diff options
author | Jerome Glisse <jglisse@redhat.com> | 2012-01-05 22:11:05 -0500 |
---|---|---|
committer | Dave Airlie <airlied@redhat.com> | 2012-01-06 09:15:42 +0000 |
commit | 721604a15b934f0a8d1909acb8017f029128be2f (patch) | |
tree | ac1dc0f837d70616b36c9b57d22eb9678c5e68fc /drivers/gpu/drm/radeon/evergreen_cs.c | |
parent | 09b4ea47d1041612b101c369969db123ac2c1511 (diff) | |
download | op-kernel-dev-721604a15b934f0a8d1909acb8017f029128be2f.zip op-kernel-dev-721604a15b934f0a8d1909acb8017f029128be2f.tar.gz |
drm/radeon: GPU virtual memory support v22
Virtual address space are per drm client (opener of /dev/drm).
Client are in charge of virtual address space, they need to
map bo into it by calling DRM_RADEON_GEM_VA ioctl.
First 16M of virtual address space is reserved by the kernel.
Once using 2 level page table we should be able to have a small
vram memory footprint for each pt (there would be one pt for all
gart, one for all vram and then one first level for each virtual
address space).
Plan include using the sub allocator for a common vm page table
area and using memcpy to copy vm page table in & out. Or use
a gart object and copy things in & out using dma.
v2: agd5f fixes:
- Add vram base offset for vram pages. The GPU physical address of a
vram page is FB_OFFSET + page offset. FB_OFFSET is 0 on discrete
cards and the physical bus address of the stolen memory on
integrated chips.
- VM_CONTEXT1_PROTECTION_FAULT_DEFAULT_ADDR covers all vmid's >= 1
v3: agd5f:
- integrate with the semaphore/multi-ring stuff
v4:
- rebase on top ttm dma & multi-ring stuff
- userspace is now in charge of the address space
- no more specific cs vm ioctl, instead cs ioctl has a new
chunk
v5:
- properly handle mem == NULL case from move_notify callback
- fix the vm cleanup path
v6:
- fix update of page table to only happen on valid mem placement
v7:
- add tlb flush for each vm context
- add flags to define mapping property (readable, writeable, snooped)
- make ring id implicit from ib->fence->ring, up to each asic callback
to then do ring specific scheduling if vm ib scheduling function
v8:
- add query for ib limit and kernel reserved virtual space
- rename vm->size to max_pfn (maximum number of page)
- update gem_va ioctl to also allow unmap operation
- bump kernel version to allow userspace to query for vm support
v9:
- rebuild page table only when bind and incrementaly depending
on bo referenced by cs and that have been moved
- allow virtual address space to grow
- use sa allocator for vram page table
- return invalid when querying vm limit on non cayman GPU
- dump vm fault register on lockup
v10: agd5f:
- Move the vm schedule_ib callback to a standalone function, remove
the callback and use the existing ib_execute callback for VM IBs.
v11:
- rebase on top of lastest Linus
v12: agd5f:
- remove spurious backslash
- set IB vm_id to 0 in radeon_ib_get()
v13: agd5f:
- fix handling of RADEON_CHUNK_ID_FLAGS
v14:
- fix va destruction
- fix suspend resume
- forbid bo to have several different va in same vm
v15:
- rebase
v16:
- cleanup left over of vm init/fini
v17: agd5f:
- cs checker
v18: agd5f:
- reworks the CS ioctl to better support multiple rings and
VM. Rather than adding a new chunk id for VM, just re-use the
IB chunk id and add a new flags for VM mode. Also define additional
dwords for the flags chunk id to define the what ring we want to use
(gfx, compute, uvd, etc.) and the priority.
