summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/i915/intel_ringbuffer.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c')
-rw-r--r--drivers/gpu/drm/i915/intel_ringbuffer.c476
1 files changed, 234 insertions, 242 deletions
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 04402bb9d26b..8d35a3978f9b 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -34,6 +34,11 @@
#include "i915_trace.h"
#include "intel_drv.h"
+/* Rough estimate of the typical request size, performing a flush,
+ * set-context and then emitting the batch.
+ */
+#define LEGACY_REQUEST_SIZE 200
+
int __intel_ring_space(int head, int tail, int size)
{
int space = head - tail;
@@ -55,7 +60,7 @@ void intel_ring_update_space(struct intel_ringbuffer *ringbuf)
bool intel_engine_stopped(struct intel_engine_cs *engine)
{
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
return dev_priv->gpu_error.stop_rings & intel_engine_flag(engine);
}
@@ -101,7 +106,6 @@ gen4_render_ring_flush(struct drm_i915_gem_request *req,
u32 flush_domains)
{
struct intel_engine_cs *engine = req->engine;
- struct drm_device *dev = engine->dev;
u32 cmd;
int ret;
@@ -140,7 +144,7 @@ gen4_render_ring_flush(struct drm_i915_gem_request *req,
cmd |= MI_EXE_FLUSH;
if (invalidate_domains & I915_GEM_DOMAIN_COMMAND &&
- (IS_G4X(dev) || IS_GEN5(dev)))
+ (IS_G4X(req->i915) || IS_GEN5(req->i915)))
cmd |= MI_INVALIDATE_ISP;
ret = intel_ring_begin(req, 2);
@@ -426,19 +430,19 @@ gen8_render_ring_flush(struct drm_i915_gem_request *req,
static void ring_write_tail(struct intel_engine_cs *engine,
u32 value)
{
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
I915_WRITE_TAIL(engine, value);
}
u64 intel_ring_get_active_head(struct intel_engine_cs *engine)
{
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
u64 acthd;
- if (INTEL_INFO(engine->dev)->gen >= 8)
+ if (INTEL_GEN(dev_priv) >= 8)
acthd = I915_READ64_2x32(RING_ACTHD(engine->mmio_base),
RING_ACTHD_UDW(engine->mmio_base));
- else if (INTEL_INFO(engine->dev)->gen >= 4)
+ else if (INTEL_GEN(dev_priv) >= 4)
acthd = I915_READ(RING_ACTHD(engine->mmio_base));
else
acthd = I915_READ(ACTHD);
@@ -448,25 +452,24 @@ u64 intel_ring_get_active_head(struct intel_engine_cs *engine)
static void ring_setup_phys_status_page(struct intel_engine_cs *engine)
{
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
u32 addr;
addr = dev_priv->status_page_dmah->busaddr;
- if (INTEL_INFO(engine->dev)->gen >= 4)
+ if (INTEL_GEN(dev_priv) >= 4)
addr |= (dev_priv->status_page_dmah->busaddr >> 28) & 0xf0;
I915_WRITE(HWS_PGA, addr);
}
static void intel_ring_setup_status_page(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
i915_reg_t mmio;
/* The ring status page addresses are no longer next to the rest of
* the ring registers as of gen7.
*/
- if (IS_GEN7(dev)) {
+ if (IS_GEN7(dev_priv)) {
switch (engine->id) {
case RCS:
mmio = RENDER_HWS_PGA_GEN7;
@@ -486,7 +489,7 @@ static void intel_ring_setup_status_page(struct intel_engine_cs *engine)
mmio = VEBOX_HWS_PGA_GEN7;
break;
}
- } else if (IS_GEN6(engine->dev)) {
+ } else if (IS_GEN6(dev_priv)) {
mmio = RING_HWS_PGA_GEN6(engine->mmio_base);
} else {
/* XXX: gen8 returns to sanity */
@@ -503,7 +506,7 @@ static void intel_ring_setup_status_page(struct intel_engine_cs *engine)
* arises: do we still need this and if so how should we go about
* invalidating the TLB?
*/
- if (INTEL_INFO(dev)->gen >= 6 && INTEL_INFO(dev)->gen < 8) {
+ if (IS_GEN(dev_priv, 6, 7)) {
i915_reg_t reg = RING_INSTPM(engine->mmio_base);
/* ring should be idle before issuing a sync flush*/
@@ -521,9 +524,9 @@ static void intel_ring_setup_status_page(struct intel_engine_cs *engine)
static bool stop_ring(struct intel_engine_cs *engine)
{
- struct drm_i915_private *dev_priv = to_i915(engine->dev);
+ struct drm_i915_private *dev_priv = engine->i915;
- if (!IS_GEN2(engine->dev)) {
+ if (!IS_GEN2(dev_priv)) {
I915_WRITE_MODE(engine, _MASKED_BIT_ENABLE(STOP_RING));
if (wait_for((I915_READ_MODE(engine) & MODE_IDLE) != 0, 1000)) {
DRM_ERROR("%s : timed out trying to stop ring\n",
@@ -541,7 +544,7 @@ static bool stop_ring(struct intel_engine_cs *engine)
I915_WRITE_HEAD(engine, 0);
engine->write_tail(engine, 0);
- if (!IS_GEN2(engine->dev)) {
+ if (!IS_GEN2(dev_priv)) {
(void)I915_READ_CTL(engine);
I915_WRITE_MODE(engine, _MASKED_BIT_DISABLE(STOP_RING));
}
@@ -556,8 +559,7 @@ void intel_engine_init_hangcheck(struct intel_engine_cs *engine)
static int init_ring_common(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
struct intel_ringbuffer *ringbuf = engine->buffer;
struct drm_i915_gem_object *obj = ringbuf->obj;
int ret = 0;
@@ -587,7 +589,7 @@ static int init_ring_common(struct intel_engine_cs *engine)
}
}
- if (I915_NEED_GFX_HWS(dev))
+ if (I915_NEED_GFX_HWS(dev_priv))
intel_ring_setup_status_page(engine);
else
ring_setup_phys_status_page(engine);
@@ -644,12 +646,10 @@ out:
void
intel_fini_pipe_control(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
-
if (engine->scratch.obj == NULL)
return;
- if (INTEL_INFO(dev)->gen >= 5) {
+ if (INTEL_GEN(engine->i915) >= 5) {
kunmap(sg_page(engine->scratch.obj->pages->sgl));
i915_gem_object_ggtt_unpin(engine->scratch.obj);
}
@@ -665,10 +665,11 @@ intel_init_pipe_control(struct intel_engine_cs *engine)
WARN_ON(engine->scratch.obj);
- engine->scratch.obj = i915_gem_alloc_object(engine->dev, 4096);
- if (engine->scratch.obj == NULL) {
+ engine->scratch.obj = i915_gem_object_create(engine->i915->dev, 4096);
+ if (IS_ERR(engine->scratch.obj)) {
