static void
__gen9_force_wake_get(struct drm_i915_private *dev_priv, int fw_engine)
{
/* Check for Render Engine */
if (FORCEWAKE_RENDER & fw_engine) {
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_RENDER_GEN9) &
FORCEWAKE_KERNEL) == 0,
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: Render forcewake old ack to clear.\n");
__raw_i915_write32(dev_priv, FORCEWAKE_RENDER_GEN9,
_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL));
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_RENDER_GEN9) &
FORCEWAKE_KERNEL),
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: waiting for Render to ack.\n");
}
/* Check for Media Engine */
if (FORCEWAKE_MEDIA & fw_engine) {
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_MEDIA_GEN9) &
FORCEWAKE_KERNEL) == 0,
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: Media forcewake old ack to clear.\n");
__raw_i915_write32(dev_priv, FORCEWAKE_MEDIA_GEN9,
_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL));
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_MEDIA_GEN9) &
FORCEWAKE_KERNEL),
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: waiting for Media to ack.\n");
}
/* Check for Blitter Engine */
if (FORCEWAKE_BLITTER & fw_engine) {
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_BLITTER_GEN9) &
FORCEWAKE_KERNEL) == 0,
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: Blitter forcewake old ack to clear.\n");
__raw_i915_write32(dev_priv, FORCEWAKE_BLITTER_GEN9,
_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL));
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_BLITTER_GEN9) &
FORCEWAKE_KERNEL),
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: waiting for Blitter to ack.\n");
}
}
static void gen9_gt_workarounds_apply(struct drm_i915_private *dev_priv)
{
/* WaContextSwitchWithConcurrentTLBInvalidate:skl,bxt,kbl,glk,cfl */
I915_WRITE(GEN9_CSFE_CHICKEN1_RCS,
_MASKED_BIT_ENABLE(GEN9_PREEMPT_GPGPU_SYNC_SWITCH_DISABLE));
/* WaEnableLbsSlaRetryTimerDecrement:skl,bxt,kbl,glk,cfl */
I915_WRITE(BDW_SCRATCH1, I915_READ(BDW_SCRATCH1) |
GEN9_LBS_SLA_RETRY_TIMER_DECREMENT_ENABLE);
/* WaDisableKillLogic:bxt,skl,kbl */
if (!IS_COFFEELAKE(dev_priv))
I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) |
ECOCHK_DIS_TLB);
if (HAS_LLC(dev_priv)) {
/* WaCompressedResourceSamplerPbeMediaNewHashMode:skl,kbl
*
* Must match Display Engine. See
* WaCompressedResourceDisplayNewHashMode.
*/
I915_WRITE(MMCD_MISC_CTRL,
I915_READ(MMCD_MISC_CTRL) |
MMCD_PCLA |
MMCD_HOTSPOT_EN);
}
/* WaDisableHDCInvalidation:skl,bxt,kbl,cfl */
I915_WRITE(GAM_ECOCHK, I915_READ(GAM_ECOCHK) |
BDW_DISABLE_HDC_INVALIDATION);
/* WaProgramL3SqcReg1DefaultForPerf:bxt,glk */
if (IS_GEN9_LP(dev_priv)) {
u32 val = I915_READ(GEN8_L3SQCREG1);
val &= ~L3_PRIO_CREDITS_MASK;
val |= L3_GENERAL_PRIO_CREDITS(62) | L3_HIGH_PRIO_CREDITS(2);
I915_WRITE(GEN8_L3SQCREG1, val);
}
/* WaOCLCoherentLineFlush:skl,bxt,kbl,cfl */
I915_WRITE(GEN8_L3SQCREG4,
I915_READ(GEN8_L3SQCREG4) | GEN8_LQSC_FLUSH_COHERENT_LINES);
/* WaEnablePreemptionGranularityControlByUMD:skl,bxt,kbl,cfl,[cnl] */
I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1,
_MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL));
}
static void direct_interrupts_to_guc(struct drm_i915_private *dev_priv)
{
struct intel_engine_cs *engine;
int irqs;
u32 tmp;
/* tell all command streamers to forward interrupts (but not vblank) to GuC */
irqs = _MASKED_BIT_ENABLE(GFX_INTERRUPT_STEERING);
for_each_engine(engine, dev_priv)
I915_WRITE(RING_MODE_GEN7(engine), irqs);
/* route USER_INTERRUPT to Host, all others are sent to GuC. */
irqs = GT_RENDER_USER_INTERRUPT << GEN8_RCS_IRQ_SHIFT |
GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT;
/* These three registers have the same bit definitions */
I915_WRITE(GUC_BCS_RCS_IER, ~irqs);
I915_WRITE(GUC_VCS2_VCS1_IER, ~irqs);
I915_WRITE(GUC_WD_VECS_IER, ~irqs);
/*
* If GuC has routed PM interrupts to itself, don't keep it.
