diff options
Diffstat (limited to 'drivers/gpu/drm/radeon/r600.c')
| -rw-r--r-- | drivers/gpu/drm/radeon/r600.c | 1436 |
1 files changed, 723 insertions, 713 deletions
diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index 3cb9d608937..3c69f58e46e 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -28,7 +28,6 @@ #include <linux/slab.h> #include <linux/seq_file.h> #include <linux/firmware.h> -#include <linux/platform_device.h> #include <linux/module.h> #include <drm/drmP.h> #include <drm/radeon_drm.h> @@ -38,18 +37,7 @@ #include "r600d.h" #include "atom.h" #include "avivod.h" - -#define PFP_UCODE_SIZE 576 -#define PM4_UCODE_SIZE 1792 -#define RLC_UCODE_SIZE 768 -#define R700_PFP_UCODE_SIZE 848 -#define R700_PM4_UCODE_SIZE 1360 -#define R700_RLC_UCODE_SIZE 1024 -#define EVERGREEN_PFP_UCODE_SIZE 1120 -#define EVERGREEN_PM4_UCODE_SIZE 1376 -#define EVERGREEN_RLC_UCODE_SIZE 768 -#define CAYMAN_RLC_UCODE_SIZE 1024 -#define ARUBA_RLC_UCODE_SIZE 1536 +#include "radeon_ucode.h" /* Firmware Names */ MODULE_FIRMWARE("radeon/R600_pfp.bin"); @@ -68,24 +56,32 @@ MODULE_FIRMWARE("radeon/RS780_pfp.bin"); MODULE_FIRMWARE("radeon/RS780_me.bin"); MODULE_FIRMWARE("radeon/RV770_pfp.bin"); MODULE_FIRMWARE("radeon/RV770_me.bin"); +MODULE_FIRMWARE("radeon/RV770_smc.bin"); MODULE_FIRMWARE("radeon/RV730_pfp.bin"); MODULE_FIRMWARE("radeon/RV730_me.bin"); +MODULE_FIRMWARE("radeon/RV730_smc.bin"); +MODULE_FIRMWARE("radeon/RV740_smc.bin"); MODULE_FIRMWARE("radeon/RV710_pfp.bin"); MODULE_FIRMWARE("radeon/RV710_me.bin"); +MODULE_FIRMWARE("radeon/RV710_smc.bin"); MODULE_FIRMWARE("radeon/R600_rlc.bin"); MODULE_FIRMWARE("radeon/R700_rlc.bin"); MODULE_FIRMWARE("radeon/CEDAR_pfp.bin"); MODULE_FIRMWARE("radeon/CEDAR_me.bin"); MODULE_FIRMWARE("radeon/CEDAR_rlc.bin"); +MODULE_FIRMWARE("radeon/CEDAR_smc.bin"); MODULE_FIRMWARE("radeon/REDWOOD_pfp.bin"); MODULE_FIRMWARE("radeon/REDWOOD_me.bin"); MODULE_FIRMWARE("radeon/REDWOOD_rlc.bin"); +MODULE_FIRMWARE("radeon/REDWOOD_smc.bin"); MODULE_FIRMWARE("radeon/JUNIPER_pfp.bin"); MODULE_FIRMWARE("radeon/JUNIPER_me.bin"); MODULE_FIRMWARE("radeon/JUNIPER_rlc.bin"); +MODULE_FIRMWARE("radeon/JUNIPER_smc.bin"); MODULE_FIRMWARE("radeon/CYPRESS_pfp.bin"); MODULE_FIRMWARE("radeon/CYPRESS_me.bin"); MODULE_FIRMWARE("radeon/CYPRESS_rlc.bin"); +MODULE_FIRMWARE("radeon/CYPRESS_smc.bin"); MODULE_FIRMWARE("radeon/PALM_pfp.bin"); MODULE_FIRMWARE("radeon/PALM_me.bin"); MODULE_FIRMWARE("radeon/SUMO_rlc.bin"); @@ -94,6 +90,12 @@ MODULE_FIRMWARE("radeon/SUMO_me.bin"); MODULE_FIRMWARE("radeon/SUMO2_pfp.bin"); MODULE_FIRMWARE("radeon/SUMO2_me.bin"); +static const u32 crtc_offsets[2] = +{ + 0, + AVIVO_D2CRTC_H_TOTAL - AVIVO_D1CRTC_H_TOTAL +}; + int r600_debugfs_mc_info_init(struct radeon_device *rdev); /* r600,rv610,rv630,rv620,rv635,rv670 */ @@ -102,6 +104,79 @@ static void r600_gpu_init(struct radeon_device *rdev); void r600_fini(struct radeon_device *rdev); void r600_irq_disable(struct radeon_device *rdev); static void r600_pcie_gen2_enable(struct radeon_device *rdev); +extern int evergreen_rlc_resume(struct radeon_device *rdev); +extern void rv770_set_clk_bypass_mode(struct radeon_device *rdev); + +/** + * r600_get_xclk - get the xclk + * + * @rdev: radeon_device pointer + * + * Returns the reference clock used by the gfx engine + * (r6xx, IGPs, APUs). + */ +u32 r600_get_xclk(struct radeon_device *rdev) +{ + return rdev->clock.spll.reference_freq; +} + +int r600_set_uvd_clocks(struct radeon_device *rdev, u32 vclk, u32 dclk) +{ + return 0; +} + +void dce3_program_fmt(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); + struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + int bpc = 0; + u32 tmp = 0; + enum radeon_connector_dither dither = RADEON_FMT_DITHER_DISABLE; + + if (connector) { + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + bpc = radeon_get_monitor_bpc(connector); + dither = radeon_connector->dither; + } + + /* LVDS FMT is set up by atom */ + if (radeon_encoder->devices & ATOM_DEVICE_LCD_SUPPORT) + return; + + /* not needed for analog */ + if ((radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1) || + (radeon_encoder->encoder_id == ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2)) + return; + + if (bpc == 0) + return; + + switch (bpc) { + case 6: + if (dither == RADEON_FMT_DITHER_ENABLE) + /* XXX sort out optimal dither settings */ + tmp |= FMT_SPATIAL_DITHER_EN; + else + tmp |= FMT_TRUNCATE_EN; + break; + case 8: + if (dither == RADEON_FMT_DITHER_ENABLE) + /* XXX sort out optimal dither settings */ + tmp |= (FMT_SPATIAL_DITHER_EN | FMT_SPATIAL_DITHER_DEPTH); + else + tmp |= (FMT_TRUNCATE_EN | FMT_TRUNCATE_DEPTH); + break; + case 10: + default: + /* not needed */ + break; + } + + WREG32(FMT_BIT_DEPTH_CONTROL + radeon_crtc->crtc_offset, tmp); +} /* get temperature in millidegrees */ int rv6xx_get_temp(struct radeon_device *rdev) @@ -1027,6 +1102,31 @@ int r600_mc_wait_for_idle(struct radeon_device *rdev) return -1; } +uint32_t rs780_mc_rreg(struct radeon_device *rdev, uint32_t reg) +{ + unsigned long flags; + uint32_t r; + + spin_lock_irqsave(&rdev->mc_idx_lock, flags); + WREG32(R_0028F8_MC_INDEX, S_0028F8_MC_IND_ADDR(reg)); + r = RREG32(R_0028FC_MC_DATA); + WREG32(R_0028F8_MC_INDEX, ~C_0028F8_MC_IND_ADDR); + spin_unlock_irqrestore(&rdev->mc_idx_lock, flags); + return r; +} + +void rs780_mc_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) +{ + unsigned long flags; + + spin_lock_irqsave(&rdev->mc_idx_lock, flags); + WREG32(R_0028F8_MC_INDEX, S_0028F8_MC_IND_ADDR(reg) | + S_0028F8_MC_IND_WR_EN(1)); + WREG32(R_0028FC_MC_DATA, v); + WREG32(R_0028F8_MC_INDEX, 0x7F); + spin_unlock_irqrestore(&rdev->mc_idx_lock, flags); +} + static void r600_mc_program(struct radeon_device *rdev) { struct rv515_mc_save save; @@ -1126,7 +1226,7 @@ static void r600_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc } if (rdev->flags & RADEON_IS_AGP) { size_bf = mc->gtt_start; - size_af = 0xFFFFFFFF - mc->gtt_end; + size_af = mc->mc_mask - mc->gtt_end; if (size_bf > size_af) { if (mc->mc_vram_size > size_bf) { dev_warn(rdev->dev, "limiting VRAM\n"); @@ -1162,6 +1262,8 @@ static int r600_mc_init(struct radeon_device *rdev) { u32 tmp; int chansize, numchan; + uint32_t h_addr, l_addr; + unsigned long long k8_addr; /* Get VRAM informations */ rdev->mc.vram_is_ddr = true; @@ -1202,7 +1304,30 @@ static int r600_mc_init(struct radeon_device *rdev) if (rdev->flags & RADEON_IS_IGP) { rs690_pm_info(rdev); rdev->mc.igp_sideport_enabled = radeon_atombios_sideport_present(rdev); + + if (rdev->family == CHIP_RS780 || rdev->family == CHIP_RS880) { + /* Use K8 direct mapping for fast fb access. */ + rdev->fastfb_working = false; + h_addr = G_000012_K8_ADDR_EXT(RREG32_MC(R_000012_MC_MISC_UMA_CNTL)); + l_addr = RREG32_MC(R_000011_K8_FB_LOCATION); + k8_addr = ((unsigned long long)h_addr) << 32 | l_addr; +#if defined(CONFIG_X86_32) && !defined(CONFIG_X86_PAE) + if (k8_addr + rdev->mc.visible_vram_size < 0x100000000ULL) +#endif + { + /* FastFB shall be used with UMA memory. Here it is simply disabled when sideport + * memory is present. + */ + if (rdev->mc.igp_sideport_enabled == false && radeon_fastfb == 1) { + DRM_INFO("Direct mapping: aper base at 0x%llx, replaced by direct mapping base 0x%llx.\n", + (unsigned long long)rdev->mc.aper_base, k8_addr); + rdev->mc.aper_base = (resource_size_t)k8_addr; + rdev->fastfb_working = true; + } + } + } } + radeon_update_bandwidth_info(rdev); return 0; } @@ -1254,207 +1379,381 @@ void r600_vram_scratch_fini(struct radeon_device *rdev) radeon_bo_unref(&rdev->vram_scratch.robj); } -/* We doesn't check that the GPU really needs a reset we simply do the - * reset, it's up to the caller to determine if the GPU needs one. We - * might add an helper function to check that. - */ -static void r600_gpu_soft_reset_gfx(struct radeon_device *rdev) +void r600_set_bios_scratch_engine_hung(struct radeon_device *rdev, bool hung) { - u32 grbm_busy_mask = S_008010_VC_BUSY(1) | S_008010_VGT_BUSY_NO_DMA(1) | - S_008010_VGT_BUSY(1) | S_008010_TA03_BUSY(1) | - S_008010_TC_BUSY(1) | S_008010_SX_BUSY(1) | - S_008010_SH_BUSY(1) | S_008010_SPI03_BUSY(1) | - S_008010_SMX_BUSY(1) | S_008010_SC_BUSY(1) | - S_008010_PA_BUSY(1) | S_008010_DB03_BUSY(1) | - S_008010_CR_BUSY(1) | S_008010_CB03_BUSY(1) | - S_008010_GUI_ACTIVE(1); - u32 grbm2_busy_mask = S_008014_SPI0_BUSY(1) | S_008014_SPI1_BUSY(1) | - S_008014_SPI2_BUSY(1) | S_008014_SPI3_BUSY(1) | - S_008014_TA0_BUSY(1) | S_008014_TA1_BUSY(1) | - S_008014_TA2_BUSY(1) | S_008014_TA3_BUSY(1) | - S_008014_DB0_BUSY(1) | S_008014_DB1_BUSY(1) | - S_008014_DB2_BUSY(1) | S_008014_DB3_BUSY(1) | - S_008014_CB0_BUSY(1) | S_008014_CB1_BUSY(1) | - S_008014_CB2_BUSY(1) | S_008014_CB3_BUSY(1); - u32 tmp; + u32 tmp = RREG32(R600_BIOS_3_SCRATCH); - if (!