diff options
Diffstat (limited to 'drivers/gpu/drm/radeon')
62 files changed, 3656 insertions, 1858 deletions
diff --git a/drivers/gpu/drm/radeon/Makefile b/drivers/gpu/drm/radeon/Makefile index 306364a1ecda..09433534dc47 100644 --- a/drivers/gpu/drm/radeon/Makefile +++ b/drivers/gpu/drm/radeon/Makefile @@ -80,7 +80,7 @@ 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 + ci_dpm.o dce6_afmt.o radeon_vm.o # add async DMA block radeon-y += \ @@ -99,6 +99,12 @@ radeon-y += \ uvd_v3_1.o \ uvd_v4_2.o +# add VCE block +radeon-y += \ + radeon_vce.o \ + vce_v1_0.o \ + vce_v2_0.o \ + radeon-$(CONFIG_COMPAT) += radeon_ioc32.o radeon-$(CONFIG_VGA_SWITCHEROO) += radeon_atpx_handler.o radeon-$(CONFIG_ACPI) += radeon_acpi.o diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c index daa4dd375ab1..fb187c78978f 100644 --- a/drivers/gpu/drm/radeon/atombios_crtc.c +++ b/drivers/gpu/drm/radeon/atombios_crtc.c @@ -1106,7 +1106,7 @@ static int dce4_crtc_do_set_base(struct drm_crtc *crtc, int r; /* no fb bound */ - if (!atomic && !crtc->fb) { + if (!atomic && !crtc->primary->fb) { DRM_DEBUG_KMS("No FB bound\n"); return 0; } @@ -1116,8 +1116,8 @@ static int dce4_crtc_do_set_base(struct drm_crtc *crtc, target_fb = fb; } else { - radeon_fb = to_radeon_framebuffer(crtc->fb); - target_fb = crtc->fb; + radeon_fb = to_radeon_framebuffer(crtc->primary->fb); + target_fb = crtc->primary->fb; } /* If atomic, assume fb object is pinned & idle & fenced and @@ -1316,7 +1316,7 @@ static int dce4_crtc_do_set_base(struct drm_crtc *crtc, /* set pageflip to happen anywhere in vblank interval */ WREG32(EVERGREEN_MASTER_UPDATE_MODE + radeon_crtc->crtc_offset, 0); - if (!atomic && fb && fb != crtc->fb) { + if (!atomic && fb && fb != crtc->primary->fb) { radeon_fb = to_radeon_framebuffer(fb); rbo = gem_to_radeon_bo(radeon_fb->obj); r = radeon_bo_reserve(rbo, false); @@ -1350,7 +1350,7 @@ static int avivo_crtc_do_set_base(struct drm_crtc *crtc, int r; /* no fb bound */ - if (!atomic && !crtc->fb) { + if (!atomic && !crtc->primary->fb) { DRM_DEBUG_KMS("No FB bound\n"); return 0; } @@ -1360,8 +1360,8 @@ static int avivo_crtc_do_set_base(struct drm_crtc *crtc, target_fb = fb; } else { - radeon_fb = to_radeon_framebuffer(crtc->fb); - target_fb = crtc->fb; + radeon_fb = to_radeon_framebuffer(crtc->primary->fb); + target_fb = crtc->primary->fb; } obj = radeon_fb->obj; @@ -1485,7 +1485,7 @@ static int avivo_crtc_do_set_base(struct drm_crtc *crtc, /* set pageflip to happen anywhere in vblank interval */ WREG32(AVIVO_D1MODE_MASTER_UPDATE_MODE + radeon_crtc->crtc_offset, 0); - if (!atomic && fb && fb != crtc->fb) { + if (!atomic && fb && fb != crtc->primary->fb) { radeon_fb = to_radeon_framebuffer(fb); rbo = gem_to_radeon_bo(radeon_fb->obj); r = radeon_bo_reserve(rbo, false); @@ -1972,12 +1972,12 @@ static void atombios_crtc_disable(struct drm_crtc *crtc) int i; atombios_crtc_dpms(crtc, DRM_MODE_DPMS_OFF); - if (crtc->fb) { + if (crtc->primary->fb) { int r; struct radeon_framebuffer *radeon_fb; struct radeon_bo *rbo; - radeon_fb = to_radeon_framebuffer(crtc->fb); + radeon_fb = to_radeon_framebuffer(crtc->primary->fb); rbo = gem_to_radeon_bo(radeon_fb->obj); r = radeon_bo_reserve(rbo, false); if (unlikely(r)) diff --git a/drivers/gpu/drm/radeon/atombios_dp.c b/drivers/gpu/drm/radeon/atombios_dp.c index 4ad7643fce5f..8b0ab170cef9 100644 --- a/drivers/gpu/drm/radeon/atombios_dp.c +++ b/drivers/gpu/drm/radeon/atombios_dp.c @@ -142,101 +142,69 @@ static int radeon_process_aux_ch(struct radeon_i2c_chan *chan, return recv_bytes; } -static int radeon_dp_aux_native_write(struct radeon_connector *radeon_connector, - u16 address, u8 *send, u8 send_bytes, u8 delay) -{ - struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; - int ret; - u8 msg[20]; - int msg_bytes = send_bytes + 4; - u8 ack; - unsigned retry; - - if (send_bytes > 16) - return -1; +#define HEADER_SIZE 4 - msg[0] = address; - msg[1] = address >> 8; - msg[2] = DP_AUX_NATIVE_WRITE << 4; - msg[3] = (msg_bytes << 4) | (send_bytes - 1); - memcpy(&msg[4], send, send_bytes); - - for (retry = 0; retry < 7; retry++) { - ret = radeon_process_aux_ch(dig_connector->dp_i2c_bus, - msg, msg_bytes, NULL, 0, delay, &ack); - if (ret == -EBUSY) - continue; - else if (ret < 0) - return ret; - ack >>= 4; - if ((ack & DP_AUX_NATIVE_REPLY_MASK) == DP_AUX_NATIVE_REPLY_ACK) - return send_bytes; - else if ((ack & DP_AUX_NATIVE_REPLY_MASK) == DP_AUX_NATIVE_REPLY_DEFER) - usleep_range(400, 500); - else - return -EIO; - } - - return -EIO; -} - -static int radeon_dp_aux_native_read(struct radeon_connector *radeon_connector, - u16 address, u8 *recv, int recv_bytes, u8 delay) +static ssize_t +radeon_dp_aux_transfer(struct drm_dp_aux *aux, struct drm_dp_aux_msg *msg) { - struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; - u8 msg[4]; - int msg_bytes = 4; - u8 ack; + struct radeon_i2c_chan *chan = + container_of(aux, struct radeon_i2c_chan, aux); int ret; - unsigned retry; - - msg[0] = address; - msg[1] = address >> 8; - msg[2] = DP_AUX_NATIVE_READ << 4; - msg[3] = (msg_bytes << 4) | (recv_bytes - 1); - - for (retry = 0; retry < 7; retry++) { - ret = radeon_process_aux_ch(dig_connector->dp_i2c_bus, - msg, msg_bytes, recv, recv_bytes, delay, &ack); - if (ret == -EBUSY) - continue; - else if (ret < 0) - return ret; - ack >>= 4; - if ((ack & DP_AUX_NATIVE_REPLY_MASK) == DP_AUX_NATIVE_REPLY_ACK) - return ret; - else if ((ack & DP_AUX_NATIVE_REPLY_MASK) == DP_AUX_NATIVE_REPLY_DEFER) - usleep_range(400, 500); - else if (ret == 0) - return -EPROTO; - else - return -EIO; + u8 tx_buf[20]; + size_t tx_size; + u8 ack, delay = 0; + + if (WARN_ON(msg->size > 16)) + return -E2BIG; + + tx_buf[0] = msg->address & 0xff; + tx_buf[1] = msg->address >> 8; + tx_buf[2] = msg->request << 4; + tx_buf[3] = msg->size - 1; + + switch (msg->request & ~DP_AUX_I2C_MOT) { + case DP_AUX_NATIVE_WRITE: + case DP_AUX_I2C_WRITE: + tx_size = HEADER_SIZE + msg->size; + tx_buf[3] |= tx_size << 4; + memcpy(tx_buf + HEADER_SIZE, msg->buffer, msg->size); + ret = radeon_process_aux_ch(chan, + tx_buf, tx_size, NULL, 0, delay, &ack); + if (ret >= 0) + /* Return payload size. */ + ret = msg->size; + break; + case DP_AUX_NATIVE_READ: + case DP_AUX_I2C_READ: + tx_size = HEADER_SIZE; + tx_buf[3] |= tx_size << 4; + ret = radeon_process_aux_ch(chan, + tx_buf, tx_size, msg->buffer, msg->size, delay, &ack); + break; + default: + ret = -EINVAL; + break; } - return -EIO; -} + if (ret > 0) + msg->reply = ack >> 4; -static void radeon_write_dpcd_reg(struct radeon_connector *radeon_connector, - u16 reg, u8 val) -{ - radeon_dp_aux_native_write(radeon_connector, reg, &val, 1, 0); + return ret; } -static u8 radeon_read_dpcd_reg(struct radeon_connector *radeon_connector, - u16 reg) +void radeon_dp_aux_init(struct radeon_connector *radeon_connector) { - u8 val = 0; - - radeon_dp_aux_native_read(radeon_connector, reg, &val, 1, 0); + struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; - return val; + dig_connector->dp_i2c_bus->aux.dev = radeon_connector->base.kdev; + dig_connector->dp_i2c_bus->aux.transfer = radeon_dp_aux_transfer; } int radeon_dp_i2c_aux_ch(struct i2c_adapter *adapter, int mode, u8 write_byte, u8 *read_byte) { struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; - struct radeon_i2c_chan *auxch = (struct radeon_i2c_chan *)adapter; + struct radeon_i2c_chan *auxch = i2c_get_adapdata(adapter); u16 address = algo_data->address; u8 msg[5]; u8 reply[2]; @@ -246,34 +214,30 @@ int radeon_dp_i2c_aux_ch(struct i2c_adapter *adapter, int mode, int ret; u8 ack; - /* Set up the command byte */ - if (mode & MODE_I2C_READ) - msg[2] = DP_AUX_I2C_READ << 4; - else - msg[2] = DP_AUX_I2C_WRITE << 4; - - if (!(mode & MODE_I2C_STOP)) - msg[2] |= DP_AUX_I2C_MOT << 4; - + /* Set up the address */ msg[0] = address; msg[1] = address >> 8; - switch (mode) { - case MODE_I2C_WRITE: + /* Set up the command byte */ + if (mode & MODE_I2C_READ) { + msg[2] = DP_AUX_I2C_READ << 4; + msg_bytes = 4; + msg[3] = msg_bytes << 4; + } else { + msg[2] = DP_AUX_I2C_WRITE << 4; msg_bytes = 5; msg[3] = msg_bytes << 4; msg[4] = write_byte; - break; - case MODE_I2C_READ: - msg_bytes = 4; - msg[3] = msg_bytes << 4; - break; - default: - msg_bytes = 4; - msg[3] = 3 << 4; - break; } + /* special handling for start/stop */ + if (mode & (MODE_I2C_START | MODE_I2C_STOP)) + msg[3] = 3 << 4; + + /* Set MOT bit for all but stop */ + if ((mode & MODE_I2C_STOP) == 0) + msg[2] |= DP_AUX_I2C_MOT << 4; + for (retry = 0; retry < 7; retry++) { ret = radeon_process_aux_ch(auxch, msg, msg_bytes, reply, reply_bytes, 0, &ack); @@ -472,11 +436,11 @@ static void radeon_dp_probe_oui(struct radeon_connector *radeon_connector) if (!(dig_connector->dpcd[DP_DOWN_STREAM_PORT_COUNT] & DP_OUI_SUPPORT)) return; - if (radeon_dp_aux_native_read(radeon_connector, DP_SINK_OUI, buf, 3, 0)) + if (drm_dp_dpcd_read(&dig_connector->dp_i2c_bus->aux, DP_SINK_OUI, buf, 3)) DRM_DEBUG_KMS("Sink OUI: %02hx%02hx%02hx\n", buf[0], buf[1], buf[2]); - if (radeon_dp_aux_native_read(radeon_connector, DP_BRANCH_OUI, buf, 3, 0)) + if (drm_dp_dpcd_read(&dig_connector->dp_i2c_bus->aux, DP_BRANCH_OUI, buf, 3)) DRM_DEBUG_KMS("Branch OUI: %02hx%02hx%02hx\n", buf[0], buf[1], buf[2]); } @@ -487,8 +451,8 @@ bool radeon_dp_getdpcd(struct radeon_connector *radeon_connector) u8 msg[DP_DPCD_SIZE]; int ret, i; - ret = radeon_dp_aux_native_read(radeon_connector, DP_DPCD_REV, msg, - DP_DPCD_SIZE, 0); + ret = drm_dp_dpcd_read(&dig_connector->dp_i2c_bus->aux, DP_DPCD_REV, msg, + DP_DPCD_SIZE); if (ret > 0) { memcpy(dig_connector->dpcd, msg, DP_DPCD_SIZE); DRM_DEBUG_KMS("DPCD: "); @@ -510,6 +474,7 @@ int radeon_dp_get_panel_mode(struct drm_encoder *encoder, struct drm_device *dev = encoder->dev; struct radeon_device *rdev = dev->dev_private; struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector; int panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE; u16 dp_bridge = radeon_connector_encoder_get_dp_bridge_encoder_id(connector); u8 tmp; @@ -517,9 +482,15 @@ int radeon_dp_get_panel_mode(struct drm_encoder *encoder, if (!ASIC_IS_DCE4(rdev)) return panel_mode; + if (!radeon_connector->con_priv) + return panel_mode; + + dig_connector = radeon_connector->con_priv; + if (dp_bridge != ENCODER_OBJECT_ID_NONE) { /* DP bridge chips */ - tmp = radeon_read_dpcd_reg(radeon_connector, DP_EDP_CONFIGURATION_CAP); + drm_dp_dpcd_readb(&dig_connector->dp_i2c_bus->aux, + DP_EDP_CONFIGURATION_CAP, &tmp); if (tmp & 1) panel_mode = DP_PANEL_MODE_INTERNAL_DP2_MODE; else if ((dp_bridge == ENCODER_OBJECT_ID_NUTMEG) || @@ -529,7 +500,8 @@ int radeon_dp_get_panel_mode(struct drm_encoder *encoder, panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE; } else if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) { /* eDP */ - tmp = radeon_read_dpcd_reg(radeon_connector, DP_EDP_CONFIGURATION_CAP); + drm_dp_dpcd_readb(&dig_connector->dp_i2c_bus->aux, + DP_EDP_CONFIGURATION_CAP, &tmp); if (tmp & 1) panel_mode = DP_PANEL_MODE_INTERNAL_DP2_MODE; } @@ -577,37 +549,42 @@ int radeon_dp_mode_valid_helper(struct drm_connector *connector, return MODE_OK; } -static bool radeon_dp_get_link_status(struct radeon_connector *radeon_connector, - u8 link_status[DP_LINK_STATUS_SIZE]) -{ - int ret; - ret = radeon_dp_aux_native_read(radeon_connector, DP_LANE0_1_STATUS, - link_status, DP_LINK_STATUS_SIZE, 100); - if (ret <= 0) { - return false; - } - - DRM_DEBUG_KMS("link status %6ph\n", link_status); - return true; -} - bool radeon_dp_needs_link_train(struct radeon_connector *radeon_connector) { u8 link_status[DP_LINK_STATUS_SIZE]; struct radeon_connector_atom_dig *dig = radeon_connector->con_priv; - if (!radeon_dp_get_link_status(radeon_connector, link_status)) + if (drm_dp_dpcd_read_link_status(&dig->dp_i2c_bus->aux, link_status) <= 0) return false; if (drm_dp_channel_eq_ok(link_status, dig->dp_lane_count)) return false; return true; } +void radeon_dp_set_rx_power_state(struct drm_connector *connector, + u8 power_state) +{ + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector; + + if (!radeon_connector->con_priv) + return; + + dig_connector = radeon_connector->con_priv; + + /* power up/down the sink */ + if (dig_connector->dpcd[0] >= 0x11) { + drm_dp_dpcd_writeb(&dig_connector->dp_i2c_bus->aux, + DP_SET_POWER, power_state); + usleep_range(1000, 2000); + } +} + + struct radeon_dp_link_train_info { struct radeon_device *rdev; struct drm_encoder *encoder; struct drm_connector *connector; - struct radeon_connector *radeon_connector; int enc_id; int dp_clock; int dp_lane_count; @@ -617,6 +594,7 @@ struct radeon_dp_link_train_info { u8 link_status[DP_LINK_STATUS_SIZE]; u8 tries; bool use_dpencoder; + struct drm_dp_aux *aux; }; static void radeon_dp_update_vs_emph(struct radeon_dp_link_train_info *dp_info) @@ -627,8 +605,8 @@ static void radeon_dp_update_vs_emph(struct radeon_dp_link_train_info *dp_info) 0, dp_info->train_set[0]); /* sets all lanes at once */ /* set the vs/emph on the sink */ - radeon_dp_aux_native_write(dp_info->radeon_connector, DP_TRAINING_LANE0_SET, - dp_info->train_set, dp_info->dp_lane_count, 0); + drm_dp_dpcd_write(dp_info->aux, DP_TRAINING_LANE0_SET, + dp_info->train_set, dp_info->dp_lane_count); } static void radeon_dp_set_tp(struct radeon_dp_link_train_info *dp_info, int tp) @@ -663,7 +641,7 @@ static void radeon_dp_set_tp(struct radeon_dp_link_train_info *dp_info, int tp) } /* enable training pattern on the sink */ - radeon_write_dpcd_reg(dp_info->radeon_connector, DP_TRAINING_PATTERN_SET, tp); + drm_dp_dpcd_writeb(dp_info->aux, DP_TRAINING_PATTERN_SET, tp); } static int radeon_dp_link_train_init(struct radeon_dp_link_train_info *dp_info) @@ -673,34 +651,30 @@ static int radeon_dp_link_train_init(struct radeon_dp_link_train_info *dp_info) u8 tmp; /* power up the sink */ - if (dp_info->dpcd[0] >= 0x11) { - radeon_write_dpcd_reg(dp_info->radeon_connector, - DP_SET_POWER, DP_SET_POWER_D0); - usleep_range(1000, 2000); - } + radeon_dp_set_rx_power_state(dp_info->connector, DP_SET_POWER_D0); /* possibly enable downspread on the sink */ if (dp_info->dpcd[3] & 0x1) - radeon_write_dpcd_reg(dp_info->radeon_connector, - DP_DOWNSPREAD_CTRL, DP_SPREAD_AMP_0_5); + drm_dp_dpcd_writeb(dp_info->aux, + DP_DOWNSPREAD_CTRL, DP_SPREAD_AMP_0_5); else - radeon_write_dpcd_reg(dp_info->radeon_connector, - DP_DOWNSPREAD_CTRL, 0); + drm_dp_dpcd_writeb(dp_info->aux, + DP_DOWNSPREAD_CTRL, 0); if ((dp_info->connector->connector_type == DRM_MODE_CONNECTOR_eDP) && (dig->panel_mode == DP_PANEL_MODE_INTERNAL_DP2_MODE)) { - radeon_write_dpcd_reg(dp_info->radeon_connector, DP_EDP_CONFIGURATION_SET, 1); + drm_dp_dpcd_writeb(dp_info->aux, DP_EDP_CONFIGURATION_SET, 1); } /* set the lane count on the sink */ tmp = dp_info->dp_lane_count; if (drm_dp_enhanced_frame_cap(dp_info->dpcd)) tmp |= DP_LANE_COUNT_ENHANCED_FRAME_EN; - radeon_write_dpcd_reg(dp_info->radeon_connector, DP_LANE_COUNT_SET, tmp); + drm_dp_dpcd_writeb(dp_info->aux, DP_LANE_COUNT_SET, tmp); /* set the link rate on the sink */ tmp = drm_dp_link_rate_to_bw_code(dp_info->dp_clock); - radeon_write_dpcd_reg(dp_info->radeon_connector, DP_LINK_BW_SET, tmp); + drm_dp_dpcd_writeb(dp_info->aux, DP_LINK_BW_SET, tmp); /* start training on the source */ if (ASIC_IS_DCE4(dp_info->rdev) || !dp_info->use_dpencoder) @@ -711,9 +685,9 @@ static int radeon_dp_link_train_init(struct radeon_dp_link_train_info *dp_info) dp_info->dp_clock, dp_info->enc_id, 0); /* disable the training pattern on the sink */ - radeon_write_dpcd_reg(dp_info->radeon_connector, - DP_TRAINING_PATTERN_SET, - DP_TRAINING_PATTERN_DISABLE); + drm_dp_dpcd_writeb(dp_info->aux, + DP_TRAINING_PATTERN_SET, + DP_TRAINING_PATTERN_DISABLE); return 0; } @@ -723,9 +697,9 @@ static int radeon_dp_link_train_finish(struct radeon_dp_link_train_info *dp_info udelay(400); /* disable the training pattern on the sink */ - radeon_write_dpcd_reg(dp_info->radeon_connector, - DP_TRAINING_PATTERN_SET, - DP_TRAINING_PATTERN_DISABLE); + drm_dp_dpcd_writeb(dp_info->aux, + DP_TRAINING_PATTERN_SET, + DP_TRAINING_PATTERN_DISABLE); /* disable the training pattern on the source */ if (ASIC_IS_DCE4(dp_info->rdev) || !dp_info->use_dpencoder) @@ -757,7 +731,8 @@ static int radeon_dp_link_train_cr(struct radeon_dp_link_train_info *dp_info) while (1) { drm_dp_link_train_clock_recovery_delay(dp_info->dpcd); - if (!radeon_dp_get_link_status(dp_info->radeon_connector, dp_info->link_status)) { + if (drm_dp_dpcd_read_link_status(dp_info->aux, + dp_info->link_status) <= 0) { DRM_ERROR("displayport link status failed\n"); break; } @@ -819,7 +794,8 @@ static int radeon_dp_link_train_ce(struct radeon_dp_link_train_info *dp_info) while (1) { drm_dp_link_train_channel_eq_delay(dp_info->dpcd); - if (!radeon_dp_get_link_status(dp_info->radeon_connector, dp_info->link_status)) { + if (drm_dp_dpcd_read_link_status(dp_info->aux, + dp_info->link_status) <= 0) { DRM_ERROR("displayport link status failed\n"); break; } @@ -902,7 +878,7 @@ void radeon_dp_link_train(struct drm_encoder *encoder, else dp_info.enc_id |= ATOM_DP_CONFIG_LINK_A; - tmp = radeon_read_dpcd_reg(radeon_connector, DP_MAX_LANE_COUNT); + drm_dp_dpcd_readb(&dig_connector->dp_i2c_bus->aux, DP_MAX_LANE_COUNT, &tmp); if (ASIC_IS_DCE5(rdev) && (tmp & DP_TPS3_SUPPORTED)) dp_info.tp3_supported = true; else @@ -912,9 +888,9 @@ void radeon_dp_link_train(struct drm_encoder *encoder, dp_info.rdev = rdev; dp_info.encoder = encoder; dp_info.connector = connector; - dp_info.radeon_connector = radeon_connector; dp_info.dp_lane_count = dig_connector->dp_lane_count; dp_info.dp_clock = dig_connector->dp_clock; + dp_info.aux = &dig_connector->dp_i2c_bus->aux; if (radeon_dp_link_train_init(&dp_info)) goto done; diff --git a/drivers/gpu/drm/radeon/atombios_encoders.c b/drivers/gpu/drm/radeon/atombios_encoders.c index 607dc14d195e..e6eb5097597f 100644 --- a/drivers/gpu/drm/radeon/atombios_encoders.c +++ b/drivers/gpu/drm/radeon/atombios_encoders.c @@ -1633,10 +1633,16 @@ radeon_atom_encoder_dpms_dig(struct drm_encoder *encoder, int mode) struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); struct radeon_connector *radeon_connector = NULL; struct radeon_connector_atom_dig *radeon_dig_connector = NULL; + bool travis_quirk = false; if (connector) { radeon_connector = to_radeon_connector(connector); radeon_dig_connector = radeon_connector->con_priv; + if ((radeon_connector_encoder_get_dp_bridge_encoder_id(connector) == + ENCODER_OBJECT_ID_TRAVIS) && + (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) && + !ASIC_IS_DCE5(rdev)) + travis_quirk = true; } switch (mode) { @@ -1657,17 +1663,13 @@ radeon_atom_encoder_dpms_dig(struct drm_encoder *encoder, int mode) atombios_external_encoder_setup(encoder, ext_encoder, EXTERNAL_ENCODER_ACTION_V3_ENCODER_SETUP); } - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); } else if (ASIC_IS_DCE4(rdev)) { /* setup and enable the encoder */ atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_SETUP, 0); - /* enable the transmitter */ - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); } else { /* setup and enable the encoder and transmitter */ atombios_dig_encoder_setup(encoder, ATOM_ENABLE, 0); atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_SETUP, 0, 0); - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); } if (ENCODER_MODE_IS_DP(atombios_get_encoder_mode(encoder)) && connector) { if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) { @@ -1675,68 +1677,56 @@ radeon_atom_encoder_dpms_dig(struct drm_encoder *encoder, int mode) ATOM_TRANSMITTER_ACTION_POWER_ON); radeon_dig_connector->edp_on = true; } + } + /* enable the transmitter */ + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); + if (ENCODER_MODE_IS_DP(atombios_get_encoder_mode(encoder)) && connector) { + /* DP_SET_POWER_D0 is set in radeon_dp_link_train */ radeon_dp_link_train(encoder, connector); if (ASIC_IS_DCE4(rdev)) atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_ON, 0); } if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_LCD_BLON, 0, 0); + atombios_dig_transmitter_setup(encoder, + ATOM_TRANSMITTER_ACTION_LCD_BLON, 0, 0); + if (ext_encoder) + atombios_external_encoder_setup(encoder, ext_encoder, ATOM_ENABLE); break; case DRM_MODE_DPMS_STANDBY: case DRM_MODE_DPMS_SUSPEND: case DRM_MODE_DPMS_OFF: if (ASIC_IS_DCE4(rdev)) { + if (ENCODER_MODE_IS_DP(atombios_get_encoder_mode(encoder)) && connector) + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_OFF, 0); + } + if (ext_encoder) + atombios_external_encoder_setup(encoder, ext_encoder, ATOM_DISABLE); + if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) + atombios_dig_transmitter_setup(encoder, + ATOM_TRANSMITTER_ACTION_LCD_BLOFF, 0, 0); + + if (ENCODER_MODE_IS_DP(atombios_get_encoder_mode(encoder)) && + connector && !travis_quirk) + radeon_dp_set_rx_power_state(connector, DP_SET_POWER_D3); + if (ASIC_IS_DCE4(rdev)) { /* disable the transmitter */ - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); + atombios_dig_transmitter_setup(encoder, + ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); } else { /* disable the encoder and transmitter */ - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); + atombios_dig_transmitter_setup(encoder, + ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); atombios_dig_encoder_setup(encoder, ATOM_DISABLE, 0); } if (ENCODER_MODE_IS_DP(atombios_get_encoder_mode(encoder)) && connector) { - if (ASIC_IS_DCE4(rdev)) - atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_OFF, 0); + if (travis_quirk) + radeon_dp_set_rx_power_state(connector, DP_SET_POWER_D3); if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) { atombios_set_edp_panel_power(connector, ATOM_TRANSMITTER_ACTION_POWER_OFF); radeon_dig_connector->edp_on = false; } } - if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_LCD_BLOFF, 0, 0); - break; - } -} - -static void -radeon_atom_encoder_dpms_ext(struct drm_encoder *encoder, - struct drm_encoder *ext_encoder, - int mode) -{ - struct drm_device *dev = encoder->dev; - struct radeon_device *rdev = dev->dev_private; - - switch (mode) { - case DRM_MODE_DPMS_ON: - default: - if (ASIC_IS_DCE41(rdev) || ASIC_IS_DCE61(rdev)) { - atombios_external_encoder_setup(encoder, ext_encoder, - EXTERNAL_ENCODER_ACTION_V3_ENABLE_OUTPUT); - atombios_external_encoder_setup(encoder, ext_encoder, - EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING_OFF); - } else - atombios_external_encoder_setup(encoder, ext_encoder, ATOM_ENABLE); - break; - case DRM_MODE_DPMS_STANDBY: - case DRM_MODE_DPMS_SUSPEND: - case DRM_MODE_DPMS_OFF: - if (ASIC_IS_DCE41(rdev) || ASIC_IS_DCE61(rdev)) { - atombios_external_encoder_setup(encoder, ext_encoder, - EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING); - atombios_external_encoder_setup(encoder, ext_encoder, - EXTERNAL_ENCODER_ACTION_V3_DISABLE_OUTPUT); - } else - atombios_external_encoder_setup(encoder, ext_encoder, ATOM_DISABLE); break; } } @@ -1747,7 +1737,6 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) struct drm_device *dev = encoder->dev; struct radeon_device *rdev = dev->dev_private; struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); - struct drm_encoder *ext_encoder = radeon_get_external_encoder(encoder); DRM_DEBUG_KMS("encoder dpms %d to mode %d, devices %08x, active_devices %08x\n", radeon_encoder->encoder_id, mode, radeon_encoder->devices, @@ -1807,9 +1796,6 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) return; } - if (ext_encoder) - radeon_atom_encoder_dpms_ext(encoder, ext_encoder, mode); - radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); } diff --git a/drivers/gpu/drm/radeon/btc_dpm.c b/drivers/gpu/drm/radeon/btc_dpm.c index ea103ccdf4bd..f81d7ca134db 100644 --- a/drivers/gpu/drm/radeon/btc_dpm.c +++ b/drivers/gpu/drm/radeon/btc_dpm.c @@ -2601,6 +2601,10 @@ int btc_dpm_init(struct radeon_device *rdev) pi->min_vddc_in_table = 0; pi->max_vddc_in_table = 0; + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = rv7xx_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/ci_dpm.c b/drivers/gpu/drm/radeon/ci_dpm.c index 8d49104ca6c2..cad89a977527 100644 --- a/drivers/gpu/drm/radeon/ci_dpm.c +++ b/drivers/gpu/drm/radeon/ci_dpm.c @@ -172,6 +172,8 @@ extern void si_trim_voltage_table_to_fit_state_table(struct radeon_device *rdev, extern void cik_enter_rlc_safe_mode(struct radeon_device *rdev); extern void cik_exit_rlc_safe_mode(struct radeon_device *rdev); extern int ci_mc_load_microcode(struct radeon_device *rdev); +extern void cik_update_cg(struct radeon_device *rdev, + u32 block, bool enable); static int ci_get_std_voltage_value_sidd(struct radeon_device *rdev, struct atom_voltage_table_entry *voltage_table, @@ -746,6 +748,14 @@ static void ci_apply_state_adjust_rules(struct radeon_device *rdev, u32 max_sclk_vddc, max_mclk_vddci, max_mclk_vddc; int i; + if (rps->vce_active) { + rps->evclk = rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].evclk; + rps->ecclk = rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].ecclk; + } else { + rps->evclk = 0; + rps->ecclk = 0; + } + if ((rdev->pm.dpm.new_active_crtc_count > 1) || ci_dpm_vblank_too_short(rdev)) disable_mclk_switching = true; @@ -804,6 +814,13 @@ static void ci_apply_state_adjust_rules(struct radeon_device *rdev, sclk = ps->performance_levels[0].sclk; } + if (rps->vce_active) { + if (sclk < rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].sclk) + sclk = rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].sclk; + if (mclk < rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].mclk) + mclk = rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].mclk; + } + ps->performance_levels[0].sclk = sclk; ps->performance_levels[0].mclk = mclk; @@ -3468,7 +3485,6 @@ static int ci_enable_uvd_dpm(struct radeon_device *rdev, bool enable) 0 : -EINVAL; } -#if 0 static int ci_enable_vce_dpm(struct radeon_device *rdev, bool enable) { struct ci_power_info *pi = ci_get_pi(rdev); @@ -3501,6 +3517,7 @@ static int ci_enable_vce_dpm(struct radeon_device *rdev, bool enable) 0 : -EINVAL; } +#if 0 static int ci_enable_samu_dpm(struct radeon_device *rdev, bool enable) { struct ci_power_info *pi = ci_get_pi(rdev); @@ -3587,7 +3604,6 @@ static int ci_update_uvd_dpm(struct radeon_device *rdev, bool gate) return ci_enable_uvd_dpm(rdev, !gate); } -#if 0 static u8 ci_get_vce_boot_level(struct radeon_device *rdev) { u8 i; @@ -3608,15 +3624,15 @@ static int ci_update_vce_dpm(struct radeon_device *rdev, struct radeon_ps *radeon_current_state) { struct ci_power_info *pi = ci_get_pi(rdev); - bool new_vce_clock_non_zero = (radeon_new_state->evclk != 0); - bool old_vce_clock_non_zero = (radeon_current_state->evclk != 0); int ret = 0; u32 tmp; - if (new_vce_clock_non_zero != old_vce_clock_non_zero) { - if (new_vce_clock_non_zero) { - pi->smc_state_table.VceBootLevel = ci_get_vce_boot_level(rdev); + if (radeon_current_state->evclk != radeon_new_state->evclk) { + if (radeon_new_state->evclk) { + /* turn the clocks on when encoding */ + cik_update_cg(rdev, RADEON_CG_BLOCK_VCE, false); + pi->smc_state_table.VceBootLevel = ci_get_vce_boot_level(rdev); tmp = RREG32_SMC(DPM_TABLE_475); tmp &= ~VceBootLevel_MASK; tmp |= VceBootLevel(pi->smc_state_table.VceBootLevel); @@ -3624,12 +3640,16 @@ static int ci_update_vce_dpm(struct radeon_device *rdev, ret = ci_enable_vce_dpm(rdev, true); } else { + /* turn the clocks off when not encoding */ + cik_update_cg(rdev, RADEON_CG_BLOCK_VCE, true); + ret = ci_enable_vce_dpm(rdev, false); } } return ret; } +#if 0 static int ci_update_samu_dpm(struct radeon_device *rdev, bool gate) { return ci_enable_samu_dpm(rdev, gate); @@ -4752,13 +4772,13 @@ int ci_dpm_set_power_state(struct radeon_device *rdev) DRM_ERROR("ci_generate_dpm_level_enable_mask failed\n"); return ret; } -#if 0 + ret = ci_update_vce_dpm(rdev, new_ps, old_ps); if (ret) { DRM_ERROR("ci_update_vce_dpm failed\n"); return ret; } -#endif + ret = ci_update_sclk_t(rdev); if (ret) { DRM_ERROR("ci_update_sclk_t failed\n"); @@ -4959,9 +4979,6 @@ static int ci_parse_power_table(struct radeon_device *rdev) if (!rdev->pm.dpm.ps) return -ENOMEM; power_state_offset = (u8 *)state_array->states; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < state_array->ucNumEntries; i++) { u8 *idx; power_state = (union pplib_power_state *)power_state_offset; @@ -4998,6 +5015,21 @@ static int ci_parse_power_table(struct radeon_device *rdev) power_state_offset += 2 + power_state->v2.ucNumDPMLevels; } rdev->pm.dpm.num_ps = state_array->ucNumEntries; + + /* fill in the vce power states */ + for (i = 0; i < RADEON_MAX_VCE_LEVELS; i++) { + u32 sclk, mclk; + clock_array_index = rdev->pm.dpm.vce_states[i].clk_idx; + clock_info = (union pplib_clock_info *) + &clock_info_array->clockInfo[clock_array_index * clock_info_array->ucEntrySize]; + sclk = le16_to_cpu(clock_info->ci.usEngineClockLow); + sclk |= clock_info->ci.ucEngineClockHigh << 16; + mclk = le16_to_cpu(clock_info->ci.usMemoryClockLow); + mclk |= clock_info->ci.ucMemoryClockHigh << 16; + rdev->pm.dpm.vce_states[i].sclk = sclk; + rdev->pm.dpm.vce_states[i].mclk = mclk; + } + return 0; } @@ -5077,17 +5109,25 @@ int ci_dpm_init(struct radeon_device *rdev) ci_dpm_fini(rdev); return ret; } - ret = ci_parse_power_table(rdev); + + ret = r600_get_platform_caps(rdev); if (ret) { ci_dpm_fini(rdev); return ret; } + ret = r600_parse_extended_power_table(rdev); if (ret) { ci_dpm_fini(rdev); return ret; } + ret = ci_parse_power_table(rdev); + if (ret) { + ci_dpm_fini(rdev); + return ret; + } + pi->dll_default_on = false; pi->sram_end = SMC_RAM_END; @@ -5120,6 +5160,7 @@ int ci_dpm_init(struct radeon_device *rdev) pi->caps_sclk_throttle_low_notification = false; pi->caps_uvd_dpm = true; + pi->caps_vce_dpm = true; ci_get_leakage_voltages(rdev); ci_patch_dependency_tables_with_leakage(rdev); diff --git a/drivers/gpu/drm/radeon/cik.c b/drivers/gpu/drm/radeon/cik.c index bbb17841a9e5..745143c2358f 100644 --- a/drivers/gpu/drm/radeon/cik.c +++ b/drivers/gpu/drm/radeon/cik.c @@ -75,6 +75,7 @@ extern void si_init_uvd_internal_cg(struct radeon_device *rdev); extern int cik_sdma_resume(struct radeon_device *rdev); extern void cik_sdma_enable(struct radeon_device *rdev, bool enable); extern void cik_sdma_fini(struct radeon_device *rdev); +extern void vce_v2_0_enable_mgcg(struct radeon_device *rdev, bool enable); static void cik_rlc_stop(struct radeon_device *rdev); static void cik_pcie_gen3_enable(struct radeon_device *rdev); static void cik_program_aspm(struct radeon_device *rdev); @@ -1095,7 +1096,7 @@ static const u32 spectre_golden_registers[] = 0x8a14, 0xf000003f, 0x00000007, 0x8b24, 0xffffffff, 0x00ffffff, 0x28350, 0x3f3f3fff, 0x00000082, - 0x28355, 0x0000003f, 0x00000000, + 0x28354, 0x0000003f, 0x00000000, 0x3e78, 0x00000001, 0x00000002, 0x913c, 0xffff03df, 0x00000004, 0xc768, 0x00000008, 0x00000008, @@ -2028,6 +2029,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 5: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); break; case 6: @@ -2048,6 +2050,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 9: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING)); break; case 10: @@ -2070,6 +2073,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 13: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING)); break; case 14: @@ -2092,6 +2096,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 27: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P16_32x32_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING)); break; case 28: @@ -2246,6 +2251,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 5: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); break; case 6: @@ -2266,6 +2272,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 9: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING)); break; case 10: @@ -2288,6 +2295,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 13: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING)); break; case 14: @@ -2310,6 +2318,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 27: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P8_32x32_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING)); break; case 28: @@ -2466,6 +2475,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 5: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); break; case 6: @@ -2486,6 +2496,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 9: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING)); break; case 10: @@ -2508,6 +2519,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 13: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING)); break; case 14: @@ -2530,6 +2542,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 27: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_16x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING)); break; case 28: @@ -2592,6 +2605,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 5: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_8x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); break; case 6: @@ -2612,6 +2626,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 9: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_8x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING)); break; case 10: @@ -2634,6 +2649,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 13: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_8x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING)); break; case 14: @@ -2656,6 +2672,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 27: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P4_8x16) | MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING)); break; case 28: @@ -2812,6 +2829,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 5: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | MICRO_TILE_MODE_NEW(ADDR_SURF_DEPTH_MICRO_TILING)); break; case 6: @@ -2827,11 +2845,13 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) TILE_SPLIT(split_equal_to_row_size)); break; case 8: - gb_tile_moden = ARRAY_MODE(ARRAY_LINEAR_ALIGNED); + gb_tile_moden = ARRAY_MODE(ARRAY_LINEAR_ALIGNED) | + PIPE_CONFIG(ADDR_SURF_P2); break; case 9: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING)); + MICRO_TILE_MODE_NEW(ADDR_SURF_DISPLAY_MICRO_TILING) | + PIPE_CONFIG(ADDR_SURF_P2)); break; case 10: gb_tile_moden = (ARRAY_MODE(ARRAY_2D_TILED_THIN1) | @@ -2853,6 +2873,7 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 13: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | + PIPE_CONFIG(ADDR_SURF_P2) | MICRO_TILE_MODE_NEW(ADDR_SURF_THIN_MICRO_TILING)); break; case 14: @@ -2875,7 +2896,8 @@ static void cik_tiling_mode_table_init(struct radeon_device *rdev) break; case 27: gb_tile_moden = (ARRAY_MODE(ARRAY_1D_TILED_THIN1) | - MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING)); + MICRO_TILE_MODE_NEW(ADDR_SURF_ROTATED_MICRO_TILING) | + PIPE_CONFIG(ADDR_SURF_P2)); break; case 28: gb_tile_moden = (ARRAY_MODE(ARRAY_PRT_2D_TILED_THIN1) | @@ -4030,8 +4052,6 @@ static int cik_cp_gfx_resume(struct radeon_device *rdev) WREG32(CP_RB0_BASE, rb_addr); WREG32(CP_RB0_BASE_HI, upper_32_bits(rb_addr)); - ring->rptr = RREG32(CP_RB0_RPTR); - /* start the ring */ cik_cp_gfx_start(rdev); rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready = true; @@ -4589,8 +4609,7 @@ static int cik_cp_compute_resume(struct radeon_device *rdev) rdev->ring[idx].wptr = 0; mqd->queue_state.cp_hqd_pq_wptr = rdev->ring[idx].wptr; WREG32(CP_HQD_PQ_WPTR, mqd->queue_state.cp_hqd_pq_wptr); - rdev->ring[idx].rptr = RREG32(CP_HQD_PQ_RPTR); - mqd->queue_state.cp_hqd_pq_rptr = rdev->ring[idx].rptr; + mqd->queue_state.cp_hqd_pq_rptr = RREG32(CP_HQD_PQ_RPTR); /* set the vmid for the queue */ mqd->queue_state.cp_hqd_vmid = 0; @@ -5120,11 +5139,9 @@ bool cik_gfx_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) if (!(reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE | RADEON_RESET_CP))) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force CP activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } @@ -6144,6 +6161,10 @@ void cik_update_cg(struct radeon_device *rdev, cik_enable_hdp_mgcg(rdev, enable); cik_enable_hdp_ls(rdev, enable); } + + if (block & RADEON_CG_BLOCK_VCE) { + vce_v2_0_enable_mgcg(rdev, enable); + } } static void cik_init_cg(struct radeon_device *rdev) @@ -6521,8 +6542,8 @@ void cik_get_csb_buffer(struct radeon_device *rdev, volatile u32 *buffer) buffer[count++] = cpu_to_le32(0x00000000); break; case CHIP_HAWAII: - buffer[count++] = 0x3a00161a; - buffer[count++] = 0x0000002e; + buffer[count++] = cpu_to_le32(0x3a00161a); + buffer[count++] = cpu_to_le32(0x0000002e); break; default: buffer[count++] = cpu_to_le32(0x00000000); @@ -7493,6 +7514,20 @@ restart_ih: /* reset addr and status */ WREG32_P(VM_CONTEXT1_CNTL2, 1, ~1); break; + case 167: /* VCE */ + DRM_DEBUG("IH: VCE int: 0x%08x\n", src_data); + switch (src_data) { + case 0: + radeon_fence_process(rdev, TN_RING_TYPE_VCE1_INDEX); + break; + case 1: + radeon_fence_process(rdev, TN_RING_TYPE_VCE2_INDEX); + break; + default: + DRM_ERROR("Unhandled interrupt: %d %d\n", src_id, src_data); + break; + } + break; case 176: /* GFX RB CP_INT */ case 177: /* GFX IB CP_INT */ radeon_fence_process(rdev, RADEON_RING_TYPE_GFX_INDEX); @@ -7792,6 +7827,22 @@ static int cik_startup(struct radeon_device *rdev) if (r) rdev->ring[R600_RING_TYPE_UVD_INDEX].ring_size = 0; + r = radeon_vce_resume(rdev); + if (!r) { + r = vce_v2_0_resume(rdev); + if (!r) + r = radeon_fence_driver_start_ring(rdev, + TN_RING_TYPE_VCE1_INDEX); + if (!r) + r = radeon_fence_driver_start_ring(rdev, + TN_RING_TYPE_VCE2_INDEX); + } + if (r) { + dev_err(rdev->dev, "VCE init error (%d).\n", r); + rdev->ring[TN_RING_TYPE_VCE1_INDEX].ring_size = 0; + rdev->ring[TN_RING_TYPE_VCE2_INDEX].ring_size = 0; + } + /* Enable IRQ */ if (!rdev->irq.installed) { r = radeon_irq_kms_init(rdev); @@ -7867,6 +7918,23 @@ static int cik_startup(struct radeon_device *rdev) DRM_ERROR("radeon: failed initializing UVD (%d).\n", r); } + r = -ENOENT; + + ring = &rdev->ring[TN_RING_TYPE_VCE1_INDEX]; + if (ring->ring_size) + r = radeon_ring_init(rdev, ring, ring->ring_size, 0, + VCE_CMD_NO_OP); + + ring = &rdev->ring[TN_RING_TYPE_VCE2_INDEX]; + if (ring->ring_size) + r = radeon_ring_init(rdev, ring, ring->ring_size, 0, + VCE_CMD_NO_OP); + + if (!r) + r = vce_v1_0_init(rdev); + else if (r != -ENOENT) + DRM_ERROR("radeon: failed initializing VCE (%d).\n", r); + r = radeon_ib_pool_init(rdev); if (r) { dev_err(rdev->dev, "IB initialization failed (%d).\n", r); @@ -7938,6 +8006,7 @@ int cik_suspend(struct radeon_device *rdev) cik_sdma_enable(rdev, false); uvd_v1_0_fini(rdev); radeon_uvd_suspend(rdev); + radeon_vce_suspend(rdev); cik_fini_pg(rdev); cik_fini_cg(rdev); cik_irq_suspend(rdev); @@ -8070,6 +8139,17 @@ int cik_init(struct radeon_device *rdev) r600_ring_init(rdev, ring, 4096); } + r = radeon_vce_init(rdev); + if (!r) { + ring = &rdev->ring[TN_RING_TYPE_VCE1_INDEX]; + ring->ring_obj = NULL; + r600_ring_init(rdev, ring, 4096); + + ring = &rdev->ring[TN_RING_TYPE_VCE2_INDEX]; + ring->ring_obj = NULL; + r600_ring_init(rdev, ring, 4096); + } + rdev->ih.ring_obj = NULL; r600_ih_ring_init(rdev, 64 * 1024); @@ -8131,6 +8211,7 @@ void cik_fini(struct radeon_device *rdev) radeon_irq_kms_fini(rdev); uvd_v1_0_fini(rdev); radeon_uvd_fini(rdev); + radeon_vce_fini(rdev); cik_pcie_gart_fini(rdev); r600_vram_scratch_fini(rdev); radeon_gem_fini(rdev); @@ -8869,6 +8950,41 @@ int cik_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk) return r; } +int cik_set_vce_clocks(struct radeon_device *rdev, u32 evclk, u32 ecclk) +{ + int r, i; + struct atom_clock_dividers dividers; + u32 tmp; + + r = radeon_atom_get_clock_dividers(rdev, COMPUTE_GPUCLK_INPUT_FLAG_DEFAULT_GPUCLK, + ecclk, false, ÷rs); + if (r) + return r; + + for (i = 0; i < 100; i++) { + if (RREG32_SMC(CG_ECLK_STATUS) & ECLK_STATUS) + break; + mdelay(10); + } + if (i == 100) + return -ETIMEDOUT; + + tmp = RREG32_SMC(CG_ECLK_CNTL); + tmp &= ~(ECLK_DIR_CNTL_EN|ECLK_DIVIDER_MASK); + tmp |= dividers.post_divider; + WREG32_SMC(CG_ECLK_CNTL, tmp); + + for (i = 0; i < 100; i++) { + if (RREG32_SMC(CG_ECLK_STATUS) & ECLK_STATUS) + break; + mdelay(10); + } + if (i == 100) + return -ETIMEDOUT; + + return 0; +} + static void cik_pcie_gen3_enable(struct radeon_device *rdev) { struct pci_dev *root = rdev->pdev->bus->self; diff --git a/drivers/gpu/drm/radeon/cik_sdma.c b/drivers/gpu/drm/radeon/cik_sdma.c index 94626ea90fa5..89b4afa5041c 100644 --- a/drivers/gpu/drm/radeon/cik_sdma.c +++ b/drivers/gpu/drm/radeon/cik_sdma.c @@ -369,8 +369,6 @@ static int cik_sdma_gfx_resume(struct radeon_device *rdev) ring->wptr = 0; WREG32(SDMA0_GFX_RB_WPTR + reg_offset, ring->wptr << 2); - ring->rptr = RREG32(SDMA0_GFX_RB_RPTR + reg_offset) >> 2; - /* enable DMA RB */ WREG32(SDMA0_GFX_RB_CNTL + reg_offset, rb_cntl | SDMA_RB_ENABLE); @@ -713,11 +711,9 @@ bool cik_sdma_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) mask = RADEON_RESET_DMA1; if (!(reset_mask & mask)) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force ring activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } diff --git a/drivers/gpu/drm/radeon/cikd.h b/drivers/gpu/drm/radeon/cikd.h index 98bae9d7b74d..213873270d5f 100644 --- a/drivers/gpu/drm/radeon/cikd.h +++ b/drivers/gpu/drm/radeon/cikd.h @@ -203,6 +203,12 @@ #define CTF_TEMP_MASK 0x0003fe00 #define CTF_TEMP_SHIFT 9 +#define CG_ECLK_CNTL 0xC05000AC +# define ECLK_DIVIDER_MASK 0x7f +# define ECLK_DIR_CNTL_EN (1 << 8) +#define CG_ECLK_STATUS 0xC05000B0 +# define ECLK_STATUS (1 << 0) + #define CG_SPLL_FUNC_CNTL 0xC0500140 #define SPLL_RESET (1 << 0) #define SPLL_PWRON (1 << 1) @@ -2010,4 +2016,47 @@ /* UVD CTX indirect */ #define UVD_CGC_MEM_CTRL 0xC0 +/* VCE */ + +#define VCE_VCPU_CACHE_OFFSET0 0x20024 +#define VCE_VCPU_CACHE_SIZE0 0x20028 +#define VCE_VCPU_CACHE_OFFSET1 0x2002c +#define VCE_VCPU_CACHE_SIZE1 0x20030 +#define VCE_VCPU_CACHE_OFFSET2 0x20034 +#define VCE_VCPU_CACHE_SIZE2 0x20038 +#define VCE_RB_RPTR2 0x20178 +#define VCE_RB_WPTR2 0x2017c +#define VCE_RB_RPTR 0x2018c +#define VCE_RB_WPTR 0x20190 +#define VCE_CLOCK_GATING_A 0x202f8 +# define CGC_CLK_GATE_DLY_TIMER_MASK (0xf << 0) +# define CGC_CLK_GATE_DLY_TIMER(x) ((x) << 0) +# define CGC_CLK_GATER_OFF_DLY_TIMER_MASK (0xff << 4) +# define CGC_CLK_GATER_OFF_DLY_TIMER(x) ((x) << 4) +# define CGC_UENC_WAIT_AWAKE (1 << 18) +#define VCE_CLOCK_GATING_B 0x202fc +#define VCE_CGTT_CLK_OVERRIDE 0x207a0 +#define VCE_UENC_CLOCK_GATING 0x207bc +# define CLOCK_ON_DELAY_MASK (0xf << 0) +# define CLOCK_ON_DELAY(x) ((x) << 0) +# define CLOCK_OFF_DELAY_MASK (0xff << 4) +# define CLOCK_OFF_DELAY(x) ((x) << 4) +#define VCE_UENC_REG_CLOCK_GATING 0x207c0 +#define VCE_SYS_INT_EN 0x21300 +# define VCE_SYS_INT_TRAP_INTERRUPT_EN (1 << 3) +#define VCE_LMI_CTRL2 0x21474 +#define VCE_LMI_CTRL 0x21498 +#define VCE_LMI_VM_CTRL 0x214a0 +#define VCE_LMI_SWAP_CNTL 0x214b4 +#define VCE_LMI_SWAP_CNTL1 0x214b8 +#define VCE_LMI_CACHE_CTRL 0x214f4 + +#define VCE_CMD_NO_OP 0x00000000 +#define VCE_CMD_END 0x00000001 +#define VCE_CMD_IB 0x00000002 +#define VCE_CMD_FENCE 0x00000003 +#define VCE_CMD_TRAP 0x00000004 +#define VCE_CMD_IB_AUTO 0x00000005 +#define VCE_CMD_SEMAPHORE 0x00000006 + #endif diff --git a/drivers/gpu/drm/radeon/cypress_dpm.c b/drivers/gpu/drm/radeon/cypress_dpm.c index cf783fc0ef21..5a9a5f4d7888 100644 --- a/drivers/gpu/drm/radeon/cypress_dpm.c +++ b/drivers/gpu/drm/radeon/cypress_dpm.c @@ -2036,6 +2036,10 @@ int cypress_dpm_init(struct radeon_device *rdev) pi->min_vddc_in_table = 0; pi->max_vddc_in_table = 0; + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = rv7xx_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c index 27b0ff16082e..b406546440da 100644 --- a/drivers/gpu/drm/radeon/evergreen.c +++ b/drivers/gpu/drm/radeon/evergreen.c @@ -2990,8 +2990,6 @@ static int evergreen_cp_resume(struct radeon_device *rdev) WREG32(CP_RB_BASE, ring->gpu_addr >> 8); WREG32(CP_DEBUG, (1 << 27) | (1 << 28)); - ring->rptr = RREG32(CP_RB_RPTR); - evergreen_cp_start(rdev); ring->ready = true; r = radeon_ring_test(rdev, RADEON_RING_TYPE_GFX_INDEX, ring); @@ -3952,11 +3950,9 @@ bool evergreen_gfx_is_lockup(struct radeon_device *rdev, struct radeon_ring *rin if (!(reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE | RADEON_RESET_CP))) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force CP activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c index c7cac07f139b..5c8b358f9fba 100644 --- a/drivers/gpu/drm/radeon/evergreen_cs.c +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -1165,7 +1165,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); break; case DB_DEPTH_CONTROL: track->db_depth_control = radeon_get_ib_value(p, idx); @@ -1196,12 +1196,12 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } ib[idx] &= ~Z_ARRAY_MODE(0xf); track->db_z_info &= ~Z_ARRAY_MODE(0xf); - ib[idx] |= Z_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags)); - track->db_z_info |= Z_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags)); - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + ib[idx] |= Z_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->tiling_flags)); + track->db_z_info |= Z_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->tiling_flags)); + if (reloc->tiling_flags & RADEON_TILING_MACRO) { unsigned bankw, bankh, mtaspect, tile_split; - evergreen_tiling_fields(reloc->lobj.tiling_flags, + evergreen_tiling_fields(reloc->tiling_flags, &bankw, &bankh, &mtaspect, &tile_split); ib[idx] |= DB_NUM_BANKS(evergreen_cs_get_num_banks(track->nbanks)); @@ -1237,7 +1237,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } track->db_z_read_offset = radeon_get_ib_value(p, idx); - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->db_z_read_bo = reloc->robj; track->db_dirty = true; break; @@ -1249,7 +1249,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } track->db_z_write_offset = radeon_get_ib_value(p, idx); - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->db_z_write_bo = reloc->robj; track->db_dirty = true; break; @@ -1261,7 +1261,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } track->db_s_read_offset = radeon_get_ib_value(p, idx); - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->db_s_read_bo = reloc->robj; track->db_dirty = true; break; @@ -1273,7 +1273,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } track->db_s_write_offset = radeon_get_ib_value(p, idx); - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->db_s_write_bo = reloc->robj; track->db_dirty = true; break; @@ -1297,7 +1297,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } tmp = (reg - VGT_STRMOUT_BUFFER_BASE_0) / 16; track->vgt_strmout_bo_offset[tmp] = radeon_get_ib_value(p, idx) << 8; - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->vgt_strmout_bo[tmp] = reloc->robj; track->streamout_dirty = true; break; @@ -1317,7 +1317,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); case CB_TARGET_MASK: track->cb_target_mask = radeon_get_ib_value(p, idx); track->cb_dirty = true; @@ -1381,8 +1381,8 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] |= CB_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags)); - track->cb_color_info[tmp] |= CB_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags)); + ib[idx] |= CB_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->tiling_flags)); + track->cb_color_info[tmp] |= CB_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->tiling_flags)); } track->cb_dirty = true; break; @@ -1399,8 +1399,8 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] |= CB_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags)); - track->cb_color_info[tmp] |= CB_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags)); + ib[idx] |= CB_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->tiling_flags)); + track->cb_color_info[tmp] |= CB_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->tiling_flags)); } track->cb_dirty = true; break; @@ -1461,10 +1461,10 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + if (reloc->tiling_flags & RADEON_TILING_MACRO) { unsigned bankw, bankh, mtaspect, tile_split; - evergreen_tiling_fields(reloc->lobj.tiling_flags, + evergreen_tiling_fields(reloc->tiling_flags, &bankw, &bankh, &mtaspect, &tile_split); ib[idx] |= CB_NUM_BANKS(evergreen_cs_get_num_banks(track->nbanks)); @@ -1489,10 +1489,10 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + if (reloc->tiling_flags & RADEON_TILING_MACRO) { unsigned bankw, bankh, mtaspect, tile_split; - evergreen_tiling_fields(reloc->lobj.tiling_flags, + evergreen_tiling_fields(reloc->tiling_flags, &bankw, &bankh, &mtaspect, &tile_split); ib[idx] |= CB_NUM_BANKS(evergreen_cs_get_num_banks(track->nbanks)); @@ -1520,7 +1520,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) dev_err(p->dev, "bad SET_CONTEXT_REG 0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->cb_color_fmask_bo[tmp] = reloc->robj; break; case CB_COLOR0_CMASK: @@ -1537,7 +1537,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) dev_err(p->dev, "bad SET_CONTEXT_REG 0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->cb_color_cmask_bo[tmp] = reloc->robj; break; case CB_COLOR0_FMASK_SLICE: @@ -1578,7 +1578,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } tmp = (reg - CB_COLOR0_BASE) / 0x3c; track->cb_color_bo_offset[tmp] = radeon_get_ib_value(p, idx); - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->cb_color_bo[tmp] = reloc->robj; track->cb_dirty = true; break; @@ -1594,7 +1594,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } tmp = ((reg - CB_COLOR8_BASE) / 0x1c) + 8; track->cb_color_bo_offset[tmp] = radeon_get_ib_value(p, idx); - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->cb_color_bo[tmp] = reloc->robj; track->cb_dirty = true; break; @@ -1606,7 +1606,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } track->htile_offset = radeon_get_ib_value(p, idx); - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->htile_bo = reloc->robj; track->db_dirty = true; break; @@ -1723,7 +1723,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); break; case SX_MEMORY_EXPORT_BASE: if (p->rdev->family >= CHIP_CAYMAN) { @@ -1737,7 +1737,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); break; case CAYMAN_SX_SCATTER_EXPORT_BASE: if (p->rdev->family < CHIP_CAYMAN) { @@ -1751,7 +1751,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); break; case SX_MISC: track->sx_misc_kill_all_prims = (radeon_get_ib_value(p, idx) & 0x1) != 0; @@ -1836,7 +1836,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (idx_value & 0xfffffff0) + ((u64)(tmp & 0xff) << 32); @@ -1882,7 +1882,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + idx_value + ((u64)(radeon_get_ib_value(p, idx+1) & 0xff) << 32); @@ -1909,7 +1909,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + idx_value + ((u64)(radeon_get_ib_value(p, idx+1) & 0xff) << 32); @@ -1937,7 +1937,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + radeon_get_ib_value(p, idx+1) + ((u64)(radeon_get_ib_value(p, idx+2) & 0xff) << 32); @@ -2027,7 +2027,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad DISPATCH_INDIRECT\n"); return -EINVAL; } - ib[idx+0] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + ib[idx+0] = idx_value + (u32)(reloc->gpu_offset & 0xffffffff); r = evergreen_cs_track_check(p); if (r) { dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); @@ -2049,7 +2049,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (radeon_get_ib_value(p, idx+1) & 0xfffffffc) + ((u64)(radeon_get_ib_value(p, idx+2) & 0xff) << 32); @@ -2106,7 +2106,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, tmp = radeon_get_ib_value(p, idx) + ((u64)(radeon_get_ib_value(p, idx+1) & 0xff) << 32); - offset = reloc->lobj.gpu_offset + tmp; + offset = reloc->gpu_offset + tmp; if ((tmp + size) > radeon_bo_size(reloc->robj)) { dev_warn(p->dev, "CP DMA src buffer too small (%llu %lu)\n", @@ -2144,7 +2144,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, tmp = radeon_get_ib_value(p, idx+2) + ((u64)(radeon_get_ib_value(p, idx+3) & 0xff) << 32); - offset = reloc->lobj.gpu_offset + tmp; + offset = reloc->gpu_offset + tmp; if ((tmp + size) > radeon_bo_size(reloc->robj)) { dev_warn(p->dev, "CP DMA dst buffer too small (%llu %lu)\n", @@ -2174,7 +2174,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad SURFACE_SYNC\n"); return -EINVAL; } - ib[idx+2] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx+2] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); } break; case PACKET3_EVENT_WRITE: @@ -2190,7 +2190,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad EVENT_WRITE\n"); return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (radeon_get_ib_value(p, idx+1) & 0xfffffff8) + ((u64)(radeon_get_ib_value(p, idx+2) & 0xff) << 32); @@ -2212,7 +2212,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (radeon_get_ib_value(p, idx+1) & 0xfffffffc) + ((u64)(radeon_get_ib_value(p, idx+2) & 0xff) << 32); @@ -2234,7 +2234,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (radeon_get_ib_value(p, idx+1) & 0xfffffffc) + ((u64)(radeon_get_ib_value(p, idx+2) & 0xff) << 32); @@ -2302,11 +2302,11 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, } if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { ib[idx+1+(i*8)+1] |= - TEX_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags)); - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + TEX_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->tiling_flags)); + if (reloc->tiling_flags & RADEON_TILING_MACRO) { unsigned bankw, bankh, mtaspect, tile_split; - evergreen_tiling_fields(reloc->lobj.tiling_flags, + evergreen_tiling_fields(reloc->tiling_flags, &bankw, &bankh, &mtaspect, &tile_split); ib[idx+1+(i*8)+6] |= TEX_TILE_SPLIT(tile_split); @@ -2318,7 +2318,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, } } texture = reloc->robj; - toffset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + toffset = (u32)((reloc->gpu_offset >> 8) & 0xffffffff); /* tex mip base */ tex_dim = ib[idx+1+(i*8)+0] & 0x7; @@ -2337,7 +2337,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad SET_RESOURCE (tex)\n"); return -EINVAL; } - moffset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + moffset = (u32)((reloc->gpu_offset >> 8) & 0xffffffff); mipmap = reloc->robj; } @@ -2364,7 +2364,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, ib[idx+1+(i*8)+1] = radeon_bo_size(reloc->robj) - offset; } - offset64 = reloc->lobj.gpu_offset + offset; + offset64 = reloc->gpu_offset + offset; ib[idx+1+(i*8)+0] = offset64; ib[idx+1+(i*8)+2] = (ib[idx+1+(i*8)+2] & 0xffffff00) | (upper_32_bits(offset64) & 0xff); @@ -2445,7 +2445,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+1] = offset; ib[idx+2] = upper_32_bits(offset) & 0xff; } @@ -2464,7 +2464,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+3] = offset; ib[idx+4] = upper_32_bits(offset) & 0xff; } @@ -2493,7 +2493,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, offset + 8, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+0] = offset; ib[idx+1] = upper_32_bits(offset) & 0xff; break; @@ -2518,7 +2518,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+1] = offset; ib[idx+2] = upper_32_bits(offset) & 0xff; } else { @@ -2542,7 +2542,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+3] = offset; ib[idx+4] = upper_32_bits(offset) & 0xff; } else { @@ -2717,7 +2717,7 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst_offset = radeon_get_ib_value(p, idx+1); dst_offset <<= 8; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); p->idx += count + 7; break; /* linear */ @@ -2725,8 +2725,8 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst_offset = radeon_get_ib_value(p, idx+1); dst_offset |= ((u64)(radeon_get_ib_value(p, idx+2) & 0xff)) << 32; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+2] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+2] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; p->idx += count + 3; break; default: @@ -2768,10 +2768,10 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst_offset + (count * 4), radeon_bo_size(dst_reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+2] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+3] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; - ib[idx+4] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+2] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+3] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; + ib[idx+4] += upper_32_bits(src_reloc->gpu_offset) & 0xff; p->idx += 5; break; /* Copy L2T/T2L */ @@ -2781,22 +2781,22 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) /* tiled src, linear dst */ src_offset = radeon_get_ib_value(p, idx+1); src_offset <<= 8; - ib[idx+1] += (u32)(src_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(src_reloc->gpu_offset >> 8); dst_offset = radeon_get_ib_value(p, idx + 7); dst_offset |= ((u64)(radeon_get_ib_value(p, idx+8) & 0xff)) << 32; - ib[idx+7] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+8] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; + ib[idx+7] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+8] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; } else { /* linear src, tiled dst */ src_offset = radeon_get_ib_value(p, idx+7); src_offset |= ((u64)(radeon_get_ib_value(p, idx+8) & 0xff)) << 32; - ib[idx+7] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+8] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+7] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+8] += upper_32_bits(src_reloc->gpu_offset) & 0xff; dst_offset = radeon_get_ib_value(p, idx+1); dst_offset <<= 8; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); } if ((src_offset + (count * 4)) > radeon_bo_size(src_reloc->robj)) { dev_warn(p->dev, "DMA L2T, src buffer too small (%llu %lu)\n", @@ -2827,10 +2827,10 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst_offset + count, radeon_bo_size(dst_reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xffffffff); - ib[idx+2] += (u32)(src_reloc->lobj.gpu_offset & 0xffffffff); - ib[idx+3] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; - ib[idx+4] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xffffffff); + ib[idx+2] += (u32)(src_reloc->gpu_offset & 0xffffffff); + ib[idx+3] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; + ib[idx+4] += upper_32_bits(src_reloc->gpu_offset) & 0xff; p->idx += 5; break; /* Copy L2L, partial */ @@ -2840,10 +2840,10 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) DRM_ERROR("L2L Partial is cayman only !\n"); return -EINVAL; } - ib[idx+1] += (u32)(src_reloc->lobj.gpu_offset & 0xffffffff); - ib[idx+2] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; - ib[idx+4] += (u32)(dst_reloc->lobj.gpu_offset & 0xffffffff); - ib[idx+5] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(src_reloc->gpu_offset & 0xffffffff); + ib[idx+2] += upper_32_bits(src_reloc->gpu_offset) & 0xff; + ib[idx+4] += (u32)(dst_reloc->gpu_offset & 0xffffffff); + ib[idx+5] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; p->idx += 9; break; @@ -2876,12 +2876,12 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst2_offset + (count * 4), radeon_bo_size(dst2_reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+2] += (u32)(dst2_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+3] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+4] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; - ib[idx+5] += upper_32_bits(dst2_reloc->lobj.gpu_offset) & 0xff; - ib[idx+6] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+2] += (u32)(dst2_reloc->gpu_offset & 0xfffffffc); + ib[idx+3] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+4] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; + ib[idx+5] += upper_32_bits(dst2_reloc->gpu_offset) & 0xff; + ib[idx+6] += upper_32_bits(src_reloc->gpu_offset) & 0xff; p->idx += 7; break; /* Copy L2T Frame to Field */ @@ -2916,10 +2916,10 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst2_offset + (count * 4), radeon_bo_size(dst2_reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); - ib[idx+2] += (u32)(dst2_reloc->lobj.gpu_offset >> 8); - ib[idx+8] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+9] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); + ib[idx+2] += (u32)(dst2_reloc->gpu_offset >> 8); + ib[idx+8] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+9] += upper_32_bits(src_reloc->gpu_offset) & 0xff; p->idx += 10; break; /* Copy L2T/T2L, partial */ @@ -2932,16 +2932,16 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) /* detile bit */ if (radeon_get_ib_value(p, idx + 2) & (1 << 31)) { /* tiled src, linear dst */ - ib[idx+1] += (u32)(src_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(src_reloc->gpu_offset >> 8); - ib[idx+7] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+8] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; + ib[idx+7] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+8] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; } else { /* linear src, tiled dst */ - ib[idx+7] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+8] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+7] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+8] += upper_32_bits(src_reloc->gpu_offset) & 0xff; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); } p->idx += 12; break; @@ -2978,10 +2978,10 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst2_offset + (count * 4), radeon_bo_size(dst2_reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); - ib[idx+2] += (u32)(dst2_reloc->lobj.gpu_offset >> 8); - ib[idx+8] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+9] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); + ib[idx+2] += (u32)(dst2_reloc->gpu_offset >> 8); + ib[idx+8] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+9] += upper_32_bits(src_reloc->gpu_offset) & 0xff; p->idx += 10; break; /* Copy L2T/T2L (tile units) */ @@ -2992,22 +2992,22 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) /* tiled src, linear dst */ src_offset = radeon_get_ib_value(p, idx+1); src_offset <<= 8; - ib[idx+1] += (u32)(src_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(src_reloc->gpu_offset >> 8); dst_offset = radeon_get_ib_value(p, idx+7); dst_offset |= ((u64)(radeon_get_ib_value(p, idx+8) & 0xff)) << 32; - ib[idx+7] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+8] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; + ib[idx+7] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+8] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; } else { /* linear src, tiled dst */ src_offset = radeon_get_ib_value(p, idx+7); src_offset |= ((u64)(radeon_get_ib_value(p, idx+8) & 0xff)) << 32; - ib[idx+7] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+8] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+7] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+8] += upper_32_bits(src_reloc->gpu_offset) & 0xff; dst_offset = radeon_get_ib_value(p, idx+1); dst_offset <<= 8; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); } if ((src_offset + (count * 4)) > radeon_bo_size(src_reloc->robj)) { dev_warn(p->dev, "DMA L2T, T2L src buffer too small (%llu %lu)\n", @@ -3028,8 +3028,8 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) DRM_ERROR("L2T, T2L Partial is cayman only !\n"); return -EINVAL; } - ib[idx+1] += (u32)(src_reloc->lobj.gpu_offset >> 8); - ib[idx+4] += (u32)(dst_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(src_reloc->gpu_offset >> 8); + ib[idx+4] += (u32)(dst_reloc->gpu_offset >> 8); p->idx += 13; break; /* Copy L2T broadcast (tile units) */ @@ -3065,10 +3065,10 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst2_offset + (count * 4), radeon_bo_size(dst2_reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); - ib[idx+2] += (u32)(dst2_reloc->lobj.gpu_offset >> 8); - ib[idx+8] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+9] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); + ib[idx+2] += (u32)(dst2_reloc->gpu_offset >> 8); + ib[idx+8] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+9] += upper_32_bits(src_reloc->gpu_offset) & 0xff; p->idx += 10; break; default: @@ -3089,8 +3089,8 @@ int evergreen_dma_cs_parse(struct radeon_cs_parser *p) dst_offset, radeon_bo_size(dst_reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+3] += (upper_32_bits(dst_reloc->lobj.gpu_offset) << 16) & 0x00ff0000; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+3] += (upper_32_bits(dst_reloc->gpu_offset) << 16) & 0x00ff0000; p->idx += 4; break; case DMA_PACKET_NOP: diff --git a/drivers/gpu/drm/radeon/evergreen_dma.c b/drivers/gpu/drm/radeon/evergreen_dma.c index a37b54436382..287fe966d7de 100644 --- a/drivers/gpu/drm/radeon/evergreen_dma.c +++ b/drivers/gpu/drm/radeon/evergreen_dma.c @@ -174,11 +174,9 @@ bool evergreen_dma_is_lockup(struct radeon_device *rdev, struct radeon_ring *rin u32 reset_mask = evergreen_gpu_check_soft_reset(rdev); if (!