summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/i915
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/i915')
-rw-r--r--drivers/gpu/drm/i915/gvt/aperture_gm.c8
-rw-r--r--drivers/gpu/drm/i915/gvt/cfg_space.c54
-rw-r--r--drivers/gpu/drm/i915/gvt/cmd_parser.c119
-rw-r--r--drivers/gpu/drm/i915/gvt/debug.h8
-rw-r--r--drivers/gpu/drm/i915/gvt/display.c139
-rw-r--r--drivers/gpu/drm/i915/gvt/display.h20
-rw-r--r--drivers/gpu/drm/i915/gvt/edid.c16
-rw-r--r--drivers/gpu/drm/i915/gvt/execlist.c32
-rw-r--r--drivers/gpu/drm/i915/gvt/firmware.c11
-rw-r--r--drivers/gpu/drm/i915/gvt/gtt.c122
-rw-r--r--drivers/gpu/drm/i915/gvt/gvt.c2
-rw-r--r--drivers/gpu/drm/i915/gvt/gvt.h19
-rw-r--r--drivers/gpu/drm/i915/gvt/handlers.c484
-rw-r--r--drivers/gpu/drm/i915/gvt/kvmgt.c56
-rw-r--r--drivers/gpu/drm/i915/gvt/mmio.c104
-rw-r--r--drivers/gpu/drm/i915/gvt/mmio.h3
-rw-r--r--drivers/gpu/drm/i915/gvt/opregion.c15
-rw-r--r--drivers/gpu/drm/i915/gvt/render.c20
-rw-r--r--drivers/gpu/drm/i915/gvt/sched_policy.c4
-rw-r--r--drivers/gpu/drm/i915/gvt/scheduler.c141
-rw-r--r--drivers/gpu/drm/i915/gvt/vgpu.c115
-rw-r--r--drivers/gpu/drm/i915/i915_drv.c9
-rw-r--r--drivers/gpu/drm/i915/i915_drv.h7
-rw-r--r--drivers/gpu/drm/i915/i915_gem.c123
-rw-r--r--drivers/gpu/drm/i915/i915_gem_context.c1
-rw-r--r--drivers/gpu/drm/i915/i915_gem_context.h3
-rw-r--r--drivers/gpu/drm/i915/i915_gem_evict.c8
-rw-r--r--drivers/gpu/drm/i915/i915_gem_execbuffer.c56
-rw-r--r--drivers/gpu/drm/i915/i915_gem_gtt.c2
-rw-r--r--drivers/gpu/drm/i915/i915_gem_object.h3
-rw-r--r--drivers/gpu/drm/i915/i915_gem_request.c11
-rw-r--r--drivers/gpu/drm/i915/i915_gem_shrinker.c28
-rw-r--r--drivers/gpu/drm/i915/i915_irq.c83
-rw-r--r--drivers/gpu/drm/i915/i915_pci.c5
-rw-r--r--drivers/gpu/drm/i915/i915_perf.c11
-rw-r--r--drivers/gpu/drm/i915/i915_vma.c57
-rw-r--r--drivers/gpu/drm/i915/intel_csr.c1
-rw-r--r--drivers/gpu/drm/i915/intel_display.c67
-rw-r--r--drivers/gpu/drm/i915/intel_engine_cs.c2
-rw-r--r--drivers/gpu/drm/i915/intel_fbdev.c10
-rw-r--r--drivers/gpu/drm/i915/intel_gvt.c5
-rw-r--r--drivers/gpu/drm/i915/intel_hdmi.c24
-rw-r--r--drivers/gpu/drm/i915/intel_hotplug.c14
-rw-r--r--drivers/gpu/drm/i915/intel_lrc.c60
-rw-r--r--drivers/gpu/drm/i915/intel_pm.c23
-rw-r--r--drivers/gpu/drm/i915/intel_ringbuffer.c2
-rw-r--r--drivers/gpu/drm/i915/intel_ringbuffer.h11
-rw-r--r--drivers/gpu/drm/i915/intel_sprite.c3
-rw-r--r--drivers/gpu/drm/i915/intel_uncore.c13
49 files changed, 1416 insertions, 718 deletions
diff --git a/drivers/gpu/drm/i915/gvt/aperture_gm.c b/drivers/gpu/drm/i915/gvt/aperture_gm.c
index 3b6caaca9751..325618d969fe 100644
--- a/drivers/gpu/drm/i915/gvt/aperture_gm.c
+++ b/drivers/gpu/drm/i915/gvt/aperture_gm.c
@@ -242,7 +242,7 @@ static int alloc_resource(struct intel_vgpu *vgpu,
const char *item;
if (!param->low_gm_sz || !param->high_gm_sz || !param->fence_sz) {
- gvt_err("Invalid vGPU creation params\n");
+ gvt_vgpu_err("Invalid vGPU creation params\n");
return -EINVAL;
}
@@ -285,9 +285,9 @@ static int alloc_resource(struct intel_vgpu *vgpu,
return 0;
no_enough_resource:
- gvt_err("vgpu%d: fail to allocate resource %s\n", vgpu->id, item);
- gvt_err("vgpu%d: request %luMB avail %luMB max %luMB taken %luMB\n",
- vgpu->id, BYTES_TO_MB(request), BYTES_TO_MB(avail),
+ gvt_vgpu_err("fail to allocate resource %s\n", item);
+ gvt_vgpu_err("request %luMB avail %luMB max %luMB taken %luMB\n",
+ BYTES_TO_MB(request), BYTES_TO_MB(avail),
BYTES_TO_MB(max), BYTES_TO_MB(taken));
return -ENOSPC;
}
diff --git a/drivers/gpu/drm/i915/gvt/cfg_space.c b/drivers/gpu/drm/i915/gvt/cfg_space.c
index 4a6a2ed65732..40af17ec6312 100644
--- a/drivers/gpu/drm/i915/gvt/cfg_space.c
+++ b/drivers/gpu/drm/i915/gvt/cfg_space.c
@@ -41,6 +41,54 @@ enum {
INTEL_GVT_PCI_BAR_MAX,
};
+/* bitmap for writable bits (RW or RW1C bits, but cannot co-exist in one
+ * byte) byte by byte in standard pci configuration space. (not the full
+ * 256 bytes.)
+ */
+static const u8 pci_cfg_space_rw_bmp[PCI_INTERRUPT_LINE + 4] = {
+ [PCI_COMMAND] = 0xff, 0x07,
+ [PCI_STATUS] = 0x00, 0xf9, /* the only one RW1C byte */
+ [PCI_CACHE_LINE_SIZE] = 0xff,
+ [PCI_BASE_ADDRESS_0 ... PCI_CARDBUS_CIS - 1] = 0xff,
+ [PCI_ROM_ADDRESS] = 0x01, 0xf8, 0xff, 0xff,
+ [PCI_INTERRUPT_LINE] = 0xff,
+};
+
+/**
+ * vgpu_pci_cfg_mem_write - write virtual cfg space memory
+ *
+ * Use this function to write virtual cfg space memory.
+ * For standard cfg space, only RW bits can be changed,
+ * and we emulates the RW1C behavior of PCI_STATUS register.
+ */
+static void vgpu_pci_cfg_mem_write(struct intel_vgpu *vgpu, unsigned int off,
+ u8 *src, unsigned int bytes)
+{
+ u8 *cfg_base = vgpu_cfg_space(vgpu);
+ u8 mask, new, old;
+ int i = 0;
+
+ for (; i < bytes && (off + i < sizeof(pci_cfg_space_rw_bmp)); i++) {
+ mask = pci_cfg_space_rw_bmp[off + i];
+ old = cfg_base[off + i];
+ new = src[i] & mask;
+
+ /**
+ * The PCI_STATUS high byte has RW1C bits, here
+ * emulates clear by writing 1 for these bits.
+ * Writing a 0b to RW1C bits has no effect.
+ */
+ if (off + i == PCI_STATUS + 1)
+ new = (~new & old) & mask;
+
+ cfg_base[off + i] = (old & ~mask) | new;
+ }
+
+ /* For other configuration space directly copy as it is. */
+ if (i < bytes)
+ memcpy(cfg_base + off + i, src + i, bytes - i);
+}
+
/**
* intel_vgpu_emulate_cfg_read - emulate vGPU configuration space read
*
@@ -123,7 +171,7 @@ static int emulate_pci_command_write(struct intel_vgpu *vgpu,
u8 changed = old ^ new;
int ret;
- memcpy(vgpu_cfg_space(vgpu) + offset, p_data, bytes);
+ vgpu_pci_cfg_mem_write(vgpu, offset, p_data, bytes);
if (!(changed & PCI_COMMAND_MEMORY))
return 0;
@@ -274,10 +322,10 @@ int intel_vgpu_emulate_cfg_write(struct intel_vgpu *vgpu, unsigned int offset,
if (ret)
return ret;
- memcpy(vgpu_cfg_space(vgpu) + offset, p_data, bytes);
+ vgpu_pci_cfg_mem_write(vgpu, offset, p_data, bytes);
break;
default:
- memcpy(vgpu_cfg_space(vgpu) + offset, p_data, bytes);
+ vgpu_pci_cfg_mem_write(vgpu, offset, p_data, bytes);
break;
}
return 0;
diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c
index b9c8e2407682..2b92cc8a7d1a 100644
--- a/drivers/gpu/drm/i915/gvt/cmd_parser.c
+++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c
@@ -668,7 +668,7 @@ static inline void print_opcode(u32 cmd, int ring_id)
if (d_info == NULL)
return;
- gvt_err("opcode=0x%x %s sub_ops:",
+ gvt_dbg_cmd("opcode=0x%x %s sub_ops:",
cmd >> (32 - d_info->op_len), d_info->name);
for (i = 0; i < d_info->nr_sub_op; i++)
@@ -693,23 +693,23 @@ static void parser_exec_state_dump(struct parser_exec_state *s)
int cnt = 0;
int i;
- gvt_err(" vgpu%d RING%d: ring_start(%08lx) ring_end(%08lx)"
+ gvt_dbg_cmd(" vgpu%d RING%d: ring_start(%08lx) ring_end(%08lx)"
" ring_head(%08lx) ring_tail(%08lx)\n", s->vgpu->id,
s->ring_id, s->ring_start, s->ring_start + s->ring_size,
s->ring_head, s->ring_tail);
- gvt_err(" %s %s ip_gma(%08lx) ",
+ gvt_dbg_cmd(" %s %s ip_gma(%08lx) ",
s->buf_type == RING_BUFFER_INSTRUCTION ?
"RING_BUFFER" : "BATCH_BUFFER",
s->buf_addr_type == GTT_BUFFER ?
"GTT" : "PPGTT", s->ip_gma);
if (s->ip_va == NULL) {
- gvt_err(" ip_va(NULL)");
+ gvt_dbg_cmd(" ip_va(NULL)");
return;
}
- gvt_err(" ip_va=%p: %08x %08x %08x %08x\n",
+ gvt_dbg_cmd(" ip_va=%p: %08x %08x %08x %08x\n",
s->ip_va, cmd_val(s, 0), cmd_val(s, 1),
cmd_val(s, 2), cmd_val(s, 3));
@@ -817,6 +817,25 @@ static bool is_shadowed_mmio(unsigned int offset)
return ret;
}
+static inline bool is_force_nonpriv_mmio(unsigned int offset)
+{
+ return (offset >= 0x24d0 && offset < 0x2500);
+}
+
+static int force_nonpriv_reg_handler(struct parser_exec_state *s,
+ unsigned int offset, unsigned int index)
+{
+ struct intel_gvt *gvt = s->vgpu->gvt;
+ unsigned int data = cmd_val(s, index + 1);
+
+ if (!intel_gvt_in_force_nonpriv_whitelist(gvt, data)) {
+ gvt_err("Unexpected forcenonpriv 0x%x LRI write, value=0x%x\n",
+ offset, data);
+ return -EINVAL;
+ }
+ return 0;
+}
+
static int cmd_reg_handler(struct parser_exec_state *s,
unsigned int offset, unsigned int index, char *cmd)
{
@@ -824,23 +843,26 @@ static int cmd_reg_handler(struct parser_exec_state *s,
struct intel_gvt *gvt = vgpu->gvt;
if (offset + 4 > gvt->device_info.mmio_size) {
- gvt_err("%s access to (%x) outside of MMIO range\n",
+ gvt_vgpu_err("%s access to (%x) outside of MMIO range\n",
cmd, offset);
return -EINVAL;
}
if (!intel_gvt_mmio_is_cmd_access(gvt, offset)) {
- gvt_err("vgpu%d: %s access to non-render register (%x)\n",
- s->vgpu->id, cmd, offset);
+ gvt_vgpu_err("%s access to non-render register (%x)\n",
+ cmd, offset);
return 0;
}
if (is_shadowed_mmio(offset)) {
- gvt_err("vgpu%d: found access of shadowed MMIO %x\n",
- s->vgpu->id, offset);
+ gvt_vgpu_err("found access of shadowed MMIO %x\n", offset);
return 0;
}
+ if (is_force_nonpriv_mmio(offset) &&
+ force_nonpriv_reg_handler(s, offset, index))
+ return -EINVAL;
+
if (offset == i915_mmio_reg_offset(DERRMR) ||
offset == i915_mmio_reg_offset(FORCEWAKE_MT)) {
/* Writing to HW VGT_PVINFO_PAGE offset will be discarded */
@@ -1008,7 +1030,7 @@ static int cmd_handler_pipe_control(struct parser_exec_state *s)
ret = cmd_reg_handler(s, 0x2358, 1, "pipe_ctrl");
else if (post_sync == 1) {
/* check ggtt*/
- if ((cmd_val(s, 2) & (1 << 2))) {
+ if ((cmd_val(s, 1) & PIPE_CONTROL_GLOBAL_GTT_IVB)) {
gma = cmd_val(s, 2) & GENMASK(31, 3);
if (gmadr_bytes == 8)
gma |= (cmd_gma_hi(s, 3)) << 32;
@@ -1129,6 +1151,7 @@ static int skl_decode_mi_display_flip(struct parser_exec_state *s,
struct mi_display_flip_command_info *info)
{
struct drm_i915_private *dev_priv = s->vgpu->gvt->dev_priv;
+ struct intel_vgpu *vgpu = s->vgpu;
u32 dword0 = cmd_val(s, 0);
u32 dword1 = cmd_val(s, 1);
u32 dword2 = cmd_val(s, 2);
@@ -1167,7 +1190,7 @@ static int skl_decode_mi_display_flip(struct parser_exec_state *s,
break;
default:
- gvt_err("unknown plane code %d\n", plane);
+ gvt_vgpu_err("unknown plane code %d\n", plane);
return -EINVAL;
}
@@ -1274,25 +1297,26 @@ static int update_plane_mmio_from_mi_display_flip(
static int cmd_handler_mi_display_flip(struct parser_exec_state *s)
{
struct mi_display_flip_command_info info;
+ struct intel_vgpu *vgpu = s->vgpu;
int ret;
int i;
int len = cmd_length(s);
ret = decode_mi_display_flip(s, &info);
if (ret) {
- gvt_err("fail to decode MI display flip command\n");
+ gvt_vgpu_err("fail to decode MI display flip command\n");
return ret;
}
ret = check_mi_display_flip(s, &info);
if (ret) {
- gvt_err("invalid MI display flip command\n");
+ gvt_vgpu_err("invalid MI display flip command\n");
return ret;
}
ret = update_plane_mmio_from_mi_display_flip(s, &info);
if (ret) {
- gvt_err("fail to update plane mmio\n");
+ gvt_vgpu_err("fail to update plane mmio\n");
return ret;
}
@@ -1350,7 +1374,8 @@ static inline int cmd_address_audit(struct parser_exec_state *s,
int ret;
if (op_size > max_surface_size) {
- gvt_err("command address audit fail name %s\n", s->info->name);
+ gvt_vgpu_err("command address audit fail name %s\n",
+ s->info->name);
return -EINVAL;
}
@@ -1367,7 +1392,7 @@ static inline int cmd_address_audit(struct parser_exec_state *s,
}
return 0;
err:
- gvt_err("cmd_parser: Malicious %s detected, addr=0x%lx, len=%d!\n",
+ gvt_vgpu_err("cmd_parser: Malicious %s detected, addr=0x%lx, len=%d!\n",
s->info->name, guest_gma, op_size);
pr_err("cmd dump: ");
@@ -1412,8 +1437,10 @@ static int cmd_handler_mi_store_data_imm(struct parser_exec_state *s)
static inline int unexpected_cmd(struct parser_exec_state *s)
{
- gvt_err("vgpu%d: Unexpected %s in command buffer!\n",
- s->vgpu->id, s->info->name);
+ struct intel_vgpu *vgpu = s->vgpu;
+
+ gvt_vgpu_err("Unexpected %s in command buffer!\n", s->info->name);
+
return -EINVAL;
}
@@ -1516,7 +1543,7 @@ static int copy_gma_to_hva(struct intel_vgpu *vgpu, struct intel_vgpu_mm *mm,
while (gma != end_gma) {
gpa = intel_vgpu_gma_to_gpa(mm, gma);
if (gpa == INTEL_GVT_INVALID_ADDR) {
- gvt_err("invalid gma address: %lx\n", gma);
+ gvt_vgpu_err("invalid gma address: %lx\n", gma);
return -EFAULT;
}
@@ -1557,6 +1584,7 @@ static uint32_t find_bb_size(struct parser_exec_state *s)
uint32_t bb_size = 0;
uint32_t cmd_len = 0;
bool met_bb_end = false;
+ struct intel_vgpu *vgpu = s->vgpu;
u32 cmd;
/* get the start gm address of the batch buffer */
@@ -1565,7 +1593,7 @@ static uint32_t find_bb_size(struct parser_exec_state *s)
info = get_cmd_info(s->vgpu->gvt, cmd, s->ring_id);
if (info == NULL) {
- gvt_err("unknown cmd 0x%x, opcode=0x%x\n",
+ gvt_vgpu_err("unknown cmd 0x%x, opcode=0x%x\n",
cmd, get_opcode(cmd, s->ring_id));
return -EINVAL;
}
@@ -1574,7 +1602,7 @@ static uint32_t find_bb_size(struct parser_exec_state *s)
gma, gma + 4, &cmd);
info = get_cmd_info(s->vgpu->gvt, cmd, s->ring_id);
if (info == NULL) {
- gvt_err("unknown cmd 0x%x, opcode=0x%x\n",
+ gvt_vgpu_err("unknown cmd 0x%x, opcode=0x%x\n",
cmd, get_opcode(cmd, s->ring_id));
return -EINVAL;
}
@@ -1599,6 +1627,7 @@ static uint32_t find_bb_size(struct parser_exec_state *s)
static int perform_bb_shadow(struct parser_exec_state *s)
{
struct intel_shadow_bb_entry *entry_obj;
+ struct intel_vgpu *vgpu = s->vgpu;
unsigned long gma = 0;
uint32_t bb_size;
void *dst = NULL;
@@ -1633,7 +1662,7 @@ static int perform_bb_shadow(struct parser_exec_state *s)
ret = i915_gem_object_set_to_cpu_domain(entry_obj->obj, false);
if (ret) {
- gvt_err("failed to set shadow batch to CPU\n");
+ gvt_vgpu_err("failed to set shadow batch to CPU\n");
goto unmap_src;
}
@@ -1645,7 +1674,7 @@ static int perform_bb_shadow(struct parser_exec_state *s)
gma, gma + bb_size,
dst);
if (ret) {
- gvt_err("fail to copy guest ring buffer\n");
+ gvt_vgpu_err("fail to copy guest ring buffer\n");
goto unmap_src;
}
@@ -1676,15 +1705,16 @@ static int cmd_handler_mi_batch_buffer_start(struct parser_exec_state *s)
{
bool second_level;
int ret = 0;
+ struct intel_vgpu *vgpu = s->vgpu;
if (s->buf_type == BATCH_BUFFER_2ND_LEVEL) {
- gvt_err("Found MI_BATCH_BUFFER_START in 2nd level BB\n");
+ gvt_vgpu_err("Found MI_BATCH_BUFFER_START in 2nd level BB\n");
return -EINVAL;
}
second_level = BATCH_BUFFER_2ND_LEVEL_BIT(cmd_val(s, 0)) == 1;
if (second_level && (s->buf_type != BATCH_BUFFER_INSTRUCTION)) {
- gvt_err("Jumping to 2nd level BB from RB is not allowed\n");
+ gvt_vgpu_err("Jumping to 2nd level BB from RB is not allowed\n");
return -EINVAL;
}
@@ -1702,7 +1732,7 @@ static int cmd_handler_mi_batch_buffer_start(struct parser_exec_state *s)
if (batch_buffer_needs_scan(s)) {
ret = perform_bb_shadow(s);
if (ret < 0)
- gvt_err("invalid shadow batch buffer\n");
+ gvt_vgpu_err("invalid shadow batch buffer\n");
} else {
/* emulate a batch buffer end to do return right */
ret = cmd_handler_mi_batch_buffer_end(s);
@@ -2429,6 +2459,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
int ret = 0;
cycles_t t0, t1, t2;
struct parser_exec_state s_before_advance_custom;
+ struct intel_vgpu *vgpu = s->vgpu;
t0 = get_cycles();
@@ -2436,7 +2467,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
info = get_cmd_info(s->vgpu->gvt, cmd, s->ring_id);
if (info == NULL) {
- gvt_err("unknown cmd 0x%x, opcode=0x%x\n",
+ gvt_vgpu_err("unknown cmd 0x%x, opcode=0x%x\n",
cmd, get_opcode(cmd, s->ring_id));
return -EINVAL;
}
@@ -2452,7 +2483,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
if (info->handler) {
ret = info->handler(s);
if (ret < 0) {
- gvt_err("%s handler error\n", info->name);
+ gvt_vgpu_err("%s handler error\n", info->name);
return ret;
}
}
@@ -2463,7 +2494,7 @@ static int cmd_parser_exec(struct parser_exec_state *s)
if (!(info->flag & F_IP_ADVANCE_CUSTOM)) {
ret = cmd_advance_default(s);
if (ret) {
- gvt_err("%s IP advance error\n", info->name);
+ gvt_vgpu_err("%s IP advance error\n", info->name);
return ret;
}
}
@@ -2486,6 +2517,7 @@ static int command_scan(struct parser_exec_state *s,
unsigned long gma_head, gma_tail, gma_bottom;
int ret = 0;
+ struct intel_vgpu *vgpu = s->vgpu;
gma_head = rb_start + rb_head;
gma_tail = rb_start + rb_tail;
@@ -2497,7 +2529,7 @@ static int command_scan(struct parser_exec_state *s,
if (s->buf_type == RING_BUFFER_INSTRUCTION) {
if (!(s->ip_gma >= rb_start) ||
!(s->ip_gma < gma_bottom)) {
- gvt_err("ip_gma %lx out of ring scope."
+ gvt_vgpu_err("ip_gma %lx out of ring scope."
"(base:0x%lx, bottom: 0x%lx)\n",
s->ip_gma, rb_start,
gma_bottom);
@@ -2505,7 +2537,7 @@ static int command_scan(struct parser_exec_state *s,
return -EINVAL;
}
if (gma_out_of_range(s->ip_gma, gma_head, gma_tail)) {
- gvt_err("ip_gma %lx out of range."
+ gvt_vgpu_err("ip_gma %lx out of range."
"base 0x%lx head 0x%lx tail 0x%lx\n",
s->ip_gma, rb_start,
rb_head, rb_tail);
@@ -2515,7 +2547,7 @@ static int command_scan(struct parser_exec_state *s,
}
ret = cmd_parser_exec(s);
if (ret) {
- gvt_err("cmd parser error\n");
+ gvt_vgpu_err("cmd parser error\n");
parser_exec_state_dump(s);
break;
}
@@ -2639,7 +2671,7 @@ static int shadow_workload_ring_buffer(struct intel_vgpu_workload *workload)
gma_head, gma_top,
workload->shadow_ring_buffer_va);
if (ret) {
- gvt_err("fail to copy guest ring buffer\n");
+ gvt_vgpu_err("fail to copy guest ring buffer\n");
return ret;
}
copy_len = gma_top - gma_head;
@@ -2651,7 +2683,7 @@ static int shadow_workload_ring_buffer(struct intel_vgpu_workload *workload)
gma_head, gma_tail,
workload->shadow_ring_buffer_va + copy_len);
if (ret) {
- gvt_err("fail to copy guest ring buffer\n");
+ gvt_vgpu_err("fail to copy guest ring buffer\n");
return ret;
}
ring->tail += workload->rb_len;
@@ -2662,16 +2694,17 @@ static int shadow_workload_ring_buffer(struct intel_vgpu_workload *workload)
int intel_gvt_scan_and_shadow_workload(struct intel_vgpu_workload *workload)
{
int ret;
+ struct intel_vgpu *vgpu = workload->vgpu;
ret = shadow_workload_ring_buffer(workload);
if (ret) {
- gvt_err("fail to shadow workload ring_buffer\n");
+ gvt_vgpu_err("fail to shadow workload ring_buffer\n");
return ret;
}
ret = scan_workload(workload);
if (ret) {
- gvt_err("scan workload error\n");
+ gvt_vgpu_err("scan workload error\n");
return ret;
}
return 0;
@@ -2681,6 +2714,7 @@ static int shadow_indirect_ctx(struct intel_shadow_wa_ctx *wa_ctx)
{
int ctx_size = wa_ctx->indirect_ctx.size;
unsigned long guest_gma = wa_ctx->indirect_ctx.guest_gma;
+ struct intel_vgpu *vgpu = wa_ctx->workload->vgpu;
struct drm_i915_gem_object *obj;
int ret = 0;
void *map;
@@ -2694,14 +2728,14 @@ static int shadow_indirect_ctx(struct intel_shadow_wa_ctx *wa_ctx)
/* get the va of the shadow batch buffer */
map = i915_gem_object_pin_map(obj, I915_MAP_WB);
if (IS_ERR(map)) {
- gvt_err("failed to vmap shadow indirect ctx\n");
+ gvt_vgpu_err("failed to vmap shadow indirect ctx\n");
ret = PTR_ERR(map);
goto put_obj;
}
ret = i915_gem_object_set_to_cpu_domain(obj, false);
if (ret) {
- gvt_err("failed to set shadow indirect ctx to CPU\n");
+ gvt_vgpu_err("failed to set shadow indirect ctx to CPU\n");
goto unmap_src;
}
@@ -2710,7 +2744,7 @@ static int shadow_indirect_ctx(struct intel_shadow_wa_ctx *wa_ctx)
guest_gma, guest_gma + ctx_size,
map);
if (ret) {
- gvt_err("fail to copy guest indirect ctx\n");
+ gvt_vgpu_err("fail to copy guest indirect ctx\n");
goto unmap_src;
}
@@ -2744,13 +2778,14 @@ static int combine_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
int intel_gvt_scan_and_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
{
int ret;
+ struct intel_vgpu *vgpu = wa_ctx->workload->vgpu;
if (wa_ctx->indirect_ctx.size == 0)
return 0;
ret = shadow_indirect_ctx(wa_ctx);
if (ret) {
- gvt_err("fail to shadow indirect ctx\n");
+ gvt_vgpu_err("fail to shadow indirect ctx\n");
return ret;
}
@@ -2758,7 +2793,7 @@ int intel_gvt_scan_and_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
ret = scan_wa_ctx(wa_ctx);
if (ret) {
- gvt_err("scan wa ctx error\n");
+ gvt_vgpu_err("scan wa ctx error\n");
return ret;
}
diff --git a/drivers/gpu/drm/i915/gvt/debug.h b/drivers/gpu/drm/i915/gvt/debug.h
index 68cba7bd980a..b0cff4dc2684 100644
--- a/drivers/gpu/drm/i915/gvt/debug.h
+++ b/drivers/gpu/drm/i915/gvt/debug.h
@@ -27,6 +27,14 @@
#define gvt_err(fmt, args...) \
DRM_ERROR("gvt: "fmt, ##args)
+#define gvt_vgpu_err(fmt, args...) \
+do { \
+ if (IS_ERR_OR_NULL(vgpu)) \
+ DRM_DEBUG_DRIVER("gvt: "fmt, ##args); \
+ else \
+ DRM_DEBUG_DRIVER("gvt: vgpu %d: "fmt, vgpu->id, ##args);\
+} while (0)
+
#define gvt_dbg_core(fmt, args...) \
DRM_DEBUG_DRIVER("gvt: core: "fmt, ##args)
diff --git a/drivers/gpu/drm/i915/gvt/display.c b/drivers/gpu/drm/i915/gvt/display.c
index 6d8fde880c39..5419ae6ec633 100644
--- a/drivers/gpu/drm/i915/gvt/display.c
+++ b/drivers/gpu/drm/i915/gvt/display.c
@@ -83,44 +83,80 @@ static int pipe_is_enabled(struct intel_vgpu *vgpu, int pipe)
return 0;
}
+static unsigned char virtual_dp_monitor_edid[GVT_EDID_NUM][EDID_SIZE] = {
+ {
+/* EDID with 1024x768 as its resolution */
+ /*Header*/
+ 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00,
+ /* Vendor & Product Identification */
+ 0x22, 0xf0, 0x54, 0x29, 0x00, 0x00, 0x00, 0x00, 0x04, 0x17,
+ /* Version & Revision */
+ 0x01, 0x04,
+ /* Basic Display Parameters & Features */
+ 0xa5, 0x34, 0x20, 0x78, 0x23,
+ /* Color Characteristics */
+ 0xfc, 0x81, 0xa4, 0x55, 0x4d, 0x9d, 0x25, 0x12, 0x50, 0x54,
+ /* Established Timings: maximum resolution is 1024x768 */
+ 0x21, 0x08, 0x00,
+ /* Standard Timings. All invalid */
+ 0x00, 0xc0, 0x00, 0xc0, 0x00, 0x40, 0x00, 0x80, 0x00, 0x00,
+ 0x00, 0x40, 0x00, 0x00, 0x00, 0x01,
+ /* 18 Byte Data Blocks 1: invalid */
+ 0x00, 0x00, 0x80, 0xa0, 0x70, 0xb0,
+ 0x23, 0x40, 0x30, 0x20, 0x36, 0x00, 0x06, 0x44, 0x21, 0x00, 0x00, 0x1a,
+ /* 18 Byte Data Blocks 2: invalid */
+ 0x00, 0x00, 0x00, 0xfd, 0x00, 0x18, 0x3c, 0x18, 0x50, 0x11, 0x00, 0x0a,
+ 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
+ /* 18 Byte Data Blocks 3: invalid */
+ 0x00, 0x00, 0x00, 0xfc, 0x00, 0x48,
+ 0x50, 0x20, 0x5a, 0x52, 0x32, 0x34, 0x34, 0x30, 0x77, 0x0a, 0x20, 0x20,
+ /* 18 Byte Data Blocks 4: invalid */
+ 0x00, 0x00, 0x00, 0xff, 0x00, 0x43, 0x4e, 0x34, 0x33, 0x30, 0x34, 0x30,
+ 0x44, 0x58, 0x51, 0x0a, 0x20, 0x20,
+ /* Extension Block Count */
+ 0x00,
+ /* Checksum */
+ 0xef,
+ },
+ {
/* EDID with 1920x1200 as its resolution */
-static unsigned char virtual_dp_monitor_edid[] = {
- /*Header*/
- 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00,
- /* Vendor & Product Identification */
- 0x22, 0xf0, 0x54, 0x29, 0x00, 0x00, 0x00, 0x00, 0x04, 0x17,
- /* Version & Revision */
- 0x01, 0x04,
- /* Basic Display Parameters & Features */
- 0xa5, 0x34, 0x20, 0x78, 0x23,
- /* Color Characteristics */
- 0xfc, 0x81, 0xa4, 0x55, 0x4d, 0x9d, 0x25, 0x12, 0x50, 0x54,
- /* Established Timings: maximum resolution is 1024x768 */
- 0x21, 0x08, 0x00,
- /*
- * Standard Timings.