DRM_ERROR("Failed to allocate seqno page\n");
- ret = -ENOMEM;
+ ret = PTR_ERR(engine->scratch.obj);
+ engine->scratch.obj = NULL;
goto err;
}
@@ -702,11 +703,9 @@ err:
static int intel_ring_workarounds_emit(struct drm_i915_gem_request *req)
{
- int ret, i;
struct intel_engine_cs *engine = req->engine;
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
- struct i915_workarounds *w = &dev_priv->workarounds;
+ struct i915_workarounds *w = &req->i915->workarounds;
+ int ret, i;
if (w->count == 0)
return 0;
@@ -795,7 +794,7 @@ static int wa_add(struct drm_i915_private *dev_priv,
static int wa_ring_whitelist_reg(struct intel_engine_cs *engine,
i915_reg_t reg)
{
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
struct i915_workarounds *wa = &dev_priv->workarounds;
const uint32_t index = wa->hw_whitelist_count[engine->id];
@@ -811,8 +810,7 @@ static int wa_ring_whitelist_reg(struct intel_engine_cs *engine,
static int gen8_init_workarounds(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
WA_SET_BIT_MASKED(INSTPM, INSTPM_FORCE_ORDERING);
@@ -863,9 +861,8 @@ static int gen8_init_workarounds(struct intel_engine_cs *engine)
static int bdw_init_workarounds(struct intel_engine_cs *engine)
{
+ struct drm_i915_private *dev_priv = engine->i915;
int ret;
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
ret = gen8_init_workarounds(engine);
if (ret)
@@ -885,16 +882,15 @@ static int bdw_init_workarounds(struct intel_engine_cs *engine)
/* WaForceContextSaveRestoreNonCoherent:bdw */
HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT |
/* WaDisableFenceDestinationToSLM:bdw (pre-prod) */
- (IS_BDW_GT3(dev) ? HDC_FENCE_DEST_SLM_DISABLE : 0));
+ (IS_BDW_GT3(dev_priv) ? HDC_FENCE_DEST_SLM_DISABLE : 0));
return 0;
}
static int chv_init_workarounds(struct intel_engine_cs *engine)
{
+ struct drm_i915_private *dev_priv = engine->i915;
int ret;
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
ret = gen8_init_workarounds(engine);
if (ret)
@@ -911,8 +907,7 @@ static int chv_init_workarounds(struct intel_engine_cs *engine)
static int gen9_init_workarounds(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
uint32_t tmp;
int ret;
@@ -935,14 +930,14 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
GEN9_DISABLE_OCL_OOB_SUPPRESS_LOGIC);
/* WaDisableDgMirrorFixInHalfSliceChicken5:skl,bxt */
- if (IS_SKL_REVID(dev, 0, SKL_REVID_B0) ||
- IS_BXT_REVID(dev, 0, BXT_REVID_A1))
+ if (IS_SKL_REVID(dev_priv, 0, SKL_REVID_B0) ||
+ IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
WA_CLR_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN5,
GEN9_DG_MIRROR_FIX_ENABLE);
/* WaSetDisablePixMaskCammingAndRhwoInCommonSliceChicken:skl,bxt */
- if (IS_SKL_REVID(dev, 0, SKL_REVID_B0) ||
- IS_BXT_REVID(dev, 0, BXT_REVID_A1)) {
+ if (IS_SKL_REVID(dev_priv, 0, SKL_REVID_B0) ||
+ IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
WA_SET_BIT_MASKED(GEN7_COMMON_SLICE_CHICKEN1,
GEN9_RHWO_OPTIMIZATION_DISABLE);
/*
@@ -968,20 +963,20 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
GEN9_CCS_TLB_PREFETCH_ENABLE);
/* WaDisableMaskBasedCammingInRCC:skl,bxt */
- if (IS_SKL_REVID(dev, SKL_REVID_C0, SKL_REVID_C0) ||
- IS_BXT_REVID(dev, 0, BXT_REVID_A1))
+ if (IS_SKL_REVID(dev_priv, SKL_REVID_C0, SKL_REVID_C0) ||
+ IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
WA_SET_BIT_MASKED(SLICE_ECO_CHICKEN0,
PIXEL_MASK_CAMMING_DISABLE);
/* WaForceContextSaveRestoreNonCoherent:skl,bxt */
tmp = HDC_FORCE_CONTEXT_SAVE_RESTORE_NON_COHERENT;
- if (IS_SKL_REVID(dev, SKL_REVID_F0, REVID_FOREVER) ||
- IS_BXT_REVID(dev, BXT_REVID_B0, REVID_FOREVER))
+ if (IS_SKL_REVID(dev_priv, SKL_REVID_F0, REVID_FOREVER) ||
+ IS_BXT_REVID(dev_priv, BXT_REVID_B0, REVID_FOREVER))
tmp |= HDC_FORCE_CSR_NON_COHERENT_OVR_DISABLE;
WA_SET_BIT_MASKED(HDC_CHICKEN0, tmp);
/* WaDisableSamplerPowerBypassForSOPingPong:skl,bxt */
- if (IS_SKYLAKE(dev) || IS_BXT_REVID(dev, 0, BXT_REVID_B0))
+ if (IS_SKYLAKE(dev_priv) || IS_BXT_REVID(dev_priv, 0, BXT_REVID_B0))
WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3,
GEN8_SAMPLER_POWER_BYPASS_DIS);
@@ -1007,8 +1002,7 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
static int skl_tune_iz_hashing(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
u8 vals[3] = { 0, 0, 0 };
unsigned int i;
@@ -1049,9 +1043,8 @@ static int skl_tune_iz_hashing(struct intel_engine_cs *engine)
static int skl_init_workarounds(struct intel_engine_cs *engine)
{
+ struct drm_i915_private *dev_priv = engine->i915;
int ret;
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
ret = gen9_init_workarounds(engine);
if (ret)
@@ -1062,12 +1055,12 @@ static int skl_init_workarounds(struct intel_engine_cs *engine)
* until D0 which is the default case so this is equivalent to
* !WaDisablePerCtxtPreemptionGranularityControl:skl
*/
- if (IS_SKL_REVID(dev, SKL_REVID_E0, REVID_FOREVER)) {
+ if (IS_SKL_REVID(dev_priv, SKL_REVID_E0, REVID_FOREVER)) {
I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1,
_MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL));
}
- if (IS_SKL_REVID(dev, 0, SKL_REVID_D0)) {
+ if (IS_SKL_REVID(dev_priv, 0, SKL_REVID_D0)) {
/* WaDisableChickenBitTSGBarrierAckForFFSliceCS:skl */
I915_WRITE(FF_SLICE_CS_CHICKEN2,
_MASKED_BIT_ENABLE(GEN9_TSG_BARRIER_ACK_DISABLE));
@@ -1076,24 +1069,24 @@ static int skl_init_workarounds(struct intel_engine_cs *engine)
/* GEN8_L3SQCREG4 has a dependency with WA batch so any new changes
* involving this register should also be added to WA batch as required.
