From 4b8dc71de5696679f13f8863bc5775a23e786b34 Mon Sep 17 00:00:00 2001 From: Peter Daifuku Date: Fri, 6 Oct 2017 15:38:11 -0700 Subject: gpu: nvgpu: vgpu: flatten out t19x vgpu hal Instead of calling the native HAL init function then adding multiple layers of modification for VGPU, flatten out the sequence so that all entry points are set statically and visible in a single file. JIRA ESRM-30 Change-Id: I8d277aaccb0e63b2d504e7aba32eb31ef82f4ec0 Signed-off-by: Peter Daifuku Reviewed-on: https://git-master.nvidia.com/r/1574619 Reviewed-by: svc-mobile-coverity Reviewed-by: Richard Zhao GVS: Gerrit_Virtual_Submit Reviewed-by: Terje Bergstrom --- drivers/gpu/nvgpu/gv11b/hal_gv11b.c | 2 +- drivers/gpu/nvgpu/gv11b/hal_gv11b.h | 3 +- drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.c | 16 +- drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.h | 5 +- drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.c | 10 +- drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.h | 4 +- drivers/gpu/nvgpu/vgpu/gv11b/vgpu_hal_gv11b.c | 618 ++++++++++++++++++++++++- 7 files changed, 619 insertions(+), 39 deletions(-) diff --git a/drivers/gpu/nvgpu/gv11b/hal_gv11b.c b/drivers/gpu/nvgpu/gv11b/hal_gv11b.c index 4a289d8f..4241145a 100644 --- a/drivers/gpu/nvgpu/gv11b/hal_gv11b.c +++ b/drivers/gpu/nvgpu/gv11b/hal_gv11b.c @@ -94,7 +94,7 @@ #include #include -static int gv11b_get_litter_value(struct gk20a *g, int value) +int gv11b_get_litter_value(struct gk20a *g, int value) { int ret = EINVAL; switch (value) { diff --git a/drivers/gpu/nvgpu/gv11b/hal_gv11b.h b/drivers/gpu/nvgpu/gv11b/hal_gv11b.h index 0fdda6f3..668353dc 100644 --- a/drivers/gpu/nvgpu/gv11b/hal_gv11b.h +++ b/drivers/gpu/nvgpu/gv11b/hal_gv11b.h @@ -1,7 +1,7 @@ /* * GV11B Tegra HAL interface * - * Copyright (c) 2016, NVIDIA CORPORATION. All rights reserved. + * Copyright (c) 2016-2017, NVIDIA CORPORATION. All rights reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -27,4 +27,5 @@ struct gk20a; int gv11b_init_hal(struct gk20a *gops); +int gv11b_get_litter_value(struct gk20a *g, int value); #endif diff --git a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.c b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.c index ff13b11f..048a4c64 100644 --- a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.c +++ b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.c @@ -21,14 +21,10 @@ */ #include -#include #include "vgpu/vgpu.h" -#include "vgpu_fifo_gv11b.h" -#include "vgpu_subctx_gv11b.h" - -static int vgpu_gv11b_init_fifo_setup_hw(struct gk20a *g) +int vgpu_gv11b_init_fifo_setup_hw(struct gk20a *g) { struct fifo_gk20a *f = &g->fifo; int err; @@ -43,13 +39,3 @@ static int vgpu_gv11b_init_fifo_setup_hw(struct gk20a *g) return 0; } - -void vgpu_gv11b_init_fifo_ops(struct gpu_ops *gops) -{ - vgpu_gp10b_init_fifo_ops(gops); - - gops->fifo.init_fifo_setup_hw = vgpu_gv11b_init_fifo_setup_hw; - gops->fifo.free_channel_ctx_header = vgpu_gv11b_free_subctx_header; - /* TODO: implement it for CE fault */ - gops->fifo.tsg_verify_status_faulted = NULL; -} diff --git a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.h b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.h index b3c7f729..03404542 100644 --- a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.h +++ b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_fifo_gv11b.h @@ -23,8 +23,7 @@ #ifndef _VGPU_FIFO_GV11B_H_ #define _VGPU_FIFO_GV11B_H_ -struct gpu_ops; - -void vgpu_gv11b_init_fifo_ops(struct gpu_ops *gops); +struct gk20a; +int vgpu_gv11b_init_fifo_setup_hw(struct