aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/nouveau
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/nouveau')
-rw-r--r--drivers/gpu/drm/nouveau/Kconfig5
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_acpi.c12
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_bios.c205
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_bios.h2
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_bo.c245
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_channel.c54
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_connector.c36
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_dma.c108
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_dma.h10
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_dp.c8
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_drv.c14
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_drv.h78
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_fbcon.c59
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_fbcon.h7
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_fence.c2
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_gem.c55
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_grctx.c4
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_irq.c161
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_mem.c215
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_notifier.c13
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_object.c5
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_reg.h17
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_sgdma.c7
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_state.c77
-rw-r--r--drivers/gpu/drm/nouveau/nouveau_ttm.c30
-rw-r--r--drivers/gpu/drm/nouveau/nv04_dac.c41
-rw-r--r--drivers/gpu/drm/nouveau/nv04_fbcon.c50
-rw-r--r--drivers/gpu/drm/nouveau/nv04_fifo.c34
-rw-r--r--drivers/gpu/drm/nouveau/nv04_graph.c159
-rw-r--r--drivers/gpu/drm/nouveau/nv04_instmem.c2
-rw-r--r--drivers/gpu/drm/nouveau/nv10_fb.c32
-rw-r--r--drivers/gpu/drm/nouveau/nv10_graph.c28
-rw-r--r--drivers/gpu/drm/nouveau/nv17_tv.c117
-rw-r--r--drivers/gpu/drm/nouveau/nv20_graph.c61
-rw-r--r--drivers/gpu/drm/nouveau/nv40_fb.c53
-rw-r--r--drivers/gpu/drm/nouveau/nv40_graph.c116
-rw-r--r--drivers/gpu/drm/nouveau/nv50_crtc.c33
-rw-r--r--drivers/gpu/drm/nouveau/nv50_display.c17
-rw-r--r--drivers/gpu/drm/nouveau/nv50_fbcon.c32
-rw-r--r--drivers/gpu/drm/nouveau/nv50_fifo.c17
-rw-r--r--drivers/gpu/drm/nouveau/nv50_graph.c13
-rw-r--r--drivers/gpu/drm/nouveau/nv50_instmem.c58
-rw-r--r--drivers/gpu/drm/nouveau/nv50_sor.c14
43 files changed, 1468 insertions, 838 deletions
diff --git a/drivers/gpu/drm/nouveau/Kconfig b/drivers/gpu/drm/nouveau/Kconfig
index b1bc1ea182b..1175429da10 100644
--- a/drivers/gpu/drm/nouveau/Kconfig
+++ b/drivers/gpu/drm/nouveau/Kconfig
@@ -30,12 +30,11 @@ config DRM_NOUVEAU_DEBUG
via debugfs.
menu "I2C encoder or helper chips"
- depends on DRM && I2C
+ depends on DRM && DRM_KMS_HELPER && I2C
config DRM_I2C_CH7006
tristate "Chrontel ch7006 TV encoder"
- depends on DRM_NOUVEAU
- default m
+ default m if DRM_NOUVEAU
help
Support for Chrontel ch7006 and similar TV encoders, found
on some nVidia video cards.
diff --git a/drivers/gpu/drm/nouveau/nouveau_acpi.c b/drivers/gpu/drm/nouveau/nouveau_acpi.c
index 1cf488247a1..48227e74475 100644
--- a/drivers/gpu/drm/nouveau/nouveau_acpi.c
+++ b/drivers/gpu/drm/nouveau/nouveau_acpi.c
@@ -90,21 +90,21 @@ int nouveau_hybrid_setup(struct drm_device *dev)
{
int result;
- if (nouveau_dsm(dev, NOUVEAU_DSM_ACTIVE, NOUVEAU_DSM_ACTIVE_QUERY,
+ if (nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_STATE,
&result))
return -ENODEV;
NV_INFO(dev, "_DSM hardware status gave 0x%x\n", result);
- if (result & 0x1) { /* Stamina mode - disable the external GPU */
+ if (result) { /* Ensure that the external GPU is enabled */
+ nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_SPEED, NULL);
+ nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_SPEED,
+ NULL);
+ } else { /* Stamina mode - disable the external GPU */
nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_STAMINA,
NULL);
nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_STAMINA,
NULL);
- } else { /* Ensure that the external GPU is enabled */
- nouveau_dsm(dev, NOUVEAU_DSM_LED, NOUVEAU_DSM_LED_SPEED, NULL);
- nouveau_dsm(dev, NOUVEAU_DSM_POWER, NOUVEAU_DSM_POWER_SPEED,
- NULL);
}
return 0;
diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.c b/drivers/gpu/drm/nouveau/nouveau_bios.c
index ba143972769..0e9cd1d4913 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bios.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bios.c
@@ -310,63 +310,22 @@ valid_reg(struct nvbios *bios, uint32_t reg)
struct drm_device *dev = bios->dev;
/* C51 has misaligned regs on purpose. Marvellous */
- if (reg & 0x2 || (reg & 0x1 && dev_priv->VBIOS.pub.chip_version != 0x51)) {
- NV_ERROR(dev, "========== misaligned reg 0x%08X ==========\n",
- reg);
- return 0;
- }
- /*
- * Warn on C51 regs that have not been verified accessible in
- * mmiotracing
- */
+ if (reg & 0x2 ||
+ (reg & 0x1 && dev_priv->VBIOS.pub.chip_version != 0x51))
+ NV_ERROR(dev, "======= misaligned reg 0x%08X =======\n", reg);
+
+ /* warn on C51 regs that haven't been verified accessible in tracing */
if (reg & 0x1 && dev_priv->VBIOS.pub.chip_version == 0x51 &&
reg != 0x130d && reg != 0x1311 && reg != 0x60081d)
NV_WARN(dev, "=== C51 misaligned reg 0x%08X not verified ===\n",
reg);
- /* Trust the init scripts on G80 */
- if (dev_priv->card_type >= NV_50)
- return 1;
-
- #define WITHIN(x, y, z) ((x >= y) && (x < y + z))
- if (WITHIN(reg, NV_PMC_OFFSET, NV_PMC_SIZE))
- return 1;
- if (WITHIN(reg, NV_PBUS_OFFSET, NV_PBUS_SIZE))
- return 1;
- if (WITHIN(reg, NV_PFIFO_OFFSET, NV_PFIFO_SIZE))
- return 1;
- if (dev_priv->VBIOS.pub.chip_version >= 0x30 &&
- (WITHIN(reg, 0x4000, 0x600) || reg == 0x00004600))
- return 1;
- if (dev_priv->VBIOS.pub.chip_version >= 0x40 &&
- WITHIN(reg, 0xc000, 0x48))
- return 1;
- if (dev_priv->VBIOS.pub.chip_version >= 0x17 && reg == 0x0000d204)
- return 1;
- if (dev_priv->VBIOS.pub.chip_version >= 0x40) {
- if (reg == 0x00011014 || reg == 0x00020328)
- return 1;
- if (WITHIN(reg, 0x88000, NV_PBUS_SIZE)) /* new PBUS */
- return 1;
+ if (reg >= (8*1024*1024)) {
+ NV_ERROR(dev, "=== reg 0x%08x out of mapped bounds ===\n", reg);
+ return 0;
}
- if (WITHIN(reg, NV_PFB_OFFSET, NV_PFB_SIZE))
- return 1;
- if (WITHIN(reg, NV_PEXTDEV_OFFSET, NV_PEXTDEV_SIZE))
- return 1;
- if (WITHIN(reg, NV_PCRTC0_OFFSET, NV_PCRTC0_SIZE * 2))
- return 1;
- if (WITHIN(reg, NV_PRAMDAC0_OFFSET, NV_PRAMDAC0_SIZE * 2))
- return 1;
- if (dev_priv->VBIOS.pub.chip_version >= 0x17 && reg == 0x0070fff0)
- return 1;
- if (dev_priv->VBIOS.pub.chip_version == 0x51 &&
- WITHIN(reg, NV_PRAMIN_OFFSET, NV_PRAMIN_SIZE))
- return 1;
- #undef WITHIN
-
- NV_ERROR(dev, "========== unknown reg 0x%08X ==========\n", reg);
- return 0;
+ return 1;
}
static bool
@@ -1906,7 +1865,7 @@ init_compute_mem(struct nvbios *bios, uint16_t offset, struct init_exec *iexec)
struct drm_nouveau_private *dev_priv = bios->dev->dev_private;
- if (dev_priv->card_type >= NV_50)
+ if (dev_priv->card_type >= NV_40)
return 1;
/*
@@ -3196,16 +3155,25 @@ static int call_lvds_manufacturer_script(struct drm_device *dev, struct dcb_entr
}
#ifdef __powerpc__
/* Powerbook specific quirks */
- if (script == LVDS_RESET && ((dev->pci_device & 0xffff) == 0x0179 || (dev->pci_device & 0xffff) == 0x0329))
- nv_write_tmds(dev, dcbent->or, 0, 0x02, 0x72);
- if ((dev->pci_device & 0xffff) == 0x0179 || (dev->pci_device & 0xffff) == 0x0189 || (dev->pci_device & 0xffff) == 0x0329) {
- if (script == LVDS_PANEL_ON) {
- bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL, bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL) | (1 << 31));
- bios_wr32(bios, NV_PCRTC_GPIO_EXT, bios_rd32(bios, NV_PCRTC_GPIO_EXT) | 1);
- }
- if (script == LVDS_PANEL_OFF) {
- bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL, bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL) & ~(1 << 31));
- bios_wr32(bios, NV_PCRTC_GPIO_EXT, bios_rd32(bios, NV_PCRTC_GPIO_EXT) & ~3);
+ if ((dev->pci_device & 0xffff) == 0x0179 ||
+ (dev->pci_device & 0xffff) == 0x0189 ||
+ (dev->pci_device & 0xffff) == 0x0329) {
+ if (script == LVDS_RESET) {
+ nv_write_tmds(dev, dcbent->or, 0, 0x02, 0x72);
+
+ } else if (script == LVDS_PANEL_ON) {
+ bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL,
+ bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL)
+ | (1 << 31));
+ bios_wr32(bios, NV_PCRTC_GPIO_EXT,
+ bios_rd32(bios, NV_PCRTC_GPIO_EXT) | 1);
+
+ } else if (script == LVDS_PANEL_OFF) {
+ bios_wr32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL,
+ bios_rd32(bios, NV_PBUS_DEBUG_DUALHEAD_CTL)
+ & ~(1 << 31));
+ bios_wr32(bios, NV_PCRTC_GPIO_EXT,
+ bios_rd32(bios, NV_PCRTC_GPIO_EXT) & ~3);
}
}
#endif
@@ -3797,7 +3765,6 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
*/
struct drm_nouveau_private *dev_priv = dev->dev_private;
- struct init_exec iexec = {true, false};
struct nvbios *bios = &dev_priv->VBIOS;
uint8_t *table = &bios->data[bios->display.script_table_ptr];
uint8_t *otable = NULL;
@@ -3877,8 +3844,6 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
}
}
- bios->display.output = dcbent;
-
if (pxclk == 0) {
script = ROM16(otable[6]);
if (!script) {
@@ -3887,7 +3852,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
}
NV_TRACE(dev, "0x%04X: parsing output script 0\n", script);
- parse_init_table(bios, script, &iexec);
+ nouveau_bios_run_init_table(dev, script, dcbent);
} else
if (pxclk == -1) {
script = ROM16(otable[8]);
@@ -3897,7 +3862,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
}
NV_TRACE(dev, "0x%04X: parsing output script 1\n", script);
- parse_init_table(bios, script, &iexec);
+ nouveau_bios_run_init_table(dev, script, dcbent);
} else
if (pxclk == -2) {
if (table[4] >= 12)
@@ -3910,7 +3875,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
}
NV_TRACE(dev, "0x%04X: parsing output script 2\n", script);
- parse_init_table(bios, script, &iexec);
+ nouveau_bios_run_init_table(dev, script, dcbent);
} else
if (pxclk > 0) {
script = ROM16(otable[table[4] + i*6 + 2]);
@@ -3922,7 +3887,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
}
NV_TRACE(dev, "0x%04X: parsing clock script 0\n", script);
- parse_init_table(bios, script, &iexec);
+ nouveau_bios_run_init_table(dev, script, dcbent);
} else
if (pxclk < 0) {
script = ROM16(otable[table[4] + i*6 + 4]);
@@ -3934,7 +3899,7 @@ nouveau_bios_run_display_table(struct drm_device *dev, struct dcb_entry *dcbent,
}
NV_TRACE(dev, "0x%04X: parsing clock script 1\n", script);
- parse_init_table(bios, script, &iexec);
+ nouveau_bios_run_init_table(dev, script, dcbent);
}
return 0;
@@ -5434,52 +5399,49 @@ static bool
parse_dcb15_entry(struct drm_device *dev, struct parsed_dcb *dcb,
uint32_t conn, uint32_t conf, struct dcb_entry *entry)
{
- if (conn != 0xf0003f00 && conn != 0xf2247f10 && conn != 0xf2204001 &&
- conn != 0xf2204301 && conn != 0xf2204311 && conn != 0xf2208001 &&
- conn != 0xf2244001 && conn != 0xf2244301 && conn != 0xf2244311 &&
- conn != 0xf4204011 && conn != 0xf4208011 && conn != 0xf4248011 &&
- conn != 0xf2045ff2 && conn != 0xf2045f14 && conn != 0xf207df14 &&
- conn != 0xf2205004 && conn != 0xf2209004) {
- NV_ERROR(dev, "Unknown DCB 1.5 entry, please report\n");
-
- /* cause output setting to fail for !TV, so message is seen */
- if ((conn & 0xf) != 0x1)
- dcb->entries = 0;
-
- return false;
- }
- /* most of the below is a "best guess" atm */
- entry->type = conn & 0xf;
- if (entry->type == 2)
- /* another way of specifying straps based lvds... */
+ switch (conn & 0x0000000f) {
+ case 0:
+ entry->type = OUTPUT_ANALOG;
+ break;
+ case 1:
+ entry->type = OUTPUT_TV;
+ break;
+ case 2:
+ case 3:
entry->type = OUTPUT_LVDS;
- if (entry->type == 4) { /* digital */
- if (conn & 0x10)
- entry->type = OUTPUT_LVDS;
- else
+ break;
+ case 4:
+ switch ((conn & 0x000000f0) >> 4) {
+ case 0:
entry->type = OUTPUT_TMDS;
+ break;
+ case 1:
+ entry->type = OUTPUT_LVDS;
+ break;
+ default:
+ NV_ERROR(dev, "Unknown DCB subtype 4/%d\n",
+ (conn & 0x000000f0) >> 4);
+ return false;
+ }
+ break;
+ default:
+ NV_ERROR(dev, "Unknown DCB type %d\n", conn & 0x0000000f);
+ return false;
}
- /* what's in bits 5-13? could be some encoder maker thing, in tv case */
- entry->i2c_index = (conn >> 14) & 0xf;
- /* raw heads field is in range 0-1, so move to 1-2 */
- entry->heads = ((conn >> 18) & 0x7) + 1;
- entry->location = (conn >> 21) & 0xf;
- /* unused: entry->bus = (conn >> 25) & 0x7; */
- /* set or to be same as heads -- hopefully safe enough */
- entry->or = entry->heads;
+
+ entry->i2c_index = (conn & 0x0003c000) >> 14;
+ entry->heads = ((conn & 0x001c0000) >> 18) + 1;
+ entry->or = entry->heads; /* same as heads, hopefully safe enough */
+ entry->location = (conn & 0x01e00000) >> 21;
+ entry->bus = (conn & 0x0e000000) >> 25;
entry->duallink_possible = false;
switch (entry->type) {
case OUTPUT_ANALOG:
entry->crtconf.maxfreq = (conf & 0xffff) * 10;
break;
- case OUTPUT_LVDS:
- /*
- * This is probably buried in conn's unknown bits.
- * This will upset EDID-ful models, if they exist
- */
- entry->lvdsconf.use_straps_for_mode = true;
- entry->lvdsconf.use_power_scripts = true;
+ case OUTPUT_TV:
+ entry->tvconf.has_component_output = false;
break;
case OUTPUT_TMDS:
/*
@@ -5488,8 +5450,12 @@ parse_dcb15_entry(struct drm_device *dev, struct parsed_dcb *dcb,
*/
fabricate_vga_output(dcb, entry->i2c_index, entry->heads);
break;
- case OUTPUT_TV:
- entry->tvconf.has_component_output = false;
+ case OUTPUT_LVDS:
+ if ((conn & 0x00003f00) != 0x10)
+ entry->lvdsconf.use_straps_for_mode = true;
+ entry->lvdsconf.use_power_scripts = true;
+ break;
+ default:
break;
}
@@ -5564,11 +5530,13 @@ void merge_like_dcb_entries(struct drm_device *dev, struct parsed_dcb *dcb)
dcb->entries = newentries;
}
-static int parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads)
+static int
+parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool twoHeads)
{
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
struct bios_parsed_dcb *bdcb = &bios->bdcb;
struct parsed_dcb *dcb;
- uint16_t dcbptr, i2ctabptr = 0;
+ uint16_t dcbptr = 0, i2ctabptr = 0;
uint8_t *dcbtable;
uint8_t headerlen = 0x4, entries = DCB_MAX_NUM_ENTRIES;
bool configblock = true;
@@ -5579,16 +5547,18 @@ static int parse_dcb_table(struct drm_device *dev, struct nvbios *bios, bool two
dcb->entries = 0;
/* get the offset from 0x36 */
- dcbptr = ROM16(bios->data[0x36]);
+ if (dev_priv->card_type > NV_04) {
+ dcbptr = ROM16(bios->data[0x36]);
+ if (dcbptr == 0x0000)
+ NV_WARN(dev, "No output data (DCB) found in BIOS\n");
+ }
+ /* this situation likely means a really old card, pre DCB */
if (dcbptr == 0x0) {
- NV_WARN(dev, "No output data (DCB) found in BIOS, "
- "assuming a CRT output exists\n");
- /* this situation likely means a really old card, pre DCB */
+ NV_INFO(dev, "Assuming a CRT output exists\n");
fabricate_vga_output(dcb, LEGACY_I2C_CRT, 1);
- if (nv04_tv_identify(dev,
- bios->legacy.i2c_indices.tv) >= 0)
+ if (nv04_tv_identify(dev, bios->legacy.i2c_indices.tv) >= 0)
fabricate_tv_output(dcb, twoHeads);
return 0;
@@ -5892,9 +5862,11 @@ nouveau_bios_run_init_table(struct drm_device *dev, uint16_t table,
struct nvbios *bios = &dev_priv->VBIOS;
struct init_exec iexec = { true, false };
+ mutex_lock(&bios->lock);
bios->display.output = dcbent;
parse_init_table(bios, table, &iexec);
bios->display.output = NULL;
+ mutex_unlock(&bios->lock);
}
static bool NVInitVBIOS(struct drm_device *dev)
@@ -5903,6 +5875,7 @@ static bool NVInitVBIOS(struct drm_device *dev)
struct nvbios *bios = &dev_priv->VBIOS;
memset(bios, 0, sizeof(struct nvbios));
+ mutex_init(&bios->lock);
bios->dev = dev;
if (!NVShadowVBIOS(dev, bios->data))
diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.h b/drivers/gpu/drm/nouveau/nouveau_bios.h
index 058e98c76d8..fd94bd6dc26 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bios.h
+++ b/drivers/gpu/drm/nouveau/nouveau_bios.h
@@ -205,6 +205,8 @@ struct nvbios {
struct drm_device *dev;
struct nouveau_bios_info pub;
+ struct mutex lock;
+
uint8_t data[NV_PROM_SIZE];
unsigned int length;
bool execute;
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c
index 0cad6d834eb..028719fddf7 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo.c
@@ -33,10 +33,13 @@
#include "nouveau_drv.h"
#include "nouveau_dma.h"
+#include <linux/log2.h>
+
static void
nouveau_bo_del_ttm(struct ttm_buffer_object *bo)
{
struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev);
+ struct drm_device *dev = dev_priv->dev;
struct nouveau_bo *nvbo = nouveau_bo(bo);
ttm_bo_kunmap(&nvbo->kmap);
@@ -44,12 +47,87 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo)
if (unlikely(nvbo->gem))
DRM_ERROR("bo %p still attached to GEM object\n", bo);
+ if (nvbo->tile)
+ nv10_mem_expire_tiling(dev, nvbo->tile, NULL);
+
spin_lock(&dev_priv->ttm.bo_list_lock);
list_del(&nvbo->head);
spin_unlock(&dev_priv->ttm.bo_list_lock);
kfree(nvbo);
}
+static void
+nouveau_bo_fixup_align(struct drm_device *dev,
+ uint32_t tile_mode, uint32_t tile_flags,
+ int *align, int *size)
+{
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+
+ /*
+ * Some of the tile_flags have a periodic structure of N*4096 bytes,
+ * align to to that as well as the page size. Align the size to the
+ * appropriate boundaries. This does imply that sizes are rounded up
+ * 3-7 pages, so be aware of this and do not waste memory by allocating
+ * many small buffers.
+ */
+ if (dev_priv->card_type == NV_50) {
+ uint32_t block_size = nouveau_mem_fb_amount(dev) >> 15;
+ int i;
+
+ switch (tile_flags) {
+ case 0x1800:
+ case 0x2800:
+ case 0x4800:
+ case 0x7a00:
+ if (is_power_of_2(block_size)) {
+ for (i = 1; i < 10; i++) {
+ *align = 12 * i * block_size;
+ if (!(*align % 65536))
+ break;
+ }
+ } else {
+ for (i = 1; i < 10; i++) {
+ *align = 8 * i * block_size;
+ if (!(*align % 65536))
+ break;
+ }
+ }
+ *size = roundup(*size, *align);
+ break;
+ default:
+ break;
+ }
+
+ } else {
+ if (tile_mode) {
+ if (dev_priv->chipset >= 0x40) {
+ *align = 65536;
+ *size = roundup(*size, 64 * tile_mode);
+
+ } else if (dev_priv->chipset >= 0x30) {
+ *align = 32768;
+ *size = roundup(*size, 64 * tile_mode);
+
+ } else if (dev_priv->chipset >= 0x20) {
+ *align = 16384;
+ *size = roundup(*size, 64 * tile_mode);
+
+ } else if (dev_priv->chipset >= 0x10) {
+ *align = 16384;
+ *size = roundup(*size, 32 * tile_mode);
+ }
+ }
+ }
+
+ /* ALIGN works only on powers of two. */
+ *size = roundup(*size, PAGE_SIZE);
+
+ if (dev_priv->card_type == NV_50) {
+ *size = roundup(*size, 65536);
+ *align = max(65536, *align);
+ }
+}
+
int
nouveau_bo_new(struct drm_device *dev, struct nouveau_channel *chan,
int size, int align, uint32_t flags, uint32_t tile_mode,
@@ -58,7 +136,7 @@ nouveau_bo_new(struct drm_device *dev, struct nouveau_channel *chan,
{
struct drm_nouveau_private *dev_priv = dev->dev_private;
struct nouveau_bo *nvbo;
- int ret, n = 0;
+ int ret = 0;
nvbo = kzalloc(sizeof(struct nouveau_bo), GFP_KERNEL);
if (!nvbo)
@@ -70,59 +148,14 @@ nouveau_bo_new(struct drm_device *dev, struct nouveau_channel *chan,
nvbo->tile_mode = tile_mode;
nvbo->tile_flags = tile_flags;
- /*
- * Some of the tile_flags have a periodic structure of N*4096 bytes,
- * align to to that as well as the page size. Overallocate memory to
- * avoid corruption of other buffer objects.