*/
- if (IS_SKL_REVID(dev, 0, SKL_REVID_E0))
+ if (IS_SKL_REVID(dev_priv, 0, SKL_REVID_E0))
/* WaDisableLSQCROPERFforOCL:skl */
I915_WRITE(GEN8_L3SQCREG4, I915_READ(GEN8_L3SQCREG4) |
GEN8_LQSC_RO_PERF_DIS);
/* WaEnableGapsTsvCreditFix:skl */
- if (IS_SKL_REVID(dev, SKL_REVID_C0, REVID_FOREVER)) {
+ if (IS_SKL_REVID(dev_priv, SKL_REVID_C0, REVID_FOREVER)) {
I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) |
GEN9_GAPS_TSV_CREDIT_DISABLE));
}
/* WaDisablePowerCompilerClockGating:skl */
- if (IS_SKL_REVID(dev, SKL_REVID_B0, SKL_REVID_B0))
+ if (IS_SKL_REVID(dev_priv, SKL_REVID_B0, SKL_REVID_B0))
WA_SET_BIT_MASKED(HIZ_CHICKEN,
BDW_HIZ_POWER_COMPILER_CLOCK_GATING_DISABLE);
/* This is tied to WaForceContextSaveRestoreNonCoherent */
- if (IS_SKL_REVID(dev, 0, REVID_FOREVER)) {
+ if (IS_SKL_REVID(dev_priv, 0, REVID_FOREVER)) {
/*
*Use Force Non-Coherent whenever executing a 3D context. This
* is a workaround for a possible hang in the unlikely event
@@ -1109,13 +1102,13 @@ static int skl_init_workarounds(struct intel_engine_cs *engine)
}
/* WaBarrierPerformanceFixDisable:skl */
- if (IS_SKL_REVID(dev, SKL_REVID_C0, SKL_REVID_D0))
+ if (IS_SKL_REVID(dev_priv, SKL_REVID_C0, SKL_REVID_D0))
WA_SET_BIT_MASKED(HDC_CHICKEN0,
HDC_FENCE_DEST_SLM_DISABLE |
HDC_BARRIER_PERFORMANCE_DISABLE);
/* WaDisableSbeCacheDispatchPortSharing:skl */
- if (IS_SKL_REVID(dev, 0, SKL_REVID_F0))
+ if (IS_SKL_REVID(dev_priv, 0, SKL_REVID_F0))
WA_SET_BIT_MASKED(
GEN7_HALF_SLICE_CHICKEN1,
GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE);
@@ -1130,9 +1123,8 @@ static int skl_init_workarounds(struct intel_engine_cs *engine)
static int bxt_init_workarounds(struct intel_engine_cs *engine)
{
+ struct drm_i915_private *dev_priv = engine->i915;
int ret;
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
ret = gen9_init_workarounds(engine);
if (ret)
@@ -1140,11 +1132,11 @@ static int bxt_init_workarounds(struct intel_engine_cs *engine)
/* WaStoreMultiplePTEenable:bxt */
/* This is a requirement according to Hardware specification */
- if (IS_BXT_REVID(dev, 0, BXT_REVID_A1))
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1))
I915_WRITE(TILECTL, I915_READ(TILECTL) | TILECTL_TLBPF);
/* WaSetClckGatingDisableMedia:bxt */
- if (IS_BXT_REVID(dev, 0, BXT_REVID_A1)) {
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
I915_WRITE(GEN7_MISCCPCTL, (I915_READ(GEN7_MISCCPCTL) &
~GEN8_DOP_CLOCK_GATE_MEDIA_ENABLE));
}
@@ -1154,7 +1146,7 @@ static int bxt_init_workarounds(struct intel_engine_cs *engine)
STALL_DOP_GATING_DISABLE);
/* WaDisableSbeCacheDispatchPortSharing:bxt */
- if (IS_BXT_REVID(dev, 0, BXT_REVID_B0)) {
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_B0)) {
WA_SET_BIT_MASKED(
GEN7_HALF_SLICE_CHICKEN1,
GEN7_SBE_SS_CACHE_DISPATCH_PORT_SHARING_DISABLE);
@@ -1164,7 +1156,7 @@ static int bxt_init_workarounds(struct intel_engine_cs *engine)
/* WaDisableObjectLevelPreemptionForInstancedDraw:bxt */
/* WaDisableObjectLevelPreemtionForInstanceId:bxt */
/* WaDisableLSQCROPERFforOCL:bxt */
- if (IS_BXT_REVID(dev, 0, BXT_REVID_A1)) {
+ if (IS_BXT_REVID(dev_priv, 0, BXT_REVID_A1)) {
ret = wa_ring_whitelist_reg(engine, GEN9_CS_DEBUG_MODE1);
if (ret)
return ret;
@@ -1174,29 +1166,33 @@ static int bxt_init_workarounds(struct intel_engine_cs *engine)
return ret;
}
+ /* WaProgramL3SqcReg1DefaultForPerf:bxt */
+ if (IS_BXT_REVID(dev_priv, BXT_REVID_B0, REVID_FOREVER))
+ I915_WRITE(GEN8_L3SQCREG1, L3_GENERAL_PRIO_CREDITS(62) |
+ L3_HIGH_PRIO_CREDITS(2));
+
return 0;
}
int init_workarounds_ring(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
WARN_ON(engine->id != RCS);
dev_priv->workarounds.count = 0;
dev_priv->workarounds.hw_whitelist_count[RCS] = 0;
- if (IS_BROADWELL(dev))
+ if (IS_BROADWELL(dev_priv))
return bdw_init_workarounds(engine);
- if (IS_CHERRYVIEW(dev))
+ if (IS_CHERRYVIEW(dev_priv))
return chv_init_workarounds(engine);
- if (IS_SKYLAKE(dev))
+ if (IS_SKYLAKE(dev_priv))
return skl_init_workarounds(engine);
- if (IS_BROXTON(dev))
+ if (IS_BROXTON(dev_priv))
return bxt_init_workarounds(engine);
return 0;
@@ -1204,14 +1200,13 @@ int init_workarounds_ring(struct intel_engine_cs *engine)
static int init_render_ring(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
int ret = init_ring_common(engine);
if (ret)
return ret;
/* WaTimedSingleVertexDispatch:cl,bw,ctg,elk,ilk,snb */
- if (INTEL_INFO(dev)->gen >= 4 && INTEL_INFO(dev)->gen < 7)
+ if (IS_GEN(dev_priv, 4, 6))
I915_WRITE(MI_MODE, _MASKED_BIT_ENABLE(VS_TIMER_DISPATCH));
/* We need to disable the AsyncFlip performance optimisations in order
@@ -1220,22 +1215,22 @@ static int init_render_ring(struct intel_engine_cs *engine)
*
* WaDisableAsyncFlipPerfMode:snb,ivb,hsw,vlv
*/
- if (INTEL_INFO(dev)->gen >= 6 && INTEL_INFO(dev)->gen < 8)
+ if (IS_GEN(dev_priv, 6, 7))
I915_WRITE(MI_MODE, _MASKED_BIT_ENABLE(ASYNC_FLIP_PERF_DISABLE));
/* Required for the hardware to program scanline values for waiting */
/* WaEnableFlushTlbInvalidationMode:snb */
- if (INTEL_INFO(dev)->gen == 6)
+ if (IS_GEN6(dev_priv))
I915_WRITE(GFX_MODE,
_MASKED_BIT_ENABLE(GFX_TLB_INVALIDATE_EXPLICIT));
/* WaBCSVCSTlbInvalidationMode:ivb,vlv,hsw */
- if (IS_GEN7(dev))
+ if (IS_GEN7(dev_priv))
I915_WRITE(GFX_MODE_GEN7,
_MASKED_BIT_ENABLE(GFX_TLB_INVALIDATE_EXPLICIT) |