v19:
- fix cs fini in weird case of no ib
- semi working flush fix for ni
- rebase on top of sa allocator changes
v20: agd5f:
- further CS ioctl cleanups from Christian's comments
v21: agd5f:
- integrate CS checker improvements
v22: agd5f:
- final cleanups for release, only allow VM CS on cayman
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Diffstat (limited to 'drivers/gpu/drm/radeon/evergreen_cs.c')
-rw-r--r-- | drivers/gpu/drm/radeon/evergreen_cs.c | 246 |
1 files changed, 242 insertions, 4 deletions
diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c index cd4590a..f7442e6 100644 --- a/drivers/gpu/drm/radeon/evergreen_cs.c +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -520,7 +520,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) break; case DB_Z_INFO: track->db_z_info = radeon_get_ib_value(p, idx); - if (!p->keep_tiling_flags) { + if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { r = evergreen_cs_packet_next_reloc(p, &reloc); if (r) { dev_warn(p->dev, "bad SET_CONTEXT_REG " @@ -649,7 +649,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) case CB_COLOR7_INFO: tmp = (reg - CB_COLOR0_INFO) / 0x3c; track->cb_color_info[tmp] = radeon_get_ib_value(p, idx); - if (!p->keep_tiling_flags) { + if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { r = evergreen_cs_packet_next_reloc(p, &reloc); if (r) { dev_warn(p->dev, "bad SET_CONTEXT_REG " @@ -666,7 +666,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) case CB_COLOR11_INFO: tmp = ((reg - CB_COLOR8_INFO) / 0x1c) + 8; track->cb_color_info[tmp] = radeon_get_ib_value(p, idx); - if (!p->keep_tiling_flags) { + if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { r = evergreen_cs_packet_next_reloc(p, &reloc); if (r) { dev_warn(p->dev, "bad SET_CONTEXT_REG " @@ -1355,7 +1355,7 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } ib[idx+1+(i*8)+2] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); - if (!p->keep_tiling_flags) { + if (!(p->cs_flags & RADEON_CS_KEEP_TILING_FLAGS)) { ib[idx+1+(i*8)+1] |= TEX_ARRAY_MODE(evergreen_cs_get_aray_mode(reloc->lobj.tiling_flags)); if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) { @@ -1572,3 +1572,241 @@ int evergreen_cs_parse(struct radeon_cs_parser *p) return 0; } +/* vm parser */ +static bool evergreen_vm_reg_valid(u32 reg) +{ + /* context regs are fine */ + if (reg >= 0x28000) + return true; + + /* check config regs */ + switch (reg) { + case GRBM_GFX_INDEX: + case VGT_VTX_VECT_EJECT_REG: + case VGT_CACHE_INVALIDATION: + case VGT_GS_VERTEX_REUSE: + case VGT_PRIMITIVE_TYPE: + case VGT_INDEX_TYPE: + case VGT_NUM_INDICES: + case VGT_NUM_INSTANCES: + case VGT_COMPUTE_DIM_X: + case VGT_COMPUTE_DIM_Y: + case VGT_COMPUTE_DIM_Z: + case VGT_COMPUTE_START_X: + case VGT_COMPUTE_START_Y: + case VGT_COMPUTE_START_Z: + case VGT_COMPUTE_INDEX: + case VGT_COMPUTE_THREAD_GROUP_SIZE: + case