RREG32(CG_FDO_CTRL2) & ~FDO_PWM_MODE_MASK; + tmp |= FDO_PWM_MODE(mode); + WREG32(CG_FDO_CTRL2, tmp); +} + +static int si_thermal_setup_fan_table(struct radeon_device *rdev) +{ + struct si_power_info *si_pi = si_get_pi(rdev); + PP_SIslands_FanTable fan_table = { FDO_MODE_HARDWARE }; + u32 duty100; + u32 t_diff1, t_diff2, pwm_diff1, pwm_diff2; + u16 fdo_min, slope1, slope2; + u32 reference_clock, tmp; + int ret; + u64 tmp64; + + if (!si_pi->fan_table_start) { + rdev->pm.dpm.fan.ucode_fan_control = false; + return 0; + } + + duty100 = (RREG32(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT; + + if (duty100 == 0) { + rdev->pm.dpm.fan.ucode_fan_control = false; + return 0; + } + + tmp64 = (u64)rdev->pm.dpm.fan.pwm_min * duty100; + do_div(tmp64, 10000); + fdo_min = (u16)tmp64; + + t_diff1 = rdev->pm.dpm.fan.t_med - rdev->pm.dpm.fan.t_min; + t_diff2 = rdev->pm.dpm.fan.t_high - rdev->pm.dpm.fan.t_med; + + pwm_diff1 = rdev->pm.dpm.fan.pwm_med - rdev->pm.dpm.fan.pwm_min; + pwm_diff2 = rdev->pm.dpm.fan.pwm_high - rdev->pm.dpm.fan.pwm_med; + + slope1 = (u16)((50 + ((16 * duty100 * pwm_diff1) / t_diff1)) / 100); + slope2 = (u16)((50 + ((16 * duty100 * pwm_diff2) / t_diff2)) / 100); + + fan_table.slope1 = cpu_to_be16(slope1); + fan_table.slope2 = cpu_to_be16(slope2); + + fan_table.fdo_min = cpu_to_be16(fdo_min); + + fan_table.hys_down = cpu_to_be16(rdev->pm.dpm.fan.t_hyst); + + fan_table.hys_up = cpu_to_be16(1); + + fan_table.hys_slope = cpu_to_be16(1); + + fan_table.temp_resp_lim = cpu_to_be16(5); + + reference_clock = radeon_get_xclk(rdev); + + fan_table.refresh_period = cpu_to_be32((rdev->pm.dpm.fan.cycle_delay * + reference_clock) / 1600); + + fan_table.fdo_max = cpu_to_be16((u16)duty100); + + tmp = (RREG32(CG_MULT_THERMAL_CTRL) & TEMP_SEL_MASK) >> TEMP_SEL_SHIFT; + fan_table.temp_src = (uint8_t)tmp; + + ret = si_copy_bytes_to_smc(rdev, + si_pi->fan_table_start, + (u8 *)(&fan_table), + sizeof(fan_table), + si_pi->sram_end); + + if (ret) { + DRM_ERROR("Failed to load fan table to the SMC."); + rdev->pm.dpm.fan.ucode_fan_control = false; + } + + return 0; +} + +static int si_fan_ctrl_start_smc_fan_control(struct radeon_device *rdev) +{ + PPSMC_Result ret; + + ret = si_send_msg_to_smc(rdev, PPSMC_StartFanControl); + if (ret == PPSMC_Result_OK) + return 0; + else + return -EINVAL; +} + +static int si_fan_ctrl_stop_smc_fan_control(struct radeon_device *rdev) +{ + PPSMC_Result ret; + + ret = si_send_msg_to_smc(rdev, PPSMC_StopFanControl); + if (ret == PPSMC_Result_OK) + return 0; + else + return -EINVAL; +} + +#if 0 +static int si_fan_ctrl_get_fan_speed_percent(struct radeon_device *rdev, + u32 *speed) +{ + u32 duty, duty100; + u64 tmp64; + + if (rdev->pm.no_fan) + return -ENOENT; + + duty100 = (RREG32(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT; + duty = (RREG32(CG_THERMAL_STATUS) & FDO_PWM_DUTY_MASK) >> FDO_PWM_DUTY_SHIFT; + + if (duty100 == 0) + return -EINVAL; + + tmp64 = (u64)duty * 100; + do_div(tmp64, duty100); + *speed = (u32)tmp64; + + if (*speed > 100) + *speed = 100; + + return 0; +} + +static int si_fan_ctrl_set_fan_speed_percent(struct radeon_device *rdev, + u32 speed) +{ + u32 tmp; + u32 duty, duty100; + u64 tmp64; + + if (rdev->pm.no_fan) + return -ENOENT; + + if (speed > 100) + return -EINVAL; + + if (rdev->pm.dpm.fan.ucode_fan_control) + si_fan_ctrl_stop_smc_fan_control(rdev); + + duty100 = (RREG32(CG_FDO_CTRL1) & FMAX_DUTY100_MASK) >> FMAX_DUTY100_SHIFT; + + if (duty100 == 0) + return -EINVAL; + + tmp64 = (u64)speed * duty100; + do_div(tmp64, 100); + duty = (u32)tmp64; + + tmp = RREG32(CG_FDO_CTRL0) & ~FDO_STATIC_DUTY_MASK; + tmp |= FDO_STATIC_DUTY(duty); + WREG32(CG_FDO_CTRL0, tmp); + + si_fan_ctrl_set_static_mode(rdev, FDO_PWM_MODE_STATIC); + + return 0; +} + +static int si_fan_ctrl_get_fan_speed_rpm(struct radeon_device *rdev, + u32 *speed) +{ + u32 tach_period; + u32 xclk = radeon_get_xclk(rdev); + + if (rdev->pm.no_fan) + return -ENOENT; + + if (rdev->pm.fan_pulses_per_revolution == 0) + return -ENOENT; + + tach_period = (RREG32(CG_TACH_STATUS) & TACH_PERIOD_MASK) >> TACH_PERIOD_SHIFT; + if (tach_period == 0) + return -ENOENT; + + *speed = 60 * xclk * 10000 / tach_period; + + return 0; +} + +static int si_fan_ctrl_set_fan_speed_rpm(struct radeon_device *rdev, + u32 speed) +{ + u32 tach_period, tmp; + u32 xclk = radeon_get_xclk(rdev); + + if (rdev->pm.no_fan) + return -ENOENT; + + if (rdev->pm.fan_pulses_per_revolution == 0) + return -ENOENT; + + if ((speed < rdev->pm.fan_min_rpm) || + (speed > rdev->pm.fan_max_rpm)) + return -EINVAL; + + if (rdev->pm.dpm.fan.ucode_fan_control) + si_fan_ctrl_stop_smc_fan_control(rdev); + + tach_period = 60 * xclk * 10000 / (8 * speed); + tmp = RREG32(CG_TACH_CTRL) & ~TARGET_PERIOD_MASK; + tmp |= TARGET_PERIOD(tach_period); + WREG32(CG_TACH_CTRL, tmp); + + si_fan_ctrl_set_static_mode(rdev, FDO_PWM_MODE_STATIC_RPM); + + return 0; +} +#endif + +static void si_fan_ctrl_set_default_mode(struct radeon_device *rdev) +{ + struct si_power_info *si_pi = si_get_pi(rdev); + u32 tmp; + + if (!si_pi->fan_ctrl_is_in_default_mode) { + tmp = RREG32(CG_FDO_CTRL2) & ~FDO_PWM_MODE_MASK; + tmp |= FDO_PWM_MODE(si_pi->fan_ctrl_default_mode); + WREG32(CG_FDO_CTRL2, tmp); + + tmp = RREG32(CG_FDO_CTRL2) & ~TMIN_MASK; + tmp |= TMIN(si_pi->t_min); + WREG32(CG_FDO_CTRL2, tmp); + si_pi->fan_ctrl_is_in_default_mode = true; + } +} + +static void si_thermal_start_smc_fan_control(struct radeon_device *rdev) +{ + if (rdev->pm.dpm.fan.ucode_fan_control) { + si_fan_ctrl_start_smc_fan_control(rdev); + si_fan_ctrl_set_static_mode(rdev, FDO_PWM_MODE_STATIC); + } +} + +static void