(RREG32(GRBM_STATUS) & GUI_ACTIVE)) - return; + if (hung) + tmp |= ATOM_S3_ASIC_GUI_ENGINE_HUNG; + else + tmp &= ~ATOM_S3_ASIC_GUI_ENGINE_HUNG; + + WREG32(R600_BIOS_3_SCRATCH, tmp); +} +static void r600_print_gpu_status_regs(struct radeon_device *rdev) +{ dev_info(rdev->dev, " R_008010_GRBM_STATUS = 0x%08X\n", - RREG32(R_008010_GRBM_STATUS)); + RREG32(R_008010_GRBM_STATUS)); dev_info(rdev->dev, " R_008014_GRBM_STATUS2 = 0x%08X\n", - RREG32(R_008014_GRBM_STATUS2)); + RREG32(R_008014_GRBM_STATUS2)); dev_info(rdev->dev, " R_000E50_SRBM_STATUS = 0x%08X\n", - RREG32(R_000E50_SRBM_STATUS)); + RREG32(R_000E50_SRBM_STATUS)); dev_info(rdev->dev, " R_008674_CP_STALLED_STAT1 = 0x%08X\n", - RREG32(CP_STALLED_STAT1)); + RREG32(CP_STALLED_STAT1)); dev_info(rdev->dev, " R_008678_CP_STALLED_STAT2 = 0x%08X\n", - RREG32(CP_STALLED_STAT2)); + RREG32(CP_STALLED_STAT2)); dev_info(rdev->dev, " R_00867C_CP_BUSY_STAT = 0x%08X\n", - RREG32(CP_BUSY_STAT)); + RREG32(CP_BUSY_STAT)); dev_info(rdev->dev, " R_008680_CP_STAT = 0x%08X\n", - RREG32(CP_STAT)); + RREG32(CP_STAT)); + dev_info(rdev->dev, " R_00D034_DMA_STATUS_REG = 0x%08X\n", + RREG32(DMA_STATUS_REG)); +} - /* Disable CP parsing/prefetching */ - WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1)); +static bool r600_is_display_hung(struct radeon_device *rdev) +{ + u32 crtc_hung = 0; + u32 crtc_status[2]; + u32 i, j, tmp; + + for (i = 0; i < rdev->num_crtc; i++) { + if (RREG32(AVIVO_D1CRTC_CONTROL + crtc_offsets[i]) & AVIVO_CRTC_EN) { + crtc_status[i] = RREG32(AVIVO_D1CRTC_STATUS_HV_COUNT + crtc_offsets[i]); + crtc_hung |= (1 << i); + } + } - /* Check if any of the rendering block is busy and reset it */ - if ((RREG32(R_008010_GRBM_STATUS) & grbm_busy_mask) || - (RREG32(R_008014_GRBM_STATUS2) & grbm2_busy_mask)) { - tmp = S_008020_SOFT_RESET_CR(1) | - S_008020_SOFT_RESET_DB(1) | - S_008020_SOFT_RESET_CB(1) | - S_008020_SOFT_RESET_PA(1) | - S_008020_SOFT_RESET_SC(1) | - S_008020_SOFT_RESET_SMX(1) | - S_008020_SOFT_RESET_SPI(1) | - S_008020_SOFT_RESET_SX(1) | - S_008020_SOFT_RESET_SH(1) | - S_008020_SOFT_RESET_TC(1) | - S_008020_SOFT_RESET_TA(1) | - S_008020_SOFT_RESET_VC(1) | - S_008020_SOFT_RESET_VGT(1); - dev_info(rdev->dev, " R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp); - WREG32(R_008020_GRBM_SOFT_RESET, tmp); - RREG32(R_008020_GRBM_SOFT_RESET); - mdelay(15); - WREG32(R_008020_GRBM_SOFT_RESET, 0); + for (j = 0; j < 10; j++) { + for (i = 0; i < rdev->num_crtc; i++) { + if (crtc_hung & (1 << i)) { + tmp = RREG32(AVIVO_D1CRTC_STATUS_HV_COUNT + crtc_offsets[i]); + if (tmp != crtc_status[i]) + crtc_hung &= ~(1 << i); + } + } + if (crtc_hung == 0) + return false; + udelay(100); } - /* Reset CP (we always reset CP) */ - tmp = S_008020_SOFT_RESET_CP(1); - dev_info(rdev->dev, "R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp); - WREG32(R_008020_GRBM_SOFT_RESET, tmp); - RREG32(R_008020_GRBM_SOFT_RESET); - mdelay(15); - WREG32(R_008020_GRBM_SOFT_RESET, 0); - dev_info(rdev->dev, " R_008010_GRBM_STATUS = 0x%08X\n", - RREG32(R_008010_GRBM_STATUS)); - dev_info(rdev->dev, " R_008014_GRBM_STATUS2 = 0x%08X\n", - RREG32(R_008014_GRBM_STATUS2)); - dev_info(rdev->dev, " R_000E50_SRBM_STATUS = 0x%08X\n", - RREG32(R_000E50_SRBM_STATUS)); - dev_info(rdev->dev, " R_008674_CP_STALLED_STAT1 = 0x%08X\n", - RREG32(CP_STALLED_STAT1)); - dev_info(rdev->dev, " R_008678_CP_STALLED_STAT2 = 0x%08X\n", - RREG32(CP_STALLED_STAT2)); - dev_info(rdev->dev, " R_00867C_CP_BUSY_STAT = 0x%08X\n", - RREG32(CP_BUSY_STAT)); - dev_info(rdev->dev, " R_008680_CP_STAT = 0x%08X\n", - RREG32(CP_STAT)); + return true; +} + +u32 r600_gpu_check_soft_reset(struct radeon_device *rdev) +{ + u32 reset_mask = 0; + u32 tmp; + + /* GRBM_STATUS */ + tmp = RREG32(R_008010_GRBM_STATUS); + if (rdev->family >= CHIP_RV770) { + if (G_008010_PA_BUSY(tmp) | G_008010_SC_BUSY(tmp) | + G_008010_SH_BUSY(tmp) | G_008010_SX_BUSY(tmp) | + G_008010_TA_BUSY(tmp) | G_008010_VGT_BUSY(tmp) | + G_008010_DB03_BUSY(tmp) | G_008010_CB03_BUSY(tmp) | + G_008010_SPI03_BUSY(tmp) | G_008010_VGT_BUSY_NO_DMA(tmp)) + reset_mask |= RADEON_RESET_GFX; + } else { + if (G_008010_PA_BUSY(tmp) | G_008010_SC_BUSY(tmp) | + G_008010_SH_BUSY(tmp) | G_008010_SX_BUSY(tmp) | + G_008010_TA03_BUSY(tmp) | G_008010_VGT_BUSY(tmp) | + G_008010_DB03_BUSY(tmp) | G_008010_CB03_BUSY(tmp) | + G_008010_SPI03_BUSY(tmp) | G_008010_VGT_BUSY_NO_DMA(tmp)) + reset_mask |= RADEON_RESET_GFX; + } + + if (G_008010_CF_RQ_PENDING(tmp) | G_008010_PF_RQ_PENDING(tmp) | + G_008010_CP_BUSY(tmp) | G_008010_CP_COHERENCY_BUSY(tmp)) + reset_mask |= RADEON_RESET_CP; + + if (G_008010_GRBM_EE_BUSY(tmp)) + reset_mask |= RADEON_RESET_GRBM | RADEON_RESET_GFX | RADEON_RESET_CP; + + /* DMA_STATUS_REG */ + tmp = RREG32(DMA_STATUS_REG); + if (!(tmp & DMA_IDLE)) + reset_mask |= RADEON_RESET_DMA; + + /* SRBM_STATUS */ + tmp = RREG32(R_000E50_SRBM_STATUS); + if (G_000E50_RLC_RQ_PENDING(tmp) | G_000E50_RLC_BUSY(tmp)) + reset_mask |= RADEON_RESET_RLC; + + if (G_000E50_IH_BUSY(tmp)) + reset_mask |= RADEON_RESET_IH; + + if (G_000E50_SEM_BUSY(tmp)) + reset_mask |= RADEON_RESET_SEM; + + if (G_000E50_GRBM_RQ_PENDING(tmp)) + reset_mask |= RADEON_RESET_GRBM; + + if (G_000E50_VMC_BUSY(tmp)) + reset_mask |= RADEON_RESET_VMC; + + if (G_000E50_MCB_BUSY(tmp) | G_000E50_MCDZ_BUSY(tmp) | + G_000E50_MCDY_BUSY(tmp) | G_000E50_MCDX_BUSY(tmp) | + G_000E50_MCDW_BUSY(tmp)) + reset_mask |= RADEON_RESET_MC; + + if (r600_is_display_hung(rdev)) + reset_mask |= RADEON_RESET_DISPLAY; + /* Skip MC reset as it's mostly likely not hung, just busy */ + if (reset_mask & RADEON_RESET_MC) { + DRM_DEBUG("MC busy: 0x%08X, clearing.\n", reset_mask); + reset_mask &= ~RADEON_RESET_MC; + } + + return reset_mask; } -static void r600_gpu_soft_reset_dma(struct radeon_device *rdev) +static void r600_gpu_soft_reset(struct radeon_device *rdev, u32 reset_mask) { + struct rv515_mc_save save; + u32 grbm_soft_reset = 0, srbm_soft_reset = 0; u32 tmp; - if (RREG32(DMA_STATUS_REG) & DMA_IDLE) + if (reset_mask == 0) return; - dev_info(rdev->dev, " R_00D034_DMA_STATUS_REG = 0x%08X\n", - RREG32(DMA_STATUS_REG)); + dev_info(rdev->dev, "GPU softreset: 0x%08X\n", reset_mask); - /* Disable DMA */ - tmp = RREG32(DMA_RB_CNTL); - tmp &= ~DMA_RB_ENABLE; - WREG32(DMA_RB_CNTL, tmp); + r600_print_gpu_status_regs(rdev); - /* Reset dma */ + /* Disable CP parsing/prefetching */ if (rdev->family >= CHIP_RV770) - WREG32(SRBM_SOFT_RESET, RV770_SOFT_RESET_DMA); + WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1) | S_0086D8_CP_PFP_HALT(1)); else - WREG32(SRBM_SOFT_RESET, SOFT_RESET_DMA); - RREG32(SRBM_SOFT_RESET); + WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1)); + + /* disable the RLC */ + WREG32(RLC_CNTL, 0); + + if (reset_mask & RADEON_RESET_DMA) { + /* Disable DMA */ + tmp = RREG32(DMA_RB_CNTL); + tmp &= ~DMA_RB_ENABLE; + WREG32(DMA_RB_CNTL, tmp); + } + + mdelay(50); + + rv515_mc_stop(rdev, &save); + if (r600_mc_wait_for_idle(rdev)) { + dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); + } + + if (reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE)) { + if (rdev->family >= CHIP_RV770) + grbm_soft_reset |= S_008020_SOFT_RESET_DB(1) | + S_008020_SOFT_RESET_CB(1) | + S_008020_SOFT_RESET_PA(1) | + S_008020_SOFT_RESET_SC(1) | + S_008020_SOFT_RESET_SPI(1) | + S_008020_SOFT_RESET_SX(1) | + S_008020_SOFT_RESET_SH(1) | + S_008020_SOFT_RESET_TC(1) | + S_008020_SOFT_RESET_TA(1) | + S_008020_SOFT_RESET_VC(1) | + S_008020_SOFT_RESET_VGT(1); + else + grbm_soft_reset |= S_008020_SOFT_RESET_CR(1) | + S_008020_SOFT_RESET_DB(1) | + S_008020_SOFT_RESET_CB(1) | + S_008020_SOFT_RESET_PA(1) | + S_008020_SOFT_RESET_SC(1) | + S_008020_SOFT_RESET_SMX(1) | + S_008020_SOFT_RESET_SPI(1) | + S_008020_SOFT_RESET_SX(1) | + S_008020_SOFT_RESET_SH(1) | + S_008020_SOFT_RESET_TC(1) | + S_008020_SOFT_RESET_TA(1) | + S_008020_SOFT_RESET_VC(1) | + S_008020_SOFT_RESET_VGT(1); + } + + if (reset_mask & RADEON_RESET_CP) { + grbm_soft_reset |= S_008020_SOFT_RESET_CP(1) | + S_008020_SOFT_RESET_VGT(1); + + srbm_soft_reset |= S_000E60_SOFT_RESET_GRBM(1); + } + + if (reset_mask & RADEON_RESET_DMA) { + if (rdev->family >= CHIP_RV770) + srbm_soft_reset |= RV770_SOFT_RESET_DMA; + else + srbm_soft_reset |= SOFT_RESET_DMA; + } + + if (reset_mask & RADEON_RESET_RLC) + srbm_soft_reset |= S_000E60_SOFT_RESET_RLC(1); + + if (reset_mask & RADEON_RESET_SEM) + srbm_soft_reset |= S_000E60_SOFT_RESET_SEM(1); + + if (reset_mask & RADEON_RESET_IH) + srbm_soft_reset |= S_000E60_SOFT_RESET_IH(1); + + if (reset_mask & RADEON_RESET_GRBM) + srbm_soft_reset |= S_000E60_SOFT_RESET_GRBM(1); + + if (!