- */
- switch (tile_flags) {
- case 0x1800:
- case 0x2800:
- case 0x4800:
- case 0x7a00:
- if (dev_priv->chipset >= 0xA0) {
- /* This is based on high end cards with 448 bits
- * memory bus, could be different elsewhere.*/
- size += 6 * 28672;
- /* 8 * 28672 is the actual alignment requirement,
- * but we must also align to page size. */
- align = 2 * 8 * 28672;
- } else if (dev_priv->chipset >= 0x90) {
- size += 3 * 16384;
- align = 12 * 16834;
- } else {
- size += 3 * 8192;
- /* 12 * 8192 is the actual alignment requirement,
- * but we must also align to page size. */
- align = 2 * 12 * 8192;
- }
- break;
- default:
- break;
- }
-
+ nouveau_bo_fixup_align(dev, tile_mode, tile_flags, &align, &size);
align >>= PAGE_SHIFT;
- size = (size + (PAGE_SIZE - 1)) & ~(PAGE_SIZE - 1);
- if (dev_priv->card_type == NV_50) {
- size = (size + 65535) & ~65535;
- if (align < (65536 / PAGE_SIZE))
- align = (65536 / PAGE_SIZE);
- }
-
- if (flags & TTM_PL_FLAG_VRAM)
- nvbo->placements[n++] = TTM_PL_FLAG_VRAM | TTM_PL_MASK_CACHING;
- if (flags & TTM_PL_FLAG_TT)
- nvbo->placements[n++] = TTM_PL_FLAG_TT | TTM_PL_MASK_CACHING;
nvbo->placement.fpfn = 0;
nvbo->placement.lpfn = mappable ? dev_priv->fb_mappable_pages : 0;
- nvbo->placement.placement = nvbo->placements;
- nvbo->placement.busy_placement = nvbo->placements;
- nvbo->placement.num_placement = n;
- nvbo->placement.num_busy_placement = n;
+ nouveau_bo_placement_set(nvbo, flags);
nvbo->channel = chan;
- nouveau_bo_placement_set(nvbo, flags);
ret = ttm_bo_init(&dev_priv->ttm.bdev, &nvbo->bo, size,
ttm_bo_type_device, &nvbo->placement, align, 0,
false, NULL, size, nouveau_bo_del_ttm);
@@ -421,6 +454,7 @@ nouveau_bo_evict_flags(struct ttm_buffer_object *bo, struct ttm_placement *pl)
/* GPU-assisted copy using NV_MEMORY_TO_MEMORY_FORMAT, can access
* TTM_PL_{VRAM,TT} directly.
*/
+
static int
nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan,
struct nouveau_bo *nvbo, bool evict, bool no_wait,
@@ -435,6 +469,8 @@ nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan,
ret = ttm_bo_move_accel_cleanup(&nvbo->bo, fence, NULL,
evict, no_wait, new_mem);
+ if (nvbo->channel && nvbo->channel != chan)
+ ret = nouveau_fence_wait(fence, NULL, false, false);
nouveau_fence_unref((void *)&fence);
return ret;
}
@@ -455,11 +491,12 @@ nouveau_bo_mem_ctxdma(struct nouveau_bo *nvbo, struct nouveau_channel *chan,
}
static int
-nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, int no_wait,
- struct ttm_mem_reg *old_mem, struct ttm_mem_reg *new_mem)
+nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, bool intr,
+ int no_wait, struct ttm_mem_reg *new_mem)
{
struct nouveau_bo *nvbo = nouveau_bo(bo);
struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev);
+ struct ttm_mem_reg *old_mem = &bo->mem;
struct nouveau_channel *chan;
uint64_t src_offset, dst_offset;
uint32_t page_count;
@@ -547,7 +584,7 @@ nouveau_bo_move_flipd(struct ttm_buffer_object *bo, bool evict, bool intr,
placement.fpfn = placement.lpfn = 0;
placement.num_placement = placement.num_busy_placement = 1;
- placement.placement = &placement_memtype;
+ placement.placement = placement.busy_placement = &placement_memtype;
tmp_mem = *new_mem;
tmp_mem.mm_node = NULL;
@@ -559,7 +596,7 @@ nouveau_bo_move_flipd(struct ttm_buffer_object *bo, bool evict, bool intr,
if (ret)
goto out;
- ret = nouveau_bo_move_m2mf(bo, true, no_wait, &bo->mem, &tmp_mem);
+ ret = nouveau_bo_move_m2mf(bo, true, intr, no_wait, &tmp_mem);
if (ret)
goto out;
@@ -585,7 +622,7 @@ nouveau_bo_move_flips(struct ttm_buffer_object *bo, bool evict, bool intr,
placement.fpfn = placement.lpfn = 0;
placement.num_placement = placement.num_busy_placement = 1;
- placement.placement = &placement_memtype;
+ placement.placement = placement.busy_placement = &placement_memtype;
tmp_mem = *new_mem;
tmp_mem.mm_node = NULL;
@@ -597,7 +634,7 @@ nouveau_bo_move_flips(struct ttm_buffer_object *bo, bool evict, bool intr,
if (ret)
goto out;
- ret = nouveau_bo_move_m2mf(bo, true, no_wait, &bo->mem, new_mem);
+ ret = nouveau_bo_move_m2mf(bo, evict, intr, no_wait, new_mem);
if (ret)
goto out;
@@ -612,52 +649,106 @@ out:
}
static int
-nouveau_bo_move(struct ttm_buffer_object *bo, bool evict, bool intr,
- bool no_wait, struct ttm_mem_reg *new_mem)
+nouveau_bo_vm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *new_mem,
+ struct nouveau_tile_reg **new_tile)
{
struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev);
- struct nouveau_bo *nvbo = nouveau_bo(bo);
struct drm_device *dev = dev_priv->dev;
- struct ttm_mem_reg *old_mem = &bo->mem;
+ struct nouveau_bo *nvbo = nouveau_bo(bo);
+ uint64_t offset;
int ret;
- if (dev_priv->card_type == NV_50 && new_mem->mem_type == TTM_PL_VRAM &&
- !nvbo->no_vm) {
- uint64_t offset = new_mem->mm_node->start << PAGE_SHIFT;
+ if (nvbo->no_vm || new_mem->mem_type != TTM_PL_VRAM) {
+ /* Nothing to do. */
+ *new_tile = NULL;
+ return 0;
+ }
+
+ offset = new_mem->mm_node->start << PAGE_SHIFT;
+ if (dev_priv->card_type == NV_50) {
ret = nv50_mem_vm_bind_linear(dev,
offset + dev_priv->vm_vram_base,
new_mem->size, nvbo->tile_flags,
offset);
if (ret)
return ret;
+
+ } else if (dev_priv->card_type >= NV_10) {
+ *new_tile = nv10_mem_set_tiling(dev, offset, new_mem->size,
+ nvbo->tile_mode);
+ }
+
+ return 0;
+}
+
+static void
+nouveau_bo_vm_cleanup(struct ttm_buffer_object *bo,
+ struct nouveau_tile_reg *new_tile,
+ struct nouveau_tile_reg **old_tile)
+{
+ struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev);
+ struct drm_device *dev = dev_priv->dev;
+
+ if (dev_priv->card_type >= NV_10 &&
+ dev_priv->card_type < NV_50) {
+ if (*old_tile)
+ nv10_mem_expire_tiling(dev, *old_tile, bo->sync_obj);
+
+ *old_tile = new_tile;
}
+}
+
+static int
+nouveau_bo_move(struct ttm_buffer_object *bo, bool evict, bool intr,
+ bool no_wait, struct ttm_mem_reg *new_mem)
+{
+ struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev);
+ struct nouveau_bo *nvbo = nouveau_bo(bo);
+ struct ttm_mem_reg *old_mem = &bo->mem;
+ struct nouveau_tile_reg *new_tile = NULL;
+ int ret = 0;
+
+ ret = nouveau_bo_vm_bind(bo, new_mem, &new_tile);
+ if (ret)
+ return ret;
+ /* Software copy if the card isn't up and running yet. */
if (dev_priv->init_state != NOUVEAU_CARD_INIT_DONE ||
- !dev_priv->channel)
- return ttm_bo_move_memcpy(bo, evict, no_wait, new_mem);
+ !dev_priv->channel) {
+ ret = ttm_bo_move_memcpy(bo, evict, no_wait, new_mem);
+ goto out;
+ }
+ /* Fake bo copy. */
if (old_mem->mem_type == TTM_PL_SYSTEM && !bo->ttm) {
BUG_ON(bo->mem.mm_node != NULL);
bo->mem = *new_mem;
new_mem->mm_node = NULL;
- return 0;
+ goto out;
}
- if (new_mem->mem_type == TTM_PL_SYSTEM) {
- if (old_mem->mem_type == TTM_PL_SYSTEM)
- return ttm_bo_move_memcpy(bo, evict, no_wait, new_mem);
- if (nouveau_bo_move_flipd(bo, evict, intr, no_wait, new_mem))
- return ttm_bo_move_memcpy(bo, evict, no_wait, new_mem);
- } else if (old_mem->mem_type == TTM_PL_SYSTEM) {
- if (nouveau_bo_move_flips(bo, evict, intr, no_wait, new_mem))
- return ttm_bo_move_memcpy(bo, evict, no_wait, new_mem);
- } else {
- if (nouveau_bo_move_m2mf(bo, evict, no_wait, old_mem, new_mem))
- return ttm_bo_move_memcpy(bo, evict, no_wait, new_mem);
- }
+ /* Hardware assisted copy. */
+ if (new_mem->mem_type == TTM_PL_SYSTEM)
+ ret = nouveau_bo_move_flipd(bo, evict, intr, no_wait, new_mem);
+ else if (old_mem->mem_type == TTM_PL_SYSTEM)
+ ret = nouveau_bo_move_flips(bo, evict, intr, no_wait, new_mem);
+ else
+ ret = nouveau_bo_move_m2mf(bo, evict, intr, no_wait, new_mem);
- return 0;
+ if (!ret)
+ goto out;
+
+ /* Fallback to software copy. */
+ ret = ttm_bo_move_memcpy(bo, evict, no_wait, new_mem);
+
+out:
+ if (ret)
+ nouveau_bo_vm_cleanup(bo, NULL, &new_tile);
+ else
+ nouveau_bo_vm_cleanup(bo, new_tile, &nvbo->tile);
+
+ return ret;
}
static int
diff --git a/drivers/gpu/drm/nouveau/nouveau_channel.c b/drivers/gpu/drm/nouveau/nouveau_channel.c
index 9aaa972f882..2281f99da7f 100644
--- a/drivers/gpu/drm/nouveau/nouveau_channel.c
+++ b/drivers/gpu/drm/nouveau/nouveau_channel.c
@@ -158,6 +158,8 @@ nouveau_channel_alloc(struct drm_device *dev, struct nouveau_channel **chan_ret,
return ret;
}
+ nouveau_dma_pre_init(chan);
+
/* Locate channel's user control regs */
if (dev_priv->card_type < NV_40)
user = NV03_USER(channel);
@@ -235,47 +237,6 @@ nouveau_channel_alloc(struct drm_device *dev, struct nouveau_channel **chan_ret,
return 0;
}
-int
-nouveau_channel_idle(struct nouveau_channel *chan)
-{
- struct drm_device *dev = chan->dev;
- struct drm_nouveau_private *dev_priv = dev->dev_private;
- struct nouveau_engine *engine = &dev_priv->engine;
- uint32_t caches;
- int idle;
-
- if (!chan) {
- NV_ERROR(dev, "no channel...\n");
- return 1;
- }
-
- caches = nv_rd32(dev, NV03_PFIFO_CACHES);
- nv_wr32(dev, NV03_PFIFO_CACHES, caches & ~1);
-
- if (engine->fifo.channel_id(dev) != chan->id) {
- struct nouveau_gpuobj *ramfc =
- chan->ramfc ? chan->ramfc->gpuobj : NULL;
-
- if (!ramfc) {
- NV_ERROR(dev, "No RAMFC for channel %d\n", chan->id);
- return 1;
- }
-
- engine->instmem.prepare_access(dev, false);
- if (nv_ro32(dev, ramfc, 0) != nv_ro32(dev, ramfc, 1))
- idle = 0;
- else
- idle = 1;
- engine->instmem.finish_access(dev);
- } else {
- idle = (nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_GET) ==
- nv_rd32(dev, NV04_PFIFO_CACHE1_DMA_PUT));
- }
-
- nv_wr32(dev, NV03_PFIFO_CACHES, caches);
- return idle;
-}
-
/* stops a fifo */
void
nouveau_channel_free(struct nouveau_channel *chan)
@@ -317,12 +278,11 @@ nouveau_channel_free(struct nouveau_channel *chan)
/* Ensure the channel is no longer active on the GPU */
pfifo->reassign(dev, false);
- if (pgraph->channel(dev) == chan) {
- pgraph->fifo_access(dev, false);
+ pgraph->fifo_access(dev, false);
+ if (pgraph->channel(dev) == chan)
pgraph->unload_context(dev);
- pgraph->fifo_access(dev, true);
- }
pgraph->destroy_context(chan);
+ pgraph->fifo_access(dev, true);
if (pfifo->channel_id(dev) == chan->id) {
pfifo->disable(dev);
@@ -414,7 +374,9 @@ nouveau_ioctl_fifo_alloc(struct drm_device *dev, void *data,
init->subchan[0].grclass = 0x0039;
else
init->subchan[0].grclass = 0x5039;
- init->nr_subchan = 1;
+ init->subchan[1].handle = NvSw;
+ init->subchan[1].grclass = NV_SW;
+ init->nr_subchan = 2;
/* Named memory object area */
ret = drm_gem_handle_create(file_priv, chan->notifier_bo->gem,
diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c
index 5a10deb8bdb..d2f63353ea9 100644
--- a/drivers/gpu/drm/nouveau/nouveau_connector.c
+++ b/drivers/gpu/drm/nouveau/nouveau_connector.c
@@ -24,9 +24,12 @@
*
*/
+#include <acpi/button.h>
+
#include "drmP.h"
#include "drm_edid.h"
#include "drm_crtc_helper.h"
+
#include "nouveau_reg.h"
#include "nouveau_drv.h"
#include "nouveau_encoder.h"
@@ -83,14 +86,17 @@ nouveau_encoder_connector_get(struct nouveau_encoder *encoder)
static void
nouveau_connector_destroy(struct drm_connector *drm_connector)
{
- struct nouveau_connector *connector = nouveau_connector(drm_connector);
- struct drm_device *dev = connector->base.dev;
+ struct nouveau_connector *nv_connector =
+ nouveau_connector(drm_connector);
+ struct drm_device *dev;
- NV_DEBUG_KMS(dev, "\n");
-
- if (!connector)
+ if (!nv_connector)
return;
+ dev = nv_connector->base.dev;
+ NV_DEBUG_KMS(dev, "\n");
+
+ kfree(nv_connector->edid);
drm_sysfs_connector_remove(drm_connector);
drm_connector_cleanup(drm_connector);
kfree(drm_connector);
@@ -233,10 +239,21 @@ nouveau_connector_detect(struct drm_connector *connector)
if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)
nv_encoder = find_encoder_by_type(connector, OUTPUT_LVDS);
if (nv_encoder && nv_connector->native_mode) {
+#ifdef CONFIG_ACPI
+ if (!nouveau_ignorelid && !acpi_lid_open())
+ return connector_status_disconnected;
+#endif
nouveau_connector_set_encoder(connector, nv_encoder);
return connector_status_connected;
}
+ /* Cleanup the previous EDID block. */
+ if (nv_connector->edid) {
+ drm_mode_connector_update_edid_property(connector, NULL);
+ kfree(nv_connector->edid);
+ nv_connector->edid = NULL;
+ }
+
i2c = nouveau_connector_ddc_detect(connector, &nv_encoder);
if (i2c) {
nouveau_connector_ddc_prepare(connector, &flags);
@@ -247,7 +264,7 @@ nouveau_connector_detect(struct drm_connector *connector)
if (!nv_connector->edid) {
NV_ERROR(dev, "DDC responded, but no EDID for %s\n",
drm_get_connector_name(connector));
- return connector_status_disconnected;
+ goto detect_analog;
}
if (nv_encoder->dcb->type == OUTPUT_DP &&
@@ -281,6 +298,7 @@ nouveau_connector_detect(struct drm_connector *connector)
return connector_status_connected;
}
+detect_analog:
nv_encoder = find_encoder_by_type(connector, OUTPUT_ANALOG);
if (!nv_encoder)
nv_encoder = find_encoder_by_type(connector, OUTPUT_TV);
@@ -687,8 +705,12 @@ nouveau_connector_create_lvds(struct drm_device *dev,
*/
if (!nv_connector->edid && !nv_connector->native_mode &&
!dev_priv->VBIOS.pub.fp_no_ddc) {
- nv_connector->edid =
+ struct edid *edid =
(struct edid *)nouveau_bios_embedded_edid(dev);
+ if (edid) {
+ nv_connector->edid = kmalloc(EDID_LENGTH, GFP_KERNEL);
+ *(nv_connector->edid) = *edid;
+ }
}
if (!nv_connector->edid)
diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.c b/drivers/gpu/drm/nouveau/nouveau_dma.c
index 703553687b2..50d9e67745a 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dma.c
+++ b/drivers/gpu/drm/nouveau/nouveau_dma.c
@@ -29,12 +29,22 @@
#include "nouveau_drv.h"
#include "nouveau_dma.h"
+void
+nouveau_dma_pre_init(struct nouveau_channel *chan)
+{
+ chan->dma.max = (chan->pushbuf_bo->bo.mem.size >> 2) - 2;
+ chan->dma.put = 0;
+ chan->dma.cur = chan->dma.put;
+ chan->dma.free = chan->dma.max - chan->dma.cur;
+}
+
int
nouveau_dma_init(struct nouveau_channel *chan)
{
struct drm_device *dev = chan->dev;
struct drm_nouveau_private *dev_priv = dev->dev_private;
struct nouveau_gpuobj *m2mf = NULL;
+ struct nouveau_gpuobj *nvsw = NULL;
int ret, i;
/* Create NV_MEMORY_TO_MEMORY_FORMAT for buffer moves */
@@ -47,6 +57,15 @@ nouveau_dma_init(struct nouveau_channel *chan)
if (ret)
return ret;
+ /* Create an NV_SW object for various sync purposes */
+ ret = nouveau_gpuobj_sw_new(chan, NV_SW, &nvsw);
+ if (ret)
+ return ret;
+
+ ret = nouveau_gpuobj_ref_add(dev, chan, NvSw, nvsw, NULL);
+ if (ret)
+ return ret;
+
/* NV_MEMORY_TO_MEMORY_FORMAT requires a notifier object */
ret = nouveau_notifier_alloc(chan, NvNotify0, 32, &chan->m2mf_ntfy);
if (ret)
@@ -64,12 +83,6 @@ nouveau_dma_init(struct nouveau_channel *chan)
return ret;
}
- /* Initialise DMA vars */
- chan->dma.max = (chan->pushbuf_bo->bo.mem.size >> 2) - 2;
- chan->dma.put = 0;
- chan->dma.cur = chan->dma.put;
- chan->dma.free = chan->dma.max - chan->dma.cur;
-
/* Insert NOPS for NOUVEAU_DMA_SKIPS */
ret = RING_SPACE(chan, NOUVEAU_DMA_SKIPS);
if (ret)
@@ -87,6 +100,13 @@ nouveau_dma_init(struct nouveau_channel *chan)
BEGIN_RING(chan, NvSubM2MF, NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1);
OUT_RING(chan, NvNotify0);
+ /* Initialise NV_SW */
+ ret = RING_SPACE(chan, 2);
+ if (ret)
+ return ret;
+ BEGIN_RING(chan, NvSubSw, 0, 1);
+ OUT_RING(chan, NvSw);
+
/* Sit back and pray the channel works.. */
FIRE_RING(chan);
@@ -106,47 +126,52 @@ OUT_RINGp(struct nouveau_channel *chan, const void *data, unsigned nr_dwords)
chan->dma.cur += nr_dwords;
}
-static inline bool
-READ_GET(struct nouveau_channel *chan, uint32_t *get)
+/* Fetch and adjust GPU GET pointer
+ *
+ * Returns:
+ * value >= 0, the adjusted GET pointer
+ * -EINVAL if GET pointer currently outside main push buffer
+ * -EBUSY if timeout exceeded
+ */
+static inline int
+READ_GET(struct nouveau_channel *chan, uint32_t *prev_get, uint32_t *timeout)
{
uint32_t val;
val = nvchan_rd32(chan, chan->user_get);
- if (val < chan->pushbuf_base ||
- val >= chan->pushbuf_base + chan->pushbuf_bo->bo.mem.size) {
- /* meaningless to dma_wait() except to know whether the
- * GPU has stalled or not
- */
- *get = val;
- return false;
+
+ /* reset counter as long as GET is still advancing, this is
+ * to avoid misdetecting a GPU lockup if the GPU happens to
+ * just be processing an operation that takes a long time
+ */
+ if (val != *prev_get) {
+ *prev_get = val;
+ *timeout = 0;
+ }
+
+ if ((++*timeout & 0xff) == 0) {
+ DRM_UDELAY(1);
+ if (*timeout > 100000)
+ return -EBUSY;
}
- *get = (val - chan->pushbuf_base) >> 2;
- return true;
+ if (val < chan->pushbuf_base ||
+ val > chan->pushbuf_base + (chan->dma.max << 2))
+ return -EINVAL;
+
+ return (val - chan->pushbuf_base) >> 2;
}
int
nouveau_dma_wait(struct nouveau_channel *chan, int size)
{
- uint32_t get, prev_get = 0, cnt = 0;
- bool get_valid;
+ uint32_t prev_get = 0, cnt = 0;
+ int get;
while (chan->dma.free < size) {
- /* reset counter as long as GET is still advancing, this is
- * to avoid misdetecting a GPU lockup if the GPU happens to
- * just be processing an operation that takes a long time
- */
- get_valid = READ_GET(chan, &get);
- if (get != prev_get) {
- prev_get = get;
- cnt = 0;
- }
-
- if ((++cnt & 0xff) == 0) {
- DRM_UDELAY(1);
- if (cnt > 100000)
- return -EBUSY;
- }
+ get = READ_GET(chan, &prev_get, &cnt);
+ if (unlikely(get == -EBUSY))
+ return -EBUSY;
/* loop until we have a usable GET pointer. the value
* we read from the GPU may be outside the main ring if
@@ -157,7 +182,7 @@ nouveau_dma_wait(struct nouveau_channel *chan, int size)
* from the SKIPS area, so the code below doesn't have to deal
* with some fun corner cases.