* and keep other interrupts those are unmasked by GuC.
*/
tmp = I915_READ(GEN6_PMINTRMSK);
if (tmp & GEN8_PMINTR_REDIRECT_TO_NON_DISP) {
dev_priv->rps.pm_intr_keep |= ~(tmp & ~GEN8_PMINTR_REDIRECT_TO_NON_DISP);
dev_priv->rps.pm_intr_keep &= ~GEN8_PMINTR_REDIRECT_TO_NON_DISP;
}
}
/*
* Transfer the firmware image to RAM for execution by the microcontroller.
*
* Architecturally, the DMA engine is bidirectional, and can potentially even
* transfer between GTT locations. This functionality is left out of the API
* for now as there is no need for it.
*/
static int guc_xfer_ucode(struct intel_guc *guc)
{
struct drm_i915_private *dev_priv = guc_to_i915(guc);
struct intel_uc_fw *guc_fw = &guc->fw;
unsigned long offset;
/*
* The header plus uCode will be copied to WOPCM via DMA, excluding any
* other components
*/
I915_WRITE(DMA_COPY_SIZE, guc_fw->header_size + guc_fw->ucode_size);
/* Set the source address for the new blob */
offset = intel_uc_fw_ggtt_offset(guc_fw) + guc_fw->header_offset;
I915_WRITE(DMA_ADDR_0_LOW, lower_32_bits(offset));
I915_WRITE(DMA_ADDR_0_HIGH, upper_32_bits(offset) & 0xFFFF);
/*
* Set the DMA destination. Current uCode expects the code to be
* loaded at 8k; locations below this are used for the stack.
*/
I915_WRITE(DMA_ADDR_1_LOW, 0x2000);
I915_WRITE(DMA_ADDR_1_HIGH, DMA_ADDRESS_SPACE_WOPCM);
/* Finally start the DMA */
I915_WRITE(DMA_CTRL, _MASKED_BIT_ENABLE(UOS_MOVE | START_DMA));
return guc_wait_ucode(guc);
}
static void __gen6_gt_force_wake_mt_get(struct drm_i915_private *dev_priv,
int fw_engine)
{
u32 forcewake_ack;
if (IS_HASWELL(dev_priv->dev) || IS_GEN8(dev_priv->dev))
forcewake_ack = FORCEWAKE_ACK_HSW;
else
forcewake_ack = FORCEWAKE_MT_ACK;
if (wait_for_atomic((__raw_i915_read32(dev_priv, forcewake_ack) & FORCEWAKE_KERNEL) == 0,
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out waiting for forcewake old ack to clear.\n");
__raw_i915_write32(dev_priv, FORCEWAKE_MT,
_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL));
/* something from same cacheline, but !FORCEWAKE_MT */
__raw_posting_read(dev_priv, ECOBUS);
if (wait_for_atomic((__raw_i915_read32(dev_priv, forcewake_ack) & FORCEWAKE_KERNEL),
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out waiting for forcewake to ack request.\n");
/* WaRsForcewakeWaitTC0:ivb,hsw */
if (INTEL_INFO(dev_priv->dev)->gen < 8)
__gen6_gt_wait_for_thread_c0(dev_priv);
}
static inline void nugpgpu_ring_render_stop(struct nugpgpu_private *gpu_priv)
{
RING_WRITE_MODE(RING, _MASKED_BIT_ENABLE(STOP_RING));
if(wait_for((RING_READ_MODE(RING) & MODE_IDLE) != 0, 10000))
printk(LOG_ERR "MI_MODE failed to stop\n" LOG_END);
else
printk(LOG_INFO "MI_MODE ring stopped successfully\n" LOG_END);
}
/*
* Transfer the firmware image to RAM for execution by the microcontroller.
*
* Architecturally, the DMA engine is bidirectional, and can potentially even
* transfer between GTT locations. This functionality is left out of the API
* for now as there is no need for it.
*
* Note that GuC needs the CSS header plus uKernel code to be copied by the
* DMA engine in one operation, whereas the RSA signature is loaded via MMIO.