- * below new resolutions can be supported:
- * 1920x1080, 1280x720, 1280x960, 1280x1024,
- * 1440x900, 1600x1200, 1680x1050
- */
- 0xd1, 0xc0, 0x81, 0xc0, 0x81, 0x40, 0x81, 0x80, 0x95, 0x00,
- 0xa9, 0x40, 0xb3, 0x00, 0x01, 0x01,
- /* 18 Byte Data Blocks 1: max resolution is 1920x1200 */
- 0x28, 0x3c, 0x80, 0xa0, 0x70, 0xb0,
- 0x23, 0x40, 0x30, 0x20, 0x36, 0x00, 0x06, 0x44, 0x21, 0x00, 0x00, 0x1a,
- /* 18 Byte Data Blocks 2: invalid */
- 0x00, 0x00, 0x00, 0xfd, 0x00, 0x18, 0x3c, 0x18, 0x50, 0x11, 0x00, 0x0a,
- 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
- /* 18 Byte Data Blocks 3: invalid */
- 0x00, 0x00, 0x00, 0xfc, 0x00, 0x48,
- 0x50, 0x20, 0x5a, 0x52, 0x32, 0x34, 0x34, 0x30, 0x77, 0x0a, 0x20, 0x20,
- /* 18 Byte Data Blocks 4: invalid */
- 0x00, 0x00, 0x00, 0xff, 0x00, 0x43, 0x4e, 0x34, 0x33, 0x30, 0x34, 0x30,
- 0x44, 0x58, 0x51, 0x0a, 0x20, 0x20,
- /* Extension Block Count */
- 0x00,
- /* Checksum */
- 0x45,
+ /*Header*/
+ 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00,
+ /* Vendor & Product Identification */
+ 0x22, 0xf0, 0x54, 0x29, 0x00, 0x00, 0x00, 0x00, 0x04, 0x17,
+ /* Version & Revision */
+ 0x01, 0x04,
+ /* Basic Display Parameters & Features */
+ 0xa5, 0x34, 0x20, 0x78, 0x23,
+ /* Color Characteristics */
+ 0xfc, 0x81, 0xa4, 0x55, 0x4d, 0x9d, 0x25, 0x12, 0x50, 0x54,
+ /* Established Timings: maximum resolution is 1024x768 */
+ 0x21, 0x08, 0x00,
+ /*
+ * Standard Timings.
+ * below new resolutions can be supported:
+ * 1920x1080, 1280x720, 1280x960, 1280x1024,
+ * 1440x900, 1600x1200, 1680x1050
+ */
+ 0xd1, 0xc0, 0x81, 0xc0, 0x81, 0x40, 0x81, 0x80, 0x95, 0x00,
+ 0xa9, 0x40, 0xb3, 0x00, 0x01, 0x01,
+ /* 18 Byte Data Blocks 1: max resolution is 1920x1200 */
+ 0x28, 0x3c, 0x80, 0xa0, 0x70, 0xb0,
+ 0x23, 0x40, 0x30, 0x20, 0x36, 0x00, 0x06, 0x44, 0x21, 0x00, 0x00, 0x1a,
+ /* 18 Byte Data Blocks 2: invalid */
+ 0x00, 0x00, 0x00, 0xfd, 0x00, 0x18, 0x3c, 0x18, 0x50, 0x11, 0x00, 0x0a,
+ 0x20, 0x20, 0x20, 0x20, 0x20, 0x20,
+ /* 18 Byte Data Blocks 3: invalid */
+ 0x00, 0x00, 0x00, 0xfc, 0x00, 0x48,
+ 0x50, 0x20, 0x5a, 0x52, 0x32, 0x34, 0x34, 0x30, 0x77, 0x0a, 0x20, 0x20,
+ /* 18 Byte Data Blocks 4: invalid */
+ 0x00, 0x00, 0x00, 0xff, 0x00, 0x43, 0x4e, 0x34, 0x33, 0x30, 0x34, 0x30,
+ 0x44, 0x58, 0x51, 0x0a, 0x20, 0x20,
+ /* Extension Block Count */
+ 0x00,
+ /* Checksum */
+ 0x45,
+ },
};
#define DPCD_HEADER_SIZE 0xb
@@ -140,14 +176,20 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu)
vgpu_vreg(vgpu, SDEISR) &= ~(SDE_PORTA_HOTPLUG_SPT |
SDE_PORTE_HOTPLUG_SPT);
- if (intel_vgpu_has_monitor_on_port(vgpu, PORT_B))
+ if (intel_vgpu_has_monitor_on_port(vgpu, PORT_B)) {
vgpu_vreg(vgpu, SDEISR) |= SDE_PORTB_HOTPLUG_CPT;
+ vgpu_vreg(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDIB_DETECTED;
+ }
- if (intel_vgpu_has_monitor_on_port(vgpu, PORT_C))
+ if (intel_vgpu_has_monitor_on_port(vgpu, PORT_C)) {
vgpu_vreg(vgpu, SDEISR) |= SDE_PORTC_HOTPLUG_CPT;
+ vgpu_vreg(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDIC_DETECTED;
+ }
- if (intel_vgpu_has_monitor_on_port(vgpu, PORT_D))
+ if (intel_vgpu_has_monitor_on_port(vgpu, PORT_D)) {
vgpu_vreg(vgpu, SDEISR) |= SDE_PORTD_HOTPLUG_CPT;
+ vgpu_vreg(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDID_DETECTED;
+ }
if (IS_SKYLAKE(dev_priv) &&
intel_vgpu_has_monitor_on_port(vgpu, PORT_E)) {
@@ -160,6 +202,8 @@ static void emulate_monitor_status_change(struct intel_vgpu *vgpu)
GEN8_PORT_DP_A_HOTPLUG;
else
vgpu_vreg(vgpu, SDEISR) |= SDE_PORTA_HOTPLUG_SPT;
+
+ vgpu_vreg(vgpu, DDI_BUF_CTL(PORT_A)) |= DDI_INIT_DISPLAY_DETECTED;
}
}
@@ -175,10 +219,13 @@ static void clean_virtual_dp_monitor(struct intel_vgpu *vgpu, int port_num)
}
static int setup_virtual_dp_monitor(struct intel_vgpu *vgpu, int port_num,
- int type)
+ int type, unsigned int resolution)
{
struct intel_vgpu_port *port = intel_vgpu_port(vgpu, port_num);
+ if (WARN_ON(resolution >= GVT_EDID_NUM))
+ return -EINVAL;
+
port->edid = kzalloc(sizeof(*(port->edid)), GFP_KERNEL);
if (!port->edid)
return -ENOMEM;
@@ -189,7 +236,7 @@ static int setup_virtual_dp_monitor(struct intel_vgpu *vgpu, int port_num,
return -ENOMEM;
}
- memcpy(port->edid->edid_block, virtual_dp_monitor_edid,
+ memcpy(port->edid->edid_block, virtual_dp_monitor_edid[resolution],
EDID_SIZE);
port->edid->data_valid = true;
@@ -322,16 +369,18 @@ void intel_vgpu_clean_display(struct intel_vgpu *vgpu)
* Zero on success, negative error code if failed.
*
*/
-int intel_vgpu_init_display(struct intel_vgpu *vgpu)
+int intel_vgpu_init_display(struct intel_vgpu *vgpu, u64 resolution)
{
struct drm_i915_private *dev_priv = vgpu->gvt->dev_priv;
intel_vgpu_init_i2c_edid(vgpu);
if (IS_SKYLAKE(dev_priv))
- return setup_virtual_dp_monitor(vgpu, PORT_D, GVT_DP_D);
+ return setup_virtual_dp_monitor(vgpu, PORT_D, GVT_DP_D,
+ resolution);
else
- return setup_virtual_dp_monitor(vgpu, PORT_B, GVT_DP_B);
+ return setup_virtual_dp_monitor(vgpu, PORT_B, GVT_DP_B,
+ resolution);
}
/**
diff --git a/drivers/gpu/drm/i915/gvt/display.h b/drivers/gpu/drm/i915/gvt/display.h
index 8b234ea961f6..d73de22102e2 100644
--- a/drivers/gpu/drm/i915/gvt/display.h
+++ b/drivers/gpu/drm/i915/gvt/display.h
@@ -154,10 +154,28 @@ struct intel_vgpu_port {
int type;
};
+enum intel_vgpu_edid {
+ GVT_EDID_1024_768,
+ GVT_EDID_1920_1200,
+ GVT_EDID_NUM,
+};
+
+static inline char *vgpu_edid_str(enum intel_vgpu_edid id)
+{
+ switch (id) {
+ case GVT_EDID_1024_768:
+ return "1024x768";
+ case GVT_EDID_1920_1200:
+ return "1920x1200";
+ default:
+ return "";
+ }
+}
+
void intel_gvt_emulate_vblank(struct intel_gvt *gvt);
void intel_gvt_check_vblank_emulation(struct intel_gvt *gvt);
-int intel_vgpu_init_display(struct intel_vgpu *vgpu);
+int intel_vgpu_init_display(struct intel_vgpu *vgpu, u64 resolution);
void intel_vgpu_reset_display(struct intel_vgpu *vgpu);
void intel_vgpu_clean_display(struct intel_vgpu *vgpu);
diff --git a/drivers/gpu/drm/i915/gvt/edid.c b/drivers/gpu/drm/i915/gvt/edid.c
index bda85dff7b2a..42cd09ec63fa 100644
--- a/drivers/gpu/drm/i915/gvt/edid.c
+++ b/drivers/gpu/drm/i915/gvt/edid.c
@@ -52,16 +52,16 @@ static unsigned char edid_get_byte(struct intel_vgpu *vgpu)
unsigned char chr = 0;
if (edid->state == I2C_NOT_SPECIFIED || !edid->slave_selected) {
- gvt_err("Driver tries to read EDID without proper sequence!\n");
+ gvt_vgpu_err("Driver tries to read EDID without proper sequence!\n");
return 0;
}
if (edid->current_edid_read >= EDID_SIZE) {
- gvt_err("edid_get_byte() exceeds the size of EDID!\n");
+ gvt_vgpu_err("edid_get_byte() exceeds the size of EDID!\n");
return 0;
}
if (!edid->edid_available) {
- gvt_err("Reading EDID but EDID is not available!\n");
+ gvt_vgpu_err("Reading EDID but EDID is not available!\n");
return 0;
}
@@ -72,7 +72,7 @@ static unsigned char edid_get_byte(struct intel_vgpu *vgpu)
chr = edid_data->edid_block[edid->current_edid_read];
edid->current_edid_read++;
} else {
- gvt_err("No EDID available during the reading?\n");
+ gvt_vgpu_err("No EDID available during the reading?\n");
}
return chr;
}
@@ -223,7 +223,7 @@ static int gmbus1_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
vgpu_vreg(vgpu, PCH_GMBUS2) |= GMBUS_ACTIVE;
break;
default:
- gvt_err("Unknown/reserved GMBUS cycle detected!\n");
+ gvt_vgpu_err("Unknown/reserved GMBUS cycle detected!\n");
break;
}
/*
@@ -292,8 +292,7 @@ static int gmbus3_mmio_read(struct intel_vgpu *vgpu, unsigned int offset,
*/
} else {
memcpy(p_data, &vgpu_vreg(vgpu, offset), bytes);
- gvt_err("vgpu%d: warning: gmbus3 read with nothing returned\n",
- vgpu->id);
+ gvt_vgpu_err("warning: gmbus3 read with nothing returned\n");
}
return 0;
}
@@ -496,7 +495,8 @@ void intel_gvt_i2c_handle_aux_ch_write(struct intel_vgpu *vgpu,
unsigned char val = edid_get_byte(vgpu);
aux_data_for_write = (val << 16);
- }
+ } else
+ aux_data_for_write = (0xff << 16);
}
/* write the return value in AUX_CH_DATA reg which includes:
* ACK of I2C_WRITE
diff --git a/drivers/gpu/drm/i915/gvt/execlist.c b/drivers/gpu/drm/i915/gvt/execlist.c
index 46eb9fd3c03f..d186c157f65f 100644
--- a/drivers/gpu/drm/i915/gvt/execlist.c
+++ b/drivers/gpu/drm/i915/gvt/execlist.c
@@ -172,6 +172,7 @@ static int emulate_execlist_ctx_schedule_out(
struct intel_vgpu_execlist *execlist,
struct execlist_ctx_descriptor_format *ctx)
{
+ struct intel_vgpu *vgpu = execlist->vgpu;
struct intel_vgpu_execlist_slot *running = execlist->running_slot;
struct intel_vgpu_execlist_slot *pending = execlist->pending_slot;
struct execlist_ctx_descriptor_format *ctx0 = &running->ctx[0];
@@ -183,7 +184,7 @@ static int emulate_execlist_ctx_schedule_out(
gvt_dbg_el("schedule out context id %x\n", ctx->context_id);
if (WARN_ON(!same_context(ctx, execlist->running_context))) {
- gvt_err("schedule out context is not running context,"
+ gvt_vgpu_err("schedule out context is not running context,"
"ctx id %x running ctx id %x\n",
ctx->context_id,
execlist->running_context->context_id);
@@ -254,7 +255,7 @@ static struct intel_vgpu_execlist_slot *get_next_execlist_slot(
status.udw = vgpu_vreg(vgpu, status_reg + 4);
if (status.execlist_queue_full) {
- gvt_err("virtual execlist slots are full\n");
+ gvt_vgpu_err("virtual execlist slots are full\n");
return NULL;
}
@@ -270,11 +271,12 @@ static int emulate_execlist_schedule_in(struct intel_vgpu_execlist *execlist,
struct execlist_ctx_descriptor_format *ctx0, *ctx1;
struct execlist_context_status_format status;
+ struct intel_vgpu *vgpu = execlist->vgpu;
gvt_dbg_el("emulate schedule-in\n");
if (!slot) {
- gvt_err("no available execlist slot\n");
+ gvt_vgpu_err("no available execlist slot\n");
return -EINVAL;
}
@@ -375,7 +377,6 @@ static void prepare_shadow_batch_buffer(struct intel_vgpu_workload *workload)
vma = i915_gem_object_ggtt_pin(entry_obj->obj, NULL, 0, 4, 0);
if (IS_ERR(vma)) {
- gvt_err("Cannot pin\n");
return;
}
@@ -428,7 +429,6 @@ static void prepare_shadow_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx)
vma = i915_gem_object_ggtt_pin(wa_ctx->indirect_ctx.obj, NULL,
0, CACHELINE_BYTES, 0);
if (IS_ERR(vma)) {
- gvt_err("Cannot pin indirect ctx obj\n");
return;
}
@@ -561,6 +561,7 @@ static int prepare_mm(struct intel_vgpu_workload *workload)
{
struct execlist_ctx_descriptor_format *desc = &workload->ctx_desc;
struct intel_vgpu_mm *mm;
+ struct intel_vgpu *vgpu = workload->vgpu;
int page_table_level;
u32 pdp[8];
@@ -569,7 +570,7 @@ static int prepare_mm(struct intel_vgpu_workload *workload)
} else if (desc->addressing_mode == 3) { /* legacy 64 bit */
page_table_level = 4;
} else {
- gvt_err("Advanced Context mode(SVM) is not supported!\n");
+ gvt_vgpu_err("Advanced Context mode(SVM) is not supported!\n");
return -EINVAL;
}
@@ -583,7 +584,7 @@ static int prepare_mm(struct intel_vgpu_workload *workload)
mm = intel_vgpu_create_mm(workload->vgpu, INTEL_GVT_MM_PPGTT,
pdp, page_table_level, 0);
if (IS_ERR(mm)) {
- gvt_err("fail to create mm object.\n");
+ gvt_vgpu_err("fail to create mm object.\n");
return PTR_ERR(mm);
}
}
@@ -609,7 +610,7 @@ static int submit_context(struct intel_vgpu *vgpu, int ring_id,
ring_context_gpa = intel_vgpu_gma_to_gpa(vgpu->gtt.ggtt_mm,
(u32)((desc->lrca + 1) << GTT_PAGE_SHIFT));
if (ring_context_gpa == INTEL_GVT_INVALID_ADDR) {
- gvt_err("invalid guest context LRCA: %x\n", desc->lrca);
+ gvt_vgpu_err("invalid guest context LRCA: %x\n", desc->lrca);
return -EINVAL;
}
@@ -724,8 +725,7 @@ int intel_vgpu_submit_execlist(struct intel_vgpu *vgpu, int ring_id)
continue;
if (!desc[i]->privilege_access) {
- gvt_err("vgpu%d: unexpected GGTT elsp submission\n",
- vgpu->id);
+ gvt_vgpu_err("unexpected GGTT elsp submission\n");
return -EINVAL;
}
@@ -735,15 +735,13 @@ int intel_vgpu_submit_execlist(struct intel_vgpu *vgpu, int ring_id)
}
if (!valid_desc_bitmap) {
- gvt_err("vgpu%d: no valid desc in a elsp submission\n",
- vgpu->id);
+ gvt_vgpu_err("no valid desc in a elsp submission\n");
return -EINVAL;
}
if (!test_bit(0, (void *)&valid_desc_bitmap) &&
test_bit(1, (void *)&valid_desc_bitmap)) {
- gvt_err("vgpu%d: weird elsp submission, desc 0 is not valid\n",
- vgpu->id);
+ gvt_vgpu_err("weird elsp submission, desc 0 is not valid\n");
return -EINVAL;
}
@@ -752,8 +750,7 @@ int intel_vgpu_submit_execlist(struct intel_vgpu *vgpu, int ring_id)
ret = submit_context(vgpu, ring_id, &valid_desc[i],
emulate_schedule_in);
if (ret) {
- gvt_err("vgpu%d: fail to schedule workload\n",
- vgpu->id);
+ gvt_vgpu_err("fail to schedule workload\n");
return ret;
}
emulate_schedule_in = false;
@@ -778,7 +775,8 @@ static void init_vgpu_execlist(struct intel_vgpu *vgpu, int ring_id)
_EL_OFFSET_STATUS_PTR);
ctx_status_ptr.dw = vgpu_vreg(vgpu, ctx_status_ptr_reg);
- ctx_status_ptr.read_ptr = ctx_status_ptr.write_ptr = 0x7;
+ ctx_status_ptr.read_ptr = 0;
+ ctx_status_ptr.write_ptr = 0x7;
vgpu_vreg(vgpu, ctx_status_ptr_reg) = ctx_status_ptr.dw;
}
diff --git a/drivers/gpu/drm/i915/gvt/firmware.c b/drivers/gpu/drm/i915/gvt/firmware.c
index 1cb29b2d7dc6..dce8d15f706f 100644
--- a/drivers/gpu/drm/i915/gvt/firmware.c
+++ b/drivers/gpu/drm/i915/gvt/firmware.c
@@ -75,12 +75,12 @@ static int expose_firmware_sysfs(struct intel_gvt *gvt)
struct gvt_firmware_header *h;
void *firmware;
void *p;
- unsigned long size;
+ unsigned long size, crc32_start;
int i;
int ret;
- size = sizeof(*h) + info->mmio_size + info->cfg_space_size - 1;
- firmware = vmalloc(size);
+ size = sizeof(*h) + info->mmio_size + info->cfg_space_size;
+ firmware = vzalloc(size);
if (!firmware)
return -ENOMEM;
@@ -112,6 +112,9 @@ static int expose_firmware_sysfs(struct intel_gvt *gvt)
memcpy(gvt->firmware.mmio, p, info->mmio_size);
+ crc32_start = offsetof(struct gvt_firmware_header, crc32) + 4;
+ h->crc32 = crc32_le(0, firmware + crc32_start, size - crc32_start);
+
firmware_attr.size = size;
firmware_attr.private = firmware;
@@ -234,7 +237,7 @@ int intel_gvt_load_firmware(struct intel_gvt *gvt)
firmware->mmio = mem;
- sprintf(path, "%s/vid_0x%04x_did_0x%04x_rid_0x%04x.golden_hw_state",
+ sprintf(path, "%s/vid_0x%04x_did_0x%04x_rid_0x%02x.golden_hw_state",
GVT_FIRMWARE_PATH, pdev->vendor, pdev->device,
pdev->revision);
diff --git a/drivers/gpu/drm/i915/gvt/gtt.c b/drivers/gpu/drm/i915/gvt/gtt.c
index 28c92346db0e..b832bea64e03 100644
--- a/drivers/gpu/drm/i915/gvt/gtt.c
+++ b/drivers/gpu/drm/i915/gvt/gtt.c
@@ -49,8 +49,8 @@ bool intel_gvt_ggtt_validate_range(struct intel_vgpu *vgpu, u64 addr, u32 size)
{
if ((!vgpu_gmadr_is_valid(vgpu, addr)) || (size
&& !vgpu_gmadr_is_valid(vgpu, addr + size - 1))) {
- gvt_err("vgpu%d: invalid range gmadr 0x%llx size 0x%x\n",
- vgpu->id, addr, size);
+ gvt_vgpu_err("invalid range gmadr 0x%llx size 0x%x\n",
+ addr, size);
return false;
}
return true;
@@ -430,7 +430,7 @@ static int gtt_entry_p2m(struct intel_vgpu *vgpu, struct intel_gvt_gtt_entry *p,
mfn = intel_gvt_hypervisor_gfn_to_mfn(vgpu, gfn);
if (mfn == INTEL_GVT_INVALID_ADDR) {
- gvt_err("fail to translate gfn: 0x%lx\n", gfn);
+ gvt_vgpu_err("fail to translate gfn: 0x%lx\n", gfn);
return -ENXIO;
}
@@ -611,7 +611,7 @@ static inline int init_shadow_page(struct intel_vgpu *vgpu,
daddr = dma_map_page(kdev, p->page, 0, 4096, PCI_DMA_BIDIRECTIONAL);
if (dma_mapping_error(kdev, daddr)) {
- gvt_err("fail to map dma addr\n");
+ gvt_vgpu_err("fail to map dma addr\n");
return -EINVAL;
}
@@ -735,7 +735,7 @@ retry:
if (reclaim_one_mm(vgpu->gvt))
goto retry;
- gvt_err("fail to allocate ppgtt shadow page\n");
+ gvt_vgpu_err("fail to allocate ppgtt shadow page\n");
return ERR_PTR(-ENOMEM);
}
@@ -750,14 +750,14 @@ retry:
*/
ret = init_shadow_page(vgpu, &spt->shadow_page, type);
if (ret) {
- gvt_err("fail to initialize shadow page for spt\n");
+ gvt_vgpu_err("fail to initialize shadow page for spt\n");
goto err;
}
ret = intel_vgpu_init_guest_page(vgpu, &spt->guest_page,
gfn, ppgtt_write_protection_handler, NULL);
if (ret) {
- gvt_err("fail to initialize guest page for spt\n");
+ gvt_vgpu_err("fail to initialize guest page for spt\n");
goto err;
}
@@ -776,8 +776,7 @@ static struct intel_vgpu_ppgtt_spt *ppgtt_find_shadow_page(
if (p)
return shadow_page_to_ppgtt_spt(p);
- gvt_err("vgpu%d: fail to find ppgtt shadow page: 0x%lx\n",
- vgpu->id, mfn);
+ gvt_vgpu_err("fail to find ppgtt shadow page: 0x%lx\n", mfn);
return NULL;
}
@@ -827,8 +826,8 @@ static int ppgtt_invalidate_shadow_page_by_shadow_entry(struct intel_vgpu *vgpu,
}
s = ppgtt_find_shadow_page(vgpu, ops->get_pfn(e));
if (!s) {
- gvt_err("vgpu%d: fail to find shadow page: mfn: 0x%lx\n",
- vgpu->id, ops->get_pfn(e));
+ gvt_vgpu_err("fail to find shadow page: mfn: 0x%lx\n",
+ ops->get_pfn(e));
return -ENXIO;
}
return ppgtt_invalidate_shadow_page(s);
@@ -836,6 +835,7 @@ static int ppgtt_invalidate_shadow_page_by_shadow_entry(struct intel_vgpu *vgpu,
static int ppgtt_invalidate_shadow_page(struct intel_vgpu_ppgtt_spt *spt)
{
+ struct intel_vgpu *vgpu = spt->vgpu;
struct intel_gvt_gtt_entry e;
unsigned long index;
int ret;
@@ -854,7 +854,7 @@ static int ppgtt_invalidate_shadow_page(struct intel_vgpu_ppgtt_spt *spt)
for_each_present_shadow_entry(spt, &e, index) {
if (!gtt_type_is_pt(get_next_pt_type(e.type))) {
- gvt_err("GVT doesn't support pse bit for now\n");
+ gvt_vgpu_err("GVT doesn't support pse bit for now\n");
return -EINVAL;
}
ret = ppgtt_invalidate_shadow_page_by_shadow_entry(
@@ -868,8 +868,8 @@ release:
ppgtt_free_shadow_page(spt);
return 0;
fail:
- gvt_err("vgpu%d: fail: shadow page %p shadow entry 0x%llx type %d\n",
- spt->vgpu->id, spt, e.val64, e.type);
+ gvt_vgpu_err("fail: shadow page %p shadow entry 0x%llx type %d\n",
+ spt, e.val64, e.type);
return ret;
}
@@ -914,8 +914,8 @@ static struct intel_vgpu_ppgtt_spt *ppgtt_populate_shadow_page_by_guest_entry(
}
return s;
fail:
- gvt_err("vgpu%d: fail: shadow page %p guest entry 0x%llx type %d\n",
- vgpu->id, s, we->val64, we->type);
+ gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d\n",
+ s, we->val64, we->type);
return ERR_PTR(ret);
}
@@ -953,7 +953,7 @@ static int ppgtt_populate_shadow_page(struct intel_vgpu_ppgtt_spt *spt)
for_each_present_guest_entry(spt, &ge, i) {
if (!gtt_type_is_pt(get_next_pt_type(ge.type))) {
- gvt_err("GVT doesn't support pse bit now\n");
+ gvt_vgpu_err("GVT doesn't support pse bit now\n");