_MASKED_BIT_ENABLE(GFX_REPLAY_MODE));
- if (IS_GEN6(dev)) {
+ if (IS_GEN6(dev_priv)) {
/* From the Sandybridge PRM, volume 1 part 3, page 24:
* "If this bit is set, STCunit will have LRA as replacement
* policy. [...] This bit must be reset. LRA replacement
@@ -1245,19 +1240,18 @@ static int init_render_ring(struct intel_engine_cs *engine)
_MASKED_BIT_DISABLE(CM0_STC_EVICT_DISABLE_LRA_SNB));
}
- if (INTEL_INFO(dev)->gen >= 6 && INTEL_INFO(dev)->gen < 8)
+ if (IS_GEN(dev_priv, 6, 7))
I915_WRITE(INSTPM, _MASKED_BIT_ENABLE(INSTPM_FORCE_ORDERING));
- if (HAS_L3_DPF(dev))
- I915_WRITE_IMR(engine, ~GT_PARITY_ERROR(dev));
+ if (HAS_L3_DPF(dev_priv))
+ I915_WRITE_IMR(engine, ~GT_PARITY_ERROR(dev_priv));
return init_workarounds_ring(engine);
}
static void render_ring_cleanup(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
if (dev_priv->semaphore_obj) {
i915_gem_object_ggtt_unpin(dev_priv->semaphore_obj);
@@ -1273,13 +1267,12 @@ static int gen8_rcs_signal(struct drm_i915_gem_request *signaller_req,
{
#define MBOX_UPDATE_DWORDS 8
struct intel_engine_cs *signaller = signaller_req->engine;
- struct drm_device *dev = signaller->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = signaller_req->i915;
struct intel_engine_cs *waiter;
enum intel_engine_id id;
int ret, num_rings;
- num_rings = hweight32(INTEL_INFO(dev)->ring_mask);
+ num_rings = hweight32(INTEL_INFO(dev_priv)->ring_mask);
num_dwords += (num_rings-1) * MBOX_UPDATE_DWORDS;
#undef MBOX_UPDATE_DWORDS
@@ -1297,7 +1290,7 @@ static int gen8_rcs_signal(struct drm_i915_gem_request *signaller_req,
intel_ring_emit(signaller, GFX_OP_PIPE_CONTROL(6));
intel_ring_emit(signaller, PIPE_CONTROL_GLOBAL_GTT_IVB |
PIPE_CONTROL_QW_WRITE |
- PIPE_CONTROL_FLUSH_ENABLE);
+ PIPE_CONTROL_CS_STALL);
intel_ring_emit(signaller, lower_32_bits(gtt_offset));
intel_ring_emit(signaller, upper_32_bits(gtt_offset));
intel_ring_emit(signaller, seqno);
@@ -1315,13 +1308,12 @@ static int gen8_xcs_signal(struct drm_i915_gem_request *signaller_req,
{
#define MBOX_UPDATE_DWORDS 6
struct intel_engine_cs *signaller = signaller_req->engine;
- struct drm_device *dev = signaller->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = signaller_req->i915;
struct intel_engine_cs *waiter;
enum intel_engine_id id;
int ret, num_rings;
- num_rings = hweight32(INTEL_INFO(dev)->ring_mask);
+ num_rings = hweight32(INTEL_INFO(dev_priv)->ring_mask);
num_dwords += (num_rings-1) * MBOX_UPDATE_DWORDS;
#undef MBOX_UPDATE_DWORDS
@@ -1354,14 +1346,13 @@ static int gen6_signal(struct drm_i915_gem_request *signaller_req,
unsigned int num_dwords)
{
struct intel_engine_cs *signaller = signaller_req->engine;
- struct drm_device *dev = signaller->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = signaller_req->i915;
struct intel_engine_cs *useless;
enum intel_engine_id id;
int ret, num_rings;
#define MBOX_UPDATE_DWORDS 3
- num_rings = hweight32(INTEL_INFO(dev)->ring_mask);
+ num_rings = hweight32(INTEL_INFO(dev_priv)->ring_mask);
num_dwords += round_up((num_rings-1) * MBOX_UPDATE_DWORDS, 2);
#undef MBOX_UPDATE_DWORDS
@@ -1420,10 +1411,38 @@ gen6_add_request(struct drm_i915_gem_request *req)
return 0;
}
-static inline bool i915_gem_has_seqno_wrapped(struct drm_device *dev,
+static int
+gen8_render_add_request(struct drm_i915_gem_request *req)
+{
+ struct intel_engine_cs *engine = req->engine;
+ int ret;
+
+ if (engine->semaphore.signal)
+ ret = engine->semaphore.signal(req, 8);
+ else
+ ret = intel_ring_begin(req, 8);
+ if (ret)
+ return ret;
+
+ intel_ring_emit(engine, GFX_OP_PIPE_CONTROL(6));
+ intel_ring_emit(engine, (PIPE_CONTROL_GLOBAL_GTT_IVB |
+ PIPE_CONTROL_CS_STALL |
+ PIPE_CONTROL_QW_WRITE));
+ intel_ring_emit(engine, intel_hws_seqno_address(req->engine));
+ intel_ring_emit(engine, 0);
+ intel_ring_emit(engine, i915_gem_request_get_seqno(req));
+ /* We're thrashing one dword of HWS. */
+ intel_ring_emit(engine, 0);
+ intel_ring_emit(engine, MI_USER_INTERRUPT);
+ intel_ring_emit(engine, MI_NOOP);
+ __intel_ring_advance(engine);
+
+ return 0;
+}
+
+static inline bool i915_gem_has_seqno_wrapped(struct drm_i915_private *dev_priv,
u32 seqno)
{
- struct drm_i915_private *dev_priv = dev->dev_private;
return dev_priv->last_seqno < seqno;
}
@@ -1441,7 +1460,8 @@ gen8_ring_sync(struct drm_i915_gem_request *waiter_req,
u32 seqno)
{
struct intel_engine_cs *waiter = waiter_req->engine;
- struct drm_i915_private *dev_priv = waiter->dev->dev_private;
+ struct drm_i915_private *dev_priv = waiter_req->i915;
+ struct i915_hw_ppgtt *ppgtt;
int ret;
ret = intel_ring_begin(waiter_req, 4);
@@ -1450,7 +1470,6 @@ gen8_ring_sync(struct drm_i915_gem_request *waiter_req,
intel_ring_emit(waiter, MI_SEMAPHORE_WAIT |
MI_SEMAPHORE_GLOBAL_GTT |
- MI_SEMAPHORE_POLL |
MI_SEMAPHORE_SAD_GTE_SDD);
intel_ring_emit(waiter, seqno);
intel_ring_emit(waiter,
@@ -1458,6 +1477,15 @@ gen8_ring_sync(struct drm_i915_gem_request *waiter_req,
intel_ring_emit(waiter,
upper_32_bits(GEN8_WAIT_OFFSET(waiter, signaller->id)));
intel_ring_advance(waiter);
+
+ /* When the !RCS engines idle waiting upon a semaphore, they lose their
+ * pagetables and we must reload them before executing the batch.
+ * We do this on the i915_switch_context() following the wait and
+ * before the dispatch.