*/
static int guc_ucode_xfer_dma(struct drm_i915_private *dev_priv,
struct i915_vma *vma)
{
struct intel_uc_fw *guc_fw = &dev_priv->guc.fw;
unsigned long offset;
struct sg_table *sg = vma->pages;
u32 status, rsa[UOS_RSA_SCRATCH_MAX_COUNT];
int i, ret = 0;
/* where RSA signature starts */
offset = guc_fw->rsa_offset;
/* Copy RSA signature from the fw image to HW for verification */
sg_pcopy_to_buffer(sg->sgl, sg->nents, rsa, sizeof(rsa), offset);
for (i = 0; i < UOS_RSA_SCRATCH_MAX_COUNT; i++)
I915_WRITE(UOS_RSA_SCRATCH(i), rsa[i]);
/* The header plus uCode will be copied to WOPCM via DMA, excluding any
* other components */
I915_WRITE(DMA_COPY_SIZE, guc_fw->header_size + guc_fw->ucode_size);
/* Set the source address for the new blob */
offset = guc_ggtt_offset(vma) + guc_fw->header_offset;
I915_WRITE(DMA_ADDR_0_LOW, lower_32_bits(offset));
I915_WRITE(DMA_ADDR_0_HIGH, upper_32_bits(offset) & 0xFFFF);
/*
* Set the DMA destination. Current uCode expects the code to be
* loaded at 8k; locations below this are used for the stack.
*/
I915_WRITE(DMA_ADDR_1_LOW, 0x2000);
I915_WRITE(DMA_ADDR_1_HIGH, DMA_ADDRESS_SPACE_WOPCM);
/* Finally start the DMA */
I915_WRITE(DMA_CTRL, _MASKED_BIT_ENABLE(UOS_MOVE | START_DMA));
/*
* Wait for the DMA to complete & the GuC to start up.
* NB: Docs recommend not using the interrupt for completion.
* Measurements indicate this should take no more than 20ms, so a
* timeout here indicates that the GuC has failed and is unusable.
* (Higher levels of the driver will attempt to fall back to
* execlist mode if this happens.)
*/
ret = wait_for(guc_ucode_response(dev_priv, &status), 100);
DRM_DEBUG_DRIVER("DMA status 0x%x, GuC status 0x%x\n",
I915_READ(DMA_CTRL), status);
if ((status & GS_BOOTROM_MASK) == GS_BOOTROM_RSA_FAILED) {
DRM_ERROR("GuC firmware signature verification failed\n");
ret = -ENOEXEC;
}
DRM_DEBUG_DRIVER("returning %d\n", ret);
return ret;
}
static void __vlv_force_wake_get(struct drm_i915_private *dev_priv,
int fw_engine)
{
/* Check for Render Engine */
if (FORCEWAKE_RENDER & fw_engine) {
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_VLV) &
FORCEWAKE_KERNEL) == 0,
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: Render forcewake old ack to clear.\n");
__raw_i915_write32(dev_priv, FORCEWAKE_VLV,
_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL));
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_VLV) &
FORCEWAKE_KERNEL),
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: waiting for Render to ack.\n");
}
/* Check for Media Engine */
if (FORCEWAKE_MEDIA & fw_engine) {
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_MEDIA_VLV) &
FORCEWAKE_KERNEL) == 0,
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: Media forcewake old ack to clear.\n");
__raw_i915_write32(dev_priv, FORCEWAKE_MEDIA_VLV,
_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL));
if (wait_for_atomic((__raw_i915_read32(dev_priv,
FORCEWAKE_ACK_MEDIA_VLV) &
FORCEWAKE_KERNEL),
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out: waiting for media to ack.\n");
}
/* WaRsForcewakeWaitTC0:vlv */
__gen6_gt_wait_for_thread_c0(dev_priv);
}
static inline void flushtlb(struct nugpgpu_private *gpu_priv)
{
RING_WRITE_INSTPM(RING,
_MASKED_BIT_ENABLE(INSTPM_TLB_INVALIDATE | INSTPM_SYNC_FLUSH));
if( wait_for((RING_READ_INSTPM(RING) & INSTPM_SYNC_FLUSH) == 0, 1000))
printk(LOG_ERR "Wait for SyncFlush timedout\n" LOG_END);
else
printk(LOG_INFO "TLB invalidated successfully\n" LOG_END);
}
static void bxt_gt_workarounds_apply(struct drm_i915_private *dev_priv)
{
gen9_gt_workarounds_apply(dev_priv);