*/
- if (!get_valid || get < NOUVEAU_DMA_SKIPS)
+ if (unlikely(get == -EINVAL) || get < NOUVEAU_DMA_SKIPS)
continue;
if (get <= chan->dma.cur) {
@@ -183,6 +208,19 @@ nouveau_dma_wait(struct nouveau_channel *chan, int size)
* after processing the currently pending commands.
*/
OUT_RING(chan, chan->pushbuf_base | 0x20000000);
+
+ /* wait for GET to depart from the skips area.
+ * prevents writing GET==PUT and causing a race
+ * condition that causes us to think the GPU is
+ * idle when it's not.
+ */
+ do {
+ get = READ_GET(chan, &prev_get, &cnt);
+ if (unlikely(get == -EBUSY))
+ return -EBUSY;
+ if (unlikely(get == -EINVAL))
+ continue;
+ } while (get <= NOUVEAU_DMA_SKIPS);
WRITE_PUT(NOUVEAU_DMA_SKIPS);
/* we're now submitting commands at the start of
diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.h b/drivers/gpu/drm/nouveau/nouveau_dma.h
index 04e85d8f757..dabfd655f93 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dma.h
+++ b/drivers/gpu/drm/nouveau/nouveau_dma.h
@@ -46,10 +46,11 @@
/* Hardcoded object assignments to subchannels (subchannel id). */
enum {
NvSubM2MF = 0,
- NvSub2D = 1,
- NvSubCtxSurf2D = 1,
- NvSubGdiRect = 2,
- NvSubImageBlit = 3
+ NvSubSw = 1,
+ NvSub2D = 2,
+ NvSubCtxSurf2D = 2,
+ NvSubGdiRect = 3,
+ NvSubImageBlit = 4
};
/* Object handles. */
@@ -67,6 +68,7 @@ enum {
NvClipRect = 0x8000000b,
NvGdiRect = 0x8000000c,
NvImageBlit = 0x8000000d,
+ NvSw = 0x8000000e,
/* G80+ display objects */
NvEvoVRAM = 0x01000000,
diff --git a/drivers/gpu/drm/nouveau/nouveau_dp.c b/drivers/gpu/drm/nouveau/nouveau_dp.c
index 9e2926c4857..f954ad93e81 100644
--- a/drivers/gpu/drm/nouveau/nouveau_dp.c
+++ b/drivers/gpu/drm/nouveau/nouveau_dp.c
@@ -490,7 +490,8 @@ nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr,
if (!nv_wait(NV50_AUXCH_CTRL(index), 0x00010000, 0x00000000)) {
NV_ERROR(dev, "expected bit 16 == 0, got 0x%08x\n",
nv_rd32(dev, NV50_AUXCH_CTRL(index)));
- return -EBUSY;
+ ret = -EBUSY;
+ goto out;
}
udelay(400);
@@ -502,6 +503,11 @@ nouveau_dp_auxch(struct nouveau_i2c_chan *auxch, int cmd, int addr,
}
if (cmd & 1) {
+ if ((stat & NV50_AUXCH_STAT_COUNT) != data_nr) {
+ ret = -EREMOTEIO;
+ goto out;
+ }
+
for (i = 0; i < 4; i++) {
data32[i] = nv_rd32(dev, NV50_AUXCH_DATA_IN(index, i));
NV_DEBUG_KMS(dev, "rd %d: 0x%08x\n", i, data32[i]);
diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.c b/drivers/gpu/drm/nouveau/nouveau_drv.c
index 06eb993e088..da3b93b8450 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.c
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.c
@@ -56,7 +56,7 @@ int nouveau_vram_pushbuf;
module_param_named(vram_pushbuf, nouveau_vram_pushbuf, int, 0400);
MODULE_PARM_DESC(vram_notify, "Force DMA notifiers to be in VRAM");
-int nouveau_vram_notify;
+int nouveau_vram_notify = 1;
module_param_named(vram_notify, nouveau_vram_notify, int, 0400);
MODULE_PARM_DESC(duallink, "Allow dual-link TMDS (>=GeForce 8)");
@@ -71,6 +71,18 @@ MODULE_PARM_DESC(uscript_tmds, "TMDS output script table ID (>=GeForce 8)");
int nouveau_uscript_tmds = -1;
module_param_named(uscript_tmds, nouveau_uscript_tmds, int, 0400);
+MODULE_PARM_DESC(ignorelid, "Ignore ACPI lid status");
+int nouveau_ignorelid = 0;
+module_param_named(ignorelid, nouveau_ignorelid, int, 0400);
+
+MODULE_PARM_DESC(noagp, "Disable all acceleration");
+int nouveau_noaccel = 0;
+module_param_named(noaccel, nouveau_noaccel, int, 0400);
+
+MODULE_PARM_DESC(noagp, "Disable fbcon acceleration");
+int nouveau_nofbaccel = 0;
+module_param_named(nofbaccel, nouveau_nofbaccel, int, 0400);
+
MODULE_PARM_DESC(tv_norm, "Default TV norm.\n"
"\t\tSupported: PAL, PAL-M, PAL-N, PAL-Nc, NTSC-M, NTSC-J,\n"
"\t\t\thd480i, hd480p, hd576i, hd576p, hd720p, hd1080i.\n"
diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h
index 5f8cbb79c49..1c15ef37b71 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.h
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.h
@@ -59,11 +59,19 @@ struct nouveau_grctx;
#define MAX_NUM_DCB_ENTRIES 16
#define NOUVEAU_MAX_CHANNEL_NR 128
+#define NOUVEAU_MAX_TILE_NR 15
#define NV50_VM_MAX_VRAM (2*1024*1024*1024ULL)
#define NV50_VM_BLOCK (512*1024*1024ULL)
#define NV50_VM_VRAM_NR (NV50_VM_MAX_VRAM / NV50_VM_BLOCK)
+struct nouveau_tile_reg {
+ struct nouveau_fence *fence;
+ uint32_t addr;
+ uint32_t size;
+ bool used;
+};
+
struct nouveau_bo {
struct ttm_buffer_object bo;
struct ttm_placement placement;
@@ -83,6 +91,7 @@ struct nouveau_bo {
uint32_t tile_mode;
uint32_t tile_flags;
+ struct nouveau_tile_reg *tile;
struct drm_gem_object *gem;
struct drm_file *cpu_filp;
@@ -277,8 +286,13 @@ struct nouveau_timer_engine {
};
struct nouveau_fb_engine {
+ int num_tiles;
+
int (*init)(struct drm_device *dev);
void (*takedown)(struct drm_device *dev);
+
+ void (*set_region_tiling)(struct drm_device *dev, int i, uint32_t addr,
+ uint32_t size, uint32_t pitch);
};
struct nouveau_fifo_engine {
@@ -292,6 +306,8 @@ struct nouveau_fifo_engine {
void (*disable)(struct drm_device *);
void (*enable)(struct drm_device *);
bool (*reassign)(struct drm_device *, bool enable);
+ bool (*cache_flush)(struct drm_device *dev);
+ bool (*cache_pull)(struct drm_device *dev, bool enable);
int (*channel_id)(struct drm_device *);
@@ -330,6 +346,9 @@ struct nouveau_pgraph_engine {
void (*destroy_context)(struct nouveau_channel *);
int (*load_context)(struct nouveau_channel *);
int (*unload_context)(struct drm_device *);
+
+ void (*set_region_tiling)(struct drm_device *dev, int i, uint32_t addr,
+ uint32_t size, uint32_t pitch);
};
struct nouveau_engine {
@@ -490,6 +509,8 @@ struct drm_nouveau_private {
void __iomem *ramin;
uint32_t ramin_size;
+ struct nouveau_bo *vga_ram;
+
struct workqueue_struct *wq;
struct work_struct irq_work;
@@ -548,6 +569,12 @@ struct drm_nouveau_private {
unsigned long sg_handle;
} gart_info;
+ /* nv10-nv40 tiling regions */
+ struct {
+ struct nouveau_tile_reg reg[NOUVEAU_MAX_TILE_NR];
+ spinlock_t lock;
+ } tile;
+
/* G8x/G9x virtual address space */
uint64_t vm_gart_base;
uint64_t vm_gart_size;
@@ -556,6 +583,7 @@ struct drm_nouveau_private {
uint64_t vm_end;
struct nouveau_gpuobj *vm_vram_pt[NV50_VM_VRAM_NR];
int vm_vram_pt_nr;
+ uint64_t vram_sys_base;
/* the mtrr covering the FB */
int fb_mtrr;
@@ -650,6 +678,9 @@ extern char *nouveau_tv_norm;
extern int nouveau_reg_debug;
extern char *nouveau_vbios;
extern int nouveau_ctxfw;
+extern int nouveau_ignorelid;
+extern int nouveau_nofbaccel;
+extern int nouveau_noaccel;
/* nouveau_state.c */
extern void nouveau_preclose(struct drm_device *dev, struct drm_file *);
@@ -685,6 +716,13 @@ extern void nouveau_mem_release(struct drm_file *, struct mem_block *heap);
extern int nouveau_mem_init(struct drm_device *);
extern int nouveau_mem_init_agp(struct drm_device *);
extern void nouveau_mem_close(struct drm_device *);
+extern struct nouveau_tile_reg *nv10_mem_set_tiling(struct drm_device *dev,
+ uint32_t addr,
+ uint32_t size,
+ uint32_t pitch);
+extern void nv10_mem_expire_tiling(struct drm_device *dev,
+ struct nouveau_tile_reg *tile,
+ struct nouveau_fence *fence);
extern int nv50_mem_vm_bind_linear(struct drm_device *, uint64_t virt,
uint32_t size, uint32_t flags,
uint64_t phys);
@@ -713,7 +751,6 @@ extern int nouveau_channel_alloc(struct drm_device *dev,
struct drm_file *file_priv,
uint32_t fb_ctxdma, uint32_t tt_ctxdma);
extern void nouveau_channel_free(struct nouveau_channel *);
-extern int nouveau_channel_idle(struct nouveau_channel *chan);
/* nouveau_object.c */
extern int nouveau_gpuobj_early_init(struct drm_device *);
@@ -756,6 +793,8 @@ extern int nouveau_gpuobj_gart_dma_new(struct nouveau_channel *,
uint32_t *o_ret);
extern int nouveau_gpuobj_gr_new(struct nouveau_channel *, int class,
struct nouveau_gpuobj **);
+extern int nouveau_gpuobj_sw_new(struct nouveau_channel *, int class,
+ struct nouveau_gpuobj **);
extern int nouveau_ioctl_grobj_alloc(struct drm_device *, void *data,
struct drm_file *);
extern int nouveau_ioctl_gpuobj_free(struct drm_device *, void *data,
@@ -804,6 +843,7 @@ nouveau_debugfs_channel_fini(struct nouveau_channel *chan)
#endif
/* nouveau_dma.c */
+extern void nouveau_dma_pre_init(struct nouveau_channel *);
extern int nouveau_dma_init(struct nouveau_channel *);
extern int nouveau_dma_wait(struct nouveau_channel *, int size);
@@ -879,16 +919,22 @@ extern void nv04_fb_takedown(struct drm_device *);
/* nv10_fb.c */
extern int nv10_fb_init(struct drm_device *);
extern void nv10_fb_takedown(struct drm_device *);
+extern void nv10_fb_set_region_tiling(struct drm_device *, int, uint32_t,
+ uint32_t, uint32_t);
/* nv40_fb.c */
extern int nv40_fb_init(struct drm_device *);
extern void nv40_fb_takedown(struct drm_device *);
+extern void nv40_fb_set_region_tiling(struct drm_device *, int, uint32_t,
+ uint32_t, uint32_t);
/* nv04_fifo.c */
extern int nv04_fifo_init(struct drm_device *);
extern void nv04_fifo_disable(struct drm_device *);
extern void nv04_fifo_enable(struct drm_device *);
extern bool nv04_fifo_reassign(struct drm_device *, bool);
+extern bool nv04_fifo_cache_flush(struct drm_device *);
+extern bool nv04_fifo_cache_pull(struct drm_device *, bool);
extern int nv04_fifo_channel_id(struct drm_device *);
extern int nv04_fifo_create_context(struct nouveau_channel *);
extern void nv04_fifo_destroy_context(struct nouveau_channel *);
@@ -941,6 +987,8 @@ extern void nv10_graph_destroy_context(struct nouveau_channel *);
extern int nv10_graph_load_context(struct nouveau_channel *);
extern int nv10_graph_unload_context(struct drm_device *);
extern void nv10_graph_context_switch(struct drm_device *);
+extern void nv10_graph_set_region_tiling(struct drm_device *, int, uint32_t,
+ uint32_t, uint32_t);
/* nv20_graph.c */
extern struct nouveau_pgraph_object_class nv20_graph_grclass[];
@@ -952,6 +1000,8 @@ extern int nv20_graph_unload_context(struct drm_device *);
extern int nv20_graph_init(struct drm_device *);
extern void nv20_graph_takedown(struct drm_device *);
extern int nv30_graph_init(struct drm_device *);
+extern void nv20_graph_set_region_tiling(struct drm_device *, int, uint32_t,
+ uint32_t, uint32_t);
/* nv40_graph.c */
extern struct nouveau_pgraph_object_class nv40_graph_grclass[];
@@ -963,6 +1013,8 @@ extern void nv40_graph_destroy_context(struct nouveau_channel *);
extern int nv40_graph_load_context(struct nouveau_channel *);
extern int nv40_graph_unload_context(struct drm_device *);
extern void nv40_grctx_init(struct nouveau_grctx *);
+extern void nv40_graph_set_region_tiling(struct drm_device *, int, uint32_t,
+ uint32_t, uint32_t);
/* nv50_graph.c */
extern struct nouveau_pgraph_object_class nv50_graph_grclass[];
@@ -1030,8 +1082,7 @@ extern long nouveau_compat_ioctl(struct file *file, unsigned int cmd,
/* nv04_dac.c */
extern int nv04_dac_create(struct drm_device *dev, struct dcb_entry *entry);
-extern enum drm_connector_status nv17_dac_detect(struct drm_encoder *encoder,
- struct drm_connector *connector);
+extern uint32_t nv17_dac_sample_load(struct drm_encoder *encoder);
extern int nv04_dac_output_offset(struct drm_encoder *encoder);
extern void nv04_dac_update_dacclk(struct drm_encoder *encoder, bool enable);
@@ -1049,9 +1100,6 @@ extern int nv04_tv_create(struct drm_device *dev, struct dcb_entry *entry);
/* nv17_tv.c */
extern int nv17_tv_create(struct drm_device *dev, struct dcb_entry *entry);
-extern enum drm_connector_status nv17_tv_detect(struct drm_encoder *encoder,
- struct drm_connector *connector,
- uint32_t pin_mask);
/* nv04_display.c */
extern int nv04_display_create(struct drm_device *);
@@ -1290,14 +1338,14 @@ nv_two_reg_pll(struct drm_device *dev)
return false;
}
-#define NV50_NVSW 0x0000506e
-#define NV50_NVSW_DMA_SEMAPHORE 0x00000060
-#define NV50_NVSW_SEMAPHORE_OFFSET 0x00000064
-#define NV50_NVSW_SEMAPHORE_ACQUIRE 0x00000068
-#define NV50_NVSW_SEMAPHORE_RELEASE 0x0000006c
-#define NV50_NVSW_DMA_VBLSEM 0x0000018c
-#define NV50_NVSW_VBLSEM_OFFSET 0x00000400
-#define NV50_NVSW_VBLSEM_RELEASE_VALUE 0x00000404
-#define NV50_NVSW_VBLSEM_RELEASE 0x00000408
+#define NV_SW 0x0000506e
+#define NV_SW_DMA_SEMAPHORE 0x00000060
+#define NV_SW_SEMAPHORE_OFFSET 0x00000064
+#define NV_SW_SEMAPHORE_ACQUIRE 0x00000068
+#define NV_SW_SEMAPHORE_RELEASE 0x0000006c
+#define NV_SW_DMA_VBLSEM 0x0000018c
+#define NV_SW_VBLSEM_OFFSET 0x00000400
+#define NV_SW_VBLSEM_RELEASE_VALUE 0x00000404
+#define NV_SW_VBLSEM_RELEASE 0x00000408
#endif /* __NOUVEAU_DRV_H__ */
diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.c b/drivers/gpu/drm/nouveau/nouveau_fbcon.c
index 84af25c238b..ea879a2efef 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fbcon.c
+++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.c
@@ -64,8 +64,7 @@ nouveau_fbcon_sync(struct fb_info *info)
return 0;
if (RING_SPACE(chan, 4)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
return 0;
}
@@ -86,8 +85,7 @@ nouveau_fbcon_sync(struct fb_info *info)
}
if (ret) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
return 0;
}
@@ -109,6 +107,34 @@ static struct fb_ops nouveau_fbcon_ops = {
.fb_setcmap = drm_fb_helper_setcmap,
};
+static struct fb_ops nv04_fbcon_ops = {
+ .owner = THIS_MODULE,
+ .fb_check_var = drm_fb_helper_check_var,
+ .fb_set_par = drm_fb_helper_set_par,
+ .fb_setcolreg = drm_fb_helper_setcolreg,
+ .fb_fillrect = nv04_fbcon_fillrect,
+ .fb_copyarea = nv04_fbcon_copyarea,
+ .fb_imageblit = nv04_fbcon_imageblit,
+ .fb_sync = nouveau_fbcon_sync,
+ .fb_pan_display = drm_fb_helper_pan_display,
+ .fb_blank = drm_fb_helper_blank,
+ .fb_setcmap = drm_fb_helper_setcmap,
+};
+
+static struct fb_ops nv50_fbcon_ops = {
+ .owner = THIS_MODULE,
+ .fb_check_var = drm_fb_helper_check_var,
+ .fb_set_par = drm_fb_helper_set_par,
+ .fb_setcolreg = drm_fb_helper_setcolreg,
+ .fb_fillrect = nv50_fbcon_fillrect,
+ .fb_copyarea = nv50_fbcon_copyarea,
+ .fb_imageblit = nv50_fbcon_imageblit,
+ .fb_sync = nouveau_fbcon_sync,
+ .fb_pan_display = drm_fb_helper_pan_display,
+ .fb_blank = drm_fb_helper_blank,
+ .fb_setcmap = drm_fb_helper_setcmap,
+};
+
static void nouveau_fbcon_gamma_set(struct drm_crtc *crtc, u16 red, u16 green,
u16 blue, int regno)
{
@@ -212,11 +238,11 @@ nouveau_fbcon_create(struct drm_device *dev, uint32_t fb_width,
mode_cmd.bpp = surface_bpp;
mode_cmd.pitch = mode_cmd.width * (mode_cmd.bpp >> 3);
- mode_cmd.pitch = ALIGN(mode_cmd.pitch, 256);
+ mode_cmd.pitch = roundup(mode_cmd.pitch, 256);
mode_cmd.depth = surface_depth;
size = mode_cmd.pitch * mode_cmd.height;
- size = ALIGN(size, PAGE_SIZE);
+ size = roundup(size, PAGE_SIZE);
ret = nouveau_gem_new(dev, dev_priv->channel, size, 0, TTM_PL_FLAG_VRAM,
0, 0x0000, false, true, &nvbo);
@@ -269,8 +295,12 @@ nouveau_fbcon_create(struct drm_device *dev, uint32_t fb_width,
dev_priv->fbdev_info = info;
strcpy(info->fix.id, "nouveaufb");
- info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_COPYAREA |
- FBINFO_HWACCEL_FILLRECT | FBINFO_HWACCEL_IMAGEBLIT;
+ if (nouveau_nofbaccel)
+ info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_DISABLED;
+ else
+ info->flags = FBINFO_DEFAULT | FBINFO_HWACCEL_COPYAREA |
+ FBINFO_HWACCEL_FILLRECT |
+ FBINFO_HWACCEL_IMAGEBLIT;
info->fbops = &nouveau_fbcon_ops;
info->fix.smem_start = dev->mode_config.fb_base + nvbo->bo.offset -
dev_priv->vm_vram_base;
@@ -318,13 +348,15 @@ nouveau_fbcon_create(struct drm_device *dev, uint32_t fb_width,
par->nouveau_fb = nouveau_fb;
par->dev = dev;
- if (dev_priv->channel) {
+ if (dev_priv->channel && !nouveau_nofbaccel) {
switch (dev_priv->card_type) {
case NV_50:
nv50_fbcon_accel_init(info);
+ info->fbops = &nv50_fbcon_ops;
break;
default:
nv04_fbcon_accel_init(info);
+ info->fbops = &nv04_fbcon_ops;
break;
};
}
@@ -380,3 +412,12 @@ nouveau_fbcon_remove(struct drm_device *dev, struct drm_framebuffer *fb)
return 0;
}
+
+void nouveau_fbcon_gpu_lockup(struct fb_info *info)
+{
+ struct nouveau_fbcon_par *par = info->par;
+ struct drm_device *dev = par->dev;
+
+ NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
+ info->flags |= FBINFO_HWACCEL_DISABLED;
+}
diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.h b/drivers/gpu/drm/nouveau/nouveau_fbcon.h
index 8531140fedb..f9c34e1a8c1 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fbcon.h
+++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.h
@@ -40,8 +40,15 @@ int nouveau_fbcon_remove(struct drm_device *dev, struct drm_framebuffer *fb);
void nouveau_fbcon_restore(void);
void nouveau_fbcon_zfill(struct drm_device *dev);
+void nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region);
+void nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect);
+void nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image);
int nv04_fbcon_accel_init(struct fb_info *info);
+void nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect);
+void nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region);
+void nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image);
int nv50_fbcon_accel_init(struct fb_info *info);
+void nouveau_fbcon_gpu_lockup(struct fb_info *info);
#endif /* __NV50_FBCON_H__ */
diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.c b/drivers/gpu/drm/nouveau/nouveau_fence.c
index dacac9a0842..faddf53ff9e 100644
--- a/drivers/gpu/drm/nouveau/nouveau_fence.c
+++ b/drivers/gpu/drm/nouveau/nouveau_fence.c
@@ -142,7 +142,7 @@ nouveau_fence_emit(struct nouveau_fence *fence)
list_add_tail(&fence->entry, &chan->fence.pending);
spin_unlock_irqrestore(&chan->fence.lock, flags);
- BEGIN_RING(chan, NvSubM2MF, USE_REFCNT ? 0x0050 : 0x0150, 1);
+ BEGIN_RING(chan, NvSubSw, USE_REFCNT ? 0x0050 : 0x0150, 1);
OUT_RING(chan, fence->sequence);
FIRE_RING(chan);
diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c
index 18fd8ac9fca..70cc30803e3 100644
--- a/drivers/gpu/drm/nouveau/nouveau_gem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_gem.c
@@ -220,7 +220,6 @@ nouveau_gem_set_domain(struct drm_gem_object *gem, uint32_t read_domains,
}
struct validate_op {
- struct nouveau_fence *fence;
struct list_head vram_list;
struct list_head gart_list;
struct list_head both_list;
@@ -252,17 +251,11 @@ validate_fini_list(struct list_head *list, struct nouveau_fence *fence)
}
static void
-validate_fini(struct validate_op *op, bool success)
+validate_fini(struct validate_op *op, struct nouveau_fence* fence)
{
- struct nouveau_fence *fence = op->fence;
-
- if (unlikely(!success))
- op->fence = NULL;
-
- validate_fini_list(&op->vram_list, op->fence);
- validate_fini_list(&op->gart_list, op->fence);
- validate_fini_list(&op->both_list, op->fence);
- nouveau_fence_unref((void *)&fence);
+ validate_fini_list(&op->vram_list, fence);
+ validate_fini_list(&op->gart_list, fence);
+ validate_fini_list(&op->both_list, fence);
}
static int