ret = -EINVAL;
goto fail;
}
@@ -969,8 +969,8 @@ static int ppgtt_populate_shadow_page(struct intel_vgpu_ppgtt_spt *spt)
}
return 0;
fail:
- gvt_err("vgpu%d: fail: shadow page %p guest entry 0x%llx type %d\n",
- vgpu->id, spt, ge.val64, ge.type);
+ gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d\n",
+ spt, ge.val64, ge.type);
return ret;
}
@@ -999,7 +999,7 @@ static int ppgtt_handle_guest_entry_removal(struct intel_vgpu_guest_page *gpt,
struct intel_vgpu_ppgtt_spt *s =
ppgtt_find_shadow_page(vgpu, ops->get_pfn(&e));
if (!s) {
- gvt_err("fail to find guest page\n");
+ gvt_vgpu_err("fail to find guest page\n");
ret = -ENXIO;
goto fail;
}
@@ -1011,8 +1011,8 @@ static int ppgtt_handle_guest_entry_removal(struct intel_vgpu_guest_page *gpt,
ppgtt_set_shadow_entry(spt, &e, index);
return 0;
fail:
- gvt_err("vgpu%d: fail: shadow page %p guest entry 0x%llx type %d\n",
- vgpu->id, spt, e.val64, e.type);
+ gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d\n",
+ spt, e.val64, e.type);
return ret;
}
@@ -1046,8 +1046,8 @@ static int ppgtt_handle_guest_entry_add(struct intel_vgpu_guest_page *gpt,
}
return 0;
fail:
- gvt_err("vgpu%d: fail: spt %p guest entry 0x%llx type %d\n", vgpu->id,
- spt, we->val64, we->type);
+ gvt_vgpu_err("fail: spt %p guest entry 0x%llx type %d\n",
+ spt, we->val64, we->type);
return ret;
}
@@ -1250,8 +1250,8 @@ static int ppgtt_handle_guest_write_page_table(
}
return 0;
fail:
- gvt_err("vgpu%d: fail: shadow page %p guest entry 0x%llx type %d.\n",
- vgpu->id, spt, we->val64, we->type);
+ gvt_vgpu_err("fail: shadow page %p guest entry 0x%llx type %d.\n",
+ spt, we->val64, we->type);
return ret;
}
@@ -1493,7 +1493,7 @@ static int shadow_mm(struct intel_vgpu_mm *mm)
spt = ppgtt_populate_shadow_page_by_guest_entry(vgpu, &ge);
if (IS_ERR(spt)) {
- gvt_err("fail to populate guest root pointer\n");
+ gvt_vgpu_err("fail to populate guest root pointer\n");
ret = PTR_ERR(spt);
goto fail;
}
@@ -1566,7 +1566,7 @@ struct intel_vgpu_mm *intel_vgpu_create_mm(struct intel_vgpu *vgpu,
ret = gtt->mm_alloc_page_table(mm);
if (ret) {
- gvt_err("fail to allocate page table for mm\n");
+ gvt_vgpu_err("fail to allocate page table for mm\n");
goto fail;
}
@@ -1584,7 +1584,7 @@ struct intel_vgpu_mm *intel_vgpu_create_mm(struct intel_vgpu *vgpu,
}
return mm;
fail:
- gvt_err("fail to create mm\n");
+ gvt_vgpu_err("fail to create mm\n");
if (mm)
intel_gvt_mm_unreference(mm);
return ERR_PTR(ret);
@@ -1760,7 +1760,7 @@ unsigned long intel_vgpu_gma_to_gpa(struct intel_vgpu_mm *mm, unsigned long gma)
mm->page_table_level, gma, gpa);
return gpa;
err:
- gvt_err("invalid mm type: %d gma %lx\n", mm->type, gma);
+ gvt_vgpu_err("invalid mm type: %d gma %lx\n", mm->type, gma);
return INTEL_GVT_INVALID_ADDR;
}
@@ -1825,11 +1825,8 @@ static int emulate_gtt_mmio_write(struct intel_vgpu *vgpu, unsigned int off,
gma = g_gtt_index << GTT_PAGE_SHIFT;
/* the VM may configure the whole GM space when ballooning is used */
- if (WARN_ONCE(!vgpu_gmadr_is_valid(vgpu, gma),
- "vgpu%d: found oob ggtt write, offset %x\n",
- vgpu->id, off)) {
+ if (!vgpu_gmadr_is_valid(vgpu, gma))
return 0;
- }
ggtt_get_guest_entry(ggtt_mm, &e, g_gtt_index);
@@ -1839,13 +1836,16 @@ static int emulate_gtt_mmio_write(struct intel_vgpu *vgpu, unsigned int off,
if (ops->test_present(&e)) {
ret = gtt_entry_p2m(vgpu, &e, &m);
if (ret) {
- gvt_err("vgpu%d: fail to translate guest gtt entry\n",
- vgpu->id);
- return ret;
+ gvt_vgpu_err("fail to translate guest gtt entry\n");
+ /* guest driver may read/write the entry when partial
+ * update the entry in this situation p2m will fail
+ * settting the shadow entry to point to a scratch page
+ */
+ ops->set_pfn(&m, gvt->gtt.scratch_ggtt_mfn);
}
} else {
m = e;
- m.val64 = 0;
+ ops->set_pfn(&m, gvt->gtt.scratch_ggtt_mfn);
}
ggtt_set_shadow_entry(ggtt_mm, &m, g_gtt_index);
@@ -1896,14 +1896,14 @@ static int alloc_scratch_pages(struct intel_vgpu *vgpu,
scratch_pt = (void *)get_zeroed_page(GFP_KERNEL);
if (!scratch_pt) {
- gvt_err("fail to allocate scratch page\n");
+ gvt_vgpu_err("fail to allocate scratch page\n");
return -ENOMEM;
}
daddr = dma_map_page(dev, virt_to_page(scratch_pt), 0,
4096, PCI_DMA_BIDIRECTIONAL);
if (dma_mapping_error(dev, daddr)) {
- gvt_err("fail to dmamap scratch_pt\n");
+ gvt_vgpu_err("fail to dmamap scratch_pt\n");
__free_page(virt_to_page(scratch_pt));
return -ENOMEM;
}
@@ -2006,7 +2006,7 @@ int intel_vgpu_init_gtt(struct intel_vgpu *vgpu)
ggtt_mm = intel_vgpu_create_mm(vgpu, INTEL_GVT_MM_GGTT,
NULL, 1, 0);
if (IS_ERR(ggtt_mm)) {
- gvt_err("fail to create mm for ggtt.\n");
+ gvt_vgpu_err("fail to create mm for ggtt.\n");
return PTR_ERR(ggtt_mm);
}
@@ -2015,6 +2015,22 @@ int intel_vgpu_init_gtt(struct intel_vgpu *vgpu)
return create_scratch_page_tree(vgpu);
}
+static void intel_vgpu_free_mm(struct intel_vgpu *vgpu, int type)
+{
+ struct list_head *pos, *n;
+ struct intel_vgpu_mm *mm;
+
+ list_for_each_safe(pos, n, &vgpu->gtt.mm_list_head) {
+ mm = container_of(pos, struct intel_vgpu_mm, list);
+ if (mm->type == type) {
+ vgpu->gvt->gtt.mm_free_page_table(mm);
+ list_del(&mm->list);
+ list_del(&mm->lru_list);
+ kfree(mm);
+ }
+ }
+}
+
/**
* intel_vgpu_clean_gtt - clean up per-vGPU graphics memory virulization
* @vgpu: a vGPU
@@ -2027,19 +2043,11 @@ int intel_vgpu_init_gtt(struct intel_vgpu *vgpu)
*/
void intel_vgpu_clean_gtt(struct intel_vgpu *vgpu)
{
- struct list_head *pos, *n;
- struct intel_vgpu_mm *mm;
-
ppgtt_free_all_shadow_page(vgpu);
release_scratch_page_tree(vgpu);
- list_for_each_safe(pos, n, &vgpu->gtt.mm_list_head) {
- mm = container_of(pos, struct intel_vgpu_mm, list);
- vgpu->gvt->gtt.mm_free_page_table(mm);
- list_del(&mm->list);
- list_del(&mm->lru_list);
- kfree(mm);
- }
+ intel_vgpu_free_mm(vgpu, INTEL_GVT_MM_PPGTT);
+ intel_vgpu_free_mm(vgpu, INTEL_GVT_MM_GGTT);
}
static void clean_spt_oos(struct intel_gvt *gvt)
@@ -2071,7 +2079,6 @@ static int setup_spt_oos(struct intel_gvt *gvt)
for (i = 0; i < preallocated_oos_pages; i++) {
oos_page = kzalloc(sizeof(*oos_page), GFP_KERNEL);
if (!oos_page) {
- gvt_err("fail to pre-allocate oos page\n");
ret = -ENOMEM;
goto fail;
}
@@ -2161,7 +2168,7 @@ int intel_vgpu_g2v_create_ppgtt_mm(struct intel_vgpu *vgpu,
mm = intel_vgpu_create_mm(vgpu, INTEL_GVT_MM_PPGTT,
pdp, page_table_level, 0);
if (IS_ERR(mm)) {
- gvt_err("fail to create mm\n");
+ gvt_vgpu_err("fail to create mm\n");
return PTR_ERR(mm);
}
}
@@ -2191,7 +2198,7 @@ int intel_vgpu_g2v_destroy_ppgtt_mm(struct intel_vgpu *vgpu,
mm = intel_vgpu_find_ppgtt_mm(vgpu, page_table_level, pdp);
if (!mm) {
- gvt_err("fail to find ppgtt instance.\n");
+ gvt_vgpu_err("fail to find ppgtt instance.\n");
return -EINVAL;
}
intel_gvt_mm_unreference(mm);
@@ -2322,6 +2329,13 @@ void intel_vgpu_reset_gtt(struct intel_vgpu *vgpu, bool dmlr)
int i;
ppgtt_free_all_shadow_page(vgpu);
+
+ /* Shadow pages are only created when there is no page
+ * table tracking data, so remove page tracking data after
+ * removing the shadow pages.
+ */
+ intel_vgpu_free_mm(vgpu, INTEL_GVT_MM_PPGTT);
+
if (!dmlr)
return;
diff --git a/drivers/gpu/drm/i915/gvt/gvt.c b/drivers/gpu/drm/i915/gvt/gvt.c
index 3b9d59e457ba..ef3baa0c4754 100644
--- a/drivers/gpu/drm/i915/gvt/gvt.c
+++ b/drivers/gpu/drm/i915/gvt/gvt.c
@@ -52,6 +52,8 @@ static const struct intel_gvt_ops intel_gvt_ops = {
.vgpu_create = intel_gvt_create_vgpu,
.vgpu_destroy = intel_gvt_destroy_vgpu,
.vgpu_reset = intel_gvt_reset_vgpu,
+ .vgpu_activate = intel_gvt_activate_vgpu,
+ .vgpu_deactivate = intel_gvt_deactivate_vgpu,
};
/**
diff --git a/drivers/gpu/drm/i915/gvt/gvt.h b/drivers/gpu/drm/i915/gvt/gvt.h
index e227caf5859e..becae2fa3b29 100644
--- a/drivers/gpu/drm/i915/gvt/gvt.h
+++ b/drivers/gpu/drm/i915/gvt/gvt.h
@@ -143,6 +143,8 @@ struct intel_vgpu {
int id;
unsigned long handle; /* vGPU handle used by hypervisor MPT modules */
bool active;
+ bool pv_notified;
+ bool failsafe;
bool resetting;
void *sched_data;
@@ -160,7 +162,6 @@ struct intel_vgpu {
atomic_t running_workload_num;
DECLARE_BITMAP(tlb_handle_pending, I915_NUM_ENGINES);
struct i915_gem_context *shadow_ctx;
- struct notifier_block shadow_ctx_notifier_block;
#if IS_ENABLED(CONFIG_DRM_I915_GVT_KVMGT)
struct {
@@ -203,18 +204,18 @@ struct intel_gvt_firmware {
};
struct intel_gvt_opregion {
- void __iomem *opregion_va;
+ void *opregion_va;
u32 opregion_pa;
};
#define NR_MAX_INTEL_VGPU_TYPES 20
struct intel_vgpu_type {
char name[16];
- unsigned int max_instance;
unsigned int avail_instance;
unsigned int low_gm_size;
unsigned int high_gm_size;
unsigned int fence;
+ enum intel_vgpu_edid resolution;
};
struct intel_gvt {
@@ -231,6 +232,7 @@ struct intel_gvt {
struct intel_gvt_gtt gtt;
struct intel_gvt_opregion opregion;
struct intel_gvt_workload_scheduler scheduler;
+ struct notifier_block shadow_ctx_notifier_block[I915_NUM_ENGINES];
DECLARE_HASHTABLE(cmd_table, GVT_CMD_HASH_BITS);
struct intel_vgpu_type *types;
unsigned int num_types;
@@ -317,6 +319,7 @@ struct intel_vgpu_creation_params {
__u64 low_gm_sz; /* in MB */
__u64 high_gm_sz; /* in MB */
__u64 fence_sz;
+ __u64 resolution;
__s32 primary;
__u64 vgpu_id;
};
@@ -379,7 +382,8 @@ void intel_gvt_destroy_vgpu(struct intel_vgpu *vgpu);
void intel_gvt_reset_vgpu_locked(struct intel_vgpu *vgpu, bool dmlr,
unsigned int engine_mask);
void intel_gvt_reset_vgpu(struct intel_vgpu *vgpu);
-
+void intel_gvt_activate_vgpu(struct intel_vgpu *vgpu);
+void intel_gvt_deactivate_vgpu(struct intel_vgpu *vgpu);
/* validating GM functions */
#define vgpu_gmadr_is_aperture(vgpu, gmadr) \
@@ -446,9 +450,16 @@ struct intel_gvt_ops {
struct intel_vgpu_type *);
void (*vgpu_destroy)(struct intel_vgpu *);
void (*vgpu_reset)(struct intel_vgpu *);
+ void (*vgpu_activate)(struct intel_vgpu *);
+ void (*vgpu_deactivate)(struct intel_vgpu *);
};
+enum {
+ GVT_FAILSAFE_UNSUPPORTED_GUEST,
+ GVT_FAILSAFE_INSUFFICIENT_RESOURCE,
+};
+
#include "mpt.h"
#endif
diff --git a/drivers/gpu/drm/i915/gvt/handlers.c b/drivers/gpu/drm/i915/gvt/handlers.c
index 1d450627ff65..6da9ae1618e3 100644
--- a/drivers/gpu/drm/i915/gvt/handlers.c
+++ b/drivers/gpu/drm/i915/gvt/handlers.c
@@ -121,6 +121,7 @@ static int new_mmio_info(struct intel_gvt *gvt,
info->size = size;
info->length = (i + 4) < end ? 4 : (end - i);
info->addr_mask = addr_mask;
+ info->ro_mask = ro_mask;
info->device = device;
info->read = read ? read : intel_vgpu_default_mmio_read;
info->write = write ? write : intel_vgpu_default_mmio_write;
@@ -150,15 +151,42 @@ static int render_mmio_to_ring_id(struct intel_gvt *gvt, unsigned int reg)
#define fence_num_to_offset(num) \
(num * 8 + i915_mmio_reg_offset(FENCE_REG_GEN6_LO(0)))
+
+static void enter_failsafe_mode(struct intel_vgpu *vgpu, int reason)
+{
+ switch (reason) {
+ case GVT_FAILSAFE_UNSUPPORTED_GUEST:
+ pr_err("Detected your guest driver doesn't support GVT-g.\n");
+ break;
+ case GVT_FAILSAFE_INSUFFICIENT_RESOURCE:
+ pr_err("Graphics resource is not enough for the guest\n");
+ default:
+ break;
+ }
+ pr_err("Now vgpu %d will enter failsafe mode.\n", vgpu->id);
+ vgpu->failsafe = true;
+}
+
static int sanitize_fence_mmio_access(struct intel_vgpu *vgpu,
unsigned int fence_num, void *p_data, unsigned int bytes)
{
if (fence_num >= vgpu_fence_sz(vgpu)) {
- gvt_err("vgpu%d: found oob fence register access\n",
- vgpu->id);
- gvt_err("vgpu%d: total fence num %d access fence num %d\n",
- vgpu->id, vgpu_fence_sz(vgpu), fence_num);
+
+ /* When guest access oob fence regs without access
+ * pv_info first, we treat guest not supporting GVT,
+ * and we will let vgpu enter failsafe mode.
+ */
+ if (!vgpu->pv_notified)
+ enter_failsafe_mode(vgpu,
+ GVT_FAILSAFE_UNSUPPORTED_GUEST);
+
+ if (!vgpu->mmio.disable_warn_untrack) {
+ gvt_vgpu_err("found oob fence register access\n");
+ gvt_vgpu_err("total fence %d, access fence %d\n",
+ vgpu_fence_sz(vgpu), fence_num);
+ }
memset(p_data, 0, bytes);
+ return -EINVAL;
}
return 0;
}
@@ -219,7 +247,7 @@ static int mul_force_wake_write(struct intel_vgpu *vgpu,
break;
default:
/*should not hit here*/
- gvt_err("invalid forcewake offset 0x%x\n", offset);
+ gvt_vgpu_err("invalid forcewake offset 0x%x\n", offset);
return -EINVAL;
}
} else {
@@ -369,6 +397,74 @@ static int pipeconf_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
return 0;
}
+/* ascendingly sorted */
+static i915_reg_t force_nonpriv_white_list[] = {
+ GEN9_CS_DEBUG_MODE1, //_MMIO(0x20ec)
+ GEN9_CTX_PREEMPT_REG,//_MMIO(0x2248)
+ GEN8_CS_CHICKEN1,//_MMIO(0x2580)
+ _MMIO(0x2690),
+ _MMIO(0x2694),
+ _MMIO(0x2698),
+ _MMIO(0x4de0),
+ _MMIO(0x4de4),
+ _MMIO(0x4dfc),
+ GEN7_COMMON_SLICE_CHICKEN1,//_MMIO(0x7010)
+ _MMIO(0x7014),
+ HDC_CHICKEN0,//_MMIO(0x7300)
+ GEN8_HDC_CHICKEN1,//_MMIO(0x7304)
+ _MMIO(0x7700),
+ _MMIO(0x7704),
+ _MMIO(0x7708),
+ _MMIO(0x770c),
+ _MMIO(0xb110),
+ GEN8_L3SQCREG4,//_MMIO(0xb118)
+ _MMIO(0xe100),
+ _MMIO(0xe18c),
+ _MMIO(0xe48c),
+ _MMIO(0xe5f4),
+};
+
+/* a simple bsearch */
+static inline bool in_whitelist(unsigned int reg)
+{
+ int left = 0, right = ARRAY_SIZE(force_nonpriv_white_list);
+ i915_reg_t *array = force_nonpriv_white_list;
+
+ while (left < right) {
+ int mid = (left + right)/2;
+
+ if (reg > array[mid].reg)
+ left = mid + 1;
+ else if (reg < array[mid].reg)
+ right = mid;
+ else
+ return true;
+ }
+ return false;
+}
+
+static int force_nonpriv_write(struct intel_vgpu *vgpu,
+ unsigned int offset, void *p_data, unsigned int bytes)
+{
+ u32 reg_nonpriv = *(u32 *)p_data;
+ int ret = -EINVAL;
+
+ if ((bytes != 4) || ((offset & (bytes - 1)) != 0)) {
+ gvt_err("vgpu(%d) Invalid FORCE_NONPRIV offset %x(%dB)\n",
+ vgpu->id, offset, bytes);
+ return ret;
+ }
+
+ if (in_whitelist(reg_nonpriv)) {
+ ret = intel_vgpu_default_mmio_write(vgpu, offset, p_data,
+ bytes);
+ } else {
+ gvt_err("vgpu(%d) Invalid FORCE_NONPRIV write %x\n",
+ vgpu->id, reg_nonpriv);
+ }
+ return ret;
+}
+
static int ddi_buf_ctl_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
void *p_data, unsigned int bytes)
{
@@ -432,7 +528,7 @@ static int check_fdi_rx_train_status(struct intel_vgpu *vgpu,
fdi_tx_train_bits = FDI_LINK_TRAIN_PATTERN_2;
fdi_iir_check_bits = FDI_RX_SYMBOL_LOCK;
} else {
- gvt_err("Invalid train pattern %d\n", train_pattern);
+ gvt_vgpu_err("Invalid train pattern %d\n", train_pattern);
return -EINVAL;
}
@@ -490,7 +586,7 @@ static int update_fdi_rx_iir_status(struct intel_vgpu *vgpu,
else if (FDI_RX_IMR_TO_PIPE(offset) != INVALID_INDEX)
index = FDI_RX_IMR_TO_PIPE(offset);
else {
- gvt_err("Unsupport registers %x\n", offset);
+ gvt_vgpu_err("Unsupport registers %x\n", offset);
return -EINVAL;
}
@@ -720,7 +816,7 @@ static int dp_aux_ch_ctl_mmio_write(struct intel_vgpu *vgpu,
u32 data;
if (!dpy_is_valid_port(port_index)) {
- gvt_err("GVT(%d): Unsupported DP port access!\n", vgpu->id);
+ gvt_vgpu_err("Unsupported DP port access!\n");
return 0;
}
@@ -874,6 +970,14 @@ static int dp_aux_ch_ctl_mmio_write(struct intel_vgpu *vgpu,
return 0;
}
+static int mbctl_write(struct intel_vgpu *vgpu, unsigned int offset,
+ void *p_data, unsigned int bytes)
+{
+ *(u32 *)p_data &= (~GEN6_MBCTL_ENABLE_BOOT_FETCH);
+ write_vreg(vgpu, offset, p_data, bytes);
+ return 0;
+}
+
static int vga_control_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
void *p_data, unsigned int bytes)
{
@@ -918,8 +1022,7 @@ static void write_virtual_sbi_register(struct intel_vgpu *vgpu,
if (i == num) {
if (num == SBI_REG_MAX) {
- gvt_err("vgpu%d: SBI caching meets maximum limits\n",
- vgpu->id);
+ gvt_vgpu_err("SBI caching meets maximum limits\n");
return;
}
display->sbi.number++;
@@ -999,8 +1102,9 @@ static int pvinfo_mmio_read(struct intel_vgpu *vgpu, unsigned int offset,
break;
}
if (invalid_read)
- gvt_err("invalid pvinfo read: [%x:%x] = %x\n",
+ gvt_vgpu_err("invalid pvinfo read: [%x:%x] = %x\n",
offset, bytes, *(u32 *)p_data);
+ vgpu->pv_notified = true;
return 0;
}
@@ -1026,7 +1130,7 @@ static int handle_g2v_notification(struct intel_vgpu *vgpu, int notification)
case 1: /* Remove this in guest driver. */
break;
default:
- gvt_err("Invalid PV notification %d\n", notification);
+ gvt_vgpu_err("Invalid PV notification %d\n", notification);
}
return ret;
}
@@ -1039,7 +1143,7 @@ static int send_display_ready_uevent(struct intel_vgpu *vgpu, int ready)
char vmid_str[20];
char display_ready_str[20];
- snprintf(display_ready_str, 20, "GVT_DISPLAY_READY=%d\n", ready);
+ snprintf(display_ready_str, 20, "GVT_DISPLAY_READY=%d", ready);
env[0] = display_ready_str;
snprintf(vmid_str, 20, "VMID=%d", vgpu->id);
@@ -1078,8 +1182,11 @@ static int pvinfo_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
case _vgtif_reg(execlist_context_descriptor_lo):
case _vgtif_reg(execlist_context_descriptor_hi):
break;
+ case _vgtif_reg(rsv5[0])..._vgtif_reg(rsv5[3]):
+ enter_failsafe_mode(vgpu, GVT_FAILSAFE_INSUFFICIENT_RESOURCE);
+ break;
default:
- gvt_err("invalid pvinfo write offset %x bytes %x data %x\n",
+ gvt_vgpu_err("invalid pvinfo write offset %x bytes %x data %x\n",
offset, bytes, data);
break;
}
@@ -1203,26 +1310,37 @@ static int mailbox_write(struct intel_vgpu *vgpu, unsigned int offset,
u32 *data0 = &vgpu_vreg(vgpu, GEN6_PCODE_DATA);
switch (cmd) {
- case 0x6:
- /**
- * "Read memory latency" command on gen9.
- * Below memory latency values are read
- * from skylake platform.
- */
- if (!*data0)
- *data0 = 0x1e1a1100;
- else
- *data0 = 0x61514b3d;
+ case GEN9_PCODE_READ_MEM_LATENCY:
+ if (IS_SKYLAKE(vgpu->gvt->dev_priv)) {
+ /**
+ * "Read memory latency" command on gen9.
+ * Below memory latency values are read
+ * from skylake platform.
+ */
+ if (!*data0)
+ *data0 = 0x1e1a1100;
+ else
+ *data0 = 0x61514b3d;
+ }
break;
- case 0x5:
+ case SKL_PCODE_CDCLK_CONTROL:
+ if (IS_SKYLAKE(vgpu->gvt->dev_priv))
+ *data0 = SKL_CDCLK_READY_FOR_CHANGE;
+ break;
+ case GEN6_PCODE_READ_RC6VIDS:
*data0 |= 0x1;
break;
}
gvt_dbg_core("VM(%d) write %x to mailbox, return data0 %x\n",
vgpu->id, value, *data0);
-
- value &= ~(1 << 31);
+ /**
+ * PCODE_READY clear means ready for pcode read/write,
+ * PCODE_ERROR_MASK clear means no error happened. In GVT-g we
+ * always emulate as pcode read/write success and ready for access
+ * anytime, since we don't touch real physical registers here.