(rdev->flags & RADEON_IS_IGP)) { + if (reset_mask & RADEON_RESET_MC) + srbm_soft_reset |= S_000E60_SOFT_RESET_MC(1); + } + + if (reset_mask & RADEON_RESET_VMC) + srbm_soft_reset |= S_000E60_SOFT_RESET_VMC(1); + + if (grbm_soft_reset) { + tmp = RREG32(R_008020_GRBM_SOFT_RESET); + tmp |= grbm_soft_reset; + dev_info(rdev->dev, "R_008020_GRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(R_008020_GRBM_SOFT_RESET, tmp); + tmp = RREG32(R_008020_GRBM_SOFT_RESET); + + udelay(50); + + tmp &= ~grbm_soft_reset; + WREG32(R_008020_GRBM_SOFT_RESET, tmp); + tmp = RREG32(R_008020_GRBM_SOFT_RESET); + } + + if (srbm_soft_reset) { + tmp = RREG32(SRBM_SOFT_RESET); + tmp |= srbm_soft_reset; + dev_info(rdev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(SRBM_SOFT_RESET, tmp); + tmp = RREG32(SRBM_SOFT_RESET); + + udelay(50); + + tmp &= ~srbm_soft_reset; + WREG32(SRBM_SOFT_RESET, tmp); + tmp = RREG32(SRBM_SOFT_RESET); + } + + /* Wait a little for things to settle down */ + mdelay(1); + + rv515_mc_resume(rdev, &save); udelay(50); - WREG32(SRBM_SOFT_RESET, 0); - dev_info(rdev->dev, " R_00D034_DMA_STATUS_REG = 0x%08X\n", - RREG32(DMA_STATUS_REG)); + r600_print_gpu_status_regs(rdev); } -static int r600_gpu_soft_reset(struct radeon_device *rdev, u32 reset_mask) +static void r600_gpu_pci_config_reset(struct radeon_device *rdev) { struct rv515_mc_save save; + u32 tmp, i; - if (!(RREG32(GRBM_STATUS) & GUI_ACTIVE)) - reset_mask &= ~(RADEON_RESET_GFX | RADEON_RESET_COMPUTE); + dev_info(rdev->dev, "GPU pci config reset\n"); - if (RREG32(DMA_STATUS_REG) & DMA_IDLE) - reset_mask &= ~RADEON_RESET_DMA; + /* disable dpm? */ - if (reset_mask == 0) - return 0; + /* Disable CP parsing/prefetching */ + if (rdev->family >= CHIP_RV770) + WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1) | S_0086D8_CP_PFP_HALT(1)); + else + WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1)); - dev_info(rdev->dev, "GPU softreset: 0x%08X\n", reset_mask); + /* disable the RLC */ + WREG32(RLC_CNTL, 0); + /* Disable DMA */ + tmp = RREG32(DMA_RB_CNTL); + tmp &= ~DMA_RB_ENABLE; + WREG32(DMA_RB_CNTL, tmp); + + mdelay(50); + + /* set mclk/sclk to bypass */ + if (rdev->family >= CHIP_RV770) + rv770_set_clk_bypass_mode(rdev); + /* disable BM */ + pci_clear_master(rdev->pdev); + /* disable mem access */ rv515_mc_stop(rdev, &save); if (r600_mc_wait_for_idle(rdev)) { dev_warn(rdev->dev, "Wait for MC idle timedout !\n"); } - if (reset_mask & (RADEON_RESET_GFX | RADEON_RESET_COMPUTE)) - r600_gpu_soft_reset_gfx(rdev); + /* BIF reset workaround. Not sure if this is needed on 6xx */ + tmp = RREG32(BUS_CNTL); + tmp |= VGA_COHE_SPEC_TIMER_DIS; + WREG32(BUS_CNTL, tmp); - if (reset_mask & RADEON_RESET_DMA) - r600_gpu_soft_reset_dma(rdev); + tmp = RREG32(BIF_SCRATCH0); - /* Wait a little for things to settle down */ + /* reset */ + radeon_pci_config_reset(rdev); mdelay(1); - rv515_mc_resume(rdev, &save); - return 0; + /* BIF reset workaround. Not sure if this is needed on 6xx */ + tmp = SOFT_RESET_BIF; + WREG32(SRBM_SOFT_RESET, tmp); + mdelay(1); + WREG32(SRBM_SOFT_RESET, 0); + + /* wait for asic to come out of reset */ + for (i = 0; i < rdev->usec_timeout; i++) { + if (RREG32(CONFIG_MEMSIZE) != 0xffffffff) + break; + udelay(1); + } } -bool r600_gpu_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) +int r600_asic_reset(struct radeon_device *rdev) { - u32 srbm_status; - u32 grbm_status; - u32 grbm_status2; - - srbm_status = RREG32(R_000E50_SRBM_STATUS); - grbm_status = RREG32(R_008010_GRBM_STATUS); - grbm_status2 = RREG32(R_008014_GRBM_STATUS2); - if (!G_008010_GUI_ACTIVE(grbm_status)) { - radeon_ring_lockup_update(ring); - return false; - } - /* force CP activities */ - radeon_ring_force_activity(rdev, ring); - return radeon_ring_test_lockup(rdev, ring); + u32 reset_mask; + + reset_mask = r600_gpu_check_soft_reset(rdev); + + if (reset_mask) + r600_set_bios_scratch_engine_hung(rdev, true); + + /* try soft reset */ + r600_gpu_soft_reset(rdev, reset_mask); + + reset_mask = r600_gpu_check_soft_reset(rdev); + + /* try pci config reset */ + if (reset_mask && radeon_hard_reset) + r600_gpu_pci_config_reset(rdev); + + reset_mask = r600_gpu_check_soft_reset(rdev); + + if (!reset_mask) + r600_set_bios_scratch_engine_hung(rdev, false); + + return 0; } /** - * r600_dma_is_lockup - Check if the DMA engine is locked up + * r600_gfx_is_lockup - Check if the GFX engine is locked up * * @rdev: radeon_device pointer * @ring: radeon_ring structure holding ring information * - * Check if the async DMA engine is locked up (r6xx-evergreen). + * Check if the GFX engine is locked up. * Returns true if the engine appears to be locked up, false if not. */ -bool r600_dma_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) +bool r600_gfx_is_lockup(struct radeon_device *rdev, struct radeon_ring *ring) { - u32 dma_status_reg; + u32 reset_mask = r600_gpu_check_soft_reset(rdev); - dma_status_reg = RREG32(DMA_STATUS_REG); - if (dma_status_reg & DMA_IDLE) { - radeon_ring_lockup_update(ring); + if (!(reset_mask & (RADEON_RESET_GFX | + RADEON_RESET_COMPUTE | + RADEON_RESET_CP))) { + radeon_ring_lockup_update(rdev, ring); return false; } - /* force ring activities */ - radeon_ring_force_activity(rdev, ring); return radeon_ring_test_lockup(rdev, ring); } -int r600_asic_reset(struct radeon_device *rdev) -{ - return r600_gpu_soft_reset(rdev, (RADEON_RESET_GFX | - RADEON_RESET_COMPUTE | - RADEON_RESET_DMA)); -} - u32 r6xx_remap_render_backend(struct radeon_device *rdev, u32 tiling_pipe_num, u32 max_rb_num, @@ -1462,12 +1761,15 @@ u32 r6xx_remap_render_backend(struct radeon_device *rdev, u32 disabled_rb_mask) { u32 rendering_pipe_num, rb_num_width, req_rb_num; - u32 pipe_rb_ratio, pipe_rb_remain; + u32 pipe_rb_ratio, pipe_rb_remain, tmp; u32 data = 0, mask = 1 << (max_rb_num - 1); unsigned i, j; /* mask out the RBs that don't exist on that asic */ - disabled_rb_mask |= (0xff << max_rb_num) & 0xff; + tmp = disabled_rb_mask | ((0xff << max_rb_num) & 0xff); + /* make sure at least one RB is available */ + if ((tmp & 0xff) != 0xff) + disabled_rb_mask = tmp; rendering_pipe_num = 1 << tiling_pipe_num; req_rb_num = total_max_rb_num - r600_count_pipe_bits(disabled_rb_mask); @@ -1656,6 +1958,9 @@ static void r600_gpu_init(struct radeon_device *rdev) if (tmp < rdev->config.r600.max_simds) { rdev->config.r600.max_simds = tmp; } + tmp = rdev->config.r600.max_simds - + r600_count_pipe_bits((cc_gc_shader_pipe_config >> 16) & R6XX_MAX_SIMDS_MASK); + rdev->config.r600.active_simds = tmp; disabled_rb_mask = (RREG32(CC_RB_BACKEND_DISABLE) >> 16) & R6XX_MAX_BACKENDS_MASK; tmp = (tiling_config & PIPE_TILING__MASK) >> PIPE_TILING__SHIFT; @@ -1922,20 +2227,27 @@ static void r600_gpu_init(struct radeon_device *rdev) */ u32 r600_pciep_rreg(struct radeon_device *rdev, u32 reg) { + unsigned long flags; u32 r; + spin_lock_irqsave(&rdev->pciep_idx_lock, flags); WREG32(PCIE_PORT_INDEX, ((reg) & 0xff)); (void)RREG32(PCIE_PORT_INDEX); r = RREG32(PCIE_PORT_DATA); + spin_unlock_irqrestore(&rdev->pciep_idx_lock, flags); return r; } void r600_pciep_wreg(struct radeon_device *rdev, u32 reg, u32 v) { + unsigned long flags; + + spin_lock_irqsave(&rdev->pciep_idx_lock, flags); WREG32(PCIE_PORT_INDEX, ((reg) & 