/* WaDisablePooledEuLoadBalancingFix:bxt */
I915_WRITE(FF_SLICE_CS_CHICKEN2,
_MASKED_BIT_ENABLE(GEN9_POOLED_EU_LOAD_BALANCING_FIX_DISABLE));
/* WaInPlaceDecompressionHang:bxt */
I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA,
I915_READ(GEN9_GAMT_ECO_REG_RW_IA) |
GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
}
static int gen8_ppgtt_enable(struct drm_device *dev)
{
struct drm_i915_private *dev_priv = dev->dev_private;
struct intel_ring_buffer *ring;
struct i915_hw_ppgtt *ppgtt = dev_priv->mm.aliasing_ppgtt;
int i, j, ret;
/* bit of a hack to find the actual last used pd */
int used_pd = ppgtt->num_pd_entries / GEN8_PDES_PER_PAGE;
for_each_ring(ring, dev_priv, j) {
I915_WRITE(RING_MODE_GEN7(ring),
_MASKED_BIT_ENABLE(GFX_PPGTT_ENABLE));
}
static void vlv_force_wake_get(struct drm_i915_private *dev_priv)
{
if (wait_for_atomic((__raw_i915_read32(dev_priv, FORCEWAKE_ACK_VLV) & FORCEWAKE_KERNEL) == 0,
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out waiting for forcewake old ack to clear.\n");
__raw_i915_write32(dev_priv, FORCEWAKE_VLV,
_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL));
__raw_i915_write32(dev_priv, FORCEWAKE_MEDIA_VLV,
_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL));
if (wait_for_atomic((__raw_i915_read32(dev_priv, FORCEWAKE_ACK_VLV) & FORCEWAKE_KERNEL),
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out waiting for GT to ack forcewake request.\n");
if (wait_for_atomic((__raw_i915_read32(dev_priv, FORCEWAKE_ACK_MEDIA_VLV) &
FORCEWAKE_KERNEL),
FORCEWAKE_ACK_TIMEOUT_MS))
DRM_ERROR("Timed out waiting for media to ack forcewake request.\n");
/* WaRsForcewakeWaitTC0:vlv */
__gen6_gt_wait_for_thread_c0(dev_priv);
}
void __gen6_gt_force_wake_mt_get(struct drm_i915_private *dev_priv)
{
int count;
count = 0;
while (count++ < 50 && (I915_READ_NOTRACE(FORCEWAKE_MT_ACK) & 1))
udelay(10);
I915_WRITE_NOTRACE(FORCEWAKE_MT, _MASKED_BIT_ENABLE(1));
POSTING_READ(FORCEWAKE_MT);
count = 0;
while (count++ < 50 && (I915_READ_NOTRACE(FORCEWAKE_MT_ACK) & 1) == 0)
udelay(10);
}
static void cnl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
{
/* WaDisableI2mCycleOnWRPort:cnl (pre-prod) */
if (IS_CNL_REVID(dev_priv, CNL_REVID_B0, CNL_REVID_B0))
I915_WRITE(GAMT_CHKN_BIT_REG,
I915_READ(GAMT_CHKN_BIT_REG) |
GAMT_CHKN_DISABLE_I2M_CYCLE_ON_WR_PORT);
/* WaInPlaceDecompressionHang:cnl */
I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA,
I915_READ(GEN9_GAMT_ECO_REG_RW_IA) |
GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
/* WaEnablePreemptionGranularityControlByUMD:cnl */
I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1,
_MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL));
}
static void guc_interrupts_capture(struct drm_i915_private *dev_priv)
{
struct intel_engine_cs *engine;
enum intel_engine_id id;
int irqs;
u32 tmp;
/* tell all command streamers to forward interrupts (but not vblank) to GuC */
irqs = _MASKED_BIT_ENABLE(GFX_INTERRUPT_STEERING);
for_each_engine(engine, dev_priv, id)
I915_WRITE(RING_MODE_GEN7(engine), irqs);
/* route USER_INTERRUPT to Host, all others are sent to GuC. */
irqs = GT_RENDER_USER_INTERRUPT << GEN8_RCS_IRQ_SHIFT |
GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT;
/* These three registers have the same bit definitions */
I915_WRITE(GUC_BCS_RCS_IER, ~irqs);
I915_WRITE(GUC_VCS2_VCS1_IER, ~irqs);
I915_WRITE(GUC_WD_VECS_IER, ~irqs);
/*
* The REDIRECT_TO_GUC bit of the PMINTRMSK register directs all
* (unmasked) PM interrupts to the GuC. All other bits of this
* register *disable* generation of a specific interrupt.