@@ -328,6 +321,7 @@ retry:
else {
NV_ERROR(dev, "invalid valid domains: 0x%08x\n",
b->valid_domains);
+ list_add_tail(&nvbo->entry, &op->both_list);
validate_fini(op, NULL);
return -EINVAL;
}
@@ -420,10 +414,6 @@ nouveau_gem_pushbuf_validate(struct nouveau_channel *chan,
INIT_LIST_HEAD(&op->gart_list);
INIT_LIST_HEAD(&op->both_list);
- ret = nouveau_fence_new(chan, &op->fence, false);
- if (ret)
- return ret;
-
if (nr_buffers == 0)
return 0;
@@ -477,13 +467,14 @@ u_memcpya(uint64_t user, unsigned nmemb, unsigned size)
static int
nouveau_gem_pushbuf_reloc_apply(struct nouveau_channel *chan, int nr_bo,
struct drm_nouveau_gem_pushbuf_bo *bo,
- int nr_relocs, uint64_t ptr_relocs,
- int nr_dwords, int first_dword,
+ unsigned nr_relocs, uint64_t ptr_relocs,
+ unsigned nr_dwords, unsigned first_dword,
uint32_t *pushbuf, bool is_iomem)
{
struct drm_nouveau_gem_pushbuf_reloc *reloc = NULL;
struct drm_device *dev = chan->dev;
- int ret = 0, i;
+ int ret = 0;
+ unsigned i;
reloc = u_memcpya(ptr_relocs, nr_relocs, sizeof(*reloc));
if (IS_ERR(reloc))
@@ -541,6 +532,7 @@ nouveau_gem_ioctl_pushbuf(struct drm_device *dev, void *data,
struct drm_nouveau_gem_pushbuf_bo *bo = NULL;
struct nouveau_channel *chan;
struct validate_op op;
+ struct nouveau_fence* fence = 0;
uint32_t *pushbuf = NULL;
int ret = 0, do_reloc = 0, i;
@@ -597,7 +589,7 @@ nouveau_gem_ioctl_pushbuf(struct drm_device *dev, void *data,
OUT_RINGp(chan, pushbuf, req->nr_dwords);
- ret = nouveau_fence_emit(op.fence);
+ ret = nouveau_fence_new(chan, &fence, true);
if (ret) {
NV_ERROR(dev, "error fencing pushbuf: %d\n", ret);
WIND_RING(chan);
@@ -605,7 +597,7 @@ nouveau_gem_ioctl_pushbuf(struct drm_device *dev, void *data,
}
if (nouveau_gem_pushbuf_sync(chan)) {
- ret = nouveau_fence_wait(op.fence, NULL, false, false);
+ ret = nouveau_fence_wait(fence, NULL, false, false);
if (ret) {
for (i = 0; i < req->nr_dwords; i++)
NV_ERROR(dev, "0x%08x\n", pushbuf[i]);
@@ -614,7 +606,8 @@ nouveau_gem_ioctl_pushbuf(struct drm_device *dev, void *data,
}
out:
- validate_fini(&op, ret == 0);
+ validate_fini(&op, fence);
+ nouveau_fence_unref((void**)&fence);
mutex_unlock(&dev->struct_mutex);
kfree(pushbuf);
kfree(bo);
@@ -634,6 +627,7 @@ nouveau_gem_ioctl_pushbuf_call(struct drm_device *dev, void *data,
struct drm_gem_object *gem;
struct nouveau_bo *pbbo;
struct validate_op op;
+ struct nouveau_fence* fence = 0;
int i, ret = 0, do_reloc = 0;
NOUVEAU_CHECK_INITIALISED_WITH_RETURN;
@@ -675,6 +669,18 @@ nouveau_gem_ioctl_pushbuf_call(struct drm_device *dev, void *data,
}
pbbo = nouveau_gem_object(gem);
+ if ((req->offset & 3) || req->nr_dwords < 2 ||
+ (unsigned long)req->offset > (unsigned long)pbbo->bo.mem.size ||
+ (unsigned long)req->nr_dwords >
+ ((unsigned long)(pbbo->bo.mem.size - req->offset ) >> 2)) {
+ NV_ERROR(dev, "pb call misaligned or out of bounds: "
+ "%d + %d * 4 > %ld\n",
+ req->offset, req->nr_dwords, pbbo->bo.mem.size);
+ ret = -EINVAL;
+ drm_gem_object_unreference(gem);
+ goto out;
+ }
+
ret = ttm_bo_reserve(&pbbo->bo, false, false, true,
chan->fence.sequence);
if (ret) {
@@ -772,7 +778,7 @@ nouveau_gem_ioctl_pushbuf_call(struct drm_device *dev, void *data,
OUT_RING(chan, 0);
}
- ret = nouveau_fence_emit(op.fence);
+ ret = nouveau_fence_new(chan, &fence, true);
if (ret) {
NV_ERROR(dev, "error fencing pushbuf: %d\n", ret);
WIND_RING(chan);
@@ -780,7 +786,8 @@ nouveau_gem_ioctl_pushbuf_call(struct drm_device *dev, void *data,
}
out:
- validate_fini(&op, ret == 0);
+ validate_fini(&op, fence);
+ nouveau_fence_unref((void**)&fence);
mutex_unlock(&dev->struct_mutex);
kfree(bo);
@@ -918,7 +925,9 @@ nouveau_gem_ioctl_cpu_prep(struct drm_device *dev, void *data,
}
if (req->flags & NOUVEAU_GEM_CPU_PREP_NOBLOCK) {
+ spin_lock(&nvbo->bo.lock);
ret = ttm_bo_wait(&nvbo->bo, false, false, no_wait);
+ spin_unlock(&nvbo->bo.lock);
} else {
ret = ttm_bo_synccpu_write_grab(&nvbo->bo, no_wait);
if (ret == 0)
diff --git a/drivers/gpu/drm/nouveau/nouveau_grctx.c b/drivers/gpu/drm/nouveau/nouveau_grctx.c
index 419f4c2b3b8..c7ebec69674 100644
--- a/drivers/gpu/drm/nouveau/nouveau_grctx.c
+++ b/drivers/gpu/drm/nouveau/nouveau_grctx.c
@@ -97,8 +97,8 @@ nouveau_grctx_prog_load(struct drm_device *dev)
}
pgraph->ctxvals = kmalloc(fw->size, GFP_KERNEL);
- if (!pgraph->ctxprog) {
- NV_ERROR(dev, "OOM copying ctxprog\n");
+ if (!pgraph->ctxvals) {
+ NV_ERROR(dev, "OOM copying ctxvals\n");
release_firmware(fw);
nouveau_grctx_fini(dev);
return -ENOMEM;
diff --git a/drivers/gpu/drm/nouveau/nouveau_irq.c b/drivers/gpu/drm/nouveau/nouveau_irq.c
index 370c72c968d..447f9f69d6b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_irq.c
+++ b/drivers/gpu/drm/nouveau/nouveau_irq.c
@@ -211,6 +211,20 @@ nouveau_fifo_irq_handler(struct drm_device *dev)
get + 4);
}
+ if (status & NV_PFIFO_INTR_SEMAPHORE) {
+ uint32_t sem;
+
+ status &= ~NV_PFIFO_INTR_SEMAPHORE;
+ nv_wr32(dev, NV03_PFIFO_INTR_0,
+ NV_PFIFO_INTR_SEMAPHORE);
+
+ sem = nv_rd32(dev, NV10_PFIFO_CACHE1_SEMAPHORE);
+ nv_wr32(dev, NV10_PFIFO_CACHE1_SEMAPHORE, sem | 0x1);
+
+ nv_wr32(dev, NV03_PFIFO_CACHE1_GET, get + 4);
+ nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, 1);
+ }
+
if (status) {
NV_INFO(dev, "PFIFO_INTR 0x%08x - Ch %d\n",
status, chid);
@@ -483,6 +497,13 @@ nouveau_pgraph_intr_error(struct drm_device *dev, uint32_t nsource)
if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) {
if (nouveau_pgraph_intr_swmthd(dev, &trap))
unhandled = 1;
+ } else if (nsource & NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION) {
+ uint32_t v = nv_rd32(dev, 0x402000);
+ nv_wr32(dev, 0x402000, v);
+
+ /* dump the error anyway for now: it's useful for
+ Gallium development */
+ unhandled = 1;
} else {
unhandled = 1;
}
@@ -559,85 +580,99 @@ nouveau_pgraph_irq_handler(struct drm_device *dev)
static void
nv50_pgraph_irq_handler(struct drm_device *dev)
{
- uint32_t status, nsource;
+ uint32_t status;
- status = nv_rd32(dev, NV03_PGRAPH_INTR);
- nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE);
+ while ((status = nv_rd32(dev, NV03_PGRAPH_INTR))) {
+ uint32_t nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE);
- if (status & 0x00000001) {
- nouveau_pgraph_intr_notify(dev, nsource);
- status &= ~0x00000001;
- nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000001);
- }
+ if (status & 0x00000001) {
+ nouveau_pgraph_intr_notify(dev, nsource);
+ status &= ~0x00000001;
+ nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000001);
+ }
- if (status & 0x00000010) {
- nouveau_pgraph_intr_error(dev, nsource |
- NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD);
+ if (status & 0x00000010) {
+ nouveau_pgraph_intr_error(dev, nsource |
+ NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD);
- status &= ~0x00000010;
- nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000010);
- }
+ status &= ~0x00000010;
+ nv_wr32(dev, NV03_PGRAPH_INTR, 0x00000010);
+ }
- if (status & 0x00001000) {
- nv_wr32(dev, 0x400500, 0x00000000);
- nv_wr32(dev, NV03_PGRAPH_INTR, NV_PGRAPH_INTR_CONTEXT_SWITCH);
- nv_wr32(dev, NV40_PGRAPH_INTR_EN, nv_rd32(dev,
- NV40_PGRAPH_INTR_EN) & ~NV_PGRAPH_INTR_CONTEXT_SWITCH);
- nv_wr32(dev, 0x400500, 0x00010001);
+ if (status & 0x00001000) {
+ nv_wr32(dev, 0x400500, 0x00000000);
+ nv_wr32(dev, NV03_PGRAPH_INTR,
+ NV_PGRAPH_INTR_CONTEXT_SWITCH);
+ nv_wr32(dev, NV40_PGRAPH_INTR_EN, nv_rd32(dev,
+ NV40_PGRAPH_INTR_EN) &
+ ~NV_PGRAPH_INTR_CONTEXT_SWITCH);
+ nv_wr32(dev, 0x400500, 0x00010001);
- nv50_graph_context_switch(dev);
+ nv50_graph_context_switch(dev);
- status &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH;
- }
+ status &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH;
+ }
- if (status & 0x00100000) {
- nouveau_pgraph_intr_error(dev, nsource |
- NV03_PGRAPH_NSOURCE_DATA_ERROR);
+ if (status & 0x00100000) {
+ nouveau_pgraph_intr_error(dev, nsource |
+ NV03_PGRAPH_NSOURCE_DATA_ERROR);
- status &= ~0x00100000;
- nv_wr32(dev, NV03_PGRAPH_INTR, 0x00100000);
- }
+ status &= ~0x00100000;
+ nv_wr32(dev, NV03_PGRAPH_INTR, 0x00100000);
+ }
- if (status & 0x00200000) {
- int r;
-
- nouveau_pgraph_intr_error(dev, nsource |
- NV03_PGRAPH_NSOURCE_PROTECTION_ERROR);
-
- NV_ERROR(dev, "magic set 1:\n");
- for (r = 0x408900; r <= 0x408910; r += 4)
- NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r));
- nv_wr32(dev, 0x408900, nv_rd32(dev, 0x408904) | 0xc0000000);
- for (r = 0x408e08; r <= 0x408e24; r += 4)
- NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r));
- nv_wr32(dev, 0x408e08, nv_rd32(dev, 0x408e08) | 0xc0000000);
-
- NV_ERROR(dev, "magic set 2:\n");
- for (r = 0x409900; r <= 0x409910; r += 4)
- NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r));
- nv_wr32(dev, 0x409900, nv_rd32(dev, 0x409904) | 0xc0000000);
- for (r = 0x409e08; r <= 0x409e24; r += 4)
- NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r, nv_rd32(dev, r));
- nv_wr32(dev, 0x409e08, nv_rd32(dev, 0x409e08) | 0xc0000000);
-
- status &= ~0x00200000;
- nv_wr32(dev, NV03_PGRAPH_NSOURCE, nsource);
- nv_wr32(dev, NV03_PGRAPH_INTR, 0x00200000);
- }
+ if (status & 0x00200000) {
+ int r;
+
+ nouveau_pgraph_intr_error(dev, nsource |
+ NV03_PGRAPH_NSOURCE_PROTECTION_ERROR);
+
+ NV_ERROR(dev, "magic set 1:\n");
+ for (r = 0x408900; r <= 0x408910; r += 4)
+ NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r,
+ nv_rd32(dev, r));
+ nv_wr32(dev, 0x408900,
+ nv_rd32(dev, 0x408904) | 0xc0000000);
+ for (r = 0x408e08; r <= 0x408e24; r += 4)
+ NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r,
+ nv_rd32(dev, r));
+ nv_wr32(dev, 0x408e08,
+ nv_rd32(dev, 0x408e08) | 0xc0000000);
+
+ NV_ERROR(dev, "magic set 2:\n");
+ for (r = 0x409900; r <= 0x409910; r += 4)
+ NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r,
+ nv_rd32(dev, r));
+ nv_wr32(dev, 0x409900,
+ nv_rd32(dev, 0x409904) | 0xc0000000);
+ for (r = 0x409e08; r <= 0x409e24; r += 4)
+ NV_ERROR(dev, "\t0x%08x: 0x%08x\n", r,
+ nv_rd32(dev, r));
+ nv_wr32(dev, 0x409e08,
+ nv_rd32(dev, 0x409e08) | 0xc0000000);
+
+ status &= ~0x00200000;
+ nv_wr32(dev, NV03_PGRAPH_NSOURCE, nsource);
+ nv_wr32(dev, NV03_PGRAPH_INTR, 0x00200000);
+ }
- if (status) {
- NV_INFO(dev, "Unhandled PGRAPH_INTR - 0x%08x\n", status);
- nv_wr32(dev, NV03_PGRAPH_INTR, status);
- }
+ if (status) {
+ NV_INFO(dev, "Unhandled PGRAPH_INTR - 0x%08x\n",
+ status);
+ nv_wr32(dev, NV03_PGRAPH_INTR, status);
+ }
- {
- const int isb = (1 << 16) | (1 << 0);
+ {
+ const int isb = (1 << 16) | (1 << 0);
- if ((nv_rd32(dev, 0x400500) & isb) != isb)
- nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) | isb);
+ if ((nv_rd32(dev, 0x400500) & isb) != isb)
+ nv_wr32(dev, 0x400500,
+ nv_rd32(dev, 0x400500) | isb);
+ }
}
nv_wr32(dev, NV03_PMC_INTR_0, NV_PMC_INTR_0_PGRAPH_PENDING);
+ nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) & ~(1 << 31));
}
static void
diff --git a/drivers/gpu/drm/nouveau/nouveau_mem.c b/drivers/gpu/drm/nouveau/nouveau_mem.c
index 5158a12f784..2dc09dbd817 100644
--- a/drivers/gpu/drm/nouveau/nouveau_mem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_mem.c
@@ -192,6 +192,92 @@ void nouveau_mem_release(struct drm_file *file_priv, struct mem_block *heap)
}
/*
+ * NV10-NV40 tiling helpers
+ */
+
+static void
+nv10_mem_set_region_tiling(struct drm_device *dev, int i, uint32_t addr,
+ uint32_t size, uint32_t pitch)
+{
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo;
+ struct nouveau_fb_engine *pfb = &dev_priv->engine.fb;
+ struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph;
+ struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i];
+
+ tile->addr = addr;
+ tile->size = size;
+ tile->used = !!pitch;
+ nouveau_fence_unref((void **)&tile->fence);
+
+ if (!pfifo->cache_flush(dev))
+ return;
+
+ pfifo->reassign(dev, false);
+ pfifo->cache_flush(dev);
+ pfifo->cache_pull(dev, false);
+
+ nouveau_wait_for_idle(dev);
+
+ pgraph->set_region_tiling(dev, i, addr, size, pitch);
+ pfb->set_region_tiling(dev, i, addr, size, pitch);
+
+ pfifo->cache_pull(dev, true);
+ pfifo->reassign(dev, true);
+}
+
+struct nouveau_tile_reg *
+nv10_mem_set_tiling(struct drm_device *dev, uint32_t addr, uint32_t size,
+ uint32_t pitch)
+{
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ struct nouveau_fb_engine *pfb = &dev_priv->engine.fb;
+ struct nouveau_tile_reg *tile = dev_priv->tile.reg, *found = NULL;
+ int i;
+
+ spin_lock(&dev_priv->tile.lock);
+
+ for (i = 0; i < pfb->num_tiles; i++) {
+ if (tile[i].used)
+ /* Tile region in use. */
+ continue;
+
+ if (tile[i].fence &&
+ !nouveau_fence_signalled(tile[i].fence, NULL))
+ /* Pending tile region. */
+ continue;
+
+ if (max(tile[i].addr, addr) <
+ min(tile[i].addr + tile[i].size, addr + size))
+ /* Kill an intersecting tile region. */
+ nv10_mem_set_region_tiling(dev, i, 0, 0, 0);
+
+ if (pitch && !found) {
+ /* Free tile region. */
+ nv10_mem_set_region_tiling(dev, i, addr, size, pitch);
+ found = &tile[i];
+ }
+ }
+
+ spin_unlock(&dev_priv->tile.lock);
+
+ return found;
+}
+
+void
+nv10_mem_expire_tiling(struct drm_device *dev, struct nouveau_tile_reg *tile,
+ struct nouveau_fence *fence)
+{
+ if (fence) {
+ /* Mark it as pending. */
+ tile->fence = fence;
+ nouveau_fence_ref(fence);
+ }
+
+ tile->used = false;
+}
+
+/*
* NV50 VM helpers
*/
int
@@ -199,53 +285,50 @@ nv50_mem_vm_bind_linear(struct drm_device *dev, uint64_t virt, uint32_t size,
uint32_t flags, uint64_t phys)
{
struct drm_nouveau_private *dev_priv = dev->dev_private;
- struct nouveau_gpuobj **pgt;
- unsigned psz, pfl, pages;
-
- if (virt >= dev_priv->vm_gart_base &&
- (virt + size) < (dev_priv->vm_gart_base + dev_priv->vm_gart_size)) {
- psz = 12;
- pgt = &dev_priv->gart_info.sg_ctxdma;
- pfl = 0x21;
- virt -= dev_priv->vm_gart_base;
- } else
- if (virt >= dev_priv->vm_vram_base &&
- (virt + size) < (dev_priv->vm_vram_base + dev_priv->vm_vram_size)) {
- psz = 16;
- pgt = dev_priv->vm_vram_pt;
- pfl = 0x01;
- virt -= dev_priv->vm_vram_base;
- } else {
- NV_ERROR(dev, "Invalid address: 0x%16llx-0x%16llx\n",
- virt, virt + size - 1);
- return -EINVAL;
+ struct nouveau_gpuobj *pgt;
+ unsigned block;
+ int i;
+
+ virt = ((virt - dev_priv->vm_vram_base) >> 16) << 1;
+ size = (size >> 16) << 1;
+
+ phys |= ((uint64_t)flags << 32);
+ phys |= 1;
+ if (dev_priv->vram_sys_base) {
+ phys += dev_priv->vram_sys_base;
+ phys |= 0x30;
}
- pages = size >> psz;
-
dev_priv->engine.instmem.prepare_access(dev, true);
- if (flags & 0x80000000) {
- while (pages--) {
- struct nouveau_gpuobj *pt = pgt[virt >> 29];
- unsigned pte = ((virt & 0x1fffffffULL) >> psz) << 1;
+ while (size) {
+ unsigned offset_h = upper_32_bits(phys);
+ unsigned offset_l = lower_32_bits(phys);
+ unsigned pte, end;
+
+ for (i = 7; i >= 0; i--) {
+ block = 1 << (i + 1);
+ if (size >= block && !(virt & (block - 1)))
+ break;
+ }
+ offset_l |= (i << 7);
- nv_wo32(dev, pt, pte++, 0x00000000);
- nv_wo32(dev, pt, pte++, 0x00000000);
+ phys += block << 15;
+ size -= block;
- virt += (1 << psz);
- }
- } else {
- while (pages--) {
- struct nouveau_gpuobj *pt = pgt[virt >> 29];
- unsigned pte = ((virt & 0x1fffffffULL) >> psz) << 1;
- unsigned offset_h = upper_32_bits(phys) & 0xff;
- unsigned offset_l = lower_32_bits(phys);
+ while (block) {
+ pgt = dev_priv->vm_vram_pt[virt >> 14];
+ pte = virt & 0x3ffe;
- nv_wo32(dev, pt, pte++, offset_l | pfl);
- nv_wo32(dev, pt, pte++, offset_h | flags);
+ end = pte + block;
+ if (end > 16384)
+ end = 16384;
+ block -= (end - pte);
+ virt += (end - pte);
- phys += (1 << psz);
- virt += (1 << psz);
+ while (pte < end) {
+ nv_wo32(dev, pgt, pte++, offset_l);
+ nv_wo32(dev, pgt, pte++, offset_h);
+ }
}
}
dev_priv->engine.instmem.finish_access(dev);
@@ -270,7 +353,41 @@ nv50_mem_vm_bind_linear(struct drm_device *dev, uint64_t virt, uint32_t size,
void
nv50_mem_vm_unbind(struct drm_device *dev, uint64_t virt, uint32_t size)
{
- nv50_mem_vm_bind_linear(dev, virt, size, 0x80000000, 0);
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ struct nouveau_gpuobj *pgt;
+ unsigned pages, pte, end;
+
+ virt -= dev_priv->vm_vram_base;
+ pages = (size >> 16) << 1;
+
+ dev_priv->engine.instmem.prepare_access(dev, true);
+ while (pages) {
+ pgt = dev_priv->vm_vram_pt[virt >> 29];
+ pte = (virt & 0x1ffe0000ULL) >> 15;
+
+ end = pte + pages;
+ if (end > 16384)
+ end = 16384;
+ pages -= (end - pte);
+ virt += (end - pte) << 15;
+
+ while (pte < end)
+ nv_wo32(dev, pgt, pte++, 0);
+ }
+ dev_priv->engine.instmem.finish_access(dev);
+
+ nv_wr32(dev, 0x100c80, 0x00050001);
+ if (!nv_wait(0x100c80, 0x00000001, 0x00000000)) {
+ NV_ERROR(dev, "timeout: (0x100c80 & 1) == 0 (2)\n");
+ NV_ERROR(dev, "0x100c80 = 0x%08x\n", nv_rd32(dev, 0x100c80));
+ return;
+ }
+
+ nv_wr32(dev, 0x100c80, 0x00000001);
+ if (!nv_wait(0x100c80, 0x00000001, 0x00000000)) {
+ NV_ERROR(dev, "timeout: (0x100c80 & 1) == 0 (2)\n");
+ NV_ERROR(dev, "0x100c80 = 0x%08x\n", nv_rd32(dev, 0x100c80));
+ }
}
/*
@@ -297,9 +414,8 @@ void nouveau_mem_close(struct drm_device *dev)
{
struct drm_nouveau_private *dev_priv = dev->dev_private;
- if (dev_priv->ttm.bdev.man[TTM_PL_PRIV0].has_type)
- ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_PRIV0);
- ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM);
+ nouveau_bo_unpin(dev_priv->vga_ram);
+ nouveau_bo_ref(NULL, &dev_priv->vga_ram);
ttm_bo_device_release(&dev_priv->ttm.bdev);
@@ -513,6 +629,7 @@ nouveau_mem_init(struct drm_device *dev)
INIT_LIST_HEAD(&dev_priv->ttm.bo_list);
spin_lock_init(&dev_priv->ttm.bo_list_lock);
+ spin_lock_init(&dev_priv->tile.lock);
dev_priv->fb_available_size = nouveau_mem_fb_amount(dev);
@@ -535,6 +652,15 @@ nouveau_mem_init(struct drm_device *dev)
return ret;
}
+ ret = nouveau_bo_new(dev, NULL, 256*1024, 0, TTM_PL_FLAG_VRAM,
+ 0, 0, true, true, &dev_priv->vga_ram);
+ if (ret == 0)
+ ret = nouveau_bo_pin(dev_priv->vga_ram, TTM_PL_FLAG_VRAM);
+ if (ret) {
+ NV_WARN(dev, "failed to reserve VGA memory\n");
+ nouveau_bo_ref(NULL, &dev_priv->vga_ram);
+ }
+
/* GART */
#if !defined(__powerpc__) && !defined(__ia64__)
if (drm_device_is_agp(dev) && dev->agp) {
@@ -566,6 +692,7 @@ nouveau_mem_init(struct drm_device *dev)
dev_priv->fb_mtrr = drm_mtrr_add(drm_get_resource_start(dev, 1),
drm_get_resource_len(dev, 1),
DRM_MTRR_WC);
+
return 0;
}
diff --git a/drivers/gpu/drm/nouveau/nouveau_notifier.c b/drivers/gpu/drm/nouveau/nouveau_notifier.c
index 6c66a34b634..d99dc087f9b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_notifier.c
+++ b/drivers/gpu/drm/nouveau/nouveau_notifier.c
@@ -34,15 +34,20 @@ nouveau_notifier_init_channel(struct nouveau_channel *chan)
{
struct drm_device *dev = chan->dev;
struct nouveau_bo *ntfy = NULL;
+ uint32_t flags;
int ret;
- ret = nouveau_gem_new(dev, NULL, PAGE_SIZE, 0, nouveau_vram_notify ?