0xff)); (void)RREG32(PCIE_PORT_INDEX); WREG32(PCIE_PORT_DATA, (v)); (void)RREG32(PCIE_PORT_DATA); + spin_unlock_irqrestore(&rdev->pciep_idx_lock, flags); } /* @@ -1943,7 +2255,8 @@ void r600_pciep_wreg(struct radeon_device *rdev, u32 reg, u32 v) */ void r600_cp_stop(struct radeon_device *rdev) { - radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); + if (rdev->asic->copy.copy_ring_index == RADEON_RING_TYPE_GFX_INDEX) + radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); WREG32(R_0086D8_CP_ME_CNTL, S_0086D8_CP_ME_HALT(1)); WREG32(SCRATCH_UMSK, 0); rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ready = false; @@ -1951,22 +2264,15 @@ void r600_cp_stop(struct radeon_device *rdev) int r600_init_microcode(struct radeon_device *rdev) { - struct platform_device *pdev; const char *chip_name; const char *rlc_chip_name; - size_t pfp_req_size, me_req_size, rlc_req_size; + const char *smc_chip_name = "RV770"; + size_t pfp_req_size, me_req_size, rlc_req_size, smc_req_size = 0; char fw_name[30]; int err; DRM_DEBUG("\n"); - pdev = platform_device_register_simple("radeon_cp", 0, NULL, 0); - err = IS_ERR(pdev); - if (err) { - printk(KERN_ERR "radeon_cp: Failed to register firmware\n"); - return -EINVAL; - } - switch (rdev->family) { case CHIP_R600: chip_name = "R600"; @@ -2000,32 +2306,51 @@ int r600_init_microcode(struct radeon_device *rdev) case CHIP_RV770: chip_name = "RV770"; rlc_chip_name = "R700"; + smc_chip_name = "RV770"; + smc_req_size = ALIGN(RV770_SMC_UCODE_SIZE, 4); break; case CHIP_RV730: - case CHIP_RV740: chip_name = "RV730"; rlc_chip_name = "R700"; + smc_chip_name = "RV730"; + smc_req_size = ALIGN(RV730_SMC_UCODE_SIZE, 4); break; case CHIP_RV710: chip_name = "RV710"; rlc_chip_name = "R700"; + smc_chip_name = "RV710"; + smc_req_size = ALIGN(RV710_SMC_UCODE_SIZE, 4); + break; + case CHIP_RV740: + chip_name = "RV730"; + rlc_chip_name = "R700"; + smc_chip_name = "RV740"; + smc_req_size = ALIGN(RV740_SMC_UCODE_SIZE, 4); break; case CHIP_CEDAR: chip_name = "CEDAR"; rlc_chip_name = "CEDAR"; + smc_chip_name = "CEDAR"; + smc_req_size = ALIGN(CEDAR_SMC_UCODE_SIZE, 4); break; case CHIP_REDWOOD: chip_name = "REDWOOD"; rlc_chip_name = "REDWOOD"; + smc_chip_name = "REDWOOD"; + smc_req_size = ALIGN(REDWOOD_SMC_UCODE_SIZE, 4); break; case CHIP_JUNIPER: chip_name = "JUNIPER"; rlc_chip_name = "JUNIPER"; + smc_chip_name = "JUNIPER"; + smc_req_size = ALIGN(JUNIPER_SMC_UCODE_SIZE, 4); break; case CHIP_CYPRESS: case CHIP_HEMLOCK: chip_name = "CYPRESS"; rlc_chip_name = "CYPRESS"; + smc_chip_name = "CYPRESS"; + smc_req_size = ALIGN(CYPRESS_SMC_UCODE_SIZE, 4); break; case CHIP_PALM: chip_name = "PALM"; @@ -2051,15 +2376,15 @@ int r600_init_microcode(struct radeon_device *rdev) me_req_size = R700_PM4_UCODE_SIZE * 4; rlc_req_size = R700_RLC_UCODE_SIZE * 4; } else { - pfp_req_size = PFP_UCODE_SIZE * 4; - me_req_size = PM4_UCODE_SIZE * 12; - rlc_req_size = RLC_UCODE_SIZE * 4; + pfp_req_size = R600_PFP_UCODE_SIZE * 4; + me_req_size = R600_PM4_UCODE_SIZE * 12; + rlc_req_size = R600_RLC_UCODE_SIZE * 4; } DRM_INFO("Loading %s Microcode\n", chip_name); snprintf(fw_name, sizeof(fw_name), "radeon/%s_pfp.bin", chip_name); - err = request_firmware(&rdev->pfp_fw, fw_name, &pdev->dev); + err = request_firmware(&rdev->pfp_fw, fw_name, rdev->dev); if (err) goto out; if (rdev->pfp_fw->size != pfp_req_size) { @@ -2071,7 +2396,7 @@ int r600_init_microcode(struct radeon_device *rdev) } snprintf(fw_name, sizeof(fw_name), "radeon/%s_me.bin", chip_name); - err = request_firmware(&rdev->me_fw, fw_name, &pdev->dev); + err = request_firmware(&rdev->me_fw, fw_name, rdev->dev); if (err) goto out; if (rdev->me_fw->size != me_req_size) { @@ -2082,7 +2407,7 @@ int r600_init_microcode(struct radeon_device *rdev) } snprintf(fw_name, sizeof(fw_name), "radeon/%s_rlc.bin", rlc_chip_name); - err = request_firmware(&rdev->rlc_fw, fw_name, &pdev->dev); + err = request_firmware(&rdev->rlc_fw, fw_name, rdev->dev); if (err) goto out; if (rdev->rlc_fw->size != rlc_req_size) { @@ -2092,9 +2417,25 @@ int r600_init_microcode(struct radeon_device *rdev) err = -EINVAL; } -out: - platform_device_unregister(pdev); + if ((rdev->family >= CHIP_RV770) && (rdev->family <= CHIP_HEMLOCK)) { + snprintf(fw_name, sizeof(fw_name), "radeon/%s_smc.bin", smc_chip_name); + err = request_firmware(&rdev->smc_fw, fw_name, rdev->dev); + if (err) { + printk(KERN_ERR + "smc: error loading firmware \"%s\"\n", + fw_name); + release_firmware(rdev->smc_fw); + rdev->smc_fw = NULL; + err = 0; + } else if (rdev->smc_fw->size != smc_req_size) { + printk(KERN_ERR + "smc: Bogus length %zu in firmware \"%s\"\n", + rdev->smc_fw->size, fw_name); + err = -EINVAL; + } + } +out: if (err) { if (err != -EINVAL) printk(KERN_ERR @@ -2106,10 +2447,42 @@ out: rdev->me_fw = NULL; release_firmware(rdev->rlc_fw); rdev->rlc_fw = NULL; + release_firmware(rdev->smc_fw); + rdev->smc_fw = NULL; } return err; } +u32 r600_gfx_get_rptr(struct radeon_device *rdev, + struct radeon_ring *ring) +{ + u32 rptr; + + if (rdev->wb.enabled) + rptr = rdev->wb.wb[ring->rptr_offs/4]; + else + rptr = RREG32(R600_CP_RB_RPTR); + + return rptr; +} + +u32 r600_gfx_get_wptr(struct radeon_device *rdev, + struct radeon_ring *ring) +{ + u32 wptr; + + wptr = RREG32(R600_CP_RB_WPTR); + + return wptr; +} + +void r600_gfx_set_wptr(struct radeon_device *rdev, + struct radeon_ring *ring) +{ + WREG32(R600_CP_RB_WPTR, ring->wptr); + (void)RREG32(R600_CP_RB_WPTR); +} + static int r600_cp_load_microcode(struct radeon_device *rdev) { const __be32 *fw_data; @@ -2136,13 +2509,13 @@ static int r600_cp_load_microcode(struct radeon_device *rdev) fw_data = (const __be32 *)rdev->me_fw->data; WREG32(CP_ME_RAM_WADDR, 0); - for (i = 0; i < PM4_UCODE_SIZE * 3; i++) + for (i = 0; i < R600_PM4_UCODE_SIZE * 3; i++) WREG32(CP_ME_RAM_DATA, be32_to_cpup(fw_data++)); fw_data = (const __be32 *)rdev->pfp_fw->data; WREG32(CP_PFP_UCODE_ADDR, 0); - for (i = 0; i < PFP_UCODE_SIZE; i++) + for (i = 0; i < R600_PFP_UCODE_SIZE; i++) WREG32(CP_PFP_UCODE_DATA, be32_to_cpup(fw_data++)); @@ -2196,8 +2569,8 @@ int r600_cp_resume(struct radeon_device *rdev) WREG32(GRBM_SOFT_RESET, 0); /* Set ring buffer size */ - rb_bufsz = drm_order(ring->ring_size / 8); - tmp = (drm_order(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz; + rb_bufsz = order_base_2(ring->ring_size / 8); + tmp = (order_base_2(RADEON_GPU_PAGE_SIZE/8) << 8) | rb_bufsz; #ifdef __BIG_ENDIAN tmp |= BUF_SWAP_32BIT; #endif @@ -2232,8 +2605,6 @@ 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); @@ -2241,6 +2612,10 @@ int r600_cp_resume(struct radeon_device *rdev) ring->ready = false; return r; } + + if (rdev->asic->copy.copy_ring_index == RADEON_RING_TYPE_GFX_INDEX) + radeon_ttm_set_active_vram_size(rdev, rdev->mc.real_vram_size); + return 0; } @@ -2250,7 +2625,7 @@ void r600_ring_init(struct radeon_device *rdev, struct radeon_ring *ring, unsign int r; /* Align ring size */ - rb_bufsz = drm_order(ring_size / 8); + rb_bufsz = order_base_2(ring_size / 8); ring_size = (1 << (rb_bufsz + 1)) * 4; ring->ring_size = ring_size; ring->align_mask = 16 - 1; @@ -2273,129 +2648,6 @@ void r600_cp_fini(struct radeon_device *rdev) } /* - * DMA - * Starting with R600, the GPU has an asynchronous - * DMA engine. The programming model is very similar - * to the 3D engine (ring buffer, IBs, etc.), but the - * DMA controller has it's own packet format that is - * different form the PM4 format used by the 3D engine. - * It supports copying data, writing embedded data, - * solid fills, and a number of other things. It also - * has support for tiling/detiling of buffers. - */ -/** - * r600_dma_stop - stop the async dma engine - * - * @rdev: radeon_device pointer - * - * Stop the async dma engine (r6xx-evergreen). - */ -void r600_dma_stop(struct radeon_device *rdev) -{ - u32 rb_cntl = RREG32(DMA_RB_CNTL); - - radeon_ttm_set_active_vram_size(rdev, rdev->mc.visible_vram_size); - - rb_cntl &= ~DMA_RB_ENABLE; - WREG32(DMA_RB_CNTL, rb_cntl); - - rdev->ring[R600_RING_TYPE_DMA_INDEX].ready = false; -} - -/** - * r600_dma_resume - setup and start the async dma engine - * - * @rdev: radeon_device pointer - * - * Set up the DMA ring buffer