+ */
+ ppgtt = waiter_req->ctx->ppgtt;
+ if (ppgtt && waiter_req->engine->id != RCS)
+ ppgtt->pd_dirty_rings |= intel_engine_flag(waiter_req->engine);
return 0;
}
@@ -1486,7 +1514,7 @@ gen6_ring_sync(struct drm_i915_gem_request *waiter_req,
return ret;
/* If seqno wrap happened, omit the wait with no-ops */
- if (likely(!i915_gem_has_seqno_wrapped(waiter->dev, seqno))) {
+ if (likely(!i915_gem_has_seqno_wrapped(waiter_req->i915, seqno))) {
intel_ring_emit(waiter, dw1 | wait_mbox);
intel_ring_emit(waiter, seqno);
intel_ring_emit(waiter, 0);
@@ -1567,7 +1595,7 @@ pc_render_add_request(struct drm_i915_gem_request *req)
static void
gen6_seqno_barrier(struct intel_engine_cs *engine)
{
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
/* Workaround to force correct ordering between irq and seqno writes on
* ivb (and maybe also on snb) by reading from a CS register (like
@@ -1616,8 +1644,7 @@ pc_render_set_seqno(struct intel_engine_cs *engine, u32 seqno)
static bool
gen5_ring_get_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
if (WARN_ON(!intel_irqs_enabled(dev_priv)))
@@ -1634,8 +1661,7 @@ gen5_ring_get_irq(struct intel_engine_cs *engine)
static void
gen5_ring_put_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
spin_lock_irqsave(&dev_priv->irq_lock, flags);
@@ -1647,8 +1673,7 @@ gen5_ring_put_irq(struct intel_engine_cs *engine)
static bool
i9xx_ring_get_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
if (!intel_irqs_enabled(dev_priv))
@@ -1668,8 +1693,7 @@ i9xx_ring_get_irq(struct intel_engine_cs *engine)
static void
i9xx_ring_put_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
spin_lock_irqsave(&dev_priv->irq_lock, flags);
@@ -1684,8 +1708,7 @@ i9xx_ring_put_irq(struct intel_engine_cs *engine)
static bool
i8xx_ring_get_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
if (!intel_irqs_enabled(dev_priv))
@@ -1705,8 +1728,7 @@ i8xx_ring_get_irq(struct intel_engine_cs *engine)
static void
i8xx_ring_put_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
spin_lock_irqsave(&dev_priv->irq_lock, flags);
@@ -1759,8 +1781,7 @@ i9xx_add_request(struct drm_i915_gem_request *req)
static bool
gen6_ring_get_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
if (WARN_ON(!intel_irqs_enabled(dev_priv)))
@@ -1768,10 +1789,10 @@ gen6_ring_get_irq(struct intel_engine_cs *engine)
spin_lock_irqsave(&dev_priv->irq_lock, flags);
if (engine->irq_refcount++ == 0) {
- if (HAS_L3_DPF(dev) && engine->id == RCS)
+ if (HAS_L3_DPF(dev_priv) && engine->id == RCS)
I915_WRITE_IMR(engine,
~(engine->irq_enable_mask |
- GT_PARITY_ERROR(dev)));
+ GT_PARITY_ERROR(dev_priv)));
else
I915_WRITE_IMR(engine, ~engine->irq_enable_mask);
gen5_enable_gt_irq(dev_priv, engine->irq_enable_mask);
@@ -1784,14 +1805,13 @@ gen6_ring_get_irq(struct intel_engine_cs *engine)
static void
gen6_ring_put_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
spin_lock_irqsave(&dev_priv->irq_lock, flags);
if (--engine->irq_refcount == 0) {
- if (HAS_L3_DPF(dev) && engine->id == RCS)
- I915_WRITE_IMR(engine, ~GT_PARITY_ERROR(dev));
+ if (HAS_L3_DPF(dev_priv) && engine->id == RCS)
+ I915_WRITE_IMR(engine, ~GT_PARITY_ERROR(dev_priv));
else
I915_WRITE_IMR(engine, ~0);
gen5_disable_gt_irq(dev_priv, engine->irq_enable_mask);
@@ -1802,8 +1822,7 @@ gen6_ring_put_irq(struct intel_engine_cs *engine)
static bool
hsw_vebox_get_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
if (WARN_ON(!intel_irqs_enabled(dev_priv)))
@@ -1822,8 +1841,7 @@ hsw_vebox_get_irq(struct intel_engine_cs *engine)
static void
hsw_vebox_put_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
spin_lock_irqsave(&dev_priv->irq_lock, flags);
@@ -1837,8 +1855,7 @@ hsw_vebox_put_irq(struct intel_engine_cs *engine)
static bool
gen8_ring_get_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
if (WARN_ON(!intel_irqs_enabled(dev_priv)))
@@ -1846,7 +1863,7 @@ gen8_ring_get_irq(struct intel_engine_cs *engine)
spin_lock_irqsave(&dev_priv->irq_lock, flags);
if (engine->irq_refcount++ == 0) {
- if (HAS_L3_DPF(dev) && engine->id == RCS) {
+ if (HAS_L3_DPF(dev_priv) && engine->id == RCS) {
I915_WRITE_IMR(engine,
~(engine->irq_enable_mask |
GT_RENDER_L3_PARITY_ERROR_INTERRUPT));
@@ -1863,13 +1880,12 @@ gen8_ring_get_irq(struct intel_engine_cs *engine)
static void
gen8_ring_put_irq(struct intel_engine_cs *engine)
{
- struct drm_device *dev = engine->dev;
- struct drm_i915_private *dev_priv = dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
unsigned long flags;
spin_lock_irqsave(&dev_priv->irq_lock, flags);
if (--engine->irq_refcount == 0) {
- if (HAS_L3_DPF(dev) && engine->id == RCS) {
+ if (HAS_L3_DPF(dev_priv) && engine->id == RCS) {
I915_WRITE_IMR(engine,
~GT_RENDER_L3_PARITY_ERROR_INTERRUPT);
} else {
@@ -1991,12 +2007,12 @@ i915_dispatch_execbuffer(struct drm_i915_gem_request *req,
static void cleanup_phys_status_page(struct intel_engine_cs *engine)
{
- struct drm_i915_private *dev_priv = to_i915(engine->dev);
+ struct drm_i915_private *dev_priv = engine->i915;
if (!dev_priv->status_page_dmah)
return;
- drm_pci_free(engine->dev, dev_priv->status_page_dmah);
+ drm_pci_free(dev_priv->dev, dev_priv->status_page_dmah);
engine->status_page.page_addr = NULL;
}
@@ -2022,10 +2038,10 @@ static int init_status_page(struct intel_engine_cs *engine)
unsigned flags;
int ret;
- obj = i915_gem_alloc_object(engine->dev, 4096);
- if (obj == NULL) {
+ obj = i915_gem_object_create(engine->i915->dev, 4096);
+ if (IS_ERR(obj)) {
DRM_ERROR("Failed to allocate status page\n");
- return -ENOMEM;
+ return PTR_ERR(obj);
}
ret = i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
@@ -2033,7 +2049,7 @@ static int init_status_page(struct intel_engine_cs *engine)
goto err_unref;
flags = 0;
- if (!HAS_LLC(engine->dev))
+ if (!HAS_LLC(engine->i915))
/* On g33, we cannot place HWS above 256MiB, so
* restrict its pinning to the low mappable arena.
* Though this restriction is not documented for
@@ -2067,11 +2083,11 @@ err_unref:
static int init_phys_status_page(struct intel_engine_cs *engine)
{
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
if (!dev_priv->status_page_dmah) {
dev_priv->status_page_dmah =
- drm_pci_alloc(engine->dev, PAGE_SIZE, PAGE_SIZE);
+ drm_pci_alloc(dev_priv->dev, PAGE_SIZE, PAGE_SIZE);
if (!dev_priv->status_page_dmah)
return -ENOMEM;
}
@@ -2084,20 +2100,22 @@ static int init_phys_status_page(struct intel_engine_cs *engine)
void intel_unpin_ringbuffer_obj(struct intel_ringbuffer *ringbuf)
{
+ GEM_BUG_ON(ringbuf->vma == NULL);
+ GEM_BUG_ON(ringbuf->virtual_start == NULL);
+
if (HAS_LLC(ringbuf->obj->base.dev) && !ringbuf->obj->stolen)
i915_gem_object_unpin_map(ringbuf->obj);
else
- iounmap(ringbuf->virtual_start);
+ i915_vma_unpin_iomap(ringbuf->vma);
ringbuf->virtual_start = NULL;
- ringbuf->vma = NULL;
+
i915_gem_object_ggtt_unpin(ringbuf->obj);
+ ringbuf->vma = NULL;
}
-int intel_pin_and_map_ringbuffer_obj(struct drm_device *dev,
+int intel_pin_and_map_ringbuffer_obj(struct drm_i915_private *dev_priv,
struct intel_ringbuffer *ringbuf)
{
- struct drm_i915_private *dev_priv = to_i915(dev);
- struct i915_ggtt *ggtt = &dev_priv->ggtt;
struct drm_i915_gem_object *obj = ringbuf->obj;
/* Ring wraparound at offset 0 sometimes hangs. No idea why. */
unsigned flags = PIN_OFFSET_BIAS | 4096;
@@ -2131,10 +2149,9 @@ int intel_pin_and_map_ringbuffer_obj(struct drm_device *dev,
/* Access through the GTT requires the device to be awake. */
assert_rpm_wakelock_held(dev_priv);
- addr = ioremap_wc(ggtt->mappable_base +
- i915_gem_obj_ggtt_offset(obj), ringbuf->size);
- if (addr == NULL) {
- ret = -ENOMEM;
+ addr = i915_vma_pin_iomap(i915_gem_obj_to_ggtt(obj));
+ if (IS_ERR(addr)) {
+ ret = PTR_ERR(addr);
goto err_unpin;
}
}
@@ -2163,9 +2180,9 @@ static int intel_alloc_ringbuffer_obj(struct drm_device *dev,
if (!HAS_LLC(dev))
obj = i915_gem_object_create_stolen(dev, ringbuf->size);
if (obj == NULL)
- obj = i915_gem_alloc_object(dev, ringbuf->size);
- if (obj == NULL)
- return -ENOMEM;
+ obj = i915_gem_object_create(dev, ringbuf->size);
+ if (IS_ERR(obj))
+ return PTR_ERR(obj);
/* mark ring buffers as read-only from GPU side by default */
obj->gt_ro = 1;
@@ -2197,13 +2214,13 @@ intel_engine_create_ringbuffer(struct intel_engine_cs *engine, int size)
* of the buffer.