- TTM_PL_FLAG_VRAM : TTM_PL_FLAG_TT,
+ if (nouveau_vram_notify)
+ flags = TTM_PL_FLAG_VRAM;
+ else
+ flags = TTM_PL_FLAG_TT;
+
+ ret = nouveau_gem_new(dev, NULL, PAGE_SIZE, 0, flags,
0, 0x0000, false, true, &ntfy);
if (ret)
return ret;
- ret = nouveau_bo_pin(ntfy, TTM_PL_FLAG_VRAM);
+ ret = nouveau_bo_pin(ntfy, flags);
if (ret)
goto out_err;
@@ -128,6 +133,8 @@ nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle,
target = NV_DMA_TARGET_PCI;
} else {
target = NV_DMA_TARGET_AGP;
+ if (dev_priv->card_type >= NV_50)
+ offset += dev_priv->vm_gart_base;
}
} else {
NV_ERROR(dev, "Bad DMA target, mem_type %d!\n",
diff --git a/drivers/gpu/drm/nouveau/nouveau_object.c b/drivers/gpu/drm/nouveau/nouveau_object.c
index 93379bb81be..e7c100ba63a 100644
--- a/drivers/gpu/drm/nouveau/nouveau_object.c
+++ b/drivers/gpu/drm/nouveau/nouveau_object.c
@@ -881,15 +881,16 @@ nouveau_gpuobj_gr_new(struct nouveau_channel *chan, int class,
return 0;
}
-static int
+int
nouveau_gpuobj_sw_new(struct nouveau_channel *chan, int class,
struct nouveau_gpuobj **gpuobj_ret)
{
- struct drm_nouveau_private *dev_priv = chan->dev->dev_private;
+ struct drm_nouveau_private *dev_priv;
struct nouveau_gpuobj *gpuobj;
if (!chan || !gpuobj_ret || *gpuobj_ret != NULL)
return -EINVAL;
+ dev_priv = chan->dev->dev_private;
gpuobj = kzalloc(sizeof(*gpuobj), GFP_KERNEL);
if (!gpuobj)
diff --git a/drivers/gpu/drm/nouveau/nouveau_reg.h b/drivers/gpu/drm/nouveau/nouveau_reg.h
index fa1b0e7165b..aa9b310e41b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_reg.h
+++ b/drivers/gpu/drm/nouveau/nouveau_reg.h
@@ -99,6 +99,7 @@
* the card will hang early on in the X init process.
*/
# define NV_PMC_ENABLE_UNK13 (1<<13)
+#define NV40_PMC_GRAPH_UNITS 0x00001540
#define NV40_PMC_BACKLIGHT 0x000015f0
# define NV40_PMC_BACKLIGHT_MASK 0x001f0000
#define NV40_PMC_1700 0x00001700
@@ -349,19 +350,19 @@
#define NV04_PGRAPH_BLEND 0x00400824
#define NV04_PGRAPH_STORED_FMT 0x00400830
#define NV04_PGRAPH_PATT_COLORRAM 0x00400900
-#define NV40_PGRAPH_TILE0(i) (0x00400900 + (i*16))
-#define NV40_PGRAPH_TLIMIT0(i) (0x00400904 + (i*16))
-#define NV40_PGRAPH_TSIZE0(i) (0x00400908 + (i*16))
-#define NV40_PGRAPH_TSTATUS0(i) (0x0040090C + (i*16))
+#define NV20_PGRAPH_TILE(i) (0x00400900 + (i*16))
+#define NV20_PGRAPH_TLIMIT(i) (0x00400904 + (i*16))
+#define NV20_PGRAPH_TSIZE(i) (0x00400908 + (i*16))
+#define NV20_PGRAPH_TSTATUS(i) (0x0040090C + (i*16))
#define NV10_PGRAPH_TILE(i) (0x00400B00 + (i*16))
#define NV10_PGRAPH_TLIMIT(i) (0x00400B04 + (i*16))
#define NV10_PGRAPH_TSIZE(i) (0x00400B08 + (i*16))
#define NV10_PGRAPH_TSTATUS(i) (0x00400B0C + (i*16))
#define NV04_PGRAPH_U_RAM 0x00400D00
-#define NV47_PGRAPH_TILE0(i) (0x00400D00 + (i*16))
-#define NV47_PGRAPH_TLIMIT0(i) (0x00400D04 + (i*16))
-#define NV47_PGRAPH_TSIZE0(i) (0x00400D08 + (i*16))
-#define NV47_PGRAPH_TSTATUS0(i) (0x00400D0C + (i*16))
+#define NV47_PGRAPH_TILE(i) (0x00400D00 + (i*16))
+#define NV47_PGRAPH_TLIMIT(i) (0x00400D04 + (i*16))
+#define NV47_PGRAPH_TSIZE(i) (0x00400D08 + (i*16))
+#define NV47_PGRAPH_TSTATUS(i) (0x00400D0C + (i*16))
#define NV04_PGRAPH_V_RAM 0x00400D40
#define NV04_PGRAPH_W_RAM 0x00400D80
#define NV10_PGRAPH_COMBINER0_IN_ALPHA 0x00400E40
diff --git a/drivers/gpu/drm/nouveau/nouveau_sgdma.c b/drivers/gpu/drm/nouveau/nouveau_sgdma.c
index 4c7f1e403e8..ed1590577b6 100644
--- a/drivers/gpu/drm/nouveau/nouveau_sgdma.c
+++ b/drivers/gpu/drm/nouveau/nouveau_sgdma.c
@@ -54,11 +54,12 @@ static void
nouveau_sgdma_clear(struct ttm_backend *be)
{
struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)be;
- struct drm_device *dev = nvbe->dev;
-
- NV_DEBUG(nvbe->dev, "\n");
+ struct drm_device *dev;
if (nvbe && nvbe->pages) {
+ dev = nvbe->dev;
+ NV_DEBUG(dev, "\n");
+
if (nvbe->bound)
be->func->unbind(be);
diff --git a/drivers/gpu/drm/nouveau/nouveau_state.c b/drivers/gpu/drm/nouveau/nouveau_state.c
index e76ec2d207a..a4851af5b05 100644
--- a/drivers/gpu/drm/nouveau/nouveau_state.c
+++ b/drivers/gpu/drm/nouveau/nouveau_state.c
@@ -76,6 +76,8 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->fifo.disable = nv04_fifo_disable;
engine->fifo.enable = nv04_fifo_enable;
engine->fifo.reassign = nv04_fifo_reassign;
+ engine->fifo.cache_flush = nv04_fifo_cache_flush;
+ engine->fifo.cache_pull = nv04_fifo_cache_pull;
engine->fifo.channel_id = nv04_fifo_channel_id;
engine->fifo.create_context = nv04_fifo_create_context;
engine->fifo.destroy_context = nv04_fifo_destroy_context;
@@ -100,6 +102,7 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->timer.takedown = nv04_timer_takedown;
engine->fb.init = nv10_fb_init;
engine->fb.takedown = nv10_fb_takedown;
+ engine->fb.set_region_tiling = nv10_fb_set_region_tiling;
engine->graph.grclass = nv10_graph_grclass;
engine->graph.init = nv10_graph_init;
engine->graph.takedown = nv10_graph_takedown;
@@ -109,12 +112,15 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->graph.fifo_access = nv04_graph_fifo_access;
engine->graph.load_context = nv10_graph_load_context;
engine->graph.unload_context = nv10_graph_unload_context;
+ engine->graph.set_region_tiling = nv10_graph_set_region_tiling;
engine->fifo.channels = 32;
engine->fifo.init = nv10_fifo_init;
engine->fifo.takedown = nouveau_stub_takedown;
engine->fifo.disable = nv04_fifo_disable;
engine->fifo.enable = nv04_fifo_enable;
engine->fifo.reassign = nv04_fifo_reassign;
+ engine->fifo.cache_flush = nv04_fifo_cache_flush;
+ engine->fifo.cache_pull = nv04_fifo_cache_pull;
engine->fifo.channel_id = nv10_fifo_channel_id;
engine->fifo.create_context = nv10_fifo_create_context;
engine->fifo.destroy_context = nv10_fifo_destroy_context;
@@ -139,6 +145,7 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->timer.takedown = nv04_timer_takedown;
engine->fb.init = nv10_fb_init;
engine->fb.takedown = nv10_fb_takedown;
+ engine->fb.set_region_tiling = nv10_fb_set_region_tiling;
engine->graph.grclass = nv20_graph_grclass;
engine->graph.init = nv20_graph_init;
engine->graph.takedown = nv20_graph_takedown;
@@ -148,12 +155,15 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->graph.fifo_access = nv04_graph_fifo_access;
engine->graph.load_context = nv20_graph_load_context;
engine->graph.unload_context = nv20_graph_unload_context;
+ engine->graph.set_region_tiling = nv20_graph_set_region_tiling;
engine->fifo.channels = 32;
engine->fifo.init = nv10_fifo_init;
engine->fifo.takedown = nouveau_stub_takedown;
engine->fifo.disable = nv04_fifo_disable;
engine->fifo.enable = nv04_fifo_enable;
engine->fifo.reassign = nv04_fifo_reassign;
+ engine->fifo.cache_flush = nv04_fifo_cache_flush;
+ engine->fifo.cache_pull = nv04_fifo_cache_pull;
engine->fifo.channel_id = nv10_fifo_channel_id;
engine->fifo.create_context = nv10_fifo_create_context;
engine->fifo.destroy_context = nv10_fifo_destroy_context;
@@ -178,6 +188,7 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->timer.takedown = nv04_timer_takedown;
engine->fb.init = nv10_fb_init;
engine->fb.takedown = nv10_fb_takedown;
+ engine->fb.set_region_tiling = nv10_fb_set_region_tiling;
engine->graph.grclass = nv30_graph_grclass;
engine->graph.init = nv30_graph_init;
engine->graph.takedown = nv20_graph_takedown;
@@ -187,12 +198,15 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->graph.destroy_context = nv20_graph_destroy_context;
engine->graph.load_context = nv20_graph_load_context;
engine->graph.unload_context = nv20_graph_unload_context;
+ engine->graph.set_region_tiling = nv20_graph_set_region_tiling;
engine->fifo.channels = 32;
engine->fifo.init = nv10_fifo_init;
engine->fifo.takedown = nouveau_stub_takedown;
engine->fifo.disable = nv04_fifo_disable;
engine->fifo.enable = nv04_fifo_enable;
engine->fifo.reassign = nv04_fifo_reassign;
+ engine->fifo.cache_flush = nv04_fifo_cache_flush;
+ engine->fifo.cache_pull = nv04_fifo_cache_pull;
engine->fifo.channel_id = nv10_fifo_channel_id;
engine->fifo.create_context = nv10_fifo_create_context;
engine->fifo.destroy_context = nv10_fifo_destroy_context;
@@ -218,6 +232,7 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->timer.takedown = nv04_timer_takedown;
engine->fb.init = nv40_fb_init;
engine->fb.takedown = nv40_fb_takedown;
+ engine->fb.set_region_tiling = nv40_fb_set_region_tiling;
engine->graph.grclass = nv40_graph_grclass;
engine->graph.init = nv40_graph_init;
engine->graph.takedown = nv40_graph_takedown;
@@ -227,12 +242,15 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
engine->graph.destroy_context = nv40_graph_destroy_context;
engine->graph.load_context = nv40_graph_load_context;
engine->graph.unload_context = nv40_graph_unload_context;
+ engine->graph.set_region_tiling = nv40_graph_set_region_tiling;
engine->fifo.channels = 32;
engine->fifo.init = nv40_fifo_init;
engine->fifo.takedown = nouveau_stub_takedown;
engine->fifo.disable = nv04_fifo_disable;
engine->fifo.enable = nv04_fifo_enable;
engine->fifo.reassign = nv04_fifo_reassign;
+ engine->fifo.cache_flush = nv04_fifo_cache_flush;
+ engine->fifo.cache_pull = nv04_fifo_cache_pull;
engine->fifo.channel_id = nv10_fifo_channel_id;
engine->fifo.create_context = nv40_fifo_create_context;
engine->fifo.destroy_context = nv40_fifo_destroy_context;
@@ -292,6 +310,14 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev)
static unsigned int
nouveau_vga_set_decode(void *priv, bool state)
{
+ struct drm_device *dev = priv;
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+
+ if (dev_priv->chipset >= 0x40)
+ nv_wr32(dev, 0x88054, state);
+ else
+ nv_wr32(dev, 0x1854, state);
+
if (state)
return VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM |
VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM;
@@ -409,15 +435,19 @@ nouveau_card_init(struct drm_device *dev)
if (ret)
goto out_timer;
- /* PGRAPH */
- ret = engine->graph.init(dev);
- if (ret)
- goto out_fb;
+ if (nouveau_noaccel)
+ engine->graph.accel_blocked = true;
+ else {
+ /* PGRAPH */
+ ret = engine->graph.init(dev);
+ if (ret)
+ goto out_fb;
- /* PFIFO */
- ret = engine->fifo.init(dev);
- if (ret)
- goto out_graph;
+ /* PFIFO */
+ ret = engine->fifo.init(dev);
+ if (ret)
+ goto out_graph;
+ }
/* this call irq_preinstall, register irq handler and
* call irq_postinstall
@@ -461,9 +491,11 @@ nouveau_card_init(struct drm_device *dev)
out_irq:
drm_irq_uninstall(dev);
out_fifo:
- engine->fifo.takedown(dev);
+ if (!nouveau_noaccel)
+ engine->fifo.takedown(dev);
out_graph:
- engine->graph.takedown(dev);
+ if (!nouveau_noaccel)
+ engine->graph.takedown(dev);
out_fb:
engine->fb.takedown(dev);
out_timer:
@@ -500,13 +532,16 @@ static void nouveau_card_takedown(struct drm_device *dev)
dev_priv->channel = NULL;
}
- engine->fifo.takedown(dev);
- engine->graph.takedown(dev);
+ if (!nouveau_noaccel) {
+ engine->fifo.takedown(dev);
+ engine->graph.takedown(dev);
+ }
engine->fb.takedown(dev);
engine->timer.takedown(dev);
engine->mc.takedown(dev);
mutex_lock(&dev->struct_mutex);
+ ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_VRAM);
ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_TT);
mutex_unlock(&dev->struct_mutex);
nouveau_sgdma_takedown(dev);
@@ -624,7 +659,10 @@ int nouveau_load(struct drm_device *dev, unsigned long flags)
dev_priv->chipset = (reg0 & 0xff00000) >> 20;
/* NV04 or NV05 */
} else if ((reg0 & 0xff00fff0) == 0x20004000) {
- dev_priv->chipset = 0x04;
+ if (reg0 & 0x00f00000)
+ dev_priv->chipset = 0x05;
+ else
+ dev_priv->chipset = 0x04;
} else
dev_priv->chipset = 0xff;
@@ -704,8 +742,8 @@ static void nouveau_close(struct drm_device *dev)
{
struct drm_nouveau_private *dev_priv = dev->dev_private;
- /* In the case of an error dev_priv may not be be allocated yet */
- if (dev_priv && dev_priv->card_type)
+ /* In the case of an error dev_priv may not be allocated yet */
+ if (dev_priv)
nouveau_card_takedown(dev);
}
@@ -795,6 +833,15 @@ int nouveau_ioctl_getparam(struct drm_device *dev, void *data,
case NOUVEAU_GETPARAM_VM_VRAM_BASE:
getparam->value = dev_priv->vm_vram_base;
break;
+ case NOUVEAU_GETPARAM_GRAPH_UNITS:
+ /* NV40 and NV50 versions are quite different, but register
+ * address is the same. User is supposed to know the card
+ * family anyway... */
+ if (dev_priv->chipset >= 0x40) {
+ getparam->value = nv_rd32(dev, NV40_PMC_GRAPH_UNITS);
+ break;
+ }
+ /* FALLTHRU */
default:
NV_ERROR(dev, "unknown parameter %lld\n", getparam->param);
return -EINVAL;
diff --git a/drivers/gpu/drm/nouveau/nouveau_ttm.c b/drivers/gpu/drm/nouveau/nouveau_ttm.c
index 187eb84e4da..c385d50f041 100644
--- a/drivers/gpu/drm/nouveau/nouveau_ttm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_ttm.c
@@ -28,45 +28,17 @@
#include "nouveau_drv.h"
-static struct vm_operations_struct nouveau_ttm_vm_ops;
-static const struct vm_operations_struct *ttm_vm_ops;
-
-static int
-nouveau_ttm_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
-{
- struct ttm_buffer_object *bo = vma->vm_private_data;
- int ret;
-
- if (unlikely(bo == NULL))
- return VM_FAULT_NOPAGE;
-
- ret = ttm_vm_ops->fault(vma, vmf);
- return ret;
-}
-
int
nouveau_ttm_mmap(struct file *filp, struct vm_area_struct *vma)
{
struct drm_file *file_priv = filp->private_data;
struct drm_nouveau_private *dev_priv =
file_priv->minor->dev->dev_private;
- int ret;
if (unlikely(vma->vm_pgoff < DRM_FILE_PAGE_OFFSET))
return drm_mmap(filp, vma);
- ret = ttm_bo_mmap(filp, vma, &dev_priv->ttm.bdev);
- if (unlikely(ret != 0))
- return ret;
-
- if (unlikely(ttm_vm_ops == NULL)) {
- ttm_vm_ops = vma->vm_ops;
- nouveau_ttm_vm_ops = *ttm_vm_ops;
- nouveau_ttm_vm_ops.fault = &nouveau_ttm_fault;
- }
-
- vma->vm_ops = &nouveau_ttm_vm_ops;
- return 0;
+ return ttm_bo_mmap(filp, vma, &dev_priv->ttm.bdev);
}
static int
diff --git a/drivers/gpu/drm/nouveau/nv04_dac.c b/drivers/gpu/drm/nouveau/nv04_dac.c
index d9f32879ba3..1d73b15d70d 100644
--- a/drivers/gpu/drm/nouveau/nv04_dac.c
+++ b/drivers/gpu/drm/nouveau/nv04_dac.c
@@ -119,7 +119,7 @@ static enum drm_connector_status nv04_dac_detect(struct drm_encoder *encoder,
struct drm_connector *connector)
{
struct drm_device *dev = encoder->dev;
- uint8_t saved_seq1, saved_pi, saved_rpc1;
+ uint8_t saved_seq1, saved_pi, saved_rpc1, saved_cr_mode;
uint8_t saved_palette0[3], saved_palette_mask;
uint32_t saved_rtest_ctrl, saved_rgen_ctrl;
int i;
@@ -135,6 +135,9 @@ static enum drm_connector_status nv04_dac_detect(struct drm_encoder *encoder,
/* only implemented for head A for now */
NVSetOwner(dev, 0);
+ saved_cr_mode = NVReadVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX);
+ NVWriteVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX, saved_cr_mode | 0x80);
+
saved_seq1 = NVReadVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX);
NVWriteVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX, saved_seq1 & ~0x20);
@@ -203,6 +206,7 @@ out:
NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_PIXEL_INDEX, saved_pi);
NVWriteVgaCrtc(dev, 0, NV_CIO_CRE_RPC1_INDEX, saved_rpc1);
NVWriteVgaSeq(dev, 0, NV_VIO_SR_CLOCK_INDEX, saved_seq1);
+ NVWriteVgaCrtc(dev, 0, NV_CIO_CR_MODE_INDEX, saved_cr_mode);
if (blue == 0x18) {
NV_INFO(dev, "Load detected on head A\n");
@@ -212,16 +216,15 @@ out:
return connector_status_disconnected;
}
-enum drm_connector_status nv17_dac_detect(struct drm_encoder *encoder,
- struct drm_connector *connector)
+uint32_t nv17_dac_sample_load(struct drm_encoder *encoder)
{
struct drm_device *dev = encoder->dev;
struct drm_nouveau_private *dev_priv = dev->dev_private;
struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb;
- uint32_t testval, regoffset = nv04_dac_output_offset(encoder);
+ uint32_t sample, testval, regoffset = nv04_dac_output_offset(encoder);
uint32_t saved_powerctrl_2 = 0, saved_powerctrl_4 = 0, saved_routput,
saved_rtest_ctrl, saved_gpio0, saved_gpio1, temp, routput;
- int head, present = 0;
+ int head;
#define RGB_TEST_DATA(r, g, b) (r << 0 | g << 10 | b << 20)