and enable it. (r6xx-evergreen). - * Returns 0 for success, error for failure. - */ -int r600_dma_resume(struct radeon_device *rdev) -{ - struct radeon_ring *ring = &rdev->ring[R600_RING_TYPE_DMA_INDEX]; - u32 rb_cntl, dma_cntl; - u32 rb_bufsz; - int r; - - /* Reset dma */ - if (rdev->family >= CHIP_RV770) - WREG32(SRBM_SOFT_RESET, RV770_SOFT_RESET_DMA); - else - WREG32(SRBM_SOFT_RESET, SOFT_RESET_DMA); - RREG32(SRBM_SOFT_RESET); - udelay(50); - WREG32(SRBM_SOFT_RESET, 0); - - WREG32(DMA_SEM_INCOMPLETE_TIMER_CNTL, 0); - WREG32(DMA_SEM_WAIT_FAIL_TIMER_CNTL, 0); - - /* Set ring buffer size in dwords */ - rb_bufsz = drm_order(ring->ring_size / 4); - rb_cntl = rb_bufsz << 1; -#ifdef __BIG_ENDIAN - rb_cntl |= DMA_RB_SWAP_ENABLE | DMA_RPTR_WRITEBACK_SWAP_ENABLE; -#endif - WREG32(DMA_RB_CNTL, rb_cntl); - - /* Initialize the ring buffer's read and write pointers */ - WREG32(DMA_RB_RPTR, 0); - WREG32(DMA_RB_WPTR, 0); - - /* set the wb address whether it's enabled or not */ - WREG32(DMA_RB_RPTR_ADDR_HI, - upper_32_bits(rdev->wb.gpu_addr + R600_WB_DMA_RPTR_OFFSET) & 0xFF); - WREG32(DMA_RB_RPTR_ADDR_LO, - ((rdev->wb.gpu_addr + R600_WB_DMA_RPTR_OFFSET) & 0xFFFFFFFC)); - - if (rdev->wb.enabled) - rb_cntl |= DMA_RPTR_WRITEBACK_ENABLE; - - WREG32(DMA_RB_BASE, ring->gpu_addr >> 8); - - /* enable DMA IBs */ - WREG32(DMA_IB_CNTL, DMA_IB_ENABLE); - - dma_cntl = RREG32(DMA_CNTL); - dma_cntl &= ~CTXEMPTY_INT_ENABLE; - WREG32(DMA_CNTL, dma_cntl); - - if (rdev->family >= CHIP_RV770) - WREG32(DMA_MODE, 1); - - 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; - - r = radeon_ring_test(rdev, R600_RING_TYPE_DMA_INDEX, ring); - if (r) { - ring->ready = false; - return r; - } - - radeon_ttm_set_active_vram_size(rdev, rdev->mc.real_vram_size); - - return 0; -} - -/** - * r600_dma_fini - tear down the async dma engine - * - * @rdev: radeon_device pointer - * - * Stop the async dma engine and free the ring (r6xx-evergreen). - */ -void r600_dma_fini(struct radeon_device *rdev) -{ - r600_dma_stop(rdev); - radeon_ring_fini(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX]); -} - -/* * GPU scratch registers helpers function. */ void r600_scratch_init(struct radeon_device *rdev) @@ -2450,60 +2702,6 @@ int r600_ring_test(struct radeon_device *rdev, struct radeon_ring *ring) return r; } -/** - * r600_dma_ring_test - simple async dma engine test - * - * @rdev: radeon_device pointer - * @ring: radeon_ring structure holding ring information - * - * Test the DMA engine by writing using it to write an - * value to memory. (r6xx-SI). - * Returns 0 for success, error for failure. - */ -int r600_dma_ring_test(struct radeon_device *rdev, - struct radeon_ring *ring) -{ - unsigned i; - int r; - void __iomem *ptr = (void *)rdev->vram_scratch.ptr; - u32 tmp; - - if (!ptr) { - DRM_ERROR("invalid vram scratch pointer\n"); - return -EINVAL; - } - - tmp = 0xCAFEDEAD; - writel(tmp, ptr); - - r = radeon_ring_lock(rdev, ring, 4); - if (r) { - DRM_ERROR("radeon: dma failed to lock ring %d (%d).\n", ring->idx, r); - return r; - } - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 1)); - radeon_ring_write(ring, rdev->vram_scratch.gpu_addr & 0xfffffffc); - radeon_ring_write(ring, upper_32_bits(rdev->vram_scratch.gpu_addr) & 0xff); - radeon_ring_write(ring, 0xDEADBEEF); - radeon_ring_unlock_commit(rdev, ring); - - for (i = 0; i < rdev->usec_timeout; i++) { - tmp = readl(ptr); - if (tmp == 0xDEADBEEF) - 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 (0x%08X)\n", - ring->idx, tmp); - r = -EINVAL; - } - return r; -} - /* * CP fences/semaphores */ @@ -2512,30 +2710,31 @@ void r600_fence_ring_emit(struct radeon_device *rdev, struct radeon_fence *fence) { struct radeon_ring *ring = &rdev->ring[fence->ring]; + u32 cp_coher_cntl = PACKET3_TC_ACTION_ENA | PACKET3_VC_ACTION_ENA | + PACKET3_SH_ACTION_ENA; + + if (rdev->family >= CHIP_RV770) + cp_coher_cntl |= PACKET3_FULL_CACHE_ENA; if (rdev->wb.use_event) { u64 addr = rdev->fence_drv[fence->ring].gpu_addr; /* flush read cache over gart */ radeon_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3)); - radeon_ring_write(ring, PACKET3_TC_ACTION_ENA | - PACKET3_VC_ACTION_ENA | - PACKET3_SH_ACTION_ENA); + radeon_ring_write(ring, cp_coher_cntl); radeon_ring_write(ring, 0xFFFFFFFF); radeon_ring_write(ring, 0); radeon_ring_write(ring, 10); /* poll interval */ /* EVENT_WRITE_EOP - flush caches, send int */ radeon_ring_write(ring, PACKET3(PACKET3_EVENT_WRITE_EOP, 4)); radeon_ring_write(ring, EVENT_TYPE(CACHE_FLUSH_AND_INV_EVENT_TS) | EVENT_INDEX(5)); - radeon_ring_write(ring, addr & 0xffffffff); + radeon_ring_write(ring, lower_32_bits(addr)); radeon_ring_write(ring, (upper_32_bits(addr) & 0xff) | DATA_SEL(1) | INT_SEL(2)); radeon_ring_write(ring, fence->seq); radeon_ring_write(ring, 0); } else { /* flush read cache over gart */ radeon_ring_write(ring, PACKET3(PACKET3_SURFACE_SYNC, 3)); - radeon_ring_write(ring, PACKET3_TC_ACTION_ENA | - PACKET3_VC_ACTION_ENA | - PACKET3_SH_ACTION_ENA); + radeon_ring_write(ring, cp_coher_cntl); radeon_ring_write(ring, 0xFFFFFFFF); radeon_ring_write(ring, 0); radeon_ring_write(ring, 10); /* poll interval */ @@ -2555,7 +2754,7 @@ void r600_fence_ring_emit(struct radeon_device *rdev, } } -void r600_semaphore_ring_emit(struct radeon_device *rdev, +bool r600_semaphore_ring_emit(struct radeon_device *rdev, struct radeon_ring *ring, struct radeon_semaphore *semaphore, bool emit_wait) @@ -2567,84 +2766,14 @@ void r600_semaphore_ring_emit(struct radeon_device *rdev, sel |= PACKET3_SEM_WAIT_ON_SIGNAL; radeon_ring_write(ring, PACKET3(PACKET3_MEM_SEMAPHORE, 1)); - radeon_ring_write(ring, addr & 0xffffffff); + radeon_ring_write(ring, lower_32_bits(addr)); radeon_ring_write(ring, (upper_32_bits(addr) & 0xff) | sel); -} -/* - * DMA fences/semaphores - */ - -/** - * r600_dma_fence_ring_emit - emit a fence on the DMA ring - * - * @rdev: radeon_device pointer - * @fence: radeon fence object - * - * Add a DMA fence packet to the ring to write - * the fence seq number and DMA trap packet to generate - * an interrupt if needed (r6xx-r7xx). - */ -void r600_dma_fence_ring_emit(struct radeon_device *rdev, - struct radeon_fence *fence) -{ - struct radeon_ring *ring = &rdev->ring[fence->ring]; - u64 addr = rdev->fence_drv[fence->ring].gpu_addr; - - /* write the fence */ - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_FENCE, 0, 0, 0)); - radeon_ring_write(ring, addr & 0xfffffffc); - radeon_ring_write(ring, (upper_32_bits(addr) & 0xff)); - radeon_ring_write(ring, lower_32_bits(fence->seq)); - /* generate an interrupt */ - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_TRAP, 0, 0, 0)); + return true; } /** - * r600_dma_semaphore_ring_emit - emit a semaphore on the dma ring - * - * @rdev: radeon_device pointer - * @ring: radeon_ring structure holding ring information - * @semaphore: radeon semaphore object - * @emit_wait: wait or signal semaphore - * - * Add a DMA semaphore packet to the ring wait on or signal - * other rings (r6xx-SI). - */ -void r600_dma_semaphore_ring_emit(struct radeon_device *rdev, - struct radeon_ring *ring, - struct radeon_semaphore *semaphore, - bool emit_wait) -{ - u64 addr = semaphore->gpu_addr; - u32 s = emit_wait ? 