gk20a *g); #endif diff --git a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.c b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.c index a3d1cd90..89952221 100644 --- a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.c +++ b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.c @@ -22,12 +22,10 @@ #include #include -#include -#include "vgpu_gr_gv11b.h" #include "vgpu_subctx_gv11b.h" -static int vgpu_gr_gv11b_commit_inst(struct channel_gk20a *c, u64 gpu_va) +int vgpu_gr_gv11b_commit_inst(struct channel_gk20a *c, u64 gpu_va) { int err; @@ -41,9 +39,3 @@ static int vgpu_gr_gv11b_commit_inst(struct channel_gk20a *c, u64 gpu_va) return err; } - -void vgpu_gv11b_init_gr_ops(struct gpu_ops *gops) -{ - vgpu_gp10b_init_gr_ops(gops); - gops->gr.commit_inst = vgpu_gr_gv11b_commit_inst; -} diff --git a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.h b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.h index 933aa06b..562198ca 100644 --- a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.h +++ b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_gr_gv11b.h @@ -23,6 +23,8 @@ #ifndef _VGPU_GR_GV11B_H_ #define _VGPU_GR_GV11B_H_ -void vgpu_gv11b_init_gr_ops(struct gpu_ops *gops); +struct channel_gk20a; + +int vgpu_gr_gv11b_commit_inst(struct channel_gk20a *c, u64 gpu_va); #endif diff --git a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_hal_gv11b.c b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_hal_gv11b.c index 34f1198f..aa661d66 100644 --- a/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_hal_gv11b.c +++ b/drivers/gpu/nvgpu/vgpu/gv11b/vgpu_hal_gv11b.c @@ -23,27 +23,627 @@ #include #include #include +#include +#include +#include +#include +#include +#include +#include #include +#include #include +#include + +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include #include "vgpu_gr_gv11b.h" #include "vgpu_fifo_gv11b.h" +#include "vgpu_subctx_gv11b.h" + +#include +#include +#include +#include +#include + +static const struct gpu_ops vgpu_gv11b_ops = { + .ltc = { + .determine_L2_size_bytes = vgpu_determine_L2_size_bytes, + .set_zbc_s_entry = gv11b_ltc_set_zbc_stencil_entry, + .set_zbc_color_entry = gm20b_ltc_set_zbc_color_entry, + .set_zbc_depth_entry = gm20b_ltc_set_zbc_depth_entry, + .init_cbc = NULL, + .init_fs_state = vgpu_ltc_init_fs_state, + .init_comptags = vgpu_ltc_init_comptags, + .cbc_ctrl = NULL, + .isr = gv11b_ltc_isr, + .cbc_fix_config = gv11b_ltc_cbc_fix_config, + .flush = gm20b_flush_ltc, + .set_enabled = gp10b_ltc_set_enabled, + }, + .ce2 = { + .isr_stall = gv11b_ce_isr, + .isr_nonstall = gp10b_ce_nonstall_isr, + .get_num_pce = vgpu_ce_get_num_pce, + }, + .gr = { + .init_gpc_mmu = gr_gv11b_init_gpc_mmu, + .bundle_cb_defaults = gr_gv11b_bundle_cb_defaults, + .cb_size_default = gr_gv11b_cb_size_default, + .calc_global_ctx_buffer_size = + gr_gv11b_calc_global_ctx_buffer_size, + .commit_global_attrib_cb = gr_gv11b_commit_global_attrib_cb, + .commit_global_bundle_cb = gr_gp10b_commit_global_bundle_cb, + .commit_global_cb_manager = gr_gp10b_commit_global_cb_manager, + .commit_global_pagepool = gr_gp10b_commit_global_pagepool, + .handle_sw_method = gr_gv11b_handle_sw_method, + .set_alpha_circular_buffer_size = + gr_gv11b_set_alpha_circular_buffer_size, + .set_circular_buffer_size = gr_gv11b_set_circular_buffer_size, + .enable_hww_exceptions = gr_gv11b_enable_hww_exceptions, + .is_valid_class = gr_gv11b_is_valid_class, + .is_valid_gfx_class = gr_gv11b_is_valid_gfx_class, + .is_valid_compute_class = gr_gv11b_is_valid_compute_class, + .get_sm_dsm_perf_regs = gv11b_gr_get_sm_dsm_perf_regs, + .get_sm_dsm_perf_ctrl_regs = gv11b_gr_get_sm_dsm_perf_ctrl_regs, + .init_fs_state = vgpu_gm20b_init_fs_state, + .set_hww_esr_report_mask = gv11b_gr_set_hww_esr_report_mask, + .falcon_load_ucode = gr_gm20b_load_ctxsw_ucode_segments, + .load_ctxsw_ucode = gr_gk20a_load_ctxsw_ucode, + .set_gpc_tpc_mask = gr_gv11b_set_gpc_tpc_mask, + .get_gpc_tpc_mask = vgpu_gr_get_gpc_tpc_mask, + .free_channel_ctx = vgpu_gr_free_channel_ctx, + .alloc_obj_ctx = vgpu_gr_alloc_obj_ctx, + .bind_ctxsw_zcull = vgpu_gr_bind_ctxsw_zcull, + .get_zcull_info = vgpu_gr_get_zcull_info, + .is_tpc_addr = gr_gm20b_is_tpc_addr, + .get_tpc_num = gr_gm20b_get_tpc_num, + .detect_sm_arch = vgpu_gr_detect_sm_arch, + .add_zbc_color = gr_gp10b_add_zbc_color, + .add_zbc_depth = gr_gp10b_add_zbc_depth, + .zbc_set_table = vgpu_gr_add_zbc, + .zbc_query_table = vgpu_gr_query_zbc, + .pmu_save_zbc = gk20a_pmu_save_zbc, + .add_zbc = gr_gk20a_add_zbc, + .pagepool_default_size = gr_gv11b_pagepool_default_size, + .init_ctx_state = vgpu_gr_gp10b_init_ctx_state, + .alloc_gr_ctx = vgpu_gr_gp10b_alloc_gr_ctx, + .free_gr_ctx = vgpu_gr_gp10b_free_gr_ctx, + .update_ctxsw_preemption_mode = + gr_gp10b_update_ctxsw_preemption_mode, + .dump_gr_regs = NULL, + .update_pc_sampling = gr_gm20b_update_pc_sampling, + .get_fbp_en_mask = vgpu_gr_get_fbp_en_mask, + .get_max_ltc_per_fbp = vgpu_gr_get_max_ltc_per_fbp, + .get_max_lts_per_ltc = vgpu_gr_get_max_lts_per_ltc, + .get_rop_l2_en_mask = vgpu_gr_rop_l2_en_mask, + .get_max_fbps_count = vgpu_gr_get_max_fbps_count, + .init_sm_dsm_reg_info = gv11b_gr_init_sm_dsm_reg_info, + .wait_empty = gr_gv11b_wait_empty, + .init_cyclestats = gr_gv11b_init_cyclestats, + .set_sm_debug_mode = vgpu_gr_set_sm_debug_mode, + .enable_cde_in_fecs = gr_gm20b_enable_cde_in_fecs, + .bpt_reg_info = gv11b_gr_bpt_reg_info, + .get_access_map = gr_gv11b_get_access_map, + .handle_fecs_error = gr_gv11b_handle_fecs_error, + .handle_sm_exception = gr_gk20a_handle_sm_exception, + .handle_tex_exception = gr_gv11b_handle_tex_exception, + .enable_gpc_exceptions = gr_gv11b_enable_gpc_exceptions, + .enable_exceptions = gr_gv11b_enable_exceptions, + .get_lrf_tex_ltc_dram_override = get_ecc_override_val, + .update_smpc_ctxsw_mode = vgpu_gr_update_smpc_ctxsw_mode, + .update_hwpm_ctxsw_mode = vgpu_gr_update_hwpm_ctxsw_mode, + .record_sm_error_state = gv11b_gr_record_sm_error_state, + .update_sm_error_state = gv11b_gr_update_sm_error_state, + .clear_sm_error_state = vgpu_gr_clear_sm_error_state, + .suspend_contexts = vgpu_gr_suspend_contexts, + .resume_contexts = vgpu_gr_resume_contexts, + .get_preemption_mode_flags = gr_gp10b_get_preemption_mode_flags, + .fuse_override = gp10b_gr_fuse_override, + .init_sm_id_table = gr_gv11b_init_sm_id_table, + .load_smid_config = gr_gv11b_load_smid_config, + .program_sm_id_numbering = gr_gv11b_program_sm_id_numbering, + .is_ltcs_ltss_addr = gr_gm20b_is_ltcs_ltss_addr, + .is_ltcn_ltss_addr = gr_gm20b_is_ltcn_ltss_addr, + .split_lts_broadcast_addr = gr_gm20b_split_lts_broadcast_addr, + .split_ltc_broadcast_addr = gr_gm20b_split_ltc_broadcast_addr, + .setup_rop_mapping = gr_gv11b_setup_rop_mapping, + .program_zcull_mapping = gr_gv11b_program_zcull_mapping, + .commit_global_timeslice = gr_gv11b_commit_global_timeslice, + .commit_inst = vgpu_gr_gv11b_commit_inst, + .write_zcull_ptr = gr_gv11b_write_zcull_ptr, + .write_pm_ptr = gr_gv11b_write_pm_ptr, + .init_elcg_mode = gr_gv11b_init_elcg_mode, + .load_tpc_mask = gr_gv11b_load_tpc_mask, + .inval_icache = gr_gk20a_inval_icache, + .trigger_suspend = gv11b_gr_sm_trigger_suspend, + .wait_for_pause = gr_gk20a_wait_for_pause, + .resume_from_pause = gv11b_gr_resume_from_pause, + .clear_sm_errors = gr_gk20a_clear_sm_errors, + .tpc_enabled_exceptions = gr_gk20a_tpc_enabled_exceptions, + .get_esr_sm_sel = gv11b_gr_get_esr_sm_sel, + .sm_debugger_attached = gv11b_gr_sm_debugger_attached, + .suspend_single_sm = gv11b_gr_suspend_single_sm, + .suspend_all_sms = gv11b_gr_suspend_all_sms, + .resume_single_sm = gv11b_gr_resume_single_sm, + .resume_all_sms = gv11b_gr_resume_all_sms, + .get_sm_hww_warp_esr = gv11b_gr_get_sm_hww_warp_esr, + .get_sm_hww_global_esr = gv11b_gr_get_sm_hww_global_esr, + .get_sm_no_lock_down_hww_global_esr_mask = + gv11b_gr_get_sm_no_lock_down_hww_global_esr_mask, + .lock_down_sm = gv11b_gr_lock_down_sm, + .wait_for_sm_lock_down = gv11b_gr_wait_for_sm_lock_down, + .clear_sm_hww = gv11b_gr_clear_sm_hww, + .init_ovr_sm_dsm_perf = gv11b_gr_init_ovr_sm_dsm_perf, + .get_ovr_perf_regs = gv11b_gr_get_ovr_perf_regs, + .disable_rd_coalesce = gm20a_gr_disable_rd_coalesce, + .set_boosted_ctx = NULL, + .set_preemption_mode = vgpu_gr_gp10b_set_preemption_mode, + .set_czf_bypass = NULL, + .pre_process_sm_exception = gr_gv11b_pre_process_sm_exception, + .set_preemption_buffer_va = gr_gv11b_set_preemption_buffer_va, + .init_preemption_state = NULL, + .update_boosted_ctx = NULL, + .set_bes_crop_debug3 = gr_gp10b_set_bes_crop_debug3, + .create_gr_sysfs = gr_gv11b_create_sysfs, + .set_ctxsw_preemption_mode = vgpu_gr_gp10b_set_ctxsw_preemption_mode, + .is_etpc_addr = gv11b_gr_pri_is_etpc_addr, + .egpc_etpc_priv_addr_table = gv11b_gr_egpc_etpc_priv_addr_table, + .handle_tpc_mpc_exception = gr_gv11b_handle_tpc_mpc_exception, + .zbc_s_query_table = gr_gv11b_zbc_s_query_table, + .load_zbc_s_default_tbl = gr_gv11b_load_stencil_default_tbl, + .restore_context_header = gv11b_restore_context_header, + .handle_gpc_gpcmmu_exception = + gr_gv11b_handle_gpc_gpcmmu_exception, + .add_zbc_type_s = gr_gv11b_add_zbc_type_s, + .get_egpc_base = gv11b_gr_get_egpc_base, + .get_egpc_etpc_num = gv11b_gr_get_egpc_etpc_num, + .handle_gpc_gpccs_exception = + gr_gv11b_handle_gpc_gpccs_exception, + .load_zbc_s_tbl = gr_gv11b_load_stencil_tbl, + .access_smpc_reg = gv11b_gr_access_smpc_reg, + .is_egpc_addr = gv11b_gr_pri_is_egpc_addr, + .add_zbc_s = gr_gv11b_add_zbc_stencil, + .handle_gcc_exception = gr_gv11b_handle_gcc_exception, + .init_sw_veid_bundle = gr_gv11b_init_sw_veid_bundle, + .handle_tpc_sm_ecc_exception = + gr_gv11b_handle_tpc_sm_ecc_exception, + .decode_egpc_addr = gv11b_gr_decode_egpc_addr, + .init_ctxsw_hdr_data = gr_gp10b_init_ctxsw_hdr_data, + }, + .fb = { + .reset = gv11b_fb_reset, + .init_hw = gk20a_fb_init_hw, + .init_fs_state = gv11b_fb_init_fs_state, + .init_cbc = gv11b_fb_init_cbc, + .set_mmu_page_size = gm20b_fb_set_mmu_page_size, + .set_use_full_comp_tag_line = + gm20b_fb_set_use_full_comp_tag_line, + .compression_page_size = gp10b_fb_compression_page_size, + .compressible_page_size = gp10b_fb_compressible_page_size, + .vpr_info_fetch = gm20b_fb_vpr_info_fetch, + .dump_vpr_wpr_info = gm20b_fb_dump_vpr_wpr_info, + .read_wpr_info = gm20b_fb_read_wpr_info, + .is_debug_mode_enabled = NULL, + .set_debug_mode = vgpu_mm_mmu_set_debug_mode, + .tlb_invalidate = vgpu_mm_tlb_invalidate, + .hub_isr = gv11b_fb_hub_isr, + }, + .clock_gating = { + .slcg_bus_load_gating_prod = + gv11b_slcg_bus_load_gating_prod, + .slcg_ce2_load_gating_prod = + gv11b_slcg_ce2_load_gating_prod, + .slcg_chiplet_load_gating_prod = + gv11b_slcg_chiplet_load_gating_prod, + .slcg_ctxsw_firmware_load_gating_prod = + gv11b_slcg_ctxsw_firmware_load_gating_prod, + .slcg_fb_load_gating_prod = + gv11b_slcg_fb_load_gating_prod, + .slcg_fifo_load_gating_prod = + gv11b_slcg_fifo_load_gating_prod, + .slcg_gr_load_gating_prod = + gr_gv11b_slcg_gr_load_gating_prod, + .slcg_ltc_load_gating_prod = + ltc_gv11b_slcg_ltc_load_gating_prod, + .slcg_perf_load_gating_prod = + gv11b_slcg_perf_load_gating_prod, + .slcg_priring_load_gating_prod = + gv11b_slcg_priring_load_gating_prod, + .slcg_pmu_load_gating_prod = + gv11b_slcg_pmu_load_gating_prod, + .slcg_therm_load_gating_prod = + gv11b_slcg_therm_load_gating_prod, + .slcg_xbar_load_gating_prod = + gv11b_slcg_xbar_load_gating_prod, + .blcg_bus_load_gating_prod = + gv11b_blcg_bus_load_gating_prod, + .blcg_ce_load_gating_prod = + gv11b_blcg_ce_load_gating_prod, + .blcg_ctxsw_firmware_load_gating_prod = + gv11b_blcg_ctxsw_firmware_load_gating_prod, + .blcg_fb_load_gating_prod = + gv11b_blcg_fb_load_gating_prod, + .blcg_fifo_load_gating_prod = + gv11b_blcg_fifo_load_gating_prod, + .blcg_gr_load_gating_prod = + gv11b_blcg_gr_load_gating_prod, + .blcg_ltc_load_gating_prod = + gv11b_blcg_ltc_load_gating_prod, + .blcg_pwr_csb_load_gating_prod = + gv11b_blcg_pwr_csb_load_gating_prod, + .blcg_pmu_load_gating_prod = + gv11b_blcg_pmu_load_gating_prod, + .blcg_xbar_load_gating_prod = + gv11b_blcg_xbar_load_gating_prod, + .pg_gr_load_gating_prod = + gr_gv11b_pg_gr_load_gating_prod, + }, + .fifo = { + .init_fifo_setup_hw = vgpu_gv11b_init_fifo_setup_hw, + .bind_channel = vgpu_channel_bind, + .unbind_channel = vgpu_channel_unbind, + .disable_channel = vgpu_channel_disable, + .enable_channel = vgpu_channel_enable, + .alloc_inst = vgpu_channel_alloc_inst, + .free_inst = vgpu_channel_free_inst, + .setup_ramfc = vgpu_channel_setup_ramfc, + .channel_set_priority = vgpu_channel_set_priority, + .channel_set_timeslice = vgpu_channel_set_timeslice, + .default_timeslice_us = vgpu_fifo_default_timeslice_us, + .setup_userd = gk20a_fifo_setup_userd, + .userd_gp_get = gv11b_userd_gp_get, + .userd_gp_put = gv11b_userd_gp_put, + .userd_pb_get = gv11b_userd_pb_get, + .pbdma_acquire_val = gk20a_fifo_pbdma_acquire_val, + .preempt_channel = vgpu_fifo_preempt_channel, + .preempt_tsg = vgpu_fifo_preempt_tsg, + .enable_tsg = gk20a_enable_tsg, + .disable_tsg = gk20a_disable_tsg, + .tsg_verify_channel_status = NULL, + .tsg_verify_status_ctx_reload = gm20b_fifo_tsg_verify_status_ctx_reload, + /* TODO: implement it for CE fault */ + .tsg_verify_status_faulted = NULL, + .update_runlist = vgpu_fifo_update_runlist, + .trigger_mmu_fault = NULL, + .get_mmu_fault_info = NULL, + .wait_engine_idle = vgpu_fifo_wait_engine_idle, + .get_num_fifos = gv11b_fifo_get_num_fifos, + .get_pbdma_signature = gp10b_fifo_get_pbdma_signature, + .set_runlist_interleave = vgpu_fifo_set_runlist_interleave, + .tsg_set_timeslice = vgpu_tsg_set_timeslice, + .tsg_open = vgpu_tsg_open, + .force_reset_ch = vgpu_fifo_force_reset_ch, + .engine_enum_from_type = gp10b_fifo_engine_enum_from_type, + .device_info_data_parse = gp10b_device_info_data_parse, + .eng_runlist_base_size = fifo_eng_runlist_base__size_1_v, + .init_engine_info = vgpu_fifo_init_engine_info, + .runlist_entry_size = ram_rl_entry_size_v, + .get_tsg_runlist_entry = gv11b_get_tsg_runlist_entry, + .get_ch_runlist_entry = gv11b_get_ch_runlist_entry, + .is_fault_engine_subid_gpc = gv11b_is_fault_engine_subid_gpc, + .dump_pbdma_status = gk20a_dump_pbdma_status, + .dump_eng_status = gv11b_dump_eng_status, + .dump_channel_status_ramfc = gv11b_dump_channel_status_ramfc, + .intr_0_error_mask = gv11b_fifo_intr_0_error_mask, + .is_preempt_pending = gv11b_fifo_is_preempt_pending, + .init_pbdma_intr_descs = gv11b_fifo_init_pbdma_intr_descs, + .reset_enable_hw = gv11b_init_fifo_reset_enable_hw, + .teardown_ch_tsg = gv11b_fifo_teardown_ch_tsg, + .handle_sched_error = gv11b_fifo_handle_sched_error, + .handle_pbdma_intr_0 = gv11b_fifo_handle_pbdma_intr_0, + .handle_pbdma_intr_1 = gv11b_fifo_handle_pbdma_intr_1, + .init_eng_method_buffers = gv11b_fifo_init_eng_method_buffers, + .deinit_eng_method_buffers = + gv11b_fifo_deinit_eng_method_buffers, + .tsg_bind_channel = vgpu_tsg_bind_channel, + .tsg_unbind_channel = vgpu_tsg_unbind_channel, +#ifdef CONFIG_TEGRA_GK20A_NVHOST + .alloc_syncpt_buf = gv11b_fifo_alloc_syncpt_buf, + .free_syncpt_buf = gv11b_fifo_free_syncpt_buf, + .add_syncpt_wait_cmd = gv11b_fifo_add_syncpt_wait_cmd, + .get_syncpt_wait_cmd_size = gv11b_fifo_get_syncpt_wait_cmd_size, + .add_syncpt_incr_cmd = gv11b_fifo_add_syncpt_incr_cmd, + .get_syncpt_incr_cmd_size = gv11b_fifo_get_syncpt_incr_cmd_size, +#endif + .resetup_ramfc = NULL, + .reschedule_runlist = NULL, + .device_info_fault_id = top_device_info_data_fault_id_enum_v, + .free_channel_ctx_header = vgpu_gv11b_free_subctx_header, + .preempt_ch_tsg = gv11b_fifo_preempt_ch_tsg, + .handle_ctxsw_timeout = gv11b_fifo_handle_ctxsw_timeout, + }, + .gr_ctx = { + .get_netlist_name = gr_gv11b_get_netlist_name, + .is_fw_defined = gr_gv11b_is_firmware_defined, + }, +#ifdef CONFIG_GK20A_CTXSW_TRACE + .fecs_trace = { + .alloc_user_buffer = NULL, + .free_user_buffer = NULL, + .mmap_user_buffer = NULL, + .init = NULL, + .deinit = NULL, + .enable = NULL, + .disable = NULL, + .is_enabled = NULL, + .reset = NULL, + .flush = NULL, + .poll = NULL, + .bind_channel = NULL, + .unbind_channel = NULL, + .max_entries = NULL, + }, +#endif /* CONFIG_GK20A_CTXSW_TRACE */ + .mm = { + /* FIXME: add support for sparse mappings */ + .support_sparse = NULL, + .gmmu_map = vgpu_gp10b_locked_gmmu_map, + .gmmu_unmap = vgpu_locked_gmmu_unmap, + .vm_bind_channel = vgpu_vm_bind_channel, + .fb_flush = vgpu_mm_fb_flush, + .l2_invalidate = vgpu_mm_l2_invalidate, + .l2_flush = vgpu_mm_l2_flush, + .cbc_clean = gk20a_mm_cbc_clean, + .set_big_page_size = gm20b_mm_set_big_page_size, + .get_big_page_sizes = gm20b_mm_get_big_page_sizes, + .get_default_big_page_size = gp10b_mm_get_default_big_page_size, + .gpu_phys_addr = gm20b_gpu_phys_addr, + .get_iommu_bit = gk20a_mm_get_iommu_bit, + .get_mmu_levels = gp10b_mm_get_mmu_levels, + .init_pdb = gp10b_mm_init_pdb, + .init_mm_setup_hw = vgpu_gp10b_init_mm_setup_hw, + .is_bar1_supported = gv11b_mm_is_bar1_supported, + .init_inst_block = gv11b_init_inst_block, + .mmu_fault_pending = gv11b_mm_mmu_fault_pending, + .init_bar2_vm = gb10b_init_bar2_vm, + .init_bar2_mm_hw_setup = gv11b_init_bar2_mm_hw_setup, + .remove_bar2_vm = gv11b_mm_remove_bar2_vm, + .fault_info_mem_destroy = gv11b_mm_fault_info_mem_destroy, + }, + .therm = { + .init_therm_setup_hw = gp10b_init_therm_setup_hw, + .elcg_init_idle_filters = gv11b_elcg_init_idle_filters, + }, + .pmu = { + .pmu_setup_elpg = gp10b_pmu_setup_elpg, + .pmu_get_queue_head = pwr_pmu_queue_head_r, + .pmu_get_queue_head_size = pwr_pmu_queue_head__size_1_v, + .pmu_get_queue_tail = pwr_pmu_queue_tail_r, + .pmu_get_queue_tail_size = pwr_pmu_queue_tail__size_1_v, + .pmu_queue_head = gk20a_pmu_queue_head, + .pmu_queue_tail = gk20a_pmu_queue_tail, + .pmu_msgq_tail = gk20a_pmu_msgq_tail, + .pmu_mutex_size = pwr_pmu_mutex__size_1_v, + .pmu_mutex_acquire = gk20a_pmu_mutex_acquire, + .pmu_mutex_release = gk20a_pmu_mutex_release, + .write_dmatrfbase = gp10b_write_dmatrfbase, + .pmu_elpg_statistics = gp106_pmu_elpg_statistics, + .pmu_pg_init_param = gv11b_pg_gr_init, + .pmu_pg_supported_engines_list = gk20a_pmu_pg_engines_list, + .pmu_pg_engines_feature_list = gk20a_pmu_pg_feature_list, + .dump_secure_fuses = pmu_dump_security_fuses_gp10b, + .reset_engine = gp106_pmu_engine_reset, + .is_engine_in_reset = gp106_pmu_is_engine_in_reset, + .pmu_nsbootstrap = gv11b_pmu_bootstrap, + .pmu_pg_set_sub_feature_mask = gv11b_pg_set_subfeature_mask, + .is_pmu_supported = gv11b_is_pmu_supported, + }, + .regops = { + .get_global_whitelist_ranges = + gv11b_get_global_whitelist_ranges, + .get_global_whitelist_ranges_count = + gv11b_get_global_whitelist_ranges_count, + .get_context_whitelist_ranges = + gv11b_get_context_whitelist_ranges, + .get_context_whitelist_ranges_count = + gv11b_get_context_whitelist_ranges_count, + .get_runcontrol_whitelist = gv11b_get_runcontrol_whitelist, + .get_runcontrol_whitelist_count = + gv11b_get_runcontrol_whitelist_count, + .get_runcontrol_whitelist_ranges = + gv11b_get_runcontrol_whitelist_ranges, + .get_runcontrol_whitelist_ranges_count = + gv11b_get_runcontrol_whitelist_ranges_count, + .get_qctl_whitelist = gv11b_get_qctl_whitelist, + .get_qctl_whitelist_count = gv11b_get_qctl_whitelist_count, + .get_qctl_whitelist_ranges = gv11b_get_qctl_whitelist_ranges, + .get_qctl_whitelist_ranges_count = + gv11b_get_qctl_whitelist_ranges_count, + .apply_smpc_war = gv11b_apply_smpc_war, + }, + .mc = { + .intr_enable = mc_gv11b_intr_enable, + .intr_unit_config = mc_gp10b_intr_unit_config, + .isr_stall = mc_gp10b_isr_stall, + .intr_stall = mc_gp10b_intr_stall, + .intr_stall_pause = mc_gp10b_intr_stall_pause, + .intr_stall_resume = mc_gp10b_intr_stall_resume, + .intr_nonstall = mc_gp10b_intr_nonstall, + .intr_nonstall_pause = mc_gp10b_intr_nonstall_pause, + .intr_nonstall_resume = mc_gp10b_intr_nonstall_resume, + .enable = gk20a_mc_enable, + .disable = gk20a_mc_disable, + .reset = gk20a_mc_reset, + .boot_0 = gk20a_mc_boot_0, + .is_intr1_pending = mc_gp10b_is_intr1_pending, + .is_intr_hub_pending = gv11b_mc_is_intr_hub_pending, + }, + .debug = { + .show_dump = NULL, + }, + .dbg_session_ops = { + .exec_reg_ops = vgpu_exec_regops, + .dbg_set_powergate = vgpu_dbg_set_powergate, + .check_and_set_global_reservation = + vgpu_check_and_set_global_reservation, + .check_and_set_context_reservation = + vgpu_check_and_set_context_reservation, + .release_profiler_reservation = + vgpu_release_profiler_reservation, + .perfbuffer_enable = vgpu_perfbuffer_enable, + .perfbuffer_disable = vgpu_perfbuffer_disable, + }, + .bus = { + .init_hw = gk20a_bus_init_hw, + .isr = gk20a_bus_isr, + .read_ptimer = vgpu_read_ptimer, + .get_timestamps_zipper = vgpu_get_timestamps_zipper, + .bar1_bind = NULL, + }, +#if defined(CONFIG_GK20A_CYCLE_STATS) + .css = { + .enable_snapshot = vgpu_css_enable_snapshot_buffer, + .disable_snapshot = vgpu_css_release_snapshot_buffer, + .check_data_available = vgpu_css_flush_snapshots, + .set_handled_snapshots = NULL, + .allocate_perfmon_ids = NULL, + .release_perfmon_ids = NULL, + }, +#endif + .falcon = { + .falcon_hal_sw_init = gk20a_falcon_hal_sw_init, + }, + .priv_ring = { + .isr = gp10b_priv_ring_isr, + }, + .chip_init_gpu_characteristics = vgpu_init_gpu_characteristics, + .get_litter_value = gv11b_get_litter_value, +}; int