*/
ring->effective_size = size;
- if (IS_I830(engine->dev) || IS_845G(engine->dev))
+ if (IS_I830(engine->i915) || IS_845G(engine->i915))
ring->effective_size -= 2 * CACHELINE_BYTES;
ring->last_retired_head = -1;
intel_ring_update_space(ring);
- ret = intel_alloc_ringbuffer_obj(engine->dev, ring);
+ ret = intel_alloc_ringbuffer_obj(engine->i915->dev, ring);
if (ret) {
DRM_DEBUG_DRIVER("Failed to allocate ringbuffer %s: %d\n",
engine->name, ret);
@@ -2226,12 +2243,13 @@ intel_ringbuffer_free(struct intel_ringbuffer *ring)
static int intel_init_ring_buffer(struct drm_device *dev,
struct intel_engine_cs *engine)
{
+ struct drm_i915_private *dev_priv = to_i915(dev);
struct intel_ringbuffer *ringbuf;
int ret;
WARN_ON(engine->buffer);
- engine->dev = dev;
+ engine->i915 = dev_priv;
INIT_LIST_HEAD(&engine->active_list);
INIT_LIST_HEAD(&engine->request_list);
INIT_LIST_HEAD(&engine->execlist_queue);
@@ -2249,7 +2267,7 @@ static int intel_init_ring_buffer(struct drm_device *dev,
}
engine->buffer = ringbuf;
- if (I915_NEED_GFX_HWS(dev)) {
+ if (I915_NEED_GFX_HWS(dev_priv)) {
ret = init_status_page(engine);
if (ret)
goto error;
@@ -2260,7 +2278,7 @@ static int intel_init_ring_buffer(struct drm_device *dev,
goto error;
}
- ret = intel_pin_and_map_ringbuffer_obj(dev, ringbuf);
+ ret = intel_pin_and_map_ringbuffer_obj(dev_priv, ringbuf);
if (ret) {
DRM_ERROR("Failed to pin and map ringbuffer %s: %d\n",
engine->name, ret);
@@ -2286,11 +2304,11 @@ void intel_cleanup_engine(struct intel_engine_cs *engine)
if (!intel_engine_initialized(engine))
return;
- dev_priv = to_i915(engine->dev);
+ dev_priv = engine->i915;
if (engine->buffer) {
intel_stop_engine(engine);
- WARN_ON(!IS_GEN2(engine->dev) && (I915_READ_MODE(engine) & MODE_IDLE) == 0);
+ WARN_ON(!IS_GEN2(dev_priv) && (I915_READ_MODE(engine) & MODE_IDLE) == 0);
intel_unpin_ringbuffer_obj(engine->buffer);
intel_ringbuffer_free(engine->buffer);
@@ -2300,7 +2318,7 @@ void intel_cleanup_engine(struct intel_engine_cs *engine)
if (engine->cleanup)
engine->cleanup(engine);
- if (I915_NEED_GFX_HWS(engine->dev)) {
+ if (I915_NEED_GFX_HWS(dev_priv)) {
cleanup_status_page(engine);
} else {
WARN_ON(engine->id != RCS);
@@ -2309,7 +2327,7 @@ void intel_cleanup_engine(struct intel_engine_cs *engine)
i915_cmd_parser_fini_ring(engine);
i915_gem_batch_pool_fini(&engine->batch_pool);
- engine->dev = NULL;
+ engine->i915 = NULL;
}
int intel_engine_idle(struct intel_engine_cs *engine)
@@ -2332,46 +2350,22 @@ int intel_engine_idle(struct intel_engine_cs *engine)
int intel_ring_alloc_request_extras(struct drm_i915_gem_request *request)
{
- request->ringbuf = request->engine->buffer;
- return 0;
-}
+ int ret;
-int intel_ring_reserve_space(struct drm_i915_gem_request *request)
-{
- /*
- * The first call merely notes the reserve request and is common for
- * all back ends. The subsequent localised _begin() call actually
- * ensures that the reservation is available. Without the begin, if
- * the request creator immediately submitted the request without
- * adding any commands to it then there might not actually be
- * sufficient room for the submission commands.
+ /* Flush enough space to reduce the likelihood of waiting after
+ * we start building the request - in which case we will just
+ * have to repeat work.
*/
- intel_ring_reserved_space_reserve(request->ringbuf, MIN_SPACE_FOR_ADD_REQUEST);
+ request->reserved_space += LEGACY_REQUEST_SIZE;
- return intel_ring_begin(request, 0);
-}
-
-void intel_ring_reserved_space_reserve(struct intel_ringbuffer *ringbuf, int size)
-{
- GEM_BUG_ON(ringbuf->reserved_size);
- ringbuf->reserved_size = size;
-}
-
-void intel_ring_reserved_space_cancel(struct intel_ringbuffer *ringbuf)
-{
- GEM_BUG_ON(!ringbuf->reserved_size);
- ringbuf->reserved_size = 0;
-}
+ request->ringbuf = request->engine->buffer;
-void intel_ring_reserved_space_use(struct intel_ringbuffer *ringbuf)
-{
- GEM_BUG_ON(!ringbuf->reserved_size);
- ringbuf->reserved_size = 0;
-}
+ ret = intel_ring_begin(request, 0);
+ if (ret)
+ return ret;
-void intel_ring_reserved_space_end(struct intel_ringbuffer *ringbuf)
-{
- GEM_BUG_ON(ringbuf->reserved_size);
+ request->reserved_space -= LEGACY_REQUEST_SIZE;
+ return 0;
}
static int wait_for_space(struct drm_i915_gem_request *req, int bytes)
@@ -2393,7 +2387,7 @@ static int wait_for_space(struct drm_i915_gem_request *req, int bytes)
*
* See also i915_gem_request_alloc() and i915_add_request().