(reset_mask & RADEON_RESET_DMA)) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force ring activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } diff --git a/drivers/gpu/drm/radeon/kv_dpm.c b/drivers/gpu/drm/radeon/kv_dpm.c index 351db361239d..16ec9d56a234 100644 --- a/drivers/gpu/drm/radeon/kv_dpm.c +++ b/drivers/gpu/drm/radeon/kv_dpm.c @@ -1338,13 +1338,11 @@ static int kv_enable_uvd_dpm(struct radeon_device *rdev, bool enable) PPSMC_MSG_UVDDPM_Enable : PPSMC_MSG_UVDDPM_Disable); } -#if 0 static int kv_enable_vce_dpm(struct radeon_device *rdev, bool enable) { return kv_notify_message_to_smu(rdev, enable ? PPSMC_MSG_VCEDPM_Enable : PPSMC_MSG_VCEDPM_Disable); } -#endif static int kv_enable_samu_dpm(struct radeon_device *rdev, bool enable) { @@ -1389,7 +1387,6 @@ static int kv_update_uvd_dpm(struct radeon_device *rdev, bool gate) return kv_enable_uvd_dpm(rdev, !gate); } -#if 0 static u8 kv_get_vce_boot_level(struct radeon_device *rdev) { u8 i; @@ -1414,6 +1411,9 @@ static int kv_update_vce_dpm(struct radeon_device *rdev, int ret; if (radeon_new_state->evclk > 0 && radeon_current_state->evclk == 0) { + kv_dpm_powergate_vce(rdev, false); + /* turn the clocks on when encoding */ + cik_update_cg(rdev, RADEON_CG_BLOCK_VCE, false); if (pi->caps_stable_p_state) pi->vce_boot_level = table->count - 1; else @@ -1436,11 +1436,13 @@ static int kv_update_vce_dpm(struct radeon_device *rdev, kv_enable_vce_dpm(rdev, true); } else if (radeon_new_state->evclk == 0 && radeon_current_state->evclk > 0) { kv_enable_vce_dpm(rdev, false); + /* turn the clocks off when not encoding */ + cik_update_cg(rdev, RADEON_CG_BLOCK_VCE, true); + kv_dpm_powergate_vce(rdev, true); } return 0; } -#endif static int kv_update_samu_dpm(struct radeon_device *rdev, bool gate) { @@ -1575,11 +1577,16 @@ static void kv_dpm_powergate_vce(struct radeon_device *rdev, bool gate) pi->vce_power_gated = gate; if (gate) { - if (pi->caps_vce_pg) + if (pi->caps_vce_pg) { + /* XXX do we need a vce_v1_0_stop() ? */ kv_notify_message_to_smu(rdev, PPSMC_MSG_VCEPowerOFF); + } } else { - if (pi->caps_vce_pg) + if (pi->caps_vce_pg) { kv_notify_message_to_smu(rdev, PPSMC_MSG_VCEPowerON); + vce_v2_0_resume(rdev); + vce_v1_0_start(rdev); + } } } @@ -1768,7 +1775,7 @@ int kv_dpm_set_power_state(struct radeon_device *rdev) { struct kv_power_info *pi = kv_get_pi(rdev); struct radeon_ps *new_ps = &pi->requested_rps; - /*struct radeon_ps *old_ps = &pi->current_rps;*/ + struct radeon_ps *old_ps = &pi->current_rps; int ret; if (pi->bapm_enable) { @@ -1798,13 +1805,12 @@ int kv_dpm_set_power_state(struct radeon_device *rdev) kv_set_enabled_levels(rdev); kv_force_lowest_valid(rdev); kv_unforce_levels(rdev); -#if 0 + ret = kv_update_vce_dpm(rdev, new_ps, old_ps); if (ret) { DRM_ERROR("kv_update_vce_dpm failed\n"); return ret; } -#endif kv_update_sclk_t(rdev); } } else { @@ -1823,13 +1829,11 @@ int kv_dpm_set_power_state(struct radeon_device *rdev) kv_program_nbps_index_settings(rdev, new_ps); kv_freeze_sclk_dpm(rdev, false); kv_set_enabled_levels(rdev); -#if 0 ret = kv_update_vce_dpm(rdev, new_ps, old_ps); if (ret) { DRM_ERROR("kv_update_vce_dpm failed\n"); return ret; } -#endif kv_update_acp_boot_level(rdev); kv_update_sclk_t(rdev); kv_enable_nb_dpm(rdev); @@ -2037,6 +2041,14 @@ static void kv_apply_state_adjust_rules(struct radeon_device *rdev, struct radeon_clock_and_voltage_limits *max_limits = &rdev->pm.dpm.dyn_state.max_clock_voltage_on_ac; + if (new_rps->vce_active) { + new_rps->evclk = rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].evclk; + new_rps->ecclk = rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].ecclk; + } else { + new_rps->evclk = 0; + new_rps->ecclk = 0; + } + mclk = max_limits->mclk; sclk = min_sclk; @@ -2056,6 +2068,11 @@ static void kv_apply_state_adjust_rules(struct radeon_device *rdev, sclk = stable_p_state_sclk; } + if (new_rps->vce_active) { + if (sclk < rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].sclk) + sclk = rdev->pm.dpm.vce_states[rdev->pm.dpm.vce_level].sclk; + } + ps->need_dfs_bypass = true; for (i = 0; i < ps->num_levels; i++) { @@ -2092,7 +2109,8 @@ static void kv_apply_state_adjust_rules(struct radeon_device *rdev, } } - pi->video_start = new_rps->dclk || new_rps->vclk; + pi->video_start = new_rps->dclk || new_rps->vclk || + new_rps->evclk || new_rps->ecclk; if ((new_rps->class & ATOM_PPLIB_CLASSIFICATION_UI_MASK) == ATOM_PPLIB_CLASSIFICATION_UI_BATTERY) @@ -2538,9 +2556,6 @@ static int kv_parse_power_table(struct radeon_device *rdev) if (!rdev->pm.dpm.ps) return -ENOMEM; power_state_offset = (u8 *)state_array->states; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < state_array->ucNumEntries; i++) { u8 *idx; power_state = (union pplib_power_state *)power_state_offset; @@ -2577,6 +2592,19 @@ static int kv_parse_power_table(struct radeon_device *rdev) power_state_offset += 2 + power_state->v2.ucNumDPMLevels; } rdev->pm.dpm.num_ps = state_array->ucNumEntries; + + /* fill in the vce power states */ + for (i = 0; i < RADEON_MAX_VCE_LEVELS; i++) { + u32 sclk; + clock_array_index = rdev->pm.dpm.vce_states[i].clk_idx; + clock_info = (union pplib_clock_info *) + &clock_info_array->clockInfo[clock_array_index * clock_info_array->ucEntrySize]; + sclk = le16_to_cpu(clock_info->sumo.usEngineClockLow); + sclk |= clock_info->sumo.ucEngineClockHigh << 16; + rdev->pm.dpm.vce_states[i].sclk = sclk; + rdev->pm.dpm.vce_states[i].mclk = 0; + } + return 0; } @@ -2590,6 +2618,10 @@ int kv_dpm_init(struct radeon_device *rdev) return -ENOMEM; rdev->pm.dpm.priv = pi; + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = r600_parse_extended_power_table(rdev); if (ret) return ret; @@ -2623,7 +2655,7 @@ int kv_dpm_init(struct radeon_device *rdev) pi->caps_fps = false; /* true? */ pi->caps_uvd_pg = true; pi->caps_uvd_dpm = true; - pi->caps_vce_pg = false; + pi->caps_vce_pg = false; /* XXX true */ pi->caps_samu_pg = false; pi->caps_acp_pg = false; pi->caps_stable_p_state = false; diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c index bf6300cfd62d..d246e043421a 100644 --- a/drivers/gpu/drm/radeon/ni.c +++ b/drivers/gpu/drm/radeon/ni.c @@ -1642,8 +1642,8 @@ static int cayman_cp_resume(struct radeon_device *rdev) ring = &rdev->ring[ridx[i]]; WREG32_P(cp_rb_cntl[i], RB_RPTR_WR_ENA, ~RB_RPTR_WR_ENA); - ring->rptr = ring->wptr = 0; - WREG32(cp_rb_rptr[i], ring->rptr); + ring->wptr = 0; + WREG32(cp_rb_rptr[i], 0); WREG32(cp_rb_wptr[i], ring->wptr); mdelay(1); @@ -1917,11 +1917,9 @@ bool cayman_gfx_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) if (!(reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE | RADEON_RESET_CP))) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force CP activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } diff --git a/drivers/gpu/drm/radeon/ni_dma.c b/drivers/gpu/drm/radeon/ni_dma.c index 7cf96b15377f..6378e0276691 100644 --- a/drivers/gpu/drm/radeon/ni_dma.c +++ b/drivers/gpu/drm/radeon/ni_dma.c @@ -248,8 +248,6 @@ int cayman_dma_resume(struct radeon_device *rdev) ring->wptr = 0; WREG32(DMA_RB_WPTR + reg_offset, ring->wptr << 2); - ring->rptr = RREG32(DMA_RB_RPTR + reg_offset) >> 2; - WREG32(DMA_RB_CNTL + reg_offset, rb_cntl | DMA_RB_ENABLE); ring->ready = true; @@ -302,11 +300,9 @@ bool cayman_dma_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) mask = RADEON_RESET_DMA1; if (!(reset_mask & mask)) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force ring activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } diff --git a/drivers/gpu/drm/radeon/ni_dpm.c b/drivers/gpu/drm/radeon/ni_dpm.c index ca814276b075..004c931606c4 100644 --- a/drivers/gpu/drm/radeon/ni_dpm.c +++ b/drivers/gpu/drm/radeon/ni_dpm.c @@ -4025,9 +4025,6 @@ static int ni_parse_power_table(struct radeon_device *rdev) power_info->pplib.ucNumStates, GFP_KERNEL); if (!rdev->pm.dpm.ps) return -ENOMEM; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < power_info->pplib.ucNumStates; i++) { power_state = (union pplib_power_state *) @@ -4089,6 +4086,10 @@ int ni_dpm_init(struct radeon_device *rdev) pi->min_vddc_in_table = 0; pi->max_vddc_in_table = 0; + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = ni_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c index 3cc78bb66042..b6c32640df20 100644 --- a/drivers/gpu/drm/radeon/r100.c +++ b/drivers/gpu/drm/radeon/r100.c @@ -1193,7 +1193,6 @@ int r100_cp_init(struct radeon_device *rdev, unsigned ring_size) WREG32(RADEON_CP_RB_CNTL, tmp); udelay(10); - ring->rptr = RREG32(RADEON_CP_RB_RPTR); /* Set cp mode to bus mastering & enable cp*/ WREG32(RADEON_CP_CSQ_MODE, REG_SET(RADEON_INDIRECT2_START, indirect2_start) | @@ -1275,12 +1274,12 @@ int r100_reloc_pitch_offset(struct radeon_cs_parser *p, value = radeon_get_ib_value(p, idx); tmp = value & 0x003fffff; - tmp += (((u32)reloc->lobj.gpu_offset) >> 10); + tmp += (((u32)reloc->gpu_offset) >> 10); if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) tile_flags |= RADEON_DST_TILE_MACRO; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { + if (reloc->tiling_flags & RADEON_TILING_MICRO) { if (reg == RADEON_SRC_PITCH_OFFSET) { DRM_ERROR("Cannot src blit from microtiled surface\n"); radeon_cs_dump_packet(p, pkt); @@ -1326,7 +1325,7 @@ int r100_packet3_load_vbpntr(struct radeon_cs_parser *p, return r; } idx_value = radeon_get_ib_value(p, idx); - ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->gpu_offset); track->arrays[i + 0].esize = idx_value >> 8; track->arrays[i + 0].robj = reloc->robj; @@ -1338,7 +1337,7 @@ int r100_packet3_load_vbpntr(struct radeon_cs_parser *p, radeon_cs_dump_packet(p, pkt); return r; } - ib[idx+2] = radeon_get_ib_value(p, idx + 2) + ((u32)reloc->lobj.gpu_offset); + ib[idx+2] = radeon_get_ib_value(p, idx + 2) + ((u32)reloc->gpu_offset); track->arrays[i + 1].robj = reloc->robj; track->arrays[i + 1].esize = idx_value >> 24; track->arrays[i + 1].esize &= 0x7F; @@ -1352,7 +1351,7 @@ int r100_packet3_load_vbpntr(struct radeon_cs_parser *p, return r; } idx_value = radeon_get_ib_value(p, idx); - ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->gpu_offset); track->arrays[i + 0].robj = reloc->robj; track->arrays[i + 0].esize = idx_value >> 8; track->arrays[i + 0].esize &= 0x7F; @@ -1595,7 +1594,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, track->zb.robj = reloc->robj; track->zb.offset = idx_value; track->zb_dirty = true; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case RADEON_RB3D_COLOROFFSET: r = radeon_cs_packet_next_reloc(p, &reloc, 0); @@ -1608,7 +1607,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, track->cb[0].robj = reloc->robj; track->cb[0].offset = idx_value; track->cb_dirty = true; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case RADEON_PP_TXOFFSET_0: case RADEON_PP_TXOFFSET_1: @@ -1622,16 +1621,16 @@ static int r100_packet0_check(struct radeon_cs_parser *p, return r; } if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) tile_flags |= RADEON_TXO_MACRO_TILE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + if (reloc->tiling_flags & RADEON_TILING_MICRO) tile_flags |= RADEON_TXO_MICRO_TILE_X2; tmp = idx_value & ~(0x7 << 2); tmp |= tile_flags; - ib[idx] = tmp + ((u32)reloc->lobj.gpu_offset); + ib[idx] = tmp + ((u32)reloc->gpu_offset); } else - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); track->textures[i].robj = reloc->robj; track->tex_dirty = true; break; @@ -1649,7 +1648,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, return r; } track->textures[0].cube_info[i].offset = idx_value; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); track->textures[0].cube_info[i].robj = reloc->robj; track->tex_dirty = true; break; @@ -1667,7 +1666,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, return r; } track->textures[1].cube_info[i].offset = idx_value; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); track->textures[1].cube_info[i].robj = reloc->robj; track->tex_dirty = true; break; @@ -1685,7 +1684,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, return r; } track->textures[2].cube_info[i].offset = idx_value; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); track->textures[2].cube_info[i].robj = reloc->robj; track->tex_dirty = true; break; @@ -1703,9 +1702,9 @@ static int r100_packet0_check(struct radeon_cs_parser *p, return r; } if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) tile_flags |= RADEON_COLOR_TILE_ENABLE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + if (reloc->tiling_flags & RADEON_TILING_MICRO) tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; tmp = idx_value & ~(0x7 << 16); @@ -1773,7 +1772,7 @@ static int r100_packet0_check(struct radeon_cs_parser *p, radeon_cs_dump_packet(p, pkt); return r; } - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case RADEON_PP_CNTL: { @@ -1933,7 +1932,7 @@ static int r100_packet3_check(struct radeon_cs_parser *p, radeon_cs_dump_packet(p, pkt); return r; } - ib[idx+1] = radeon_get_ib_value(p, idx+1) + ((u32)reloc->lobj.gpu_offset); + ib[idx+1] = radeon_get_ib_value(p, idx+1) + ((u32)reloc->gpu_offset); r = r100_cs_track_check_pkt3_indx_buffer(p, pkt, reloc->robj); if (r) { return r; @@ -1947,7 +1946,7 @@ static int r100_packet3_check(struct radeon_cs_parser *p, radeon_cs_dump_packet(p, pkt); return r; } - ib[idx] = radeon_get_ib_value(p, idx) + ((u32)reloc->lobj.gpu_offset); + ib[idx] = radeon_get_ib_value(p, idx) + ((u32)reloc->gpu_offset); track->num_arrays = 1; track->vtx_size = r100_get_vtx_size(radeon_get_ib_value(p, idx + 2)); @@ -2523,11 +2522,9 @@ bool r100_gpu_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) rbbm_status = RREG32(R_000E40_RBBM_STATUS); if (!G_000E40_GUI_ACTIVE(rbbm_status)) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force CP activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } @@ -3223,12 +3220,12 @@ void r100_bandwidth_update(struct radeon_device *rdev) if (rdev->mode_info.crtcs[0]->base.enabled) { mode1 = &rdev->mode_info.crtcs[0]->base.mode; - pixel_bytes1 = rdev->mode_info.crtcs[0]->base.fb->bits_per_pixel / 8; + pixel_bytes1 = rdev->mode_info.crtcs[0]->base.primary->fb->bits_per_pixel / 8; } if (!(rdev->flags & RADEON_SINGLE_CRTC)) { if (rdev->mode_info.crtcs[1]->base.enabled) { mode2 = &rdev->mode_info.crtcs[1]->base.mode; - pixel_bytes2 = rdev->mode_info.crtcs[1]->base.fb->bits_per_pixel / 8; + pixel_bytes2 = rdev->mode_info.crtcs[1]->base.primary->fb->bits_per_pixel / 8; } } diff --git a/drivers/gpu/drm/radeon/r200.c b/drivers/gpu/drm/radeon/r200.c index b3807edb1936..58f0473aa73f 100644 --- a/drivers/gpu/drm/radeon/r200.c +++ b/drivers/gpu/drm/radeon/r200.c @@ -185,7 +185,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, track->zb.robj = reloc->robj; track->zb.offset = idx_value; track->zb_dirty = true; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case RADEON_RB3D_COLOROFFSET: r = radeon_cs_packet_next_reloc(p, &reloc, 0); @@ -198,7 +198,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, track->cb[0].robj = reloc->robj; track->cb[0].offset = idx_value; track->cb_dirty = true; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case R200_PP_TXOFFSET_0: case R200_PP_TXOFFSET_1: @@ -215,16 +215,16 @@ int r200_packet0_check(struct radeon_cs_parser *p, return r; } if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) tile_flags |= R200_TXO_MACRO_TILE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + if (reloc->tiling_flags & RADEON_TILING_MICRO) tile_flags |= R200_TXO_MICRO_TILE; tmp = idx_value & ~(0x7 << 2); tmp |= tile_flags; - ib[idx] = tmp + ((u32)reloc->lobj.gpu_offset); + ib[idx] = tmp + ((u32)reloc->gpu_offset); } else - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); track->textures[i].robj = reloc->robj; track->tex_dirty = true; break; @@ -268,7 +268,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, return r; } track->textures[i].cube_info[face - 1].offset = idx_value; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); track->textures[i].cube_info[face - 1].robj = reloc->robj; track->tex_dirty = true; break; @@ -287,9 +287,9 @@ int r200_packet0_check(struct radeon_cs_parser *p, } if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) tile_flags |= RADEON_COLOR_TILE_ENABLE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + if (reloc->tiling_flags & RADEON_TILING_MICRO) tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; tmp = idx_value & ~(0x7 << 16); @@ -362,7 +362,7 @@ int r200_packet0_check(struct radeon_cs_parser *p, radeon_cs_dump_packet(p, pkt); return r; } - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case RADEON_PP_CNTL: { diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c index 0b658b34b33a..206caf9700b7 100644 --- a/drivers/gpu/drm/radeon/r300.c +++ b/drivers/gpu/drm/radeon/r300.c @@ -640,7 +640,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, track->cb[i].robj = reloc->robj; track->cb[i].offset = idx_value; track->cb_dirty = true; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case R300_ZB_DEPTHOFFSET: r = radeon_cs_packet_next_reloc(p, &reloc, 0); @@ -653,7 +653,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, track->zb.robj = reloc->robj; track->zb.offset = idx_value; track->zb_dirty = true; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case R300_TX_OFFSET_0: case R300_TX_OFFSET_0+4: @@ -682,16 +682,16 @@ static int r300_packet0_check(struct radeon_cs_parser *p, if (p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS) { ib[idx] = (idx_value & 31) | /* keep the 1st 5 bits */ - ((idx_value & ~31) + (u32)reloc->lobj.gpu_offset); + ((idx_value & ~31) + (u32)reloc->gpu_offset); } else { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) tile_flags |= R300_TXO_MACRO_TILE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + if (reloc->tiling_flags & RADEON_TILING_MICRO) tile_flags |= R300_TXO_MICRO_TILE; - else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO_SQUARE) + else if (reloc->tiling_flags & RADEON_TILING_MICRO_SQUARE) tile_flags |= R300_TXO_MICRO_TILE_SQUARE; - tmp = idx_value + ((u32)reloc->lobj.gpu_offset); + tmp = idx_value + ((u32)reloc->gpu_offset); tmp |= tile_flags; ib[idx] = tmp; } @@ -753,11 +753,11 @@ static int r300_packet0_check(struct radeon_cs_parser *p, return r; } - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) tile_flags |= R300_COLOR_TILE_ENABLE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + if (reloc->tiling_flags & RADEON_TILING_MICRO) tile_flags |= R300_COLOR_MICROTILE_ENABLE; - else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO_SQUARE) + else if (reloc->tiling_flags & RADEON_TILING_MICRO_SQUARE) tile_flags |= R300_COLOR_MICROTILE_SQUARE_ENABLE; tmp = idx_value & ~(0x7 << 16); @@ -838,11 +838,11 @@ static int r300_packet0_check(struct radeon_cs_parser *p, return r; } - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) tile_flags |= R300_DEPTHMACROTILE_ENABLE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + if (reloc->tiling_flags & RADEON_TILING_MICRO) tile_flags |= R300_DEPTHMICROTILE_TILED; - else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO_SQUARE) + else if (reloc->tiling_flags & RADEON_TILING_MICRO_SQUARE) tile_flags |= R300_DEPTHMICROTILE_TILED_SQUARE; tmp = idx_value & ~(0x7 << 16); @@ -1052,7 +1052,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, radeon_cs_dump_packet(p, pkt); return r; } - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case 0x4e0c: /* RB3D_COLOR_CHANNEL_MASK */ @@ -1097,7 +1097,7 @@ static int r300_packet0_check(struct radeon_cs_parser *p, track->aa.robj = reloc->robj; track->aa.offset = idx_value; track->aa_dirty = true; - ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + ib[idx] = idx_value + ((u32)reloc->gpu_offset); break; case R300_RB3D_AARESOLVE_PITCH: track->aa.pitch = idx_value & 0x3FFE; @@ -1162,7 +1162,7 @@ static int r300_packet3_check(struct radeon_cs_parser *p, radeon_cs_dump_packet(p, pkt); return r; } - ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->lobj.gpu_offset); + ib[idx+1] = radeon_get_ib_value(p, idx + 1) + ((u32)reloc->gpu_offset); r = r100_cs_track_check_pkt3_indx_buffer(p, pkt, reloc->robj); if (r) { return r; diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index 647ef4079217..6e887d004eba 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -1748,11 +1748,9 @@ bool r600_gfx_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) if (!(reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE | RADEON_RESET_CP))) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force CP activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } @@ -2604,8 +2602,6 @@ int r600_cp_resume(struct radeon_device *rdev) WREG32(CP_RB_BASE, ring->gpu_addr >> 8); WREG32(CP_DEBUG, (1 << 27) | (1 << 28)); - ring->rptr = RREG32(CP_RB_RPTR); - r600_cp_start(rdev); ring->ready = true; r = radeon_ring_test(rdev, RADEON_RING_TYPE_GFX_INDEX, ring); diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c index 2812c7d1ae6f..12511bb5fd6f 100644 --- a/drivers/gpu/drm/radeon/r600_cs.c +++ b/drivers/gpu/drm/radeon/r600_cs.c @@ -1022,7 +1022,7 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); break; case SQ_CONFIG: track->sq_config = radeon_get_ib_value(p, idx); @@ -1043,7 +1043,7 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) track->db_depth_info = radeon_get_ib_value(p, idx); ib[idx] &= C_028010_ARRAY_MODE; track->db_depth_info &= C_028010_ARRAY_MODE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + if (reloc->tiling_flags & RADEON_TILING_MACRO) { ib[idx] |= S_028010_ARRAY_MODE(V_028010_ARRAY_2D_TILED_THIN1); track->db_depth_info |= S_028010_ARRAY_MODE(V_028010_ARRAY_2D_TILED_THIN1); } else { @@ -1084,9 +1084,9 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } tmp = (reg - VGT_STRMOUT_BUFFER_BASE_0) / 16; track->vgt_strmout_bo_offset[tmp] = radeon_get_ib_value(p, idx) << 8; - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->vgt_strmout_bo[tmp] = reloc->robj; - track->vgt_strmout_bo_mc[tmp] = reloc->lobj.gpu_offset; + track->vgt_strmout_bo_mc[tmp] = reloc->gpu_offset; track->streamout_dirty = true; break; case VGT_STRMOUT_BUFFER_SIZE_0: @@ -1105,7 +1105,7 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); break; case R_028238_CB_TARGET_MASK: track->cb_target_mask = radeon_get_ib_value(p, idx); @@ -1142,10 +1142,10 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } tmp = (reg - R_0280A0_CB_COLOR0_INFO) / 4; track->cb_color_info[tmp] = radeon_get_ib_value(p, idx); - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { + if (reloc->tiling_flags & RADEON_TILING_MACRO) { ib[idx] |= S_0280A0_ARRAY_MODE(V_0280A0_ARRAY_2D_TILED_THIN1); track->cb_color_info[tmp] |= S_0280A0_ARRAY_MODE(V_0280A0_ARRAY_2D_TILED_THIN1); - } else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { + } else if (reloc->tiling_flags & RADEON_TILING_MICRO) { ib[idx] |= S_0280A0_ARRAY_MODE(V_0280A0_ARRAY_1D_TILED_THIN1); track->cb_color_info[tmp] |= S_0280A0_ARRAY_MODE(V_0280A0_ARRAY_1D_TILED_THIN1); } @@ -1214,7 +1214,7 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } track->cb_color_frag_bo[tmp] = reloc->robj; track->cb_color_frag_offset[tmp] = (u64)ib[idx] << 8; - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); } if (G_0280A0_TILE_MODE(track->cb_color_info[tmp])) { track->cb_dirty = true; @@ -1245,7 +1245,7 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } track->cb_color_tile_bo[tmp] = reloc->robj; track->cb_color_tile_offset[tmp] = (u64)ib[idx] << 8; - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); } if (G_0280A0_TILE_MODE(track->cb_color_info[tmp])) { track->cb_dirty = true; @@ -1281,10 +1281,10 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) } tmp = (reg - CB_COLOR0_BASE) / 4; track->cb_color_bo_offset[tmp] = radeon_get_ib_value(p, idx) << 8; - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->cb_color_base_last[tmp] = ib[idx]; track->cb_color_bo[tmp] = reloc->robj; - track->cb_color_bo_mc[tmp] = reloc->lobj.gpu_offset; + track->cb_color_bo_mc[tmp] = reloc->gpu_offset; track->cb_dirty = true; break; case DB_DEPTH_BASE: @@ -1295,9 +1295,9 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } track->db_offset = radeon_get_ib_value(p, idx) << 8; - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->db_bo = reloc->robj; - track->db_bo_mc = reloc->lobj.gpu_offset; + track->db_bo_mc = reloc->gpu_offset; track->db_dirty = true; break; case DB_HTILE_DATA_BASE: @@ -1308,7 +1308,7 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) return -EINVAL; } track->htile_offset = radeon_get_ib_value(p, idx) << 8; - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); track->htile_bo = reloc->robj; track->db_dirty = true; break; @@ -1377,7 +1377,7 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); break; case SX_MEMORY_EXPORT_BASE: r = radeon_cs_packet_next_reloc(p, &reloc, r600_nomm); @@ -1386,7 +1386,7 @@ static int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) "0x%04X\n", reg); return -EINVAL; } - ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); break; case SX_MISC: track->sx_misc_kill_all_prims = (radeon_get_ib_value(p, idx) & 0x1) != 0; @@ -1672,7 +1672,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (idx_value & 0xfffffff0) + ((u64)(tmp & 0xff) << 32); @@ -1713,7 +1713,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + idx_value + ((u64)(radeon_get_ib_value(p, idx+1) & 0xff) << 32); @@ -1765,7 +1765,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (radeon_get_ib_value(p, idx+1) & 0xfffffff0) + ((u64)(radeon_get_ib_value(p, idx+2) & 0xff) << 32); @@ -1805,7 +1805,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, tmp = radeon_get_ib_value(p, idx) + ((u64)(radeon_get_ib_value(p, idx+1) & 0xff) << 32); - offset = reloc->lobj.gpu_offset + tmp; + offset = reloc->gpu_offset + tmp; if ((tmp + size) > radeon_bo_size(reloc->robj)) { dev_warn(p->dev, "CP DMA src buffer too small (%llu %lu)\n", @@ -1835,7 +1835,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, tmp = radeon_get_ib_value(p, idx+2) + ((u64)(radeon_get_ib_value(p, idx+3) & 0xff) << 32); - offset = reloc->lobj.gpu_offset + tmp; + offset = reloc->gpu_offset + tmp; if ((tmp + size) > radeon_bo_size(reloc->robj)) { dev_warn(p->dev, "CP DMA dst buffer too small (%llu %lu)\n", @@ -1861,7 +1861,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad SURFACE_SYNC\n"); return -EINVAL; } - ib[idx+2] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx+2] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); } break; case PACKET3_EVENT_WRITE: @@ -1877,7 +1877,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad EVENT_WRITE\n"); return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (radeon_get_ib_value(p, idx+1) & 0xfffffff8) + ((u64)(radeon_get_ib_value(p, idx+2) & 0xff) << 32); @@ -1899,7 +1899,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } - offset = reloc->lobj.gpu_offset + + offset = reloc->gpu_offset + (radeon_get_ib_value(p, idx+1) & 0xfffffffc) + ((u64)(radeon_get_ib_value(p, idx+2) & 0xff) << 32); @@ -1964,11 +1964,11 @@ static int r600_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad SET_RESOURCE\n"); return -EINVAL; } - base_offset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + base_offset = (u32)((reloc->gpu_offset >> 8) & 0xffffffff); if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + if (reloc->tiling_flags & RADEON_TILING_MACRO) ib[idx+1+(i*7)+0] |= S_038000_TILE_MODE(V_038000_ARRAY_2D_TILED_THIN1); - else if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + else if (reloc->tiling_flags & RADEON_TILING_MICRO) ib[idx+1+(i*7)+0] |= S_038000_TILE_MODE(V_038000_ARRAY_1D_TILED_THIN1); } texture = reloc->robj; @@ -1978,13 +1978,13 @@ static int r600_packet3_check(struct radeon_cs_parser *p, DRM_ERROR("bad SET_RESOURCE\n"); return -EINVAL; } - mip_offset = (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + mip_offset = (u32)((reloc->gpu_offset >> 8) & 0xffffffff); mipmap = reloc->robj; r = r600_check_texture_resource(p, idx+(i*7)+1, texture, mipmap, base_offset + radeon_get_ib_value(p, idx+1+(i*7)+2), mip_offset + radeon_get_ib_value(p, idx+1+(i*7)+3), - reloc->lobj.tiling_flags); + reloc->tiling_flags); if (r) return r; ib[idx+1+(i*7)+2] += base_offset; @@ -2008,7 +2008,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, ib[idx+1+(i*7)+1] = radeon_bo_size(reloc->robj) - offset; } - offset64 = reloc->lobj.gpu_offset + offset; + offset64 = reloc->gpu_offset + offset; ib[idx+1+(i*8)+0] = offset64; ib[idx+1+(i*8)+2] = (ib[idx+1+(i*8)+2] & 0xffffff00) | (upper_32_bits(offset64) & 0xff); @@ -2118,7 +2118,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + ib[idx+1] += (u32)((reloc->gpu_offset >> 8) & 0xffffffff); } break; case PACKET3_SURFACE_BASE_UPDATE: @@ -2151,7 +2151,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+1] = offset; ib[idx+2] = upper_32_bits(offset) & 0xff; } @@ -2170,7 +2170,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+3] = offset; ib[idx+4] = upper_32_bits(offset) & 0xff; } @@ -2199,7 +2199,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, offset + 8, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+0] = offset; ib[idx+1] = upper_32_bits(offset) & 0xff; break; @@ -2224,7 +2224,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+1] = offset; ib[idx+2] = upper_32_bits(offset) & 0xff; } else { @@ -2248,7 +2248,7 @@ static int r600_packet3_check(struct radeon_cs_parser *p, offset + 4, radeon_bo_size(reloc->robj)); return -EINVAL; } - offset += reloc->lobj.gpu_offset; + offset += reloc->gpu_offset; ib[idx+3] = offset; ib[idx+4] = upper_32_bits(offset) & 0xff; } else { @@ -2505,14 +2505,14 @@ int r600_dma_cs_parse(struct radeon_cs_parser *p) dst_offset = radeon_get_ib_value(p, idx+1); dst_offset <<= 8; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); p->idx += count + 5; } else { dst_offset = radeon_get_ib_value(p, idx+1); dst_offset |= ((u64)(radeon_get_ib_value(p, idx+2) & 0xff)) << 32; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+2] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+2] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; p->idx += count + 3; } if ((dst_offset + (count * 4)) > radeon_bo_size(dst_reloc->robj)) { @@ -2539,22 +2539,22 @@ int r600_dma_cs_parse(struct radeon_cs_parser *p) /* tiled src, linear dst */ src_offset = radeon_get_ib_value(p, idx+1); src_offset <<= 8; - ib[idx+1] += (u32)(src_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(src_reloc->gpu_offset >> 8); dst_offset = radeon_get_ib_value(p, idx+5); dst_offset |= ((u64)(radeon_get_ib_value(p, idx+6) & 0xff)) << 32; - ib[idx+5] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+6] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; + ib[idx+5] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+6] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; } else { /* linear src, tiled dst */ src_offset = radeon_get_ib_value(p, idx+5); src_offset |= ((u64)(radeon_get_ib_value(p, idx+6) & 0xff)) << 32; - ib[idx+5] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+6] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+5] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+6] += upper_32_bits(src_reloc->gpu_offset) & 0xff; dst_offset = radeon_get_ib_value(p, idx+1); dst_offset <<= 8; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset >> 8); + ib[idx+1] += (u32)(dst_reloc->gpu_offset >> 8); } p->idx += 7; } else { @@ -2564,10 +2564,10 @@ int r600_dma_cs_parse(struct radeon_cs_parser *p) dst_offset = radeon_get_ib_value(p, idx+1); dst_offset |= ((u64)(radeon_get_ib_value(p, idx+3) & 0xff)) << 32; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+2] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+3] += upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff; - ib[idx+4] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+2] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+3] += upper_32_bits(dst_reloc->gpu_offset) & 0xff; + ib[idx+4] += upper_32_bits(src_reloc->gpu_offset) & 0xff; p->idx += 5; } else { src_offset = radeon_get_ib_value(p, idx+2); @@ -2575,10 +2575,10 @@ int r600_dma_cs_parse(struct radeon_cs_parser *p) dst_offset = radeon_get_ib_value(p, idx+1); dst_offset |= ((u64)(radeon_get_ib_value(p, idx+3) & 0xff0000)) << 16; - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+2] += (u32)(src_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+3] += upper_32_bits(src_reloc->lobj.gpu_offset) & 0xff; - ib[idx+3] += (upper_32_bits(dst_reloc->lobj.gpu_offset) & 0xff) << 16; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+2] += (u32)(src_reloc->gpu_offset & 0xfffffffc); + ib[idx+3] += upper_32_bits(src_reloc->gpu_offset) & 0xff; + ib[idx+3] += (upper_32_bits(dst_reloc->gpu_offset) & 0xff) << 16; p->idx += 4; } } @@ -2610,8 +2610,8 @@ int r600_dma_cs_parse(struct radeon_cs_parser *p) dst_offset + (count * 4), radeon_bo_size(dst_reloc->robj)); return -EINVAL; } - ib[idx+1] += (u32)(dst_reloc->lobj.gpu_offset & 0xfffffffc); - ib[idx+3] += (upper_32_bits(dst_reloc->lobj.gpu_offset) << 16) & 0x00ff0000; + ib[idx+1] += (u32)(dst_reloc->gpu_offset & 0xfffffffc); + ib[idx+3] += (upper_32_bits(dst_reloc->gpu_offset) << 16) & 0x00ff0000; p->idx += 4; break; case DMA_PACKET_NOP: diff --git a/drivers/gpu/drm/radeon/r600_dma.c b/drivers/gpu/drm/radeon/r600_dma.c index b2d4c91e6272..53fcb28f5578 100644 --- a/drivers/gpu/drm/radeon/r600_dma.c +++ b/drivers/gpu/drm/radeon/r600_dma.c @@ -176,8 +176,6 @@ int r600_dma_resume(struct radeon_device *rdev) ring->wptr = 0; WREG32(DMA_RB_WPTR, ring->wptr << 2); - ring->rptr = RREG32(DMA_RB_RPTR) >> 2; - WREG32(DMA_RB_CNTL, rb_cntl | DMA_RB_ENABLE); ring->ready = true; @@ -221,11 +219,9 @@ bool r600_dma_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) u32 reset_mask = r600_gpu_check_soft_reset(rdev); if (!