+ */
+ value &= ~(GEN6_PCODE_READY | GEN6_PCODE_ERROR_MASK);
return intel_vgpu_default_mmio_write(vgpu, offset, &value, bytes);
}
@@ -1302,7 +1420,8 @@ static int elsp_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
if (execlist->elsp_dwords.index == 3) {
ret = intel_vgpu_submit_execlist(vgpu, ring_id);
if(ret)
- gvt_err("fail submit workload on ring %d\n", ring_id);
+ gvt_vgpu_err("fail submit workload on ring %d\n",
+ ring_id);
}
++execlist->elsp_dwords.index;
@@ -1318,6 +1437,17 @@ static int ring_mode_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
bool enable_execlist;
write_vreg(vgpu, offset, p_data, bytes);
+
+ /* when PPGTT mode enabled, we will check if guest has called
+ * pvinfo, if not, we will treat this guest as non-gvtg-aware
+ * guest, and stop emulating its cfg space, mmio, gtt, etc.
+ */
+ if (((data & _MASKED_BIT_ENABLE(GFX_PPGTT_ENABLE)) ||
+ (data & _MASKED_BIT_ENABLE(GFX_RUN_LIST_ENABLE)))
+ && !vgpu->pv_notified) {
+ enter_failsafe_mode(vgpu, GVT_FAILSAFE_UNSUPPORTED_GUEST);
+ return 0;
+ }
if ((data & _MASKED_BIT_ENABLE(GFX_RUN_LIST_ENABLE))
|| (data & _MASKED_BIT_DISABLE(GFX_RUN_LIST_ENABLE))) {
enable_execlist = !!(data & GFX_RUN_LIST_ENABLE);
@@ -1400,6 +1530,9 @@ static int ring_reset_ctl_write(struct intel_vgpu *vgpu,
#define MMIO_GM(reg, d, r, w) \
MMIO_F(reg, 4, F_GMADR, 0xFFFFF000, 0, d, r, w)
+#define MMIO_GM_RDR(reg, d, r, w) \
+ MMIO_F(reg, 4, F_GMADR | F_CMD_ACCESS, 0xFFFFF000, 0, d, r, w)
+
#define MMIO_RO(reg, d, f, rm, r, w) \
MMIO_F(reg, 4, F_RO | f, 0, rm, d, r, w)
@@ -1419,6 +1552,9 @@ static int ring_reset_ctl_write(struct intel_vgpu *vgpu,
#define MMIO_RING_GM(prefix, d, r, w) \
MMIO_RING_F(prefix, 4, F_GMADR, 0xFFFF0000, 0, d, r, w)
+#define MMIO_RING_GM_RDR(prefix, d, r, w) \
+ MMIO_RING_F(prefix, 4, F_GMADR | F_CMD_ACCESS, 0xFFFF0000, 0, d, r, w)
+
#define MMIO_RING_RO(prefix, d, f, rm, r, w) \
MMIO_RING_F(prefix, 4, F_RO | f, 0, rm, d, r, w)
@@ -1427,73 +1563,81 @@ static int init_generic_mmio_info(struct intel_gvt *gvt)
struct drm_i915_private *dev_priv = gvt->dev_priv;
int ret;
- MMIO_RING_DFH(RING_IMR, D_ALL, 0, NULL, intel_vgpu_reg_imr_handler);
+ MMIO_RING_DFH(RING_IMR, D_ALL, F_CMD_ACCESS, NULL,
+ intel_vgpu_reg_imr_handler);
MMIO_DFH(SDEIMR, D_ALL, 0, NULL, intel_vgpu_reg_imr_handler);
MMIO_DFH(SDEIER, D_ALL, 0, NULL, intel_vgpu_reg_ier_handler);
MMIO_DFH(SDEIIR, D_ALL, 0, NULL, intel_vgpu_reg_iir_handler);
MMIO_D(SDEISR, D_ALL);
- MMIO_RING_D(RING_HWSTAM, D_ALL);
+ MMIO_RING_DFH(RING_HWSTAM, D_ALL, F_CMD_ACCESS, NULL, NULL);
- MMIO_GM(RENDER_HWS_PGA_GEN7, D_ALL, NULL, NULL);
- MMIO_GM(BSD_HWS_PGA_GEN7, D_ALL, NULL, NULL);
- MMIO_GM(BLT_HWS_PGA_GEN7, D_ALL, NULL, NULL);
- MMIO_GM(VEBOX_HWS_PGA_GEN7, D_ALL, NULL, NULL);
+ MMIO_GM_RDR(RENDER_HWS_PGA_GEN7, D_ALL, NULL, NULL);
+ MMIO_GM_RDR(BSD_HWS_PGA_GEN7, D_ALL, NULL, NULL);
+ MMIO_GM_RDR(BLT_HWS_PGA_GEN7, D_ALL, NULL, NULL);
+ MMIO_GM_RDR(VEBOX_HWS_PGA_GEN7, D_ALL, NULL, NULL);
#define RING_REG(base) (base + 0x28)
- MMIO_RING_D(RING_REG, D_ALL);
+ MMIO_RING_DFH(RING_REG, D_ALL, F_CMD_ACCESS, NULL, NULL);
#undef RING_REG
#define RING_REG(base) (base + 0x134)
- MMIO_RING_D(RING_REG, D_ALL);
+ MMIO_RING_DFH(RING_REG, D_ALL, F_CMD_ACCESS, NULL, NULL);
#undef RING_REG
- MMIO_GM(0x2148, D_ALL, NULL, NULL);
- MMIO_GM(CCID, D_ALL, NULL, NULL);
- MMIO_GM(0x12198, D_ALL, NULL, NULL);
+ MMIO_GM_RDR(0x2148, D_ALL, NULL, NULL);
+ MMIO_GM_RDR(CCID, D_ALL, NULL, NULL);
+ MMIO_GM_RDR(0x12198, D_ALL, NULL, NULL);
MMIO_D(GEN7_CXT_SIZE, D_ALL);
- MMIO_RING_D(RING_TAIL, D_ALL);
- MMIO_RING_D(RING_HEAD, D_ALL);
- MMIO_RING_D(RING_CTL, D_ALL);
- MMIO_RING_D(RING_ACTHD, D_ALL);
- MMIO_RING_GM(RING_START, D_ALL, NULL, NULL);
+ MMIO_RING_DFH(RING_TAIL, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_RING_DFH(RING_HEAD, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_RING_DFH(RING_CTL, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_RING_DFH(RING_ACTHD, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_RING_GM_RDR(RING_START, D_ALL, NULL, NULL);
/* RING MODE */
#define RING_REG(base) (base + 0x29c)
- MMIO_RING_DFH(RING_REG, D_ALL, F_MODE_MASK, NULL, ring_mode_mmio_write);
+ MMIO_RING_DFH(RING_REG, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL,
+ ring_mode_mmio_write);
#undef RING_REG
- MMIO_RING_DFH(RING_MI_MODE, D_ALL, F_MODE_MASK, NULL, NULL);
- MMIO_RING_DFH(RING_INSTPM, D_ALL, F_MODE_MASK, NULL, NULL);
+ MMIO_RING_DFH(RING_MI_MODE, D_ALL, F_MODE_MASK | F_CMD_ACCESS,
+ NULL, NULL);
+ MMIO_RING_DFH(RING_INSTPM, D_ALL, F_MODE_MASK | F_CMD_ACCESS,
+ NULL, NULL);
MMIO_RING_DFH(RING_TIMESTAMP, D_ALL, F_CMD_ACCESS,
ring_timestamp_mmio_read, NULL);
MMIO_RING_DFH(RING_TIMESTAMP_UDW, D_ALL, F_CMD_ACCESS,
ring_timestamp_mmio_read, NULL);
- MMIO_DFH(GEN7_GT_MODE, D_ALL, F_MODE_MASK, NULL, NULL);
- MMIO_DFH(CACHE_MODE_0_GEN7, D_ALL, F_MODE_MASK, NULL, NULL);
+ MMIO_DFH(GEN7_GT_MODE, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(CACHE_MODE_0_GEN7, D_ALL, F_MODE_MASK | F_CMD_ACCESS,
+ NULL, NULL);
MMIO_DFH(CACHE_MODE_1, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
-
- MMIO_DFH(0x20dc, D_ALL, F_MODE_MASK, NULL, NULL);
- MMIO_DFH(_3D_CHICKEN3, D_ALL, F_MODE_MASK, NULL, NULL);
- MMIO_DFH(0x2088, D_ALL, F_MODE_MASK, NULL, NULL);
- MMIO_DFH(0x20e4, D_ALL, F_MODE_MASK, NULL, NULL);
- MMIO_DFH(0x2470, D_ALL, F_MODE_MASK, NULL, NULL);
- MMIO_D(GAM_ECOCHK, D_ALL);
- MMIO_DFH(GEN7_COMMON_SLICE_CHICKEN1, D_ALL, F_MODE_MASK, NULL, NULL);
+ MMIO_DFH(CACHE_MODE_0, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x2124, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+
+ MMIO_DFH(0x20dc, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(_3D_CHICKEN3, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x2088, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x20e4, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x2470, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(GAM_ECOCHK, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(GEN7_COMMON_SLICE_CHICKEN1, D_ALL, F_MODE_MASK | F_CMD_ACCESS,
+ NULL, NULL);
MMIO_DFH(COMMON_SLICE_CHICKEN2, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
- MMIO_D(0x9030, D_ALL);
- MMIO_D(0x20a0, D_ALL);
- MMIO_D(0x2420, D_ALL);
- MMIO_D(0x2430, D_ALL);
- MMIO_D(0x2434, D_ALL);
- MMIO_D(0x2438, D_ALL);
- MMIO_D(0x243c, D_ALL);
- MMIO_DFH(0x7018, D_ALL, F_MODE_MASK, NULL, NULL);
+ MMIO_DFH(0x9030, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x20a0, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x2420, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x2430, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x2434, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x2438, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x243c, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x7018, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(HALF_SLICE_CHICKEN3, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
- MMIO_DFH(0xe100, D_ALL, F_MODE_MASK, NULL, NULL);
+ MMIO_DFH(GEN7_HALF_SLICE_CHICKEN1, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
/* display */
MMIO_F(0x60220, 0x20, 0, 0, 0, D_ALL, NULL, NULL);
@@ -2022,8 +2166,8 @@ static int init_generic_mmio_info(struct intel_gvt *gvt)
MMIO_D(FORCEWAKE_ACK, D_ALL);
MMIO_D(GEN6_GT_CORE_STATUS, D_ALL);
MMIO_D(GEN6_GT_THREAD_STATUS_REG, D_ALL);
- MMIO_D(GTFIFODBG, D_ALL);
- MMIO_D(GTFIFOCTL, D_ALL);
+ MMIO_DFH(GTFIFODBG, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(GTFIFOCTL, D_ALL, F_CMD_ACCESS, NULL, NULL);
MMIO_DH(FORCEWAKE_MT, D_PRE_SKL, NULL, mul_force_wake_write);
MMIO_DH(FORCEWAKE_ACK_HSW, D_HSW | D_BDW, NULL, NULL);
MMIO_D(ECOBUS, D_ALL);
@@ -2080,7 +2224,7 @@ static int init_generic_mmio_info(struct intel_gvt *gvt)
MMIO_F(0x4f000, 0x90, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_D(GEN6_PCODE_MAILBOX, D_PRE_SKL);
+ MMIO_D(GEN6_PCODE_MAILBOX, D_PRE_BDW);
MMIO_D(GEN6_PCODE_DATA, D_ALL);
MMIO_D(0x13812c, D_ALL);
MMIO_DH(GEN7_ERR_INT, D_ALL, NULL, NULL);
@@ -2102,7 +2246,7 @@ static int init_generic_mmio_info(struct intel_gvt *gvt)
MMIO_D(0x7180, D_ALL);
MMIO_D(0x7408, D_ALL);
MMIO_D(0x7c00, D_ALL);
- MMIO_D(GEN6_MBCTL, D_ALL);
+ MMIO_DH(GEN6_MBCTL, D_ALL, NULL, mbctl_write);
MMIO_D(0x911c, D_ALL);
MMIO_D(0x9120, D_ALL);
MMIO_DFH(GEN7_UCGCTL4, D_ALL, F_CMD_ACCESS, NULL, NULL);
@@ -2159,36 +2303,35 @@ static int init_generic_mmio_info(struct intel_gvt *gvt)
MMIO_D(0x1a054, D_ALL);
MMIO_D(0x44070, D_ALL);
-
- MMIO_D(0x215c, D_HSW_PLUS);
+ MMIO_DFH(0x215c, D_HSW_PLUS, F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(0x2178, D_ALL, F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(0x217c, D_ALL, F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(0x12178, D_ALL, F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(0x1217c, D_ALL, F_CMD_ACCESS, NULL, NULL);
- MMIO_F(0x2290, 8, 0, 0, 0, D_HSW_PLUS, NULL, NULL);
- MMIO_D(GEN7_OACONTROL, D_HSW);
+ MMIO_F(0x2290, 8, F_CMD_ACCESS, 0, 0, D_HSW_PLUS, NULL, NULL);
+ MMIO_DFH(GEN7_OACONTROL, D_HSW, F_CMD_ACCESS, NULL, NULL);
MMIO_D(0x2b00, D_BDW_PLUS);
MMIO_D(0x2360, D_BDW_PLUS);
- MMIO_F(0x5200, 32, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(0x5240, 32, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(0x5280, 16, 0, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(0x5200, 32, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(0x5240, 32, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(0x5280, 16, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
MMIO_DFH(0x1c17c, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(0x1c178, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
- MMIO_D(BCS_SWCTRL, D_ALL);
-
- MMIO_F(HS_INVOCATION_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(DS_INVOCATION_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(IA_VERTICES_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(IA_PRIMITIVES_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(VS_INVOCATION_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(GS_INVOCATION_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(GS_PRIMITIVES_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(CL_INVOCATION_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(CL_PRIMITIVES_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(PS_INVOCATION_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
- MMIO_F(PS_DEPTH_COUNT, 8, 0, 0, 0, D_ALL, NULL, NULL);
+ MMIO_DFH(BCS_SWCTRL, D_ALL, F_CMD_ACCESS, NULL, NULL);
+
+ MMIO_F(HS_INVOCATION_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(DS_INVOCATION_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(IA_VERTICES_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(IA_PRIMITIVES_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(VS_INVOCATION_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(GS_INVOCATION_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(GS_PRIMITIVES_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(CL_INVOCATION_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(CL_PRIMITIVES_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(PS_INVOCATION_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
+ MMIO_F(PS_DEPTH_COUNT, 8, F_CMD_ACCESS, 0, 0, D_ALL, NULL, NULL);
MMIO_DH(0x4260, D_BDW_PLUS, NULL, gvt_reg_tlb_control_handler);
MMIO_DH(0x4264, D_BDW_PLUS, NULL, gvt_reg_tlb_control_handler);
MMIO_DH(0x4268, D_BDW_PLUS, NULL, gvt_reg_tlb_control_handler);
@@ -2196,6 +2339,17 @@ static int init_generic_mmio_info(struct intel_gvt *gvt)
MMIO_DH(0x4270, D_BDW_PLUS, NULL, gvt_reg_tlb_control_handler);
MMIO_DFH(0x4094, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(ARB_MODE, D_ALL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_RING_GM_RDR(RING_BBADDR, D_ALL, NULL, NULL);
+ MMIO_DFH(0x2220, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x12220, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x22220, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_RING_DFH(RING_SYNC_1, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_RING_DFH(RING_SYNC_0, D_ALL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x22178, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x1a178, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x1a17c, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x2217c, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
return 0;
}
@@ -2204,7 +2358,7 @@ static int init_broadwell_mmio_info(struct intel_gvt *gvt)
struct drm_i915_private *dev_priv = gvt->dev_priv;
int ret;
- MMIO_DH(RING_IMR(GEN8_BSD2_RING_BASE), D_BDW_PLUS, NULL,
+ MMIO_DFH(RING_IMR(GEN8_BSD2_RING_BASE), D_BDW_PLUS, F_CMD_ACCESS, NULL,
intel_vgpu_reg_imr_handler);
MMIO_DH(GEN8_GT_IMR(0), D_BDW_PLUS, NULL, intel_vgpu_reg_imr_handler);
@@ -2269,24 +2423,31 @@ static int init_broadwell_mmio_info(struct intel_gvt *gvt)
MMIO_DH(GEN8_MASTER_IRQ, D_BDW_PLUS, NULL,
intel_vgpu_reg_master_irq_handler);
- MMIO_D(RING_HWSTAM(GEN8_BSD2_RING_BASE), D_BDW_PLUS);
- MMIO_D(0x1c134, D_BDW_PLUS);
-
- MMIO_D(RING_TAIL(GEN8_BSD2_RING_BASE), D_BDW_PLUS);
- MMIO_D(RING_HEAD(GEN8_BSD2_RING_BASE), D_BDW_PLUS);
- MMIO_GM(RING_START(GEN8_BSD2_RING_BASE), D_BDW_PLUS, NULL, NULL);
- MMIO_D(RING_CTL(GEN8_BSD2_RING_BASE), D_BDW_PLUS);
- MMIO_D(RING_ACTHD(GEN8_BSD2_RING_BASE), D_BDW_PLUS);
- MMIO_D(RING_ACTHD_UDW(GEN8_BSD2_RING_BASE), D_BDW_PLUS);
- MMIO_DFH(0x1c29c, D_BDW_PLUS, F_MODE_MASK, NULL, ring_mode_mmio_write);
- MMIO_DFH(RING_MI_MODE(GEN8_BSD2_RING_BASE), D_BDW_PLUS, F_MODE_MASK,
- NULL, NULL);
- MMIO_DFH(RING_INSTPM(GEN8_BSD2_RING_BASE), D_BDW_PLUS, F_MODE_MASK,
- NULL, NULL);
+ MMIO_DFH(RING_HWSTAM(GEN8_BSD2_RING_BASE), D_BDW_PLUS,
+ F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x1c134, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+
+ MMIO_DFH(RING_TAIL(GEN8_BSD2_RING_BASE), D_BDW_PLUS, F_CMD_ACCESS,
+ NULL, NULL);
+ MMIO_DFH(RING_HEAD(GEN8_BSD2_RING_BASE), D_BDW_PLUS,
+ F_CMD_ACCESS, NULL, NULL);
+ MMIO_GM_RDR(RING_START(GEN8_BSD2_RING_BASE), D_BDW_PLUS, NULL, NULL);
+ MMIO_DFH(RING_CTL(GEN8_BSD2_RING_BASE), D_BDW_PLUS, F_CMD_ACCESS,
+ NULL, NULL);
+ MMIO_DFH(RING_ACTHD(GEN8_BSD2_RING_BASE), D_BDW_PLUS,
+ F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(RING_ACTHD_UDW(GEN8_BSD2_RING_BASE), D_BDW_PLUS,
+ F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x1c29c, D_BDW_PLUS, F_MODE_MASK | F_CMD_ACCESS, NULL,
+ ring_mode_mmio_write);
+ MMIO_DFH(RING_MI_MODE(GEN8_BSD2_RING_BASE), D_BDW_PLUS,
+ F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(RING_INSTPM(GEN8_BSD2_RING_BASE), D_BDW_PLUS,
+ F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(RING_TIMESTAMP(GEN8_BSD2_RING_BASE), D_BDW_PLUS, F_CMD_ACCESS,
ring_timestamp_mmio_read, NULL);
- MMIO_RING_D(RING_ACTHD_UDW, D_BDW_PLUS);
+ MMIO_RING_DFH(RING_ACTHD_UDW, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
#define RING_REG(base) (base + 0xd0)
MMIO_RING_F(RING_REG, 4, F_RO, 0,
@@ -2303,13 +2464,16 @@ static int init_broadwell_mmio_info(struct intel_gvt *gvt)
#undef RING_REG
#define RING_REG(base) (base + 0x234)
- MMIO_RING_F(RING_REG, 8, F_RO, 0, ~0, D_BDW_PLUS, NULL, NULL);
- MMIO_F(RING_REG(GEN8_BSD2_RING_BASE), 4, F_RO, 0, ~0LL, D_BDW_PLUS, NULL, NULL);
+ MMIO_RING_F(RING_REG, 8, F_RO | F_CMD_ACCESS, 0, ~0, D_BDW_PLUS,
+ NULL, NULL);
+ MMIO_F(RING_REG(GEN8_BSD2_RING_BASE), 4, F_RO | F_CMD_ACCESS, 0,
+ ~0LL, D_BDW_PLUS, NULL, NULL);
#undef RING_REG
#define RING_REG(base) (base + 0x244)
- MMIO_RING_D(RING_REG, D_BDW_PLUS);
- MMIO_D(RING_REG(GEN8_BSD2_RING_BASE), D_BDW_PLUS);
+ MMIO_RING_DFH(RING_REG, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(RING_REG(GEN8_BSD2_RING_BASE), D_BDW_PLUS, F_CMD_ACCESS,
+ NULL, NULL);
#undef RING_REG
#define RING_REG(base) (base + 0x370)
@@ -2331,6 +2495,8 @@ static int init_broadwell_mmio_info(struct intel_gvt *gvt)
MMIO_D(GEN7_MISCCPCTL, D_BDW_PLUS);
MMIO_D(0x1c054, D_BDW_PLUS);
+ MMIO_DH(GEN6_PCODE_MAILBOX, D_BDW_PLUS, NULL, mailbox_write);
+
MMIO_D(GEN8_PRIVATE_PAT_LO, D_BDW_PLUS);
MMIO_D(GEN8_PRIVATE_PAT_HI, D_BDW_PLUS);
@@ -2341,14 +2507,14 @@ static int init_broadwell_mmio_info(struct intel_gvt *gvt)
MMIO_F(RING_REG(GEN8_BSD2_RING_BASE), 32, 0, 0, 0, D_BDW_PLUS, NULL, NULL);
#undef RING_REG
- MMIO_RING_GM(RING_HWS_PGA, D_BDW_PLUS, NULL, NULL);
- MMIO_GM(0x1c080, D_BDW_PLUS, NULL, NULL);
+ MMIO_RING_GM_RDR(RING_HWS_PGA, D_BDW_PLUS, NULL, NULL);
+ MMIO_GM_RDR(RING_HWS_PGA(GEN8_BSD2_RING_BASE), D_BDW_PLUS, NULL, NULL);
MMIO_DFH(HDC_CHICKEN0, D_BDW_PLUS, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
- MMIO_D(CHICKEN_PIPESL_1(PIPE_A), D_BDW);
- MMIO_D(CHICKEN_PIPESL_1(PIPE_B), D_BDW);
- MMIO_D(CHICKEN_PIPESL_1(PIPE_C), D_BDW);
+ MMIO_D(CHICKEN_PIPESL_1(PIPE_A), D_BDW_PLUS);
+ MMIO_D(CHICKEN_PIPESL_1(PIPE_B), D_BDW_PLUS);
+ MMIO_D(CHICKEN_PIPESL_1(PIPE_C), D_BDW_PLUS);
MMIO_D(WM_MISC, D_BDW);
MMIO_D(BDW_EDP_PSR_BASE, D_BDW);
@@ -2362,27 +2528,31 @@ static int init_broadwell_mmio_info(struct intel_gvt *gvt)
MMIO_D(GEN8_EU_DISABLE1, D_BDW_PLUS);
MMIO_D(GEN8_EU_DISABLE2, D_BDW_PLUS);
- MMIO_D(0xfdc, D_BDW);
- MMIO_DFH(GEN8_ROW_CHICKEN, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
- MMIO_D(GEN7_ROW_CHICKEN2, D_BDW_PLUS);
- MMIO_D(GEN8_UCGCTL6, D_BDW_PLUS);
+ MMIO_D(0xfdc, D_BDW_PLUS);
+ MMIO_DFH(GEN8_ROW_CHICKEN, D_BDW_PLUS, F_MODE_MASK | F_CMD_ACCESS,
+ NULL, NULL);
+ MMIO_DFH(GEN7_ROW_CHICKEN2, D_BDW_PLUS, F_MODE_MASK | F_CMD_ACCESS,
+ NULL, NULL);
+ MMIO_DFH(GEN8_UCGCTL6, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
- MMIO_D(0xb1f0, D_BDW);
- MMIO_D(0xb1c0, D_BDW);
+ MMIO_DFH(0xb1f0, D_BDW, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xb1c0, D_BDW, F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(GEN8_L3SQCREG4, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
- MMIO_D(0xb100, D_BDW);
- MMIO_D(0xb10c, D_BDW);
+ MMIO_DFH(0xb100, D_BDW, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xb10c, D_BDW, F_CMD_ACCESS, NULL, NULL);
MMIO_D(0xb110, D_BDW);
- MMIO_DFH(0x24d0, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
- MMIO_DFH(0x24d4, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
- MMIO_DFH(0x24d8, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
- MMIO_DFH(0x24dc, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_F(0x24d0, 48, F_CMD_ACCESS, 0, 0, D_BDW_PLUS,
+ NULL, force_nonpriv_write);
- MMIO_D(0x83a4, D_BDW);
+ MMIO_D(0x22040, D_BDW_PLUS);
+ MMIO_D(0x44484, D_BDW_PLUS);
+ MMIO_D(0x4448c, D_BDW_PLUS);
+
+ MMIO_DFH(0x83a4, D_BDW, F_CMD_ACCESS, NULL, NULL);
MMIO_D(GEN8_L3_LRA_1_GPGPU, D_BDW_PLUS);
- MMIO_D(0x8430, D_BDW);
+ MMIO_DFH(0x8430, D_BDW, F_CMD_ACCESS, NULL, NULL);
MMIO_D(0x110000, D_BDW_PLUS);
@@ -2394,10 +2564,19 @@ static int init_broadwell_mmio_info(struct intel_gvt *gvt)
MMIO_DFH(0xe194, D_BDW_PLUS, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(0xe188, D_BDW_PLUS, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
MMIO_DFH(HALF_SLICE_CHICKEN2, D_BDW_PLUS, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
- MMIO_DFH(0x2580, D_BDW_PLUS, F_MODE_MASK, NULL, NULL);
-
- MMIO_D(0x2248, D_BDW);
-
+ MMIO_DFH(0x2580, D_BDW_PLUS, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
+
+ MMIO_DFH(0x2248, D_BDW, F_CMD_ACCESS, NULL, NULL);
+
+ MMIO_DFH(0xe220, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xe230, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xe240, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xe260, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xe270, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xe280, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xe2a0, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xe2b0, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0xe2c0, D_BDW_PLUS, F_CMD_ACCESS, NULL, NULL);
return 0;
}
@@ -2420,7 +2599,6 @@ static int init_skl_mmio_info(struct intel_gvt *gvt)
MMIO_D(HSW_PWR_WELL_BIOS, D_SKL);
MMIO_DH(HSW_PWR_WELL_DRIVER, D_SKL, NULL, skl_power_well_ctl_write);
- MMIO_DH(GEN6_PCODE_MAILBOX, D_SKL, NULL, mailbox_write);
MMIO_D(0xa210, D_SKL_PLUS);
MMIO_D(GEN9_MEDIA_PG_IDLE_HYSTERESIS, D_SKL_PLUS);
MMIO_D(GEN9_RENDER_PG_IDLE_HYSTERESIS, D_SKL_PLUS);
@@ -2578,16 +2756,16 @@ static int init_skl_mmio_info(struct intel_gvt *gvt)
MMIO_F(0xb020, 0x80, F_CMD_ACCESS, 0, 0, D_SKL, NULL, NULL);
MMIO_D(0xd08, D_SKL);
- MMIO_D(0x20e0, D_SKL);
- MMIO_D(0x20ec, D_SKL);
+ MMIO_DFH(0x20e0, D_SKL, F_MODE_MASK, NULL, NULL);
+ MMIO_DFH(0x20ec, D_SKL, F_MODE_MASK | F_CMD_ACCESS, NULL, NULL);
/* TRTT */
- MMIO_D(0x4de0, D_SKL);
- MMIO_D(0x4de4, D_SKL);
- MMIO_D(0x4de8, D_SKL);
- MMIO_D(0x4dec, D_SKL);
- MMIO_D(0x4df0, D_SKL);
- MMIO_DH(0x4df4, D_SKL, NULL, gen9_trtte_write);
+ MMIO_DFH(0x4de0, D_SKL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x4de4, D_SKL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x4de8, D_SKL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x4dec, D_SKL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x4df0, D_SKL, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(0x4df4, D_SKL, F_CMD_ACCESS, NULL, gen9_trtte_write);
MMIO_DH(0x4dfc, D_SKL, NULL, gen9_trtt_chicken_write);
MMIO_D(0x45008, D_SKL);
@@ -2611,7 +2789,7 @@ static int init_skl_mmio_info(struct intel_gvt *gvt)
MMIO_D(0x65f08, D_SKL);
MMIO_D(0x320f0, D_SKL);
- MMIO_D(_REG_VCS2_EXCC, D_SKL);
+ MMIO_DFH(_REG_VCS2_EXCC, D_SKL, F_CMD_ACCESS, NULL, NULL);
MMIO_D(0x70034, D_SKL);
MMIO_D(0x71034, D_SKL);
MMIO_D(0x72034, D_SKL);
@@ -2624,6 +2802,9 @@ static int init_skl_mmio_info(struct intel_gvt *gvt)
MMIO_D(_PLANE_KEYMSK_1(PIPE_C), D_SKL);
MMIO_D(0x44500, D_SKL);
+ MMIO_DFH(GEN9_CSFE_CHICKEN1_RCS, D_SKL_PLUS, F_CMD_ACCESS, NULL, NULL);
+ MMIO_DFH(GEN8_HDC_CHICKEN1, D_SKL, F_MODE_MASK | F_CMD_ACCESS,
+ NULL, NULL);
return 0;
}
@@ -2813,3 +2994,20 @@ int intel_vgpu_default_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
write_vreg(vgpu, offset, p_data, bytes);
return 0;
}
+
+/**