*
* 'pm_intr_keep' indicates bits that are NOT to be set when
* writing to the PM interrupt mask register, i.e. interrupts
* that must not be disabled.
*
* If the GuC is handling these interrupts, then we must not let
* the PM code disable ANY interrupt that the GuC is expecting.
* So for each ENABLED (0) bit in this register, we must SET the
* bit in pm_intr_keep so that it's left enabled for the GuC.
*
* OTOH the REDIRECT_TO_GUC bit is initially SET in pm_intr_keep
* (so interrupts go to the DISPLAY unit at first); but here we
* need to CLEAR that bit, which will result in the register bit
* being left SET!
*/
tmp = I915_READ(GEN6_PMINTRMSK);
if (tmp & GEN8_PMINTR_REDIRECT_TO_GUC) {
dev_priv->rps.pm_intr_keep |= ~tmp;
dev_priv->rps.pm_intr_keep &= ~GEN8_PMINTR_REDIRECT_TO_GUC;
}
}
static void bdw_forcewake_get(void __iomem *mmio)
{
WR(_MASKED_BIT_DISABLE(0xffff), FORCEWAKE_MT);
RD(ECOBUS);
if (wait_for((RD(FORCEWAKE_ACK_HSW) & FORCEWAKE_KERNEL) == 0, 50))
gvt_err("fail to wait forcewake idle\n");
WR(_MASKED_BIT_ENABLE(FORCEWAKE_KERNEL), FORCEWAKE_MT);
if (wait_for((RD(FORCEWAKE_ACK_HSW) & FORCEWAKE_KERNEL), 50))
gvt_err("fail to wait forcewake ack\n");
if (wait_for((RD(GEN6_GT_THREAD_STATUS_REG) &
GEN6_GT_THREAD_STATUS_CORE_MASK) == 0, 50))
gvt_err("fail to wait c0 wake up\n");
}
static void direct_interrupts_to_guc(struct drm_i915_private *dev_priv)
{
struct intel_engine_cs *ring;
int i, irqs;
/* tell all command streamers to forward interrupts and vblank to GuC */
irqs = _MASKED_FIELD(GFX_FORWARD_VBLANK_MASK, GFX_FORWARD_VBLANK_ALWAYS);
irqs |= _MASKED_BIT_ENABLE(GFX_INTERRUPT_STEERING);
for_each_ring(ring, dev_priv, i)
I915_WRITE(RING_MODE_GEN7(ring), irqs);
/* route USER_INTERRUPT to Host, all others are sent to GuC. */
irqs = GT_RENDER_USER_INTERRUPT << GEN8_RCS_IRQ_SHIFT |
GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT;
/* These three registers have the same bit definitions */
I915_WRITE(GUC_BCS_RCS_IER, ~irqs);
I915_WRITE(GUC_VCS2_VCS1_IER, ~irqs);
I915_WRITE(GUC_WD_VECS_IER, ~irqs);
}
/**
* huc_fw_xfer() - DMA's the firmware
* @huc_fw: the firmware descriptor
* @vma: the firmware image (bound into the GGTT)
*
* Transfer the firmware image to RAM for execution by the microcontroller.
*
* Return: 0 on success, non-zero on failure
*/
static int huc_fw_xfer(struct intel_uc_fw *huc_fw, struct i915_vma *vma)
{
struct intel_huc *huc = container_of(huc_fw, struct intel_huc, fw);
struct drm_i915_private *dev_priv = huc_to_i915(huc);
unsigned long offset = 0;
u32 size;
int ret;
GEM_BUG_ON(huc_fw->type != INTEL_UC_FW_TYPE_HUC);
intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
/* Set the source address for the uCode */
offset = intel_guc_ggtt_offset(&dev_priv->guc, vma) +
huc_fw->header_offset;
I915_WRITE(DMA_ADDR_0_LOW, lower_32_bits(offset));
I915_WRITE(DMA_ADDR_0_HIGH, upper_32_bits(offset) & 0xFFFF);
/* Hardware doesn't look at destination address for HuC. Set it to 0,
* but still program the correct address space.