*/
- GEM_BUG_ON(!ringbuf->reserved_size);
+ GEM_BUG_ON(!req->reserved_space);
list_for_each_entry(target, &engine->request_list, list) {
unsigned space;
@@ -2428,7 +2422,7 @@ int intel_ring_begin(struct drm_i915_gem_request *req, int num_dwords)
int total_bytes, wait_bytes;
bool need_wrap = false;
- total_bytes = bytes + ringbuf->reserved_size;
+ total_bytes = bytes + req->reserved_space;
if (unlikely(bytes > remain_usable)) {
/*
@@ -2444,7 +2438,7 @@ int intel_ring_begin(struct drm_i915_gem_request *req, int num_dwords)
* and only need to effectively wait for the reserved
* size space from the start of ringbuffer.
*/
- wait_bytes = remain_actual + ringbuf->reserved_size;
+ wait_bytes = remain_actual + req->reserved_space;
} else {
/* No wrapping required, just waiting. */
wait_bytes = total_bytes;
@@ -2501,7 +2495,7 @@ int intel_ring_cacheline_align(struct drm_i915_gem_request *req)
void intel_ring_init_seqno(struct intel_engine_cs *engine, u32 seqno)
{
- struct drm_i915_private *dev_priv = to_i915(engine->dev);
+ struct drm_i915_private *dev_priv = engine->i915;
/* Our semaphore implementation is strictly monotonic (i.e. we proceed
* so long as the semaphore value in the register/page is greater
@@ -2511,7 +2505,7 @@ void intel_ring_init_seqno(struct intel_engine_cs *engine, u32 seqno)
* the semaphore value, then when the seqno moves backwards all
* future waits will complete instantly (causing rendering corruption).
*/
- if (INTEL_INFO(dev_priv)->gen == 6 || INTEL_INFO(dev_priv)->gen == 7) {
+ if (IS_GEN6(dev_priv) || IS_GEN7(dev_priv)) {
I915_WRITE(RING_SYNC_0(engine->mmio_base), 0);
I915_WRITE(RING_SYNC_1(engine->mmio_base), 0);
if (HAS_VEBOX(dev_priv))
@@ -2537,7 +2531,7 @@ void intel_ring_init_seqno(struct intel_engine_cs *engine, u32 seqno)
static void gen6_bsd_ring_write_tail(struct intel_engine_cs *engine,
u32 value)
{
- struct drm_i915_private *dev_priv = engine->dev->dev_private;
+ struct drm_i915_private *dev_priv = engine->i915;
/* Every tail move must follow the sequence below */
@@ -2579,7 +2573,7 @@ static int gen6_bsd_ring_flush(struct drm_i915_gem_request *req,
return ret;
cmd = MI_FLUSH_DW;
- if (INTEL_INFO(engine->dev)->gen >= 8)
+ if (INTEL_GEN(req->i915) >= 8)
cmd += 1;
/* We always require a command barrier so that subsequent
@@ -2601,7 +2595,7 @@ static int gen6_bsd_ring_flush(struct drm_i915_gem_request *req,
intel_ring_emit(engine, cmd);
intel_ring_emit(engine,
I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);
- if (INTEL_INFO(engine->dev)->gen >= 8) {
+ if (INTEL_GEN(req->i915) >= 8) {
intel_ring_emit(engine, 0); /* upper addr */
intel_ring_emit(engine, 0); /* value */
} else {
@@ -2692,7 +2686,6 @@ static int gen6_ring_flush(struct drm_i915_gem_request *req,
u32 invalidate, u32 flush)
{
struct intel_engine_cs *engine = req->engine;
- struct drm_device *dev = engine->dev;
uint32_t cmd;
int ret;
@@ -2701,7 +2694,7 @@ static int gen6_ring_flush(struct drm_i915_gem_request *req,
return ret;
cmd = MI_FLUSH_DW;
- if (INTEL_INFO(dev)->gen >= 8)
+ if (INTEL_GEN(req->i915) >= 8)
cmd += 1;
/* We always require a command barrier so that subsequent
@@ -2722,7 +2715,7 @@ static int gen6_ring_flush(struct drm_i915_gem_request *req,
intel_ring_emit(engine, cmd);
intel_ring_emit(engine,
I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT);
- if (INTEL_INFO(dev)->gen >= 8) {
+ if (INTEL_GEN(req->i915) >= 8) {
intel_ring_emit(engine, 0); /* upper addr */
intel_ring_emit(engine, 0); /* value */
} else {
@@ -2747,10 +2740,10 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
engine->hw_id = 0;
engine->mmio_base = RENDER_RING_BASE;
- if (INTEL_INFO(dev)->gen >= 8) {
- if (i915_semaphore_is_enabled(dev)) {
- obj = i915_gem_alloc_object(dev, 4096);
- if (obj == NULL) {
+ if (INTEL_GEN(dev_priv) >= 8) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
+ obj = i915_gem_object_create(dev, 4096);
+ if (IS_ERR(obj)) {
DRM_ERROR("Failed to allocate semaphore bo. Disabling semaphores\n");
i915.semaphores = 0;
} else {
@@ -2766,25 +2759,24 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
}
engine->init_context = intel_rcs_ctx_init;
- engine->add_request = gen6_add_request;
+ engine->add_request = gen8_render_add_request;
engine->flush = gen8_render_ring_flush;
engine->irq_get = gen8_ring_get_irq;
engine->irq_put = gen8_ring_put_irq;
engine->irq_enable_mask = GT_RENDER_USER_INTERRUPT;
- engine->irq_seqno_barrier = gen6_seqno_barrier;
engine->get_seqno = ring_get_seqno;
engine->set_seqno = ring_set_seqno;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
WARN_ON(!dev_priv->semaphore_obj);
engine->semaphore.sync_to = gen8_ring_sync;
engine->semaphore.signal = gen8_rcs_signal;
GEN8_RING_SEMAPHORE_INIT(engine);
}
- } else if (INTEL_INFO(dev)->gen >= 6) {
+ } else if (INTEL_GEN(dev_priv) >= 6) {
engine->init_context = intel_rcs_ctx_init;
engine->add_request = gen6_add_request;
engine->flush = gen7_render_ring_flush;
- if (INTEL_INFO(dev)->gen == 6)
+ if (IS_GEN6(dev_priv))
engine->flush = gen6_render_ring_flush;
engine->irq_get = gen6_ring_get_irq;
engine->irq_put = gen6_ring_put_irq;
@@ -2792,7 +2784,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
engine->irq_seqno_barrier = gen6_seqno_barrier;
engine->get_seqno = ring_get_seqno;
engine->set_seqno = ring_set_seqno;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
engine->semaphore.sync_to = gen6_ring_sync;
engine->semaphore.signal = gen6_signal;
/*
@@ -2813,7 +2805,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
engine->semaphore.mbox.signal[VECS] = GEN6_VERSYNC;
engine->semaphore.mbox.signal[VCS2] = GEN6_NOSYNC;
}
- } else if (IS_GEN5(dev)) {
+ } else if (IS_GEN5(dev_priv)) {
engine->add_request = pc_render_add_request;
engine->flush = gen4_render_ring_flush;
engine->get_seqno = pc_render_get_seqno;
@@ -2824,13 +2816,13 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
GT_RENDER_PIPECTL_NOTIFY_INTERRUPT;
} else {
engine->add_request = i9xx_add_request;
- if (INTEL_INFO(dev)->gen < 4)
+ if (INTEL_GEN(dev_priv) < 4)
engine->flush = gen2_render_ring_flush;