(reset_mask & RADEON_RESET_DMA)) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force ring activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } diff --git a/drivers/gpu/drm/radeon/r600_dpm.c b/drivers/gpu/drm/radeon/r600_dpm.c index e4cc9b314ce9..cbf7e3269f84 100644 --- a/drivers/gpu/drm/radeon/r600_dpm.c +++ b/drivers/gpu/drm/radeon/r600_dpm.c @@ -834,6 +834,26 @@ static int r600_parse_clk_voltage_dep_table(struct radeon_clock_voltage_dependen return 0; } +int r600_get_platform_caps(struct radeon_device *rdev) +{ + struct radeon_mode_info *mode_info = &rdev->mode_info; + union power_info *power_info; + int index = GetIndexIntoMasterTable(DATA, PowerPlayInfo); + u16 data_offset; + u8 frev, crev; + + if (!atom_parse_data_header(mode_info->atom_context, index, NULL, + &frev, &crev, &data_offset)) + return -EINVAL; + power_info = (union power_info *)(mode_info->atom_context->bios + data_offset); + + rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); + rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); + rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); + + return 0; +} + /* sizeof(ATOM_PPLIB_EXTENDEDHEADER) */ #define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V2 12 #define SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3 14 @@ -1043,7 +1063,15 @@ int r600_parse_extended_power_table(struct radeon_device *rdev) (mode_info->atom_context->bios + data_offset + le16_to_cpu(ext_hdr->usVCETableOffset) + 1 + 1 + array->ucNumEntries * sizeof(VCEClockInfo)); + ATOM_PPLIB_VCE_State_Table *states = + (ATOM_PPLIB_VCE_State_Table *) + (mode_info->atom_context->bios + data_offset + + le16_to_cpu(ext_hdr->usVCETableOffset) + 1 + + 1 + (array->ucNumEntries * sizeof (VCEClockInfo)) + + 1 + (limits->numEntries * sizeof(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record))); ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record *entry; + ATOM_PPLIB_VCE_State_Record *state_entry; + VCEClockInfo *vce_clk; u32 size = limits->numEntries * sizeof(struct radeon_vce_clock_voltage_dependency_entry); rdev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries = @@ -1055,8 +1083,9 @@ int r600_parse_extended_power_table(struct radeon_device *rdev) rdev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.count = limits->numEntries; entry = &limits->entries[0]; + state_entry = &states->entries[0]; for (i = 0; i < limits->numEntries; i++) { - VCEClockInfo *vce_clk = (VCEClockInfo *) + vce_clk = (VCEClockInfo *) ((u8 *)&array->entries[0] + (entry->ucVCEClockInfoIndex * sizeof(VCEClockInfo))); rdev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table.entries[i].evclk = @@ -1068,6 +1097,23 @@ int r600_parse_extended_power_table(struct radeon_device *rdev) entry = (ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record *) ((u8 *)entry + sizeof(ATOM_PPLIB_VCE_Clock_Voltage_Limit_Record)); } + for (i = 0; i < states->numEntries; i++) { + if (i >= RADEON_MAX_VCE_LEVELS) + break; + vce_clk = (VCEClockInfo *) + ((u8 *)&array->entries[0] + + (state_entry->ucVCEClockInfoIndex * sizeof(VCEClockInfo))); + rdev->pm.dpm.vce_states[i].evclk = + le16_to_cpu(vce_clk->usEVClkLow) | (vce_clk->ucEVClkHigh << 16); + rdev->pm.dpm.vce_states[i].ecclk = + le16_to_cpu(vce_clk->usECClkLow) | (vce_clk->ucECClkHigh << 16); + rdev->pm.dpm.vce_states[i].clk_idx = + state_entry->ucClockInfoIndex & 0x3f; + rdev->pm.dpm.vce_states[i].pstate = + (state_entry->ucClockInfoIndex & 0xc0) >> 6; + state_entry = (ATOM_PPLIB_VCE_State_Record *) + ((u8 *)state_entry + sizeof(ATOM_PPLIB_VCE_State_Record)); + } } if ((le16_to_cpu(ext_hdr->usSize) >= SIZE_OF_ATOM_PPLIB_EXTENDEDHEADER_V3) && ext_hdr->usUVDTableOffset) { diff --git a/drivers/gpu/drm/radeon/r600_dpm.h b/drivers/gpu/drm/radeon/r600_dpm.h index 07eab2b04e81..46b9d2a03018 100644 --- a/drivers/gpu/drm/radeon/r600_dpm.h +++ b/drivers/gpu/drm/radeon/r600_dpm.h @@ -215,6 +215,8 @@ void r600_stop_dpm(struct radeon_device *rdev); bool r600_is_internal_thermal_sensor(enum radeon_int_thermal_type sensor); +int r600_get_platform_caps(struct radeon_device *rdev); + int r600_parse_extended_power_table(struct radeon_device *rdev); void r600_free_extended_power_table(struct radeon_device *rdev); diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index e887d027b6d0..f21db7a0b34d 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -113,19 +113,16 @@ extern int radeon_hard_reset; #define RADEONFB_CONN_LIMIT 4 #define RADEON_BIOS_NUM_SCRATCH 8 -/* max number of rings */ -#define RADEON_NUM_RINGS 6 - /* fence seq are set to this number when signaled */ #define RADEON_FENCE_SIGNALED_SEQ 0LL /* internal ring indices */ /* r1xx+ has gfx CP ring */ -#define RADEON_RING_TYPE_GFX_INDEX 0 +#define RADEON_RING_TYPE_GFX_INDEX 0 /* cayman has 2 compute CP rings */ -#define CAYMAN_RING_TYPE_CP1_INDEX 1 -#define CAYMAN_RING_TYPE_CP2_INDEX 2 +#define CAYMAN_RING_TYPE_CP1_INDEX 1 +#define CAYMAN_RING_TYPE_CP2_INDEX 2 /* R600+ has an async dma ring */ #define R600_RING_TYPE_DMA_INDEX 3 @@ -133,7 +130,17 @@ extern int radeon_hard_reset; #define CAYMAN_RING_TYPE_DMA1_INDEX 4 /* R600+ */ -#define R600_RING_TYPE_UVD_INDEX 5 +#define R600_RING_TYPE_UVD_INDEX 5 + +/* TN+ */ +#define TN_RING_TYPE_VCE1_INDEX 6 +#define TN_RING_TYPE_VCE2_INDEX 7 + +/* max number of rings */ +#define RADEON_NUM_RINGS 8 + +/* 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 @@ -356,9 +363,8 @@ int radeon_fence_emit(struct radeon_device *rdev, struct radeon_fence **fence, i void radeon_fence_process(struct radeon_device *rdev, int ring); bool radeon_fence_signaled(struct radeon_fence *fence); int radeon_fence_wait(struct radeon_fence *fence, bool interruptible); -int radeon_fence_wait_locked(struct radeon_fence *fence); -int radeon_fence_wait_next_locked(struct radeon_device *rdev, int ring); -int radeon_fence_wait_empty_locked(struct radeon_device *rdev, int ring); +int radeon_fence_wait_next(struct radeon_device *rdev, int ring); +int radeon_fence_wait_empty(struct radeon_device *rdev, int ring); int radeon_fence_wait_any(struct radeon_device *rdev, struct radeon_fence **fences, bool intr); @@ -450,6 +456,7 @@ struct radeon_bo { /* Protected by gem.mutex */ struct list_head list; /* Protected by tbo.reserved */ + u32 initial_domain; u32 placements[3]; struct ttm_placement placement; struct ttm_buffer_object tbo; @@ -472,16 +479,6 @@ struct radeon_bo { }; #define gem_to_radeon_bo(gobj) container_of((gobj), struct radeon_bo, gem_base) -struct radeon_bo_list { - struct ttm_validate_buffer tv; - struct radeon_bo *bo; - uint64_t gpu_offset; - bool written; - unsigned domain; - unsigned alt_domain; - u32 tiling_flags; -}; - int radeon_gem_debugfs_init(struct radeon_device *rdev); /* sub-allocation manager, it has to be protected by another lock. @@ -789,7 +786,6 @@ struct radeon_ib { struct radeon_ring { struct radeon_bo *ring_obj; volatile uint32_t *ring; - unsigned rptr; unsigned rptr_offs; unsigned rptr_save_reg; u64 next_rptr_gpu_addr; @@ -799,8 +795,8 @@ struct radeon_ring { unsigned ring_size; unsigned ring_free_dw; int count_dw; - unsigned long last_activity; - unsigned last_rptr; + atomic_t last_rptr; + atomic64_t last_activity; uint64_t gpu_addr; uint32_t align_mask; uint32_t ptr_mask; @@ -852,17 +848,22 @@ struct radeon_mec { #define R600_PTE_READABLE (1 << 5) #define R600_PTE_WRITEABLE (1 << 6) +struct radeon_vm_pt { + struct radeon_bo *bo; + uint64_t addr; +}; + struct radeon_vm { - struct list_head list; struct list_head va; unsigned id; /* contains the page directory */ - struct radeon_sa_bo *page_directory; + struct radeon_bo *page_directory; uint64_t pd_gpu_addr; + unsigned max_pde_used; /* array of page tables, one for each page directory entry */ - struct radeon_sa_bo **page_tables; + struct radeon_vm_pt *page_tables; struct mutex mutex; /* last fence for cs using this vm */ @@ -874,10 +875,7 @@ struct radeon_vm { }; struct radeon_vm_manager { - struct mutex lock; - struct list_head lru_vm; struct radeon_fence *active[RADEON_NUM_VM]; - struct radeon_sa_manager sa_manager; uint32_t max_pfn; /* number of VMIDs */ unsigned nvm; @@ -953,8 +951,8 @@ void radeon_ring_unlock_commit(struct radeon_device *rdev, struct radeon_ring *c void radeon_ring_undo(struct radeon_ring *ring); void radeon_ring_unlock_undo(struct radeon_device *rdev, struct radeon_ring *cp); int radeon_ring_test(struct radeon_device *rdev, struct radeon_ring *cp); -void radeon_ring_force_activity(struct radeon_device *rdev, struct radeon_ring *ring); -void radeon_ring_lockup_update(struct radeon_ring *ring); +void radeon_ring_lockup_update(struct radeon_device *rdev, + struct radeon_ring *ring); bool radeon_ring_test_lockup(struct radeon_device *rdev, struct radeon_ring *ring); unsigned radeon_ring_backup(struct radeon_device *rdev, struct radeon_ring *ring, uint32_t **data); @@ -980,9 +978,12 @@ void cayman_dma_fini(struct radeon_device *rdev); struct radeon_cs_reloc { struct drm_gem_object *gobj; struct radeon_bo *robj; - struct radeon_bo_list lobj; + struct ttm_validate_buffer tv; + uint64_t gpu_offset; + unsigned domain; + unsigned alt_domain; + uint32_t tiling_flags; uint32_t handle; - uint32_t flags; }; struct radeon_cs_chunk { @@ -1006,6 +1007,7 @@ struct radeon_cs_parser { unsigned nrelocs; struct radeon_cs_reloc *relocs; struct radeon_cs_reloc **relocs_ptr; + struct radeon_cs_reloc *vm_bos; struct list_head validated; unsigned dma_reloc_idx; /* indices of various chunks */ @@ -1255,6 +1257,17 @@ enum radeon_dpm_event_src { RADEON_DPM_EVENT_SRC_DIGIAL_OR_EXTERNAL = 4 }; +#define RADEON_MAX_VCE_LEVELS 6 + +enum radeon_vce_level { + RADEON_VCE_LEVEL_AC_ALL = 0, /* AC, All cases */ + RADEON_VCE_LEVEL_DC_EE = 1, /* DC, entropy encoding */ + RADEON_VCE_LEVEL_DC_LL_LOW = 2, /* DC, low latency queue, res <= 720 */ + RADEON_VCE_LEVEL_DC_LL_HIGH = 3, /* DC, low latency queue, 1080 >= res > 720 */ + RADEON_VCE_LEVEL_DC_GP_LOW = 4, /* DC, general purpose queue, res <= 720 */ + RADEON_VCE_LEVEL_DC_GP_HIGH = 5, /* DC, general purpose queue, 1080 >= res > 720 */ +}; + struct radeon_ps { u32 caps; /* vbios flags */ u32 class; /* vbios flags */ @@ -1265,6 +1278,8 @@ struct radeon_ps { /* VCE clocks */ u32 evclk; u32 ecclk; + bool vce_active; + enum radeon_vce_level vce_level; /* asic priv */ void *ps_priv; }; @@ -1439,6 +1454,17 @@ enum radeon_dpm_forced_level { RADEON_DPM_FORCED_LEVEL_HIGH = 2, }; +struct radeon_vce_state { + /* vce clocks */ + u32 evclk; + u32 ecclk; + /* gpu clocks */ + u32 sclk; + u32 mclk; + u8 clk_idx; + u8 pstate; +}; + struct radeon_dpm { struct radeon_ps *ps; /* number of valid power states */ @@ -1451,6 +1477,9 @@ struct radeon_dpm { struct radeon_ps *boot_ps; /* default uvd power state */ struct radeon_ps *uvd_ps; + /* vce requirements */ + struct radeon_vce_state vce_states[RADEON_MAX_VCE_LEVELS]; + enum radeon_vce_level vce_level; enum radeon_pm_state_type state; enum radeon_pm_state_type user_state; u32 platform_caps; @@ -1476,6 +1505,7 @@ struct radeon_dpm { /* special states active */ bool thermal_active; bool uvd_active; + bool vce_active; /* thermal handling */ struct radeon_dpm_thermal thermal; /* forced levels */ @@ -1486,6 +1516,7 @@ struct radeon_dpm { }; void radeon_dpm_enable_uvd(struct radeon_device *rdev, bool enable); +void radeon_dpm_enable_vce(struct radeon_device *rdev, bool enable); struct radeon_pm { struct mutex mutex; @@ -1591,6 +1622,45 @@ int radeon_uvd_calc_upll_dividers(struct radeon_device *rdev, int radeon_uvd_send_upll_ctlreq(struct radeon_device *rdev, unsigned cg_upll_func_cntl); +/* + * VCE + */ +#define RADEON_MAX_VCE_HANDLES 16 +#define RADEON_VCE_STACK_SIZE (1024*1024) +#define RADEON_VCE_HEAP_SIZE (4*1024*1024) + +struct radeon_vce { + struct radeon_bo *vcpu_bo; + uint64_t gpu_addr; + unsigned fw_version; + unsigned fb_version; + atomic_t handles[RADEON_MAX_VCE_HANDLES]; + struct drm_file *filp[RADEON_MAX_VCE_HANDLES]; + struct delayed_work idle_work; +}; + +int radeon_vce_init(struct radeon_device *rdev); +void radeon_vce_fini(struct radeon_device *rdev); +int radeon_vce_suspend(struct radeon_device *rdev); +int radeon_vce_resume(struct radeon_device *rdev); +int radeon_vce_get_create_msg(struct radeon_device *rdev, int ring, + uint32_t handle, struct radeon_fence **fence); +int radeon_vce_get_destroy_msg(struct radeon_device *rdev, int ring, + uint32_t handle, struct radeon_fence **fence); +void radeon_vce_free_handles(struct radeon_device *rdev, struct drm_file *filp); +void radeon_vce_note_usage(struct radeon_device *rdev); +int radeon_vce_cs_reloc(struct radeon_cs_parser *p, int lo, int hi); +int radeon_vce_cs_parse(struct radeon_cs_parser *p); +bool radeon_vce_semaphore_emit(struct radeon_device *rdev, + struct radeon_ring *ring, + struct radeon_semaphore *semaphore, + bool emit_wait); +void radeon_vce_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib); +void radeon_vce_fence_emit(struct radeon_device *rdev, + struct radeon_fence *fence); +int radeon_vce_ring_test(struct radeon_device *rdev, struct radeon_ring *ring); +int radeon_vce_ib_test(struct radeon_device *rdev, struct radeon_ring *ring); + struct r600_audio_pin { int channels; int rate; @@ -1780,6 +1850,7 @@ struct radeon_asic { void (*set_pcie_lanes)(struct radeon_device *rdev, int lanes); void (*set_clock_gating)(struct radeon_device *rdev, int enable); int (*set_uvd_clocks)(struct radeon_device *rdev, u32 vclk, u32 dclk); + int (*set_vce_clocks)(struct radeon_device *rdev, u32 evclk, u32 ecclk); int (*get_temperature)(struct radeon_device *rdev); } pm; /* dynamic power management */ @@ -2041,6 +2112,8 @@ int radeon_gem_wait_idle_ioctl(struct drm_device *dev, void *data, struct drm_file *filp); int radeon_gem_va_ioctl(struct drm_device *dev, void *data, struct drm_file *filp); +int radeon_gem_op_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp); int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp); int radeon_gem_set_tiling_ioctl(struct drm_device *dev, void *data, struct drm_file *filp); @@ -2186,6 +2259,7 @@ struct radeon_device { struct radeon_gem gem; struct radeon_pm pm; struct radeon_uvd uvd; + struct radeon_vce vce; uint32_t bios_scratch[RADEON_BIOS_NUM_SCRATCH]; struct radeon_wb wb; struct radeon_dummy_page dummy_page; @@ -2205,6 +2279,7 @@ struct radeon_device { const struct firmware *sdma_fw; /* CIK SDMA firmware */ const struct firmware *smc_fw; /* SMC firmware */ const struct firmware *uvd_fw; /* UVD firmware */ + const struct firmware *vce_fw; /* VCE firmware */ struct r600_vram_scratch vram_scratch; int msi_enabled; /* msi enabled */ struct r600_ih ih; /* r6/700 interrupt ring */ @@ -2229,6 +2304,10 @@ struct radeon_device { /* virtual memory */ struct radeon_vm_manager vm_manager; struct mutex gpu_clock_mutex; + /* memory stats */ + atomic64_t vram_usage; + atomic64_t gtt_usage; + atomic64_t num_bytes_moved; /* ACPI interface */ struct radeon_atif atif; struct radeon_atcs atcs; @@ -2639,6 +2718,7 @@ void radeon_ring_write(struct radeon_ring *ring, uint32_t v); #define radeon_set_pcie_lanes(rdev, l) (rdev)->asic->pm.set_pcie_lanes((rdev), (l)) #define radeon_set_clock_gating(rdev, e) (rdev)->asic->pm.set_clock_gating((rdev), (e)) #define radeon_set_uvd_clocks(rdev, v, d) (rdev)->asic->pm.set_uvd_clocks((rdev), (v), (d)) +#define radeon_set_vce_clocks(rdev, ev, ec) (rdev)->asic->pm.set_vce_clocks((rdev), (ev), (ec)) #define radeon_get_temperature(rdev) (rdev)->asic->pm.get_temperature((rdev)) #define radeon_set_surface_reg(rdev, r, f, p, o, s) ((rdev)->asic->surface.set_reg((rdev), (r), (f), (p), (o), (s))) #define radeon_clear_surface_reg(rdev, r) ((rdev)->asic->surface.clear_reg((rdev), (r))) @@ -2715,16 +2795,22 @@ extern void radeon_program_register_sequence(struct radeon_device *rdev, */ int radeon_vm_manager_init(struct radeon_device *rdev); void radeon_vm_manager_fini(struct radeon_device *rdev); -void radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm); +int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm); void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm); -int radeon_vm_alloc_pt(struct radeon_device *rdev, struct radeon_vm *vm); -void radeon_vm_add_to_lru(struct radeon_device *rdev, struct radeon_vm *vm); +struct radeon_cs_reloc *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); void radeon_vm_fence(struct radeon_device *rdev, struct radeon_vm *vm, struct radeon_fence *fence); uint64_t radeon_vm_map_gart(struct radeon_device *rdev, uint64_t addr); +int radeon_vm_update_page_directory(struct radeon_device *rdev, + struct radeon_vm *vm); int radeon_vm_bo_update(struct radeon_device *rdev, struct radeon_vm *vm, struct radeon_bo *bo, diff --git a/drivers/gpu/drm/radeon/radeon_asic.c b/drivers/gpu/drm/radeon/radeon_asic.c index dda02bfc10a4..b8a24a75d4ff 100644 --- a/drivers/gpu/drm/radeon/radeon_asic.c +++ b/drivers/gpu/drm/radeon/radeon_asic.c @@ -1987,6 +1987,19 @@ static struct radeon_asic_ring ci_dma_ring = { .set_wptr = &cik_sdma_set_wptr, }; +static struct radeon_asic_ring ci_vce_ring = { + .ib_execute = &radeon_vce_ib_execute, + .emit_fence = &radeon_vce_fence_emit, + .emit_semaphore = &radeon_vce_semaphore_emit, + .cs_parse = &radeon_vce_cs_parse, + .ring_test = &radeon_vce_ring_test, + .ib_test = &radeon_vce_ib_test, + .is_lockup = &radeon_ring_test_lockup, + .get_rptr = &vce_v1_0_get_rptr, + .get_wptr = &vce_v1_0_get_wptr, + .set_wptr = &vce_v1_0_set_wptr, +}; + static struct radeon_asic ci_asic = { .init = &cik_init, .fini = &cik_fini, @@ -2015,6 +2028,8 @@ static struct radeon_asic ci_asic = { [R600_RING_TYPE_DMA_INDEX] = &ci_dma_ring, [CAYMAN_RING_TYPE_DMA1_INDEX] = &ci_dma_ring, [R600_RING_TYPE_UVD_INDEX] = &cayman_uvd_ring, + [TN_RING_TYPE_VCE1_INDEX] = &ci_vce_ring, + [TN_RING_TYPE_VCE2_INDEX] = &ci_vce_ring, }, .irq = { .set = &cik_irq_set, @@ -2061,6 +2076,7 @@ static struct radeon_asic ci_asic = { .set_pcie_lanes = NULL, .set_clock_gating = NULL, .set_uvd_clocks = &cik_set_uvd_clocks, + .set_vce_clocks = &cik_set_vce_clocks, .get_temperature = &ci_get_temp, }, .dpm = { @@ -2117,6 +2133,8 @@ static struct radeon_asic kv_asic = { [R600_RING_TYPE_DMA_INDEX] = &ci_dma_ring, [CAYMAN_RING_TYPE_DMA1_INDEX] = &ci_dma_ring, [R600_RING_TYPE_UVD_INDEX] = &cayman_uvd_ring, + [TN_RING_TYPE_VCE1_INDEX] = &ci_vce_ring, + [TN_RING_TYPE_VCE2_INDEX] = &ci_vce_ring, }, .irq = { .set = &cik_irq_set, @@ -2163,6 +2181,7 @@ static struct radeon_asic kv_asic = { .set_pcie_lanes = NULL, .set_clock_gating = NULL, .set_uvd_clocks = &cik_set_uvd_clocks, + .set_vce_clocks = &cik_set_vce_clocks, .get_temperature = &kv_get_temp, }, .dpm = { diff --git a/drivers/gpu/drm/radeon/radeon_asic.h b/drivers/gpu/drm/radeon/radeon_asic.h index ae637cfda783..3d55a3a39e82 100644 --- a/drivers/gpu/drm/radeon/radeon_asic.h +++ b/drivers/gpu/drm/radeon/radeon_asic.h @@ -717,6 +717,7 @@ u32 cik_get_xclk(struct radeon_device *rdev); uint32_t cik_pciep_rreg(struct radeon_device *rdev, uint32_t reg); void cik_pciep_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v); int cik_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk); +int cik_set_vce_clocks(struct radeon_device *rdev, u32 evclk, u32 ecclk); void cik_sdma_fence_ring_emit(struct radeon_device *rdev, struct radeon_fence *fence); bool cik_sdma_semaphore_ring_emit(struct radeon_device *rdev, @@ -863,4 +864,17 @@ bool uvd_v3_1_semaphore_emit(struct radeon_device *rdev, /* uvd v4.2 */ int uvd_v4_2_resume(struct radeon_device *rdev); +/* vce v1.0 */ +uint32_t vce_v1_0_get_rptr(struct radeon_device *rdev, + struct radeon_ring *ring); +uint32_t vce_v1_0_get_wptr(struct radeon_device *rdev, + struct radeon_ring *ring); +void vce_v1_0_set_wptr(struct radeon_device *rdev, + struct radeon_ring *ring); +int vce_v1_0_init(struct radeon_device *rdev); +int vce_v1_0_start(struct radeon_device *rdev); + +/* vce v2.0 */ +int vce_v2_0_resume(struct radeon_device *rdev); + #endif diff --git a/drivers/gpu/drm/radeon/radeon_connectors.c b/drivers/gpu/drm/radeon/radeon_connectors.c index 82d4f865546e..c566b486ca08 100644 --- a/drivers/gpu/drm/radeon/radeon_connectors.c +++ b/drivers/gpu/drm/radeon/radeon_connectors.c @@ -89,7 +89,7 @@ static void radeon_property_change_mode(struct drm_encoder *encoder) if (crtc && crtc->enabled) { drm_crtc_helper_set_mode(crtc, &crtc->mode, - crtc->x, crtc->y, crtc->fb); + crtc->x, crtc->y, crtc->primary->fb); } } @@ -1595,6 +1595,7 @@ radeon_add_atom_connector(struct drm_device *dev, uint32_t subpixel_order = SubPixelNone; bool shared_ddc = false; bool is_dp_bridge = false; + bool has_aux = false; if (connector_type == DRM_MODE_CONNECTOR_Unknown) return; @@ -1672,7 +1673,9 @@ radeon_add_atom_connector(struct drm_device *dev, radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "eDP-auxch"); else radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "DP-auxch"); - if (!radeon_dig_connector->dp_i2c_bus) + if (radeon_dig_connector->dp_i2c_bus) + has_aux = true; + else DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); if (!radeon_connector->ddc_bus) @@ -1895,7 +1898,9 @@ radeon_add_atom_connector(struct drm_device *dev, if (!radeon_dig_connector->dp_i2c_bus) DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); - if (!radeon_connector->ddc_bus) + if (radeon_connector->ddc_bus) + has_aux = true; + else DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); } subpixel_order = SubPixelHorizontalRGB; @@ -1939,7 +1944,9 @@ radeon_add_atom_connector(struct drm_device *dev, if (i2c_bus->valid) { /* add DP i2c bus */ radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "eDP-auxch"); - if (!radeon_dig_connector->dp_i2c_bus) + if (radeon_dig_connector->dp_i2c_bus) + has_aux = true; + else DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); if (!radeon_connector->ddc_bus) @@ -2000,6 +2007,10 @@ radeon_add_atom_connector(struct drm_device *dev, connector->display_info.subpixel_order = subpixel_order; drm_sysfs_connector_add(connector); + + if (has_aux) + radeon_dp_aux_init(radeon_connector); + return; failed: diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c index dfb5a1db87d4..2b6e0ebcc13a 100644 --- a/drivers/gpu/drm/radeon/radeon_cs.c +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -24,16 +24,59 @@ * Authors: * Jerome Glisse <glisse@freedesktop.org> */ +#include <linux/list_sort.h> #include <drm/drmP.h> #include <drm/radeon_drm.h> #include "radeon_reg.h" #include "radeon.h" #include "radeon_trace.h" +#define RADEON_CS_MAX_PRIORITY 32u +#define RADEON_CS_NUM_BUCKETS (RADEON_CS_MAX_PRIORITY + 1) + +/* This is based on the bucket sort with O(n) time complexity. + * An item with priority "i" is added to bucket[i]. The lists are then + * concatenated in descending order. + */ +struct radeon_cs_buckets { + struct list_head bucket[RADEON_CS_NUM_BUCKETS]; +}; + +static void radeon_cs_buckets_init(struct radeon_cs_buckets *b) +{ + unsigned i; + + for (i = 0; i < RADEON_CS_NUM_BUCKETS; i++) + INIT_LIST_HEAD(&b->bucket[i]); +} + +static void radeon_cs_buckets_add(struct radeon_cs_buckets *b, + struct list_head *item, unsigned priority) +{ + /* Since buffers which appear sooner in the relocation list are + * likely to be used more often than buffers which appear later + * in the list, the sort mustn't change the ordering of buffers + * with the same priority, i.e. it must be stable. + */ + list_add_tail(item, &b->bucket[min(priority, RADEON_CS_MAX_PRIORITY)]); +} + +static void radeon_cs_buckets_get_list(struct radeon_cs_buckets *b, + struct list_head *out_list) +{ + unsigned i; + + /* Connect the sorted buckets in the output list. */ + for (i = 0; i < RADEON_CS_NUM_BUCKETS; i++) { + list_splice(&b->bucket[i], out_list); + } +} + 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; @@ -52,8 +95,12 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p) if (p->relocs == NULL) { return -ENOMEM; } + + radeon_cs_buckets_init(&buckets); + for (i = 0; i < p->nrelocs; i++) { struct drm_radeon_cs_reloc *r; + unsigned priority; duplicate = false; r = (struct drm_radeon_cs_reloc *)&chunk->kdata[i*4]; @@ -78,8 +125,14 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p) } p->relocs_ptr[i] = &p->relocs[i]; p->relocs[i].robj = gem_to_radeon_bo(p->relocs[i].gobj); - p->relocs[i].lobj.bo = p->relocs[i].robj; - p->relocs[i].lobj.written = !!r->write_domain; + + /* The userspace buffer priorities are from 0 to 15. A higher + * number means the buffer is more important. + * Also, the buffers used for write have a higher priority than + * the buffers used for read only, which doubles the range + * to 0 to 31. 32 is reserved for the kernel driver. + */ + priority = (r->flags & 0xf) * 2 + !!r->write_domain; /* the first reloc of an UVD job is the msg and that must be in VRAM, also but everything into VRAM on AGP cards to avoid @@ -87,29 +140,38 @@ static int radeon_cs_parser_relocs(struct radeon_cs_parser *p) if (p->ring == R600_RING_TYPE_UVD_INDEX && (i == 0 || drm_pci_device_is_agp(p->rdev->ddev))) { /* TODO: is this still needed for NI+ ? */ - p->relocs[i].lobj.domain = + p->relocs[i].domain = RADEON_GEM_DOMAIN_VRAM; - p->relocs[i].lobj.alt_domain = + p->relocs[i].alt_domain = RADEON_GEM_DOMAIN_VRAM; + /* prioritize this over any other relocation */ + priority = RADEON_CS_MAX_PRIORITY; } else { uint32_t domain = r->write_domain ? r->write_domain : r->read_domains; - p->relocs[i].lobj.domain = domain; + p->relocs[i].domain = domain; if (domain == RADEON_GEM_DOMAIN_VRAM) domain |= RADEON_GEM_DOMAIN_GTT; - p->relocs[i].lobj.alt_domain = domain; + p->relocs[i].alt_domain = domain; } - p->relocs[i].lobj.tv.bo = &p->relocs[i].robj->tbo; + p->relocs[i].tv.bo = &p->relocs[i].robj->tbo; p->relocs[i].handle = r->handle; - radeon_bo_list_add_object(&p->relocs[i].lobj, - &p->validated); + radeon_cs_buckets_add(&buckets, &p->relocs[i].tv.head, + priority); } - return radeon_bo_list_validate(&p->ticket, &p->validated, p->ring); + + radeon_cs_buckets_get_list(&buckets, &p->validated); + + if (p->cs_flags & RADEON_CS_USE_VM) + p->vm_bos = radeon_vm_get_bos(p->rdev, p->ib.vm, + &p->validated); + + return radeon_bo_list_validate(p->rdev, &p->ticket, &p->validated, p->ring); } static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority) @@ -147,6 +209,10 @@ static int radeon_cs_get_ring(struct radeon_cs_parser *p, u32 ring, s32 priority case RADEON_CS_RING_UVD: p->ring = R600_RING_TYPE_UVD_INDEX; break; + case RADEON_CS_RING_VCE: + /* TODO: only use the low priority ring for now */ + p->ring = TN_RING_TYPE_VCE1_INDEX; + break; } return 0; } @@ -286,6 +352,16 @@ int radeon_cs_parser_init(struct radeon_cs_parser *p, void *data) return 0; } +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); + + /* Sort A before B if A is smaller. */ + return (int)la->robj->tbo.num_pages - (int)lb->robj->tbo.num_pages; +} + /** * cs_parser_fini() - clean parser states * @parser: parser structure holding parsing context. @@ -299,6 +375,18 @@ static void radeon_cs_parser_fini(struct radeon_cs_parser *parser, int error, bo unsigned i; if (!error) { + /* Sort the buffer list from the smallest to largest buffer, + * which affects the order of buffers in the LRU list. + * This assures that the smallest buffers are added first + * to the LRU list, so they are likely to be later evicted + * first, instead of large buffers whose eviction is more + * expensive. + * + * This slightly lowers the number of bytes moved by TTM + * per frame under memory pressure. + */ + list_sort(NULL, &parser->validated, cmp_size_smaller_first); + ttm_eu_fence_buffer_objects(&parser->ticket, &parser->validated, parser->ib.fence); @@ -316,6 +404,7 @@ static void radeon_cs_parser_fini(struct radeon_cs_parser *parser, int error, bo kfree(parser->track); kfree(parser->relocs); kfree(parser->relocs_ptr); + kfree(parser->vm_bos); for (i = 0; i < parser->nchunks; i++) drm_free_large(parser->chunks[i].kdata); kfree(parser->chunks); @@ -343,6 +432,9 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev, if (parser->ring == R600_RING_TYPE_UVD_INDEX) radeon_uvd_note_usage(rdev); + else if ((parser->ring == TN_RING_TYPE_VCE1_INDEX) || + (parser->ring == TN_RING_TYPE_VCE2_INDEX)) + radeon_vce_note_usage(rdev); radeon_cs_sync_rings(parser); r = radeon_ib_schedule(rdev, &parser->ib, NULL); @@ -352,24 +444,32 @@ static int radeon_cs_ib_chunk(struct radeon_device *rdev, return r; } -static int radeon_bo_vm_update_pte(struct radeon_cs_parser *parser, +static int radeon_bo_vm_update_pte(struct radeon_cs_parser *p, struct radeon_vm *vm) { - struct radeon_device *rdev = parser->rdev; - struct radeon_bo_list *lobj; - struct radeon_bo *bo; - int r; + struct radeon_device *rdev = p->rdev; + int i, r; - r = radeon_vm_bo_update(rdev, vm, rdev->ring_tmp_bo.bo, &rdev->ring_tmp_bo.bo->tbo.mem); - if (r) { + r = radeon_vm_update_page_directory(rdev, vm); + if (r) return r; - } - list_for_each_entry(lobj, &parser->validated, tv.head) { - bo = lobj->bo; - r = radeon_vm_bo_update(parser->rdev, vm, bo, &bo->tbo.mem); - if (r) { + + r = radeon_vm_bo_update(rdev, vm, rdev->ring_tmp_bo.bo, + &rdev->ring_tmp_bo.bo->tbo.mem); + if (r) + return r; + + 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; + r = radeon_vm_bo_update(rdev, vm, bo, &bo->tbo.mem); + if (r) return r; - } } return 0; } @@ -401,20 +501,13 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev, if (parser->ring == R600_RING_TYPE_UVD_INDEX) radeon_uvd_note_usage(rdev); - mutex_lock(&rdev->vm_manager.lock); mutex_lock(&vm->mutex); - r = radeon_vm_alloc_pt(rdev, vm); - if (r) { - goto out; - } r = radeon_bo_vm_update_pte(parser, vm); if (r) { goto out; } radeon_cs_sync_rings(parser); radeon_semaphore_sync_to(parser->ib.semaphore, vm->fence); - radeon_semaphore_sync_to(parser->ib.semaphore, - radeon_vm_grab_id(rdev, vm, parser->ring)); if ((rdev->family >= CHIP_TAHITI) && (parser->chunk_const_ib_idx != -1)) { @@ -423,14 +516,8 @@ static int radeon_cs_ib_vm_chunk(struct radeon_device *rdev, r = radeon_ib_schedule(rdev, &parser->ib, NULL); } - if (!r) { - radeon_vm_fence(rdev, vm, parser->ib.fence); - } - out: - radeon_vm_add_to_lru(rdev, vm); mutex_unlock(&vm->mutex); - mutex_unlock(&rdev->vm_manager.lock); return r; } @@ -698,9 +785,9 @@ int radeon_cs_packet_next_reloc(struct radeon_cs_parser *p, /* FIXME: we assume reloc size is 4 dwords */ if (nomm) { *cs_reloc = p->relocs; - (*cs_reloc)->lobj.gpu_offset = + (*cs_reloc)->gpu_offset = (u64)relocs_chunk->kdata[idx + 3] << 32; - (*cs_reloc)->lobj.gpu_offset |= relocs_chunk->kdata[idx + 0]; + (*cs_reloc)->gpu_offset |= relocs_chunk->kdata[idx + 0]; } else *cs_reloc = p->relocs_ptr[(idx / 4)]; return 0; diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c index 044bc98fb459..835516d2d257 100644 --- a/drivers/gpu/drm/radeon/radeon_device.c +++ b/drivers/gpu/drm/radeon/radeon_device.c @@ -1191,14 +1191,12 @@ int radeon_device_init(struct radeon_device *rdev, r = radeon_gem_init(rdev); if (r) return r; - /* initialize vm here */ - mutex_init(&rdev->vm_manager.lock); + /* Adjust VM size here. * Currently set to 4GB ((1 << 20) 4k pages). * Max GPUVM size for cayman and SI is 40 bits. */ rdev->vm_manager.max_pfn = 1 << 20; - INIT_LIST_HEAD(&rdev->vm_manager.lru_vm); /* Set asic functions */ r = radeon_asic_init(rdev); @@ -1426,7 +1424,7 @@ int radeon_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon) /* unpin the front buffers */ list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { - struct radeon_framebuffer *rfb = to_radeon_framebuffer(crtc->fb); + struct radeon_framebuffer *rfb = to_radeon_framebuffer(crtc->primary->fb); struct radeon_bo *robj; if (rfb == NULL || rfb->obj == NULL) { @@ -1445,10 +1443,9 @@ int radeon_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon) /* evict vram memory */ radeon_bo_evict_vram(rdev); - mutex_lock(&rdev->ring_lock); /* wait for gpu to finish processing current batch */ for (i = 0; i < RADEON_NUM_RINGS; i++) { - r = radeon_fence_wait_empty_locked(rdev, i); + r = radeon_fence_wait_empty(rdev, i); if (r) { /* delay GPU reset to resume */ force_completion = true; @@ -1457,7 +1454,6 @@ int radeon_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon) if (force_completion) { radeon_fence_driver_force_completion(rdev); } - mutex_unlock(&rdev->ring_lock); radeon_save_bios_scratch_regs(rdev); @@ -1555,10 +1551,12 @@ int radeon_resume_kms(struct drm_device *dev, bool resume, bool fbcon) /* reset hpd state */ radeon_hpd_init(rdev); /* blat the mode back in */ - drm_helper_resume_force_mode(dev); - /* turn on display hw */ - list_for_each_entry(connector, &dev->mode_config.connector_list, head) { - drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON); + if (fbcon) { + drm_helper_resume_force_mode(dev); + /* turn on display hw */ + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON); + } } drm_kms_helper_poll_enable(dev); diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c index fbd8b930f2be..386cfa4c194d 100644 --- a/drivers/gpu/drm/radeon/radeon_display.c +++ b/drivers/gpu/drm/radeon/radeon_display.c @@ -34,6 +34,8 @@ #include <drm/drm_crtc_helper.h> #include <drm/drm_edid.h> +#include <linux/gcd.h> + static void avivo_crtc_load_lut(struct drm_crtc *crtc) { struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); @@ -369,7 +371,7 @@ static int radeon_crtc_page_flip(struct drm_crtc *crtc, work->event = event; work->rdev = rdev; work->crtc_id = radeon_crtc->crtc_id; - old_radeon_fb = to_radeon_framebuffer(crtc->fb); + old_radeon_fb = to_radeon_framebuffer(crtc->primary->fb); new_radeon_fb = to_radeon_framebuffer(fb); /* schedule unpin of the old buffer */ obj = old_radeon_fb->obj; @@ -460,7 +462,7 @@ static int radeon_crtc_page_flip(struct drm_crtc *crtc, spin_unlock_irqrestore(&dev->event_lock, flags); /* update crtc fb */ - crtc->fb = fb; + crtc->primary->fb = fb; r = drm_vblank_get(dev, radeon_crtc->crtc_id); if (r) { @@ -792,6 +794,7 @@ int radeon_ddc_get_modes(struct radeon_connector *radeon_connector) if (radeon_connector->edid) { drm_mode_connector_update_edid_property(&radeon_connector->base, radeon_connector->edid); ret = drm_add_edid_modes(&radeon_connector->base, radeon_connector->edid); + drm_edid_to_eld(&radeon_connector->base, radeon_connector->edid); return ret; } drm_mode_connector_update_edid_property(&radeon_connector->base, NULL); @@ -799,66 +802,57 @@ int radeon_ddc_get_modes(struct radeon_connector *radeon_connector) } /* avivo */ -static void avivo_get_fb_div(struct radeon_pll *pll, - u32 target_clock, - u32 post_div, - u32 ref_div, - u32 *fb_div, - u32 *frac_fb_div) -{ - u32 tmp = post_div * ref_div; - tmp *= target_clock; - *fb_div = tmp / pll->reference_freq; - *frac_fb_div = tmp % pll->reference_freq; - - if (*fb_div > pll->max_feedback_div) - *fb_div = pll->max_feedback_div; - else if (*fb_div < pll->min_feedback_div) - *fb_div = pll->min_feedback_div; -} - -static u32 avivo_get_post_div(struct radeon_pll *pll, - u32 target_clock) +/** + * avivo_reduce_ratio - fractional number reduction + * + * @nom: nominator + * @den: denominator + * @nom_min: minimum value for nominator + * @den_min: minimum value for denominator + * + * Find the greatest common divisor and apply it on both nominator and + * denominator, but make nominator and denominator are at least as large + * as their minimum values. + */ +static void avivo_reduce_ratio(unsigned *nom, unsigned *den, + unsigned nom_min, unsigned den_min) { - u32 vco, post_div, tmp; - - if (pll->flags & RADEON_PLL_USE_POST_DIV) - return pll->post_div; - - if (pll->flags & RADEON_PLL_PREFER_MINM_OVER_MAXP) { - if (pll->flags & RADEON_PLL_IS_LCD) - vco = pll->lcd_pll_out_min; - else - vco = pll->pll_out_min; - } else { - if (pll->flags & RADEON_PLL_IS_LCD) - vco = pll->lcd_pll_out_max; - else - vco = pll->pll_out_max; + unsigned tmp; + + /* reduce the numbers to a simpler ratio */ + tmp = gcd(*nom, *den); + *nom /= tmp; + *den /= tmp; + + /* make sure nominator is large enough */ + if (*nom < nom_min) { + tmp = (nom_min + *nom - 1) / *nom; + *nom *= tmp; + *den *= tmp; } - post_div = vco / target_clock; - tmp = vco % target_clock; - - if (pll->flags & RADEON_PLL_PREFER_MINM_OVER_MAXP) { - if (tmp) - post_div++; - } else { - if (!tmp) - post_div--; + /* make sure the denominator is large enough */ + if (*den < den_min) { + tmp = (den_min + *den - 1) / *den; + *nom *= tmp; + *den *= tmp; } - - if (post_div > pll->max_post_div) - post_div = pll->max_post_div; - else if (post_div < pll->min_post_div) - post_div = pll->min_post_div; - - return post_div; } -#define MAX_TOLERANCE 10 - +/** + * radeon_compute_pll_avivo - compute PLL paramaters + * + * @pll: information about the PLL + * @dot_clock_p: resulting pixel clock + * fb_div_p: resulting feedback divider + * frac_fb_div_p: fractional part of the feedback divider + * ref_div_p: resulting reference divider + * post_div_p: resulting reference divider + * + * Try to calculate the PLL parameters to generate the given frequency: + * dot_clock = (ref_freq * feedback_div) / (ref_div * post_div) + */ void radeon_compute_pll_avivo(struct radeon_pll *pll, u32 freq, u32 *dot_clock_p, @@ -867,53 +861,123 @@ void radeon_compute_pll_avivo(struct radeon_pll *pll, u32 *ref_div_p, u32 *post_div_p) { - u32 target_clock = freq / 10; - u32 post_div = avivo_get_post_div(pll, target_clock); - u32 ref_div = pll->min_ref_div; - u32 fb_div = 0, frac_fb_div = 0, tmp; + unsigned fb_div_min, fb_div_max, fb_div; + unsigned post_div_min, post_div_max, post_div; + unsigned ref_div_min, ref_div_max, ref_div; + unsigned post_div_best, diff_best; + unsigned nom, den, tmp; - if (pll->flags & RADEON_PLL_USE_REF_DIV) - ref_div = pll->reference_div; + /* determine allowed feedback divider range */ + fb_div_min = pll->min_feedback_div; + fb_div_max = pll->max_feedback_div; if (pll->flags & RADEON_PLL_USE_FRAC_FB_DIV) { - avivo_get_fb_div(pll, target_clock, post_div, ref_div, &fb_div, &frac_fb_div); - frac_fb_div = (100 * frac_fb_div) / pll->reference_freq; - if (frac_fb_div >= 5) { - frac_fb_div -= 5; - frac_fb_div = frac_fb_div / 10; - frac_fb_div++; + fb_div_min *= 10; + fb_div_max *= 10; + } + + /* determine allowed ref divider range */ + if (pll->flags & RADEON_PLL_USE_REF_DIV) + ref_div_min = pll->reference_div; + else + ref_div_min = pll->min_ref_div; + ref_div_max = pll->max_ref_div; + + /* determine allowed post divider range */ + if (pll->flags & RADEON_PLL_USE_POST_DIV) { + post_div_min = pll->post_div; + post_div_max = pll->post_div; + } else { + unsigned target_clock = freq / 10; + unsigned vco_min, vco_max; + + if (pll->flags & RADEON_PLL_IS_LCD) { + vco_min = pll->lcd_pll_out_min; + vco_max = pll->lcd_pll_out_max; + } else { + vco_min = pll->pll_out_min; + vco_max = pll->pll_out_max; } - if (frac_fb_div >= 10) { - fb_div++; - frac_fb_div = 0; + + post_div_min = vco_min / target_clock; + if ((target_clock * post_div_min) < vco_min) + ++post_div_min; + if (post_div_min < pll->min_post_div) + post_div_min = pll->min_post_div; + + post_div_max = vco_max / target_clock; + if ((target_clock * post_div_max) > vco_max) + --post_div_max; + if (post_div_max > pll->max_post_div) + post_div_max = pll->max_post_div; + } + + /* represent the searched ratio as fractional number */ + nom = pll->flags & RADEON_PLL_USE_FRAC_FB_DIV ? freq : freq / 10; + den = pll->reference_freq; + + /* reduce the numbers to a simpler ratio */ + avivo_reduce_ratio(&nom, &den, fb_div_min, post_div_min); + + /* now search for a post divider */ + if (pll->flags & RADEON_PLL_PREFER_MINM_OVER_MAXP) + post_div_best = post_div_min; + else + post_div_best = post_div_max; + diff_best = ~0; + + for (post_div = post_div_min; post_div <= post_div_max; ++post_div) { + unsigned diff = abs(den - den / post_div * post_div); + if (diff < diff_best || (diff == diff_best && + !(pll->flags & RADEON_PLL_PREFER_MINM_OVER_MAXP))) { + + post_div_best = post_div; + diff_best = diff; } + } + post_div = post_div_best; + + /* get matching reference and feedback divider */ + ref_div = max(den / post_div, 1u); + fb_div = nom; + + /* we're almost done, but reference and feedback + divider might be to large now */ + + tmp = ref_div; + + if (fb_div > fb_div_max) { + ref_div = ref_div * fb_div_max / fb_div; + fb_div = fb_div_max; + } + + if (ref_div > ref_div_max) { + ref_div = ref_div_max; + fb_div = nom * ref_div_max / tmp; + } + + /* reduce the numbers to a simpler ratio once more */ + /* this also makes sure that the reference divider is large enough */ + avivo_reduce_ratio(&fb_div, &ref_div, fb_div_min, ref_div_min); + + /* and finally save the result */ + if (pll->flags & RADEON_PLL_USE_FRAC_FB_DIV) { + *fb_div_p = fb_div / 10; + *frac_fb_div_p = fb_div % 10; } else { - while (ref_div <= pll->max_ref_div) { - avivo_get_fb_div(pll, target_clock, post_div, ref_div, - &fb_div, &frac_fb_div); - if (frac_fb_div >= (pll->reference_freq / 2)) - fb_div++; - frac_fb_div = 0; - tmp = (pll->reference_freq * fb_div) / (post_div * ref_div); - tmp = (tmp * 10000) / target_clock; - - if (tmp > (10000 + MAX_TOLERANCE)) - ref_div++; - else if (tmp >= (10000 - MAX_TOLERANCE)) - break; - else - ref_div++; - } + *fb_div_p = fb_div; + *frac_fb_div_p = 0; } - *dot_clock_p = ((pll->reference_freq * fb_div * 10) + (pll->reference_freq * frac_fb_div)) / - (ref_div * post_div * 10); - *fb_div_p = fb_div; - *frac_fb_div_p = frac_fb_div; + *dot_clock_p = ((pll->reference_freq * *fb_div_p * 10) + + (pll->reference_freq * *frac_fb_div_p)) / + (ref_div * post_div * 10); *ref_div_p = ref_div; *post_div_p = post_div; - DRM_DEBUG_KMS("%d, pll dividers - fb: %d.%d ref: %d, post %d\n", - *dot_clock_p, fb_div, frac_fb_div, ref_div, post_div); + + DRM_DEBUG_KMS("%d - %d, pll dividers - fb: %d.%d ref: %d, post %d\n", + freq, *dot_clock_p, *fb_div_p, *frac_fb_div_p, + ref_div, post_div); } /* pre-avivo */ diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index f633c2782170..d0eba48dd74e 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -79,9 +79,11 @@ * 2.35.0 - Add CIK macrotile mode array query * 2.36.0 - Fix CIK DCE tiling setup * 2.37.0 - allow GS ring setup on r6xx/r7xx + * 2.38.0 - RADEON_GEM_OP (GET_INITIAL_DOMAIN, SET_INITIAL_DOMAIN), + * CIK: 1D and linear tiling modes contain valid PIPE_CONFIG */ #define KMS_DRIVER_MAJOR 2 -#define KMS_DRIVER_MINOR 37 +#define KMS_DRIVER_MINOR 38 #define KMS_DRIVER_PATCHLEVEL 0 int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags); int radeon_driver_unload_kms(struct drm_device *dev); diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c index c37cb79a9489..a77b1c13ea43 100644 --- a/drivers/gpu/drm/radeon/radeon_fence.c +++ b/drivers/gpu/drm/radeon/radeon_fence.c @@ -288,7 +288,6 @@ static bool radeon_fence_any_seq_signaled(struct radeon_device *rdev, u64 *seq) * @rdev: radeon device pointer * @target_seq: sequence number(s) we want to wait for * @intr: use interruptable sleep - * @lock_ring: whether the ring should be locked or not * * Wait for the requested sequence number(s) to be written by any ring * (all asics). Sequnce number array is indexed by ring id. @@ -299,7 +298,7 @@ static bool radeon_fence_any_seq_signaled(struct radeon_device *rdev, u64 *seq) * -EDEADLK is returned when a GPU lockup has been detected. */ static int radeon_fence_wait_seq(struct radeon_device *rdev, u64 *target_seq, - bool intr, bool lock_ring) + bool intr) { uint64_t last_seq[RADEON_NUM_RINGS]; bool signaled; @@ -358,9 +357,6 @@ static int radeon_fence_wait_seq(struct radeon_device *rdev, u64 *target_seq, if (i != RADEON_NUM_RINGS) continue; - if (lock_ring) - mutex_lock(&rdev->ring_lock); - for (i = 0; i < RADEON_NUM_RINGS; ++i) { if (!target_seq[i]) continue; @@ -378,14 +374,9 @@ static int radeon_fence_wait_seq(struct radeon_device *rdev, u64 *target_seq, /* remember that we need an reset */ rdev->needs_reset = true; - if (lock_ring) - mutex_unlock(&rdev->ring_lock); wake_up_all(&rdev->fence_queue); return -EDEADLK; } - - if (lock_ring) - mutex_unlock(&rdev->ring_lock); } } return 0; @@ -416,7 +407,7 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr) if (seq[fence->ring] == RADEON_FENCE_SIGNALED_SEQ) return 0; - r = radeon_fence_wait_seq(fence->rdev, seq, intr, true); + r = radeon_fence_wait_seq(fence->rdev, seq, intr); if (r) return r; @@ -464,7 +455,7 @@ int radeon_fence_wait_any(struct radeon_device *rdev, if (num_rings == 0) return -ENOENT; - r = radeon_fence_wait_seq(rdev, seq, intr, true); + r = radeon_fence_wait_seq(rdev, seq, intr); if (r) { return r; } @@ -472,37 +463,7 @@ int radeon_fence_wait_any(struct radeon_device *rdev, } /** - * radeon_fence_wait_locked - wait for a fence to signal - * - * @fence: radeon fence object - * - * Wait for the requested fence to signal (all asics). - * Returns 0 if the fence has passed, error for all other cases. - */ -int radeon_fence_wait_locked(struct radeon_fence *fence) -{ - uint64_t seq[RADEON_NUM_RINGS] = {}; - int r; - - if (fence == NULL) { - WARN(1, "Querying an invalid fence : %p !\n", fence); - return -EINVAL; - } - - seq[fence->ring] = fence->seq; - if (seq[fence->ring] == RADEON_FENCE_SIGNALED_SEQ) - return 0; - - r = radeon_fence_wait_seq(fence->rdev, seq, false, false); - if (r) - return r; - - fence->seq = RADEON_FENCE_SIGNALED_SEQ; - return 0; -} - -/** - * radeon_fence_wait_next_locked - wait for the next fence to signal + * radeon_fence_wait_next - wait for the next fence to signal * * @rdev: radeon device pointer * @ring: ring index the fence is associated with @@ -511,7 +472,7 @@ int radeon_fence_wait_locked(struct radeon_fence *fence) * Returns 0 if the next fence has passed, error for all other cases. * Caller must hold ring lock. */ -int radeon_fence_wait_next_locked(struct radeon_device *rdev, int ring) +int radeon_fence_wait_next(struct radeon_device *rdev, int ring) { uint64_t seq[RADEON_NUM_RINGS] = {}; @@ -521,11 +482,11 @@ int radeon_fence_wait_next_locked(struct radeon_device *rdev, int ring) already the last emited fence */ return -ENOENT; } - return radeon_fence_wait_seq(rdev, seq, false, false); + return radeon_fence_wait_seq(rdev, seq, false); } /** - * radeon_fence_wait_empty_locked - wait for all fences to signal + * radeon_fence_wait_empty - wait for all fences to signal * * @rdev: radeon device pointer * @ring: ring index the fence is associated with @@ -534,7 +495,7 @@ int radeon_fence_wait_next_locked(struct radeon_device *rdev, int ring) * Returns 0 if the fences have passed, error for all other cases. * Caller must hold ring lock. */ -int radeon_fence_wait_empty_locked(struct radeon_device *rdev, int ring) +int radeon_fence_wait_empty(struct radeon_device *rdev, int ring) { uint64_t seq[RADEON_NUM_RINGS] = {}; int r; @@ -543,7 +504,7 @@ int radeon_fence_wait_empty_locked(struct radeon_device *rdev, int ring) if (!seq[ring]) return 0; - r = radeon_fence_wait_seq(rdev, seq, false, false); + r = radeon_fence_wait_seq(rdev, seq, false); if (r) { if (r == -EDEADLK) return -EDEADLK; @@ -794,7 +755,7 @@ void radeon_fence_driver_fini(struct radeon_device *rdev) for (ring = 0; ring < RADEON_NUM_RINGS; ring++) { if (!rdev->fence_drv[ring].initialized) continue; - r = radeon_fence_wait_empty_locked(rdev, ring); + r = radeon_fence_wait_empty(rdev, ring); if (r) { /* no need to trigger GPU reset as we are unloading */ radeon_fence_driver_force_completion(rdev); diff --git a/drivers/gpu/drm/radeon/radeon_gart.c b/drivers/gpu/drm/radeon/radeon_gart.c index a8f9b463bf2a..2e723651069b 100644 --- a/drivers/gpu/drm/radeon/radeon_gart.c +++ b/drivers/gpu/drm/radeon/radeon_gart.c @@ -28,8 +28,6 @@ #include <drm/drmP.h> #include <drm/radeon_drm.h> #include "radeon.h" -#include "radeon_reg.h" -#include "radeon_trace.h" /* * GART @@ -394,959 +392,3 @@ void radeon_gart_fini(struct radeon_device *rdev) radeon_dummy_page_fini(rdev); } - -/* - * GPUVM - * GPUVM is similar to the legacy gart on older asics, however - * rather than there being a single global gart table - * for the entire GPU, there are multiple VM page tables active - * at any given time. The VM page tables can contain a mix - * vram pages and system memory pages and system memory pages - * can be mapped as snooped (cached system pages) or unsnooped - * (uncached system pages). - * Each VM has an ID associated with it and there is a page table - * associated with each VMID. When execting a command buffer, - * the kernel tells the the ring what VMID to use for that command - * buffer. VMIDs are allocated dynamically as commands are submitted. - * The userspace drivers maintain their own address space and the kernel - * sets up their pages tables accordingly when they submit their - * command buffers and a VMID is assigned. - * Cayman/Trinity support up to 8 active VMs at any given time; - * SI supports 16. - */ - -/* - * vm helpers - * - * TODO bind a default page at vm initialization for default address - */ - -/** - * radeon_vm_num_pde - return the number of page directory entries - * - * @rdev: radeon_device pointer - * - * Calculate the number of page directory entries (cayman+). - */ -static unsigned radeon_vm_num_pdes(struct radeon_device *rdev) -{ - return rdev->vm_manager.max_pfn >> RADEON_VM_BLOCK_SIZE; -} - -/** - * radeon_vm_directory_size - returns the size of the page directory in bytes - * - * @rdev: radeon_device pointer - * - * Calculate the size of the page directory in bytes (cayman+). - */ -static unsigned radeon_vm_directory_size(struct radeon_device *rdev) -{ - return RADEON_GPU_PAGE_ALIGN(radeon_vm_num_pdes(rdev) * 8); -} - -/** - * radeon_vm_manager_init - init the vm manager - * - * @rdev: radeon_device pointer - * - * Init the vm manager (cayman+). - * Returns 0 for success, error for failure. - */ -int radeon_vm_manager_init(struct radeon_device *rdev) -{ - struct radeon_vm *vm; - struct radeon_bo_va *bo_va; - int r; - unsigned size; - - if (!rdev->vm_manager.enabled) { - /* allocate enough for 2 full VM pts */ - size = radeon_vm_directory_size(rdev); - size += rdev->vm_manager.max_pfn * 8; - size *= 2; - r = radeon_sa_bo_manager_init(rdev, &rdev->vm_manager.sa_manager, - RADEON_GPU_PAGE_ALIGN(size), - RADEON_VM_PTB_ALIGN_SIZE, - RADEON_GEM_DOMAIN_VRAM); - if (r) { - dev_err(rdev->dev, "failed to allocate vm bo (%dKB)\n", - (rdev->vm_manager.max_pfn * 8) >> 10); - return r; - } - - r = radeon_asic_vm_init(rdev); - if (r) - return r; - - rdev->vm_manager.enabled = true; - - r = radeon_sa_bo_manager_start(rdev, &rdev->vm_manager.sa_manager); - if (r) - return r; - } - - /* restore page table */ - list_for_each_entry(vm, &rdev->vm_manager.lru_vm, list) { - if (vm->page_directory == NULL) - continue; - - list_for_each_entry(bo_va, &vm->va, vm_list) { - bo_va->valid = false; - } - } - return 0; -} - -/** - * radeon_vm_free_pt - free the page table for a specific vm - * - * @rdev: radeon_device pointer - * @vm: vm to unbind - * - * Free the page table of a specific vm (cayman+). - * - * Global and local mutex must be lock! - */ -static void radeon_vm_free_pt(struct radeon_device *rdev, - struct radeon_vm *vm) -{ - struct radeon_bo_va *bo_va; - int i; - - if (!vm->page_directory) - return; - - list_del_init(&vm->list); - radeon_sa_bo_free(rdev, &vm->page_directory, vm->fence); - - list_for_each_entry(bo_va, &vm->va, vm_list) { - bo_va->valid = false; - } - - if (vm->page_tables == NULL) - return; - - for (i = 0; i < radeon_vm_num_pdes(rdev); i++) - radeon_sa_bo_free(rdev, &vm->page_tables[i], vm->fence); - - kfree(vm->page_tables); -} - -/** - * radeon_vm_manager_fini - tear down the vm manager - * - * @rdev: radeon_device pointer - * - * Tear down the VM manager (cayman+). - */ -void radeon_vm_manager_fini(struct radeon_device *rdev) -{ - struct radeon_vm *vm, *tmp; - int i; - - if (!rdev->vm_manager.enabled) - return; - - mutex_lock(&rdev->vm_manager.lock); - /* free all allocated page tables */ - list_for_each_entry_safe(vm, tmp, &rdev->vm_manager.lru_vm, list) { - mutex_lock(&vm->mutex); - radeon_vm_free_pt(rdev, vm); - mutex_unlock(&vm->mutex); - } - for (i = 0; i < RADEON_NUM_VM; ++i) { - radeon_fence_unref(&rdev->vm_manager.active[i]); - } - radeon_asic_vm_fini(rdev); - mutex_unlock(&rdev->vm_manager.lock); - - radeon_sa_bo_manager_suspend(rdev, &rdev->vm_manager.sa_manager); - radeon_sa_bo_manager_fini(rdev, &rdev->vm_manager.sa_manager); - rdev->vm_manager.enabled = false; -} - -/** - * radeon_vm_evict - evict page table to make room for new one - * - * @rdev: radeon_device pointer - * @vm: VM we want to allocate something for - * - * Evict a VM from the lru, making sure that it isn't @vm. (cayman+). - * Returns 0 for success, -ENOMEM for failure. - * - * Global and local mutex must be locked! - */ -static int radeon_vm_evict(struct radeon_device *rdev, struct radeon_vm *vm) -{ - struct radeon_vm *vm_evict; - - if (list_empty(&rdev->vm_manager.lru_vm)) - return -ENOMEM; - - vm_evict = list_first_entry(&rdev->vm_manager.lru_vm, - struct radeon_vm, list); - if (vm_evict == vm) - return -ENOMEM; - - mutex_lock(&vm_evict->mutex); - radeon_vm_free_pt(rdev, vm_evict); - mutex_unlock(&vm_evict->mutex); - return 0; -} - -/** - * radeon_vm_alloc_pt - allocates a page table for a VM - * - * @rdev: radeon_device pointer - * @vm: vm to bind - * - * Allocate a page table for the requested vm (cayman+). - * Returns 0 for success, error for failure. - * - * Global and local mutex must be locked! - */ -int radeon_vm_alloc_pt(struct radeon_device *rdev, struct radeon_vm *vm) -{ - unsigned pd_size, pd_entries, pts_size; - struct radeon_ib ib; - int r; - - if (vm == NULL) { - return -EINVAL; - } - - if (vm->page_directory != NULL) { - return 0; - } - - pd_size = radeon_vm_directory_size(rdev); - pd_entries = radeon_vm_num_pdes(rdev); - -retry: - r = radeon_sa_bo_new(rdev, &rdev->vm_manager.sa_manager, - &vm->page_directory, pd_size, - RADEON_VM_PTB_ALIGN_SIZE, false); - if (r == -ENOMEM) { - r = radeon_vm_evict(rdev, vm); - if (r) - return r; - goto retry; - - } else if (r) { - return r; - } - - vm->pd_gpu_addr = radeon_sa_bo_gpu_addr(vm->page_directory); - - /* Initially clear the page directory */ - r = radeon_ib_get(rdev, R600_RING_TYPE_DMA_INDEX, &ib, - NULL, pd_entries * 2 + 64); - if (r) { - radeon_sa_bo_free(rdev, &vm->page_directory, vm->fence); - return r; - } - - ib.length_dw = 0; - - radeon_asic_vm_set_page(rdev, &ib, vm->pd_gpu_addr, - 0, pd_entries, 0, 0); - - radeon_semaphore_sync_to(ib.semaphore, vm->fence); - r = radeon_ib_schedule(rdev, &ib, NULL); - if (r) { - radeon_ib_free(rdev, &ib); - radeon_sa_bo_free(rdev, &vm->page_directory, vm->fence); - return r; - } - radeon_fence_unref(&vm->fence); - vm->fence = radeon_fence_ref(ib.fence); - radeon_ib_free(rdev, &ib); - radeon_fence_unref(&vm->last_flush); - - /* allocate page table array */ - pts_size = radeon_vm_num_pdes(rdev) * sizeof(struct radeon_sa_bo *); - vm->page_tables = kzalloc(pts_size, GFP_KERNEL); - - if (vm->page_tables == NULL) { - DRM_ERROR("Cannot allocate memory for page table array\n"); - radeon_sa_bo_free(rdev, &vm->page_directory, vm->fence); - return -ENOMEM; - } - - return 0; -} - -/** - * radeon_vm_add_to_lru - add VMs page table to LRU list - * - * @rdev: radeon_device pointer - * @vm: vm to add to LRU - * - * Add the allocated page table to the LRU list (cayman+). - * - * Global mutex must be locked! - */ -void radeon_vm_add_to_lru(struct radeon_device *rdev, struct radeon_vm *vm) -{ - list_del_init(&vm->list); - list_add_tail(&vm->list, &rdev->vm_manager.lru_vm); -} - -/** - * radeon_vm_grab_id - allocate the next free VMID - * - * @rdev: radeon_device pointer - * @vm: vm to allocate id for - * @ring: ring we want to submit job to - * - * Allocate an id for the vm (cayman+). - * Returns the fence we need to sync to (if any). - * - * Global and local mutex must be locked! - */ -struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, - struct radeon_vm *vm, int ring) -{ - struct radeon_fence *best[RADEON_NUM_RINGS] = {}; - 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]) - return NULL; - - /* we definately need to flush */ - radeon_fence_unref(&vm->last_flush); - - /* skip over VMID 0, since it is the system VM */ - for (i = 1; i < rdev->vm_manager.nvm; ++i) { - struct radeon_fence *fence = rdev->vm_manager.active[i]; - - if (fence == NULL) { - /* found a free one */ - vm->id = i; - trace_radeon_vm_grab_id(vm->id, ring); - return NULL; - } - - if (radeon_fence_is_earlier(fence, best[fence->ring])) { - best[fence->ring] = fence; - choices[fence->ring == ring ? 0 : 1] = i; - } - } - - for (i = 0; i < 2; ++i) { - if (choices[i]) { - vm->id = choices[i]; - trace_radeon_vm_grab_id(vm->id, ring); - return rdev->vm_manager.active[choices[i]]; - } - } - - /* should never happen */ - BUG(); - return NULL; -} - -/** - * radeon_vm_fence - remember fence for vm - * - * @rdev: radeon_device pointer - * @vm: vm we want to fence - * @fence: fence to remember - * - * Fence the vm (cayman+). - * Set the fence used to protect page table and id. - * - * Global and local mutex must be locked! - */ -void radeon_vm_fence(struct radeon_device *rdev, - struct radeon_vm *vm, - struct radeon_fence *fence) -{ - radeon_fence_unref(&rdev->vm_manager.active[vm->id]); - rdev->vm_manager.active[vm->id] = radeon_fence_ref(fence); - - radeon_fence_unref(&vm->fence); - vm->fence = radeon_fence_ref(fence); - - radeon_fence_unref(&vm->last_id_use); - vm->last_id_use = radeon_fence_ref(fence); -} - -/** - * radeon_vm_bo_find - find the bo_va for a specific vm & bo - * - * @vm: requested vm - * @bo: requested buffer object - * - * Find @bo inside the requested vm (cayman+). - * Search inside the @bos vm list for the requested vm - * Returns the found bo_va or NULL if none is found - * - * Object has to be reserved! - */ -struct radeon_bo_va *radeon_vm_bo_find(struct radeon_vm *vm, - struct radeon_bo *bo) -{ - struct radeon_bo_va *bo_va; - - list_for_each_entry(bo_va, &bo->va, bo_list) { - if (bo_va->vm == vm) { - return bo_va; - } - } - return NULL; -} - -/** - * radeon_vm_bo_add - add a bo to a specific vm - * - * @rdev: radeon_device pointer - * @vm: requested vm - * @bo: radeon buffer object - * - * Add @bo into the requested vm (cayman+). - * Add @bo to the list of bos associated with the vm - * Returns newly added bo_va or NULL for failure - * - * Object has to be reserved! - */ -struct radeon_bo_va *radeon_vm_bo_add(struct radeon_device *rdev, - struct radeon_vm *vm, - struct radeon_bo *bo) -{ - struct radeon_bo_va *bo_va; - - bo_va = kzalloc(sizeof(struct radeon_bo_va), GFP_KERNEL); - if (bo_va == NULL) { - return NULL; - } - bo_va->vm = vm; - bo_va->bo = bo; - bo_va->soffset = 0; - bo_va->eoffset = 0; - bo_va->flags = 0; - bo_va->valid = false; - bo_va->ref_count = 1; - INIT_LIST_HEAD(&bo_va->bo_list); - INIT_LIST_HEAD(&bo_va->vm_list); - - mutex_lock(&vm->mutex); - list_add(&bo_va->vm_list, &vm->va); - list_add_tail(&bo_va->bo_list, &bo->va); - mutex_unlock(&vm->mutex); - - return bo_va; -} - -/** - * radeon_vm_bo_set_addr - set bos virtual address inside a vm - * - * @rdev: radeon_device pointer - * @bo_va: bo_va to store the address - * @soffset: requested offset of the buffer in the VM address space - * @flags: attributes of pages (read/write/valid/etc.) - * - * Set offset of @bo_va (cayman+). - * Validate and set the offset requested within the vm address space. - * Returns 0 for success, error for failure. - * - * Object has to be reserved! - */ -int radeon_vm_bo_set_addr(struct radeon_device *rdev, - struct radeon_bo_va *bo_va, - uint64_t soffset, - uint32_t flags) -{ - uint64_t size = radeon_bo_size(bo_va->bo); - uint64_t eoffset, last_offset = 0; - struct radeon_vm *vm = bo_va->vm; - struct radeon_bo_va *tmp; - struct list_head *head; - unsigned last_pfn; - - if (soffset) { - /* make sure object fit at this offset */ - eoffset = soffset + size; - if (soffset >= eoffset) { - return -EINVAL; - } - - last_pfn = eoffset / RADEON_GPU_PAGE_SIZE; - if (last_pfn > rdev->vm_manager.max_pfn) { - dev_err(rdev->dev, "va above limit (0x%08X > 0x%08X)\n", - last_pfn, rdev->vm_manager.max_pfn); - return -EINVAL; - } - - } else { - eoffset = last_pfn = 0; - } - - mutex_lock(&vm->mutex); - head = &vm->va; - last_offset = 0; - list_for_each_entry(tmp, &vm->va, vm_list) { - if (bo_va == tmp) { - /* skip over currently modified bo */ - continue; - } - - if (soffset >= last_offset && eoffset <= tmp->soffset) { - /* bo can be added before this one */ - break; - } - if (eoffset > tmp->soffset && soffset < tmp->eoffset) { - /* bo and tmp overlap, invalid offset */ - dev_err(rdev->dev, "bo %p va 0x%08X conflict with (bo %p 0x%08X 0x%08X)\n", - bo_va->bo, (unsigned)bo_va->soffset, tmp->bo, - (unsigned)tmp->soffset, (unsigned)tmp->eoffset); - mutex_unlock(&vm->mutex); - return -EINVAL; - } - last_offset = tmp->eoffset; - head = &tmp->vm_list; - } - - bo_va->soffset = soffset; - bo_va->eoffset = eoffset; - bo_va->flags = flags; - bo_va->valid = false; - list_move(&bo_va->vm_list, head); - - mutex_unlock(&vm->mutex); - return 0; -} - -/** - * radeon_vm_map_gart - get the physical address of a gart page - * - * @rdev: radeon_device pointer - * @addr: the unmapped addr - * - * Look up the physical address of the page that the pte resolves - * to (cayman+). - * Returns the physical address of the page. - */ -uint64_t radeon_vm_map_gart(struct radeon_device *rdev, uint64_t addr) -{ - uint64_t result; - - /* page table offset */ - result = rdev->gart.pages_addr[addr >> PAGE_SHIFT]; - - /* in case cpu page size != gpu page size*/ - result |= addr & (~PAGE_MASK); - - return result; -} - -/** - * radeon_vm_page_flags - translate page flags to what the hw uses - * - * @flags: flags comming from userspace - * - * Translate the flags the userspace ABI uses to hw flags. - */ -static uint32_t radeon_vm_page_flags(uint32_t flags) -{ - uint32_t hw_flags = 0; - hw_flags |= (flags & RADEON_VM_PAGE_VALID) ? R600_PTE_VALID : 0; - hw_flags |= (flags & RADEON_VM_PAGE_READABLE) ? R600_PTE_READABLE : 0; - hw_flags |= (flags & RADEON_VM_PAGE_WRITEABLE) ? R600_PTE_WRITEABLE : 0; - if (flags & RADEON_VM_PAGE_SYSTEM) { - hw_flags |= R600_PTE_SYSTEM; - hw_flags |= (flags & RADEON_VM_PAGE_SNOOPED) ? R600_PTE_SNOOPED : 0; - } - return hw_flags; -} - -/** - * radeon_vm_update_pdes - make sure that page directory is valid - * - * @rdev: radeon_device pointer - * @vm: requested vm - * @start: start of GPU address range - * @end: end of GPU address range - * - * Allocates new page tables if necessary - * and updates the page directory (cayman+). - * Returns 0 for success, error for failure. - * - * Global and local mutex must be locked! - */ -static int radeon_vm_update_pdes(struct radeon_device *rdev, - struct radeon_vm *vm, - struct radeon_ib *ib, - uint64_t start, uint64_t end) -{ - static const uint32_t incr = RADEON_VM_PTE_COUNT * 8; - - uint64_t last_pde = ~0, last_pt = ~0; - unsigned count = 0; - uint64_t pt_idx; - int r; - - start = (start / RADEON_GPU_PAGE_SIZE) >> RADEON_VM_BLOCK_SIZE; - end = (end / RADEON_GPU_PAGE_SIZE) >> RADEON_VM_BLOCK_SIZE; - - /* walk over the address space and update the page directory */ - for (pt_idx = start; pt_idx <= end; ++pt_idx) { - uint64_t pde, pt; - - if (vm->page_tables[pt_idx]) - continue; - -retry: - r = radeon_sa_bo_new(rdev, &rdev->vm_manager.sa_manager, - &vm->page_tables[pt_idx], - RADEON_VM_PTE_COUNT * 8, - RADEON_GPU_PAGE_SIZE, false); - - if (r == -ENOMEM) { - r = radeon_vm_evict(rdev, vm); - if (r) - return r; - goto retry; - } else if (r) { - return r; - } - - pde = vm->pd_gpu_addr + pt_idx * 8; - - pt = radeon_sa_bo_gpu_addr(vm->page_tables[pt_idx]); - - if (((last_pde + 8 * count) != pde) || - ((last_pt + incr * count) != pt)) { - - if (count) { - radeon_asic_vm_set_page(rdev, ib, last_pde, - last_pt, count, incr, - R600_PTE_VALID); - - count *= RADEON_VM_PTE_COUNT; - radeon_asic_vm_set_page(rdev, ib, last_pt, 0, - count, 0, 0); - } - - count = 1; - last_pde = pde; - last_pt = pt; - } else { - ++count; - } - } - - if (count) { - radeon_asic_vm_set_page(rdev, ib, last_pde, last_pt, count, - incr, R600_PTE_VALID); - - count *= RADEON_VM_PTE_COUNT; - radeon_asic_vm_set_page(rdev, ib, last_pt, 0, - count, 0, 0); - } - - return 0; -} - -/** - * radeon_vm_update_ptes - make sure that page tables are valid - * - * @rdev: radeon_device pointer - * @vm: requested vm - * @start: start of GPU address range - * @end: end of GPU address range - * @dst: destination address to map to - * @flags: mapping flags - * - * Update the page tables in the range @start - @end (cayman+). - * - * 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 const uint64_t mask = RADEON_VM_PTE_COUNT - 1; - - uint64_t last_pte = ~0, last_dst = ~0; - unsigned count = 0; - uint64_t addr; - - start = start / RADEON_GPU_PAGE_SIZE; - end = end / RADEON_GPU_PAGE_SIZE; - - /* walk over the address space and update the page tables */ - for (addr = start; addr < end; ) { - uint64_t pt_idx = addr >> RADEON_VM_BLOCK_SIZE; - unsigned nptes; - uint64_t pte; - - if ((addr & ~mask) == (end & ~mask)) - nptes = end - addr; - else - nptes = RADEON_VM_PTE_COUNT - (addr & mask); - - pte = radeon_sa_bo_gpu_addr(vm->page_tables[pt_idx]); - pte += (addr & mask) * 8; - - if ((last_pte + 8 * count) != pte) { - - if (count) { - radeon_asic_vm_set_page(rdev, ib, last_pte, - last_dst, count, - RADEON_GPU_PAGE_SIZE, - flags); - } - - count = nptes; - last_pte = pte; - last_dst = dst; - } else { - count += nptes; - } - - addr += nptes; - dst += nptes * RADEON_GPU_PAGE_SIZE; - } - - if (count) { - radeon_asic_vm_set_page(rdev, ib, last_pte, - last_dst, count, - RADEON_GPU_PAGE_SIZE, flags); - } -} - -/** - * radeon_vm_bo_update - map a bo into the vm page table - * - * @rdev: radeon_device pointer - * @vm: requested vm - * @bo: radeon buffer object - * @mem: ttm mem - * - * Fill in the page table entries for @bo (cayman+). - * Returns 0 for success, -EINVAL for failure. - * - * Object have to be reserved & global and local mutex must be locked! - */ -int radeon_vm_bo_update(struct radeon_device *rdev, - struct radeon_vm *vm, - struct radeon_bo *bo, - struct ttm_mem_reg *mem) -{ - struct radeon_ib ib; - struct radeon_bo_va *bo_va; - unsigned nptes, npdes, ndw; - uint64_t addr; - int r; - - /* nothing to do if vm isn't bound */ - if (vm->page_directory == NULL) - return 0; - - bo_va = radeon_vm_bo_find(vm, bo); - if (bo_va == NULL) { - dev_err(rdev->dev, "bo %p not in vm %p\n", bo, vm); - return -EINVAL; - } - - if (!bo_va->soffset) { - dev_err(rdev->dev, "bo %p don't has a mapping in vm %p\n", - bo, vm); - return -EINVAL; - } - - if ((bo_va->valid && mem) || (!bo_va->valid && mem == NULL)) - return 0; - - bo_va->flags &= ~RADEON_VM_PAGE_VALID; - bo_va->flags &= ~RADEON_VM_PAGE_SYSTEM; - if (mem) { - addr = mem->start << PAGE_SHIFT; - if (mem->mem_type != TTM_PL_SYSTEM) { - bo_va->flags |= RADEON_VM_PAGE_VALID; - bo_va->valid = true; - } - if (mem->mem_type == TTM_PL_TT) { - bo_va->flags |= RADEON_VM_PAGE_SYSTEM; - } else { - addr += rdev->vm_manager.vram_base_offset; - } - } else { - addr = 0; - bo_va->valid = false; - } - - trace_radeon_vm_bo_update(bo_va); - - nptes = radeon_bo_ngpu_pages(bo); - - /* assume two extra pdes in case the mapping overlaps the borders */ - npdes = (nptes >> RADEON_VM_BLOCK_SIZE) + 2; - - /* padding, etc. */ - ndw = 64; - - if (RADEON_VM_BLOCK_SIZE > 11) - /* reserve space for one header for every 2k dwords */ - ndw += (nptes >> 11) * 4; - else - /* reserve space for one header for - every (1 << BLOCK_SIZE) entries */ - ndw += (nptes >> RADEON_VM_BLOCK_SIZE) * 4; - - /* reserve space for pte addresses */ - ndw += nptes * 2; - - /* reserve space for one header for every 2k dwords */ - ndw += (npdes >> 11) * 4; - - /* reserve space for pde addresses */ - ndw += npdes * 2; - - /* reserve space for clearing new page tables */ - ndw += npdes * 2 * RADEON_VM_PTE_COUNT; - - /* update too big for an IB */ - if (ndw > 0xfffff) - return -ENOMEM; - - r = radeon_ib_get(rdev, R600_RING_TYPE_DMA_INDEX, &ib, NULL, ndw * 4); - if (r) - return r; - ib.length_dw = 0; - - r = radeon_vm_update_pdes(rdev, vm, &ib, bo_va->soffset, bo_va->eoffset); - if (r) { - radeon_ib_free(rdev, &ib); - return r; - } - - radeon_vm_update_ptes(rdev, vm, &ib, bo_va->soffset, bo_va->eoffset, - addr, radeon_vm_page_flags(bo_va->flags)); - - radeon_semaphore_sync_to(ib.semaphore, vm->fence); - r = radeon_ib_schedule(rdev, &ib, NULL); - if (r) { - radeon_ib_free(rdev, &ib); - return r; - } - radeon_fence_unref(&vm->fence); - vm->fence = radeon_fence_ref(ib.fence); - radeon_ib_free(rdev, &ib); - radeon_fence_unref(&vm->last_flush); - - return 0; -} - -/** - * radeon_vm_bo_rmv - remove a bo to a specific vm - * - * @rdev: radeon_device pointer - * @bo_va: requested bo_va - * - * Remove @bo_va->bo from the requested vm (cayman+). - * Remove @bo_va->bo from the list of bos associated with the bo_va->vm and - * remove the ptes for @bo_va in the page table. - * Returns 0 for success. - * - * Object have to be reserved! - */ -int radeon_vm_bo_rmv(struct radeon_device *rdev, - struct radeon_bo_va *bo_va) -{ - int r = 0; - - mutex_lock(&rdev->vm_manager.lock); - mutex_lock(&bo_va->vm->mutex); - if (bo_va->soffset) { - r = radeon_vm_bo_update(rdev, bo_va->vm, bo_va->bo, NULL); - } - mutex_unlock(&rdev->vm_manager.lock); - list_del(&bo_va->vm_list); - mutex_unlock(&bo_va->vm->mutex); - list_del(&bo_va->bo_list); - - kfree(bo_va); - return r; -} - -/** - * radeon_vm_bo_invalidate - mark the bo as invalid - * - * @rdev: radeon_device pointer - * @vm: requested vm - * @bo: radeon buffer object - * - * Mark @bo as invalid (cayman+). - */ -void radeon_vm_bo_invalidate(struct radeon_device *rdev, - struct radeon_bo *bo) -{ - struct radeon_bo_va *bo_va; - - list_for_each_entry(bo_va, &bo->va, bo_list) { - bo_va->valid = false; - } -} - -/** - * radeon_vm_init - initialize a vm instance - * - * @rdev: radeon_device pointer - * @vm: requested vm - * - * Init @vm fields (cayman+). - */ -void radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) -{ - vm->id = 0; - vm->fence = NULL; - vm->last_flush = NULL; - vm->last_id_use = NULL; - mutex_init(&vm->mutex); - INIT_LIST_HEAD(&vm->list); - INIT_LIST_HEAD(&vm->va); -} - -/** - * radeon_vm_fini - tear down a vm instance - * - * @rdev: radeon_device pointer - * @vm: requested vm - * - * Tear down @vm (cayman+). - * Unbind the VM and remove all bos from the vm bo list - */ -void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm) -{ - struct radeon_bo_va *bo_va, *tmp; - int r; - - mutex_lock(&rdev->vm_manager.lock); - mutex_lock(&vm->mutex); - radeon_vm_free_pt(rdev, vm); - mutex_unlock(&rdev->vm_manager.lock); - - if (!list_empty(&vm->va)) { - dev_err(rdev->dev, "still active bo inside vm\n"); - } - list_for_each_entry_safe(bo_va, tmp, &vm->va, vm_list) { - list_del_init(&bo_va->vm_list); - r = radeon_bo_reserve(bo_va->bo, false); - if (!r) { - list_del_init(&bo_va->bo_list); - radeon_bo_unreserve(bo_va->bo); - kfree(bo_va); - } - } - radeon_fence_unref(&vm->fence); - radeon_fence_unref(&vm->last_flush); - radeon_fence_unref(&vm->last_id_use); - mutex_unlock(&vm->mutex); -} diff --git a/drivers/gpu/drm/radeon/radeon_gem.c b/drivers/gpu/drm/radeon/radeon_gem.c index b96c819024b3..d09650c1d720 100644 --- a/drivers/gpu/drm/radeon/radeon_gem.c +++ b/drivers/gpu/drm/radeon/radeon_gem.c @@ -344,18 +344,7 @@ int radeon_gem_busy_ioctl(struct drm_device *dev, void *data, } robj = gem_to_radeon_bo(gobj); r = radeon_bo_wait(robj, &cur_placement, true); - switch (cur_placement) { - case TTM_PL_VRAM: - args->domain = RADEON_GEM_DOMAIN_VRAM; - break; - case TTM_PL_TT: - args->domain = RADEON_GEM_DOMAIN_GTT; - break; - case TTM_PL_SYSTEM: - args->domain = RADEON_GEM_DOMAIN_CPU; - default: - break; - } + args->domain = radeon_mem_type_to_domain(cur_placement); drm_gem_object_unreference_unlocked(gobj); r = radeon_gem_handle_lockup(rdev, r); return r; @@ -533,6 +522,42 @@ out: return r; } +int radeon_gem_op_ioctl(struct drm_device *dev, void *data, + struct drm_file *filp) +{ + struct drm_radeon_gem_op *args = data; + struct drm_gem_object *gobj; + struct radeon_bo *robj; + int r; + + gobj = drm_gem_object_lookup(dev, filp, args->handle); + if (gobj == NULL) { + return -ENOENT; + } + robj = gem_to_radeon_bo(gobj); + r = radeon_bo_reserve(robj, false); + if (unlikely(r)) + goto out; + + switch (args->op) { + case RADEON_GEM_OP_GET_INITIAL_DOMAIN: + args->value = robj->initial_domain; + break; + case RADEON_GEM_OP_SET_INITIAL_DOMAIN: + robj->initial_domain = args->value & (RADEON_GEM_DOMAIN_VRAM | + RADEON_GEM_DOMAIN_GTT | + RADEON_GEM_DOMAIN_CPU); + break; + default: + r = -EINVAL; + } + + radeon_bo_unreserve(robj); +out: + drm_gem_object_unreference_unlocked(gobj); + return r; +} + int radeon_mode_dumb_create(struct drm_file *file_priv, struct drm_device *dev, struct drm_mode_create_dumb *args) diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index 66ed3ea71440..3e49342a20e6 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -441,6 +441,9 @@ static int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file case RADEON_CS_RING_UVD: *value = rdev->ring[R600_RING_TYPE_UVD_INDEX].ready; break; + case RADEON_CS_RING_VCE: + *value = rdev->ring[TN_RING_TYPE_VCE1_INDEX].ready; + break; default: return -EINVAL; } @@ -485,6 +488,27 @@ static int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file else *value = rdev->pm.default_sclk * 10; break; + case RADEON_INFO_VCE_FW_VERSION: + *value = rdev->vce.fw_version; + break; + case RADEON_INFO_VCE_FB_VERSION: + *value = rdev->vce.fb_version; + break; + case RADEON_INFO_NUM_BYTES_MOVED: + value = (uint32_t*)&value64; + value_size = sizeof(uint64_t); + value64 = atomic64_read(&rdev->num_bytes_moved); + break; + case RADEON_INFO_VRAM_USAGE: + value = (uint32_t*)&value64; + value_size = sizeof(uint64_t); + value64 = atomic64_read(&rdev->vram_usage); + break; + case RADEON_INFO_GTT_USAGE: + value = (uint32_t*)&value64; + value_size = sizeof(uint64_t); + value64 = atomic64_read(&rdev->gtt_usage); + break; default: DRM_DEBUG_KMS("Invalid request %d\n", info->request); return -EINVAL; @@ -543,7 +567,9 @@ int radeon_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv) return -ENOMEM; } - radeon_vm_init(rdev, &fpriv->vm); + r = radeon_vm_init(rdev, &fpriv->vm); + if (r) + return r; r = radeon_bo_reserve(rdev->ring_tmp_bo.bo, false); if (r) @@ -624,6 +650,7 @@ void radeon_driver_preclose_kms(struct drm_device *dev, if (rdev->cmask_filp == file_priv) rdev->cmask_filp = NULL; radeon_uvd_free_handles(rdev, file_priv); + radeon_vce_free_handles(rdev, file_priv); } /* @@ -818,5 +845,6 @@ const struct drm_ioctl_desc radeon_ioctls_kms[] = { DRM_IOCTL_DEF_DRV(RADEON_GEM_GET_TILING, radeon_gem_get_tiling_ioctl, DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(RADEON_GEM_BUSY, radeon_gem_busy_ioctl, DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(RADEON_GEM_VA, radeon_gem_va_ioctl, DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(RADEON_GEM_OP, radeon_gem_op_ioctl, DRM_AUTH|DRM_UNLOCKED|DRM_RENDER_ALLOW), }; int radeon_max_kms_ioctl = DRM_ARRAY_SIZE(radeon_ioctls_kms); diff --git a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c index 0b158f98d287..cafb1ccf2ec3 100644 --- a/drivers/gpu/drm/radeon/radeon_legacy_crtc.c +++ b/drivers/gpu/drm/radeon/radeon_legacy_crtc.c @@ -385,7 +385,7 @@ int radeon_crtc_do_set_base(struct drm_crtc *crtc, DRM_DEBUG_KMS("\n"); /* no fb bound */ - if (!atomic && !crtc->fb) { + if (!atomic && !crtc->primary->fb) { DRM_DEBUG_KMS("No FB bound\n"); return 0; } @@ -395,8 +395,8 @@ int radeon_crtc_do_set_base(struct drm_crtc *crtc, target_fb = fb; } else { - radeon_fb = to_radeon_framebuffer(crtc->fb); - target_fb = crtc->fb; + radeon_fb = to_radeon_framebuffer(crtc->primary->fb); + target_fb = crtc->primary->fb; } switch (target_fb->bits_per_pixel) { @@ -444,7 +444,7 @@ retry: * We don't shutdown the display controller because new buffer * will end up in same spot. */ - if (!atomic && fb && fb != crtc->fb) { + if (!atomic && fb && fb != crtc->primary->fb) { struct radeon_bo *old_rbo; unsigned long nsize, osize; @@ -555,7 +555,7 @@ retry: WREG32(RADEON_CRTC_OFFSET + radeon_crtc->crtc_offset, crtc_offset); WREG32(RADEON_CRTC_PITCH + radeon_crtc->crtc_offset, crtc_pitch); - if (!atomic && fb && fb != crtc->fb) { + if (!atomic && fb && fb != crtc->primary->fb) { radeon_fb = to_radeon_framebuffer(fb); rbo = gem_to_radeon_bo(radeon_fb->obj); r = radeon_bo_reserve(rbo, false); @@ -599,7 +599,7 @@ static bool radeon_set_crtc_timing(struct drm_crtc *crtc, struct drm_display_mod } } - switch (crtc->fb->bits_per_pixel) { + switch (crtc->primary->fb->bits_per_pixel) { case 8: format = 2; break; @@ -1087,12 +1087,12 @@ static void radeon_crtc_commit(struct drm_crtc *crtc) static void radeon_crtc_disable(struct drm_crtc *crtc) { radeon_crtc_dpms(crtc, DRM_MODE_DPMS_OFF); - if (crtc->fb) { + if (crtc->primary->fb) { int r; struct radeon_framebuffer *radeon_fb; struct radeon_bo *rbo; - radeon_fb = to_radeon_framebuffer(crtc->fb); + radeon_fb = to_radeon_framebuffer(crtc->primary->fb); rbo = gem_to_radeon_bo(radeon_fb->obj); r = radeon_bo_reserve(rbo, false); if (unlikely(r)) diff --git a/drivers/gpu/drm/radeon/radeon_mode.h b/drivers/gpu/drm/radeon/radeon_mode.h index 402dbe32c234..832d9fa1a4c4 100644 --- a/drivers/gpu/drm/radeon/radeon_mode.h +++ b/drivers/gpu/drm/radeon/radeon_mode.h @@ -192,6 +192,7 @@ struct radeon_i2c_chan { struct i2c_algo_dp_aux_data dp; } algo; struct radeon_i2c_bus_rec rec; + struct drm_dp_aux aux; }; /* mostly for macs, but really any system without connector tables */ @@ -690,6 +691,9 @@ extern u8 radeon_dp_getsinktype(struct radeon_connector *radeon_connector); extern bool radeon_dp_getdpcd(struct radeon_connector *radeon_connector); extern int radeon_dp_get_panel_mode(struct drm_encoder *encoder, struct drm_connector *connector); +extern void radeon_dp_set_rx_power_state(struct drm_connector *connector, + u8 power_state); +extern void radeon_dp_aux_init(struct radeon_connector *radeon_connector); extern void atombios_dig_encoder_setup(struct drm_encoder *encoder, int action, int panel_mode); extern void radeon_atom_encoder_init(struct radeon_device *rdev); extern void radeon_atom_disp_eng_pll_init(struct radeon_device *rdev); diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 08595cf90b01..19bec0dbfa38 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -56,11 +56,36 @@ static void radeon_bo_clear_va(struct radeon_bo *bo) } } +static void radeon_update_memory_usage(struct radeon_bo *bo, + unsigned mem_type, int sign) +{ + struct radeon_device *rdev = bo->rdev; + u64 size = (u64)bo->tbo.num_pages << PAGE_SHIFT; + + switch (mem_type) { + case TTM_PL_TT: + if (sign > 0) + atomic64_add(size, &rdev->gtt_usage); + else + atomic64_sub(size, &rdev->gtt_usage); + break; + case TTM_PL_VRAM: + if (sign > 0) + atomic64_add(size, &rdev->vram_usage); + else + atomic64_sub(size, &rdev->vram_usage); + break; + } +} + static void radeon_ttm_bo_destroy(struct ttm_buffer_object *tbo) { struct radeon_bo *bo; bo = container_of(tbo, struct radeon_bo, tbo); + + radeon_update_memory_usage(bo, bo->tbo.mem.mem_type, -1); + mutex_lock(&bo->rdev->gem.mutex); list_del_init(&bo->list); mutex_unlock(&bo->rdev->gem.mutex); @@ -79,7 +104,7 @@ bool radeon_ttm_bo_is_radeon_bo(struct ttm_buffer_object *bo) void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) { - u32 c = 0; + u32 c = 0, i; rbo->placement.fpfn = 0; rbo->placement.lpfn = 0; @@ -106,6 +131,17 @@ void radeon_ttm_placement_from_domain(struct radeon_bo *rbo, u32 domain) rbo->placements[c++] = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM; rbo->placement.num_placement = c; rbo->placement.num_busy_placement = c; + + /* + * Use two-ended allocation depending on the buffer size to + * improve fragmentation quality. + * 512kb was measured as the most optimal number. + */ + if (rbo->tbo.mem.size > 512 * 1024) { + for (i = 0; i < c; i++) { + rbo->placements[i] |= TTM_PL_FLAG_TOPDOWN; + } + } } int radeon_bo_create(struct radeon_device *rdev, @@ -120,7 +156,6 @@ int radeon_bo_create(struct radeon_device *rdev, size = ALIGN(size, PAGE_SIZE); - rdev->mman.bdev.dev_mapping = rdev->ddev->dev_mapping; if (kernel) { type = ttm_bo_type_kernel; } else if (sg) { @@ -145,6 +180,9 @@ int radeon_bo_create(struct radeon_device *rdev, bo->surface_reg = -1; INIT_LIST_HEAD(&bo->list); INIT_LIST_HEAD(&bo->va); + bo->initial_domain = domain & (RADEON_GEM_DOMAIN_VRAM | + RADEON_GEM_DOMAIN_GTT | + RADEON_GEM_DOMAIN_CPU); radeon_ttm_placement_from_domain(bo, domain); /* Kernel allocation are uninterruptible */ down_read(&rdev->pm.mclk_lock); @@ -338,39 +376,105 @@ void radeon_bo_fini(struct radeon_device *rdev) arch_phys_wc_del(rdev->mc.vram_mtrr); } -void radeon_bo_list_add_object(struct radeon_bo_list *lobj, - struct list_head *head) +/* Returns how many bytes TTM can move per IB. + */ +static u64 radeon_bo_get_threshold_for_moves(struct radeon_device *rdev) { - if (lobj->written) { - list_add(&lobj->tv.head, head); - } else { - list_add_tail(&lobj->tv.head, head); - } + u64 real_vram_size = rdev->mc.real_vram_size; + u64 vram_usage = atomic64_read(&rdev->vram_usage); + + /* This function is based on the current VRAM usage. + * + * - If all of VRAM is free, allow relocating the number of bytes that + * is equal to 1/4 of the size of VRAM for this IB. + + * - If more than one half of VRAM is occupied, only allow relocating + * 1 MB of data for this IB. + * + * - From 0 to one half of used VRAM, the threshold decreases + * linearly. + * __________________ + * 1/4 of -|\ | + * VRAM | \ | + * | \ | + * | \ | + * | \ | + * | \ | + * | \ | + * | \________|1 MB + * |----------------| + * VRAM 0 % 100 % + * used used + * + * Note: It's a threshold, not a limit. The threshold must be crossed + * for buffer relocations to stop, so any buffer of an arbitrary size + * can be moved as long as the threshold isn't crossed before + * the relocation takes place. We don't want to disable buffer + * relocations completely. + * + * The idea is that buffers should be placed in VRAM at creation time + * and TTM should only do a minimum number of relocations during + * command submission. In practice, you need to submit at least + * a dozen IBs to move all buffers to VRAM if they are in GTT. + * + * Also, things can get pretty crazy under memory pressure and actual + * VRAM usage can change a lot, so playing safe even at 50% does + * consistently increase performance. + */ + + u64 half_vram = real_vram_size >> 1; + u64 half_free_vram = vram_usage >= half_vram ? 0 : half_vram - vram_usage; + u64 bytes_moved_threshold = half_free_vram >> 1; + return max(bytes_moved_threshold, 1024*1024ull); } -int radeon_bo_list_validate(struct ww_acquire_ctx *ticket, +int radeon_bo_list_validate(struct radeon_device *rdev, + struct ww_acquire_ctx *ticket, struct list_head *head, int ring) { - struct radeon_bo_list *lobj; + struct radeon_cs_reloc *lobj; struct radeon_bo *bo; - u32 domain; 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); if (unlikely(r != 0)) { return r; } + list_for_each_entry(lobj, head, tv.head) { - bo = lobj->bo; + bo = lobj->robj; if (!bo->pin_count) { - domain = lobj->domain; - + u32 domain = lobj->domain; + u32 current_domain = + radeon_mem_type_to_domain(bo->tbo.mem.mem_type); + + /* Check if this buffer will be moved and don't move it + * if we have moved too many buffers for this IB already. + * + * Note that this allows moving at least one buffer of + * any size, because it doesn't take the current "bo" + * into account. We don't want to disallow buffer moves + * completely. + */ + if (current_domain != RADEON_GEM_DOMAIN_CPU && + (domain & current_domain) == 0 && /* will be moved */ + bytes_moved > bytes_moved_threshold) { + /* don't move it */ + domain = current_domain; + } + retry: radeon_ttm_placement_from_domain(bo, domain); if (ring == R600_RING_TYPE_UVD_INDEX) radeon_uvd_force_into_uvd_segment(bo); - r = ttm_bo_validate(&bo->tbo, &bo->placement, - true, false); + + initial_bytes_moved = atomic64_read(&rdev->num_bytes_moved); + r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false); + bytes_moved += atomic64_read(&rdev->num_bytes_moved) - + initial_bytes_moved; + if (unlikely(r)) { if (r != -ERESTARTSYS && domain != lobj->alt_domain) { domain = lobj->alt_domain; @@ -564,14 +668,23 @@ int radeon_bo_check_tiling(struct radeon_bo *bo, bool has_moved, } void radeon_bo_move_notify(struct ttm_buffer_object *bo, - struct ttm_mem_reg *mem) + struct ttm_mem_reg *new_mem) { struct radeon_bo *rbo; + if (!radeon_ttm_bo_is_radeon_bo(bo)) return; + rbo = container_of(bo, struct radeon_bo, tbo); radeon_bo_check_tiling(rbo, 0, 1); radeon_vm_bo_invalidate(rbo->rdev, rbo); + + /* update statistics */ + if (!new_mem) + return; + + radeon_update_memory_usage(rbo, bo->mem.mem_type, -1); + radeon_update_memory_usage(rbo, new_mem->mem_type, 1); } int radeon_bo_fault_reserve_notify(struct ttm_buffer_object *bo) diff --git a/drivers/gpu/drm/radeon/radeon_object.h b/drivers/gpu/drm/radeon/radeon_object.h index 209b11150263..9e7b25a0629d 100644 --- a/drivers/gpu/drm/radeon/radeon_object.h +++ b/drivers/gpu/drm/radeon/radeon_object.h @@ -138,9 +138,8 @@ extern int radeon_bo_evict_vram(struct radeon_device *rdev); extern void radeon_bo_force_delete(struct radeon_device *rdev); extern int radeon_bo_init(struct radeon_device *rdev); extern void radeon_bo_fini(struct radeon_device *rdev); -extern void radeon_bo_list_add_object(struct radeon_bo_list *lobj, - struct list_head *head); -extern int radeon_bo_list_validate(struct ww_acquire_ctx *ticket, +extern int radeon_bo_list_validate(struct radeon_device *rdev, + struct ww_acquire_ctx *ticket, struct list_head *head, int ring); extern int radeon_bo_fbdev_mmap(struct radeon_bo *bo, struct vm_area_struct *vma); @@ -151,7 +150,7 @@ extern void radeon_bo_get_tiling_flags(struct radeon_bo *bo, extern int radeon_bo_check_tiling(struct radeon_bo *bo, bool has_moved, bool force_drop); extern void radeon_bo_move_notify(struct ttm_buffer_object *bo, - struct ttm_mem_reg *mem); + 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); @@ -181,7 +180,7 @@ extern int radeon_sa_bo_manager_suspend(struct radeon_device *rdev, extern int radeon_sa_bo_new(struct radeon_device *rdev, struct radeon_sa_manager *sa_manager, struct radeon_sa_bo **sa_bo, - unsigned size, unsigned align, bool block); + unsigned size, unsigned align); extern void radeon_sa_bo_free(struct radeon_device *rdev, struct radeon_sa_bo **sa_bo, struct radeon_fence *fence); diff --git a/drivers/gpu/drm/radeon/radeon_pm.c b/drivers/gpu/drm/radeon/radeon_pm.c index 8e8153e471c2..ee738a524639 100644 --- a/drivers/gpu/drm/radeon/radeon_pm.c +++ b/drivers/gpu/drm/radeon/radeon_pm.c @@ -260,7 +260,7 @@ static void radeon_pm_set_clocks(struct radeon_device *rdev) if (!ring->ready) { continue; } - r = radeon_fence_wait_empty_locked(rdev, i); + r = radeon_fence_wait_empty(rdev, i); if (r) { /* needs a GPU reset dont reset here */ mutex_unlock(&rdev->ring_lock); @@ -826,6 +826,9 @@ static void radeon_dpm_change_power_state_locked(struct radeon_device *rdev) /* no need to reprogram if nothing changed unless we are on BTC+ */ if (rdev->pm.dpm.current_ps == rdev->pm.dpm.requested_ps) { + /* vce just modifies an existing state so force a change */ + if (ps->vce_active != rdev->pm.dpm.vce_active) + goto force; if ((rdev->family < CHIP_BARTS) || (rdev->flags & RADEON_IS_IGP)) { /* for pre-BTC and APUs if the num crtcs changed but state is the same, * all we need to do is update the display configuration. @@ -862,16 +865,21 @@ static void radeon_dpm_change_power_state_locked(struct radeon_device *rdev) } } +force: if (radeon_dpm == 1) { printk("switching from power state:\n"); radeon_dpm_print_power_state(rdev, rdev->pm.dpm.current_ps); printk("switching to power state:\n"); radeon_dpm_print_power_state(rdev, rdev->pm.dpm.requested_ps); } + mutex_lock(&rdev->ddev->struct_mutex); down_write(&rdev->pm.mclk_lock); mutex_lock(&rdev->ring_lock); + /* update whether vce is active */ + ps->vce_active = rdev->pm.dpm.vce_active; + ret = radeon_dpm_pre_set_power_state(rdev); if (ret) goto done; @@ -888,7 +896,7 @@ static void radeon_dpm_change_power_state_locked(struct radeon_device *rdev) for (i = 0; i < RADEON_NUM_RINGS; i++) { struct radeon_ring *ring = &rdev->ring[i]; if (ring->ready) - radeon_fence_wait_empty_locked(rdev, i); + radeon_fence_wait_empty(rdev, i); } /* program the new power state */ @@ -935,8 +943,6 @@ void radeon_dpm_enable_uvd(struct radeon_device *rdev, bool enable) if (enable) { mutex_lock(&rdev->pm.mutex); rdev->pm.dpm.uvd_active = true; - /* disable this for now */ -#if 0 if ((rdev->pm.dpm.sd == 1) && (rdev->pm.dpm.hd == 0)) dpm_state = POWER_STATE_TYPE_INTERNAL_UVD_SD; else if ((rdev->pm.dpm.sd == 2) && (rdev->pm.dpm.hd == 0)) @@ -946,7 +952,6 @@ void radeon_dpm_enable_uvd(struct radeon_device *rdev, bool enable) else if ((rdev->pm.dpm.sd == 0) && (rdev->pm.dpm.hd == 2)) dpm_state = POWER_STATE_TYPE_INTERNAL_UVD_HD2; else -#endif dpm_state = POWER_STATE_TYPE_INTERNAL_UVD; rdev->pm.dpm.state = dpm_state; mutex_unlock(&rdev->pm.mutex); @@ -960,6 +965,23 @@ void radeon_dpm_enable_uvd(struct radeon_device *rdev, bool enable) } } +void radeon_dpm_enable_vce(struct radeon_device *rdev, bool enable) +{ + if (enable) { + mutex_lock(&rdev->pm.mutex); + rdev->pm.dpm.vce_active = true; + /* XXX select vce level based on ring/task */ + rdev->pm.dpm.vce_level = RADEON_VCE_LEVEL_AC_ALL; + mutex_unlock(&rdev->pm.mutex); + } else { + mutex_lock(&rdev->pm.mutex); + rdev->pm.dpm.vce_active = false; + mutex_unlock(&rdev->pm.mutex); + } + + radeon_pm_compute_clocks(rdev); +} + static void radeon_pm_suspend_old(struct radeon_device *rdev) { mutex_lock(&rdev->pm.mutex); diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c index 15e44a7281ab..f8050f5429e2 100644 --- a/drivers/gpu/drm/radeon/radeon_ring.c +++ b/drivers/gpu/drm/radeon/radeon_ring.c @@ -63,7 +63,7 @@ int radeon_ib_get(struct radeon_device *rdev, int ring, { int r; - r = radeon_sa_bo_new(rdev, &rdev->ring_tmp_bo, &ib->sa_bo, size, 256, true); + r = radeon_sa_bo_new(rdev, &rdev->ring_tmp_bo, &ib->sa_bo, size, 256); if (r) { dev_err(rdev->dev, "failed to get a new IB (%d)\n", r); return r; @@ -145,6 +145,13 @@ int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib, return r; } + /* grab a vm id if necessary */ + if (ib->vm) { + struct radeon_fence *vm_id_fence; + vm_id_fence = radeon_vm_grab_id(rdev, ib->vm, ib->ring); + radeon_semaphore_sync_to(ib->semaphore, vm_id_fence); + } + /* sync with other rings */ r = radeon_semaphore_sync_rings(rdev, ib->semaphore, ib->ring); if (r) { @@ -153,11 +160,9 @@ int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib, return r; } - /* if we can't remember our last VM flush then flush now! */ - /* XXX figure out why we have to flush for every IB */ - if (ib->vm /*&& !ib->vm->last_flush*/) { - radeon_ring_vm_flush(rdev, ib->ring, ib->vm); - } + if (ib->vm) + radeon_vm_flush(rdev, ib->vm, ib->ring); + if (const_ib) { radeon_ring_ib_execute(rdev, const_ib->ring, const_ib); radeon_semaphore_free(rdev, &const_ib->semaphore, NULL); @@ -172,10 +177,10 @@ int radeon_ib_schedule(struct radeon_device *rdev, struct radeon_ib *ib, if (const_ib) { const_ib->fence = radeon_fence_ref(ib->fence); } - /* we just flushed the VM, remember that */ - if (ib->vm && !ib->vm->last_flush) { - ib->vm->last_flush = radeon_fence_ref(ib->fence); - } + + if (ib->vm) + radeon_vm_fence(rdev, ib->vm, ib->fence); + radeon_ring_unlock_commit(rdev, ring); return 0; } @@ -257,6 +262,7 @@ int radeon_ib_ring_tests(struct radeon_device *rdev) r = radeon_ib_test(rdev, i, ring); if (r) { ring->ready = false; + rdev->needs_reset = false; if (i == RADEON_RING_TYPE_GFX_INDEX) { /* oh, oh, that's really bad */ @@ -342,13 +348,17 @@ bool radeon_ring_supports_scratch_reg(struct radeon_device *rdev, */ void radeon_ring_free_size(struct radeon_device *rdev, struct radeon_ring *ring) { - ring->rptr = radeon_ring_get_rptr(rdev, ring); + uint32_t rptr = radeon_ring_get_rptr(rdev, ring); + /* This works because ring_size is a power of 2 */ - ring->ring_free_dw = (ring->rptr + (ring->ring_size / 4)); + ring->ring_free_dw = rptr + (ring->ring_size / 4); ring->ring_free_dw -= ring->wptr; ring->ring_free_dw &= ring->ptr_mask; if (!ring->ring_free_dw) { + /* this is an empty ring */ ring->ring_free_dw = ring->ring_size / 4; + /* update lockup info to avoid false positive */ + radeon_ring_lockup_update(rdev, ring); } } @@ -372,19 +382,13 @@ int radeon_ring_alloc(struct radeon_device *rdev, struct radeon_ring *ring, unsi /* Align requested size with padding so unlock_commit can * pad safely */ radeon_ring_free_size(rdev, ring); - if (ring->ring_free_dw == (ring->ring_size / 4)) { - /* This is an empty ring update lockup info to avoid - * false positive. - */ - radeon_ring_lockup_update(ring); - } ndw = (ndw + ring->align_mask) & ~ring->align_mask; while (ndw > (ring->ring_free_dw - 1)) { radeon_ring_free_size(rdev, ring); if (ndw < ring->ring_free_dw) { break; } - r = radeon_fence_wait_next_locked(rdev, ring->idx); + r = radeon_fence_wait_next(rdev, ring->idx); if (r) return r; } @@ -478,39 +482,17 @@ void radeon_ring_unlock_undo(struct radeon_device *rdev, struct radeon_ring *rin } /** - * radeon_ring_force_activity - add some nop packets to the ring - * - * @rdev: radeon_device pointer - * @ring: radeon_ring structure holding ring information - * - * Add some nop packets to the ring to force activity (all asics). - * Used for lockup detection to see if the rptr is advancing. - */ -void radeon_ring_force_activity(struct radeon_device *rdev, struct radeon_ring *ring) -{ - int r; - - radeon_ring_free_size(rdev, ring); - if (ring->rptr == ring->wptr) { - r = radeon_ring_alloc(rdev, ring, 1); - if (!r) { - radeon_ring_write(ring, ring->nop); - radeon_ring_commit(rdev, ring); - } - } -} - -/** * radeon_ring_lockup_update - update lockup variables * * @ring: radeon_ring structure holding ring information * * Update the last rptr value and timestamp (all asics). */ -void radeon_ring_lockup_update(struct radeon_ring *ring) +void radeon_ring_lockup_update(struct radeon_device *rdev, + struct radeon_ring *ring) { - ring->last_rptr = ring->rptr; - ring->last_activity = jiffies; + atomic_set(&ring->last_rptr, radeon_ring_get_rptr(rdev, ring)); + atomic64_set(&ring->last_activity, jiffies_64); } /** @@ -518,40 +500,23 @@ void radeon_ring_lockup_update(struct radeon_ring *ring) * @rdev: radeon device structure * @ring: radeon_ring structure holding ring information * - * We don't need to initialize the lockup tracking information as we will either - * have CP rptr to a different value of jiffies wrap around which will force - * initialization of the lockup tracking informations. - * - * A possible false positivie is if we get call after while and last_cp_rptr == - * the current CP rptr, even if it's unlikely it might happen. To avoid this - * if the elapsed time since last call is bigger than 2 second than we return - * false and update the tracking information. Due to this the caller must call - * radeon_ring_test_lockup several time in less than 2sec for lockup to be reported - * the fencing code should be cautious about that. - * - * Caller should write to the ring to force CP to do something so we don't get - * false positive when CP is just gived nothing to do. - * - **/ + */ bool radeon_ring_test_lockup(struct radeon_device *rdev, struct radeon_ring *ring) { - unsigned long cjiffies, elapsed; + uint32_t rptr = radeon_ring_get_rptr(rdev, ring); + uint64_t last = atomic64_read(&ring->last_activity); + uint64_t elapsed; - cjiffies = jiffies; - if (!time_after(cjiffies, ring->last_activity)) { - /* likely a wrap around */ - radeon_ring_lockup_update(ring); + if (rptr != atomic_read(&ring->last_rptr)) { + /* ring is still working, no lockup */ + radeon_ring_lockup_update(rdev, ring); return false; } - ring->rptr = radeon_ring_get_rptr(rdev, ring); - if (ring->rptr != ring->last_rptr) { - /* CP is still working no lockup */ - radeon_ring_lockup_update(ring); - return false; - } - elapsed = jiffies_to_msecs(cjiffies - ring->last_activity); + + elapsed = jiffies_to_msecs(jiffies_64 - last); if (radeon_lockup_timeout && elapsed >= radeon_lockup_timeout) { - dev_err(rdev->dev, "GPU lockup CP stall for more than %lumsec\n", elapsed); + dev_err(rdev->dev, "ring %d stalled for more than %llumsec\n", + ring->idx, elapsed); return true; } /* give a chance to the GPU ... */ @@ -709,7 +674,7 @@ int radeon_ring_init(struct radeon_device *rdev, struct radeon_ring *ring, unsig if (radeon_debugfs_ring_init(rdev, ring)) { DRM_ERROR("Failed to register debugfs file for rings !