*/
I915_WRITE(DMA_ADDR_1_LOW, 0);
I915_WRITE(DMA_ADDR_1_HIGH, DMA_ADDRESS_SPACE_WOPCM);
size = huc_fw->header_size + huc_fw->ucode_size;
I915_WRITE(DMA_COPY_SIZE, size);
/* Start the DMA */
I915_WRITE(DMA_CTRL, _MASKED_BIT_ENABLE(HUC_UKERNEL | START_DMA));
/* Wait for DMA to finish */
ret = intel_wait_for_register_fw(dev_priv, DMA_CTRL, START_DMA, 0, 100);
DRM_DEBUG_DRIVER("HuC DMA transfer wait over with ret %d\n", ret);
/* Disable the bits once DMA is over */
I915_WRITE(DMA_CTRL, _MASKED_BIT_DISABLE(HUC_UKERNEL));
intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
return ret;
}
/*
* Transfer the firmware image to RAM for execution by the microcontroller.
*
* Architecturally, the DMA engine is bidirectional, and can potentially even
* transfer between GTT locations. This functionality is left out of the API
* for now as there is no need for it.
*/
static int guc_xfer_ucode(struct intel_guc *guc, struct i915_vma *vma)
{
struct drm_i915_private *dev_priv = guc_to_i915(guc);
struct intel_uc_fw *guc_fw = &guc->fw;
unsigned long offset;
u32 status;
int ret;
/*
* The header plus uCode will be copied to WOPCM via DMA, excluding any
* other components
*/
I915_WRITE(DMA_COPY_SIZE, guc_fw->header_size + guc_fw->ucode_size);
/* Set the source address for the new blob */
offset = guc_ggtt_offset(vma) + guc_fw->header_offset;
I915_WRITE(DMA_ADDR_0_LOW, lower_32_bits(offset));
I915_WRITE(DMA_ADDR_0_HIGH, upper_32_bits(offset) & 0xFFFF);
/*
* Set the DMA destination. Current uCode expects the code to be
* loaded at 8k; locations below this are used for the stack.
*/
I915_WRITE(DMA_ADDR_1_LOW, 0x2000);
I915_WRITE(DMA_ADDR_1_HIGH, DMA_ADDRESS_SPACE_WOPCM);
/* Finally start the DMA */
I915_WRITE(DMA_CTRL, _MASKED_BIT_ENABLE(UOS_MOVE | START_DMA));
/* Wait for DMA to finish */
ret = __intel_wait_for_register_fw(dev_priv, DMA_CTRL, START_DMA, 0,
2, 100, &status);
DRM_DEBUG_DRIVER("GuC DMA status %#x\n", status);
return ret;
}
static void icl_gt_workarounds_apply(struct drm_i915_private *dev_priv)
{
/* This is not an Wa. Enable for better image quality */
I915_WRITE(_3D_CHICKEN3,
_MASKED_BIT_ENABLE(_3D_CHICKEN3_AA_LINE_QUALITY_FIX_ENABLE));
/* WaInPlaceDecompressionHang:icl */
I915_WRITE(GEN9_GAMT_ECO_REG_RW_IA, I915_READ(GEN9_GAMT_ECO_REG_RW_IA) |
GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS);
/* WaPipelineFlushCoherentLines:icl */
I915_WRITE(GEN8_L3SQCREG4, I915_READ(GEN8_L3SQCREG4) |
GEN8_LQSC_FLUSH_COHERENT_LINES);
/* Wa_1405543622:icl
* Formerly known as WaGAPZPriorityScheme
*/
I915_WRITE(GEN8_GARBCNTL, I915_READ(GEN8_GARBCNTL) |
GEN11_ARBITRATION_PRIO_ORDER_MASK);
/* Wa_1604223664:icl
* Formerly known as WaL3BankAddressHashing
*/
I915_WRITE(GEN8_GARBCNTL,
(I915_READ(GEN8_GARBCNTL) & ~GEN11_HASH_CTRL_EXCL_MASK) |
GEN11_HASH_CTRL_EXCL_BIT0);
I915_WRITE(GEN11_GLBLINVL,
(I915_READ(GEN11_GLBLINVL) & ~GEN11_BANK_HASH_ADDR_EXCL_MASK) |