0 : 1; - - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_SEMAPHORE, 0, s, 0)); - radeon_ring_write(ring, addr & 0xfffffffc); - radeon_ring_write(ring, upper_32_bits(addr) & 0xff); -} - -int r600_copy_blit(struct radeon_device *rdev, - uint64_t src_offset, - uint64_t dst_offset, - unsigned num_gpu_pages, - struct radeon_fence **fence) -{ - struct radeon_semaphore *sem = NULL; - struct radeon_sa_bo *vb = NULL; - int r; - - r = r600_blit_prepare_copy(rdev, num_gpu_pages, fence, &vb, &sem); - if (r) { - return r; - } - r600_kms_blit_copy(rdev, src_offset, dst_offset, num_gpu_pages, vb); - r600_blit_done_copy(rdev, fence, vb, sem); - return 0; -} - -/** - * r600_copy_dma - copy pages using the DMA engine + * r600_copy_cpdma - copy pages using the CP DMA engine * * @rdev: radeon_device pointer * @src_offset: src GPU address @@ -2652,19 +2781,19 @@ int r600_copy_blit(struct radeon_device *rdev, * @num_gpu_pages: number of GPU pages to xfer * @fence: radeon fence object * - * Copy GPU paging using the DMA engine (r6xx). + * Copy GPU paging using the CP DMA engine (r6xx+). * Used by the radeon ttm implementation to move pages if * registered as the asic copy callback. */ -int r600_copy_dma(struct radeon_device *rdev, - uint64_t src_offset, uint64_t dst_offset, - unsigned num_gpu_pages, - struct radeon_fence **fence) +int r600_copy_cpdma(struct radeon_device *rdev, + uint64_t src_offset, uint64_t dst_offset, + unsigned num_gpu_pages, + struct radeon_fence **fence) { struct radeon_semaphore *sem = NULL; - int ring_index = rdev->asic->copy.dma_ring_index; + int ring_index = rdev->asic->copy.blit_ring_index; struct radeon_ring *ring = &rdev->ring[ring_index]; - u32 size_in_dw, cur_size_in_dw; + u32 size_in_bytes, cur_size_in_bytes, tmp; int i, num_loops; int r = 0; @@ -2674,40 +2803,46 @@ int r600_copy_dma(struct radeon_device *rdev, return r; } - size_in_dw = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT) / 4; - num_loops = DIV_ROUND_UP(size_in_dw, 0xFFFE); - r = radeon_ring_lock(rdev, ring, num_loops * 4 + 8); + size_in_bytes = (num_gpu_pages << RADEON_GPU_PAGE_SHIFT); + num_loops = DIV_ROUND_UP(size_in_bytes, 0x1fffff); + r = radeon_ring_lock(rdev, ring, num_loops * 6 + 24); if (r) { DRM_ERROR("radeon: moving bo (%d).\n", r); radeon_semaphore_free(rdev, &sem, NULL); return r; } - if (radeon_fence_need_sync(*fence, ring->idx)) { - radeon_semaphore_sync_rings(rdev, sem, (*fence)->ring, - ring->idx); - radeon_fence_note_sync(*fence, ring->idx); - } else { - radeon_semaphore_free(rdev, &sem, NULL); - } + radeon_semaphore_sync_to(sem, *fence); + radeon_semaphore_sync_rings(rdev, sem, ring->idx); + radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(ring, (WAIT_UNTIL - PACKET3_SET_CONFIG_REG_OFFSET) >> 2); + radeon_ring_write(ring, WAIT_3D_IDLE_bit); for (i = 0; i < num_loops; i++) { - cur_size_in_dw = size_in_dw; - if (cur_size_in_dw > 0xFFFE) - cur_size_in_dw = 0xFFFE; - size_in_dw -= cur_size_in_dw; - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_COPY, 0, 0, cur_size_in_dw)); - radeon_ring_write(ring, dst_offset & 0xfffffffc); - radeon_ring_write(ring, src_offset & 0xfffffffc); - radeon_ring_write(ring, (((upper_32_bits(dst_offset) & 0xff) << 16) | - (upper_32_bits(src_offset) & 0xff))); - src_offset += cur_size_in_dw * 4; - dst_offset += cur_size_in_dw * 4; + cur_size_in_bytes = size_in_bytes; + if (cur_size_in_bytes > 0x1fffff) + cur_size_in_bytes = 0x1fffff; + size_in_bytes -= cur_size_in_bytes; + tmp = upper_32_bits(src_offset) & 0xff; + if (size_in_bytes == 0) + tmp |= PACKET3_CP_DMA_CP_SYNC; + radeon_ring_write(ring, PACKET3(PACKET3_CP_DMA, 4)); + radeon_ring_write(ring, lower_32_bits(src_offset)); + radeon_ring_write(ring, tmp); + radeon_ring_write(ring, lower_32_bits(dst_offset)); + radeon_ring_write(ring, upper_32_bits(dst_offset) & 0xff); + radeon_ring_write(ring, cur_size_in_bytes); + src_offset += cur_size_in_bytes; + dst_offset += cur_size_in_bytes; } + radeon_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(ring, (WAIT_UNTIL - PACKET3_SET_CONFIG_REG_OFFSET) >> 2); + radeon_ring_write(ring, WAIT_CP_DMA_IDLE_bit); r = radeon_fence_emit(rdev, fence, ring->idx); if (r) { radeon_ring_unlock_undo(rdev, ring); + radeon_semaphore_free(rdev, &sem, NULL); return r; } @@ -2738,19 +2873,13 @@ static int r600_startup(struct radeon_device *rdev) /* enable pcie gen2 link */ r600_pcie_gen2_enable(rdev); - if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw) { - r = r600_init_microcode(rdev); - if (r) { - DRM_ERROR("Failed to load firmware!\n"); - return r; - } - } - + /* scratch needs to be initialized before MC */ r = r600_vram_scratch_init(rdev); if (r) return r; r600_mc_program(rdev); + if (rdev->flags & RADEON_IS_AGP) { r600_agp_enable(rdev); } else { @@ -2759,12 +2888,6 @@ static int r600_startup(struct radeon_device *rdev) return r; } r600_gpu_init(rdev); - r = r600_blit_init(rdev); - if (r) { - r600_blit_fini(rdev); - rdev->asic->copy.copy = NULL; - dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r); - } /* allocate wb buffer */ r = radeon_wb_init(rdev); @@ -2777,13 +2900,13 @@ static int r600_startup(struct radeon_device *rdev) return r; } - r = radeon_fence_driver_start_ring(rdev, R600_RING_TYPE_DMA_INDEX); - if (r) { - dev_err(rdev->dev, "failed initializing DMA fences (%d).\n", r); - return r; + /* Enable IRQ */ + if (!rdev->irq.installed) { + r = radeon_irq_kms_init(rdev); + if (r) + return r; } - /* Enable IRQ */ r = r600_irq_init(rdev); if (r) { DRM_ERROR("radeon: IH init failed (%d).\n", r); @@ -2794,15 +2917,7 @@ static int r600_startup(struct radeon_device *rdev) ring = &rdev->ring[RADEON_RING_TYPE_GFX_INDEX]; r = radeon_ring_init(rdev, ring, ring->ring_size, RADEON_WB_CP_RPTR_OFFSET, - R600_CP_RB_RPTR, R600_CP_RB_WPTR, - 0, 0xfffff, RADEON_CP_PACKET2); - if (r) - return r; - - ring = &rdev->ring[R600_RING_TYPE_DMA_INDEX]; - r = radeon_ring_init(rdev, ring, ring->ring_size, R600_WB_DMA_RPTR_OFFSET, - DMA_RB_RPTR, DMA_RB_WPTR, - 2, 0x3fffc, DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0)); + RADEON_CP_PACKET2); if (r) return r; @@ -2813,10 +2928,6 @@ static int r600_startup(struct radeon_device *rdev) if (r) return r; - r = r600_dma_resume(rdev); - if (r) - return r; - r = radeon_ib_pool_init(rdev); if (r) { dev_err(rdev->dev, "IB initialization failed (%d).\n", r); @@ -2857,6 +2968,9 @@ int r600_resume(struct radeon_device *rdev) /* post card */ atom_asic_init(rdev->mode_info.atom_context); + if (rdev->pm.pm_method == PM_METHOD_DPM) + radeon_pm_resume(rdev); + rdev->accel_working = true; r = r600_startup(rdev); if (r) { @@ -2870,9 +2984,9 @@ int r600_resume(struct radeon_device *rdev) int r600_suspend(struct radeon_device *rdev) { + radeon_pm_suspend(rdev); r600_audio_fini(rdev); r600_cp_stop(rdev); - r600_dma_stop(rdev); r600_irq_suspend(rdev); radeon_wb_disable(rdev); r600_pcie_gart_disable(rdev); @@ -2938,16 +3052,20 @@ int r600_init(struct radeon_device *rdev) if (r) return r; - r = radeon_irq_kms_init(rdev); - if (r) - return r; + if (!rdev->me_fw || !rdev->pfp_fw || !rdev->rlc_fw) { + r = r600_init_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load firmware!\n"); + return r; + } + } + + /* Initialize power management */ + radeon_pm_init(rdev); rdev->ring[RADEON_RING_TYPE_GFX_INDEX].ring_obj = NULL; r600_ring_init(rdev, &rdev->ring[RADEON_RING_TYPE_GFX_INDEX], 1024 * 1024); - rdev->ring[R600_RING_TYPE_DMA_INDEX].ring_obj = NULL; - r600_ring_init(rdev, &rdev->ring[R600_RING_TYPE_DMA_INDEX], 64 * 1024); - rdev->ih.ring_obj = NULL; r600_ih_ring_init(rdev, 64 * 1024); @@ -2960,7 +3078,6 @@ int r600_init(struct radeon_device *rdev) if (r) { dev_err(rdev->dev, "disabling GPU acceleration\n"); r600_cp_fini(rdev); - r600_dma_fini(rdev); r600_irq_fini(rdev); radeon_wb_fini(rdev); radeon_ib_pool_fini(rdev); @@ -2974,10 +3091,9 @@ int r600_init(struct radeon_device *rdev) void r600_fini(struct radeon_device *rdev) { + radeon_pm_fini(rdev); r600_audio_fini(rdev); - r600_blit_fini(rdev); r600_cp_fini(rdev); - r600_dma_fini(rdev); r600_irq_fini(rdev); radeon_wb_fini(rdev); radeon_ib_pool_fini(rdev); @@ -3080,104 +3196,6 @@ free_scratch: return r; } -/** - * r600_dma_ib_test - test an IB on the DMA engine - * - * @rdev: radeon_device pointer - * @ring: radeon_ring structure holding ring information - * - * Test a simple IB in the DMA ring (r6xx-SI). - * Returns 0 on success, error on failure. - */ -int r600_dma_ib_test(struct radeon_device *rdev, struct radeon_ring *ring) -{ - struct radeon_ib ib; - unsigned i; - int r; - void __iomem *ptr = (void *)rdev->vram_scratch.ptr; - u32 tmp = 0; - - if (!ptr) { - DRM_ERROR("invalid vram scratch pointer\n"); - return -EINVAL; - } - - tmp = 0xCAFEDEAD; - writel(tmp, ptr); - - r = radeon_ib_get(rdev, ring->idx, &ib, NULL, 256); - if (r) { - DRM_ERROR("radeon: failed to get ib (%d).\n", r); - return r; - } - - ib.ptr[0] = DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 1); - ib.ptr[1] = rdev->vram_scratch.gpu_addr & 0xfffffffc; - ib.ptr[2] = upper_32_bits(rdev->vram_scratch.gpu_addr) & 0xff; - ib.ptr[3] = 0xDEADBEEF; - ib.length_dw = 4; - - r = radeon_ib_schedule(rdev, &ib, NULL); - if (r) { - radeon_ib_free(rdev, &ib); - DRM_ERROR("radeon: failed to schedule ib (%d).