+ * intel_gvt_in_force_nonpriv_whitelist - if a mmio is in whitelist to be
+ * force-nopriv register
+ *
+ * @gvt: a GVT device
+ * @offset: register offset
+ *
+ * Returns:
+ * True if the register is in force-nonpriv whitelist;
+ * False if outside;
+ */
+bool intel_gvt_in_force_nonpriv_whitelist(struct intel_gvt *gvt,
+ unsigned int offset)
+{
+ return in_whitelist(offset);
+}
diff --git a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvmgt.c
index 0f7f5d97f582..e466259034e2 100644
--- a/drivers/gpu/drm/i915/gvt/kvmgt.c
+++ b/drivers/gpu/drm/i915/gvt/kvmgt.c
@@ -96,10 +96,10 @@ static int gvt_dma_map_iova(struct intel_vgpu *vgpu, kvm_pfn_t pfn,
struct device *dev = &vgpu->gvt->dev_priv->drm.pdev->dev;
dma_addr_t daddr;
- page = pfn_to_page(pfn);
- if (is_error_page(page))
+ if (unlikely(!pfn_valid(pfn)))
return -EFAULT;
+ page = pfn_to_page(pfn);
daddr = dma_map_page(dev, page, 0, PAGE_SIZE,
PCI_DMA_BIDIRECTIONAL);
if (dma_mapping_error(dev, daddr))
@@ -295,10 +295,10 @@ static ssize_t description_show(struct kobject *kobj, struct device *dev,
return 0;
return sprintf(buf, "low_gm_size: %dMB\nhigh_gm_size: %dMB\n"
- "fence: %d\n",
- BYTES_TO_MB(type->low_gm_size),
- BYTES_TO_MB(type->high_gm_size),
- type->fence);
+ "fence: %d\nresolution: %s\n",
+ BYTES_TO_MB(type->low_gm_size),
+ BYTES_TO_MB(type->high_gm_size),
+ type->fence, vgpu_edid_str(type->resolution));
}
static MDEV_TYPE_ATTR_RO(available_instances);
@@ -426,7 +426,7 @@ static void kvmgt_protect_table_del(struct kvmgt_guest_info *info,
static int intel_vgpu_create(struct kobject *kobj, struct mdev_device *mdev)
{
- struct intel_vgpu *vgpu;
+ struct intel_vgpu *vgpu = NULL;
struct intel_vgpu_type *type;
struct device *pdev;
void *gvt;
@@ -437,7 +437,7 @@ static int intel_vgpu_create(struct kobject *kobj, struct mdev_device *mdev)
type = intel_gvt_find_vgpu_type(gvt, kobject_name(kobj));
if (!type) {
- gvt_err("failed to find type %s to create\n",
+ gvt_vgpu_err("failed to find type %s to create\n",
kobject_name(kobj));
ret = -EINVAL;
goto out;
@@ -446,7 +446,7 @@ static int intel_vgpu_create(struct kobject *kobj, struct mdev_device *mdev)
vgpu = intel_gvt_ops->vgpu_create(gvt, type);
if (IS_ERR_OR_NULL(vgpu)) {
ret = vgpu == NULL ? -EFAULT : PTR_ERR(vgpu);
- gvt_err("failed to create intel vgpu: %d\n", ret);
+ gvt_vgpu_err("failed to create intel vgpu: %d\n", ret);
goto out;
}
@@ -526,7 +526,8 @@ static int intel_vgpu_open(struct mdev_device *mdev)
ret = vfio_register_notifier(mdev_dev(mdev), VFIO_IOMMU_NOTIFY, &events,
&vgpu->vdev.iommu_notifier);
if (ret != 0) {
- gvt_err("vfio_register_notifier for iommu failed: %d\n", ret);
+ gvt_vgpu_err("vfio_register_notifier for iommu failed: %d\n",
+ ret);
goto out;
}
@@ -534,7 +535,8 @@ static int intel_vgpu_open(struct mdev_device *mdev)
ret = vfio_register_notifier(mdev_dev(mdev), VFIO_GROUP_NOTIFY, &events,
&vgpu->vdev.group_notifier);
if (ret != 0) {
- gvt_err("vfio_register_notifier for group failed: %d\n", ret);
+ gvt_vgpu_err("vfio_register_notifier for group failed: %d\n",
+ ret);
goto undo_iommu;
}
@@ -542,6 +544,8 @@ static int intel_vgpu_open(struct mdev_device *mdev)
if (ret)
goto undo_group;
+ intel_gvt_ops->vgpu_activate(vgpu);
+
atomic_set(&vgpu->vdev.released, 0);
return ret;
@@ -567,6 +571,8 @@ static void __intel_vgpu_release(struct intel_vgpu *vgpu)
if (atomic_cmpxchg(&vgpu->vdev.released, 0, 1))
return;
+ intel_gvt_ops->vgpu_deactivate(vgpu);
+
ret = vfio_unregister_notifier(mdev_dev(vgpu->vdev.mdev), VFIO_IOMMU_NOTIFY,
&vgpu->vdev.iommu_notifier);
WARN(ret, "vfio_unregister_notifier for iommu failed: %d\n", ret);
@@ -635,7 +641,7 @@ static ssize_t intel_vgpu_rw(struct mdev_device *mdev, char *buf,
if (index >= VFIO_PCI_NUM_REGIONS) {
- gvt_err("invalid index: %u\n", index);
+ gvt_vgpu_err("invalid index: %u\n", index);
return -EINVAL;
}
@@ -669,7 +675,7 @@ static ssize_t intel_vgpu_rw(struct mdev_device *mdev, char *buf,
case VFIO_PCI_VGA_REGION_INDEX:
case VFIO_PCI_ROM_REGION_INDEX:
default:
- gvt_err("unsupported region: %u\n", index);
+ gvt_vgpu_err("unsupported region: %u\n", index);
}
return ret == 0 ? count : ret;
@@ -861,7 +867,7 @@ static int intel_vgpu_set_msi_trigger(struct intel_vgpu *vgpu,
trigger = eventfd_ctx_fdget(fd);
if (IS_ERR(trigger)) {
- gvt_err("eventfd_ctx_fdget failed\n");
+ gvt_vgpu_err("eventfd_ctx_fdget failed\n");
return PTR_ERR(trigger);
}
vgpu->vdev.msi_trigger = trigger;
@@ -1120,7 +1126,7 @@ static long intel_vgpu_ioctl(struct mdev_device *mdev, unsigned int cmd,
ret = vfio_set_irqs_validate_and_prepare(&hdr, max,
VFIO_PCI_NUM_IRQS, &data_size);
if (ret) {
- gvt_err("intel:vfio_set_irqs_validate_and_prepare failed\n");
+ gvt_vgpu_err("intel:vfio_set_irqs_validate_and_prepare failed\n");
return -EINVAL;
}
if (data_size) {
@@ -1310,7 +1316,7 @@ static int kvmgt_guest_init(struct mdev_device *mdev)
kvm = vgpu->vdev.kvm;
if (!kvm || kvm->mm != current->mm) {
- gvt_err("KVM is required to use Intel vGPU\n");
+ gvt_vgpu_err("KVM is required to use Intel vGPU\n");
return -ESRCH;
}
@@ -1324,6 +1330,7 @@ static int kvmgt_guest_init(struct mdev_device *mdev)
vgpu->handle = (unsigned long)info;
info->vgpu = vgpu;
info->kvm = kvm;
+ kvm_get_kvm(info->kvm);
kvmgt_protect_table_init(info);
gvt_cache_init(vgpu);
@@ -1337,12 +1344,8 @@ static int kvmgt_guest_init(struct mdev_device *mdev)
static bool kvmgt_guest_exit(struct kvmgt_guest_info *info)
{
- if (!info) {
- gvt_err("kvmgt_guest_info invalid\n");
- return false;
- }
-
kvm_page_track_unregister_notifier(info->kvm, &info->track_node);
+ kvm_put_kvm(info->kvm);
kvmgt_protect_table_destroy(info);
gvt_cache_destroy(info->vgpu);
vfree(info);
@@ -1383,12 +1386,14 @@ static unsigned long kvmgt_gfn_to_pfn(unsigned long handle, unsigned long gfn)
unsigned long iova, pfn;
struct kvmgt_guest_info *info;
struct device *dev;
+ struct intel_vgpu *vgpu;
int rc;
if (!handle_valid(handle))
return INTEL_GVT_INVALID_ADDR;
info = (struct kvmgt_guest_info *)handle;
+ vgpu = info->vgpu;
iova = gvt_cache_find(info->vgpu, gfn);
if (iova != INTEL_GVT_INVALID_ADDR)
return iova;
@@ -1397,13 +1402,14 @@ static unsigned long kvmgt_gfn_to_pfn(unsigned long handle, unsigned long gfn)
dev = mdev_dev(info->vgpu->vdev.mdev);
rc = vfio_pin_pages(dev, &gfn, 1, IOMMU_READ | IOMMU_WRITE, &pfn);
if (rc != 1) {
- gvt_err("vfio_pin_pages failed for gfn 0x%lx: %d\n", gfn, rc);
+ gvt_vgpu_err("vfio_pin_pages failed for gfn 0x%lx: %d\n",
+ gfn, rc);
return INTEL_GVT_INVALID_ADDR;
}
/* transfer to host iova for GFX to use DMA */
rc = gvt_dma_map_iova(info->vgpu, pfn, &iova);
if (rc) {
- gvt_err("gvt_dma_map_iova failed for gfn: 0x%lx\n", gfn);
+ gvt_vgpu_err("gvt_dma_map_iova failed for gfn: 0x%lx\n", gfn);
vfio_unpin_pages(dev, &gfn, 1);
return INTEL_GVT_INVALID_ADDR;
}
@@ -1417,7 +1423,7 @@ static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa,
{
struct kvmgt_guest_info *info;
struct kvm *kvm;
- int ret;
+ int idx, ret;
bool kthread = current->mm == NULL;
if (!handle_valid(handle))
@@ -1429,8 +1435,10 @@ static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa,
if (kthread)
use_mm(kvm->mm);
+ idx = srcu_read_lock(&kvm->srcu);
ret = write ? kvm_write_guest(kvm, gpa, buf, len) :
kvm_read_guest(kvm, gpa, buf, len);
+ srcu_read_unlock(&kvm->srcu, idx);
if (kthread)
unuse_mm(kvm->mm);
diff --git a/drivers/gpu/drm/i915/gvt/mmio.c b/drivers/gpu/drm/i915/gvt/mmio.c
index 4df078bc5d04..1ba3bdb09341 100644
--- a/drivers/gpu/drm/i915/gvt/mmio.c
+++ b/drivers/gpu/drm/i915/gvt/mmio.c
@@ -57,6 +57,58 @@ int intel_vgpu_gpa_to_mmio_offset(struct intel_vgpu *vgpu, u64 gpa)
(reg >= gvt->device_info.gtt_start_offset \
&& reg < gvt->device_info.gtt_start_offset + gvt_ggtt_sz(gvt))
+static void failsafe_emulate_mmio_rw(struct intel_vgpu *vgpu, uint64_t pa,
+ void *p_data, unsigned int bytes, bool read)
+{
+ struct intel_gvt *gvt = NULL;
+ void *pt = NULL;
+ unsigned int offset = 0;
+
+ if (!vgpu || !p_data)
+ return;
+
+ gvt = vgpu->gvt;
+ mutex_lock(&gvt->lock);
+ offset = intel_vgpu_gpa_to_mmio_offset(vgpu, pa);
+ if (reg_is_mmio(gvt, offset)) {
+ if (read)
+ intel_vgpu_default_mmio_read(vgpu, offset, p_data,
+ bytes);
+ else
+ intel_vgpu_default_mmio_write(vgpu, offset, p_data,
+ bytes);
+ } else if (reg_is_gtt(gvt, offset) &&
+ vgpu->gtt.ggtt_mm->virtual_page_table) {
+ offset -= gvt->device_info.gtt_start_offset;
+ pt = vgpu->gtt.ggtt_mm->virtual_page_table + offset;
+ if (read)
+ memcpy(p_data, pt, bytes);
+ else
+ memcpy(pt, p_data, bytes);
+
+ } else if (atomic_read(&vgpu->gtt.n_write_protected_guest_page)) {
+ struct intel_vgpu_guest_page *gp;
+
+ /* Since we enter the failsafe mode early during guest boot,
+ * guest may not have chance to set up its ppgtt table, so
+ * there should not be any wp pages for guest. Keep the wp
+ * related code here in case we need to handle it in furture.
+ */
+ gp = intel_vgpu_find_guest_page(vgpu, pa >> PAGE_SHIFT);
+ if (gp) {
+ /* remove write protection to prevent furture traps */
+ intel_vgpu_clean_guest_page(vgpu, gp);
+ if (read)
+ intel_gvt_hypervisor_read_gpa(vgpu, pa,
+ p_data, bytes);
+ else
+ intel_gvt_hypervisor_write_gpa(vgpu, pa,
+ p_data, bytes);
+ }
+ }
+ mutex_unlock(&gvt->lock);
+}
+
/**
* intel_vgpu_emulate_mmio_read - emulate MMIO read
* @vgpu: a vGPU
@@ -75,6 +127,11 @@ int intel_vgpu_emulate_mmio_read(struct intel_vgpu *vgpu, uint64_t pa,
unsigned int offset = 0;
int ret = -EINVAL;
+
+ if (vgpu->failsafe) {
+ failsafe_emulate_mmio_rw(vgpu, pa, p_data, bytes, true);
+ return 0;
+ }
mutex_lock(&gvt->lock);
if (atomic_read(&vgpu->gtt.n_write_protected_guest_page)) {
@@ -85,10 +142,10 @@ int intel_vgpu_emulate_mmio_read(struct intel_vgpu *vgpu, uint64_t pa,
ret = intel_gvt_hypervisor_read_gpa(vgpu, pa,
p_data, bytes);
if (ret) {
- gvt_err("vgpu%d: guest page read error %d, "
+ gvt_vgpu_err("guest page read error %d, "
"gfn 0x%lx, pa 0x%llx, var 0x%x, len %d\n",
- vgpu->id, ret,
- gp->gfn, pa, *(u32 *)p_data, bytes);
+ ret, gp->gfn, pa, *(u32 *)p_data,
+ bytes);
}
mutex_unlock(&gvt->lock);
return ret;
@@ -143,14 +200,13 @@ int intel_vgpu_emulate_mmio_read(struct intel_vgpu *vgpu, uint64_t pa,
ret = intel_vgpu_default_mmio_read(vgpu, offset, p_data, bytes);
if (!vgpu->mmio.disable_warn_untrack) {
- gvt_err("vgpu%d: read untracked MMIO %x(%dB) val %x\n",
- vgpu->id, offset, bytes, *(u32 *)p_data);
+ gvt_vgpu_err("read untracked MMIO %x(%dB) val %x\n",
+ offset, bytes, *(u32 *)p_data);
if (offset == 0x206c) {
- gvt_err("------------------------------------------\n");
- gvt_err("vgpu%d: likely triggers a gfx reset\n",
- vgpu->id);
- gvt_err("------------------------------------------\n");
+ gvt_vgpu_err("------------------------------------------\n");
+ gvt_vgpu_err("likely triggers a gfx reset\n");
+ gvt_vgpu_err("------------------------------------------\n");
vgpu->mmio.disable_warn_untrack = true;
}
}
@@ -163,8 +219,8 @@ int intel_vgpu_emulate_mmio_read(struct intel_vgpu *vgpu, uint64_t pa,
mutex_unlock(&gvt->lock);
return 0;
err:
- gvt_err("vgpu%d: fail to emulate MMIO read %08x len %d\n",
- vgpu->id, offset, bytes);
+ gvt_vgpu_err("fail to emulate MMIO read %08x len %d\n",
+ offset, bytes);
mutex_unlock(&gvt->lock);
return ret;
}
@@ -188,6 +244,11 @@ int intel_vgpu_emulate_mmio_write(struct intel_vgpu *vgpu, uint64_t pa,
u32 old_vreg = 0, old_sreg = 0;
int ret = -EINVAL;
+ if (vgpu->failsafe) {
+ failsafe_emulate_mmio_rw(vgpu, pa, p_data, bytes, false);
+ return 0;
+ }
+
mutex_lock(&gvt->lock);
if (atomic_read(&vgpu->gtt.n_write_protected_guest_page)) {
@@ -197,10 +258,11 @@ int intel_vgpu_emulate_mmio_write(struct intel_vgpu *vgpu, uint64_t pa,
if (gp) {
ret = gp->handler(gp, pa, p_data, bytes);
if (ret) {
- gvt_err("vgpu%d: guest page write error %d, "
- "gfn 0x%lx, pa 0x%llx, var 0x%x, len %d\n",
- vgpu->id, ret,
- gp->gfn, pa, *(u32 *)p_data, bytes);
+ gvt_err("guest page write error %d, "
+ "gfn 0x%lx, pa 0x%llx, "
+ "var 0x%x, len %d\n",
+ ret, gp->gfn, pa,
+ *(u32 *)p_data, bytes);
}
mutex_unlock(&gvt->lock);
return ret;
@@ -236,7 +298,7 @@ int intel_vgpu_emulate_mmio_write(struct intel_vgpu *vgpu, uint64_t pa,
mmio = intel_gvt_find_mmio_info(gvt, rounddown(offset, 4));
if (!mmio && !vgpu->mmio.disable_warn_untrack)
- gvt_err("vgpu%d: write untracked MMIO %x len %d val %x\n",
+ gvt_dbg_mmio("vgpu%d: write untracked MMIO %x len %d val %x\n",
vgpu->id, offset, bytes, *(u32 *)p_data);
if (!intel_gvt_mmio_is_unalign(gvt, offset)) {
@@ -267,8 +329,8 @@ int intel_vgpu_emulate_mmio_write(struct intel_vgpu *vgpu, uint64_t pa,
/* all register bits are RO. */
if (ro_mask == ~(u64)0) {
- gvt_err("vgpu%d: try to write RO reg %x\n",
- vgpu->id, offset);
+ gvt_vgpu_err("try to write RO reg %x\n",
+ offset);
ret = 0;
goto out;
}
@@ -298,8 +360,8 @@ out:
mutex_unlock(&gvt->lock);
return 0;
err:
- gvt_err("vgpu%d: fail to emulate MMIO write %08x len %d\n",
- vgpu->id, offset, bytes);
+ gvt_vgpu_err("fail to emulate MMIO write %08x len %d\n", offset,
+ bytes);
mutex_unlock(&gvt->lock);
return ret;
}
@@ -322,6 +384,8 @@ void intel_vgpu_reset_mmio(struct intel_vgpu *vgpu)
/* set the bit 0:2(Core C-State ) to C0 */
vgpu_vreg(vgpu, GEN6_GT_CORE_STATUS) = 0;
+
+ vgpu->mmio.disable_warn_untrack = false;
}
/**
diff --git a/drivers/gpu/drm/i915/gvt/mmio.h b/drivers/gpu/drm/i915/gvt/mmio.h
index 3bc620f56f35..a3a027025cd0 100644
--- a/drivers/gpu/drm/i915/gvt/mmio.h
+++ b/drivers/gpu/drm/i915/gvt/mmio.h
@@ -107,4 +107,7 @@ int intel_vgpu_default_mmio_read(struct intel_vgpu *vgpu, unsigned int offset,
void *p_data, unsigned int bytes);
int intel_vgpu_default_mmio_write(struct intel_vgpu *vgpu, unsigned int offset,
void *p_data, unsigned int bytes);
+
+bool intel_gvt_in_force_nonpriv_whitelist(struct intel_gvt *gvt,
+ unsigned int offset);
#endif
diff --git a/drivers/gpu/drm/i915/gvt/opregion.c b/drivers/gpu/drm/i915/gvt/opregion.c
index d9fb41ab7119..311799136d7f 100644
--- a/drivers/gpu/drm/i915/gvt/opregion.c
+++ b/drivers/gpu/drm/i915/gvt/opregion.c
@@ -27,7 +27,6 @@
static int init_vgpu_opregion(struct intel_vgpu *vgpu, u32 gpa)
{
- void __iomem *host_va = vgpu->gvt->opregion.opregion_va;
u8 *buf;
int i;
@@ -43,8 +42,8 @@ static int init_vgpu_opregion(struct intel_vgpu *vgpu, u32 gpa)
if (!vgpu_opregion(vgpu)->va)
return -ENOMEM;
- memcpy_fromio(vgpu_opregion(vgpu)->va, host_va,
- INTEL_GVT_OPREGION_SIZE);
+ memcpy(vgpu_opregion(vgpu)->va, vgpu->gvt->opregion.opregion_va,
+ INTEL_GVT_OPREGION_SIZE);
for (i = 0; i < INTEL_GVT_OPREGION_PAGES; i++)
vgpu_opregion(vgpu)->gfn[i] = (gpa >> PAGE_SHIFT) + i;
@@ -68,14 +67,15 @@ static int map_vgpu_opregion(struct intel_vgpu *vgpu, bool map)
mfn = intel_gvt_hypervisor_virt_to_mfn(vgpu_opregion(vgpu)->va
+ i * PAGE_SIZE);
if (mfn == INTEL_GVT_INVALID_ADDR) {
- gvt_err("fail to get MFN from VA\n");
+ gvt_vgpu_err("fail to get MFN from VA\n");
return -EINVAL;
}
ret = intel_gvt_hypervisor_map_gfn_to_mfn(vgpu,
vgpu_opregion(vgpu)->gfn[i],
mfn, 1, map);
if (ret) {
- gvt_err("fail to map GFN to MFN, errno: %d\n", ret);
+ gvt_vgpu_err("fail to map GFN to MFN, errno: %d\n",
+ ret);
return ret;
}
}
@@ -288,7 +288,7 @@ int intel_vgpu_emulate_opregion_request(struct intel_vgpu *vgpu, u32 swsci)
parm = vgpu_opregion(vgpu)->va + INTEL_GVT_OPREGION_PARM;
if (!(swsci & SWSCI_SCI_SELECT)) {
- gvt_err("vgpu%d: requesting SMI service\n", vgpu->id);
+ gvt_vgpu_err("requesting SMI service\n");
return 0;
}
/* ignore non 0->1 trasitions */
@@ -301,9 +301,8 @@ int intel_vgpu_emulate_opregion_request(struct intel_vgpu *vgpu, u32 swsci)
func = GVT_OPREGION_FUNC(*scic);
subfunc = GVT_OPREGION_SUBFUNC(*scic);
if (!querying_capabilities(*scic)) {
- gvt_err("vgpu%d: requesting runtime service: func \"%s\","
+ gvt_vgpu_err("requesting runtime service: func \"%s\","
" subfunc \"%s\"\n",
- vgpu->id,
opregion_func_name(func),
opregion_subfunc_name(subfunc));
/*
diff --git a/drivers/gpu/drm/i915/gvt/render.c b/drivers/gpu/drm/i915/gvt/render.c
index 2b3a642284b6..0beb83563b08 100644
--- a/drivers/gpu/drm/i915/gvt/render.c
+++ b/drivers/gpu/drm/i915/gvt/render.c
@@ -53,6 +53,14 @@ static struct render_mmio gen8_render_mmio_list[] = {
{RCS, _MMIO(0x24d4), 0, false},
{RCS, _MMIO(0x24d8), 0, false},
{RCS, _MMIO(0x24dc), 0, false},
+ {RCS, _MMIO(0x24e0), 0, false},
+ {RCS, _MMIO(0x24e4), 0, false},
+ {RCS, _MMIO(0x24e8), 0, false},
+ {RCS, _MMIO(0x24ec), 0, false},
+ {RCS, _MMIO(0x24f0), 0, false},
+ {RCS, _MMIO(0x24f4), 0, false},
+ {RCS, _MMIO(0x24f8), 0, false},
+ {RCS, _MMIO(0x24fc), 0, false},
{RCS, _MMIO(0x7004), 0xffff, true},
{RCS, _MMIO(0x7008), 0xffff, true},
{RCS, _MMIO(0x7000), 0xffff, true},
@@ -76,6 +84,14 @@ static struct render_mmio gen9_render_mmio_list[] = {
{RCS, _MMIO(0x24d4), 0, false},
{RCS, _MMIO(0x24d8), 0, false},
{RCS, _MMIO(0x24dc), 0, false},
+ {RCS, _MMIO(0x24e0), 0, false},
+ {RCS, _MMIO(0x24e4), 0, false},
+ {RCS, _MMIO(0x24e8), 0, false},
+ {RCS, _MMIO(0x24ec), 0, false},
+ {RCS, _MMIO(0x24f0), 0, false},
+ {RCS, _MMIO(0x24f4), 0, false},
+ {RCS, _MMIO(0x24f8), 0, false},
+ {RCS, _MMIO(0x24fc), 0, false},
{RCS, _MMIO(0x7004), 0xffff, true},
{RCS, _MMIO(0x7008), 0xffff, true},
{RCS, _MMIO(0x7000), 0xffff, true},
@@ -151,7 +167,7 @@ static void handle_tlb_pending_event(struct intel_vgpu *vgpu, int ring_id)
I915_WRITE_FW(reg, 0x1);
if (wait_for_atomic((I915_READ_FW(reg) == 0), 50))
- gvt_err("timeout in invalidate ring (%d) tlb\n", ring_id);
+ gvt_vgpu_err("timeout in invalidate ring (%d) tlb\n", ring_id);
else
vgpu_vreg(vgpu, regs[ring_id]) = 0;
@@ -191,7 +207,7 @@ static void load_mocs(struct intel_vgpu *vgpu, int ring_id)
l3_offset.reg = 0xb020;
for (i = 0; i < 32; i++) {
gen9_render_mocs_L3[i] = I915_READ(l3_offset);
- I915_WRITE(l3_offset, vgpu_vreg(vgpu, offset));
+ I915_WRITE(l3_offset, vgpu_vreg(vgpu, l3_offset));
POSTING_READ(l3_offset);
l3_offset.reg += 4;
}
diff --git a/drivers/gpu/drm/i915/gvt/sched_policy.c b/drivers/gpu/drm/i915/gvt/sched_policy.c
index 06c9584ac5f0..34b9acdf3479 100644
--- a/drivers/gpu/drm/i915/gvt/sched_policy.c
+++ b/drivers/gpu/drm/i915/gvt/sched_policy.c
@@ -101,7 +101,7 @@ struct tbs_sched_data {
struct list_head runq_head;
};
-#define GVT_DEFAULT_TIME_SLICE (1 * HZ / 1000)
+#define GVT_DEFAULT_TIME_SLICE (msecs_to_jiffies(1))
static void tbs_sched_func(struct work_struct *work)
{
@@ -223,7 +223,7 @@ static void tbs_sched_start_schedule(struct intel_vgpu *vgpu)
return;
list_add_tail(&vgpu_data->list, &sched_data->runq_head);
- schedule_delayed_work(&sched_data->work, sched_data->period);
+ schedule_delayed_work(&sched_data->work, 0);
}
static void tbs_sched_stop_schedule(struct intel_vgpu *vgpu)
diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c
index d6b6d0efdd1a..a44782412f2c 100644
--- a/drivers/gpu/drm/i915/gvt/scheduler.c
+++ b/drivers/gpu/drm/i915/gvt/scheduler.c
@@ -84,7 +84,7 @@ static int populate_shadow_context(struct intel_vgpu_workload *workload)
(u32)((workload->ctx_desc.lrca + i) <<
GTT_PAGE_SHIFT));
if (context_gpa == INTEL_GVT_INVALID_ADDR) {
- gvt_err("Invalid guest context descriptor\n");
+ gvt_vgpu_err("Invalid guest context descriptor\n");
return -EINVAL;
}
@@ -127,18 +127,24 @@ static int populate_shadow_context(struct intel_vgpu_workload *workload)
return 0;
}
+static inline bool is_gvt_request(struct drm_i915_gem_request *req)
+{
+ return i915_gem_context_force_single_submission(req->ctx);
+}
+
static int shadow_context_status_change(struct notifier_block *nb,
unsigned long action, void *data)
{
- struct intel_vgpu *vgpu = container_of(nb,
- struct intel_vgpu, shadow_ctx_notifier_block);
- struct drm_i915_gem_request *req =
- (struct drm_i915_gem_request *)data;
- struct intel_gvt_workload_scheduler *scheduler =
- &vgpu->gvt->scheduler;
+ struct drm_i915_gem_request *req = (struct drm_i915_gem_request *)data;
+ struct intel_gvt *gvt = container_of(nb, struct intel_gvt,
+ shadow_ctx_notifier_block[req->engine->id]);
+ struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
struct intel_vgpu_workload *workload =
scheduler->current_workload[req->engine->id];
+ if (!is_gvt_request(req) || unlikely(!workload))
+ return NOTIFY_OK;
+
switch (action) {
case INTEL_CONTEXT_SCHEDULE_IN:
intel_gvt_load_render_mmio(workload->vgpu,
@@ -148,6 +154,15 @@ static int shadow_context_status_change(struct notifier_block *nb,
case INTEL_CONTEXT_SCHEDULE_OUT:
intel_gvt_restore_render_mmio(workload->vgpu,
workload->ring_id);
+ /* If the status is -EINPROGRESS means this workload
+ * doesn't meet any issue during dispatching so when
+ * get the SCHEDULE_OUT set the status to be zero for
+ * good. If the status is NOT -EINPROGRESS means there
+ * is something wrong happened during dispatching and
+ * the status should not be set to zero
+ */
+ if (workload->status == -EINPROGRESS)
+ workload->status = 0;
atomic_set(&workload->shadow_ctx_active, 0);
break;
default:
@@ -163,7 +178,9 @@ static int dispatch_workload(struct intel_vgpu_workload *workload)
int ring_id = workload->ring_id;
struct i915_gem_context *shadow_ctx = workload->vgpu->shadow_ctx;
struct drm_i915_private *dev_priv = workload->vgpu->gvt->dev_priv;
+ struct intel_engine_cs *engine = dev_priv->engine[ring_id];
struct drm_i915_gem_request *rq;
+ struct intel_vgpu *vgpu = workload->vgpu;
int ret;
gvt_dbg_sched("ring id %d prepare to dispatch workload %p\n",
@@ -175,9 +192,24 @@ static int dispatch_workload(struct intel_vgpu_workload *workload)
mutex_lock(&dev_priv->drm.struct_mutex);
+ /* pin shadow context by gvt even the shadow context will be pinned
+ * when i915 alloc request. That is because gvt will update the guest
+ * context from shadow context when workload is completed, and at that
+ * moment, i915 may already unpined the shadow context to make the
+ * shadow_ctx pages invalid. So gvt need to pin itself. After update
+ * the guest context, gvt can unpin the shadow_ctx safely.