GEN11_BANK_HASH_ADDR_EXCL_BIT0);
/* WaModifyGamTlbPartitioning:icl */
I915_WRITE(GEN11_GACB_PERF_CTRL,
(I915_READ(GEN11_GACB_PERF_CTRL) & ~GEN11_HASH_CTRL_MASK) |
GEN11_HASH_CTRL_BIT0 | GEN11_HASH_CTRL_BIT4);
/* Wa_1405733216:icl
* Formerly known as WaDisableCleanEvicts
*/
I915_WRITE(GEN8_L3SQCREG4, I915_READ(GEN8_L3SQCREG4) |
GEN11_LQSC_CLEAN_EVICT_DISABLE);
/* Wa_1405766107:icl
* Formerly known as WaCL2SFHalfMaxAlloc
*/
I915_WRITE(GEN11_LSN_UNSLCVC, I915_READ(GEN11_LSN_UNSLCVC) |
GEN11_LSN_UNSLCVC_GAFS_HALF_SF_MAXALLOC |
GEN11_LSN_UNSLCVC_GAFS_HALF_CL2_MAXALLOC);
/* Wa_220166154:icl
* Formerly known as WaDisCtxReload
*/
I915_WRITE(GAMW_ECO_DEV_RW_IA_REG, I915_READ(GAMW_ECO_DEV_RW_IA_REG) |
GAMW_ECO_DEV_CTX_RELOAD_DISABLE);
/* Wa_1405779004:icl (pre-prod) */
if (IS_ICL_REVID(dev_priv, ICL_REVID_A0, ICL_REVID_A0))
I915_WRITE(SLICE_UNIT_LEVEL_CLKGATE,
I915_READ(SLICE_UNIT_LEVEL_CLKGATE) |
MSCUNIT_CLKGATE_DIS);
/* Wa_1406680159:icl */
I915_WRITE(SUBSLICE_UNIT_LEVEL_CLKGATE,
I915_READ(SUBSLICE_UNIT_LEVEL_CLKGATE) |
GWUNIT_CLKGATE_DIS);
/* Wa_1604302699:icl */
I915_WRITE(GEN10_L3_CHICKEN_MODE_REGISTER,
I915_READ(GEN10_L3_CHICKEN_MODE_REGISTER) |
GEN11_I2M_WRITE_DISABLE);
/* Wa_1406838659:icl (pre-prod) */
if (IS_ICL_REVID(dev_priv, ICL_REVID_A0, ICL_REVID_B0))
I915_WRITE(INF_UNIT_LEVEL_CLKGATE,
I915_READ(INF_UNIT_LEVEL_CLKGATE) |
CGPSF_CLKGATE_DIS);
/* WaForwardProgressSoftReset:icl */
I915_WRITE(GEN10_SCRATCH_LNCF2,
I915_READ(GEN10_SCRATCH_LNCF2) |
PMFLUSHDONE_LNICRSDROP |
PMFLUSH_GAPL3UNBLOCK |
PMFLUSHDONE_LNEBLK);
}
int nugpgpu_ringbuffer_render_init(struct nugpgpu_private *gpu_priv)
{
int ret;
u32 head;
printk(LOG_INFO "nugpgpu_ringbuffer_render_init\n" LOG_END);
TRACE_IN
RING->mmio_base = RENDER_RING_BASE;
RING->size = PAGE_SIZE * RING_PAGES;
/* Allocate the status page. */
ret = allocate_object(gpu_priv, &RING->status_obj, 1);
if (ret){
printk(LOG_ERR "Failed to allocate the status page\n" LOG_END);
return 1;
}
RING->gva_status = nugpgpu_gtt_insert(gpu_priv, RING->status_obj.pg_list,
NUGPGPU_CACHE_LLC);
if (RING->gva_status == (unsigned int)-1){
printk(LOG_ERR "Failed to insert the status page in gtt\n" LOG_END);
return 1;
}
printk(LOG_INFO "RING->gva_status : 0x%x\n" LOG_END, (unsigned int) RING->gva_status);
RING->page_status = kmap(sg_page(RING->status_obj.pg_list->sgl));
if (RING->page_status == NULL) {
printk(LOG_ERR "Failed to map page_status\n" LOG_END);
return 1;
}
memset(RING->page_status, 0, PAGE_SIZE);
printk(LOG_INFO "RING->page_status : 0x%lx\n" LOG_END, (unsigned long) RING->page_status);
/* Allocate the ringbuffer object */
ret = allocate_object(gpu_priv, &RING->ringbuf_obj, RING_PAGES);
if (ret){
printk(LOG_ERR "Failed to allocate the status page\n" LOG_END);
return 1;
}
RING->gva_ringbuffer = nugpgpu_gtt_insert(gpu_priv, RING->ringbuf_obj.pg_list,
NUGPGPU_CACHE_LLC);