vgpu_gv11b_init_hal(struct gk20a *g) { - int err; + struct gpu_ops *gops = &g->ops; + struct nvgpu_gpu_characteristics *c = &g->gpu_characteristics; + u32 val; + bool priv_security; + + gops->ltc = vgpu_gv11b_ops.ltc; + gops->ce2 = vgpu_gv11b_ops.ce2; + gops->gr = vgpu_gv11b_ops.gr; + gops->fb = vgpu_gv11b_ops.fb; + gops->clock_gating = vgpu_gv11b_ops.clock_gating; + gops->fifo = vgpu_gv11b_ops.fifo; + gops->gr_ctx = vgpu_gv11b_ops.gr_ctx; + gops->mm = vgpu_gv11b_ops.mm; + gops->fecs_trace = vgpu_gv11b_ops.fecs_trace; + gops->therm = vgpu_gv11b_ops.therm; + gops->pmu = vgpu_gv11b_ops.pmu; + gops->regops = vgpu_gv11b_ops.regops; + gops->mc = vgpu_gv11b_ops.mc; + gops->debug = vgpu_gv11b_ops.debug; + gops->dbg_session_ops = vgpu_gv11b_ops.dbg_session_ops; + gops->bus = vgpu_gv11b_ops.bus; +#if defined(CONFIG_GK20A_CYCLE_STATS) + gops->css = vgpu_gv11b_ops.css; +#endif + gops->falcon = vgpu_gv11b_ops.falcon; + gops->priv_ring = vgpu_gv11b_ops.priv_ring; + + /* Lone functions */ + gops->chip_init_gpu_characteristics = + vgpu_gv11b_ops.chip_init_gpu_characteristics; + gops->get_litter_value = vgpu_gv11b_ops.get_litter_value; + + val = gk20a_readl(g, fuse_opt_priv_sec_en_r()); + if (val) { + priv_security = true; + pr_err("priv security is enabled\n"); + } else { + priv_security = false; + pr_err("priv security is disabled\n"); + } + __nvgpu_set_enabled(g, NVGPU_GR_USE_DMA_FOR_FW_BOOTSTRAP, false); + __nvgpu_set_enabled(g, NVGPU_SEC_PRIVSECURITY, priv_security); + __nvgpu_set_enabled(g, NVGPU_SEC_SECUREGPCCS, priv_security); + + /* priv security dependent ops */ + if (nvgpu_is_enabled(g, NVGPU_SEC_PRIVSECURITY)) { + /* Add in ops from gm20b acr */ + gops->pmu.prepare_ucode = gp106_prepare_ucode_blob, + gops->pmu.pmu_setup_hw_and_bootstrap = gv11b_bootstrap_hs_flcn, + gops->pmu.get_wpr = gm20b_wpr_info, + gops->pmu.alloc_blob_space = gm20b_alloc_blob_space, + gops->pmu.pmu_populate_loader_cfg = + gp106_pmu_populate_loader_cfg, + gops->pmu.flcn_populate_bl_dmem_desc = + gp106_flcn_populate_bl_dmem_desc, + gops->pmu.falcon_wait_for_halt = pmu_wait_for_halt, + gops->pmu.falcon_clear_halt_interrupt_status = + clear_halt_interrupt_status, + gops->pmu.init_falcon_setup_hw = gv11b_init_pmu_setup_hw1, + + gops->pmu.init_wpr_region = gm20b_pmu_init_acr; + gops->pmu.load_lsfalcon_ucode = gp10b_load_falcon_ucode; + gops->pmu.is_lazy_bootstrap = gv11b_is_lazy_bootstrap, + gops->pmu.is_priv_load = gv11b_is_priv_load, + + gops->gr.load_ctxsw_ucode = gr_gm20b_load_ctxsw_ucode; + } else { + /* Inherit from gk20a */ + gops->pmu.prepare_ucode = nvgpu_pmu_prepare_ns_ucode_blob, + gops->pmu.pmu_setup_hw_and_bootstrap = gk20a_init_pmu_setup_hw1, + + gops->pmu.load_lsfalcon_ucode = NULL; + gops->pmu.init_wpr_region = NULL; + gops->pmu.pmu_setup_hw_and_bootstrap = gp10b_init_pmu_setup_hw1; - gk20a_dbg_fn(""); + gops->gr.load_ctxsw_ucode = gr_gk20a_load_ctxsw_ucode; + } - err = gv11b_init_hal(g); - if (err) - return err; + __nvgpu_set_enabled(g, NVGPU_PMU_FECS_BOOTSTRAP_DONE, false); + gv11b_init_uncompressed_kind_map(); + gv11b_init_kind_attr(); + g->bootstrap_owner = LSF_BOOTSTRAP_OWNER_DEFAULT; - vgpu_init_hal_common(g); - vgpu_gp10b_init_mm_ops(&g->ops); + g->name = "gv11b"; - vgpu_gv11b_init_gr_ops(&g->ops); - vgpu_gv11b_init_fifo_ops(&g->ops); + c->twod_class = FERMI_TWOD_A; + c->threed_class = VOLTA_A; + c->compute_class = VOLTA_COMPUTE_A; + c->gpfifo_class = VOLTA_CHANNEL_GPFIFO_A; + c->inline_to_memory_class = KEPLER_INLINE_TO_MEMORY_B; + c->dma_copy_class = VOLTA_DMA_COPY_A; return 0; } -- cgit v1.2.2