+ */
+ ret = engine->context_pin(engine, shadow_ctx);
+ if (ret) {
+ gvt_vgpu_err("fail to pin shadow context\n");
+ workload->status = ret;
+ mutex_unlock(&dev_priv->drm.struct_mutex);
+ return ret;
+ }
+
rq = i915_gem_request_alloc(dev_priv->engine[ring_id], shadow_ctx);
if (IS_ERR(rq)) {
- gvt_err("fail to allocate gem request\n");
+ gvt_vgpu_err("fail to allocate gem request\n");
ret = PTR_ERR(rq);
goto out;
}
@@ -190,9 +222,12 @@ static int dispatch_workload(struct intel_vgpu_workload *workload)
if (ret)
goto out;
- ret = intel_gvt_scan_and_shadow_wa_ctx(&workload->wa_ctx);
- if (ret)
- goto out;
+ if ((workload->ring_id == RCS) &&
+ (workload->wa_ctx.indirect_ctx.size != 0)) {
+ ret = intel_gvt_scan_and_shadow_wa_ctx(&workload->wa_ctx);
+ if (ret)
+ goto out;
+ }
ret = populate_shadow_context(workload);
if (ret)
@@ -215,6 +250,9 @@ out:
if (!IS_ERR_OR_NULL(rq))
i915_add_request_no_flush(rq);
+ else
+ engine->context_unpin(engine, shadow_ctx);
+
mutex_unlock(&dev_priv->drm.struct_mutex);
return ret;
}
@@ -310,7 +348,7 @@ static void update_guest_context(struct intel_vgpu_workload *workload)
(u32)((workload->ctx_desc.lrca + i) <<
GTT_PAGE_SHIFT));
if (context_gpa == INTEL_GVT_INVALID_ADDR) {
- gvt_err("invalid guest context descriptor\n");
+ gvt_vgpu_err("invalid guest context descriptor\n");
return;
}
@@ -359,15 +397,31 @@ static void complete_current_workload(struct intel_gvt *gvt, int ring_id)
workload = scheduler->current_workload[ring_id];
vgpu = workload->vgpu;
- if (!workload->status && !vgpu->resetting) {
+ /* For the workload w/ request, needs to wait for the context
+ * switch to make sure request is completed.
+ * For the workload w/o request, directly complete the workload.
+ */
+ if (workload->req) {
+ struct drm_i915_private *dev_priv =
+ workload->vgpu->gvt->dev_priv;
+ struct intel_engine_cs *engine =
+ dev_priv->engine[workload->ring_id];
wait_event(workload->shadow_ctx_status_wq,
!atomic_read(&workload->shadow_ctx_active));
- update_guest_context(workload);
+ i915_gem_request_put(fetch_and_zero(&workload->req));
+
+ if (!workload->status && !vgpu->resetting) {
+ update_guest_context(workload);
- for_each_set_bit(event, workload->pending_events,
- INTEL_GVT_EVENT_MAX)
- intel_vgpu_trigger_virtual_event(vgpu, event);
+ for_each_set_bit(event, workload->pending_events,
+ INTEL_GVT_EVENT_MAX)
+ intel_vgpu_trigger_virtual_event(vgpu, event);
+ }
+ mutex_lock(&dev_priv->drm.struct_mutex);
+ /* unpin shadow ctx as the shadow_ctx update is done */
+ engine->context_unpin(engine, workload->vgpu->shadow_ctx);
+ mutex_unlock(&dev_priv->drm.struct_mutex);
}
gvt_dbg_sched("ring id %d complete workload %p status %d\n",
@@ -397,7 +451,7 @@ static int workload_thread(void *priv)
int ring_id = p->ring_id;
struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
struct intel_vgpu_workload *workload = NULL;
- long lret;
+ struct intel_vgpu *vgpu = NULL;
int ret;
bool need_force_wake = IS_SKYLAKE(gvt->dev_priv);
DEFINE_WAIT_FUNC(wait, woken_wake_function);
@@ -440,29 +494,19 @@ static int workload_thread(void *priv)
mutex_unlock(&gvt->lock);
if (ret) {
- gvt_err("fail to dispatch workload, skip\n");
+ vgpu = workload->vgpu;
+ gvt_vgpu_err("fail to dispatch workload, skip\n");
goto complete;
}
gvt_dbg_sched("ring id %d wait workload %p\n",
workload->ring_id, workload);
-
- lret = i915_wait_request(workload->req,
- 0, MAX_SCHEDULE_TIMEOUT);
- if (lret < 0) {
- workload->status = lret;
- gvt_err("fail to wait workload, skip\n");
- } else {
- workload->status = 0;
- }
+ i915_wait_request(workload->req, 0, MAX_SCHEDULE_TIMEOUT);
complete:
gvt_dbg_sched("will complete workload %p, status: %d\n",
workload, workload->status);
- if (workload->req)
- i915_gem_request_put(fetch_and_zero(&workload->req));
-
complete_current_workload(gvt, ring_id);
if (need_force_wake)
@@ -493,15 +537,16 @@ void intel_gvt_wait_vgpu_idle(struct intel_vgpu *vgpu)
void intel_gvt_clean_workload_scheduler(struct intel_gvt *gvt)
{
struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
- int i;
+ struct intel_engine_cs *engine;
+ enum intel_engine_id i;
gvt_dbg_core("clean workload scheduler\n");
- for (i = 0; i < I915_NUM_ENGINES; i++) {
- if (scheduler->thread[i]) {
- kthread_stop(scheduler->thread[i]);
- scheduler->thread[i] = NULL;
- }
+ for_each_engine(engine, gvt->dev_priv, i) {
+ atomic_notifier_chain_unregister(
+ &engine->context_status_notifier,
+ &gvt->shadow_ctx_notifier_block[i]);
+ kthread_stop(scheduler->thread[i]);
}
}
@@ -509,18 +554,15 @@ int intel_gvt_init_workload_scheduler(struct intel_gvt *gvt)
{
struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
struct workload_thread_param *param = NULL;
+ struct intel_engine_cs *engine;
+ enum intel_engine_id i;
int ret;
- int i;
gvt_dbg_core("init workload scheduler\n");
init_waitqueue_head(&scheduler->workload_complete_wq);
- for (i = 0; i < I915_NUM_ENGINES; i++) {
- /* check ring mask at init time */
- if (!HAS_ENGINE(gvt->dev_priv, i))
- continue;
-
+ for_each_engine(engine, gvt->dev_priv, i) {
init_waitqueue_head(&scheduler->waitq[i]);
param = kzalloc(sizeof(*param), GFP_KERNEL);
@@ -539,6 +581,11 @@ int intel_gvt_init_workload_scheduler(struct intel_gvt *gvt)
ret = PTR_ERR(scheduler->thread[i]);
goto err;
}
+
+ gvt->shadow_ctx_notifier_block[i].notifier_call =
+ shadow_context_status_change;
+ atomic_notifier_chain_register(&engine->context_status_notifier,
+ &gvt->shadow_ctx_notifier_block[i]);
}
return 0;
err:
@@ -550,9 +597,6 @@ err:
void intel_vgpu_clean_gvt_context(struct intel_vgpu *vgpu)
{
- atomic_notifier_chain_unregister(&vgpu->shadow_ctx->status_notifier,
- &vgpu->shadow_ctx_notifier_block);
-
i915_gem_context_put_unlocked(vgpu->shadow_ctx);
}
@@ -567,10 +611,5 @@ int intel_vgpu_init_gvt_context(struct intel_vgpu *vgpu)
vgpu->shadow_ctx->engine[RCS].initialised = true;
- vgpu->shadow_ctx_notifier_block.notifier_call =
- shadow_context_status_change;
-
- atomic_notifier_chain_register(&vgpu->shadow_ctx->status_notifier,
- &vgpu->shadow_ctx_notifier_block);
return 0;
}
diff --git a/drivers/gpu/drm/i915/gvt/vgpu.c b/drivers/gpu/drm/i915/gvt/vgpu.c
index 95a97aa0051e..649ef280cc9a 100644
--- a/drivers/gpu/drm/i915/gvt/vgpu.c
+++ b/drivers/gpu/drm/i915/gvt/vgpu.c
@@ -64,6 +64,20 @@ void populate_pvinfo_page(struct intel_vgpu *vgpu)
WARN_ON(sizeof(struct vgt_if) != VGT_PVINFO_SIZE);
}
+static struct {
+ unsigned int low_mm;
+ unsigned int high_mm;
+ unsigned int fence;
+ enum intel_vgpu_edid edid;
+ char *name;
+} vgpu_types[] = {
+/* Fixed vGPU type table */
+ { MB_TO_BYTES(64), MB_TO_BYTES(384), 4, GVT_EDID_1024_768, "8" },
+ { MB_TO_BYTES(128), MB_TO_BYTES(512), 4, GVT_EDID_1920_1200, "4" },
+ { MB_TO_BYTES(256), MB_TO_BYTES(1024), 4, GVT_EDID_1920_1200, "2" },
+ { MB_TO_BYTES(512), MB_TO_BYTES(2048), 4, GVT_EDID_1920_1200, "1" },
+};
+
/**
* intel_gvt_init_vgpu_types - initialize vGPU type list
* @gvt : GVT device
@@ -78,9 +92,8 @@ int intel_gvt_init_vgpu_types(struct intel_gvt *gvt)
unsigned int min_low;
/* vGPU type name is defined as GVTg_Vx_y which contains
- * physical GPU generation type and 'y' means maximum vGPU
- * instances user can create on one physical GPU for this
- * type.
+ * physical GPU generation type (e.g V4 as BDW server, V5 as
+ * SKL server).
*
* Depend on physical SKU resource, might see vGPU types like
* GVTg_V4_8, GVTg_V4_4, GVTg_V4_2, etc. We can create
@@ -92,7 +105,7 @@ int intel_gvt_init_vgpu_types(struct intel_gvt *gvt)
*/
low_avail = gvt_aperture_sz(gvt) - HOST_LOW_GM_SIZE;
high_avail = gvt_hidden_sz(gvt) - HOST_HIGH_GM_SIZE;
- num_types = 4;
+ num_types = sizeof(vgpu_types) / sizeof(vgpu_types[0]);
gvt->types = kzalloc(num_types * sizeof(struct intel_vgpu_type),
GFP_KERNEL);
@@ -101,28 +114,29 @@ int intel_gvt_init_vgpu_types(struct intel_gvt *gvt)
min_low = MB_TO_BYTES(32);
for (i = 0; i < num_types; ++i) {
- if (low_avail / min_low == 0)
+ if (low_avail / vgpu_types[i].low_mm == 0)
break;
- gvt->types[i].low_gm_size = min_low;
- gvt->types[i].high_gm_size = max((min_low<<3), MB_TO_BYTES(384U));
- gvt->types[i].fence = 4;
- gvt->types[i].max_instance = min(low_avail / min_low,
- high_avail / gvt->types[i].high_gm_size);
- gvt->types[i].avail_instance = gvt->types[i].max_instance;
+
+ gvt->types[i].low_gm_size = vgpu_types[i].low_mm;
+ gvt->types[i].high_gm_size = vgpu_types[i].high_mm;
+ gvt->types[i].fence = vgpu_types[i].fence;
+ gvt->types[i].resolution = vgpu_types[i].edid;
+ gvt->types[i].avail_instance = min(low_avail / vgpu_types[i].low_mm,
+ high_avail / vgpu_types[i].high_mm);
if (IS_GEN8(gvt->dev_priv))
- sprintf(gvt->types[i].name, "GVTg_V4_%u",
- gvt->types[i].max_instance);
+ sprintf(gvt->types[i].name, "GVTg_V4_%s",
+ vgpu_types[i].name);
else if (IS_GEN9(gvt->dev_priv))
- sprintf(gvt->types[i].name, "GVTg_V5_%u",
- gvt->types[i].max_instance);
+ sprintf(gvt->types[i].name, "GVTg_V5_%s",
+ vgpu_types[i].name);
- min_low <<= 1;
- gvt_dbg_core("type[%d]: %s max %u avail %u low %u high %u fence %u\n",
- i, gvt->types[i].name, gvt->types[i].max_instance,
+ gvt_dbg_core("type[%d]: %s avail %u low %u high %u fence %u res %s\n",
+ i, gvt->types[i].name,
gvt->types[i].avail_instance,
gvt->types[i].low_gm_size,
- gvt->types[i].high_gm_size, gvt->types[i].fence);
+ gvt->types[i].high_gm_size, gvt->types[i].fence,
+ vgpu_edid_str(gvt->types[i].resolution));
}
gvt->num_types = i;
@@ -138,7 +152,7 @@ static void intel_gvt_update_vgpu_types(struct intel_gvt *gvt)
{
int i;
unsigned int low_gm_avail, high_gm_avail, fence_avail;
- unsigned int low_gm_min, high_gm_min, fence_min, total_min;
+ unsigned int low_gm_min, high_gm_min, fence_min;
/* Need to depend on maxium hw resource size but keep on
* static config for now.
@@ -154,32 +168,45 @@ static void intel_gvt_update_vgpu_types(struct intel_gvt *gvt)
low_gm_min = low_gm_avail / gvt->types[i].low_gm_size;
high_gm_min = high_gm_avail / gvt->types[i].high_gm_size;
fence_min = fence_avail / gvt->types[i].fence;
- total_min = min(min(low_gm_min, high_gm_min), fence_min);
- gvt->types[i].avail_instance = min(gvt->types[i].max_instance,
- total_min);
+ gvt->types[i].avail_instance = min(min(low_gm_min, high_gm_min),
+ fence_min);
- gvt_dbg_core("update type[%d]: %s max %u avail %u low %u high %u fence %u\n",
- i, gvt->types[i].name, gvt->types[i].max_instance,
+ gvt_dbg_core("update type[%d]: %s avail %u low %u high %u fence %u\n",
+ i, gvt->types[i].name,
gvt->types[i].avail_instance, gvt->types[i].low_gm_size,
gvt->types[i].high_gm_size, gvt->types[i].fence);
}
}
/**
- * intel_gvt_destroy_vgpu - destroy a virtual GPU
+ * intel_gvt_active_vgpu - activate a virtual GPU
* @vgpu: virtual GPU
*
- * This function is called when user wants to destroy a virtual GPU.
+ * This function is called when user wants to activate a virtual GPU.
*
*/
-void intel_gvt_destroy_vgpu(struct intel_vgpu *vgpu)
+void intel_gvt_activate_vgpu(struct intel_vgpu *vgpu)
+{
+ mutex_lock(&vgpu->gvt->lock);
+ vgpu->active = true;
+ mutex_unlock(&vgpu->gvt->lock);
+}
+
+/**
+ * intel_gvt_deactive_vgpu - deactivate a virtual GPU
+ * @vgpu: virtual GPU
+ *
+ * This function is called when user wants to deactivate a virtual GPU.
+ * All virtual GPU runtime information will be destroyed.
+ *
+ */
+void intel_gvt_deactivate_vgpu(struct intel_vgpu *vgpu)
{
struct intel_gvt *gvt = vgpu->gvt;
mutex_lock(&gvt->lock);
vgpu->active = false;
- idr_remove(&gvt->vgpu_idr, vgpu->id);
if (atomic_read(&vgpu->running_workload_num)) {
mutex_unlock(&gvt->lock);
@@ -188,6 +215,26 @@ void intel_gvt_destroy_vgpu(struct intel_vgpu *vgpu)
}
intel_vgpu_stop_schedule(vgpu);
+
+ mutex_unlock(&gvt->lock);
+}
+
+/**
+ * intel_gvt_destroy_vgpu - destroy a virtual GPU
+ * @vgpu: virtual GPU
+ *
+ * This function is called when user wants to destroy a virtual GPU.
+ *
+ */
+void intel_gvt_destroy_vgpu(struct intel_vgpu *vgpu)
+{
+ struct intel_gvt *gvt = vgpu->gvt;
+
+ mutex_lock(&gvt->lock);
+
+ WARN(vgpu->active, "vGPU is still active!\n");
+
+ idr_remove(&gvt->vgpu_idr, vgpu->id);
intel_vgpu_clean_sched_policy(vgpu);
intel_vgpu_clean_gvt_context(vgpu);
intel_vgpu_clean_execlist(vgpu);
@@ -248,7 +295,7 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
if (ret)
goto out_detach_hypervisor_vgpu;
- ret = intel_vgpu_init_display(vgpu);
+ ret = intel_vgpu_init_display(vgpu, param->resolution);
if (ret)
goto out_clean_gtt;
@@ -264,7 +311,6 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
if (ret)
goto out_clean_shadow_ctx;
- vgpu->active = true;
mutex_unlock(&gvt->lock);
return vgpu;
@@ -312,6 +358,7 @@ struct intel_vgpu *intel_gvt_create_vgpu(struct intel_gvt *gvt,
param.low_gm_sz = type->low_gm_size;
param.high_gm_sz = type->high_gm_size;
param.fence_sz = type->fence;
+ param.resolution = type->resolution;
/* XXX current param based on MB */
param.low_gm_sz = BYTES_TO_MB(param.low_gm_sz);
@@ -387,8 +434,12 @@ void intel_gvt_reset_vgpu_locked(struct intel_vgpu *vgpu, bool dmlr,
populate_pvinfo_page(vgpu);
intel_vgpu_reset_display(vgpu);
- if (dmlr)
+ if (dmlr) {
intel_vgpu_reset_cfg_space(vgpu);
+ /* only reset the failsafe mode when dmlr reset */
+ vgpu->failsafe = false;
+ vgpu->pv_notified = false;
+ }
}
vgpu->resetting = false;
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index e703556eba99..5c089b3c2a7e 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -248,6 +248,7 @@ static int i915_getparam(struct drm_device *dev, void *data,
case I915_PARAM_IRQ_ACTIVE:
case I915_PARAM_ALLOW_BATCHBUFFER:
case I915_PARAM_LAST_DISPATCH:
+ case I915_PARAM_HAS_EXEC_CONSTANTS:
/* Reject all old ums/dri params. */
return -ENODEV;
case I915_PARAM_CHIPSET_ID:
@@ -274,9 +275,6 @@ static int i915_getparam(struct drm_device *dev, void *data,
case I915_PARAM_HAS_BSD2:
value = !!dev_priv->engine[VCS2];
break;
- case I915_PARAM_HAS_EXEC_CONSTANTS:
- value = INTEL_GEN(dev_priv) >= 4;
- break;
case I915_PARAM_HAS_LLC:
value = HAS_LLC(dev_priv);
break;
@@ -1436,8 +1434,6 @@ static int i915_drm_suspend(struct drm_device *dev)
goto out;
}
- intel_guc_suspend(dev_priv);
-
intel_display_suspend(dev);
intel_dp_mst_suspend(dev);
@@ -1788,7 +1784,7 @@ void i915_reset(struct drm_i915_private *dev_priv)
goto error;
}
- i915_gem_reset_finish(dev_priv);
+ i915_gem_reset(dev_priv);
intel_overlay_reset(dev_priv);
/* Ok, now get things going again... */
@@ -1814,6 +1810,7 @@ void i915_reset(struct drm_i915_private *dev_priv)
i915_queue_hangcheck(dev_priv);
wakeup:
+ i915_gem_reset_finish(dev_priv);
enable_irq(dev_priv->drm.irq);
wake_up_bit(&error->flags, I915_RESET_IN_PROGRESS);
return;
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 0a4b42d31391..46fcd8b7080a 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -293,6 +293,7 @@ enum plane_id {
PLANE_PRIMARY,
PLANE_SPRITE0,
PLANE_SPRITE1,
+ PLANE_SPRITE2,
PLANE_CURSOR,
I915_MAX_PLANES,
};
@@ -805,6 +806,7 @@ struct intel_csr {
func(has_resource_streamer); \
func(has_runtime_pm); \
func(has_snoop); \
+ func(unfenced_needs_alignment); \
func(cursor_needs_physical); \
func(hws_needs_physical); \
func(overlay_needs_physical); \
@@ -1324,7 +1326,7 @@ struct intel_gen6_power_mgmt {
unsigned boosts;
/* manual wa residency calculations */
- struct intel_rps_ei up_ei, down_ei;
+ struct intel_rps_ei ei;
/*
* Protects RPS/RC6 register access and PCU communication.
@@ -2063,8 +2065,6 @@ struct drm_i915_private {
const struct intel_device_info info;
- int relative_constants_mode;
-
void __iomem *regs;
struct intel_uncore uncore;
@@ -3341,6 +3341,7 @@ static inline u32 i915_reset_count(struct i915_gpu_error *error)
}
int i915_gem_reset_prepare(struct drm_i915_private *dev_priv);
+void i915_gem_reset(struct drm_i915_private *dev_priv);
void i915_gem_reset_finish(struct drm_i915_private *dev_priv);
void i915_gem_set_wedged(struct drm_i915_private *dev_priv);
void i915_gem_clflush_object(struct drm_i915_gem_object *obj, bool force);
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 6908123162d1..fe531f904062 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1434,6 +1434,12 @@ i915_gem_pwrite_ioctl(struct drm_device *dev, void *data,
trace_i915_gem_object_pwrite(obj, args->offset, args->size);
+ ret = -ENODEV;
+ if (obj->ops->pwrite)
+ ret = obj->ops->pwrite(obj, args);
+ if (ret != -ENODEV)
+ goto err;
+
ret = i915_gem_object_wait(obj,
I915_WAIT_INTERRUPTIBLE |
I915_WAIT_ALL,
@@ -2119,6 +2125,7 @@ i915_gem_object_truncate(struct drm_i915_gem_object *obj)
*/
shmem_truncate_range(file_inode(obj->base.filp), 0, (loff_t)-1);
obj->mm.madv = __I915_MADV_PURGED;
+ obj->mm.pages = ERR_PTR(-EFAULT);
}
/* Try to discard unwanted pages */
@@ -2218,7 +2225,9 @@ void __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
__i915_gem_object_reset_page_iter(obj);
- obj->ops->put_pages(obj, pages);
+ if (!IS_ERR(pages))
+ obj->ops->put_pages(obj, pages);
+
unlock:
mutex_unlock(&obj->mm.lock);
}
@@ -2437,7 +2446,7 @@ int __i915_gem_object_get_pages(struct drm_i915_gem_object *obj)
if (err)
return err;
- if (unlikely(!obj->mm.pages)) {
+ if (unlikely(IS_ERR_OR_NULL(obj->mm.pages))) {
err = ____i915_gem_object_get_pages(obj);
if (err)
goto unlock;
@@ -2515,7 +2524,7 @@ void *i915_gem_object_pin_map(struct drm_i915_gem_object *obj,
pinned = true;
if (!atomic_inc_not_zero(&obj->mm.pages_pin_count)) {
- if (unlikely(!obj->mm.pages)) {
+ if (unlikely(IS_ERR_OR_NULL(obj->mm.pages))) {
ret = ____i915_gem_object_get_pages(obj);
if (ret)
goto err_unlock;
@@ -2563,6 +2572,75 @@ err_unlock:
goto out_unlock;
}
+static int
+i915_gem_object_pwrite_gtt(struct drm_i915_gem_object *obj,
+ const struct drm_i915_gem_pwrite *arg)
+{
+ struct address_space *mapping = obj->base.filp->f_mapping;
+ char __user *user_data = u64_to_user_ptr(arg->data_ptr);
+ u64 remain, offset;
+ unsigned int pg;
+
+ /* Before we instantiate/pin the backing store for our use, we
+ * can prepopulate the shmemfs filp efficiently using a write into
+ * the pagecache. We avoid the penalty of instantiating all the
+ * pages, important if the user is just writing to a few and never
+ * uses the object on the GPU, and using a direct write into shmemfs
+ * allows it to avoid the cost of retrieving a page (either swapin
+ * or clearing-before-use) before it is overwritten.