\n"); } - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return 0; } @@ -780,8 +745,6 @@ static int radeon_debugfs_ring_info(struct seq_file *m, void *data) seq_printf(m, "driver's copy of the wptr: 0x%08x [%5d]\n", ring->wptr, ring->wptr); - seq_printf(m, "driver's copy of the rptr: 0x%08x [%5d]\n", - ring->rptr, ring->rptr); seq_printf(m, "last semaphore signal addr : 0x%016llx\n", ring->last_semaphore_signal_addr); seq_printf(m, "last semaphore wait addr : 0x%016llx\n", @@ -814,6 +777,8 @@ static int cayman_cp2_index = CAYMAN_RING_TYPE_CP2_INDEX; static int radeon_dma1_index = R600_RING_TYPE_DMA_INDEX; static int radeon_dma2_index = CAYMAN_RING_TYPE_DMA1_INDEX; static int r600_uvd_index = R600_RING_TYPE_UVD_INDEX; +static int si_vce1_index = TN_RING_TYPE_VCE1_INDEX; +static int si_vce2_index = TN_RING_TYPE_VCE2_INDEX; static struct drm_info_list radeon_debugfs_ring_info_list[] = { {"radeon_ring_gfx", radeon_debugfs_ring_info, 0, &radeon_gfx_index}, @@ -822,6 +787,8 @@ static struct drm_info_list radeon_debugfs_ring_info_list[] = { {"radeon_ring_dma1", radeon_debugfs_ring_info, 0, &radeon_dma1_index}, {"radeon_ring_dma2", radeon_debugfs_ring_info, 0, &radeon_dma2_index}, {"radeon_ring_uvd", radeon_debugfs_ring_info, 0, &r600_uvd_index}, + {"radeon_ring_vce1", radeon_debugfs_ring_info, 0, &si_vce1_index}, + {"radeon_ring_vce2", radeon_debugfs_ring_info, 0, &si_vce2_index}, }; static int radeon_debugfs_sa_info(struct seq_file *m, void *data) diff --git a/drivers/gpu/drm/radeon/radeon_sa.c b/drivers/gpu/drm/radeon/radeon_sa.c index c0625805cdd7..adcf3e2f07da 100644 --- a/drivers/gpu/drm/radeon/radeon_sa.c +++ b/drivers/gpu/drm/radeon/radeon_sa.c @@ -312,7 +312,7 @@ static bool radeon_sa_bo_next_hole(struct radeon_sa_manager *sa_manager, int radeon_sa_bo_new(struct radeon_device *rdev, struct radeon_sa_manager *sa_manager, struct radeon_sa_bo **sa_bo, - unsigned size, unsigned align, bool block) + unsigned size, unsigned align) { struct radeon_fence *fences[RADEON_NUM_RINGS]; unsigned tries[RADEON_NUM_RINGS]; @@ -353,14 +353,11 @@ int radeon_sa_bo_new(struct radeon_device *rdev, r = radeon_fence_wait_any(rdev, fences, false); spin_lock(&sa_manager->wq.lock); /* if we have nothing to wait for block */ - if (r == -ENOENT && block) { + if (r == -ENOENT) { r = wait_event_interruptible_locked( sa_manager->wq, radeon_sa_event(sa_manager, size, align) ); - - } else if (r == -ENOENT) { - r = -ENOMEM; } } while (!r); diff --git a/drivers/gpu/drm/radeon/radeon_semaphore.c b/drivers/gpu/drm/radeon/radeon_semaphore.c index 9006b32d5eed..dbd6bcde92de 100644 --- a/drivers/gpu/drm/radeon/radeon_semaphore.c +++ b/drivers/gpu/drm/radeon/radeon_semaphore.c @@ -42,7 +42,7 @@ int radeon_semaphore_create(struct radeon_device *rdev, return -ENOMEM; } r = radeon_sa_bo_new(rdev, &rdev->ring_tmp_bo, &(*semaphore)->sa_bo, - 8 * RADEON_NUM_SYNCS, 8, true); + 8 * RADEON_NUM_SYNCS, 8); if (r) { kfree(*semaphore); *semaphore = NULL; @@ -147,7 +147,9 @@ int radeon_semaphore_sync_rings(struct radeon_device *rdev, if (++count > RADEON_NUM_SYNCS) { /* not enough room, wait manually */ - radeon_fence_wait_locked(fence); + r = radeon_fence_wait(fence, false); + if (r) + return r; continue; } @@ -161,7 +163,9 @@ int radeon_semaphore_sync_rings(struct radeon_device *rdev, if (!radeon_semaphore_emit_signal(rdev, i, semaphore)) { /* signaling wasn't successful wait manually */ radeon_ring_undo(&rdev->ring[i]); - radeon_fence_wait_locked(fence); + r = radeon_fence_wait(fence, false); + if (r) + return r; continue; } @@ -169,7 +173,9 @@ int radeon_semaphore_sync_rings(struct radeon_device *rdev, if (!radeon_semaphore_emit_wait(rdev, ring, semaphore)) { /* waiting wasn't successful wait manually */ radeon_ring_undo(&rdev->ring[i]); - radeon_fence_wait_locked(fence); + r = radeon_fence_wait(fence, false); + if (r) + return r; continue; } diff --git a/drivers/gpu/drm/radeon/radeon_test.c b/drivers/gpu/drm/radeon/radeon_test.c index 12e8099a0823..3a13e0d1055c 100644 --- a/drivers/gpu/drm/radeon/radeon_test.c +++ b/drivers/gpu/drm/radeon/radeon_test.c @@ -257,20 +257,36 @@ static int radeon_test_create_and_emit_fence(struct radeon_device *rdev, struct radeon_ring *ring, struct radeon_fence **fence) { + uint32_t handle = ring->idx ^ 0xdeafbeef; int r; if (ring->idx == R600_RING_TYPE_UVD_INDEX) { - r = radeon_uvd_get_create_msg(rdev, ring->idx, 1, NULL); + r = radeon_uvd_get_create_msg(rdev, ring->idx, handle, NULL); if (r) { DRM_ERROR("Failed to get dummy create msg\n"); return r; } - r = radeon_uvd_get_destroy_msg(rdev, ring->idx, 1, fence); + r = radeon_uvd_get_destroy_msg(rdev, ring->idx, handle, fence); if (r) { DRM_ERROR("Failed to get dummy destroy msg\n"); return r; } + + } else if (ring->idx == TN_RING_TYPE_VCE1_INDEX || + ring->idx == TN_RING_TYPE_VCE2_INDEX) { + r = radeon_vce_get_create_msg(rdev, ring->idx, handle, NULL); + if (r) { + DRM_ERROR("Failed to get dummy create msg\n"); + return r; + } + + r = radeon_vce_get_destroy_msg(rdev, ring->idx, handle, fence); + if (r) { + DRM_ERROR("Failed to get dummy destroy msg\n"); + return r; + } + } else { r = radeon_ring_lock(rdev, ring, 64); if (r) { @@ -486,6 +502,16 @@ out_cleanup: printk(KERN_WARNING "Error while testing ring sync (%d).\n", r); } +static bool radeon_test_sync_possible(struct radeon_ring *ringA, + struct radeon_ring *ringB) +{ + if (ringA->idx == TN_RING_TYPE_VCE2_INDEX && + ringB->idx == TN_RING_TYPE_VCE1_INDEX) + return false; + + return true; +} + void radeon_test_syncing(struct radeon_device *rdev) { int i, j, k; @@ -500,6 +526,9 @@ void radeon_test_syncing(struct radeon_device *rdev) if (!ringB->ready) continue; + if (!radeon_test_sync_possible(ringA, ringB)) + continue; + DRM_INFO("Testing syncing between rings %d and %d...\n", i, j); radeon_test_ring_sync(rdev, ringA, ringB); @@ -511,6 +540,12 @@ void radeon_test_syncing(struct radeon_device *rdev) if (!ringC->ready) continue; + if (!radeon_test_sync_possible(ringA, ringC)) + continue; + + if (!radeon_test_sync_possible(ringB, ringC)) + continue; + DRM_INFO("Testing syncing between rings %d, %d and %d...\n", i, j, k); radeon_test_ring_sync2(rdev, ringA, ringB, ringC); diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index 040a2a10ea17..c8a8a5144ec1 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -406,8 +406,14 @@ static int radeon_bo_move(struct ttm_buffer_object *bo, if (r) { memcpy: r = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); + if (r) { + return r; + } } - return r; + + /* update statistics */ + atomic64_add((u64)bo->num_pages << PAGE_SHIFT, &rdev->num_bytes_moved); + return 0; } static int radeon_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) @@ -701,7 +707,9 @@ int radeon_ttm_init(struct radeon_device *rdev) /* No others user of address space so set it to 0 */ r = ttm_bo_device_init(&rdev->mman.bdev, rdev->mman.bo_global_ref.ref.object, - &radeon_bo_driver, DRM_FILE_PAGE_OFFSET, + &radeon_bo_driver, + rdev->ddev->anon_inode->i_mapping, + DRM_FILE_PAGE_OFFSET, rdev->need_dma32); if (r) { DRM_ERROR("failed initializing buffer object driver(%d).\n", r); @@ -742,7 +750,6 @@ int radeon_ttm_init(struct radeon_device *rdev) } DRM_INFO("radeon: %uM of GTT memory ready.\n", (unsigned)(rdev->mc.gtt_size / (1024 * 1024))); - rdev->mman.bdev.dev_mapping = rdev->ddev->dev_mapping; r = radeon_ttm_debugfs_init(rdev); if (r) { diff --git a/drivers/gpu/drm/radeon/radeon_uvd.c b/drivers/gpu/drm/radeon/radeon_uvd.c index 3e6804b2b2ef..5748bdaeacce 100644 --- a/drivers/gpu/drm/radeon/radeon_uvd.c +++ b/drivers/gpu/drm/radeon/radeon_uvd.c @@ -455,7 +455,7 @@ static int radeon_uvd_cs_reloc(struct radeon_cs_parser *p, } reloc = p->relocs_ptr[(idx / 4)]; - start = reloc->lobj.gpu_offset; + start = reloc->gpu_offset; end = start + radeon_bo_size(reloc->robj); start += offset; @@ -807,8 +807,7 @@ void radeon_uvd_note_usage(struct radeon_device *rdev) (rdev->pm.dpm.hd != hd)) { rdev->pm.dpm.sd = sd; rdev->pm.dpm.hd = hd; - /* disable this for now */ - /*streams_changed = true;*/ + streams_changed = true; } } diff --git a/drivers/gpu/drm/radeon/radeon_vce.c b/drivers/gpu/drm/radeon/radeon_vce.c new file mode 100644 index 000000000000..76e9904bc537 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_vce.c @@ -0,0 +1,699 @@ +/* + * Copyright 2013 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 <linux/firmware.h> +#include <linux/module.h> +#include <drm/drmP.h> +#include <drm/drm.h> + +#include "radeon.h" +#include "radeon_asic.h" +#include "sid.h" + +/* 1 second timeout */ +#define VCE_IDLE_TIMEOUT_MS 1000 + +/* Firmware Names */ +#define FIRMWARE_BONAIRE "radeon/BONAIRE_vce.bin" + +MODULE_FIRMWARE(FIRMWARE_BONAIRE); + +static void radeon_vce_idle_work_handler(struct work_struct *work); + +/** + * radeon_vce_init - allocate memory, load vce firmware + * + * @rdev: radeon_device pointer + * + * First step to get VCE online, allocate memory and load the firmware + */ +int radeon_vce_init(struct radeon_device *rdev) +{ + static const char *fw_version = "[ATI LIB=VCEFW,"; + static const char *fb_version = "[ATI LIB=VCEFWSTATS,"; + unsigned long size; + const char *fw_name, *c; + uint8_t start, mid, end; + int i, r; + + INIT_DELAYED_WORK(&rdev->vce.idle_work, radeon_vce_idle_work_handler); + + switch (rdev->family) { + case CHIP_BONAIRE: + case CHIP_KAVERI: + case CHIP_KABINI: + fw_name = FIRMWARE_BONAIRE; + break; + + default: + return -EINVAL; + } + + r = request_firmware(&rdev->vce_fw, fw_name, rdev->dev); + if (r) { + dev_err(rdev->dev, "radeon_vce: Can't load firmware \"%s\"\n", + fw_name); + return r; + } + + /* search for firmware version */ + + size = rdev->vce_fw->size - strlen(fw_version) - 9; + c = rdev->vce_fw->data; + for (;size > 0; --size, ++c) + if (strncmp(c, fw_version, strlen(fw_version)) == 0) + break; + + if (size == 0) + return -EINVAL; + + c += strlen(fw_version); + if (sscanf(c, "%2hhd.%2hhd.%2hhd]", &start, &mid, &end) != 3) + return -EINVAL; + + /* search for feedback version */ + + size = rdev->vce_fw->size - strlen(fb_version) - 3; + c = rdev->vce_fw->data; + for (;size > 0; --size, ++c) + if (strncmp(c, fb_version, strlen(fb_version)) == 0) + break; + + if (size == 0) + return -EINVAL; + + c += strlen(fb_version); + if (sscanf(c, "%2u]", &rdev->vce.fb_version) != 1) + return -EINVAL; + + DRM_INFO("Found VCE firmware/feedback version %hhd.%hhd.%hhd / %d!\n", + start, mid, end, rdev->vce.fb_version); + + rdev->vce.fw_version = (start << 24) | (mid << 16) | (end << 8); + + /* we can only work with this fw version for now */ + if (rdev->vce.fw_version != ((40 << 24) | (2 << 16) | (2 << 8))) + return -EINVAL; + + /* allocate firmware, stack and heap BO */ + + size = RADEON_GPU_PAGE_ALIGN(rdev->vce_fw->size) + + RADEON_VCE_STACK_SIZE + RADEON_VCE_HEAP_SIZE; + r = radeon_bo_create(rdev, size, PAGE_SIZE, true, + RADEON_GEM_DOMAIN_VRAM, NULL, &rdev->vce.vcpu_bo); + if (r) { + dev_err(rdev->dev, "(%d) failed to allocate VCE bo\n", r); + return r; + } + + r = radeon_bo_reserve(rdev->vce.vcpu_bo, false); + if (r) { + radeon_bo_unref(&rdev->vce.vcpu_bo); + dev_err(rdev->dev, "(%d) failed to reserve VCE bo\n", r); + return r; + } + + r = radeon_bo_pin(rdev->vce.vcpu_bo, RADEON_GEM_DOMAIN_VRAM, + &rdev->vce.gpu_addr); + radeon_bo_unreserve(rdev->vce.vcpu_bo); + if (r) { + radeon_bo_unref(&rdev->vce.vcpu_bo); + dev_err(rdev->dev, "(%d) VCE bo pin failed\n", r); + return r; + } + + for (i = 0; i < RADEON_MAX_VCE_HANDLES; ++i) { + atomic_set(&rdev->vce.handles[i], 0); + rdev->vce.filp[i] = NULL; + } + + return 0; +} + +/** + * radeon_vce_fini - free memory + * + * @rdev: radeon_device pointer + * + * Last step on VCE teardown, free firmware memory + */ +void radeon_vce_fini(struct radeon_device *rdev) +{ + if (rdev->vce.vcpu_bo == NULL) + return; + + radeon_bo_unref(&rdev->vce.vcpu_bo); + + release_firmware(rdev->vce_fw); +} + +/** + * radeon_vce_suspend - unpin VCE fw memory + * + * @rdev: radeon_device pointer + * + */ +int radeon_vce_suspend(struct radeon_device *rdev) +{ + int i; + + if (rdev->vce.vcpu_bo == NULL) + return 0; + + for (i = 0; i < RADEON_MAX_VCE_HANDLES; ++i) + if (atomic_read(&rdev->vce.handles[i])) + break; + + if (i == RADEON_MAX_VCE_HANDLES) + return 0; + + /* TODO: suspending running encoding sessions isn't supported */ + return -EINVAL; +} + +/** + * radeon_vce_resume - pin VCE fw memory + * + * @rdev: radeon_device pointer + * + */ +int radeon_vce_resume(struct radeon_device *rdev) +{ + void *cpu_addr; + int r; + + if (rdev->vce.vcpu_bo == NULL) + return -EINVAL; + + r = radeon_bo_reserve(rdev->vce.vcpu_bo, false); + if (r) { + dev_err(rdev->dev, "(%d) failed to reserve VCE bo\n", r); + return r; + } + + r = radeon_bo_kmap(rdev->vce.vcpu_bo, &cpu_addr); + if (r) { + radeon_bo_unreserve(rdev->vce.vcpu_bo); + dev_err(rdev->dev, "(%d) VCE map failed\n", r); + return r; + } + + memcpy(cpu_addr, rdev->vce_fw->data, rdev->vce_fw->size); + + radeon_bo_kunmap(rdev->vce.vcpu_bo); + + radeon_bo_unreserve(rdev->vce.vcpu_bo); + + return 0; +} + +/** + * radeon_vce_idle_work_handler - power off VCE + * + * @work: pointer to work structure + * + * power of VCE when it's not used any more + */ +static void radeon_vce_idle_work_handler(struct work_struct *work) +{ + struct radeon_device *rdev = + container_of(work, struct radeon_device, vce.idle_work.work); + + if ((radeon_fence_count_emitted(rdev, TN_RING_TYPE_VCE1_INDEX) == 0) && + (radeon_fence_count_emitted(rdev, TN_RING_TYPE_VCE2_INDEX) == 0)) { + if ((rdev->pm.pm_method == PM_METHOD_DPM) && rdev->pm.dpm_enabled) { + radeon_dpm_enable_vce(rdev, false); + } else { + radeon_set_vce_clocks(rdev, 0, 0); + } + } else { + schedule_delayed_work(&rdev->vce.idle_work, + msecs_to_jiffies(VCE_IDLE_TIMEOUT_MS)); + } +} + +/** + * radeon_vce_note_usage - power up VCE + * + * @rdev: radeon_device pointer + * + * Make sure VCE is powerd up when we want to use it + */ +void radeon_vce_note_usage(struct radeon_device *rdev) +{ + bool streams_changed = false; + bool set_clocks = !cancel_delayed_work_sync(&rdev->vce.idle_work); + set_clocks &= schedule_delayed_work(&rdev->vce.idle_work, + msecs_to_jiffies(VCE_IDLE_TIMEOUT_MS)); + + if ((rdev->pm.pm_method == PM_METHOD_DPM) && rdev->pm.dpm_enabled) { + /* XXX figure out if the streams changed */ + streams_changed = false; + } + + if (set_clocks || streams_changed) { + if ((rdev->pm.pm_method == PM_METHOD_DPM) && rdev->pm.dpm_enabled) { + radeon_dpm_enable_vce(rdev, true); + } else { + radeon_set_vce_clocks(rdev, 53300, 40000); + } + } +} + +/** + * radeon_vce_free_handles - free still open VCE handles + * + * @rdev: radeon_device pointer + * @filp: drm file pointer + * + * Close all VCE handles still open by this file pointer + */ +void radeon_vce_free_handles(struct radeon_device *rdev, struct drm_file *filp) +{ + int i, r; + for (i = 0; i < RADEON_MAX_VCE_HANDLES; ++i) { + uint32_t handle = atomic_read(&rdev->vce.handles[i]); + if (!handle || rdev->vce.filp[i] != filp) + continue; + + radeon_vce_note_usage(rdev); + + r = radeon_vce_get_destroy_msg(rdev, TN_RING_TYPE_VCE1_INDEX, + handle, NULL); + if (r) + DRM_ERROR("Error destroying VCE handle (%d)!\n", r); + + rdev->vce.filp[i] = NULL; + atomic_set(&rdev->vce.handles[i], 0); + } +} + +/** + * radeon_vce_get_create_msg - generate a VCE create msg + * + * @rdev: radeon_device pointer + * @ring: ring we should submit the msg to + * @handle: VCE session handle to use + * @fence: optional fence to return + * + * Open up a stream for HW test + */ +int radeon_vce_get_create_msg(struct radeon_device *rdev, int ring, + uint32_t handle, struct radeon_fence **fence) +{ + const unsigned ib_size_dw = 1024; + struct radeon_ib ib; + uint64_t dummy; + int i, r; + + r = radeon_ib_get(rdev, ring, &ib, NULL, ib_size_dw * 4); + if (r) { + DRM_ERROR("radeon: failed to get ib (%d).\n", r); + return r; + } + + dummy = ib.gpu_addr + 1024; + + /* stitch together an VCE create msg */ + ib.length_dw = 0; + ib.ptr[ib.length_dw++] = 0x0000000c; /* len */ + ib.ptr[ib.length_dw++] = 0x00000001; /* session cmd */ + ib.ptr[ib.length_dw++] = handle; + + ib.ptr[ib.length_dw++] = 0x00000030; /* len */ + ib.ptr[ib.length_dw++] = 0x01000001; /* create cmd */ + ib.ptr[ib.length_dw++] = 0x00000000; + ib.ptr[ib.length_dw++] = 0x00000042; + ib.ptr[ib.length_dw++] = 0x0000000a; + ib.ptr[ib.length_dw++] = 0x00000001; + ib.ptr[ib.length_dw++] = 0x00000080; + ib.ptr[ib.length_dw++] = 0x00000060; + ib.ptr[ib.length_dw++] = 0x00000100; + ib.ptr[ib.length_dw++] = 0x00000100; + ib.ptr[ib.length_dw++] = 0x0000000c; + ib.ptr[ib.length_dw++] = 0x00000000; + + ib.ptr[ib.length_dw++] = 0x00000014; /* len */ + ib.ptr[ib.length_dw++] = 0x05000005; /* feedback buffer */ + ib.ptr[ib.length_dw++] = upper_32_bits(dummy); + ib.ptr[ib.length_dw++] = dummy; + ib.ptr[ib.length_dw++] = 0x00000001; + + for (i = ib.length_dw; i < ib_size_dw; ++i) + ib.ptr[i] = 0x0; + + r = radeon_ib_schedule(rdev, &ib, NULL); + if (r) { + DRM_ERROR("radeon: failed to schedule ib (%d).\n", r); + } + + if (fence) + *fence = radeon_fence_ref(ib.fence); + + radeon_ib_free(rdev, &ib); + + return r; +} + +/** + * radeon_vce_get_destroy_msg - generate a VCE destroy msg + * + * @rdev: radeon_device pointer + * @ring: ring we should submit the msg to + * @handle: VCE session handle to use + * @fence: optional fence to return + * + * Close up a stream for HW test or if userspace failed to do so + */ +int radeon_vce_get_destroy_msg(struct radeon_device *rdev, int ring, + uint32_t handle, struct radeon_fence **fence) +{ + const unsigned ib_size_dw = 1024; + struct radeon_ib ib; + uint64_t dummy; + int i, r; + + r = radeon_ib_get(rdev, ring, &ib, NULL, ib_size_dw * 4); + if (r) { + DRM_ERROR("radeon: failed to get ib (%d).\n", r); + return r; + } + + dummy = ib.gpu_addr + 1024; + + /* stitch together an VCE destroy msg */ + ib.length_dw = 0; + ib.ptr[ib.length_dw++] = 0x0000000c; /* len */ + ib.ptr[ib.length_dw++] = 0x00000001; /* session cmd */ + ib.ptr[ib.length_dw++] = handle; + + ib.ptr[ib.length_dw++] = 0x00000014; /* len */ + ib.ptr[ib.length_dw++] = 0x05000005; /* feedback buffer */ + ib.ptr[ib.length_dw++] = upper_32_bits(dummy); + ib.ptr[ib.length_dw++] = dummy; + ib.ptr[ib.length_dw++] = 0x00000001; + + ib.ptr[ib.length_dw++] = 0x00000008; /* len */ + ib.ptr[ib.length_dw++] = 0x02000001; /* destroy cmd */ + + for (i = ib.length_dw; i < ib_size_dw; ++i) + ib.ptr[i] = 0x0; + + r = radeon_ib_schedule(rdev, &ib, NULL); + if (r) { + DRM_ERROR("radeon: failed to schedule ib (%d).\n", r); + } + + if (fence) + *fence = radeon_fence_ref(ib.fence); + + radeon_ib_free(rdev, &ib); + + return r; +} + +/** + * radeon_vce_cs_reloc - command submission relocation + * + * @p: parser context + * @lo: address of lower dword + * @hi: address of higher dword + * + * Patch relocation inside command stream with real buffer address + */ +int radeon_vce_cs_reloc(struct radeon_cs_parser *p, int lo, int hi) +{ + struct radeon_cs_chunk *relocs_chunk; + uint64_t offset; + unsigned idx; + + relocs_chunk = &p->chunks[p->chunk_relocs_idx]; + offset = radeon_get_ib_value(p, lo); + idx = radeon_get_ib_value(p, hi); + + if (idx >= relocs_chunk->length_dw) { + DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", + idx, relocs_chunk->length_dw); + return -EINVAL; + } + + offset += p->relocs_ptr[(idx / 4)]->gpu_offset; + + p->ib.ptr[lo] = offset & 0xFFFFFFFF; + p->ib.ptr[hi] = offset >> 32; + + return 0; +} + +/** + * radeon_vce_cs_parse - parse and validate the command stream + * + * @p: parser context + * + */ +int radeon_vce_cs_parse(struct radeon_cs_parser *p) +{ + uint32_t handle = 0; + bool destroy = false; + int i, r; + + while (p->idx < p->chunks[p->chunk_ib_idx].length_dw) { + uint32_t len = radeon_get_ib_value(p, p->idx); + uint32_t cmd = radeon_get_ib_value(p, p->idx + 1); + + if ((len < 8) || (len & 3)) { + DRM_ERROR("invalid VCE command length (%d)!\n", len); + return -EINVAL; + } + + switch (cmd) { + case 0x00000001: // session + handle = radeon_get_ib_value(p, p->idx + 2); + break; + + case 0x00000002: // task info + case 0x01000001: // create + case 0x04000001: // config extension + case 0x04000002: // pic control + case 0x04000005: // rate control + case 0x04000007: // motion estimation + case 0x04000008: // rdo + break; + + case 0x03000001: // encode + r = radeon_vce_cs_reloc(p, p->idx + 10, p->idx + 9); + if (r) + return r; + + r = radeon_vce_cs_reloc(p, p->idx + 12, p->idx + 11); + if (r) + return r; + break; + + case 0x02000001: // destroy + destroy = true; + break; + + case 0x05000001: // context buffer + case 0x05000004: // video bitstream buffer + case 0x05000005: // feedback buffer + r = radeon_vce_cs_reloc(p, p->idx + 3, p->idx + 2); + if (r) + return r; + break; + + default: + DRM_ERROR("invalid VCE command (0x%x)!\n", cmd); + return -EINVAL; + } + + p->idx += len / 4; + } + + if (destroy) { + /* IB contains a destroy msg, free the handle */ + for (i = 0; i < RADEON_MAX_VCE_HANDLES; ++i) + atomic_cmpxchg(&p->rdev->vce.handles[i], handle, 0); + + return 0; + } + + /* create or encode, validate the handle */ + for (i = 0; i < RADEON_MAX_VCE_HANDLES; ++i) { + if (atomic_read(&p->rdev->vce.handles[i]) == handle) + return 0; + } + + /* handle not found try to alloc a new one */ + for (i = 0; i < RADEON_MAX_VCE_HANDLES; ++i) { + if (!atomic_cmpxchg(&p->rdev->vce.handles[i], 0, handle)) { + p->rdev->vce.filp[i] = p->filp; + return 0; + } + } + + DRM_ERROR("No more free VCE handles!\n"); + return -EINVAL; +} + +/** + * radeon_vce_semaphore_emit - emit a semaphore command + * + * @rdev: radeon_device pointer + * @ring: engine to use + * @semaphore: address of semaphore + * @emit_wait: true=emit wait, false=emit signal + * + */ +bool radeon_vce_semaphore_emit(struct radeon_device *rdev, + struct radeon_ring *ring, + struct radeon_semaphore *semaphore, + bool emit_wait) +{ + uint64_t addr = semaphore->gpu_addr; + + radeon_ring_write(ring, VCE_CMD_SEMAPHORE); + radeon_ring_write(ring, (addr >> 3) & 0x000FFFFF); + radeon_ring_write(ring, (addr >> 23) & 0x000FFFFF); + radeon_ring_write(ring, 0x01003000 | (emit_wait ? 1 : 0)); + if (!emit_wait) + radeon_ring_write(ring, VCE_CMD_END); + + return true; +} + +/** + * radeon_vce_ib_execute - execute indirect buffer + * + * @rdev: radeon_device pointer + * @ib: the IB to execute + * + */ +void radeon_vce_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) +{ + struct radeon_ring *ring = &rdev->ring[ib->ring]; + radeon_ring_write(ring, VCE_CMD_IB); + radeon_ring_write(ring, ib->gpu_addr); + radeon_ring_write(ring, upper_32_bits(ib->gpu_addr)); + radeon_ring_write(ring, ib->length_dw); +} + +/** + * radeon_vce_fence_emit - add a fence command to the ring + * + * @rdev: radeon_device pointer + * @fence: the fence + * + */ +void radeon_vce_fence_emit(struct radeon_device *rdev, + struct radeon_fence *fence) +{ + struct radeon_ring *ring = &rdev->ring[fence->ring]; + uint32_t addr = rdev->fence_drv[fence->ring].gpu_addr; + + radeon_ring_write(ring, VCE_CMD_FENCE); + radeon_ring_write(ring, addr); + radeon_ring_write(ring, upper_32_bits(addr)); + radeon_ring_write(ring, fence->seq); + radeon_ring_write(ring, VCE_CMD_TRAP); + radeon_ring_write(ring, VCE_CMD_END); +} + +/** + * radeon_vce_ring_test - test if VCE ring is working + * + * @rdev: radeon_device pointer + * @ring: the engine to test on + * + */ +int radeon_vce_ring_test(struct radeon_device *rdev, struct radeon_ring *ring) +{ + uint32_t rptr = vce_v1_0_get_rptr(rdev, ring); + unsigned i; + int r; + + r = radeon_ring_lock(rdev, ring, 16); + if (r) { + DRM_ERROR("radeon: vce failed to lock ring %d (%d).\n", + ring->idx, r); + return r; + } + radeon_ring_write(ring, VCE_CMD_END); + radeon_ring_unlock_commit(rdev, ring); + + for (i = 0; i < rdev->usec_timeout; i++) { + if (vce_v1_0_get_rptr(rdev, ring) != rptr) + break; + DRM_UDELAY(1); + } + + if (i < rdev->usec_timeout) { + DRM_INFO("ring test on %d succeeded in %d usecs\n", + ring->idx, i); + } else { + DRM_ERROR("radeon: ring %d test failed\n", + ring->idx); + r = -ETIMEDOUT; + } + + return r; +} + +/** + * radeon_vce_ib_test - test if VCE IBs are working + * + * @rdev: radeon_device pointer + * @ring: the engine to test on + * + */ +int radeon_vce_ib_test(struct radeon_device *rdev, struct radeon_ring *ring) +{ + struct radeon_fence *fence = NULL; + int r; + + r = radeon_vce_get_create_msg(rdev, ring->idx, 1, NULL); + if (r) { + DRM_ERROR("radeon: failed to get create msg (%d).\n", r); + goto error; + } + + r = radeon_vce_get_destroy_msg(rdev, ring->idx, 1, &fence); + if (r) { + DRM_ERROR("radeon: failed to get destroy ib (%d).\n", r); + goto error; + } + + r = radeon_fence_wait(fence, false); + if (r) { + DRM_ERROR("radeon: fence wait failed (%d).\n", r); + } else { + DRM_INFO("ib test on ring %d succeeded\n", ring->idx); + } +error: + radeon_fence_unref(&fence); + return r; +} diff --git a/drivers/gpu/drm/radeon/radeon_vm.c b/drivers/gpu/drm/radeon/radeon_vm.c new file mode 100644 index 000000000000..2aae6ce49d32 --- /dev/null +++ b/drivers/gpu/drm/radeon/radeon_vm.c @@ -0,0 +1,966 @@ +/* + * Copyright 2008 Advanced Micro Devices, Inc. + * Copyright 2008 Red Hat Inc. + * Copyright 2009 Jerome Glisse. + * + * 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. + * + * Authors: Dave Airlie + * Alex Deucher + * Jerome Glisse + */ +#include <drm/drmP.h> +#include <drm/radeon_drm.h> +#include "radeon.h" +#include "radeon_trace.h" + +/* + * GPUVM + * GPUVM is similar to the legacy gart on older asics, however + * rather than there being a single global gart table + * for the entire GPU, there are multiple VM page tables active + * at any given time. The VM page tables can contain a mix + * vram pages and system memory pages and system memory pages + * can be mapped as snooped (cached system pages) or unsnooped + * (uncached system pages). + * Each VM has an ID associated with it and there is a page table + * associated with each VMID. When execting a command buffer, + * the kernel tells the the ring what VMID to use for that command + * buffer. VMIDs are allocated dynamically as commands are submitted. + * The userspace drivers maintain their own address space and the kernel + * sets up their pages tables accordingly when they submit their + * command buffers and a VMID is assigned. + * Cayman/Trinity support up to 8 active VMs at any given time; + * SI supports 16. + */ + +/** + * radeon_vm_num_pde - return the number of page directory entries + * + * @rdev: radeon_device pointer + * + * Calculate the number of page directory entries (cayman+). + */ +static unsigned radeon_vm_num_pdes(struct radeon_device *rdev) +{ + return rdev->vm_manager.max_pfn >> RADEON_VM_BLOCK_SIZE; +} + +/** + * radeon_vm_directory_size - returns the size of the page directory in bytes + * + * @rdev: radeon_device pointer + * + * Calculate the size of the page directory in bytes (cayman+). + */ +static unsigned radeon_vm_directory_size(struct radeon_device *rdev) +{ + return RADEON_GPU_PAGE_ALIGN(radeon_vm_num_pdes(rdev) * 8); +} + +/** + * radeon_vm_manager_init - init the vm manager + * + * @rdev: radeon_device pointer + * + * Init the vm manager (cayman+). + * Returns 0 for success, error for failure. + */ +int radeon_vm_manager_init(struct radeon_device *rdev) +{ + int r; + + if (!rdev->vm_manager.enabled) { + r = radeon_asic_vm_init(rdev); + if (r) + return r; + + rdev->vm_manager.enabled = true; + } + return 0; +} + +/** + * radeon_vm_manager_fini - tear down the vm manager + * + * @rdev: radeon_device pointer + * + * Tear down the VM manager (cayman+). + */ +void radeon_vm_manager_fini(struct radeon_device *rdev) +{ + int i; + + if (!rdev->vm_manager.enabled) + return; + + for (i = 0; i < RADEON_NUM_VM; ++i) + radeon_fence_unref(&rdev->vm_manager.active[i]); + radeon_asic_vm_fini(rdev); + rdev->vm_manager.enabled = false; +} + +/** + * radeon_vm_get_bos - add the vm BOs to a validation list + * + * @vm: vm providing the BOs + * @head: head of validation list + * + * 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_vm *vm, + struct list_head *head) +{ + struct radeon_cs_reloc *list; + unsigned i, idx, size; + + size = (radeon_vm_num_pdes(rdev) + 1) * sizeof(struct radeon_cs_reloc); + list = kmalloc(size, GFP_KERNEL); + if (!list) + return NULL; + + /* add the vm page table to the list */ + list[0].gobj = NULL; + list[0].robj = vm->page_directory; + list[0].domain = RADEON_GEM_DOMAIN_VRAM; + list[0].alt_domain = RADEON_GEM_DOMAIN_VRAM; + list[0].tv.bo = &vm->page_directory->tbo; + 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].domain = RADEON_GEM_DOMAIN_VRAM; + list[idx].alt_domain = RADEON_GEM_DOMAIN_VRAM; + list[idx].tv.bo = &list[idx].robj->tbo; + list[idx].tiling_flags = 0; + list[idx].handle = 0; + list_add(&list[idx++].tv.head, head); + } + + return list; +} + +/** + * radeon_vm_grab_id - allocate the next free VMID + * + * @rdev: radeon_device pointer + * @vm: vm to allocate id for + * @ring: ring we want to submit job to + * + * Allocate an id for the vm (cayman+). + * Returns the fence we need to sync to (if any). + * + * Global and local mutex must be locked! + */ +struct radeon_fence *radeon_vm_grab_id(struct radeon_device *rdev, + struct radeon_vm *vm, int ring) +{ + struct radeon_fence *best[RADEON_NUM_RINGS] = {}; + 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]) + return NULL; + + /* we definately need to flush */ + radeon_fence_unref(&vm->last_flush); + + /* skip over VMID 0, since it is the system VM */ + for (i = 1; i < rdev->vm_manager.nvm; ++i) { + struct radeon_fence *fence = rdev->vm_manager.active[i]; + + if (fence == NULL) { + /* found a free one */ + vm->id = i; + trace_radeon_vm_grab_id(vm->id, ring); + return NULL; + } + + if (radeon_fence_is_earlier(fence, best[fence->ring])) { + best[fence->ring] = fence; + choices[fence->ring == ring ? 