if (RING->gva_ringbuffer == (unsigned int)-1){
printk(LOG_ERR "Failed to insert the status page in gtt\n" LOG_END);
return 1;
}
printk(LOG_INFO "RING->gva_ringbuffer : 0x%x\n" LOG_END, (unsigned int) RING->gva_ringbuffer);
RING->page_ringbuffer = kmap(sg_page(RING->ringbuf_obj.pg_list->sgl));
if (RING->page_ringbuffer == NULL) {
printk(LOG_ERR "Failed to map page_ringbuffer\n" LOG_END);
return 1;
}
RING->virtual_start = ioremap_wc(gpu_priv->gtt.mappable_base + PAGE_SIZE, RING->size);
if (RING->virtual_start == NULL) {
printk(LOG_ERR "Problem while mapping virtual start ioremap_wc\n" LOG_END);
return 1;
}
printk(LOG_INFO "Allocated the ringbuffer\n" LOG_END);
/* Initialize the ring now.*/
gpu_forcewake_get(gpu_priv);
/* Write status page register */
printk(LOG_INFO "writing status page register\n" LOG_END);
NUGPGPU_WRITE(RENDER_HWS_PGA_GEN7, RING->gva_status);
NUGPGPU_READ(RENDER_HWS_PGA_GEN7);
flushtlb(gpu_priv);
// Stop ring
printk(LOG_INFO "stopping ring\n" LOG_END);
RING_WRITE_CTL(RING, 0);
RING_WRITE_HEAD(RING, 0);
RING_WRITE_TAIL(RING, 0);
// The doc says this enforces ordering between multiple writes
head = RING_READ_HEAD(RING) & RING_HEAD_ADDR;
if ( head !=0 ){
printk(LOG_ERR "failed to set head to zero\n" LOG_END);
RING_WRITE_HEAD(RING, 0);
if (RING_READ_HEAD(RING) & RING_HEAD_ADDR) {
printk(LOG_ERR "failed to set ring head to zero "
"ctl %08x head %08x tail %08x start %08x\n"
LOG_END,
RING_READ_CTL(RING),
RING_READ_HEAD(RING),
RING_READ_TAIL(RING),
RING_READ_START(RING));
}
}
/* i915 driver says the below line...?? */
/* Enforce ordering by reading HEAD register back */
RING_READ_HEAD(RING);
/* Comment taken directly from i915 driver */
/* Initialize the ring. This must happen _after_ we've cleared the ring
* registers with the above sequence (the readback of the HEAD registers
* also enforces ordering), otherwise the hw might lose the new ring
* register values. */
RING_WRITE_START(RING, RING->gva_ringbuffer);
RING_WRITE_CTL(RING, (((RING->size - PAGE_SIZE) &
RING_NR_PAGES) |
RING_VALID));
/* If the head is still not zero, the ring is dead */
if( wait_for((RING_READ_CTL(RING) & RING_VALID) != 0 &&
RING_READ_START(RING) == RING->gva_ringbuffer &&
(RING_READ_HEAD(RING) & RING_HEAD_ADDR) == 0, 50) ){
printk(LOG_ERR "ring failed to start ring\n" LOG_END);
return -EIO;
}
RING->head = RING_READ_HEAD(RING);
RING->tail = RING_READ_TAIL(RING) & RING_TAIL_ADDR;
RING->space = ring_space(RING);
printk(LOG_INFO "ring->space = %d\n" LOG_END, RING->space);
gpu_forcewake_put(gpu_priv);
RING_WRITE_MODE(RING, _MASKED_BIT_ENABLE(VS_TIMER_DISPATCH));
RING_WRITE_MODE(RING, _MASKED_BIT_ENABLE(ASYNC_FLIP_PERF_DISABLE));
RING_WRITE_MODE_GEN7(RING, _MASKED_BIT_DISABLE(GFX_TLB_INVALIDATE_ALWAYS) |
_MASKED_BIT_ENABLE(GFX_REPLAY_MODE));
RING_WRITE_INSTPM(RING, _MASKED_BIT_ENABLE(INSTPM_FORCE_ORDERING));
dword_check(gpu_priv, RING, temp);
TRACE_OUT
return 0;
}