if (dcb->type == OUTPUT_TV) {
@@ -287,13 +290,7 @@ enum drm_connector_status nv17_dac_detect(struct drm_encoder *encoder,
temp | NV_PRAMDAC_TEST_CONTROL_TP_INS_EN_ASSERTED);
msleep(5);
- temp = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset);
-
- if (dcb->type == OUTPUT_TV)
- present = (nv17_tv_detect(encoder, connector, temp)
- == connector_status_connected);
- else
- present = temp & NV_PRAMDAC_TEST_CONTROL_SENSEB_ALLHI;
+ sample = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset);
temp = NVReadRAMDAC(dev, head, NV_PRAMDAC_TEST_CONTROL);
NVWriteRAMDAC(dev, head, NV_PRAMDAC_TEST_CONTROL,
@@ -310,15 +307,25 @@ enum drm_connector_status nv17_dac_detect(struct drm_encoder *encoder,
nv17_gpio_set(dev, DCB_GPIO_TVDAC1, saved_gpio1);
nv17_gpio_set(dev, DCB_GPIO_TVDAC0, saved_gpio0);
- if (present) {
- NV_INFO(dev, "Load detected on output %c\n", '@' + ffs(dcb->or));
+ return sample;
+}
+
+static enum drm_connector_status
+nv17_dac_detect(struct drm_encoder *encoder, struct drm_connector *connector)
+{
+ struct drm_device *dev = encoder->dev;
+ struct dcb_entry *dcb = nouveau_encoder(encoder)->dcb;
+ uint32_t sample = nv17_dac_sample_load(encoder);
+
+ if (sample & NV_PRAMDAC_TEST_CONTROL_SENSEB_ALLHI) {
+ NV_INFO(dev, "Load detected on output %c\n",
+ '@' + ffs(dcb->or));
return connector_status_connected;
+ } else {
+ return connector_status_disconnected;
}
-
- return connector_status_disconnected;
}
-
static bool nv04_dac_mode_fixup(struct drm_encoder *encoder,
struct drm_display_mode *mode,
struct drm_display_mode *adjusted_mode)
diff --git a/drivers/gpu/drm/nouveau/nv04_fbcon.c b/drivers/gpu/drm/nouveau/nv04_fbcon.c
index 09a31071ee5..fd01caabd5c 100644
--- a/drivers/gpu/drm/nouveau/nv04_fbcon.c
+++ b/drivers/gpu/drm/nouveau/nv04_fbcon.c
@@ -27,7 +27,7 @@
#include "nouveau_dma.h"
#include "nouveau_fbcon.h"
-static void
+void
nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
{
struct nouveau_fbcon_par *par = info->par;
@@ -39,8 +39,7 @@ nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
return;
if (!(info->flags & FBINFO_HWACCEL_DISABLED) && RING_SPACE(chan, 4)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
}
if (info->flags & FBINFO_HWACCEL_DISABLED) {
@@ -55,21 +54,19 @@ nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
FIRE_RING(chan);
}
-static void
+void
nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
{
struct nouveau_fbcon_par *par = info->par;
struct drm_device *dev = par->dev;
struct drm_nouveau_private *dev_priv = dev->dev_private;
struct nouveau_channel *chan = dev_priv->channel;
- uint32_t color = ((uint32_t *) info->pseudo_palette)[rect->color];
if (info->state != FBINFO_STATE_RUNNING)
return;
if (!(info->flags & FBINFO_HWACCEL_DISABLED) && RING_SPACE(chan, 7)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
}
if (info->flags & FBINFO_HWACCEL_DISABLED) {
@@ -80,14 +77,18 @@ nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
BEGIN_RING(chan, NvSubGdiRect, 0x02fc, 1);
OUT_RING(chan, (rect->rop != ROP_COPY) ? 1 : 3);
BEGIN_RING(chan, NvSubGdiRect, 0x03fc, 1);
- OUT_RING(chan, color);
+ if (info->fix.visual == FB_VISUAL_TRUECOLOR ||
+ info->fix.visual == FB_VISUAL_DIRECTCOLOR)
+ OUT_RING(chan, ((uint32_t *)info->pseudo_palette)[rect->color]);
+ else
+ OUT_RING(chan, rect->color);
BEGIN_RING(chan, NvSubGdiRect, 0x0400, 2);
OUT_RING(chan, (rect->dx << 16) | rect->dy);
OUT_RING(chan, (rect->width << 16) | rect->height);
FIRE_RING(chan);
}
-static void
+void
nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
{
struct nouveau_fbcon_par *par = info->par;
@@ -109,8 +110,7 @@ nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
}
if (!(info->flags & FBINFO_HWACCEL_DISABLED) && RING_SPACE(chan, 8)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
}
if (info->flags & FBINFO_HWACCEL_DISABLED) {
@@ -144,8 +144,7 @@ nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
int iter_len = dsize > 128 ? 128 : dsize;
if (RING_SPACE(chan, iter_len + 1)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
cfb_imageblit(info, image);
return;
}
@@ -184,6 +183,7 @@ nv04_fbcon_accel_init(struct fb_info *info)
struct drm_device *dev = par->dev;
struct drm_nouveau_private *dev_priv = dev->dev_private;
struct nouveau_channel *chan = dev_priv->channel;
+ const int sub = NvSubCtxSurf2D;
int surface_fmt, pattern_fmt, rect_fmt;
int ret;
@@ -242,30 +242,29 @@ nv04_fbcon_accel_init(struct fb_info *info)
return ret;
if (RING_SPACE(chan, 49)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
return 0;
}
- BEGIN_RING(chan, 1, 0x0000, 1);
+ BEGIN_RING(chan, sub, 0x0000, 1);
OUT_RING(chan, NvCtxSurf2D);
- BEGIN_RING(chan, 1, 0x0184, 2);
+ BEGIN_RING(chan, sub, 0x0184, 2);
OUT_RING(chan, NvDmaFB);
OUT_RING(chan, NvDmaFB);
- BEGIN_RING(chan, 1, 0x0300, 4);
+ BEGIN_RING(chan, sub, 0x0300, 4);
OUT_RING(chan, surface_fmt);
OUT_RING(chan, info->fix.line_length | (info->fix.line_length << 16));
OUT_RING(chan, info->fix.smem_start - dev->mode_config.fb_base);
OUT_RING(chan, info->fix.smem_start - dev->mode_config.fb_base);
- BEGIN_RING(chan, 1, 0x0000, 1);
+ BEGIN_RING(chan, sub, 0x0000, 1);
OUT_RING(chan, NvRop);
- BEGIN_RING(chan, 1, 0x0300, 1);
+ BEGIN_RING(chan, sub, 0x0300, 1);
OUT_RING(chan, 0x55);
- BEGIN_RING(chan, 1, 0x0000, 1);
+ BEGIN_RING(chan, sub, 0x0000, 1);
OUT_RING(chan, NvImagePatt);
- BEGIN_RING(chan, 1, 0x0300, 8);
+ BEGIN_RING(chan, sub, 0x0300, 8);
OUT_RING(chan, pattern_fmt);
#ifdef __BIG_ENDIAN
OUT_RING(chan, 2);
@@ -279,9 +278,9 @@ nv04_fbcon_accel_init(struct fb_info *info)
OUT_RING(chan, ~0);
OUT_RING(chan, ~0);
- BEGIN_RING(chan, 1, 0x0000, 1);
+ BEGIN_RING(chan, sub, 0x0000, 1);
OUT_RING(chan, NvClipRect);
- BEGIN_RING(chan, 1, 0x0300, 2);
+ BEGIN_RING(chan, sub, 0x0300, 2);
OUT_RING(chan, 0);
OUT_RING(chan, (info->var.yres_virtual << 16) | info->var.xres_virtual);
@@ -308,9 +307,6 @@ nv04_fbcon_accel_init(struct fb_info *info)
FIRE_RING(chan);
- info->fbops->fb_fillrect = nv04_fbcon_fillrect;
- info->fbops->fb_copyarea = nv04_fbcon_copyarea;
- info->fbops->fb_imageblit = nv04_fbcon_imageblit;
return 0;
}
diff --git a/drivers/gpu/drm/nouveau/nv04_fifo.c b/drivers/gpu/drm/nouveau/nv04_fifo.c
index 0c3cd53c731..f31347b8c9b 100644
--- a/drivers/gpu/drm/nouveau/nv04_fifo.c
+++ b/drivers/gpu/drm/nouveau/nv04_fifo.c
@@ -71,6 +71,40 @@ nv04_fifo_reassign(struct drm_device *dev, bool enable)
return (reassign == 1);
}
+bool
+nv04_fifo_cache_flush(struct drm_device *dev)
+{
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ struct nouveau_timer_engine *ptimer = &dev_priv->engine.timer;
+ uint64_t start = ptimer->read(dev);
+
+ do {
+ if (nv_rd32(dev, NV03_PFIFO_CACHE1_GET) ==
+ nv_rd32(dev, NV03_PFIFO_CACHE1_PUT))
+ return true;
+
+ } while (ptimer->read(dev) - start < 100000000);
+
+ NV_ERROR(dev, "Timeout flushing the PFIFO cache.\n");
+
+ return false;
+}
+
+bool
+nv04_fifo_cache_pull(struct drm_device *dev, bool enable)
+{
+ uint32_t pull = nv_rd32(dev, NV04_PFIFO_CACHE1_PULL0);
+
+ if (enable) {
+ nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, pull | 1);
+ } else {
+ nv_wr32(dev, NV04_PFIFO_CACHE1_PULL0, pull & ~1);
+ nv_wr32(dev, NV04_PFIFO_CACHE1_HASH, 0);
+ }
+
+ return !!(pull & 1);
+}
+
int
nv04_fifo_channel_id(struct drm_device *dev)
{
diff --git a/drivers/gpu/drm/nouveau/nv04_graph.c b/drivers/gpu/drm/nouveau/nv04_graph.c
index d561d773c0f..e260986ea65 100644
--- a/drivers/gpu/drm/nouveau/nv04_graph.c
+++ b/drivers/gpu/drm/nouveau/nv04_graph.c
@@ -28,6 +28,10 @@
#include "nouveau_drv.h"
static uint32_t nv04_graph_ctx_regs[] = {
+ 0x0040053c,
+ 0x00400544,
+ 0x00400540,
+ 0x00400548,
NV04_PGRAPH_CTX_SWITCH1,
NV04_PGRAPH_CTX_SWITCH2,
NV04_PGRAPH_CTX_SWITCH3,
@@ -102,69 +106,69 @@ static uint32_t nv04_graph_ctx_regs[] = {
NV04_PGRAPH_PATT_COLOR0,
NV04_PGRAPH_PATT_COLOR1,
NV04_PGRAPH_PATT_COLORRAM+0x00,
- NV04_PGRAPH_PATT_COLORRAM+0x01,
- NV04_PGRAPH_PATT_COLORRAM+0x02,
- NV04_PGRAPH_PATT_COLORRAM+0x03,
NV04_PGRAPH_PATT_COLORRAM+0x04,
- NV04_PGRAPH_PATT_COLORRAM+0x05,
- NV04_PGRAPH_PATT_COLORRAM+0x06,
- NV04_PGRAPH_PATT_COLORRAM+0x07,
NV04_PGRAPH_PATT_COLORRAM+0x08,
- NV04_PGRAPH_PATT_COLORRAM+0x09,
- NV04_PGRAPH_PATT_COLORRAM+0x0A,
- NV04_PGRAPH_PATT_COLORRAM+0x0B,
- NV04_PGRAPH_PATT_COLORRAM+0x0C,
- NV04_PGRAPH_PATT_COLORRAM+0x0D,
- NV04_PGRAPH_PATT_COLORRAM+0x0E,
- NV04_PGRAPH_PATT_COLORRAM+0x0F,
+ NV04_PGRAPH_PATT_COLORRAM+0x0c,
NV04_PGRAPH_PATT_COLORRAM+0x10,
- NV04_PGRAPH_PATT_COLORRAM+0x11,
- NV04_PGRAPH_PATT_COLORRAM+0x12,
- NV04_PGRAPH_PATT_COLORRAM+0x13,
NV04_PGRAPH_PATT_COLORRAM+0x14,
- NV04_PGRAPH_PATT_COLORRAM+0x15,
- NV04_PGRAPH_PATT_COLORRAM+0x16,
- NV04_PGRAPH_PATT_COLORRAM+0x17,
NV04_PGRAPH_PATT_COLORRAM+0x18,
- NV04_PGRAPH_PATT_COLORRAM+0x19,
- NV04_PGRAPH_PATT_COLORRAM+0x1A,
- NV04_PGRAPH_PATT_COLORRAM+0x1B,
- NV04_PGRAPH_PATT_COLORRAM+0x1C,
- NV04_PGRAPH_PATT_COLORRAM+0x1D,
- NV04_PGRAPH_PATT_COLORRAM+0x1E,
- NV04_PGRAPH_PATT_COLORRAM+0x1F,
+ NV04_PGRAPH_PATT_COLORRAM+0x1c,
NV04_PGRAPH_PATT_COLORRAM+0x20,
- NV04_PGRAPH_PATT_COLORRAM+0x21,
- NV04_PGRAPH_PATT_COLORRAM+0x22,
- NV04_PGRAPH_PATT_COLORRAM+0x23,
NV04_PGRAPH_PATT_COLORRAM+0x24,
- NV04_PGRAPH_PATT_COLORRAM+0x25,
- NV04_PGRAPH_PATT_COLORRAM+0x26,
- NV04_PGRAPH_PATT_COLORRAM+0x27,
NV04_PGRAPH_PATT_COLORRAM+0x28,
- NV04_PGRAPH_PATT_COLORRAM+0x29,
- NV04_PGRAPH_PATT_COLORRAM+0x2A,
- NV04_PGRAPH_PATT_COLORRAM+0x2B,
- NV04_PGRAPH_PATT_COLORRAM+0x2C,
- NV04_PGRAPH_PATT_COLORRAM+0x2D,
- NV04_PGRAPH_PATT_COLORRAM+0x2E,
- NV04_PGRAPH_PATT_COLORRAM+0x2F,
+ NV04_PGRAPH_PATT_COLORRAM+0x2c,
NV04_PGRAPH_PATT_COLORRAM+0x30,
- NV04_PGRAPH_PATT_COLORRAM+0x31,
- NV04_PGRAPH_PATT_COLORRAM+0x32,
- NV04_PGRAPH_PATT_COLORRAM+0x33,
NV04_PGRAPH_PATT_COLORRAM+0x34,
- NV04_PGRAPH_PATT_COLORRAM+0x35,
- NV04_PGRAPH_PATT_COLORRAM+0x36,
- NV04_PGRAPH_PATT_COLORRAM+0x37,
NV04_PGRAPH_PATT_COLORRAM+0x38,
- NV04_PGRAPH_PATT_COLORRAM+0x39,
- NV04_PGRAPH_PATT_COLORRAM+0x3A,
- NV04_PGRAPH_PATT_COLORRAM+0x3B,
- NV04_PGRAPH_PATT_COLORRAM+0x3C,
- NV04_PGRAPH_PATT_COLORRAM+0x3D,
- NV04_PGRAPH_PATT_COLORRAM+0x3E,
- NV04_PGRAPH_PATT_COLORRAM+0x3F,
+ NV04_PGRAPH_PATT_COLORRAM+0x3c,
+ NV04_PGRAPH_PATT_COLORRAM+0x40,
+ NV04_PGRAPH_PATT_COLORRAM+0x44,
+ NV04_PGRAPH_PATT_COLORRAM+0x48,
+ NV04_PGRAPH_PATT_COLORRAM+0x4c,
+ NV04_PGRAPH_PATT_COLORRAM+0x50,
+ NV04_PGRAPH_PATT_COLORRAM+0x54,
+ NV04_PGRAPH_PATT_COLORRAM+0x58,
+ NV04_PGRAPH_PATT_COLORRAM+0x5c,
+ NV04_PGRAPH_PATT_COLORRAM+0x60,
+ NV04_PGRAPH_PATT_COLORRAM+0x64,
+ NV04_PGRAPH_PATT_COLORRAM+0x68,
+ NV04_PGRAPH_PATT_COLORRAM+0x6c,
+ NV04_PGRAPH_PATT_COLORRAM+0x70,
+ NV04_PGRAPH_PATT_COLORRAM+0x74,
+ NV04_PGRAPH_PATT_COLORRAM+0x78,
+ NV04_PGRAPH_PATT_COLORRAM+0x7c,
+ NV04_PGRAPH_PATT_COLORRAM+0x80,
+ NV04_PGRAPH_PATT_COLORRAM+0x84,
+ NV04_PGRAPH_PATT_COLORRAM+0x88,
+ NV04_PGRAPH_PATT_COLORRAM+0x8c,
+ NV04_PGRAPH_PATT_COLORRAM+0x90,
+ NV04_PGRAPH_PATT_COLORRAM+0x94,
+ NV04_PGRAPH_PATT_COLORRAM+0x98,
+ NV04_PGRAPH_PATT_COLORRAM+0x9c,
+ NV04_PGRAPH_PATT_COLORRAM+0xa0,
+ NV04_PGRAPH_PATT_COLORRAM+0xa4,
+ NV04_PGRAPH_PATT_COLORRAM+0xa8,
+ NV04_PGRAPH_PATT_COLORRAM+0xac,
+ NV04_PGRAPH_PATT_COLORRAM+0xb0,
+ NV04_PGRAPH_PATT_COLORRAM+0xb4,
+ NV04_PGRAPH_PATT_COLORRAM+0xb8,
+ NV04_PGRAPH_PATT_COLORRAM+0xbc,
+ NV04_PGRAPH_PATT_COLORRAM+0xc0,
+ NV04_PGRAPH_PATT_COLORRAM+0xc4,
+ NV04_PGRAPH_PATT_COLORRAM+0xc8,
+ NV04_PGRAPH_PATT_COLORRAM+0xcc,
+ NV04_PGRAPH_PATT_COLORRAM+0xd0,
+ NV04_PGRAPH_PATT_COLORRAM+0xd4,
+ NV04_PGRAPH_PATT_COLORRAM+0xd8,
+ NV04_PGRAPH_PATT_COLORRAM+0xdc,
+ NV04_PGRAPH_PATT_COLORRAM+0xe0,
+ NV04_PGRAPH_PATT_COLORRAM+0xe4,
+ NV04_PGRAPH_PATT_COLORRAM+0xe8,
+ NV04_PGRAPH_PATT_COLORRAM+0xec,
+ NV04_PGRAPH_PATT_COLORRAM+0xf0,
+ NV04_PGRAPH_PATT_COLORRAM+0xf4,
+ NV04_PGRAPH_PATT_COLORRAM+0xf8,
+ NV04_PGRAPH_PATT_COLORRAM+0xfc,
NV04_PGRAPH_PATTERN,
0x0040080c,
NV04_PGRAPH_PATTERN_SHAPE,
@@ -247,14 +251,6 @@ static uint32_t nv04_graph_ctx_regs[] = {
0x004004f8,
0x0040047c,
0x004004fc,
- 0x0040053c,
- 0x00400544,
- 0x00400540,
- 0x00400548,
- 0x00400560,
- 0x00400568,
- 0x00400564,
- 0x0040056c,
0x00400534,
0x00400538,
0x00400514,
@@ -341,9 +337,8 @@ static uint32_t nv04_graph_ctx_regs[] = {
0x00400500,
0x00400504,
NV04_PGRAPH_VALID1,
- NV04_PGRAPH_VALID2
-
-
+ NV04_PGRAPH_VALID2,
+ NV04_PGRAPH_DEBUG_3
};
struct graph_state {
@@ -388,6 +383,18 @@ nv04_graph_context_switch(struct drm_device *dev)
pgraph->fifo_access(dev, true);
}
+static uint32_t *ctx_reg(struct graph_state *ctx, uint32_t reg)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(nv04_graph_ctx_regs); i++) {
+ if (nv04_graph_ctx_regs[i] == reg)
+ return &ctx->nv04[i];
+ }
+
+ return NULL;
+}
+
int nv04_graph_create_context(struct nouveau_channel *chan)
{
struct graph_state *pgraph_ctx;
@@ -398,15 +405,8 @@ int nv04_graph_create_context(struct nouveau_channel *chan)
if (pgraph_ctx == NULL)
return -ENOMEM;
- /* dev_priv->fifos[channel].pgraph_ctx_user = channel << 24; */
- pgraph_ctx->nv04[0] = 0x0001ffff;
- /* is it really needed ??? */
-#if 0
- dev_priv->fifos[channel].pgraph_ctx[1] =
- nv_rd32(dev, NV_PGRAPH_DEBUG_4);
- dev_priv->fifos[channel].pgraph_ctx[2] =
- nv_rd32(dev, 0x004006b0);
-#endif
+ *ctx_reg(pgraph_ctx, NV04_PGRAPH_DEBUG_3) = 0xfad4ff31;
+
return 0;
}
@@ -429,9 +429,13 @@ int nv04_graph_load_context(struct nouveau_channel *chan)
nv_wr32(dev, nv04_graph_ctx_regs[i], pgraph_ctx->nv04[i]);
nv_wr32(dev, NV04_PGRAPH_CTX_CONTROL, 0x10010100);
- nv_wr32(dev, NV04_PGRAPH_CTX_USER, chan->id << 24);
+
+ tmp = nv_rd32(dev, NV04_PGRAPH_CTX_USER) & 0x00ffffff;
+ nv_wr32(dev, NV04_PGRAPH_CTX_USER, tmp | chan->id << 24);
+
tmp = nv_rd32(dev, NV04_PGRAPH_FFINTFC_ST2);
nv_wr32(dev, NV04_PGRAPH_FFINTFC_ST2, tmp & 0x000fffff);
+
return 0;
}
@@ -494,7 +498,7 @@ int nv04_graph_init(struct drm_device *dev)
nv_wr32(dev, NV04_PGRAPH_STATE , 0xFFFFFFFF);
nv_wr32(dev, NV04_PGRAPH_CTX_CONTROL , 0x10000100);
tmp = nv_rd32(dev, NV04_PGRAPH_CTX_USER) & 0x00ffffff;
- tmp |= dev_priv->engine.fifo.channels << 24;
+ tmp |= (dev_priv->engine.fifo.channels - 1) << 24;
nv_wr32(dev, NV04_PGRAPH_CTX_USER, tmp);
/* These don't belong here, they're part of a per-channel context */
@@ -533,7 +537,7 @@ nv04_graph_mthd_set_operation(struct nouveau_channel *chan, int grclass,
int mthd, uint32_t data)
{
struct drm_device *dev = chan->dev;
- uint32_t instance = nv_rd32(dev, NV04_PGRAPH_CTX_SWITCH4) & 0xffff;
+ uint32_t instance = (nv_rd32(dev, NV04_PGRAPH_CTX_SWITCH4) & 0xffff) << 4;
int subc = (nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR) >> 13) & 0x7;
uint32_t tmp;
@@ -547,7 +551,7 @@ nv04_graph_mthd_set_operation(struct nouveau_channel *chan, int grclass,
return 0;
}
-static struct nouveau_pgraph_object_method nv04_graph_mthds_m2mf[] = {
+static struct nouveau_pgraph_object_method nv04_graph_mthds_sw[] = {
{ 0x0150, nv04_graph_mthd_set_ref },
{}
};
@@ -558,7 +562,7 @@ static struct nouveau_pgraph_object_method nv04_graph_mthds_set_operation[] = {
};
struct nouveau_pgraph_object_class nv04_graph_grclass[] = {
- { 0x0039, false, nv04_graph_mthds_m2mf },
+ { 0x0039, false, NULL },
{ 0x004a, false, nv04_graph_mthds_set_operation }, /* gdirect */
{ 0x005f, false, nv04_graph_mthds_set_operation }, /* imageblit */
{ 0x0061, false, nv04_graph_mthds_set_operation }, /* ifc */
@@ -574,6 +578,7 @@ struct nouveau_pgraph_object_class nv04_graph_grclass[] = {
{ 0x0053, false, NULL }, /* surf3d */
{ 0x0054, false, NULL }, /* tex_tri */
{ 0x0055, false, NULL }, /* multitex_tri */
+ { 0x506e, true, nv04_graph_mthds_sw },
{}
};
diff --git a/drivers/gpu/drm/nouveau/nv04_instmem.c b/drivers/gpu/drm/nouveau/nv04_instmem.c
index a20c206625a..a3b9563a6f6 100644
--- a/drivers/gpu/drm/nouveau/nv04_instmem.c
+++ b/drivers/gpu/drm/nouveau/nv04_instmem.c
@@ -30,7 +30,7 @@ nv04_instmem_determine_amount(struct drm_device *dev)
* of vram. For now, only reserve a small piece until we know
* more about what each chipset requires.