else
engine->flush = gen4_render_ring_flush;
engine->get_seqno = ring_get_seqno;
engine->set_seqno = ring_set_seqno;
- if (IS_GEN2(dev)) {
+ if (IS_GEN2(dev_priv)) {
engine->irq_get = i8xx_ring_get_irq;
engine->irq_put = i8xx_ring_put_irq;
} else {
@@ -2841,15 +2833,15 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
}
engine->write_tail = ring_write_tail;
- if (IS_HASWELL(dev))
+ if (IS_HASWELL(dev_priv))
engine->dispatch_execbuffer = hsw_ring_dispatch_execbuffer;
- else if (IS_GEN8(dev))
+ else if (IS_GEN8(dev_priv))
engine->dispatch_execbuffer = gen8_ring_dispatch_execbuffer;
- else if (INTEL_INFO(dev)->gen >= 6)
+ else if (INTEL_GEN(dev_priv) >= 6)
engine->dispatch_execbuffer = gen6_ring_dispatch_execbuffer;
- else if (INTEL_INFO(dev)->gen >= 4)
+ else if (INTEL_GEN(dev_priv) >= 4)
engine->dispatch_execbuffer = i965_dispatch_execbuffer;
- else if (IS_I830(dev) || IS_845G(dev))
+ else if (IS_I830(dev_priv) || IS_845G(dev_priv))
engine->dispatch_execbuffer = i830_dispatch_execbuffer;
else
engine->dispatch_execbuffer = i915_dispatch_execbuffer;
@@ -2857,11 +2849,11 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
engine->cleanup = render_ring_cleanup;
/* Workaround batchbuffer to combat CS tlb bug. */
- if (HAS_BROKEN_CS_TLB(dev)) {
- obj = i915_gem_alloc_object(dev, I830_WA_SIZE);
- if (obj == NULL) {
+ if (HAS_BROKEN_CS_TLB(dev_priv)) {
+ obj = i915_gem_object_create(dev, I830_WA_SIZE);
+ if (IS_ERR(obj)) {
DRM_ERROR("Failed to allocate batch bo\n");
- return -ENOMEM;
+ return PTR_ERR(obj);
}
ret = i915_gem_obj_ggtt_pin(obj, 0, 0);
@@ -2879,7 +2871,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
if (ret)
return ret;
- if (INTEL_INFO(dev)->gen >= 5) {
+ if (INTEL_GEN(dev_priv) >= 5) {
ret = intel_init_pipe_control(engine);
if (ret)
return ret;
@@ -2899,24 +2891,24 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev)
engine->hw_id = 1;
engine->write_tail = ring_write_tail;
- if (INTEL_INFO(dev)->gen >= 6) {
+ if (INTEL_GEN(dev_priv) >= 6) {
engine->mmio_base = GEN6_BSD_RING_BASE;
/* gen6 bsd needs a special wa for tail updates */
- if (IS_GEN6(dev))
+ if (IS_GEN6(dev_priv))
engine->write_tail = gen6_bsd_ring_write_tail;
engine->flush = gen6_bsd_ring_flush;
engine->add_request = gen6_add_request;
engine->irq_seqno_barrier = gen6_seqno_barrier;
engine->get_seqno = ring_get_seqno;
engine->set_seqno = ring_set_seqno;
- if (INTEL_INFO(dev)->gen >= 8) {
+ if (INTEL_GEN(dev_priv) >= 8) {
engine->irq_enable_mask =
GT_RENDER_USER_INTERRUPT << GEN8_VCS1_IRQ_SHIFT;
engine->irq_get = gen8_ring_get_irq;
engine->irq_put = gen8_ring_put_irq;
engine->dispatch_execbuffer =
gen8_ring_dispatch_execbuffer;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
engine->semaphore.sync_to = gen8_ring_sync;
engine->semaphore.signal = gen8_xcs_signal;
GEN8_RING_SEMAPHORE_INIT(engine);
@@ -2927,7 +2919,7 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev)
engine->irq_put = gen6_ring_put_irq;
engine->dispatch_execbuffer =
gen6_ring_dispatch_execbuffer;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
engine->semaphore.sync_to = gen6_ring_sync;
engine->semaphore.signal = gen6_signal;
engine->semaphore.mbox.wait[RCS] = MI_SEMAPHORE_SYNC_VR;
@@ -2948,7 +2940,7 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev)
engine->add_request = i9xx_add_request;
engine->get_seqno = ring_get_seqno;
engine->set_seqno = ring_set_seqno;
- if (IS_GEN5(dev)) {
+ if (IS_GEN5(dev_priv)) {
engine->irq_enable_mask = ILK_BSD_USER_INTERRUPT;
engine->irq_get = gen5_ring_get_irq;
engine->irq_put = gen5_ring_put_irq;
@@ -2990,7 +2982,7 @@ int intel_init_bsd2_ring_buffer(struct drm_device *dev)
engine->irq_put = gen8_ring_put_irq;
engine->dispatch_execbuffer =
gen8_ring_dispatch_execbuffer;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
engine->semaphore.sync_to = gen8_ring_sync;
engine->semaphore.signal = gen8_xcs_signal;
GEN8_RING_SEMAPHORE_INIT(engine);
@@ -3017,13 +3009,13 @@ int intel_init_blt_ring_buffer(struct drm_device *dev)
engine->irq_seqno_barrier = gen6_seqno_barrier;
engine->get_seqno = ring_get_seqno;
engine->set_seqno = ring_set_seqno;
- if (INTEL_INFO(dev)->gen >= 8) {
+ if (INTEL_GEN(dev_priv) >= 8) {
engine->irq_enable_mask =
GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT;
engine->irq_get = gen8_ring_get_irq;
engine->irq_put = gen8_ring_put_irq;
engine->dispatch_execbuffer = gen8_ring_dispatch_execbuffer;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
engine->semaphore.sync_to = gen8_ring_sync;
engine->semaphore.signal = gen8_xcs_signal;
GEN8_RING_SEMAPHORE_INIT(engine);
@@ -3033,7 +3025,7 @@ int intel_init_blt_ring_buffer(struct drm_device *dev)
engine->irq_get = gen6_ring_get_irq;
engine->irq_put = gen6_ring_put_irq;
engine->dispatch_execbuffer = gen6_ring_dispatch_execbuffer;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
engine->semaphore.signal = gen6_signal;
engine->semaphore.sync_to = gen6_ring_sync;
/*
@@ -3078,13 +3070,13 @@ int intel_init_vebox_ring_buffer(struct drm_device *dev)
engine->get_seqno = ring_get_seqno;
engine->set_seqno = ring_set_seqno;
- if (INTEL_INFO(dev)->gen >= 8) {
+ if (INTEL_GEN(dev_priv) >= 8) {
engine->irq_enable_mask =
GT_RENDER_USER_INTERRUPT << GEN8_VECS_IRQ_SHIFT;
engine->irq_get = gen8_ring_get_irq;
engine->irq_put = gen8_ring_put_irq;
engine->dispatch_execbuffer = gen8_ring_dispatch_execbuffer;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
engine->semaphore.sync_to = gen8_ring_sync;
engine->semaphore.signal = gen8_xcs_signal;
GEN8_RING_SEMAPHORE_INIT(engine);
@@ -3094,7 +3086,7 @@ int intel_init_vebox_ring_buffer(struct drm_device *dev)
engine->irq_get = hsw_vebox_get_irq;
engine->irq_put = hsw_vebox_put_irq;
engine->dispatch_execbuffer = gen6_ring_dispatch_execbuffer;
- if (i915_semaphore_is_enabled(dev)) {
+ if (i915_semaphore_is_enabled(dev_priv)) {
engine->semaphore.sync_to = gen6_ring_sync;
engine->semaphore.signal = gen6_signal;
engine->semaphore.mbox.wait[RCS] = MI_SEMAPHORE_SYNC_VER;
generated by cgit v1.2.3 (git 2.39.1) at 2024-04-27 05:42:29 +0000