+ */
+ if (READ_ONCE(obj->mm.pages))
+ return -ENODEV;
+
+ /* Before the pages are instantiated the object is treated as being
+ * in the CPU domain. The pages will be clflushed as required before
+ * use, and we can freely write into the pages directly. If userspace
+ * races pwrite with any other operation; corruption will ensue -
+ * that is userspace's prerogative!
+ */
+
+ remain = arg->size;
+ offset = arg->offset;
+ pg = offset_in_page(offset);
+
+ do {
+ unsigned int len, unwritten;
+ struct page *page;
+ void *data, *vaddr;
+ int err;
+
+ len = PAGE_SIZE - pg;
+ if (len > remain)
+ len = remain;
+
+ err = pagecache_write_begin(obj->base.filp, mapping,
+ offset, len, 0,
+ &page, &data);
+ if (err < 0)
+ return err;
+
+ vaddr = kmap(page);
+ unwritten = copy_from_user(vaddr + pg, user_data, len);
+ kunmap(page);
+
+ err = pagecache_write_end(obj->base.filp, mapping,
+ offset, len, len - unwritten,
+ page, data);
+ if (err < 0)
+ return err;
+
+ if (unwritten)
+ return -EFAULT;
+
+ remain -= len;
+ user_data += len;
+ offset += len;
+ pg = 0;
+ } while (remain);
+
+ return 0;
+}
+
static bool ban_context(const struct i915_gem_context *ctx)
{
return (i915_gem_context_is_bannable(ctx) &&
@@ -2641,7 +2719,16 @@ int i915_gem_reset_prepare(struct drm_i915_private *dev_priv)
for_each_engine(engine, dev_priv, id) {
struct drm_i915_gem_request *request;
+ /* Prevent request submission to the hardware until we have
+ * completed the reset in i915_gem_reset_finish(). If a request
+ * is completed by one engine, it may then queue a request
+ * to a second via its engine->irq_tasklet *just* as we are
+ * calling engine->init_hw() and also writing the ELSP.
+ * Turning off the engine->irq_tasklet until the reset is over
+ * prevents the race.
+ */
tasklet_kill(&engine->irq_tasklet);
+ tasklet_disable(&engine->irq_tasklet);
if (engine_stalled(engine)) {
request = i915_gem_find_active_request(engine);
@@ -2756,7 +2843,7 @@ static void i915_gem_reset_engine(struct intel_engine_cs *engine)
engine->reset_hw(engine, request);
}
-void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
+void i915_gem_reset(struct drm_i915_private *dev_priv)
{
struct intel_engine_cs *engine;
enum intel_engine_id id;
@@ -2778,6 +2865,17 @@ void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
}
}
+void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
+{
+ struct intel_engine_cs *engine;
+ enum intel_engine_id id;
+
+ lockdep_assert_held(&dev_priv->drm.struct_mutex);
+
+ for_each_engine(engine, dev_priv, id)
+ tasklet_enable(&engine->irq_tasklet);
+}
+
static void nop_submit_request(struct drm_i915_gem_request *request)
{
dma_fence_set_error(&request->fence, -EIO);
@@ -3029,6 +3127,16 @@ i915_gem_wait_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
args->timeout_ns -= ktime_to_ns(ktime_sub(ktime_get(), start));
if (args->timeout_ns < 0)
args->timeout_ns = 0;
+
+ /*
+ * Apparently ktime isn't accurate enough and occasionally has a
+ * bit of mismatch in the jiffies<->nsecs<->ktime loop. So patch
+ * things up to make the test happy. We allow up to 1 jiffy.
+ *
+ * This is a regression from the timespec->ktime conversion.
+ */
+ if (ret == -ETIME && !nsecs_to_jiffies(args->timeout_ns))
+ args->timeout_ns = 0;
}
i915_gem_object_put(obj);
@@ -3974,8 +4082,11 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,
static const struct drm_i915_gem_object_ops i915_gem_object_ops = {
.flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE |
I915_GEM_OBJECT_IS_SHRINKABLE,
+
.get_pages = i915_gem_object_get_pages_gtt,
.put_pages = i915_gem_object_put_pages_gtt,
+
+ .pwrite = i915_gem_object_pwrite_gtt,
};
struct drm_i915_gem_object *
@@ -4237,6 +4348,8 @@ int i915_gem_suspend(struct drm_i915_private *dev_priv)
i915_gem_context_lost(dev_priv);
mutex_unlock(&dev->struct_mutex);
+ intel_guc_suspend(dev_priv);
+
cancel_delayed_work_sync(&dev_priv->gpu_error.hangcheck_work);
cancel_delayed_work_sync(&dev_priv->gt.retire_work);
@@ -4583,8 +4696,6 @@ i915_gem_load_init(struct drm_i915_private *dev_priv)
init_waitqueue_head(&dev_priv->gpu_error.wait_queue);
init_waitqueue_head(&dev_priv->gpu_error.reset_queue);
- dev_priv->relative_constants_mode = I915_EXEC_CONSTANTS_REL_GENERAL;
-
init_waitqueue_head(&dev_priv->pending_flip_queue);
dev_priv->mm.interruptible = true;
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index 17f90c618208..e2d83b6d376b 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -311,7 +311,6 @@ __create_hw_context(struct drm_i915_private *dev_priv,
ctx->ring_size = 4 * PAGE_SIZE;
ctx->desc_template = GEN8_CTX_ADDRESSING_MODE(dev_priv) <<
GEN8_CTX_ADDRESSING_MODE_SHIFT;
- ATOMIC_INIT_NOTIFIER_HEAD(&ctx->status_notifier);
/* GuC requires the ring to be placed above GUC_WOPCM_TOP. If GuC is not
* present or not in use we still need a small bias as ring wraparound
diff --git a/drivers/gpu/drm/i915/i915_gem_context.h b/drivers/gpu/drm/i915/i915_gem_context.h
index 0ac750b90f3d..e9c008fe14b1 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.h
+++ b/drivers/gpu/drm/i915/i915_gem_context.h
@@ -160,9 +160,6 @@ struct i915_gem_context {
/** desc_template: invariant fields for the HW context descriptor */
u32 desc_template;
- /** status_notifier: list of callbacks for context-switch changes */
- struct atomic_notifier_head status_notifier;
-
/** guilty_count: How many times this context has caused a GPU hang. */
unsigned int guilty_count;
/**
diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c
index c181b1bb3d2c..3be2503aa042 100644
--- a/drivers/gpu/drm/i915/i915_gem_evict.c
+++ b/drivers/gpu/drm/i915/i915_gem_evict.c
@@ -293,12 +293,12 @@ int i915_gem_evict_for_node(struct i915_address_space *vm,
* those as well to make room for our guard pages.
*/
if (check_color) {
- if (vma->node.start + vma->node.size == node->start) {
- if (vma->node.color == node->color)
+ if (node->start + node->size == target->start) {
+ if (node->color == target->color)
continue;
}
- if (vma->node.start == node->start + node->size) {
- if (vma->node.color == node->color)
+ if (node->start == target->start + target->size) {
+ if (node->color == target->color)
continue;
}
}
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index d02cfaefe1c8..15a15d00a6bf 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -888,6 +888,7 @@ i915_gem_execbuffer_reserve(struct intel_engine_cs *engine,
struct list_head ordered_vmas;
struct list_head pinned_vmas;
bool has_fenced_gpu_access = INTEL_GEN(engine->i915) < 4;
+ bool needs_unfenced_map = INTEL_INFO(engine->i915)->unfenced_needs_alignment;
int retry;
vm = list_first_entry(vmas, struct i915_vma, exec_list)->vm;
@@ -908,7 +909,8 @@ i915_gem_execbuffer_reserve(struct intel_engine_cs *engine,
if (!has_fenced_gpu_access)
entry->flags &= ~EXEC_OBJECT_NEEDS_FENCE;
need_fence =
- entry->flags & EXEC_OBJECT_NEEDS_FENCE &&
+ (entry->flags & EXEC_OBJECT_NEEDS_FENCE ||
+ needs_unfenced_map) &&
i915_gem_object_is_tiled(obj);
need_mappable = need_fence || need_reloc_mappable(vma);
@@ -1408,10 +1410,7 @@ execbuf_submit(struct i915_execbuffer_params *params,
struct drm_i915_gem_execbuffer2 *args,
struct list_head *vmas)
{
- struct drm_i915_private *dev_priv = params->request->i915;
u64 exec_start, exec_len;
- int instp_mode;
- u32 instp_mask;
int ret;
ret = i915_gem_execbuffer_move_to_gpu(params->request, vmas);
@@ -1422,56 +1421,11 @@ execbuf_submit(struct i915_execbuffer_params *params,
if (ret)
return ret;
- instp_mode = args->flags & I915_EXEC_CONSTANTS_MASK;
- instp_mask = I915_EXEC_CONSTANTS_MASK;
- switch (instp_mode) {
- case I915_EXEC_CONSTANTS_REL_GENERAL:
- case I915_EXEC_CONSTANTS_ABSOLUTE:
- case I915_EXEC_CONSTANTS_REL_SURFACE:
- if (instp_mode != 0 && params->engine->id != RCS) {
- DRM_DEBUG("non-0 rel constants mode on non-RCS\n");
- return -EINVAL;
- }
-
- if (instp_mode != dev_priv->relative_constants_mode) {
- if (INTEL_INFO(dev_priv)->gen < 4) {
- DRM_DEBUG("no rel constants on pre-gen4\n");
- return -EINVAL;
- }
-
- if (INTEL_INFO(dev_priv)->gen > 5 &&
- instp_mode == I915_EXEC_CONSTANTS_REL_SURFACE) {
- DRM_DEBUG("rel surface constants mode invalid on gen5+\n");
- return -EINVAL;
- }
-
- /* The HW changed the meaning on this bit on gen6 */
- if (INTEL_INFO(dev_priv)->gen >= 6)
- instp_mask &= ~I915_EXEC_CONSTANTS_REL_SURFACE;
- }
- break;
- default:
- DRM_DEBUG("execbuf with unknown constants: %d\n", instp_mode);
+ if (args->flags & I915_EXEC_CONSTANTS_MASK) {
+ DRM_DEBUG("I915_EXEC_CONSTANTS_* unsupported\n");
return -EINVAL;
}
- if (params->engine->id == RCS &&
- instp_mode != dev_priv->relative_constants_mode) {
- struct intel_ring *ring = params->request->ring;
-
- ret = intel_ring_begin(params->request, 4);
- if (ret)
- return ret;
-
- intel_ring_emit(ring, MI_NOOP);
- intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1));
- intel_ring_emit_reg(ring, INSTPM);
- intel_ring_emit(ring, instp_mask << 16 | instp_mode);
- intel_ring_advance(ring);
-
- dev_priv->relative_constants_mode = instp_mode;
- }
-
if (args->flags & I915_EXEC_GEN7_SOL_RESET) {
ret = i915_reset_gen7_sol_offsets(params->request);
if (ret)
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 2801a4d56324..96e45a4d5441 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -2704,7 +2704,7 @@ void i915_gem_gtt_finish_pages(struct drm_i915_gem_object *obj,
struct i915_ggtt *ggtt = &dev_priv->ggtt;
if (unlikely(ggtt->do_idle_maps)) {
- if (i915_gem_wait_for_idle(dev_priv, I915_WAIT_LOCKED)) {
+ if (i915_gem_wait_for_idle(dev_priv, 0)) {
DRM_ERROR("Failed to wait for idle; VT'd may hang.\n");
/* Wait a bit, in hopes it avoids the hang */
udelay(10);
diff --git a/drivers/gpu/drm/i915/i915_gem_object.h b/drivers/gpu/drm/i915/i915_gem_object.h
index bf90b07163d1..76b80a0be797 100644
--- a/drivers/gpu/drm/i915/i915_gem_object.h
+++ b/drivers/gpu/drm/i915/i915_gem_object.h
@@ -54,6 +54,9 @@ struct drm_i915_gem_object_ops {
struct sg_table *(*get_pages)(struct drm_i915_gem_object *);
void (*put_pages)(struct drm_i915_gem_object *, struct sg_table *);
+ int (*pwrite)(struct drm_i915_gem_object *,
+ const struct drm_i915_gem_pwrite *);
+
int (*dmabuf_export)(struct drm_i915_gem_object *);
void (*release)(struct drm_i915_gem_object *);
};
diff --git a/drivers/gpu/drm/i915/i915_gem_request.c b/drivers/gpu/drm/i915/i915_gem_request.c
index e7c3c0318ff6..da70bfe97ec5 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.c
+++ b/drivers/gpu/drm/i915/i915_gem_request.c
@@ -37,6 +37,17 @@ static const char *i915_fence_get_driver_name(struct dma_fence *fence)
static const char *i915_fence_get_timeline_name(struct dma_fence *fence)
{
+ /* The timeline struct (as part of the ppgtt underneath a context)
+ * may be freed when the request is no longer in use by the GPU.
+ * We could extend the life of a context to beyond that of all
+ * fences, possibly keeping the hw resource around indefinitely,
+ * or we just give them a false name. Since
+ * dma_fence_ops.get_timeline_name is a debug feature, the occasional
+ * lie seems justifiable.
+ */
+ if (test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &fence->flags))
+ return "signaled";
+
return to_request(fence)->timeline->common->name;
}
diff --git a/drivers/gpu/drm/i915/i915_gem_shrinker.c b/drivers/gpu/drm/i915/i915_gem_shrinker.c
index 401006b4c6a3..70b3832a79dd 100644
--- a/drivers/gpu/drm/i915/i915_gem_shrinker.c
+++ b/drivers/gpu/drm/i915/i915_gem_shrinker.c
@@ -53,6 +53,17 @@ static bool i915_gem_shrinker_lock(struct drm_device *dev, bool *unlock)
BUG();
}
+static void i915_gem_shrinker_unlock(struct drm_device *dev, bool unlock)
+{
+ if (!unlock)
+ return;
+
+ mutex_unlock(&dev->struct_mutex);
+
+ /* expedite the RCU grace period to free some request slabs */
+ synchronize_rcu_expedited();
+}
+
static bool any_vma_pinned(struct drm_i915_gem_object *obj)
{
struct i915_vma *vma;
@@ -232,11 +243,8 @@ i915_gem_shrink(struct drm_i915_private *dev_priv,
intel_runtime_pm_put(dev_priv);
i915_gem_retire_requests(dev_priv);
- if (unlock)
- mutex_unlock(&dev_priv->drm.struct_mutex);
- /* expedite the RCU grace period to free some request slabs */
- synchronize_rcu_expedited();
+ i915_gem_shrinker_unlock(&dev_priv->drm, unlock);
return count;
}
@@ -263,7 +271,7 @@ unsigned long i915_gem_shrink_all(struct drm_i915_private *dev_priv)
I915_SHRINK_BOUND |
I915_SHRINK_UNBOUND |
I915_SHRINK_ACTIVE);
- rcu_barrier(); /* wait until our RCU delayed slab frees are completed */
+ synchronize_rcu(); /* wait for our earlier RCU delayed slab frees */
return freed;
}
@@ -293,8 +301,7 @@ i915_gem_shrinker_count(struct shrinker *shrinker, struct shrink_control *sc)
count += obj->base.size >> PAGE_SHIFT;
}
- if (unlock)
- mutex_unlock(&dev->struct_mutex);
+ i915_gem_shrinker_unlock(dev, unlock);
return count;
}
@@ -321,8 +328,8 @@ i915_gem_shrinker_scan(struct shrinker *shrinker, struct shrink_control *sc)
sc->nr_to_scan - freed,
I915_SHRINK_BOUND |
I915_SHRINK_UNBOUND);
- if (unlock)
- mutex_unlock(&dev->struct_mutex);
+
+ i915_gem_shrinker_unlock(dev, unlock);
return freed;
}
@@ -364,8 +371,7 @@ i915_gem_shrinker_unlock_uninterruptible(struct drm_i915_private *dev_priv,
struct shrinker_lock_uninterruptible *slu)
{
dev_priv->mm.interruptible = slu->was_interruptible;
- if (slu->unlock)
- mutex_unlock(&dev_priv->drm.struct_mutex);
+ i915_gem_shrinker_unlock(&dev_priv->drm, slu->unlock);
}
static int
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index e6ffef2f707a..b6c886ac901b 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -1046,68 +1046,51 @@ static void vlv_c0_read(struct drm_i915_private *dev_priv,
ei->media_c0 = I915_READ(VLV_MEDIA_C0_COUNT);
}
-static bool vlv_c0_above(struct drm_i915_private *dev_priv,
- const struct intel_rps_ei *old,
- const struct intel_rps_ei *now,
- int threshold)
-{
- u64 time, c0;
- unsigned int mul = 100;
-
- if (old->cz_clock == 0)
- return false;
-
- if (I915_READ(VLV_COUNTER_CONTROL) & VLV_COUNT_RANGE_HIGH)
- mul <<= 8;
-
- time = now->cz_clock - old->cz_clock;
- time *= threshold * dev_priv->czclk_freq;
-
- /* Workload can be split between render + media, e.g. SwapBuffers
- * being blitted in X after being rendered in mesa. To account for
- * this we need to combine both engines into our activity counter.
- */
- c0 = now->render_c0 - old->render_c0;
- c0 += now->media_c0 - old->media_c0;
- c0 *= mul * VLV_CZ_CLOCK_TO_MILLI_SEC;
-
- return c0 >= time;
-}
-
void gen6_rps_reset_ei(struct drm_i915_private *dev_priv)
{
- vlv_c0_read(dev_priv, &dev_priv->rps.down_ei);
- dev_priv->rps.up_ei = dev_priv->rps.down_ei;
+ memset(&dev_priv->rps.ei, 0, sizeof(dev_priv->rps.ei));
}
static u32 vlv_wa_c0_ei(struct drm_i915_private *dev_priv, u32 pm_iir)
{
+ const struct intel_rps_ei *prev = &dev_priv->rps.ei;
struct intel_rps_ei now;
u32 events = 0;
- if ((pm_iir & (GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED)) == 0)
+ if ((pm_iir & GEN6_PM_RP_UP_EI_EXPIRED) == 0)
return 0;
vlv_c0_read(dev_priv, &now);
if (now.cz_clock == 0)
return 0;
- if (pm_iir & GEN6_PM_RP_DOWN_EI_EXPIRED) {
- if (!vlv_c0_above(dev_priv,
- &dev_priv->rps.down_ei, &now,
- dev_priv->rps.down_threshold))
- events |= GEN6_PM_RP_DOWN_THRESHOLD;
- dev_priv->rps.down_ei = now;
- }
+ if (prev->cz_clock) {
+ u64 time, c0;
+ unsigned int mul;
+
+ mul = VLV_CZ_CLOCK_TO_MILLI_SEC * 100; /* scale to threshold% */
+ if (I915_READ(VLV_COUNTER_CONTROL) & VLV_COUNT_RANGE_HIGH)
+ mul <<= 8;
- if (pm_iir & GEN6_PM_RP_UP_EI_EXPIRED) {
- if (vlv_c0_above(dev_priv,
- &dev_priv->rps.up_ei, &now,
- dev_priv->rps.up_threshold))
- events |= GEN6_PM_RP_UP_THRESHOLD;
- dev_priv->rps.up_ei = now;
+ time = now.cz_clock - prev->cz_clock;
+ time *= dev_priv->czclk_freq;
+
+ /* Workload can be split between render + media,
+ * e.g. SwapBuffers being blitted in X after being rendered in
+ * mesa. To account for this we need to combine both engines
+ * into our activity counter.
+ */
+ c0 = now.render_c0 - prev->render_c0;
+ c0 += now.media_c0 - prev->media_c0;
+ c0 *= mul;
+
+ if (c0 > time * dev_priv->rps.up_threshold)
+ events = GEN6_PM_RP_UP_THRESHOLD;
+ else if (c0 < time * dev_priv->rps.down_threshold)
+ events = GEN6_PM_RP_DOWN_THRESHOLD;
}
+ dev_priv->rps.ei = now;
return events;
}
@@ -4228,7 +4211,7 @@ void intel_irq_init(struct drm_i915_private *dev_priv)
/* Let's track the enabled rps events */
if (IS_VALLEYVIEW(dev_priv))
/* WaGsvRC0ResidencyMethod:vlv */
- dev_priv->pm_rps_events = GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED;
+ dev_priv->pm_rps_events = GEN6_PM_RP_UP_EI_EXPIRED;
else
dev_priv->pm_rps_events = GEN6_PM_RPS_EVENTS;
@@ -4266,6 +4249,16 @@ void intel_irq_init(struct drm_i915_private *dev_priv)
if (!IS_GEN2(dev_priv))
dev->vblank_disable_immediate = true;
+ /* Most platforms treat the display irq block as an always-on
+ * power domain. vlv/chv can disable it at runtime and need
+ * special care to avoid writing any of the display block registers
+ * outside of the power domain. We defer setting up the display irqs
+ * in this case to the runtime pm.
+ */
+ dev_priv->display_irqs_enabled = true;
+ if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv))
+ dev_priv->display_irqs_enabled = false;
+
dev->driver->get_vblank_timestamp = i915_get_vblank_timestamp;
dev->driver->get_scanout_position = i915_get_crtc_scanoutpos;
diff --git a/drivers/gpu/drm/i915/i915_pci.c b/drivers/gpu/drm/i915/i915_pci.c
index ecb487b5356f..9bbbd4e83e3c 100644
--- a/drivers/gpu/drm/i915/i915_pci.c
+++ b/drivers/gpu/drm/i915/i915_pci.c
@@ -60,6 +60,7 @@
.has_overlay = 1, .overlay_needs_physical = 1, \
.has_gmch_display = 1, \
.hws_needs_physical = 1, \
+ .unfenced_needs_alignment = 1, \
.ring_mask = RENDER_RING, \
GEN_DEFAULT_PIPEOFFSETS, \
CURSOR_OFFSETS
@@ -101,6 +102,7 @@ static const struct intel_device_info intel_i915g_info = {
.platform = INTEL_I915G, .cursor_needs_physical = 1,
.has_overlay = 1, .overlay_needs_physical = 1,
.hws_needs_physical = 1,
+ .unfenced_needs_alignment = 1,
};
static const struct intel_device_info intel_i915gm_info = {
@@ -112,6 +114,7 @@ static const struct intel_device_info intel_i915gm_info = {
.supports_tv = 1,
.has_fbc = 1,
.hws_needs_physical = 1,
+ .unfenced_needs_alignment = 1,
};
static const struct intel_device_info intel_i945g_info = {
@@ -120,6 +123,7 @@ static const struct intel_device_info intel_i945g_info = {
.has_hotplug = 1, .cursor_needs_physical = 1,
.has_overlay = 1, .overlay_needs_physical = 1,
.hws_needs_physical = 1,
+ .unfenced_needs_alignment = 1,
};
static const struct intel_device_info intel_i945gm_info = {
@@ -130,6 +134,7 @@ static const struct intel_device_info intel_i945gm_info = {
.supports_tv = 1,
.has_fbc = 1,
.hws_needs_physical = 1,
+ .unfenced_needs_alignment = 1,
};
static const struct intel_device_info intel_g33_info = {
diff --git a/drivers/gpu/drm/i915/i915_perf.c b/drivers/gpu/drm/i915/i915_perf.c
index a1b7eec58be2..70964ca9251e 100644
--- a/drivers/gpu/drm/i915/i915_perf.c
+++ b/drivers/gpu/drm/i915/i915_perf.c
@@ -1705,7 +1705,7 @@ i915_perf_open_ioctl_locked(struct drm_i915_private *dev_priv,
*/
if (WARN_ON(stream->sample_flags != props->sample_flags)) {
ret = -ENODEV;
- goto err_alloc;
+ goto err_flags;
}
list_add(&stream->link, &dev_priv->perf.streams);
@@ -1728,6 +1728,7 @@ i915_perf_open_ioctl_locked(struct drm_i915_private *dev_priv,
err_open:
list_del(&stream->link);
+err_flags:
if (stream->ops->destroy)
stream->ops->destroy(stream);
err_alloc:
@@ -1793,6 +1794,11 @@ static int read_properties_unlocked(struct drm_i915_private *dev_priv,
if (ret)
return ret;
+ if (id == 0 || id >= DRM_I915_PERF_PROP_MAX) {
+ DRM_DEBUG("Unknown i915 perf property ID\n");
+ return -EINVAL;
+ }
+
switch ((enum drm_i915_perf_property_id)id) {
case DRM_I915_PERF_PROP_CTX_HANDLE:
props->single_context = 1;
@@ -1862,9 +1868,8 @@ static int read_properties_unlocked(struct drm_i915_private *dev_priv,
props->oa_periodic = true;
props->oa_period_exponent = value;
break;
- default:
+ case DRM_I915_PERF_PROP_MAX:
MISSING_CASE(id);
- DRM_DEBUG("Unknown i915 perf property ID\n");
return -EINVAL;
}
diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
index 155906e84812..df20e9bc1c0f 100644
--- a/drivers/gpu/drm/i915/i915_vma.c
+++ b/drivers/gpu/drm/i915/i915_vma.c
@@ -512,10 +512,36 @@ err_unpin:
return ret;
}
+static void
+i915_vma_remove(struct i915_vma *vma)
+{
+ struct drm_i915_gem_object *obj = vma->obj;
+
+ GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
+ GEM_BUG_ON(vma->flags & (I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND));
+
+ drm_mm_remove_node(&vma->node);
+ list_move_tail(&vma->vm_link, &vma->vm->unbound_list);
+
+ /* Since the unbound list is global, only move to that list if
+ * no more VMAs exist.
+ */
+ if (--obj->bind_count == 0)
+ list_move_tail(&obj->global_link,
+ &to_i915(obj->base.dev)->mm.unbound_list);
+
+ /* And finally now the object is completely decoupled from this vma,
+ * we can drop its hold on the backing storage and allow it to be
+ * reaped by the shrinker.