si_thermal_initialize(struct radeon_device *rdev) +{ + u32 tmp; + + if (rdev->pm.fan_pulses_per_revolution) { + tmp = RREG32(CG_TACH_CTRL) & ~EDGE_PER_REV_MASK; + tmp |= EDGE_PER_REV(rdev->pm.fan_pulses_per_revolution -1); + WREG32(CG_TACH_CTRL, tmp); + } + + tmp = RREG32(CG_FDO_CTRL2) & ~TACH_PWM_RESP_RATE_MASK; + tmp |= TACH_PWM_RESP_RATE(0x28); + WREG32(CG_FDO_CTRL2, tmp); +} + +static int si_thermal_start_thermal_controller(struct radeon_device *rdev) +{ + int ret; + + si_thermal_initialize(rdev); + ret = si_thermal_set_temperature_range(rdev, R600_TEMP_RANGE_MIN, R600_TEMP_RANGE_MAX); + if (ret) + return ret; + ret = si_thermal_enable_alert(rdev, true); + if (ret) + return ret; + if (rdev->pm.dpm.fan.ucode_fan_control) { + ret = si_halt_smc(rdev); + if (ret) + return ret; + ret = si_thermal_setup_fan_table(rdev); + if (ret) + return ret; + ret = si_resume_smc(rdev); + if (ret) + return ret; + si_thermal_start_smc_fan_control(rdev); + } + + return 0; +} + +static void si_thermal_stop_thermal_controller(struct radeon_device *rdev) +{ + if (!rdev->pm.no_fan) { + si_fan_ctrl_set_default_mode(rdev); + si_fan_ctrl_stop_smc_fan_control(rdev); + } +} + int si_dpm_enable(struct radeon_device *rdev) { struct rv7xx_power_info *pi = rv770_get_pi(rdev); @@ -5954,31 +6291,39 @@ int si_dpm_enable(struct radeon_device *rdev) si_enable_auto_throttle_source(rdev, RADEON_DPM_AUTO_THROTTLE_SRC_THERMAL, true); + si_thermal_start_thermal_controller(rdev); + ni_update_current_ps(rdev, boot_ps); return 0; } -int si_dpm_late_enable(struct radeon_device *rdev) +static int si_set_temperature_range(struct radeon_device *rdev) { int ret; - if (rdev->irq.installed && - r600_is_internal_thermal_sensor(rdev->pm.int_thermal_type)) { - PPSMC_Result result; + ret = si_thermal_enable_alert(rdev, false); + if (ret) + return ret; + ret = si_thermal_set_temperature_range(rdev, R600_TEMP_RANGE_MIN, R600_TEMP_RANGE_MAX); + if (ret) + return ret; + ret = si_thermal_enable_alert(rdev, true); + if (ret) + return ret; - ret = si_set_thermal_temperature_range(rdev, R600_TEMP_RANGE_MIN, R600_TEMP_RANGE_MAX); - if (ret) - return ret; - rdev->irq.dpm_thermal = true; - radeon_irq_set(rdev); - result = si_send_msg_to_smc(rdev, PPSMC_MSG_EnableThermalInterrupt); + return ret; +} - if (result != PPSMC_Result_OK) - DRM_DEBUG_KMS("Could not enable thermal interrupts.\n"); - } +int si_dpm_late_enable(struct radeon_device *rdev) +{ + int ret; - return 0; + ret = si_set_temperature_range(rdev); + if (ret) + return ret; + + return ret; } void si_dpm_disable(struct radeon_device *rdev) @@ -5988,6 +6333,7 @@ void si_dpm_disable(struct radeon_device *rdev) if (!si_is_smc_running(rdev)) return; + si_thermal_stop_thermal_controller(rdev); si_disable_ulv(rdev); si_clear_vc(rdev); if (pi->thermal_protection) @@ -6526,6 +6872,9 @@ int si_dpm_init(struct radeon_device *rdev) rdev->pm.dpm.dyn_state.max_clock_voltage_on_dc = rdev->pm.dpm.dyn_state.max_clock_voltage_on_ac; + si_pi->fan_ctrl_is_in_default_mode = true; + rdev->pm.dpm.fan.ucode_fan_control = false; + return 0; } diff --git a/drivers/gpu/drm/radeon/si_dpm.h b/drivers/gpu/drm/radeon/si_dpm.h index 8b5c06a..d16bb1b 100644 --- a/drivers/gpu/drm/radeon/si_dpm.h +++ b/drivers/gpu/drm/radeon/si_dpm.h @@ -182,6 +182,7 @@ struct si_power_info { u32 dte_table_start; u32 spll_table_start; u32 papm_cfg_table_start; + u32 fan_table_start; /* CAC stuff */ const struct si_cac_config_reg *cac_weights; const struct si_cac_config_reg *lcac_config; @@ -197,6 +198,10 @@ struct si_power_info { /* SVI2 */ u8 svd_gpio_id; u8 svc_gpio_id; + /* fan control */ + bool fan_ctrl_is_in_default_mode; + u32 t_min; + u32 fan_ctrl_default_mode; }; #define SISLANDS_INITIAL_STATE_ARB_INDEX 0 diff --git a/drivers/gpu/drm/radeon/si_smc.c b/drivers/gpu/drm/radeon/si_smc.c index 73dbc79..e5bb92f 100644 --- a/drivers/gpu/drm/radeon/si_smc.c +++ b/drivers/gpu/drm/radeon/si_smc.c @@ -135,7 +135,7 @@ void si_reset_smc(struct radeon_device *rdev) int si_program_jump_on_start(struct radeon_device *rdev) { - static u8 data[] = { 0x0E, 0x00, 0x40, 0x40 }; + static const u8 data[] = { 0x0E, 0x00, 0x40, 0x40 }; return si_copy_bytes_to_smc(rdev, 0x0, data, 4, sizeof(data)+1); } diff --git a/drivers/gpu/drm/radeon/sid.h b/drivers/gpu/drm/radeon/sid.h index 6635da9..4069be89 100644 --- a/drivers/gpu/drm/radeon/sid.h +++ b/drivers/gpu/drm/radeon/sid.h @@ -180,7 +180,10 @@ #define DIG_THERM_DPM(x) ((x) << 14) #define DIG_THERM_DPM_MASK 0x003FC000 #define DIG_THERM_DPM_SHIFT 14 - +#define CG_THERMAL_STATUS 0x704 +#define FDO_PWM_DUTY(x) ((x) << 9) +#define FDO_PWM_DUTY_MASK (0xff << 9) +#define FDO_PWM_DUTY_SHIFT 9 #define CG_THERMAL_INT 0x708 #define DIG_THERM_INTH(x) ((x) << 8) #define DIG_THERM_INTH_MASK 0x0000FF00 @@ -191,6 +194,10 @@ #define THERM_INT_MASK_HIGH (1 << 24) #define THERM_INT_MASK_LOW (1 << 25) +#define CG_MULT_THERMAL_CTRL 0x710 +#define TEMP_SEL(x) ((x) << 20) +#define TEMP_SEL_MASK (0xff << 20) +#define TEMP_SEL_SHIFT 20 #define CG_MULT_THERMAL_STATUS 0x714 #define ASIC_MAX_TEMP(x) ((x) << 0) #define ASIC_MAX_TEMP_MASK 0x000001ff @@ -199,6 +206,37 @@ #define CTF_TEMP_MASK 0x0003fe00 #define CTF_TEMP_SHIFT 9 +#define CG_FDO_CTRL0 0x754 +#define FDO_STATIC_DUTY(x) ((x) << 0) +#define FDO_STATIC_DUTY_MASK 0x000000FF +#define FDO_STATIC_DUTY_SHIFT 0 +#define CG_FDO_CTRL1 0x758 +#define FMAX_DUTY100(x) ((x) << 0) +#define FMAX_DUTY100_MASK 0x000000FF +#define FMAX_DUTY100_SHIFT 0 +#define CG_FDO_CTRL2 0x75C +#define TMIN(x) ((x) << 0) +#define TMIN_MASK 0x000000FF +#define TMIN_SHIFT 0 +#define FDO_PWM_MODE(x) ((x) << 11) +#define FDO_PWM_MODE_MASK (7 << 11) +#define FDO_PWM_MODE_SHIFT 11 +#define TACH_PWM_RESP_RATE(x) ((x) << 25) +#define TACH_PWM_RESP_RATE_MASK (0x7f << 25) +#define TACH_PWM_RESP_RATE_SHIFT 25 + +#define CG_TACH_CTRL 0x770 +# define EDGE_PER_REV(x) ((x) << 0) +# define EDGE_PER_REV_MASK (0x7 << 0) +# define EDGE_PER_REV_SHIFT 0 +# define TARGET_PERIOD(x) ((x) << 3) +# define TARGET_PERIOD_MASK 0xfffffff8 +# define TARGET_PERIOD_SHIFT 3 +#define CG_TACH_STATUS 0x774 +# define TACH_PERIOD(x) ((x) << 0) +# define TACH_PERIOD_MASK 0xffffffff +# define TACH_PERIOD_SHIFT 0 + #define GENERAL_PWRMGT 0x780 # define GLOBAL_PWRMGT_EN (1 << 0) # define STATIC_PM_EN (1 << 1) diff --git a/drivers/gpu/drm/radeon/sislands_smc.h b/drivers/gpu/drm/radeon/sislands_smc.h index 623a0b1..3c77983 100644 --- a/drivers/gpu/drm/radeon/sislands_smc.h +++ b/drivers/gpu/drm/radeon/sislands_smc.h @@ -245,6 +245,31 @@ typedef struct SISLANDS_SMC_STATETABLE SISLANDS_SMC_STATETABLE; #define SI_SMC_SOFT_REGISTER_svi_rework_gpio_id_svd 0x11c #define SI_SMC_SOFT_REGISTER_svi_rework_gpio_id_svc 0x120 +struct PP_SIslands_FanTable +{ + uint8_t fdo_mode; + uint8_t padding; + int16_t temp_min; + int16_t temp_med; + int16_t temp_max; + int16_t slope1; + int16_t slope2; + int16_t fdo_min; + int16_t hys_up; + int16_t hys_down; + int16_t hys_slope; + int16_t temp_resp_lim; + int16_t temp_curr; + int16_t slope_curr; + int16_t pwm_curr; + uint32_t refresh_period; + int16_t fdo_max; + uint8_t temp_src; + int8_t padding2; +}; + +typedef struct PP_SIslands_FanTable PP_SIslands_FanTable; + #define SMC_SISLANDS_LKGE_LUT_NUM_OF_TEMP_ENTRIES 16 #define SMC_SISLANDS_LKGE_LUT_NUM_OF_VOLT_ENTRIES 32 diff --git a/drivers/gpu/drm/radeon/smu7_discrete.h b/drivers/gpu/drm/radeon/smu7_discrete.h index 82f70c9..0b0b404 100644 --- a/drivers/gpu/drm/radeon/smu7_discrete.h +++ b/drivers/gpu/drm/radeon/smu7_discrete.h @@ -431,6 +431,31 @@ struct SMU7_Discrete_MCRegisters typedef struct SMU7_Discrete_MCRegisters SMU7_Discrete_MCRegisters; +struct SMU7_Discrete_FanTable +{ + uint16_t FdoMode; + int16_t TempMin; + int16_t TempMed; + int16_t TempMax; + int16_t Slope1; + int16_t Slope2; + int16_t FdoMin; + int16_t HystUp; + int16_t HystDown; + int16_t HystSlope; + int16_t TempRespLim; + int16_t TempCurr; + int16_t SlopeCurr; + int16_t PwmCurr; + uint32_t RefreshPeriod; + int16_t FdoMax; + uint8_t TempSrc; + int8_t Padding; +}; + +typedef struct SMU7_Discrete_FanTable SMU7_Discrete_FanTable; + + struct SMU7_Discrete_PmFuses { // dw0-dw1 uint8_t BapmVddCVidHiSidd[8]; @@ -462,7 +487,10 @@ struct SMU7_Discrete_PmFuses { uint8_t BapmVddCVidHiSidd2[8]; // dw11-dw12 - uint32_t Reserved6[2]; + int16_t FuzzyFan_ErrorSetDelta; + int16_t FuzzyFan_ErrorRateSetDelta; + int16_t FuzzyFan_PwmSetDelta; + uint16_t CalcMeasPowerBlend; // dw13-dw16 uint8_t GnbLPML[16]; |