0 : 1] = i; + } + } + + for (i = 0; i < 2; ++i) { + if (choices[i]) { + vm->id = choices[i]; + trace_radeon_vm_grab_id(vm->id, ring); + return rdev->vm_manager.active[choices[i]]; + } + } + + /* should never happen */ + BUG(); + return NULL; +} + +/** + * radeon_vm_flush - hardware flush the vm + * + * @rdev: radeon_device pointer + * @vm: vm we want to flush + * @ring: ring to use for flush + * + * Flush the vm (cayman+). + * + * Global and local mutex must be locked! + */ +void radeon_vm_flush(struct radeon_device *rdev, + struct radeon_vm *vm, + int ring) +{ + uint64_t pd_addr = radeon_bo_gpu_offset(vm->page_directory); + + /* if we can't remember our last VM flush then flush now! */ + /* XXX figure out why we have to flush all the time */ + if (!vm->last_flush || true || pd_addr != vm->pd_gpu_addr) { + vm->pd_gpu_addr = pd_addr; + radeon_ring_vm_flush(rdev, ring, vm); + } +} + +/** + * radeon_vm_fence - remember fence for vm + * + * @rdev: radeon_device pointer + * @vm: vm we want to fence + * @fence: fence to remember + * + * Fence the vm (cayman+). + * Set the fence used to protect page table and id. + * + * Global and local mutex must be locked! + */ +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); + + radeon_fence_unref(&vm->last_id_use); + vm->last_id_use = radeon_fence_ref(fence); + + /* we just flushed the VM, remember that */ + if (!vm->last_flush) + vm->last_flush = radeon_fence_ref(fence); +} + +/** + * radeon_vm_bo_find - find the bo_va for a specific vm & bo + * + * @vm: requested vm + * @bo: requested buffer object + * + * Find @bo inside the requested vm (cayman+). + * Search inside the @bos vm list for the requested vm + * Returns the found bo_va or NULL if none is found + * + * Object has to be reserved! + */ +struct radeon_bo_va *radeon_vm_bo_find(struct radeon_vm *vm, + struct radeon_bo *bo) +{ + struct radeon_bo_va *bo_va; + + list_for_each_entry(bo_va, &bo->va, bo_list) { + if (bo_va->vm == vm) { + return bo_va; + } + } + return NULL; +} + +/** + * radeon_vm_bo_add - add a bo to a specific vm + * + * @rdev: radeon_device pointer + * @vm: requested vm + * @bo: radeon buffer object + * + * Add @bo into the requested vm (cayman+). + * Add @bo to the list of bos associated with the vm + * Returns newly added bo_va or NULL for failure + * + * Object has to be reserved! + */ +struct radeon_bo_va *radeon_vm_bo_add(struct radeon_device *rdev, + struct radeon_vm *vm, + struct radeon_bo *bo) +{ + struct radeon_bo_va *bo_va; + + bo_va = kzalloc(sizeof(struct radeon_bo_va), GFP_KERNEL); + if (bo_va == NULL) { + return NULL; + } + bo_va->vm = vm; + bo_va->bo = bo; + bo_va->soffset = 0; + bo_va->eoffset = 0; + bo_va->flags = 0; + bo_va->valid = false; + bo_va->ref_count = 1; + INIT_LIST_HEAD(&bo_va->bo_list); + INIT_LIST_HEAD(&bo_va->vm_list); + + mutex_lock(&vm->mutex); + list_add(&bo_va->vm_list, &vm->va); + list_add_tail(&bo_va->bo_list, &bo->va); + mutex_unlock(&vm->mutex); + + return bo_va; +} + +/** + * radeon_vm_clear_bo - initially clear the page dir/table + * + * @rdev: radeon_device pointer + * @bo: bo to clear + */ +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; + + INIT_LIST_HEAD(&head); + list_add(&tv.head, &head); + + r = ttm_eu_reserve_buffers(&ticket, &head); + if (r) + return r; + + r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false); + if (r) + goto error; + + addr = radeon_bo_gpu_offset(bo); + entries = radeon_bo_size(bo) / 8; + + r = radeon_ib_get(rdev, R600_RING_TYPE_DMA_INDEX, &ib, + NULL, entries * 2 + 64); + if (r) + goto error; + + ib.length_dw = 0; + + radeon_asic_vm_set_page(rdev, &ib, addr, 0, entries, 0, 0); + + r = radeon_ib_schedule(rdev, &ib, NULL); + if (r) + goto error; + + ttm_eu_fence_buffer_objects(&ticket, &head, ib.fence); + radeon_ib_free(rdev, &ib); + + return 0; + +error: + ttm_eu_backoff_reservation(&ticket, &head); + return r; +} + +/** + * radeon_vm_bo_set_addr - set bos virtual address inside a vm + * + * @rdev: radeon_device pointer + * @bo_va: bo_va to store the address + * @soffset: requested offset of the buffer in the VM address space + * @flags: attributes of pages (read/write/valid/etc.) + * + * Set offset of @bo_va (cayman+). + * Validate and set the offset requested within the vm address space. + * Returns 0 for success, error for failure. + * + * Object has to be reserved! + */ +int radeon_vm_bo_set_addr(struct radeon_device *rdev, + struct radeon_bo_va *bo_va, + uint64_t soffset, + uint32_t flags) +{ + uint64_t size = radeon_bo_size(bo_va->bo); + uint64_t eoffset, last_offset = 0; + struct radeon_vm *vm = bo_va->vm; + struct radeon_bo_va *tmp; + struct list_head *head; + unsigned last_pfn, pt_idx; + int r; + + if (soffset) { + /* make sure object fit at this offset */ + eoffset = soffset + size; + if (soffset >= eoffset) { + return -EINVAL; + } + + last_pfn = eoffset / RADEON_GPU_PAGE_SIZE; + if (last_pfn > rdev->vm_manager.max_pfn) { + dev_err(rdev->dev, "va above limit (0x%08X > 0x%08X)\n", + last_pfn, rdev->vm_manager.max_pfn); + return -EINVAL; + } + + } else { + eoffset = last_pfn = 0; + } + + mutex_lock(&vm->mutex); + head = &vm->va; + last_offset = 0; + list_for_each_entry(tmp, &vm->va, vm_list) { + if (bo_va == tmp) { + /* skip over currently modified bo */ + continue; + } + + if (soffset >= last_offset && eoffset <= tmp->soffset) { + /* bo can be added before this one */ + break; + } + if (eoffset > tmp->soffset && soffset < tmp->eoffset) { + /* bo and tmp overlap, invalid offset */ + dev_err(rdev->dev, "bo %p va 0x%08X conflict with (bo %p 0x%08X 0x%08X)\n", + bo_va->bo, (unsigned)bo_va->soffset, tmp->bo, + (unsigned)tmp->soffset, (unsigned)tmp->eoffset); + mutex_unlock(&vm->mutex); + return -EINVAL; + } + last_offset = tmp->eoffset; + head = &tmp->vm_list; + } + + bo_va->soffset = soffset; + bo_va->eoffset = eoffset; + bo_va->flags = flags; + bo_va->valid = false; + list_move(&bo_va->vm_list, head); + + soffset = (soffset / RADEON_GPU_PAGE_SIZE) >> RADEON_VM_BLOCK_SIZE; + eoffset = (eoffset / RADEON_GPU_PAGE_SIZE) >> RADEON_VM_BLOCK_SIZE; + + if (eoffset > vm->max_pde_used) + vm->max_pde_used = eoffset; + + radeon_bo_unreserve(bo_va->bo); + + /* walk over the address space and allocate the page tables */ + for (pt_idx = soffset; pt_idx <= eoffset; ++pt_idx) { + struct radeon_bo *pt; + + if (vm->page_tables[pt_idx].bo) + continue; + + /* drop mutex to allocate and clear page table */ + mutex_unlock(&vm->mutex); + + r = radeon_bo_create(rdev, RADEON_VM_PTE_COUNT * 8, + RADEON_GPU_PAGE_SIZE, false, + RADEON_GEM_DOMAIN_VRAM, NULL, &pt); + if (r) + return r; + + r = radeon_vm_clear_bo(rdev, pt); + if (r) { + radeon_bo_unref(&pt); + radeon_bo_reserve(bo_va->bo, false); + return r; + } + + /* aquire mutex again */ + mutex_lock(&vm->mutex); + if (vm->page_tables[pt_idx].bo) { + /* someone else allocated the pt in the meantime */ + mutex_unlock(&vm->mutex); + radeon_bo_unref(&pt); + mutex_lock(&vm->mutex); + continue; + } + + vm->page_tables[pt_idx].addr = 0; + vm->page_tables[pt_idx].bo = pt; + } + + mutex_unlock(&vm->mutex); + return radeon_bo_reserve(bo_va->bo, false); +} + +/** + * radeon_vm_map_gart - get the physical address of a gart page + * + * @rdev: radeon_device pointer + * @addr: the unmapped addr + * + * Look up the physical address of the page that the pte resolves + * to (cayman+). + * Returns the physical address of the page. + */ +uint64_t radeon_vm_map_gart(struct radeon_device *rdev, uint64_t addr) +{ + uint64_t result; + + /* page table offset */ + result = rdev->gart.pages_addr[addr >> PAGE_SHIFT]; + + /* in case cpu page size != gpu page size*/ + result |= addr & (~PAGE_MASK); + + return result; +} + +/** + * radeon_vm_page_flags - translate page flags to what the hw uses + * + * @flags: flags comming from userspace + * + * Translate the flags the userspace ABI uses to hw flags. + */ +static uint32_t radeon_vm_page_flags(uint32_t flags) +{ + uint32_t hw_flags = 0; + hw_flags |= (flags & RADEON_VM_PAGE_VALID) ? R600_PTE_VALID : 0; + hw_flags |= (flags & RADEON_VM_PAGE_READABLE) ? R600_PTE_READABLE : 0; + hw_flags |= (flags & RADEON_VM_PAGE_WRITEABLE) ? R600_PTE_WRITEABLE : 0; + if (flags & RADEON_VM_PAGE_SYSTEM) { + hw_flags |= R600_PTE_SYSTEM; + hw_flags |= (flags & RADEON_VM_PAGE_SNOOPED) ? R600_PTE_SNOOPED : 0; + } + return hw_flags; +} + +/** + * radeon_vm_update_pdes - make sure that page directory is valid + * + * @rdev: radeon_device pointer + * @vm: requested vm + * @start: start of GPU address range + * @end: end of GPU address range + * + * Allocates new page tables if necessary + * and updates the page directory (cayman+). + * Returns 0 for success, error for failure. + * + * Global and local mutex must be locked! + */ +int radeon_vm_update_page_directory(struct radeon_device *rdev, + struct radeon_vm *vm) +{ + static const uint32_t incr = RADEON_VM_PTE_COUNT * 8; + + uint64_t pd_addr = radeon_bo_gpu_offset(vm->page_directory); + uint64_t last_pde = ~0, last_pt = ~0; + unsigned count = 0, pt_idx, ndw; + struct radeon_ib ib; + int r; + + /* padding, etc. */ + ndw = 64; + + /* assume the worst case */ + ndw += vm->max_pde_used * 12; + + /* update too big for an IB */ + if (ndw > 0xfffff) + return -ENOMEM; + + r = radeon_ib_get(rdev, R600_RING_TYPE_DMA_INDEX, &ib, NULL, ndw * 4); + if (r) + return r; + ib.length_dw = 0; + + /* walk over the address space and update the page directory */ + for (pt_idx = 0; pt_idx <= vm->max_pde_used; ++pt_idx) { + struct radeon_bo *bo = vm->page_tables[pt_idx].bo; + uint64_t pde, pt; + + if (bo == NULL) + continue; + + pt = radeon_bo_gpu_offset(bo); + if (vm->page_tables[pt_idx].addr == pt) + continue; + vm->page_tables[pt_idx].addr = pt; + + pde = pd_addr + pt_idx * 8; + if (((last_pde + 8 * count) != pde) || + ((last_pt + incr * count) != pt)) { + + if (count) { + radeon_asic_vm_set_page(rdev, &ib, last_pde, + last_pt, count, incr, + R600_PTE_VALID); + } + + count = 1; + last_pde = pde; + last_pt = pt; + } else { + ++count; + } + } + + if (count) + radeon_asic_vm_set_page(rdev, &ib, last_pde, last_pt, count, + incr, R600_PTE_VALID); + + if (ib.length_dw != 0) { + radeon_semaphore_sync_to(ib.semaphore, vm->last_id_use); + r = radeon_ib_schedule(rdev, &ib, NULL); + 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); + } + radeon_ib_free(rdev, &ib); + + return 0; +} + +/** + * radeon_vm_update_ptes - make sure that page tables are valid + * + * @rdev: radeon_device pointer + * @vm: requested vm + * @start: start of GPU address range + * @end: end of GPU address range + * @dst: destination address to map to + * @flags: mapping flags + * + * Update the page tables in the range @start - @end (cayman+). + * + * 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 const uint64_t mask = RADEON_VM_PTE_COUNT - 1; + + uint64_t last_pte = ~0, last_dst = ~0; + unsigned count = 0; + uint64_t addr; + + start = start / RADEON_GPU_PAGE_SIZE; + end = end / RADEON_GPU_PAGE_SIZE; + + /* walk over the address space and update the page tables */ + for (addr = start; addr < end; ) { + uint64_t pt_idx = addr >> RADEON_VM_BLOCK_SIZE; + unsigned nptes; + uint64_t pte; + + if ((addr & ~mask) == (end & ~mask)) + nptes = end - addr; + else + nptes = RADEON_VM_PTE_COUNT - (addr & mask); + + pte = radeon_bo_gpu_offset(vm->page_tables[pt_idx].bo); + pte += (addr & mask) * 8; + + if ((last_pte + 8 * count) != pte) { + + if (count) { + radeon_asic_vm_set_page(rdev, ib, last_pte, + last_dst, count, + RADEON_GPU_PAGE_SIZE, + flags); + } + + count = nptes; + last_pte = pte; + last_dst = dst; + } else { + count += nptes; + } + + addr += nptes; + dst += nptes * RADEON_GPU_PAGE_SIZE; + } + + if (count) { + radeon_asic_vm_set_page(rdev, ib, last_pte, + last_dst, count, + RADEON_GPU_PAGE_SIZE, flags); + } +} + +/** + * radeon_vm_bo_update - map a bo into the vm page table + * + * @rdev: radeon_device pointer + * @vm: requested vm + * @bo: radeon buffer object + * @mem: ttm mem + * + * Fill in the page table entries for @bo (cayman+). + * Returns 0 for success, -EINVAL for failure. + * + * Object have to be reserved and mutex must be locked! + */ +int radeon_vm_bo_update(struct radeon_device *rdev, + struct radeon_vm *vm, + struct radeon_bo *bo, + struct ttm_mem_reg *mem) +{ + struct radeon_ib ib; + struct radeon_bo_va *bo_va; + unsigned nptes, ndw; + uint64_t addr; + int r; + + bo_va = radeon_vm_bo_find(vm, bo); + if (bo_va == NULL) { + dev_err(rdev->dev, "bo %p not in vm %p\n", bo, vm); + return -EINVAL; + } + + if (!bo_va->soffset) { + dev_err(rdev->dev, "bo %p don't has a mapping in vm %p\n", + bo, vm); + return -EINVAL; + } + + if ((bo_va->valid && mem) || (!bo_va->valid && mem == NULL)) + return 0; + + bo_va->flags &= ~RADEON_VM_PAGE_VALID; + bo_va->flags &= ~RADEON_VM_PAGE_SYSTEM; + if (mem) { + addr = mem->start << PAGE_SHIFT; + if (mem->mem_type != TTM_PL_SYSTEM) { + bo_va->flags |= RADEON_VM_PAGE_VALID; + bo_va->valid = true; + } + if (mem->mem_type == TTM_PL_TT) { + bo_va->flags |= RADEON_VM_PAGE_SYSTEM; + } else { + addr += rdev->vm_manager.vram_base_offset; + } + } else { + addr = 0; + bo_va->valid = false; + } + + trace_radeon_vm_bo_update(bo_va); + + nptes = radeon_bo_ngpu_pages(bo); + + /* padding, etc. */ + ndw = 64; + + if (RADEON_VM_BLOCK_SIZE > 11) + /* reserve space for one header for every 2k dwords */ + ndw += (nptes >> 11) * 4; + else + /* reserve space for one header for + every (1 << BLOCK_SIZE) entries */ + ndw += (nptes >> RADEON_VM_BLOCK_SIZE) * 4; + + /* reserve space for pte addresses */ + ndw += nptes * 2; + + /* update too big for an IB */ + if (ndw > 0xfffff) + return -ENOMEM; + + r = radeon_ib_get(rdev, R600_RING_TYPE_DMA_INDEX, &ib, NULL, ndw * 4); + if (r) + return r; + ib.length_dw = 0; + + radeon_vm_update_ptes(rdev, vm, &ib, bo_va->soffset, bo_va->eoffset, + addr, radeon_vm_page_flags(bo_va->flags)); + + radeon_semaphore_sync_to(ib.semaphore, vm->fence); + r = radeon_ib_schedule(rdev, &ib, NULL); + if (r) { + radeon_ib_free(rdev, &ib); + return r; + } + radeon_fence_unref(&vm->fence); + vm->fence = radeon_fence_ref(ib.fence); + radeon_ib_free(rdev, &ib); + radeon_fence_unref(&vm->last_flush); + + return 0; +} + +/** + * radeon_vm_bo_rmv - remove a bo to a specific vm + * + * @rdev: radeon_device pointer + * @bo_va: requested bo_va + * + * Remove @bo_va->bo from the requested vm (cayman+). + * Remove @bo_va->bo from the list of bos associated with the bo_va->vm and + * remove the ptes for @bo_va in the page table. + * Returns 0 for success. + * + * Object have to be reserved! + */ +int radeon_vm_bo_rmv(struct radeon_device *rdev, + struct radeon_bo_va *bo_va) +{ + int r = 0; + + mutex_lock(&bo_va->vm->mutex); + if (bo_va->soffset) + r = radeon_vm_bo_update(rdev, bo_va->vm, bo_va->bo, NULL); + + list_del(&bo_va->vm_list); + mutex_unlock(&bo_va->vm->mutex); + list_del(&bo_va->bo_list); + + kfree(bo_va); + return r; +} + +/** + * radeon_vm_bo_invalidate - mark the bo as invalid + * + * @rdev: radeon_device pointer + * @vm: requested vm + * @bo: radeon buffer object + * + * Mark @bo as invalid (cayman+). + */ +void radeon_vm_bo_invalidate(struct radeon_device *rdev, + struct radeon_bo *bo) +{ + struct radeon_bo_va *bo_va; + + list_for_each_entry(bo_va, &bo->va, bo_list) { + bo_va->valid = false; + } +} + +/** + * radeon_vm_init - initialize a vm instance + * + * @rdev: radeon_device pointer + * @vm: requested vm + * + * Init @vm fields (cayman+). + */ +int radeon_vm_init(struct radeon_device *rdev, struct radeon_vm *vm) +{ + unsigned pd_size, pd_entries, pts_size; + int r; + + vm->id = 0; + vm->fence = NULL; + vm->last_flush = NULL; + vm->last_id_use = NULL; + mutex_init(&vm->mutex); + INIT_LIST_HEAD(&vm->va); + + pd_size = radeon_vm_directory_size(rdev); + pd_entries = radeon_vm_num_pdes(rdev); + + /* allocate page table array */ + pts_size = pd_entries * sizeof(struct radeon_vm_pt); + vm->page_tables = kzalloc(pts_size, GFP_KERNEL); + if (vm->page_tables == NULL) { + DRM_ERROR("Cannot allocate memory for page table array\n"); + return -ENOMEM; + } + + r = radeon_bo_create(rdev, pd_size, RADEON_VM_PTB_ALIGN_SIZE, false, + RADEON_GEM_DOMAIN_VRAM, NULL, + &vm->page_directory); + if (r) + return r; + + r = radeon_vm_clear_bo(rdev, vm->page_directory); + if (r) { + radeon_bo_unref(&vm->page_directory); + vm->page_directory = NULL; + return r; + } + + return 0; +} + +/** + * radeon_vm_fini - tear down a vm instance + * + * @rdev: radeon_device pointer + * @vm: requested vm + * + * Tear down @vm (cayman+). + * Unbind the VM and remove all bos from the vm bo list + */ +void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm) +{ + struct radeon_bo_va *bo_va, *tmp; + int i, r; + + if (!list_empty(&vm->va)) { + dev_err(rdev->dev, "still active bo inside vm\n"); + } + list_for_each_entry_safe(bo_va, tmp, &vm->va, vm_list) { + list_del_init(&bo_va->vm_list); + r = radeon_bo_reserve(bo_va->bo, false); + if (!r) { + list_del_init(&bo_va->bo_list); + radeon_bo_unreserve(bo_va->bo); + kfree(bo_va); + } + } + + + for (i = 0; i < radeon_vm_num_pdes(rdev); i++) + radeon_bo_unref(&vm->page_tables[i].bo); + kfree(vm->page_tables); + + radeon_bo_unref(&vm->page_directory); + + radeon_fence_unref(&vm->fence); + radeon_fence_unref(&vm->last_flush); + radeon_fence_unref(&vm->last_id_use); + + mutex_destroy(&vm->mutex); +} diff --git a/drivers/gpu/drm/radeon/rs780_dpm.c b/drivers/gpu/drm/radeon/rs780_dpm.c index 8512085b0aef..02f7710de470 100644 --- a/drivers/gpu/drm/radeon/rs780_dpm.c +++ b/drivers/gpu/drm/radeon/rs780_dpm.c @@ -807,9 +807,6 @@ static int rs780_parse_power_table(struct radeon_device *rdev) power_info->pplib.ucNumStates, GFP_KERNEL); if (!rdev->pm.dpm.ps) return -ENOMEM; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < power_info->pplib.ucNumStates; i++) { power_state = (union pplib_power_state *) @@ -859,6 +856,10 @@ int rs780_dpm_init(struct radeon_device *rdev) return -ENOMEM; rdev->pm.dpm.priv = pi; + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = rs780_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/rv6xx_dpm.c b/drivers/gpu/drm/radeon/rv6xx_dpm.c index bebf31c4d841..e7045b085715 100644 --- a/drivers/gpu/drm/radeon/rv6xx_dpm.c +++ b/drivers/gpu/drm/radeon/rv6xx_dpm.c @@ -1891,9 +1891,6 @@ static int rv6xx_parse_power_table(struct radeon_device *rdev) power_info->pplib.ucNumStates, GFP_KERNEL); if (!rdev->pm.dpm.ps) return -ENOMEM; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < power_info->pplib.ucNumStates; i++) { power_state = (union pplib_power_state *) @@ -1943,6 +1940,10 @@ int rv6xx_dpm_init(struct radeon_device *rdev) return -ENOMEM; rdev->pm.dpm.priv = pi; + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = rv6xx_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/rv770_dpm.c b/drivers/gpu/drm/radeon/rv770_dpm.c index b5f63f5e22a3..da041a43d82e 100644 --- a/drivers/gpu/drm/radeon/rv770_dpm.c +++ b/drivers/gpu/drm/radeon/rv770_dpm.c @@ -2281,9 +2281,6 @@ int rv7xx_parse_power_table(struct radeon_device *rdev) power_info->pplib.ucNumStates, GFP_KERNEL); if (!rdev->pm.dpm.ps) return -ENOMEM; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < power_info->pplib.ucNumStates; i++) { power_state = (union pplib_power_state *) @@ -2361,6 +2358,10 @@ int rv770_dpm_init(struct radeon_device *rdev) pi->min_vddc_in_table = 0; pi->max_vddc_in_table = 0; + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = rv7xx_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/si.c b/drivers/gpu/drm/radeon/si.c index 9a124d0608b3..d589475fe9e6 100644 --- a/drivers/gpu/drm/radeon/si.c +++ b/drivers/gpu/drm/radeon/si.c @@ -3434,8 +3434,6 @@ static int si_cp_resume(struct radeon_device *rdev) WREG32(CP_RB0_BASE, ring->gpu_addr >> 8); - ring->rptr = RREG32(CP_RB0_RPTR); - /* ring1 - compute only */ /* Set ring buffer size */ ring = &rdev->ring[CAYMAN_RING_TYPE_CP1_INDEX]; @@ -3460,8 +3458,6 @@ static int si_cp_resume(struct radeon_device *rdev) WREG32(CP_RB1_BASE, ring->gpu_addr >> 8); - ring->rptr = RREG32(CP_RB1_RPTR); - /* ring2 - compute only */ /* Set ring buffer size */ ring = &rdev->ring[CAYMAN_RING_TYPE_CP2_INDEX]; @@ -3486,8 +3482,6 @@ static int si_cp_resume(struct radeon_device *rdev) WREG32(CP_RB2_BASE, ring->gpu_addr >> 8); - ring->rptr = RREG32(CP_RB2_RPTR); - /* start the rings */ si_cp_start(rdev); rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready = true; @@ -3872,11 +3866,9 @@ bool si_gfx_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) if (!(reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE | RADEON_RESET_CP))) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force CP activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } diff --git a/drivers/gpu/drm/radeon/si_dma.c b/drivers/gpu/drm/radeon/si_dma.c index 59be2cfcbb47..cf0fdad8c278 100644 --- a/drivers/gpu/drm/radeon/si_dma.c +++ b/drivers/gpu/drm/radeon/si_dma.c @@ -49,11 +49,9 @@ bool si_dma_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) mask = RADEON_RESET_DMA1; if (!(reset_mask & mask)) { - radeon_ring_lockup_update(ring); + radeon_ring_lockup_update(rdev, ring); return false; } - /* force ring activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } diff --git a/drivers/gpu/drm/radeon/si_dpm.c b/drivers/gpu/drm/radeon/si_dpm.c index 0a2f5b4bca43..9a3567bedaae 100644 --- a/drivers/gpu/drm/radeon/si_dpm.c +++ b/drivers/gpu/drm/radeon/si_dpm.c @@ -6271,9 +6271,6 @@ static int si_parse_power_table(struct radeon_device *rdev) if (!rdev->pm.dpm.ps) return -ENOMEM; power_state_offset = (u8 *)state_array->states; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < state_array->ucNumEntries; i++) { u8 *idx; power_state = (union pplib_power_state *)power_state_offset; @@ -6350,6 +6347,10 @@ int si_dpm_init(struct radeon_device *rdev) pi->min_vddc_in_table = 0; pi->max_vddc_in_table = 0; + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = si_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/sid.h b/drivers/gpu/drm/radeon/sid.h index 9239a6d29128..683532f84931 100644 --- a/drivers/gpu/drm/radeon/sid.h +++ b/drivers/gpu/drm/radeon/sid.h @@ -1798,4 +1798,51 @@ #define DMA_PACKET_CONSTANT_FILL 0xd #define DMA_PACKET_NOP 0xf +#define VCE_STATUS 0x20004 +#define VCE_VCPU_CNTL 0x20014 +#define VCE_CLK_EN (1 << 0) +#define VCE_VCPU_CACHE_OFFSET0 0x20024 +#define VCE_VCPU_CACHE_SIZE0 0x20028 +#define VCE_VCPU_CACHE_OFFSET1 0x2002c +#define VCE_VCPU_CACHE_SIZE1 0x20030 +#define VCE_VCPU_CACHE_OFFSET2 0x20034 +#define VCE_VCPU_CACHE_SIZE2 0x20038 +#define VCE_SOFT_RESET 0x20120 +#define VCE_ECPU_SOFT_RESET (1 << 0) +#define VCE_FME_SOFT_RESET (1 << 2) +#define VCE_RB_BASE_LO2 0x2016c +#define VCE_RB_BASE_HI2 0x20170 +#define VCE_RB_SIZE2 0x20174 +#define VCE_RB_RPTR2 0x20178 +#define VCE_RB_WPTR2 0x2017c +#define VCE_RB_BASE_LO 0x20180 +#define VCE_RB_BASE_HI 0x20184 +#define VCE_RB_SIZE 0x20188 +#define VCE_RB_RPTR 0x2018c +#define VCE_RB_WPTR 0x20190 +#define VCE_CLOCK_GATING_A 0x202f8 +#define VCE_CLOCK_GATING_B 0x202fc +#define VCE_UENC_CLOCK_GATING 0x205bc +#define VCE_UENC_REG_CLOCK_GATING 0x205c0 +#define VCE_FW_REG_STATUS 0x20e10 +# define VCE_FW_REG_STATUS_BUSY (1 << 0) +# define VCE_FW_REG_STATUS_PASS (1 << 3) +# define VCE_FW_REG_STATUS_DONE (1 << 11) +#define VCE_LMI_FW_START_KEYSEL 0x20e18 +#define VCE_LMI_FW_PERIODIC_CTRL 0x20e20 +#define VCE_LMI_CTRL2 0x20e74 +#define VCE_LMI_CTRL 0x20e98 +#define VCE_LMI_VM_CTRL 0x20ea0 +#define VCE_LMI_SWAP_CNTL 0x20eb4 +#define VCE_LMI_SWAP_CNTL1 0x20eb8 +#define VCE_LMI_CACHE_CTRL 0x20ef4 + +#define VCE_CMD_NO_OP 0x00000000 +#define VCE_CMD_END 0x00000001 +#define VCE_CMD_IB 0x00000002 +#define VCE_CMD_FENCE 0x00000003 +#define VCE_CMD_TRAP 0x00000004 +#define VCE_CMD_IB_AUTO 0x00000005 +#define VCE_CMD_SEMAPHORE 0x00000006 + #endif diff --git a/drivers/gpu/drm/radeon/sumo_dpm.c b/drivers/gpu/drm/radeon/sumo_dpm.c index 8b47b3cd0357..3f0e8d7b8dbe 100644 --- a/drivers/gpu/drm/radeon/sumo_dpm.c +++ b/drivers/gpu/drm/radeon/sumo_dpm.c @@ -1484,9 +1484,6 @@ static int sumo_parse_power_table(struct radeon_device *rdev) if (!rdev->pm.dpm.ps) return -ENOMEM; power_state_offset = (u8 *)state_array->states; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < state_array->ucNumEntries; i++) { u8 *idx; power_state = (union pplib_power_state *)power_state_offset; @@ -1772,6 +1769,10 @@ int sumo_dpm_init(struct radeon_device *rdev) sumo_construct_boot_and_acpi_state(rdev); + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = sumo_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/trinity_dpm.c b/drivers/gpu/drm/radeon/trinity_dpm.c index 2da0e17eb960..2a2822c03329 100644 --- a/drivers/gpu/drm/radeon/trinity_dpm.c +++ b/drivers/gpu/drm/radeon/trinity_dpm.c @@ -1694,9 +1694,6 @@ static int trinity_parse_power_table(struct radeon_device *rdev) if (!rdev->pm.dpm.ps) return -ENOMEM; power_state_offset = (u8 *)state_array->states; - rdev->pm.dpm.platform_caps = le32_to_cpu(power_info->pplib.ulPlatformCaps); - rdev->pm.dpm.backbias_response_time = le16_to_cpu(power_info->pplib.usBackbiasTime); - rdev->pm.dpm.voltage_response_time = le16_to_cpu(power_info->pplib.usVoltageTime); for (i = 0; i < state_array->ucNumEntries; i++) { u8 *idx; power_state = (union pplib_power_state *)power_state_offset; @@ -1895,6 +1892,10 @@ int trinity_dpm_init(struct radeon_device *rdev) trinity_construct_boot_state(rdev); + ret = r600_get_platform_caps(rdev); + if (ret) + return ret; + ret = trinity_parse_power_table(rdev); if (ret) return ret; diff --git a/drivers/gpu/drm/radeon/uvd_v1_0.c b/drivers/gpu/drm/radeon/uvd_v1_0.c index d4a68af1a279..0a243f0e5d68 100644 --- a/drivers/gpu/drm/radeon/uvd_v1_0.c +++ b/drivers/gpu/drm/radeon/uvd_v1_0.c @@ -262,7 +262,7 @@ int uvd_v1_0_start(struct radeon_device *rdev) /* Initialize the ring buffer's read and write pointers */ WREG32(UVD_RBC_RB_RPTR, 0x0); - ring->wptr = ring->rptr = RREG32(UVD_RBC_RB_RPTR); + ring->wptr = RREG32(UVD_RBC_RB_RPTR); WREG32(UVD_RBC_RB_WPTR, ring->wptr); /* set the ring address */ diff --git a/drivers/gpu/drm/radeon/vce_v1_0.c b/drivers/gpu/drm/radeon/vce_v1_0.c new file mode 100644 index 000000000000..b44d9c842f7b --- /dev/null +++ b/drivers/gpu/drm/radeon/vce_v1_0.c @@ -0,0 +1,187 @@ +/* + * Copyright 2013 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 <linux/firmware.h> +#include <drm/drmP.h> +#include "radeon.h" +#include "radeon_asic.h" +#include "sid.h" + +/** + * vce_v1_0_get_rptr - get read pointer + * + * @rdev: radeon_device pointer + * @ring: radeon_ring pointer + * + * Returns the current hardware read pointer + */ +uint32_t vce_v1_0_get_rptr(struct radeon_device *rdev, + struct radeon_ring *ring) +{ + if (ring->idx == TN_RING_TYPE_VCE1_INDEX) + return RREG32(VCE_RB_RPTR); + else + return RREG32(VCE_RB_RPTR2); +} + +/** + * vce_v1_0_get_wptr - get write pointer + * + * @rdev: radeon_device pointer + * @ring: radeon_ring pointer + * + * Returns the current hardware write pointer + */ +uint32_t vce_v1_0_get_wptr(struct radeon_device *rdev, + struct radeon_ring *ring) +{ + if (ring->idx == TN_RING_TYPE_VCE1_INDEX) + return RREG32(VCE_RB_WPTR); + else + return RREG32(VCE_RB_WPTR2); +} + +/** + * vce_v1_0_set_wptr - set write pointer + * + * @rdev: radeon_device pointer + * @ring: radeon_ring pointer + * + * Commits the write pointer to the hardware + */ +void vce_v1_0_set_wptr(struct radeon_device *rdev, + struct radeon_ring *ring) +{ + if (ring->idx == TN_RING_TYPE_VCE1_INDEX) + WREG32(VCE_RB_WPTR, ring->wptr); + else + WREG32(VCE_RB_WPTR2, ring->wptr); +} + +/** + * vce_v1_0_start - start VCE block + * + * @rdev: radeon_device pointer + * + * Setup and start the VCE block + */ +int vce_v1_0_start(struct radeon_device *rdev) +{ + struct radeon_ring *ring; + int i, j, r; + + /* set BUSY flag */ + WREG32_P(VCE_STATUS, 1, ~1); + + ring = &rdev->ring[TN_RING_TYPE_VCE1_INDEX]; + WREG32(VCE_RB_RPTR, ring->wptr); + WREG32(VCE_RB_WPTR, ring->wptr); + WREG32(VCE_RB_BASE_LO, ring->gpu_addr); + WREG32(VCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr)); + WREG32(VCE_RB_SIZE, ring->ring_size / 4); + + ring = &rdev->ring[TN_RING_TYPE_VCE2_INDEX]; + WREG32(VCE_RB_RPTR2, ring->wptr); + WREG32(VCE_RB_WPTR2, ring->wptr); + WREG32(VCE_RB_BASE_LO2, ring->gpu_addr); + WREG32(VCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr)); + WREG32(VCE_RB_SIZE2, ring->ring_size / 4); + + WREG32_P(VCE_VCPU_CNTL, VCE_CLK_EN, ~VCE_CLK_EN); + + WREG32_P(VCE_SOFT_RESET, + VCE_ECPU_SOFT_RESET | + VCE_FME_SOFT_RESET, ~( + VCE_ECPU_SOFT_RESET | + VCE_FME_SOFT_RESET)); + + mdelay(100); + + WREG32_P(VCE_SOFT_RESET, 0, ~( + VCE_ECPU_SOFT_RESET | + VCE_FME_SOFT_RESET)); + + for (i = 0; i < 10; ++i) { + uint32_t status; + for (j = 0; j < 100; ++j) { + status = RREG32(VCE_STATUS); + if (status & 2) + break; + mdelay(10); + } + r = 0; + if (status & 2) + break; + + DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n"); + WREG32_P(VCE_SOFT_RESET, VCE_ECPU_SOFT_RESET, ~VCE_ECPU_SOFT_RESET); + mdelay(10); + WREG32_P(VCE_SOFT_RESET, 0, ~VCE_ECPU_SOFT_RESET); + mdelay(10); + r = -1; + } + + /* clear BUSY flag */ + WREG32_P(VCE_STATUS, 0, ~1); + + if (r) { + DRM_ERROR("VCE not responding, giving up!!!\n"); + return r; + } + + return 0; +} + +int vce_v1_0_init(struct radeon_device *rdev) +{ + struct radeon_ring *ring; + int r; + + r = vce_v1_0_start(rdev); + if (r) + return r; + + ring = &rdev->ring[TN_RING_TYPE_VCE1_INDEX]; + ring->ready = true; + r = radeon_ring_test(rdev, TN_RING_TYPE_VCE1_INDEX, ring); + if (r) { + ring->ready = false; + return r; + } + + ring = &rdev->ring[TN_RING_TYPE_VCE2_INDEX]; + ring->ready = true; + r = radeon_ring_test(rdev, TN_RING_TYPE_VCE2_INDEX, ring); + if (r) { + ring->ready = false; + return r; + } + + DRM_INFO("VCE initialized successfully.\n"); + + return 0; +} diff --git a/drivers/gpu/drm/radeon/vce_v2_0.c b/drivers/gpu/drm/radeon/vce_v2_0.c new file mode 100644 index 000000000000..1ac7bb825a1b --- /dev/null +++ b/drivers/gpu/drm/radeon/vce_v2_0.c @@ -0,0 +1,181 @@ +/* + * Copyright 2013 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 <linux/firmware.h> +#include <drm/drmP.h> +#include "radeon.h" +#include "radeon_asic.h" +#include "cikd.h" + +static void vce_v2_0_set_sw_cg(struct radeon_device *rdev, bool gated) +{ + u32 tmp; + + if (gated) { + tmp = RREG32(VCE_CLOCK_GATING_B); + tmp |= 0xe70000; + WREG32(VCE_CLOCK_GATING_B, tmp); + + tmp = RREG32(VCE_UENC_CLOCK_GATING); + tmp |= 0xff000000; + WREG32(VCE_UENC_CLOCK_GATING, tmp); + + tmp = RREG32(VCE_UENC_REG_CLOCK_GATING); + tmp &= ~0x3fc; + WREG32(VCE_UENC_REG_CLOCK_GATING, tmp); + + WREG32(VCE_CGTT_CLK_OVERRIDE, 0); + } else { + tmp = RREG32(VCE_CLOCK_GATING_B); + tmp |= 0xe7; + tmp &= ~0xe70000; + WREG32(VCE_CLOCK_GATING_B, tmp); + + tmp = RREG32(VCE_UENC_CLOCK_GATING); + tmp |= 0x1fe000; + tmp &= ~0xff000000; + WREG32(VCE_UENC_CLOCK_GATING, tmp); + + tmp = RREG32(VCE_UENC_REG_CLOCK_GATING); + tmp |= 0x3fc; + WREG32(VCE_UENC_REG_CLOCK_GATING, tmp); + } +} + +static void vce_v2_0_set_dyn_cg(struct radeon_device *rdev, bool gated) +{ + u32 orig, tmp; + + tmp = RREG32(VCE_CLOCK_GATING_B); + tmp &= ~0x00060006; + if (gated) { + tmp |= 0xe10000; + } else { + tmp |= 0xe1; + tmp &= ~0xe10000; + } + WREG32(VCE_CLOCK_GATING_B, tmp); + + orig = tmp = RREG32(VCE_UENC_CLOCK_GATING); + tmp &= ~0x1fe000; + tmp &= ~0xff000000; + if (tmp != orig) + WREG32(VCE_UENC_CLOCK_GATING, tmp); + + orig = tmp = RREG32(VCE_UENC_REG_CLOCK_GATING); + tmp &= ~0x3fc; + if (tmp != orig) + WREG32(VCE_UENC_REG_CLOCK_GATING, tmp); + + if (gated) + WREG32(VCE_CGTT_CLK_OVERRIDE, 0); +} + +static void vce_v2_0_disable_cg(struct radeon_device *rdev) +{ + WREG32(VCE_CGTT_CLK_OVERRIDE, 7); +} + +void vce_v2_0_enable_mgcg(struct radeon_device *rdev, bool enable) +{ + bool sw_cg = false; + + if (enable && (rdev->cg_flags & RADEON_CG_SUPPORT_VCE_MGCG)) { + if (sw_cg) + vce_v2_0_set_sw_cg(rdev, true); + else + vce_v2_0_set_dyn_cg(rdev, true); + } else { + vce_v2_0_disable_cg(rdev); + + if (sw_cg) + vce_v2_0_set_sw_cg(rdev, false); + else + vce_v2_0_set_dyn_cg(rdev, false); + } +} + +static void vce_v2_0_init_cg(struct radeon_device *rdev) +{ + u32 tmp; + + tmp = RREG32(VCE_CLOCK_GATING_A); + tmp &= ~(CGC_CLK_GATE_DLY_TIMER_MASK | CGC_CLK_GATER_OFF_DLY_TIMER_MASK); + tmp |= (CGC_CLK_GATE_DLY_TIMER(0) | CGC_CLK_GATER_OFF_DLY_TIMER(4)); + tmp |= CGC_UENC_WAIT_AWAKE; + WREG32(VCE_CLOCK_GATING_A, tmp); + + tmp = RREG32(VCE_UENC_CLOCK_GATING); + tmp &= ~(CLOCK_ON_DELAY_MASK | CLOCK_OFF_DELAY_MASK); + tmp |= (CLOCK_ON_DELAY(0) | CLOCK_OFF_DELAY(4)); + WREG32(VCE_UENC_CLOCK_GATING, tmp); + + tmp = RREG32(VCE_CLOCK_GATING_B); + tmp |= 0x10; + tmp &= ~0x100000; + WREG32(VCE_CLOCK_GATING_B, tmp); +} + +int vce_v2_0_resume(struct radeon_device *rdev) +{ + uint64_t addr = rdev->vce.gpu_addr; + uint32_t size; + + WREG32_P(VCE_CLOCK_GATING_A, 0, ~(1 << 16)); + WREG32_P(VCE_UENC_CLOCK_GATING, 0x1FF000, ~0xFF9FF000); + WREG32_P(VCE_UENC_REG_CLOCK_GATING, 0x3F, ~0x3F); + WREG32(VCE_CLOCK_GATING_B, 0xf7); + + WREG32(VCE_LMI_CTRL, 0x00398000); + WREG32_P(VCE_LMI_CACHE_CTRL, 0x0, ~0x1); + WREG32(VCE_LMI_SWAP_CNTL, 0); + WREG32(VCE_LMI_SWAP_CNTL1, 0); + WREG32(VCE_LMI_VM_CTRL, 0); + + size = RADEON_GPU_PAGE_ALIGN(rdev->vce_fw->size); + WREG32(VCE_VCPU_CACHE_OFFSET0, addr & 0x7fffffff); + WREG32(VCE_VCPU_CACHE_SIZE0, size); + + addr += size; + size = RADEON_VCE_STACK_SIZE; + WREG32(VCE_VCPU_CACHE_OFFSET1, addr & 0x7fffffff); + WREG32(VCE_VCPU_CACHE_SIZE1, size); + + addr += size; + size = RADEON_VCE_HEAP_SIZE; + WREG32(VCE_VCPU_CACHE_OFFSET2, addr & 0x7fffffff); + WREG32(VCE_VCPU_CACHE_SIZE2, size); + + WREG32_P(VCE_LMI_CTRL2, 0x0, ~0x100); + + WREG32_P(VCE_SYS_INT_EN, VCE_SYS_INT_TRAP_INTERRUPT_EN, + ~VCE_SYS_INT_TRAP_INTERRUPT_EN); + + vce_v2_0_init_cg(rdev); + + return 0; +} |