+ */
+ i915_gem_object_unpin_pages(obj);
+ GEM_BUG_ON(atomic_read(&obj->mm.pages_pin_count) < obj->bind_count);
+}
+
int __i915_vma_do_pin(struct i915_vma *vma,
u64 size, u64 alignment, u64 flags)
{
- unsigned int bound = vma->flags;
+ const unsigned int bound = vma->flags;
int ret;
lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
@@ -524,18 +550,18 @@ int __i915_vma_do_pin(struct i915_vma *vma,
if (WARN_ON(bound & I915_VMA_PIN_OVERFLOW)) {
ret = -EBUSY;
- goto err;
+ goto err_unpin;
}
if ((bound & I915_VMA_BIND_MASK) == 0) {
ret = i915_vma_insert(vma, size, alignment, flags);
if (ret)
- goto err;
+ goto err_unpin;
}
ret = i915_vma_bind(vma, vma->obj->cache_level, flags);
if (ret)
- goto err;
+ goto err_remove;
if ((bound ^ vma->flags) & I915_VMA_GLOBAL_BIND)
__i915_vma_set_map_and_fenceable(vma);
@@ -544,7 +570,12 @@ int __i915_vma_do_pin(struct i915_vma *vma,
GEM_BUG_ON(i915_vma_misplaced(vma, size, alignment, flags));
return 0;
-err:
+err_remove:
+ if ((bound & I915_VMA_BIND_MASK) == 0) {
+ GEM_BUG_ON(vma->pages);
+ i915_vma_remove(vma);
+ }
+err_unpin:
__i915_vma_unpin(vma);
return ret;
}
@@ -657,9 +688,6 @@ int i915_vma_unbind(struct i915_vma *vma)
}
vma->flags &= ~(I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND);
- drm_mm_remove_node(&vma->node);
- list_move_tail(&vma->vm_link, &vma->vm->unbound_list);
-
if (vma->pages != obj->mm.pages) {
GEM_BUG_ON(!vma->pages);
sg_free_table(vma->pages);
@@ -667,18 +695,7 @@ int i915_vma_unbind(struct i915_vma *vma)
}
vma->pages = NULL;
- /* Since the unbound list is global, only move to that list if
- * no more VMAs exist. */
- if (--obj->bind_count == 0)
- list_move_tail(&obj->global_link,
- &to_i915(obj->base.dev)->mm.unbound_list);
-
- /* And finally now the object is completely decoupled from this vma,
- * we can drop its hold on the backing storage and allow it to be
- * reaped by the shrinker.
- */
- i915_gem_object_unpin_pages(obj);
- GEM_BUG_ON(atomic_read(&obj->mm.pages_pin_count) < obj->bind_count);
+ i915_vma_remove(vma);
destroy:
if (unlikely(i915_vma_is_closed(vma)))
diff --git a/drivers/gpu/drm/i915/intel_csr.c b/drivers/gpu/drm/i915/intel_csr.c
index 0085bc745f6a..de219b71fb76 100644
--- a/drivers/gpu/drm/i915/intel_csr.c
+++ b/drivers/gpu/drm/i915/intel_csr.c
@@ -35,7 +35,6 @@
*/
#define I915_CSR_GLK "i915/glk_dmc_ver1_01.bin"
-MODULE_FIRMWARE(I915_CSR_GLK);
#define GLK_CSR_VERSION_REQUIRED CSR_VERSION(1, 1)
#define I915_CSR_KBL "i915/kbl_dmc_ver1_01.bin"
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 01341670738f..ed1f4f272b4f 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -3669,10 +3669,6 @@ static void intel_update_pipe_config(struct intel_crtc *crtc,
/* drm_atomic_helper_update_legacy_modeset_state might not be called. */
crtc->base.mode = crtc->base.state->mode;
- DRM_DEBUG_KMS("Updating pipe size %ix%i -> %ix%i\n",
- old_crtc_state->pipe_src_w, old_crtc_state->pipe_src_h,
- pipe_config->pipe_src_w, pipe_config->pipe_src_h);
-
/*
* Update pipe size and adjust fitter if needed: the reason for this is
* that in compute_mode_changes we check the native mode (not the pfit
@@ -4796,23 +4792,17 @@ static void skylake_pfit_enable(struct intel_crtc *crtc)
struct intel_crtc_scaler_state *scaler_state =
&crtc->config->scaler_state;
- DRM_DEBUG_KMS("for crtc_state = %p\n", crtc->config);
-
if (crtc->config->pch_pfit.enabled) {
int id;
- if (WARN_ON(crtc->config->scaler_state.scaler_id < 0)) {
- DRM_ERROR("Requesting pfit without getting a scaler first\n");
+ if (WARN_ON(crtc->config->scaler_state.scaler_id < 0))
return;
- }
id = scaler_state->scaler_id;
I915_WRITE(SKL_PS_CTRL(pipe, id), PS_SCALER_EN |
PS_FILTER_MEDIUM | scaler_state->scalers[id].mode);
I915_WRITE(SKL_PS_WIN_POS(pipe, id), crtc->config->pch_pfit.pos);
I915_WRITE(SKL_PS_WIN_SZ(pipe, id), crtc->config->pch_pfit.size);
-
- DRM_DEBUG_KMS("for crtc_state = %p scaler_id = %d\n", crtc->config, id);
}
}
@@ -14379,6 +14369,24 @@ static void skl_update_crtcs(struct drm_atomic_state *state,
} while (progress);
}
+static void intel_atomic_helper_free_state(struct drm_i915_private *dev_priv)
+{
+ struct intel_atomic_state *state, *next;
+ struct llist_node *freed;
+
+ freed = llist_del_all(&dev_priv->atomic_helper.free_list);
+ llist_for_each_entry_safe(state, next, freed, freed)
+ drm_atomic_state_put(&state->base);
+}
+
+static void intel_atomic_helper_free_state_worker(struct work_struct *work)
+{
+ struct drm_i915_private *dev_priv =
+ container_of(work, typeof(*dev_priv), atomic_helper.free_work);
+
+ intel_atomic_helper_free_state(dev_priv);
+}
+
static void intel_atomic_commit_tail(struct drm_atomic_state *state)
{
struct drm_device *dev = state->dev;
@@ -14545,6 +14553,8 @@ static void intel_atomic_commit_tail(struct drm_atomic_state *state)
* can happen also when the device is completely off.
*/
intel_uncore_arm_unclaimed_mmio_detection(dev_priv);
+
+ intel_atomic_helper_free_state(dev_priv);
}
static void intel_atomic_commit_work(struct work_struct *work)
@@ -14946,17 +14956,19 @@ static void intel_begin_crtc_commit(struct drm_crtc *crtc,
to_intel_atomic_state(old_crtc_state->state);
bool modeset = needs_modeset(crtc->state);
+ if (!modeset &&
+ (intel_cstate->base.color_mgmt_changed ||
+ intel_cstate->update_pipe)) {
+ intel_color_set_csc(crtc->state);
+ intel_color_load_luts(crtc->state);
+ }
+
/* Perform vblank evasion around commit operation */
intel_pipe_update_start(intel_crtc);
if (modeset)
goto out;
- if (crtc->state->color_mgmt_changed || to_intel_crtc_state(crtc->state)->update_pipe) {
- intel_color_set_csc(crtc->state);
- intel_color_load_luts(crtc->state);
- }
-
if (intel_cstate->update_pipe)
intel_update_pipe_config(intel_crtc, old_intel_cstate);
else if (INTEL_GEN(dev_priv) >= 9)
@@ -16599,18 +16611,6 @@ fail:
drm_modeset_acquire_fini(&ctx);
}
-static void intel_atomic_helper_free_state(struct work_struct *work)
-{
- struct drm_i915_private *dev_priv =
- container_of(work, typeof(*dev_priv), atomic_helper.free_work);
- struct intel_atomic_state *state, *next;
- struct llist_node *freed;
-
- freed = llist_del_all(&dev_priv->atomic_helper.free_list);
- llist_for_each_entry_safe(state, next, freed, freed)
- drm_atomic_state_put(&state->base);
-}
-
int intel_modeset_init(struct drm_device *dev)
{
struct drm_i915_private *dev_priv = to_i915(dev);
@@ -16631,7 +16631,7 @@ int intel_modeset_init(struct drm_device *dev)
dev->mode_config.funcs = &intel_mode_funcs;
INIT_WORK(&dev_priv->atomic_helper.free_work,
- intel_atomic_helper_free_state);
+ intel_atomic_helper_free_state_worker);
intel_init_quirks(dev);
@@ -16696,12 +16696,11 @@ int intel_modeset_init(struct drm_device *dev)
}
}
- intel_update_czclk(dev_priv);
- intel_update_cdclk(dev_priv);
- dev_priv->atomic_cdclk_freq = dev_priv->cdclk_freq;
-
intel_shared_dpll_init(dev);
+ intel_update_czclk(dev_priv);
+ intel_modeset_init_hw(dev);
+
if (dev_priv->max_cdclk_freq == 0)
intel_update_max_cdclk(dev_priv);
@@ -17258,8 +17257,6 @@ void intel_modeset_gem_init(struct drm_device *dev)
intel_init_gt_powersave(dev_priv);
- intel_modeset_init_hw(dev);
-
intel_setup_overlay(dev_priv);
}
diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index 371acf109e34..ab1be5c80ea5 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -105,6 +105,8 @@ intel_engine_setup(struct drm_i915_private *dev_priv,
/* Nothing to do here, execute in order of dependencies */
engine->schedule = NULL;
+ ATOMIC_INIT_NOTIFIER_HEAD(&engine->context_status_notifier);
+
dev_priv->engine[id] = engine;
return 0;
}
diff --git a/drivers/gpu/drm/i915/intel_fbdev.c b/drivers/gpu/drm/i915/intel_fbdev.c
index 1b8ba2e77539..2d449fb5d1d2 100644
--- a/drivers/gpu/drm/i915/intel_fbdev.c
+++ b/drivers/gpu/drm/i915/intel_fbdev.c
@@ -357,14 +357,13 @@ static bool intel_fb_initial_config(struct drm_fb_helper *fb_helper,
bool *enabled, int width, int height)
{
struct drm_i915_private *dev_priv = to_i915(fb_helper->dev);
- unsigned long conn_configured, mask;
+ unsigned long conn_configured, conn_seq, mask;
unsigned int count = min(fb_helper->connector_count, BITS_PER_LONG);
int i, j;
bool *save_enabled;
bool fallback = true;
int num_connectors_enabled = 0;
int num_connectors_detected = 0;
- int pass = 0;
save_enabled = kcalloc(count, sizeof(bool), GFP_KERNEL);
if (!save_enabled)
@@ -374,6 +373,7 @@ static bool intel_fb_initial_config(struct drm_fb_helper *fb_helper,
mask = BIT(count) - 1;
conn_configured = 0;
retry:
+ conn_seq = conn_configured;
for (i = 0; i < count; i++) {
struct drm_fb_helper_connector *fb_conn;
struct drm_connector *connector;
@@ -387,7 +387,7 @@ retry:
if (conn_configured & BIT(i))
continue;
- if (pass == 0 && !connector->has_tile)
+ if (conn_seq == 0 && !connector->has_tile)
continue;
if (connector->status == connector_status_connected)
@@ -498,10 +498,8 @@ retry:
conn_configured |= BIT(i);
}
- if ((conn_configured & mask) != mask) {
- pass++;
+ if ((conn_configured & mask) != mask && conn_configured != conn_seq)
goto retry;
- }
/*
* If the BIOS didn't enable everything it could, fall back to have the
diff --git a/drivers/gpu/drm/i915/intel_gvt.c b/drivers/gpu/drm/i915/intel_gvt.c
index d23c0fcff751..8c04eca84351 100644
--- a/drivers/gpu/drm/i915/intel_gvt.c
+++ b/drivers/gpu/drm/i915/intel_gvt.c
@@ -77,6 +77,11 @@ int intel_gvt_init(struct drm_i915_private *dev_priv)
goto bail;
}
+ if (!i915.enable_execlists) {
+ DRM_INFO("GPU guest virtualisation [GVT-g] disabled due to disabled execlist submission [i915.enable_execlists module parameter]\n");
+ goto bail;
+ }
+
/*
* We're not in host or fail to find a MPT module, disable GVT-g
*/
diff --git a/drivers/gpu/drm/i915/intel_hdmi.c b/drivers/gpu/drm/i915/intel_hdmi.c
index ebae2bd83918..24b2fa5b6282 100644
--- a/drivers/gpu/drm/i915/intel_hdmi.c
+++ b/drivers/gpu/drm/i915/intel_hdmi.c
@@ -1298,16 +1298,34 @@ intel_hdmi_mode_valid(struct drm_connector *connector,
static bool hdmi_12bpc_possible(struct intel_crtc_state *crtc_state)
{
- struct drm_device *dev = crtc_state->base.crtc->dev;
+ struct drm_i915_private *dev_priv =
+ to_i915(crtc_state->base.crtc->dev);
+ struct drm_atomic_state *state = crtc_state->base.state;
+ struct drm_connector_state *connector_state;
+ struct drm_connector *connector;
+ int i;
- if (HAS_GMCH_DISPLAY(to_i915(dev)))
+ if (HAS_GMCH_DISPLAY(dev_priv))
return false;
/*
* HDMI 12bpc affects the clocks, so it's only possible
* when not cloning with other encoder types.
*/
- return crtc_state->output_types == 1 << INTEL_OUTPUT_HDMI;
+ if (crtc_state->output_types != 1 << INTEL_OUTPUT_HDMI)
+ return false;
+
+ for_each_connector_in_state(state, connector, connector_state, i) {
+ const struct drm_display_info *info = &connector->display_info;
+
+ if (connector_state->crtc != crtc_state->base.crtc)
+ continue;
+
+ if ((info->edid_hdmi_dc_modes & DRM_EDID_HDMI_DC_36) == 0)
+ return false;
+ }
+
+ return true;
}
bool intel_hdmi_compute_config(struct intel_encoder *encoder,
diff --git a/drivers/gpu/drm/i915/intel_hotplug.c b/drivers/gpu/drm/i915/intel_hotplug.c
index b62e3f8ad415..54208bef7a83 100644
--- a/drivers/gpu/drm/i915/intel_hotplug.c
+++ b/drivers/gpu/drm/i915/intel_hotplug.c
@@ -219,7 +219,7 @@ static void intel_hpd_irq_storm_reenable_work(struct work_struct *work)
}
}
}
- if (dev_priv->display.hpd_irq_setup)
+ if (dev_priv->display_irqs_enabled && dev_priv->display.hpd_irq_setup)
dev_priv->display.hpd_irq_setup(dev_priv);
spin_unlock_irq(&dev_priv->irq_lock);
@@ -425,7 +425,7 @@ void intel_hpd_irq_handler(struct drm_i915_private *dev_priv,
}
}
- if (storm_detected)
+ if (storm_detected && dev_priv->display_irqs_enabled)
dev_priv->display.hpd_irq_setup(dev_priv);
spin_unlock(&dev_priv->irq_lock);
@@ -471,10 +471,12 @@ void intel_hpd_init(struct drm_i915_private *dev_priv)
* Interrupt setup is already guaranteed to be single-threaded, this is
* just to make the assert_spin_locked checks happy.
*/
- spin_lock_irq(&dev_priv->irq_lock);
- if (dev_priv->display.hpd_irq_setup)
- dev_priv->display.hpd_irq_setup(dev_priv);
- spin_unlock_irq(&dev_priv->irq_lock);
+ if (dev_priv->display_irqs_enabled && dev_priv->display.hpd_irq_setup) {
+ spin_lock_irq(&dev_priv->irq_lock);
+ if (dev_priv->display_irqs_enabled)
+ dev_priv->display.hpd_irq_setup(dev_priv);
+ spin_unlock_irq(&dev_priv->irq_lock);
+ }
}
static void i915_hpd_poll_init_work(struct work_struct *work)
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index ebf8023d21e6..47517a02f0a4 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -345,7 +345,8 @@ execlists_context_status_change(struct drm_i915_gem_request *rq,
if (!IS_ENABLED(CONFIG_DRM_I915_GVT))
return;
- atomic_notifier_call_chain(&rq->ctx->status_notifier, status, rq);
+ atomic_notifier_call_chain(&rq->engine->context_status_notifier,
+ status, rq);
}
static void
@@ -669,15 +670,14 @@ static void execlists_submit_request(struct drm_i915_gem_request *request)
static struct intel_engine_cs *
pt_lock_engine(struct i915_priotree *pt, struct intel_engine_cs *locked)
{
- struct intel_engine_cs *engine;
+ struct intel_engine_cs *engine =
+ container_of(pt, struct drm_i915_gem_request, priotree)->engine;
+
+ GEM_BUG_ON(!locked);
- engine = container_of(pt,
- struct drm_i915_gem_request,
- priotree)->engine;
if (engine != locked) {
- if (locked)
- spin_unlock_irq(&locked->timeline->lock);
- spin_lock_irq(&engine->timeline->lock);
+ spin_unlock(&locked->timeline->lock);
+ spin_lock(&engine->timeline->lock);
}
return engine;
@@ -685,7 +685,7 @@ pt_lock_engine(struct i915_priotree *pt, struct intel_engine_cs *locked)
static void execlists_schedule(struct drm_i915_gem_request *request, int prio)
{
- struct intel_engine_cs *engine = NULL;
+ struct intel_engine_cs *engine;
struct i915_dependency *dep, *p;
struct i915_dependency stack;
LIST_HEAD(dfs);
@@ -719,26 +719,23 @@ static void execlists_schedule(struct drm_i915_gem_request *request, int prio)
list_for_each_entry_safe(dep, p, &dfs, dfs_link) {
struct i915_priotree *pt = dep->signaler;
- list_for_each_entry(p, &pt->signalers_list, signal_link)
+ /* Within an engine, there can be no cycle, but we may
+ * refer to the same dependency chain multiple times
+ * (redundant dependencies are not eliminated) and across
+ * engines.
+ */
+ list_for_each_entry(p, &pt->signalers_list, signal_link) {
+ GEM_BUG_ON(p->signaler->priority < pt->priority);
if (prio > READ_ONCE(p->signaler->priority))
list_move_tail(&p->dfs_link, &dfs);
+ }
list_safe_reset_next(dep, p, dfs_link);
- if (!RB_EMPTY_NODE(&pt->node))
- continue;
-
- engine = pt_lock_engine(pt, engine);
-
- /* If it is not already in the rbtree, we can update the
- * priority inplace and skip over it (and its dependencies)
- * if it is referenced *again* as we descend the dfs.
- */
- if (prio > pt->priority && RB_EMPTY_NODE(&pt->node)) {
- pt->priority = prio;
- list_del_init(&dep->dfs_link);
- }
}
+ engine = request->engine;
+ spin_lock_irq(&engine->timeline->lock);
+
/* Fifo and depth-first replacement ensure our deps execute before us */
list_for_each_entry_safe_reverse(dep, p, &dfs, dfs_link) {
struct i915_priotree *pt = dep->signaler;
@@ -750,16 +747,15 @@ static void execlists_schedule(struct drm_i915_gem_request *request, int prio)
if (prio <= pt->priority)
continue;
- GEM_BUG_ON(RB_EMPTY_NODE(&pt->node));
-
pt->priority = prio;
- rb_erase(&pt->node, &engine->execlist_queue);
- if (insert_request(pt, &engine->execlist_queue))
- engine->execlist_first = &pt->node;
+ if (!RB_EMPTY_NODE(&pt->node)) {
+ rb_erase(&pt->node, &engine->execlist_queue);
+ if (insert_request(pt, &engine->execlist_queue))
+ engine->execlist_first = &pt->node;
+ }
}
- if (engine)
- spin_unlock_irq(&engine->timeline->lock);
+ spin_unlock_irq(&engine->timeline->lock);
/* XXX Do we need to preempt to make room for us and our deps? */
}
@@ -1439,7 +1435,9 @@ static void reset_common_ring(struct intel_engine_cs *engine,
GEM_BUG_ON(request->ctx != port[0].request->ctx);
/* Reset WaIdleLiteRestore:bdw,skl as well */
- request->tail = request->wa_tail - WA_TAIL_DWORDS * sizeof(u32);
+ request->tail =
+ intel_ring_wrap(request->ring,
+ request->wa_tail - WA_TAIL_DWORDS*sizeof(u32));
}
static int intel_logical_ring_emit_pdps(struct drm_i915_gem_request *req)
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 249623d45be0..6a29784d2b41 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -4891,6 +4891,12 @@ static void gen6_set_rps_thresholds(struct drm_i915_private *dev_priv, u8 val)
break;
}
+ /* When byt can survive without system hang with dynamic
+ * sw freq adjustments, this restriction can be lifted.
+ */
+ if (IS_VALLEYVIEW(dev_priv))
+ goto skip_hw_write;
+
I915_WRITE(GEN6_RP_UP_EI,
GT_INTERVAL_FROM_US(dev_priv, ei_up));
I915_WRITE(GEN6_RP_UP_THRESHOLD,
@@ -4911,6 +4917,7 @@ static void gen6_set_rps_thresholds(struct drm_i915_private *dev_priv, u8 val)
GEN6_RP_UP_BUSY_AVG |
GEN6_RP_DOWN_IDLE_AVG);
+skip_hw_write:
dev_priv->rps.power = new_power;
dev_priv->rps.up_threshold = threshold_up;
dev_priv->rps.down_threshold = threshold_down;
@@ -4921,8 +4928,9 @@ static u32 gen6_rps_pm_mask(struct drm_i915_private *dev_priv, u8 val)
{
u32 mask = 0;
+ /* We use UP_EI_EXPIRED interupts for both up/down in manual mode */
if (val > dev_priv->rps.min_freq_softlimit)
- mask |= GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT;
+ mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT;
if (val < dev_priv->rps.max_freq_softlimit)
mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_UP_THRESHOLD;
@@ -5032,7 +5040,7 @@ void gen6_rps_busy(struct drm_i915_private *dev_priv)
{
mutex_lock(&dev_priv->rps.hw_lock);
if (dev_priv->rps.enabled) {
- if (dev_priv->pm_rps_events & (GEN6_PM_RP_DOWN_EI_EXPIRED | GEN6_PM_RP_UP_EI_EXPIRED))
+ if (dev_priv->pm_rps_events & GEN6_PM_RP_UP_EI_EXPIRED)
gen6_rps_reset_ei(dev_priv);
I915_WRITE(GEN6_PMINTRMSK,
gen6_rps_pm_mask(dev_priv, dev_priv->rps.cur_freq));
@@ -7916,10 +7924,10 @@ static bool skl_pcode_try_request(struct drm_i915_private *dev_priv, u32 mbox,
* @timeout_base_ms: timeout for polling with preemption enabled
*
* Keep resending the @request to @mbox until PCODE acknowledges it, PCODE
- * reports an error or an overall timeout of @timeout_base_ms+10 ms expires.
+ * reports an error or an overall timeout of @timeout_base_ms+50 ms expires.
* The request is acknowledged once the PCODE reply dword equals @reply after
* applying @reply_mask. Polling is first attempted with preemption enabled
- * for @timeout_base_ms and if this times out for another 10 ms with
+ * for @timeout_base_ms and if this times out for another 50 ms with
* preemption disabled.
*
* Returns 0 on success, %-ETIMEDOUT in case of a timeout, <0 in case of some
@@ -7955,14 +7963,15 @@ int skl_pcode_request(struct drm_i915_private *dev_priv, u32 mbox, u32 request,
* worst case) _and_ PCODE was busy for some reason even after a
* (queued) request and @timeout_base_ms delay. As a workaround retry
* the poll with preemption disabled to maximize the number of
- * requests. Increase the timeout from @timeout_base_ms to 10ms to
+ * requests. Increase the timeout from @timeout_base_ms to 50ms to
* account for interrupts that could reduce the number of these
- * requests.
+ * requests, and for any quirks of the PCODE firmware that delays
+ * the request completion.
*/
DRM_DEBUG_KMS("PCODE timeout, retrying with preemption disabled\n");
WARN_ON_ONCE(timeout_base_ms > 3);
preempt_disable();
- ret = wait_for_atomic(COND, 10);
+ ret = wait_for_atomic(COND, 50);
preempt_enable();
out:
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 91bc4abf5d3e..6c5f9958197d 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -2024,6 +2024,8 @@ static int intel_ring_context_pin(struct intel_engine_cs *engine,
ret = context_pin(ctx, flags);
if (ret)
goto error;
+
+ ce->state->obj->mm.dirty = true;
}
/* The kernel context is only used as a placeholder for flushing the
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 79c2b8d72322..8cb2078c5bfc 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -403,6 +403,9 @@ struct intel_engine_cs {
*/
struct i915_gem_context *legacy_active_context;
+ /* status_notifier: list of callbacks for context-switch changes */
+ struct atomic_notifier_head context_status_notifier;
+
struct intel_engine_hangcheck hangcheck;
bool needs_cmd_parser;
@@ -518,11 +521,17 @@ static inline void intel_ring_advance(struct intel_ring *ring)
*/
}
+static inline u32
+intel_ring_wrap(const struct intel_ring *ring, u32 pos)
+{
+ return pos & (ring->size - 1);
+}
+
static inline u32 intel_ring_offset(struct intel_ring *ring, void *addr)
{
/* Don't write ring->size (equivalent to 0) as that hangs some GPUs. */
u32 offset = addr - ring->vaddr;
- return offset & (ring->size - 1);
+ return intel_ring_wrap(ring, offset);
}
int __intel_ring_space(int head, int tail, int size);
diff --git a/drivers/gpu/drm/i915/intel_sprite.c b/drivers/gpu/drm/i915/intel_sprite.c
index 9ef54688872a..9481ca9a3ae7 100644
--- a/drivers/gpu/drm/i915/intel_sprite.c
+++ b/drivers/gpu/drm/i915/intel_sprite.c
@@ -254,9 +254,6 @@ skl_update_plane(struct drm_plane *drm_plane,
int scaler_id = plane_state->scaler_id;
const struct intel_scaler *scaler;
- DRM_DEBUG_KMS("plane = %d PS_PLANE_SEL(plane) = 0x%x\n",
- plane_id, PS_PLANE_SEL(plane_id));
-
scaler = &crtc_state->scaler_state.scalers[scaler_id];
I915_WRITE(SKL_PS_CTRL(pipe, scaler_id),
diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
index abe08885a5ba..b7ff592b14f5 100644
--- a/drivers/gpu/drm/i915/intel_uncore.c
+++ b/drivers/gpu/drm/i915/intel_uncore.c
@@ -119,6 +119,8 @@ fw_domains_get(struct drm_i915_private *dev_priv, enum forcewake_domains fw_doma
for_each_fw_domain_masked(d, fw_domains, dev_priv)
fw_domain_wait_ack(d);
+
+ dev_priv->uncore.fw_domains_active |= fw_domains;
}
static void
@@ -130,6 +132,8 @@ fw_domains_put(struct drm_i915_private *dev_priv, enum forcewake_domains fw_doma
fw_domain_put(d);
fw_domain_posting_read(d);
}
+
+ dev_priv->uncore.fw_domains_active &= ~fw_domains;
}
static void
@@ -240,10 +244,8 @@ intel_uncore_fw_release_timer(struct hrtimer *timer)
if (WARN_ON(domain->wake_count == 0))
domain->wake_count++;
- if (--domain->wake_count == 0) {
+ if (--domain->wake_count == 0)
dev_priv->uncore.funcs.force_wake_put(dev_priv, domain->mask);
- dev_priv->uncore.fw_domains_active &= ~domain->mask;
- }
spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
@@ -454,10 +456,8 @@ static void __intel_uncore_forcewake_get(struct drm_i915_private *dev_priv,
fw_domains &= ~domain->mask;
}
- if (fw_domains) {
+ if (fw_domains)
dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
- dev_priv->uncore.fw_domains_active |= fw_domains;
- }
}
/**
@@ -968,7 +968,6 @@ static noinline void ___force_wake_auto(struct drm_i915_private *dev_priv,
fw_domain_arm_timer(domain);
dev_priv->uncore.funcs.force_wake_get(dev_priv, fw_domains);
- dev_priv->uncore.fw_domains_active |= fw_domains;
}
static inline void __force_wake_auto(struct drm_i915_private *dev_priv,