\n", r); - return r; - } - r = radeon_fence_wait(ib.fence, false); - if (r) { - DRM_ERROR("radeon: fence wait failed (%d).\n", r); - return r; - } - for (i = 0; i < rdev->usec_timeout; i++) { - tmp = readl(ptr); - if (tmp == 0xDEADBEEF) - break; - DRM_UDELAY(1); - } - if (i < rdev->usec_timeout) { - DRM_INFO("ib test on ring %d succeeded in %u usecs\n", ib.fence->ring, i); - } else { - DRM_ERROR("radeon: ib test failed (0x%08X)\n", tmp); - r = -EINVAL; - } - radeon_ib_free(rdev, &ib); - return r; -} - -/** - * r600_dma_ring_ib_execute - Schedule an IB on the DMA engine - * - * @rdev: radeon_device pointer - * @ib: IB object to schedule - * - * Schedule an IB in the DMA ring (r6xx-r7xx). - */ -void r600_dma_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) -{ - struct radeon_ring *ring = &rdev->ring[ib->ring]; - - if (rdev->wb.enabled) { - u32 next_rptr = ring->wptr + 4; - while ((next_rptr & 7) != 5) - next_rptr++; - next_rptr += 3; - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_WRITE, 0, 0, 1)); - radeon_ring_write(ring, ring->next_rptr_gpu_addr & 0xfffffffc); - radeon_ring_write(ring, upper_32_bits(ring->next_rptr_gpu_addr) & 0xff); - radeon_ring_write(ring, next_rptr); - } - - /* The indirect buffer packet must end on an 8 DW boundary in the DMA ring. - * Pad as necessary with NOPs. - */ - while ((ring->wptr & 7) != 5) - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_NOP, 0, 0, 0)); - radeon_ring_write(ring, DMA_PACKET(DMA_PACKET_INDIRECT_BUFFER, 0, 0, 0)); - radeon_ring_write(ring, (ib->gpu_addr & 0xFFFFFFE0)); - radeon_ring_write(ring, (ib->length_dw << 16) | (upper_32_bits(ib->gpu_addr) & 0xFF)); - -} - /* * Interrupts * @@ -3194,7 +3212,7 @@ void r600_ih_ring_init(struct radeon_device *rdev, unsigned ring_size) u32 rb_bufsz; /* Align ring size */ - rb_bufsz = drm_order(ring_size / 4); + rb_bufsz = order_base_2(ring_size / 4); ring_size = (1 << rb_bufsz) * 4; rdev->ih.ring_size = ring_size; rdev->ih.ptr_mask = rdev->ih.ring_size - 1; @@ -3274,7 +3292,7 @@ static void r600_rlc_start(struct radeon_device *rdev) WREG32(RLC_CNTL, RLC_ENABLE); } -static int r600_rlc_init(struct radeon_device *rdev) +static int r600_rlc_resume(struct radeon_device *rdev) { u32 i; const __be32 *fw_data; @@ -3286,45 +3304,22 @@ static int r600_rlc_init(struct radeon_device *rdev) WREG32(RLC_HB_CNTL, 0); - if (rdev->family == CHIP_ARUBA) { - WREG32(TN_RLC_SAVE_AND_RESTORE_BASE, rdev->rlc.save_restore_gpu_addr >> 8); - WREG32(TN_RLC_CLEAR_STATE_RESTORE_BASE, rdev->rlc.clear_state_gpu_addr >> 8); - } - if (rdev->family <= CHIP_CAYMAN) { - WREG32(RLC_HB_BASE, 0); - WREG32(RLC_HB_RPTR, 0); - WREG32(RLC_HB_WPTR, 0); - } - if (rdev->family <= CHIP_CAICOS) { - WREG32(RLC_HB_WPTR_LSB_ADDR, 0); - WREG32(RLC_HB_WPTR_MSB_ADDR, 0); - } + WREG32(RLC_HB_BASE, 0); + WREG32(RLC_HB_RPTR, 0); + WREG32(RLC_HB_WPTR, 0); + WREG32(RLC_HB_WPTR_LSB_ADDR, 0); + WREG32(RLC_HB_WPTR_MSB_ADDR, 0); WREG32(RLC_MC_CNTL, 0); WREG32(RLC_UCODE_CNTL, 0); fw_data = (const __be32 *)rdev->rlc_fw->data; - if (rdev->family >= CHIP_ARUBA) { - for (i = 0; i < ARUBA_RLC_UCODE_SIZE; i++) { - WREG32(RLC_UCODE_ADDR, i); - WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); - } - } else if (rdev->family >= CHIP_CAYMAN) { - for (i = 0; i < CAYMAN_RLC_UCODE_SIZE; i++) { - WREG32(RLC_UCODE_ADDR, i); - WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); - } - } else if (rdev->family >= CHIP_CEDAR) { - for (i = 0; i < EVERGREEN_RLC_UCODE_SIZE; i++) { - WREG32(RLC_UCODE_ADDR, i); - WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); - } - } else if (rdev->family >= CHIP_RV770) { + if (rdev->family >= CHIP_RV770) { for (i = 0; i < R700_RLC_UCODE_SIZE; i++) { WREG32(RLC_UCODE_ADDR, i); WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); } } else { - for (i = 0; i < RLC_UCODE_SIZE; i++) { + for (i = 0; i < R600_RLC_UCODE_SIZE; i++) { WREG32(RLC_UCODE_ADDR, i); WREG32(RLC_UCODE_DATA, be32_to_cpup(fw_data++)); } @@ -3432,7 +3427,10 @@ int r600_irq_init(struct radeon_device *rdev) r600_disable_interrupts(rdev); /* init rlc */ - ret = r600_rlc_init(rdev); + if (rdev->family >= CHIP_CEDAR) + ret = evergreen_rlc_resume(rdev); + else + ret = r600_rlc_resume(rdev); if (ret) { r600_ih_ring_fini(rdev); return ret; @@ -3451,7 +3449,7 @@ int r600_irq_init(struct radeon_device *rdev) WREG32(INTERRUPT_CNTL, interrupt_cntl); WREG32(IH_RB_BASE, rdev->ih.gpu_addr >> 8); - rb_bufsz = drm_order(rdev->ih.ring_size / 4); + rb_bufsz = order_base_2(rdev->ih.ring_size / 4); ih_rb_cntl = (IH_WPTR_OVERFLOW_ENABLE | IH_WPTR_OVERFLOW_CLEAR | @@ -3511,8 +3509,8 @@ int r600_irq_set(struct radeon_device *rdev) u32 hpd1, hpd2, hpd3, hpd4 = 0, hpd5 = 0, hpd6 = 0; u32 grbm_int_cntl = 0; u32 hdmi0, hdmi1; - u32 d1grph = 0, d2grph = 0; u32 dma_cntl; + u32 thermal_int = 0; if (!rdev->irq.installed) { WARN(1, "Can't enable IRQ/MSI because no handler is installed\n"); @@ -3547,8 +3545,21 @@ int r600_irq_set(struct radeon_device *rdev) hdmi0 = RREG32(HDMI0_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK; hdmi1 = RREG32(HDMI1_AUDIO_PACKET_CONTROL) & ~HDMI0_AZ_FORMAT_WTRIG_MASK; } + dma_cntl = RREG32(DMA_CNTL) & ~TRAP_ENABLE; + if ((rdev->family > CHIP_R600) && (rdev->family < CHIP_RV770)) { + thermal_int = RREG32(CG_THERMAL_INT) & + ~(THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW); + } else if (rdev->family >= CHIP_RV770) { + thermal_int = RREG32(RV770_CG_THERMAL_INT) & + ~(THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW); + } + if (rdev->irq.dpm_thermal) { + DRM_DEBUG("dpm thermal\n"); + thermal_int |= THERM_INT_MASK_HIGH | THERM_INT_MASK_LOW; + } + if (atomic_read(&rdev->irq.ring_int[RADEON_RING_TYPE_GFX_INDEX])) { DRM_DEBUG("r600_irq_set: sw int\n"); cp_int_cntl |= RB_INT_ENABLE; @@ -3606,8 +3617,8 @@ int r600_irq_set(struct radeon_device *rdev) WREG32(CP_INT_CNTL, cp_int_cntl); WREG32(DMA_CNTL, dma_cntl); WREG32(DxMODE_INT_MASK, mode_int); - WREG32(D1GRPH_INTERRUPT_CONTROL, d1grph); - WREG32(D2GRPH_INTERRUPT_CONTROL, d2grph); + WREG32(D1GRPH_INTERRUPT_CONTROL, DxGRPH_PFLIP_INT_MASK); + WREG32(D2GRPH_INTERRUPT_CONTROL, DxGRPH_PFLIP_INT_MASK); WREG32(GRBM_INT_CNTL, grbm_int_cntl); if (ASIC_IS_DCE3(rdev)) { WREG32(DC_HPD1_INT_CONTROL, hpd1); @@ -3630,6 +3641,11 @@ int r600_irq_set(struct radeon_device *rdev) WREG32(HDMI0_AUDIO_PACKET_CONTROL, hdmi0); WREG32(HDMI1_AUDIO_PACKET_CONTROL, hdmi1); } + if ((rdev->family > CHIP_R600) && (rdev->family < CHIP_RV770)) { + WREG32(CG_THERMAL_INT, thermal_int); + } else if (rdev->family >= CHIP_RV770) { + WREG32(RV770_CG_THERMAL_INT, thermal_int); + } return 0; } @@ -3779,6 +3795,7 @@ static u32 r600_get_ih_wptr(struct radeon_device *rdev) tmp = RREG32(IH_RB_CNTL); tmp |= IH_WPTR_OVERFLOW_CLEAR; WREG32(IH_RB_CNTL, tmp); + wptr &= ~RB_OVERFLOW; } return (wptr & rdev->ih.ptr_mask); } @@ -3821,6 +3838,7 @@ int r600_irq_process(struct radeon_device *rdev) u32 ring_index; bool queue_hotplug = false; bool queue_hdmi = false; + bool queue_thermal = false; if (!rdev->ih.enabled || rdev->shutdown) return IRQ_NONE; @@ -3862,7 +3880,7 @@ restart_ih: wake_up(&rdev->irq.vblank_queue); } if (atomic_read(&rdev->irq.pflip[0])) - radeon_crtc_handle_flip(rdev, 0); + radeon_crtc_handle_vblank(rdev, 0); rdev->irq.stat_regs.r600.disp_int &= ~LB_D1_VBLANK_INTERRUPT; DRM_DEBUG("IH: D1 vblank\n"); } @@ -3888,7 +3906,7 @@ restart_ih: wake_up(&rdev->irq.vblank_queue); } if (atomic_read(&rdev->irq.pflip[1])) - radeon_crtc_handle_flip(rdev, 1); + radeon_crtc_handle_vblank(rdev, 1); rdev->irq.stat_regs.r600.disp_int &= ~LB_D2_VBLANK_INTERRUPT; DRM_DEBUG("IH: D2 vblank\n"); } @@ -3904,6 +3922,14 @@ restart_ih: break; } break; + case 9: /* D1 pflip */ + DRM_DEBUG("IH: D1 flip\n"); + radeon_crtc_handle_flip(rdev, 0); + break; + case 11: /* D2 pflip */ + DRM_DEBUG("IH: D2 flip\n"); + radeon_crtc_handle_flip(rdev, 1); + break; case 19: /* HPD/DAC hotplug */ switch (src_data) { case 0: @@ -3974,6 +4000,10 @@ restart_ih: break; } break; + case 124: /* UVD */ + DRM_DEBUG("IH: UVD int: 0x%08x\n", src_data); + radeon_fence_process(rdev, R600_RING_TYPE_UVD_INDEX); + break; case 