*/
- switch (dev_priv->chipset & 0xf0) {
+ switch (dev_priv->chipset) {
case 0x40:
case 0x47:
case 0x49:
diff --git a/drivers/gpu/drm/nouveau/nv10_fb.c b/drivers/gpu/drm/nouveau/nv10_fb.c
index 79e2d104d70..cc5cda44e50 100644
--- a/drivers/gpu/drm/nouveau/nv10_fb.c
+++ b/drivers/gpu/drm/nouveau/nv10_fb.c
@@ -3,17 +3,37 @@
#include "nouveau_drv.h"
#include "nouveau_drm.h"
+void
+nv10_fb_set_region_tiling(struct drm_device *dev, int i, uint32_t addr,
+ uint32_t size, uint32_t pitch)
+{
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ uint32_t limit = max(1u, addr + size) - 1;
+
+ if (pitch) {
+ if (dev_priv->card_type >= NV_20)
+ addr |= 1;
+ else
+ addr |= 1 << 31;
+ }
+
+ nv_wr32(dev, NV10_PFB_TLIMIT(i), limit);
+ nv_wr32(dev, NV10_PFB_TSIZE(i), pitch);
+ nv_wr32(dev, NV10_PFB_TILE(i), addr);
+}
+
int
nv10_fb_init(struct drm_device *dev)
{
- uint32_t fb_bar_size;
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ struct nouveau_fb_engine *pfb = &dev_priv->engine.fb;
int i;
- fb_bar_size = drm_get_resource_len(dev, 0) - 1;
- for (i = 0; i < NV10_PFB_TILE__SIZE; i++) {
- nv_wr32(dev, NV10_PFB_TILE(i), 0);
- nv_wr32(dev, NV10_PFB_TLIMIT(i), fb_bar_size);
- }
+ pfb->num_tiles = NV10_PFB_TILE__SIZE;
+
+ /* Turn all the tiling regions off. */
+ for (i = 0; i < pfb->num_tiles; i++)
+ pfb->set_region_tiling(dev, i, 0, 0, 0);
return 0;
}
diff --git a/drivers/gpu/drm/nouveau/nv10_graph.c b/drivers/gpu/drm/nouveau/nv10_graph.c
index 6870e0ee2e7..fcf2cdd1949 100644
--- a/drivers/gpu/drm/nouveau/nv10_graph.c
+++ b/drivers/gpu/drm/nouveau/nv10_graph.c
@@ -807,6 +807,20 @@ void nv10_graph_destroy_context(struct nouveau_channel *chan)
chan->pgraph_ctx = NULL;
}
+void
+nv10_graph_set_region_tiling(struct drm_device *dev, int i, uint32_t addr,
+ uint32_t size, uint32_t pitch)
+{
+ uint32_t limit = max(1u, addr + size) - 1;
+
+ if (pitch)
+ addr |= 1 << 31;
+
+ nv_wr32(dev, NV10_PGRAPH_TLIMIT(i), limit);
+ nv_wr32(dev, NV10_PGRAPH_TSIZE(i), pitch);
+ nv_wr32(dev, NV10_PGRAPH_TILE(i), addr);
+}
+
int nv10_graph_init(struct drm_device *dev)
{
struct drm_nouveau_private *dev_priv = dev->dev_private;
@@ -838,17 +852,9 @@ int nv10_graph_init(struct drm_device *dev)
} else
nv_wr32(dev, NV10_PGRAPH_DEBUG_4, 0x00000000);
- /* copy tile info from PFB */
- for (i = 0; i < NV10_PFB_TILE__SIZE; i++) {
- nv_wr32(dev, NV10_PGRAPH_TILE(i),
- nv_rd32(dev, NV10_PFB_TILE(i)));
- nv_wr32(dev, NV10_PGRAPH_TLIMIT(i),
- nv_rd32(dev, NV10_PFB_TLIMIT(i)));
- nv_wr32(dev, NV10_PGRAPH_TSIZE(i),
- nv_rd32(dev, NV10_PFB_TSIZE(i)));
- nv_wr32(dev, NV10_PGRAPH_TSTATUS(i),
- nv_rd32(dev, NV10_PFB_TSTATUS(i)));
- }
+ /* Turn all the tiling regions off. */
+ for (i = 0; i < NV10_PFB_TILE__SIZE; i++)
+ nv10_graph_set_region_tiling(dev, i, 0, 0, 0);
nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH1, 0x00000000);
nv_wr32(dev, NV10_PGRAPH_CTX_SWITCH2, 0x00000000);
diff --git a/drivers/gpu/drm/nouveau/nv17_tv.c b/drivers/gpu/drm/nouveau/nv17_tv.c
index 81c01353a9f..21ac6e49b6e 100644
--- a/drivers/gpu/drm/nouveau/nv17_tv.c
+++ b/drivers/gpu/drm/nouveau/nv17_tv.c
@@ -33,13 +33,103 @@
#include "nouveau_hw.h"
#include "nv17_tv.h"
-enum drm_connector_status nv17_tv_detect(struct drm_encoder *encoder,
- struct drm_connector *connector,
- uint32_t pin_mask)
+static uint32_t nv42_tv_sample_load(struct drm_encoder *encoder)
{
+ struct drm_device *dev = encoder->dev;
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ uint32_t testval, regoffset = nv04_dac_output_offset(encoder);
+ uint32_t gpio0, gpio1, fp_htotal, fp_hsync_start, fp_hsync_end,
+ fp_control, test_ctrl, dacclk, ctv_14, ctv_1c, ctv_6c;
+ uint32_t sample = 0;
+ int head;
+
+#define RGB_TEST_DATA(r, g, b) (r << 0 | g << 10 | b << 20)
+ testval = RGB_TEST_DATA(0x82, 0xeb, 0x82);
+ if (dev_priv->vbios->tvdactestval)
+ testval = dev_priv->vbios->tvdactestval;
+
+ dacclk = NVReadRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset);
+ head = (dacclk & 0x100) >> 8;
+
+ /* Save the previous state. */
+ gpio1 = nv17_gpio_get(dev, DCB_GPIO_TVDAC1);
+ gpio0 = nv17_gpio_get(dev, DCB_GPIO_TVDAC0);
+ fp_htotal = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_HTOTAL);
+ fp_hsync_start = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_START);
+ fp_hsync_end = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_END);
+ fp_control = NVReadRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL);
+ test_ctrl = NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset);
+ ctv_1c = NVReadRAMDAC(dev, head, 0x680c1c);
+ ctv_14 = NVReadRAMDAC(dev, head, 0x680c14);
+ ctv_6c = NVReadRAMDAC(dev, head, 0x680c6c);
+
+ /* Prepare the DAC for load detection. */
+ nv17_gpio_set(dev, DCB_GPIO_TVDAC1, true);
+ nv17_gpio_set(dev, DCB_GPIO_TVDAC0, true);
+
+ NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HTOTAL, 1343);
+ NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_START, 1047);
+ NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_END, 1183);
+ NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL,
+ NV_PRAMDAC_FP_TG_CONTROL_DISPEN_POS |
+ NV_PRAMDAC_FP_TG_CONTROL_WIDTH_12 |
+ NV_PRAMDAC_FP_TG_CONTROL_READ_PROG |
+ NV_PRAMDAC_FP_TG_CONTROL_HSYNC_POS |
+ NV_PRAMDAC_FP_TG_CONTROL_VSYNC_POS);
+
+ NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset, 0);
+
+ NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset,
+ (dacclk & ~0xff) | 0x22);
+ msleep(1);
+ NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset,
+ (dacclk & ~0xff) | 0x21);
+
+ NVWriteRAMDAC(dev, head, 0x680c1c, 1 << 20);
+ NVWriteRAMDAC(dev, head, 0x680c14, 4 << 16);
+
+ /* Sample pin 0x4 (usually S-video luma). */
+ NVWriteRAMDAC(dev, head, 0x680c6c, testval >> 10 & 0x3ff);
+ msleep(20);
+ sample |= NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset)
+ & 0x4 << 28;
+
+ /* Sample the remaining pins. */
+ NVWriteRAMDAC(dev, head, 0x680c6c, testval & 0x3ff);
+ msleep(20);
+ sample |= NVReadRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset)
+ & 0xa << 28;
+
+ /* Restore the previous state. */
+ NVWriteRAMDAC(dev, head, 0x680c1c, ctv_1c);
+ NVWriteRAMDAC(dev, head, 0x680c14, ctv_14);
+ NVWriteRAMDAC(dev, head, 0x680c6c, ctv_6c);
+ NVWriteRAMDAC(dev, 0, NV_PRAMDAC_DACCLK + regoffset, dacclk);
+ NVWriteRAMDAC(dev, 0, NV_PRAMDAC_TEST_CONTROL + regoffset, test_ctrl);
+ NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_TG_CONTROL, fp_control);
+ NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_END, fp_hsync_end);
+ NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HSYNC_START, fp_hsync_start);
+ NVWriteRAMDAC(dev, head, NV_PRAMDAC_FP_HTOTAL, fp_htotal);
+ nv17_gpio_set(dev, DCB_GPIO_TVDAC1, gpio1);
+ nv17_gpio_set(dev, DCB_GPIO_TVDAC0, gpio0);
+
+ return sample;
+}
+
+static enum drm_connector_status
+nv17_tv_detect(struct drm_encoder *encoder, struct drm_connector *connector)
+{
+ struct drm_device *dev = encoder->dev;
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ struct drm_mode_config *conf = &dev->mode_config;
struct nv17_tv_encoder *tv_enc = to_tv_enc(encoder);
+ struct dcb_entry *dcb = tv_enc->base.dcb;
- tv_enc->pin_mask = pin_mask >> 28 & 0xe;
+ if (dev_priv->chipset == 0x42 ||
+ dev_priv->chipset == 0x43)
+ tv_enc->pin_mask = nv42_tv_sample_load(encoder) >> 28 & 0xe;
+ else
+ tv_enc->pin_mask = nv17_dac_sample_load(encoder) >> 28 & 0xe;
switch (tv_enc->pin_mask) {
case 0x2:
@@ -50,7 +140,7 @@ enum drm_connector_status nv17_tv_detect(struct drm_encoder *encoder,
tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_SVIDEO;
break;
case 0xe:
- if (nouveau_encoder(encoder)->dcb->tvconf.has_component_output)
+ if (dcb->tvconf.has_component_output)
tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_Component;
else
tv_enc->subconnector = DRM_MODE_SUBCONNECTOR_SCART;
@@ -61,11 +151,16 @@ enum drm_connector_status nv17_tv_detect(struct drm_encoder *encoder,
}
drm_connector_property_set_value(connector,
- encoder->dev->mode_config.tv_subconnector_property,
- tv_enc->subconnector);
+ conf->tv_subconnector_property,
+ tv_enc->subconnector);
- return tv_enc->subconnector ? connector_status_connected :
- connector_status_disconnected;
+ if (tv_enc->subconnector) {
+ NV_INFO(dev, "Load detected on output %c\n",
+ '@' + ffs(dcb->or));
+ return connector_status_connected;
+ } else {
+ return connector_status_disconnected;
+ }
}
static const struct {
@@ -484,6 +579,8 @@ static void nv17_tv_restore(struct drm_encoder *encoder)
nouveau_encoder(encoder)->restore.output);
nv17_tv_state_load(dev, &to_tv_enc(encoder)->saved_state);
+
+ nouveau_encoder(encoder)->last_dpms = NV_DPMS_CLEARED;
}
static int nv17_tv_create_resources(struct drm_encoder *encoder,
@@ -633,7 +730,7 @@ static struct drm_encoder_helper_funcs nv17_tv_helper_funcs = {
.prepare = nv17_tv_prepare,
.commit = nv17_tv_commit,
.mode_set = nv17_tv_mode_set,
- .detect = nv17_dac_detect,
+ .detect = nv17_tv_detect,
};
static struct drm_encoder_slave_funcs nv17_tv_slave_funcs = {
diff --git a/drivers/gpu/drm/nouveau/nv20_graph.c b/drivers/gpu/drm/nouveau/nv20_graph.c
index 18ba74f1970..d6fc0a82f03 100644
--- a/drivers/gpu/drm/nouveau/nv20_graph.c
+++ b/drivers/gpu/drm/nouveau/nv20_graph.c
@@ -514,6 +514,27 @@ nv20_graph_rdi(struct drm_device *dev)
nouveau_wait_for_idle(dev);
}
+void
+nv20_graph_set_region_tiling(struct drm_device *dev, int i, uint32_t addr,
+ uint32_t size, uint32_t pitch)
+{
+ uint32_t limit = max(1u, addr + size) - 1;
+
+ if (pitch)
+ addr |= 1;
+
+ nv_wr32(dev, NV20_PGRAPH_TLIMIT(i), limit);
+ nv_wr32(dev, NV20_PGRAPH_TSIZE(i), pitch);
+ nv_wr32(dev, NV20_PGRAPH_TILE(i), addr);
+
+ nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0030 + 4 * i);
+ nv_wr32(dev, NV10_PGRAPH_RDI_DATA, limit);
+ nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0050 + 4 * i);
+ nv_wr32(dev, NV10_PGRAPH_RDI_DATA, pitch);
+ nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0010 + 4 * i);
+ nv_wr32(dev, NV10_PGRAPH_RDI_DATA, addr);
+}
+
int
nv20_graph_init(struct drm_device *dev)
{
@@ -572,27 +593,10 @@ nv20_graph_init(struct drm_device *dev)
nv_wr32(dev, NV10_PGRAPH_RDI_DATA , 0x00000030);
}
- /* copy tile info from PFB */
- for (i = 0; i < NV10_PFB_TILE__SIZE; i++) {
- nv_wr32(dev, 0x00400904 + i * 0x10,
- nv_rd32(dev, NV10_PFB_TLIMIT(i)));
- /* which is NV40_PGRAPH_TLIMIT0(i) ?? */
- nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0030 + i * 4);
- nv_wr32(dev, NV10_PGRAPH_RDI_DATA,
- nv_rd32(dev, NV10_PFB_TLIMIT(i)));
- nv_wr32(dev, 0x00400908 + i * 0x10,
- nv_rd32(dev, NV10_PFB_TSIZE(i)));
- /* which is NV40_PGRAPH_TSIZE0(i) ?? */
- nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0050 + i * 4);
- nv_wr32(dev, NV10_PGRAPH_RDI_DATA,
- nv_rd32(dev, NV10_PFB_TSIZE(i)));
- nv_wr32(dev, 0x00400900 + i * 0x10,
- nv_rd32(dev, NV10_PFB_TILE(i)));
- /* which is NV40_PGRAPH_TILE0(i) ?? */
- nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0010 + i * 4);
- nv_wr32(dev, NV10_PGRAPH_RDI_DATA,
- nv_rd32(dev, NV10_PFB_TILE(i)));
- }
+ /* Turn all the tiling regions off. */
+ for (i = 0; i < NV10_PFB_TILE__SIZE; i++)
+ nv20_graph_set_region_tiling(dev, i, 0, 0, 0);
+
for (i = 0; i < 8; i++) {
nv_wr32(dev, 0x400980 + i * 4, nv_rd32(dev, 0x100300 + i * 4));
nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, 0x00EA0090 + i * 4);
@@ -704,18 +708,9 @@ nv30_graph_init(struct drm_device *dev)
nv_wr32(dev, 0x4000c0, 0x00000016);
- /* copy tile info from PFB */
- for (i = 0; i < NV10_PFB_TILE__SIZE; i++) {
- nv_wr32(dev, 0x00400904 + i * 0x10,
- nv_rd32(dev, NV10_PFB_TLIMIT(i)));
- /* which is NV40_PGRAPH_TLIMIT0(i) ?? */
- nv_wr32(dev, 0x00400908 + i * 0x10,
- nv_rd32(dev, NV10_PFB_TSIZE(i)));
- /* which is NV40_PGRAPH_TSIZE0(i) ?? */
- nv_wr32(dev, 0x00400900 + i * 0x10,
- nv_rd32(dev, NV10_PFB_TILE(i)));
- /* which is NV40_PGRAPH_TILE0(i) ?? */
- }
+ /* Turn all the tiling regions off. */
+ for (i = 0; i < NV10_PFB_TILE__SIZE; i++)
+ nv20_graph_set_region_tiling(dev, i, 0, 0, 0);
nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10000100);
nv_wr32(dev, NV10_PGRAPH_STATE , 0xFFFFFFFF);
diff --git a/drivers/gpu/drm/nouveau/nv40_fb.c b/drivers/gpu/drm/nouveau/nv40_fb.c
index ca1d27107a8..3cd07d8d5bd 100644
--- a/drivers/gpu/drm/nouveau/nv40_fb.c
+++ b/drivers/gpu/drm/nouveau/nv40_fb.c
@@ -3,12 +3,37 @@
#include "nouveau_drv.h"
#include "nouveau_drm.h"
+void
+nv40_fb_set_region_tiling(struct drm_device *dev, int i, uint32_t addr,
+ uint32_t size, uint32_t pitch)
+{
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ uint32_t limit = max(1u, addr + size) - 1;
+
+ if (pitch)
+ addr |= 1;
+
+ switch (dev_priv->chipset) {
+ case 0x40:
+ nv_wr32(dev, NV10_PFB_TLIMIT(i), limit);
+ nv_wr32(dev, NV10_PFB_TSIZE(i), pitch);
+ nv_wr32(dev, NV10_PFB_TILE(i), addr);
+ break;
+
+ default:
+ nv_wr32(dev, NV40_PFB_TLIMIT(i), limit);
+ nv_wr32(dev, NV40_PFB_TSIZE(i), pitch);
+ nv_wr32(dev, NV40_PFB_TILE(i), addr);
+ break;
+ }
+}
+
int
nv40_fb_init(struct drm_device *dev)
{
struct drm_nouveau_private *dev_priv = dev->dev_private;
- uint32_t fb_bar_size, tmp;
- int num_tiles;
+ struct nouveau_fb_engine *pfb = &dev_priv->engine.fb;
+ uint32_t tmp;
int i;
/* This is strictly a NV4x register (don't know about NV5x). */
@@ -23,35 +48,23 @@ nv40_fb_init(struct drm_device *dev)
case 0x45:
tmp = nv_rd32(dev, NV10_PFB_CLOSE_PAGE2);
nv_wr32(dev, NV10_PFB_CLOSE_PAGE2, tmp & ~(1 << 15));
- num_tiles = NV10_PFB_TILE__SIZE;
+ pfb->num_tiles = NV10_PFB_TILE__SIZE;
break;
case 0x46: /* G72 */
case 0x47: /* G70 */
case 0x49: /* G71 */
case 0x4b: /* G73 */
case 0x4c: /* C51 (G7X version) */
- num_tiles = NV40_PFB_TILE__SIZE_1;
+ pfb->num_tiles = NV40_PFB_TILE__SIZE_1;
break;
default:
- num_tiles = NV40_PFB_TILE__SIZE_0;
+ pfb->num_tiles = NV40_PFB_TILE__SIZE_0;
break;
}
- fb_bar_size = drm_get_resource_len(dev, 0) - 1;
- switch (dev_priv->chipset) {
- case 0x40:
- for (i = 0; i < num_tiles; i++) {
- nv_wr32(dev, NV10_PFB_TILE(i), 0);
- nv_wr32(dev, NV10_PFB_TLIMIT(i), fb_bar_size);
- }
- break;
- default:
- for (i = 0; i < num_tiles; i++) {
- nv_wr32(dev, NV40_PFB_TILE(i), 0);
- nv_wr32(dev, NV40_PFB_TLIMIT(i), fb_bar_size);
- }
- break;
- }
+ /* Turn all the tiling regions off. */
+ for (i = 0; i < pfb->num_tiles; i++)
+ pfb->set_region_tiling(dev, i, 0, 0, 0);
return 0;
}
diff --git a/drivers/gpu/drm/nouveau/nv40_graph.c b/drivers/gpu/drm/nouveau/nv40_graph.c
index 2b332bb55ac..53e8afe1dcd 100644
--- a/drivers/gpu/drm/nouveau/nv40_graph.c
+++ b/drivers/gpu/drm/nouveau/nv40_graph.c
@@ -181,6 +181,48 @@ nv40_graph_unload_context(struct drm_device *dev)
return ret;
}
+void
+nv40_graph_set_region_tiling(struct drm_device *dev, int i, uint32_t addr,
+ uint32_t size, uint32_t pitch)
+{
+ struct drm_nouveau_private *dev_priv = dev->dev_private;
+ uint32_t limit = max(1u, addr + size) - 1;
+
+ if (pitch)
+ addr |= 1;
+
+ switch (dev_priv->chipset) {
+ case 0x44:
+ case 0x4a:
+ case 0x4e:
+ nv_wr32(dev, NV20_PGRAPH_TSIZE(i), pitch);
+ nv_wr32(dev, NV20_PGRAPH_TLIMIT(i), limit);
+ nv_wr32(dev, NV20_PGRAPH_TILE(i), addr);
+ break;
+
+ case 0x46:
+ case 0x47:
+ case 0x49:
+ case 0x4b:
+ nv_wr32(dev, NV47_PGRAPH_TSIZE(i), pitch);
+ nv_wr32(dev, NV47_PGRAPH_TLIMIT(i), limit);
+ nv_wr32(dev, NV47_PGRAPH_TILE(i), addr);
+ nv_wr32(dev, NV40_PGRAPH_TSIZE1(i), pitch);
+ nv_wr32(dev, NV40_PGRAPH_TLIMIT1(i), limit);
+ nv_wr32(dev, NV40_PGRAPH_TILE1(i), addr);
+ break;
+
+ default:
+ nv_wr32(dev, NV20_PGRAPH_TSIZE(i), pitch);
+ nv_wr32(dev, NV20_PGRAPH_TLIMIT(i), limit);
+ nv_wr32(dev, NV20_PGRAPH_TILE(i), addr);
+ nv_wr32(dev, NV40_PGRAPH_TSIZE1(i), pitch);
+ nv_wr32(dev, NV40_PGRAPH_TLIMIT1(i), limit);
+ nv_wr32(dev, NV40_PGRAPH_TILE1(i), addr);
+ break;
+ }
+}
+
/*
* G70 0x47
* G71 0x49
@@ -195,7 +237,8 @@ nv40_graph_init(struct drm_device *dev)
{
struct drm_nouveau_private *dev_priv =
(struct drm_nouveau_private *)dev->dev_private;
- uint32_t vramsz, tmp;
+ struct nouveau_fb_engine *pfb = &dev_priv->engine.fb;
+ uint32_t vramsz;
int i, j;
nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) &
@@ -292,74 +335,9 @@ nv40_graph_init(struct drm_device *dev)
nv_wr32(dev, 0x400b38, 0x2ffff800);
nv_wr32(dev, 0x400b3c, 0x00006000);
- /* copy tile info from PFB */
- switch (dev_priv->chipset) {
- case 0x40: /* vanilla NV40 */
- for (i = 0; i < NV10_PFB_TILE__SIZE; i++) {
- tmp = nv_rd32(dev, NV10_PFB_TILE(i));
- nv_wr32(dev, NV40_PGRAPH_TILE0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TILE1(i), tmp);
- tmp = nv_rd32(dev, NV10_PFB_TLIMIT(i));
- nv_wr32(dev, NV40_PGRAPH_TLIMIT0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TLIMIT1(i), tmp);
- tmp = nv_rd32(dev, NV10_PFB_TSIZE(i));
- nv_wr32(dev, NV40_PGRAPH_TSIZE0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TSIZE1(i), tmp);
- tmp = nv_rd32(dev, NV10_PFB_TSTATUS(i));