VGT_HS_OFFCHIP_PARAM: + case PA_CL_ENHANCE: + case PA_SU_LINE_STIPPLE_VALUE: + case PA_SC_LINE_STIPPLE_STATE: + case PA_SC_ENHANCE: + case SQ_DYN_GPR_CNTL_PS_FLUSH_REQ: + case SQ_DYN_GPR_SIMD_LOCK_EN: + case SQ_CONFIG: + case SQ_GPR_RESOURCE_MGMT_1: + case SQ_GLOBAL_GPR_RESOURCE_MGMT_1: + case SQ_GLOBAL_GPR_RESOURCE_MGMT_2: + case SQ_CONST_MEM_BASE: + case SQ_STATIC_THREAD_MGMT_1: + case SQ_STATIC_THREAD_MGMT_2: + case SQ_STATIC_THREAD_MGMT_3: + case SPI_CONFIG_CNTL: + case SPI_CONFIG_CNTL_1: + case TA_CNTL_AUX: + case DB_DEBUG: + case DB_DEBUG2: + case DB_DEBUG3: + case DB_DEBUG4: + case DB_WATERMARKS: + case TD_PS_BORDER_COLOR_INDEX: + case TD_PS_BORDER_COLOR_RED: + case TD_PS_BORDER_COLOR_GREEN: + case TD_PS_BORDER_COLOR_BLUE: + case TD_PS_BORDER_COLOR_ALPHA: + case TD_VS_BORDER_COLOR_INDEX: + case TD_VS_BORDER_COLOR_RED: + case TD_VS_BORDER_COLOR_GREEN: + case TD_VS_BORDER_COLOR_BLUE: + case TD_VS_BORDER_COLOR_ALPHA: + case TD_GS_BORDER_COLOR_INDEX: + case TD_GS_BORDER_COLOR_RED: + case TD_GS_BORDER_COLOR_GREEN: + case TD_GS_BORDER_COLOR_BLUE: + case TD_GS_BORDER_COLOR_ALPHA: + case TD_HS_BORDER_COLOR_INDEX: + case TD_HS_BORDER_COLOR_RED: + case TD_HS_BORDER_COLOR_GREEN: + case TD_HS_BORDER_COLOR_BLUE: + case TD_HS_BORDER_COLOR_ALPHA: + case TD_LS_BORDER_COLOR_INDEX: + case TD_LS_BORDER_COLOR_RED: + case TD_LS_BORDER_COLOR_GREEN: + case TD_LS_BORDER_COLOR_BLUE: + case TD_LS_BORDER_COLOR_ALPHA: + case TD_CS_BORDER_COLOR_INDEX: + case TD_CS_BORDER_COLOR_RED: + case TD_CS_BORDER_COLOR_GREEN: + case TD_CS_BORDER_COLOR_BLUE: + case TD_CS_BORDER_COLOR_ALPHA: + case SQ_ESGS_RING_SIZE: + case SQ_GSVS_RING_SIZE: + case SQ_ESTMP_RING_SIZE: + case SQ_GSTMP_RING_SIZE: + case SQ_HSTMP_RING_SIZE: + case SQ_LSTMP_RING_SIZE: + case SQ_PSTMP_RING_SIZE: + case SQ_VSTMP_RING_SIZE: + case SQ_ESGS_RING_ITEMSIZE: + case SQ_ESTMP_RING_ITEMSIZE: + case SQ_GSTMP_RING_ITEMSIZE: + case SQ_GSVS_RING_ITEMSIZE: + case SQ_GS_VERT_ITEMSIZE: + case SQ_GS_VERT_ITEMSIZE_1: + case SQ_GS_VERT_ITEMSIZE_2: + case SQ_GS_VERT_ITEMSIZE_3: + case SQ_GSVS_RING_OFFSET_1: + case SQ_GSVS_RING_OFFSET_2: + case SQ_GSVS_RING_OFFSET_3: + case SQ_HSTMP_RING_ITEMSIZE: + case SQ_LSTMP_RING_ITEMSIZE: + case SQ_PSTMP_RING_ITEMSIZE: + case SQ_VSTMP_RING_ITEMSIZE: + case VGT_TF_RING_SIZE: + case SQ_ESGS_RING_BASE: + case SQ_GSVS_RING_BASE: + case SQ_ESTMP_RING_BASE: + case SQ_GSTMP_RING_BASE: + case SQ_HSTMP_RING_BASE: + case SQ_LSTMP_RING_BASE: + case SQ_PSTMP_RING_BASE: + case SQ_VSTMP_RING_BASE: + case CAYMAN_VGT_OFFCHIP_LDS_BASE: + case CAYMAN_SQ_EX_ALLOC_TABLE_SLOTS: + return true; + default: + return false; + } +} + +static int evergreen_vm_packet3_check(struct radeon_device *rdev, + u32 *ib, struct radeon_cs_packet *pkt) +{ + u32 idx = pkt->idx + 1; + u32 idx_value = ib[idx]; + u32 start_reg, end_reg, reg, i; + + switch (pkt->opcode) { + case PACKET3_NOP: + case