176: /* CP_INT in ring buffer */ case 177: /* CP_INT in IB1 */ case 178: /* CP_INT in IB2 */ @@ -3988,6 +4018,16 @@ restart_ih: DRM_DEBUG("IH: DMA trap\n"); radeon_fence_process(rdev, R600_RING_TYPE_DMA_INDEX); break; + case 230: /* thermal low to high */ + DRM_DEBUG("IH: thermal low to high\n"); + rdev->pm.dpm.thermal.high_to_low = false; + queue_thermal = true; + break; + case 231: /* thermal high to low */ + DRM_DEBUG("IH: thermal high to low\n"); + rdev->pm.dpm.thermal.high_to_low = true; + queue_thermal = true; + break; case 233: /* GUI IDLE */ DRM_DEBUG("IH: GUI idle\n"); break; @@ -4004,6 +4044,8 @@ restart_ih: schedule_work(&rdev->hotplug_work); if (queue_hdmi) schedule_work(&rdev->audio_work); + if (queue_thermal && rdev->pm.dpm_enabled) + schedule_work(&rdev->pm.dpm.thermal.work); rdev->ih.rptr = rptr; WREG32(IH_RB_RPTR, rdev->ih.rptr); atomic_set(&rdev->ih.lock, 0); @@ -4076,7 +4118,7 @@ void r600_ioctl_wait_idle(struct radeon_device *rdev, struct radeon_bo *bo) void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes) { - u32 link_width_cntl, mask, target_reg; + u32 link_width_cntl, mask; if (rdev->flags & RADEON_IS_IGP) return; @@ -4088,7 +4130,7 @@ void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes) if (ASIC_IS_X2(rdev)) return; - /* FIXME wait for idle */ + radeon_gui_idle(rdev); switch (lanes) { case 0: @@ -4107,53 +4149,24 @@ void r600_set_pcie_lanes(struct radeon_device *rdev, int lanes) mask = RADEON_PCIE_LC_LINK_WIDTH_X8; break; case 12: + /* not actually supported */ mask = RADEON_PCIE_LC_LINK_WIDTH_X12; break; case 16: - default: mask = RADEON_PCIE_LC_LINK_WIDTH_X16; break; - } - - link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL); - - if ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) == - (mask << RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT)) - return; - - if (link_width_cntl & R600_PCIE_LC_UPCONFIGURE_DIS) + default: + DRM_ERROR("invalid pcie lane request: %d\n", lanes); return; + } - link_width_cntl &= ~(RADEON_PCIE_LC_LINK_WIDTH_MASK | - RADEON_PCIE_LC_RECONFIG_NOW | - R600_PCIE_LC_RENEGOTIATE_EN | - R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE); - link_width_cntl |= mask; - - WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); - - /* some northbridges can renegotiate the link rather than requiring - * a complete re-config. - * e.g., AMD 780/790 northbridges (pci ids: 0x5956, 0x5957, 0x5958, etc.) - */ - if (link_width_cntl & R600_PCIE_LC_RENEGOTIATION_SUPPORT) - link_width_cntl |= R600_PCIE_LC_RENEGOTIATE_EN | R600_PCIE_LC_UPCONFIGURE_SUPPORT; - else - link_width_cntl |= R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE; - - WREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL, (link_width_cntl | - RADEON_PCIE_LC_RECONFIG_NOW)); - - if (rdev->family >= CHIP_RV770) - target_reg = R700_TARGET_AND_CURRENT_PROFILE_INDEX; - else - target_reg = R600_TARGET_AND_CURRENT_PROFILE_INDEX; - - /* wait for lane set to complete */ - link_width_cntl = RREG32(target_reg); - while (link_width_cntl == 0xffffffff) - link_width_cntl = RREG32(target_reg); + link_width_cntl = RREG32_PCIE_PORT(RADEON_PCIE_LC_LINK_WIDTH_CNTL); + link_width_cntl &= ~RADEON_PCIE_LC_LINK_WIDTH_MASK; + link_width_cntl |= mask << RADEON_PCIE_LC_LINK_WIDTH_SHIFT; + link_width_cntl |= (RADEON_PCIE_LC_RECONFIG_NOW | + R600_PCIE_LC_RECONFIG_ARC_MISSING_ESCAPE); + WREG32_PCIE_PORT(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); } int r600_get_pcie_lanes(struct radeon_device *rdev) @@ -4170,13 +4183,11 @@ int r600_get_pcie_lanes(struct radeon_device *rdev) if (ASIC_IS_X2(rdev)) return 0; - /* FIXME wait for idle */ + radeon_gui_idle(rdev); - link_width_cntl = RREG32_PCIE_P(RADEON_PCIE_LC_LINK_WIDTH_CNTL); + link_width_cntl = RREG32_PCIE_PORT(RADEON_PCIE_LC_LINK_WIDTH_CNTL); switch ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) >> RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT) { - case RADEON_PCIE_LC_LINK_WIDTH_X0: - return 0; case RADEON_PCIE_LC_LINK_WIDTH_X1: return 1; case RADEON_PCIE_LC_LINK_WIDTH_X2: @@ -4185,6 +4196,10 @@ int r600_get_pcie_lanes(struct radeon_device *rdev) return 4; case RADEON_PCIE_LC_LINK_WIDTH_X8: return 8; + case RADEON_PCIE_LC_LINK_WIDTH_X12: + /* not actually supported */ + return 12; + case RADEON_PCIE_LC_LINK_WIDTH_X0: case RADEON_PCIE_LC_LINK_WIDTH_X16: default: return 16; @@ -4195,8 +4210,6 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev) { u32 link_width_cntl, lanes, speed_cntl, training_cntl, tmp; u16 link_cntl2; - u32 mask; - int ret; if (radeon_pcie_gen2 == 0) return; @@ -4215,14 +4228,11 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev) if (rdev->family <= CHIP_R600) return; - ret = drm_pcie_get_speed_cap_mask(rdev->ddev, &mask); - if (ret != 0) - return; - - if (!(mask & DRM_PCIE_SPEED_50)) + if ((rdev->pdev->bus->max_bus_speed != PCIE_SPEED_5_0GT) && + (rdev->pdev->bus->max_bus_speed != PCIE_SPEED_8_0GT)) return; - speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL); if (speed_cntl & LC_CURRENT_DATA_RATE) { DRM_INFO("PCIE gen 2 link speeds already enabled\n"); return; @@ -4235,23 +4245,23 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev) (rdev->family == CHIP_RV620) || (rdev->family == CHIP_RV635)) { /* advertise upconfig capability */ - link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL); link_width_cntl &= ~LC_UPCONFIGURE_DIS; - WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); - link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL); if (link_width_cntl & LC_RENEGOTIATION_SUPPORT) { lanes = (link_width_cntl & LC_LINK_WIDTH_RD_MASK) >> LC_LINK_WIDTH_RD_SHIFT; link_width_cntl &= ~(LC_LINK_WIDTH_MASK | LC_RECONFIG_ARC_MISSING_ESCAPE); link_width_cntl |= lanes | LC_RECONFIG_NOW | LC_RENEGOTIATE_EN; - WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); } else { link_width_cntl |= LC_UPCONFIGURE_DIS; - WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); } } - speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL); if ((speed_cntl & LC_OTHER_SIDE_EVER_SENT_GEN2) && (speed_cntl & LC_OTHER_SIDE_SUPPORTS_GEN2)) { @@ -4272,7 +4282,7 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev) speed_cntl &= ~LC_VOLTAGE_TIMER_SEL_MASK; speed_cntl &= ~LC_FORCE_DIS_HW_SPEED_CHANGE; speed_cntl |= LC_FORCE_EN_HW_SPEED_CHANGE; - WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl); tmp = RREG32(0x541c); WREG32(0x541c, tmp | 0x8); @@ -4286,39 +4296,39 @@ static void r600_pcie_gen2_enable(struct radeon_device *rdev) if ((rdev->family == CHIP_RV670) || (rdev->family == CHIP_RV620) || (rdev->family == CHIP_RV635)) { - training_cntl = RREG32_PCIE_P(PCIE_LC_TRAINING_CNTL); + training_cntl = RREG32_PCIE_PORT(PCIE_LC_TRAINING_CNTL); training_cntl &= ~LC_POINT_7_PLUS_EN; - WREG32_PCIE_P(PCIE_LC_TRAINING_CNTL, training_cntl); + WREG32_PCIE_PORT(PCIE_LC_TRAINING_CNTL, training_cntl); } else { - speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL); speed_cntl &= ~LC_TARGET_LINK_SPEED_OVERRIDE_EN; - WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl); } - speed_cntl = RREG32_PCIE_P(PCIE_LC_SPEED_CNTL); + speed_cntl = RREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL); speed_cntl |= LC_GEN2_EN_STRAP; - WREG32_PCIE_P(PCIE_LC_SPEED_CNTL, speed_cntl); + WREG32_PCIE_PORT(PCIE_LC_SPEED_CNTL, speed_cntl); } else { - link_width_cntl = RREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL); + link_width_cntl = RREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL); /* XXX: only disable it if gen1 bridge vendor == 0x111d or 0x1106 */ if (1) link_width_cntl |= LC_UPCONFIGURE_DIS; else link_width_cntl &= ~LC_UPCONFIGURE_DIS; - WREG32_PCIE_P(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); + WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); } } /** - * r600_get_gpu_clock - return GPU clock counter snapshot + * r600_get_gpu_clock_counter - return GPU clock counter snapshot * * @rdev: radeon_device pointer * * Fetches a GPU clock counter snapshot (R6xx-cayman). * Returns the 64 bit clock counter snapshot. */ -uint64_t r600_get_gpu_clock(struct radeon_device *rdev) +uint64_t r600_get_gpu_clock_counter(struct radeon_device *rdev) { uint64_t clock; |