- nv_wr32(dev, NV40_PGRAPH_TSTATUS0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TSTATUS1(i), tmp);
- }
- break;
- case 0x44:
- case 0x4a:
- case 0x4e: /* NV44-based cores don't have 0x406900? */
- for (i = 0; i < NV40_PFB_TILE__SIZE_0; i++) {
- tmp = nv_rd32(dev, NV40_PFB_TILE(i));
- nv_wr32(dev, NV40_PGRAPH_TILE0(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TLIMIT(i));
- nv_wr32(dev, NV40_PGRAPH_TLIMIT0(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TSIZE(i));
- nv_wr32(dev, NV40_PGRAPH_TSIZE0(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TSTATUS(i));
- nv_wr32(dev, NV40_PGRAPH_TSTATUS0(i), tmp);
- }
- break;
- case 0x46:
- case 0x47:
- case 0x49:
- case 0x4b: /* G7X-based cores */
- for (i = 0; i < NV40_PFB_TILE__SIZE_1; i++) {
- tmp = nv_rd32(dev, NV40_PFB_TILE(i));
- nv_wr32(dev, NV47_PGRAPH_TILE0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TILE1(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TLIMIT(i));
- nv_wr32(dev, NV47_PGRAPH_TLIMIT0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TLIMIT1(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TSIZE(i));
- nv_wr32(dev, NV47_PGRAPH_TSIZE0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TSIZE1(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TSTATUS(i));
- nv_wr32(dev, NV47_PGRAPH_TSTATUS0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TSTATUS1(i), tmp);
- }
- break;
- default: /* everything else */
- for (i = 0; i < NV40_PFB_TILE__SIZE_0; i++) {
- tmp = nv_rd32(dev, NV40_PFB_TILE(i));
- nv_wr32(dev, NV40_PGRAPH_TILE0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TILE1(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TLIMIT(i));
- nv_wr32(dev, NV40_PGRAPH_TLIMIT0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TLIMIT1(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TSIZE(i));
- nv_wr32(dev, NV40_PGRAPH_TSIZE0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TSIZE1(i), tmp);
- tmp = nv_rd32(dev, NV40_PFB_TSTATUS(i));
- nv_wr32(dev, NV40_PGRAPH_TSTATUS0(i), tmp);
- nv_wr32(dev, NV40_PGRAPH_TSTATUS1(i), tmp);
- }
- break;
- }
+ /* Turn all the tiling regions off. */
+ for (i = 0; i < pfb->num_tiles; i++)
+ nv40_graph_set_region_tiling(dev, i, 0, 0, 0);
/* begin RAM config */
vramsz = drm_get_resource_len(dev, 0) - 1;
diff --git a/drivers/gpu/drm/nouveau/nv50_crtc.c b/drivers/gpu/drm/nouveau/nv50_crtc.c
index 118d3285fd8..d1a651e3400 100644
--- a/drivers/gpu/drm/nouveau/nv50_crtc.c
+++ b/drivers/gpu/drm/nouveau/nv50_crtc.c
@@ -298,14 +298,17 @@ nv50_crtc_set_clock(struct drm_device *dev, int head, int pclk)
static void
nv50_crtc_destroy(struct drm_crtc *crtc)
{
- struct drm_device *dev = crtc->dev;
- struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc);
-
- NV_DEBUG_KMS(dev, "\n");
+ struct drm_device *dev;
+ struct nouveau_crtc *nv_crtc;
if (!crtc)
return;
+ dev = crtc->dev;
+ nv_crtc = nouveau_crtc(crtc);
+
+ NV_DEBUG_KMS(dev, "\n");
+
drm_crtc_cleanup(&nv_crtc->base);
nv50_cursor_fini(nv_crtc);
@@ -432,6 +435,7 @@ nv50_crtc_prepare(struct drm_crtc *crtc)
struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc);
struct drm_device *dev = crtc->dev;
struct drm_encoder *encoder;
+ uint32_t dac = 0, sor = 0;
NV_DEBUG_KMS(dev, "index %d\n", nv_crtc->index);
@@ -439,9 +443,28 @@ nv50_crtc_prepare(struct drm_crtc *crtc)
list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
- if (drm_helper_encoder_in_use(encoder))
+ if (!drm_helper_encoder_in_use(encoder))
continue;
+ if (nv_encoder->dcb->type == OUTPUT_ANALOG ||
+ nv_encoder->dcb->type == OUTPUT_TV)
+ dac |= (1 << nv_encoder->or);
+ else
+ sor |= (1 << nv_encoder->or);
+ }
+
+ list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+ struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
+
+ if (nv_encoder->dcb->type == OUTPUT_ANALOG ||
+ nv_encoder->dcb->type == OUTPUT_TV) {
+ if (dac & (1 << nv_encoder->or))
+ continue;
+ } else {
+ if (sor & (1 << nv_encoder->or))
+ continue;
+ }
+
nv_encoder->disconnect(nv_encoder);
}
diff --git a/drivers/gpu/drm/nouveau/nv50_display.c b/drivers/gpu/drm/nouveau/nv50_display.c
index a9263d92a23..90f0bf59fbc 100644
--- a/drivers/gpu/drm/nouveau/nv50_display.c
+++ b/drivers/gpu/drm/nouveau/nv50_display.c
@@ -690,9 +690,21 @@ nv50_display_script_select(struct drm_device *dev, struct dcb_entry *dcbent,
int pxclk)
{
struct drm_nouveau_private *dev_priv = dev->dev_private;
+ struct nouveau_connector *nv_connector = NULL;
+ struct drm_encoder *encoder;
struct nvbios *bios = &dev_priv->VBIOS;
uint32_t mc, script = 0, or;
+ list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) {
+ struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
+
+ if (nv_encoder->dcb != dcbent)
+ continue;
+
+ nv_connector = nouveau_encoder_connector_get(nv_encoder);
+ break;
+ }
+
or = ffs(dcbent->or) - 1;
mc = nv50_display_mode_ctrl(dev, dcbent->type != OUTPUT_ANALOG, or);
switch (dcbent->type) {
@@ -711,6 +723,11 @@ nv50_display_script_select(struct drm_device *dev, struct dcb_entry *dcbent,
} else
if (bios->fp.strapless_is_24bit & 1)
script |= 0x0200;
+
+ if (nv_connector && nv_connector->edid &&
+ (nv_connector->edid->revision >= 4) &&
+ (nv_connector->edid->input & 0x70) >= 0x20)
+ script |= 0x0200;
}
if (nouveau_uscript_lvds >= 0) {
diff --git a/drivers/gpu/drm/nouveau/nv50_fbcon.c b/drivers/gpu/drm/nouveau/nv50_fbcon.c
index 6bcc6d39e9b..0f57cdf7ccb 100644
--- a/drivers/gpu/drm/nouveau/nv50_fbcon.c
+++ b/drivers/gpu/drm/nouveau/nv50_fbcon.c
@@ -3,7 +3,7 @@
#include "nouveau_dma.h"
#include "nouveau_fbcon.h"
-static void
+void
nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
{
struct nouveau_fbcon_par *par = info->par;
@@ -16,9 +16,7 @@ nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
if (!(info->flags & FBINFO_HWACCEL_DISABLED) &&
RING_SPACE(chan, rect->rop == ROP_COPY ? 7 : 11)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
-
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
}
if (info->flags & FBINFO_HWACCEL_DISABLED) {
@@ -31,7 +29,11 @@ nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
OUT_RING(chan, 1);
}
BEGIN_RING(chan, NvSub2D, 0x0588, 1);
- OUT_RING(chan, rect->color);
+ if (info->fix.visual == FB_VISUAL_TRUECOLOR ||
+ info->fix.visual == FB_VISUAL_DIRECTCOLOR)
+ OUT_RING(chan, ((uint32_t *)info->pseudo_palette)[rect->color]);
+ else
+ OUT_RING(chan, rect->color);
BEGIN_RING(chan, NvSub2D, 0x0600, 4);
OUT_RING(chan, rect->dx);
OUT_RING(chan, rect->dy);
@@ -44,7 +46,7 @@ nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
FIRE_RING(chan);
}
-static void
+void
nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
{
struct nouveau_fbcon_par *par = info->par;
@@ -56,9 +58,7 @@ nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
return;
if (!(info->flags & FBINFO_HWACCEL_DISABLED) && RING_SPACE(chan, 12)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
-
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
}
if (info->flags & FBINFO_HWACCEL_DISABLED) {
@@ -81,7 +81,7 @@ nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region)
FIRE_RING(chan);
}
-static void
+void
nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
{
struct nouveau_fbcon_par *par = info->par;
@@ -101,8 +101,7 @@ nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
}
if (!(info->flags & FBINFO_HWACCEL_DISABLED) && RING_SPACE(chan, 11)) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
}
if (info->flags & FBINFO_HWACCEL_DISABLED) {
@@ -135,9 +134,7 @@ nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image)
int push = dwords > 2047 ? 2047 : dwords;
if (RING_SPACE(chan, push + 1)) {
- NV_ERROR(dev,
- "GPU lockup - switching to software fbcon\n");
- info->flags |= FBINFO_HWACCEL_DISABLED;
+ nouveau_fbcon_gpu_lockup(info);
cfb_imageblit(info, image);
return;
}
@@ -199,7 +196,7 @@ nv50_fbcon_accel_init(struct fb_info *info)
ret = RING_SPACE(chan, 59);
if (ret) {
- NV_ERROR(dev, "GPU lockup - switching to software fbcon\n");
+ nouveau_fbcon_gpu_lockup(info);
return ret;
}
@@ -265,9 +262,6 @@ nv50_fbcon_accel_init(struct fb_info *info)
OUT_RING(chan, info->fix.smem_start - dev_priv->fb_phys +
dev_priv->vm_vram_base);
- info->fbops->fb_fillrect = nv50_fbcon_fillrect;
- info->fbops->fb_copyarea = nv50_fbcon_copyarea;
- info->fbops->fb_imageblit = nv50_fbcon_imageblit;
return 0;
}
diff --git a/drivers/gpu/drm/nouveau/nv50_fifo.c b/drivers/gpu/drm/nouveau/nv50_fifo.c
index b7282284f08..204a79ff10f 100644
--- a/drivers/gpu/drm/nouveau/nv50_fifo.c
+++ b/drivers/gpu/drm/nouveau/nv50_fifo.c
@@ -272,7 +272,7 @@ nv50_fifo_create_context(struct nouveau_channel *chan)
return ret;
ramfc = chan->ramfc->gpuobj;
- ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, 4096, 256,
+ ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, 4096, 1024,
0, &chan->cache);
if (ret)
return ret;
@@ -317,17 +317,20 @@ void
nv50_fifo_destroy_context(struct nouveau_channel *chan)
{
struct drm_device *dev = chan->dev;
+ struct nouveau_gpuobj_ref *ramfc = chan->ramfc;
NV_DEBUG(dev, "ch%d\n", chan->id);
- nouveau_gpuobj_ref_del(dev, &chan->ramfc);
- nouveau_gpuobj_ref_del(dev, &chan->cache);
-
+ /* This will ensure the channel is seen as disabled. */
+ chan->ramfc = NULL;
nv50_fifo_channel_disable(dev, chan->id, false);
/* Dummy channel, also used on ch 127 */
if (chan->id == 0)
nv50_fifo_channel_disable(dev, 127, false);
+
+ nouveau_gpuobj_ref_del(dev, &ramfc);
+ nouveau_gpuobj_ref_del(dev, &chan->cache);
}
int
@@ -384,8 +387,8 @@ nv50_fifo_load_context(struct nouveau_channel *chan)
nv_wr32(dev, NV40_PFIFO_CACHE1_DATA(ptr),
nv_ro32(dev, cache, (ptr * 2) + 1));
}
- nv_wr32(dev, 0x3210, cnt << 2);
- nv_wr32(dev, 0x3270, 0);
+ nv_wr32(dev, NV03_PFIFO_CACHE1_PUT, cnt << 2);
+ nv_wr32(dev, NV03_PFIFO_CACHE1_GET, 0);
/* guessing that all the 0x34xx regs aren't on NV50 */
if (!IS_G80) {
@@ -398,8 +401,6 @@ nv50_fifo_load_context(struct nouveau_channel *chan)
dev_priv->engine.instmem.finish_access(dev);
- nv_wr32(dev, NV03_PFIFO_CACHE1_GET, 0);
- nv_wr32(dev, NV03_PFIFO_CACHE1_PUT, 0);
nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, chan->id | (1<<16));
return 0;
}
diff --git a/drivers/gpu/drm/nouveau/nv50_graph.c b/drivers/gpu/drm/nouveau/nv50_graph.c
index ca79f32be44..6d504801b51 100644
--- a/drivers/gpu/drm/nouveau/nv50_graph.c
+++ b/drivers/gpu/drm/nouveau/nv50_graph.c
@@ -84,7 +84,7 @@ nv50_graph_init_regs__nv(struct drm_device *dev)
nv_wr32(dev, 0x400804, 0xc0000000);
nv_wr32(dev, 0x406800, 0xc0000000);
nv_wr32(dev, 0x400c04, 0xc0000000);
- nv_wr32(dev, 0x401804, 0xc0000000);
+ nv_wr32(dev, 0x401800, 0xc0000000);
nv_wr32(dev, 0x405018, 0xc0000000);
nv_wr32(dev, 0x402000, 0xc0000000);
@@ -165,6 +165,12 @@ nv50_graph_channel(struct drm_device *dev)
uint32_t inst;
int i;
+ /* Be sure we're not in the middle of a context switch or bad things
+ * will happen, such as unloading the wrong pgraph context.
+ */
+ if (!nv_wait(0x400300, 0x00000001, 0x00000000))
+ NV_ERROR(dev, "Ctxprog is still running\n");
+
inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR);
if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED))
return NULL;
@@ -275,19 +281,18 @@ nv50_graph_load_context(struct nouveau_channel *chan)
int
nv50_graph_unload_context(struct drm_device *dev)
{
- uint32_t inst, fifo = nv_rd32(dev, 0x400500);
+ uint32_t inst;
inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR);
if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED))
return 0;
inst &= NV50_PGRAPH_CTXCTL_CUR_INSTANCE;
- nv_wr32(dev, 0x400500, fifo & ~1);
+ nouveau_wait_for_idle(dev);
nv_wr32(dev, 0x400784, inst);
nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x20);
nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 0x01);
nouveau_wait_for_idle(dev);
- nv_wr32(dev, 0x400500, fifo);
nv_wr32(dev, NV50_PGRAPH_CTXCTL_CUR, inst);
return 0;
diff --git a/drivers/gpu/drm/nouveau/nv50_instmem.c b/drivers/gpu/drm/nouveau/nv50_instmem.c
index 94400f777e7..f0dc4e36ef0 100644
--- a/drivers/gpu/drm/nouveau/nv50_instmem.c
+++ b/drivers/gpu/drm/nouveau/nv50_instmem.c
@@ -76,6 +76,11 @@ nv50_instmem_init(struct drm_device *dev)
for (i = 0x1700; i <= 0x1710; i += 4)
priv->save1700[(i-0x1700)/4] = nv_rd32(dev, i);
+ if (dev_priv->chipset == 0xaa || dev_priv->chipset == 0xac)
+ dev_priv->vram_sys_base = nv_rd32(dev, 0x100e10) << 12;
+ else
+ dev_priv->vram_sys_base = 0;
+
/* Reserve the last MiB of VRAM, we should probably try to avoid
* setting up the below tables over the top of the VBIOS image at
* some point.
@@ -172,16 +177,28 @@ nv50_instmem_init(struct drm_device *dev)
* We map the entire fake channel into the start of the PRAMIN BAR
*/
ret = nouveau_gpuobj_new_ref(dev, chan, NULL, 0, pt_size, 0x1000,
- 0, &priv->pramin_pt);
+ 0, &priv->pramin_pt);
if (ret)
return ret;
- for (i = 0, v = c_offset; i < pt_size; i += 8, v += 0x1000) {
- if (v < (c_offset + c_size))
- BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, v | 1);
- else
- BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, 0x00000009);
+ v = c_offset | 1;
+ if (dev_priv->vram_sys_base) {
+ v += dev_priv->vram_sys_base;
+ v |= 0x30;
+ }
+
+ i = 0;
+ while (v < dev_priv->vram_sys_base + c_offset + c_size) {
+ BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, v);
+ BAR0_WI32(priv->pramin_pt->gpuobj, i + 4, 0x00000000);
+ v += 0x1000;
+ i += 8;
+ }
+
+ while (i < pt_size) {
+ BAR0_WI32(priv->pramin_pt->gpuobj, i + 0, 0x00000000);
BAR0_WI32(priv->pramin_pt->gpuobj, i + 4, 0x00000000);
+ i += 8;
}
BAR0_WI32(chan->vm_pd, 0x00, priv->pramin_pt->instance | 0x63);
@@ -416,7 +433,9 @@ nv50_instmem_bind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj)
{
struct drm_nouveau_private *dev_priv = dev->dev_private;
struct nv50_instmem_priv *priv = dev_priv->engine.instmem.priv;
- uint32_t pte, pte_end, vram;
+ struct nouveau_gpuobj *pramin_pt = priv->pramin_pt->gpuobj;
+ uint32_t pte, pte_end;
+ uint64_t vram;
if (!gpuobj->im_backing || !gpuobj->im_pramin || gpuobj->im_bound)
return -EINVAL;
@@ -424,20 +443,24 @@ nv50_instmem_bind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj)
NV_DEBUG(dev, "st=0x%0llx sz=0x%0llx\n",
gpuobj->im_pramin->start, gpuobj->im_pramin->size);
- pte = (gpuobj->im_pramin->start >> 12) << 3;
- pte_end = ((gpuobj->im_pramin->size >> 12) << 3) + pte;
+ pte = (gpuobj->im_pramin->start >> 12) << 1;
+ pte_end = ((gpuobj->im_pramin->size >> 12) << 1) + pte;
vram = gpuobj->im_backing_start;
NV_DEBUG(dev, "pramin=0x%llx, pte=%d, pte_end=%d\n",
gpuobj->im_pramin->start, pte, pte_end);
NV_DEBUG(dev, "first vram page: 0x%08x\n", gpuobj->im_backing_start);
+ vram |= 1;
+ if (dev_priv->vram_sys_base) {
+ vram += dev_priv->vram_sys_base;
+ vram |= 0x30;
+ }
+
dev_priv->engine.instmem.prepare_access(dev, true);
while (pte < pte_end) {
- nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 0)/4, vram | 1);
- nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 4)/4, 0x00000000);
-
- pte += 8;
+ nv_wo32(dev, pramin_pt, pte++, lower_32_bits(vram));
+ nv_wo32(dev, pramin_pt, pte++, upper_32_bits(vram));
vram += NV50_INSTMEM_PAGE_SIZE;
}
dev_priv->engine.instmem.finish_access(dev);
@@ -470,14 +493,13 @@ nv50_instmem_unbind(struct drm_device *dev, struct nouveau_gpuobj *gpuobj)
if (gpuobj->im_bound == 0)
return -EINVAL;
- pte = (gpuobj->im_pramin->start >> 12) << 3;
- pte_end = ((gpuobj->im_pramin->size >> 12) << 3) + pte;
+ pte = (gpuobj->im_pramin->start >> 12) << 1;
+ pte_end = ((gpuobj->im_pramin->size >> 12) << 1) + pte;
dev_priv->engine.instmem.prepare_access(dev, true);
while (pte < pte_end) {
- nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 0)/4, 0x00000009);
- nv_wo32(dev, priv->pramin_pt->gpuobj, (pte + 4)/4, 0x00000000);
- pte += 8;
+ nv_wo32(dev, priv->pramin_pt->gpuobj, pte++, 0x00000000);
+ nv_wo32(dev, priv->pramin_pt->gpuobj, pte++, 0x00000000);
}
dev_priv->engine.instmem.finish_access(dev);
diff --git a/drivers/gpu/drm/nouveau/nv50_sor.c b/drivers/gpu/drm/nouveau/nv50_sor.c
index e395c16d30f..c2fff543b06 100644
--- a/drivers/gpu/drm/nouveau/nv50_sor.c
+++ b/drivers/gpu/drm/nouveau/nv50_sor.c
@@ -90,11 +90,25 @@ nv50_sor_dpms(struct drm_encoder *encoder, int mode)
{
struct drm_device *dev = encoder->dev;
struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder);
+ struct drm_encoder *enc;
uint32_t val;
int or = nv_encoder->or;
NV_DEBUG_KMS(dev, "or %d mode %d\n", or, mode);
+ nv_encoder->last_dpms = mode;
+ list_for_each_entry(enc, &dev->mode_config.encoder_list, head) {
+ struct nouveau_encoder *nvenc = nouveau_encoder(enc);
+
+ if (nvenc == nv_encoder ||
+ nvenc->disconnect != nv50_sor_disconnect ||
+ nvenc->dcb->or != nv_encoder->dcb->or)
+ continue;
+
+ if (nvenc->last_dpms == DRM_MODE_DPMS_ON)
+ return;
+ }
+
/* wait for it to be done */
if (!nv_wait(NV50_PDISPLAY_SOR_DPMS_CTRL(or),
NV50_PDISPLAY_SOR_DPMS_CTRL_PENDING, 0)) {