PACKET3_SET_BASE: + case PACKET3_CLEAR_STATE: + case PACKET3_INDEX_BUFFER_SIZE: + case PACKET3_DISPATCH_DIRECT: + case PACKET3_DISPATCH_INDIRECT: + case PACKET3_MODE_CONTROL: + case PACKET3_SET_PREDICATION: + case PACKET3_COND_EXEC: + case PACKET3_PRED_EXEC: + case PACKET3_DRAW_INDIRECT: + case PACKET3_DRAW_INDEX_INDIRECT: + case PACKET3_INDEX_BASE: + case PACKET3_DRAW_INDEX_2: + case PACKET3_CONTEXT_CONTROL: + case PACKET3_DRAW_INDEX_OFFSET: + case PACKET3_INDEX_TYPE: + case PACKET3_DRAW_INDEX: + case PACKET3_DRAW_INDEX_AUTO: + case PACKET3_DRAW_INDEX_IMMD: + case PACKET3_NUM_INSTANCES: + case PACKET3_DRAW_INDEX_MULTI_AUTO: + case PACKET3_STRMOUT_BUFFER_UPDATE: + case PACKET3_DRAW_INDEX_OFFSET_2: + case PACKET3_DRAW_INDEX_MULTI_ELEMENT: + case PACKET3_MPEG_INDEX: + case PACKET3_WAIT_REG_MEM: + case PACKET3_MEM_WRITE: + case PACKET3_SURFACE_SYNC: + case PACKET3_EVENT_WRITE: + case PACKET3_EVENT_WRITE_EOP: + case PACKET3_EVENT_WRITE_EOS: + case PACKET3_SET_CONTEXT_REG: + case PACKET3_SET_BOOL_CONST: + case PACKET3_SET_LOOP_CONST: + case PACKET3_SET_RESOURCE: + case PACKET3_SET_SAMPLER: + case PACKET3_SET_CTL_CONST: + case PACKET3_SET_RESOURCE_OFFSET: + case PACKET3_SET_CONTEXT_REG_INDIRECT: + case PACKET3_SET_RESOURCE_INDIRECT: + case CAYMAN_PACKET3_DEALLOC_STATE: + break; + case PACKET3_COND_WRITE: + if (idx_value & 0x100) { + reg = ib[idx + 5] * 4; + if (!evergreen_vm_reg_valid(reg)) + return -EINVAL; + } + break; + case PACKET3_COPY_DW: + if (idx_value & 0x2) { + reg = ib[idx + 3] * 4; + if (!evergreen_vm_reg_valid(reg)) + return -EINVAL; + } + break; + case PACKET3_SET_CONFIG_REG: + start_reg = (idx_value << 2) + PACKET3_SET_CONFIG_REG_START; + end_reg = 4 * pkt->count + start_reg - 4; + if ((start_reg < PACKET3_SET_CONFIG_REG_START) || + (start_reg >= PACKET3_SET_CONFIG_REG_END) || + (end_reg >= PACKET3_SET_CONFIG_REG_END)) { + DRM_ERROR("bad PACKET3_SET_CONFIG_REG\n"); + return -EINVAL; + } + for (i = 0; i < pkt->count; i++) { + reg = start_reg + (4 * i); + if (!evergreen_vm_reg_valid(reg)) + return -EINVAL; + } + break; + default: + return -EINVAL; + } + return 0; +} + +int evergreen_ib_parse(struct radeon_device *rdev, struct radeon_ib *ib) +{ + int ret = 0; + u32 idx = 0; + struct radeon_cs_packet pkt; + + do { + pkt.idx = idx; + pkt.type = CP_PACKET_GET_TYPE(ib->ptr[idx]); + pkt.count = CP_PACKET_GET_COUNT(ib->ptr[idx]); + pkt.one_reg_wr = 0; + switch (pkt.type) { + case PACKET_TYPE0: + dev_err(rdev->dev, "Packet0 not allowed!\n"); + ret = -EINVAL; + break; + case PACKET_TYPE2: + break; + case PACKET_TYPE3: + pkt.opcode = CP_PACKET3_GET_OPCODE(ib->ptr[idx]); + ret = evergreen_vm_packet3_check(rdev, ib->ptr, &pkt); + break; + default: + dev_err(rdev->dev, "Unknown packet type %d !\n", pkt.type); + ret = -EINVAL; + break; + } + if (ret) + break; + idx += pkt.count + 2; + } while (idx < ib->length_dw); + + return ret; +} |