From 5d30a5cda37ca349b4d9cb7e1985c7a0849001b6 Mon Sep 17 00:00:00 2001 From: Mahantesh Kumbar Date: Thu, 6 Sep 2018 20:44:27 +0530 Subject: gpu: nvgpu: ACR code refactor -Created struct nvgpu_acr to hold acr module related member within single struct which are currently spread across multiple structs like nvgpu_pmu, pmu_ops & gk20a. -Created struct hs_flcn_bl struct to hold ACR HS bootloader specific members -Created struct hs_acr to hold ACR ucode specific members like bootloader data using struct hs_flcn_bl, acr type & falcon info on which ACR ucode need to run. -Created acr ops under struct nvgpu_acr to perform ACR specific operation, currently ACR ops were part PMU which caused to have always dependence on PMU even though ACR was not executing on PMU. -Added acr_remove_support ops which will be called as part of gk20a_remove_support() method, earlier acr cleanup was part of pmu remove_support method. -Created define for ACR types, -Ops acr_sw_init() function helps to set ACR properties statically for chip currently in execution & assign ops to point to needed functions as per chip. -Ops acr_sw_init execute at early as nvgpu_init_mm_support calls acr function to alloc blob space. -Created ops to fill bootloader descriptor & to patch WPR info to ACR uocde based on interfaces used to bootstrap ACR ucode. -Created function gm20b_bootstrap_hs_acr() function which is now common HAL for all chips to bootstrap ACR, earlier had 3 different function for gm20b/gp10b, gv11b & for all dgpu based on interface needed. -Removed duplicate code for falcon engine wherever common falcon code can be used. -Removed ACR code dependent on PMU & made changes to use from nvgpu_acr. JIRA NVGPU-1148 Change-Id: I39951d2fc9a0bb7ee6057e0fa06da78045d47590 Signed-off-by: Mahantesh Kumbar Reviewed-on: https://git-master.nvidia.com/r/1813231 GVS: Gerrit_Virtual_Submit Reviewed-by: svc-misra-checker Reviewed-by: Vijayakumar Subbu Reviewed-by: mobile promotions Tested-by: mobile promotions --- drivers/gpu/nvgpu/gp106/acr_gp106.c | 133 ++++++++++++++++++++++++++++++++++-- 1 file changed, 126 insertions(+), 7 deletions(-) (limited to 'drivers/gpu/nvgpu/gp106/acr_gp106.c') diff --git a/drivers/gpu/nvgpu/gp106/acr_gp106.c b/drivers/gpu/nvgpu/gp106/acr_gp106.c index 7bb099e5..b1150e29 100644 --- a/drivers/gpu/nvgpu/gp106/acr_gp106.c +++ b/drivers/gpu/nvgpu/gp106/acr_gp106.c @@ -93,7 +93,7 @@ int gp106_alloc_blob_space(struct gk20a *g, return 0; } - g->ops.pmu.get_wpr(g, &wpr_inf); + g->acr.get_wpr_info(g, &wpr_inf); /* * Even though this mem_desc wouldn't be used, the wpr region needs to @@ -456,7 +456,7 @@ int gp106_prepare_ucode_blob(struct gk20a *g) memset((void *)plsfm, 0, sizeof(struct ls_flcn_mgr_v1)); gr_gk20a_init_ctxsw_ucode(g); - g->ops.pmu.get_wpr(g, &wpr_inf); + g->acr.get_wpr_info(g, &wpr_inf); gp106_dbg_pmu(g, "wpr carveout base:%llx\n", (wpr_inf.wpr_base)); gp106_dbg_pmu(g, "wpr carveout size :%x\n", (u32)wpr_inf.size); @@ -479,7 +479,7 @@ int gp106_prepare_ucode_blob(struct gk20a *g) } /*Alloc memory to hold ucode blob contents*/ - err = g->ops.pmu.alloc_blob_space(g, plsfm->wpr_size + err = g->acr.alloc_blob_space(g, plsfm->wpr_size ,&g->acr.ucode_blob); if (err) { goto exit_err; @@ -557,7 +557,7 @@ int lsfm_discover_ucode_images(struct gk20a *g, /*0th index is always PMU which is already handled in earlier if condition*/ - for (i = 1; i < (MAX_SUPPORTED_LSFM); i++) { + for (i = 1; i < g->acr.max_supported_lsfm; i++) { memset(&ucode_img, 0, sizeof(ucode_img)); if (pmu_acr_supp_ucode_list[i](g, &ucode_img) == 0) { if (ucode_img.lsf_desc != NULL) { @@ -626,7 +626,7 @@ int gp106_pmu_populate_loader_cfg(struct gk20a *g, * physical addresses of each respective segment. */ addr_base = p_lsfm->lsb_header.ucode_off; - g->ops.pmu.get_wpr(g, &wpr_inf); + g->acr.get_wpr_info(g, &wpr_inf); addr_base += (wpr_inf.wpr_base); gp106_dbg_pmu(g, "pmu loader cfg addrbase 0x%llx\n", addr_base); @@ -701,7 +701,7 @@ int gp106_flcn_populate_bl_dmem_desc(struct gk20a *g, * physical addresses of each respective segment. */ addr_base = p_lsfm->lsb_header.ucode_off; - g->ops.pmu.get_wpr(g, &wpr_inf); + g->acr.get_wpr_info(g, &wpr_inf); addr_base += wpr_inf.wpr_base; gp106_dbg_pmu(g, "falcon ID %x", p_lsfm->wpr_header.falcon_id); @@ -1017,7 +1017,7 @@ int lsfm_add_ucode_img(struct gk20a *g, struct ls_flcn_mgr_v1 *plsfm, /* Fill in static WPR header info*/ pnode->wpr_header.falcon_id = falcon_id; - pnode->wpr_header.bootstrap_owner = g->bootstrap_owner; + pnode->wpr_header.bootstrap_owner = g->acr.bootstrap_owner; pnode->wpr_header.status = LSF_IMAGE_STATUS_COPY; pnode->wpr_header.lazy_bootstrap = @@ -1030,6 +1030,7 @@ int lsfm_add_ucode_img(struct gk20a *g, struct ls_flcn_mgr_v1 *plsfm, pnode->wpr_header.bin_version = pnode->lsb_header.signature.version; pnode->next = plsfm->ucode_img_list; plsfm->ucode_img_list = pnode; + return 0; } @@ -1191,3 +1192,121 @@ int lsf_gen_wpr_requirements(struct gk20a *g, plsfm->wpr_size = wpr_offset; return 0; } + +int gp106_acr_patch_wpr_info_to_ucode(struct gk20a *g, struct nvgpu_acr *acr, + struct hs_acr *acr_desc, bool is_recovery) +{ + struct nvgpu_firmware *acr_fw = acr_desc->acr_fw; + struct acr_fw_header *acr_fw_hdr = NULL; + struct bin_hdr *acr_fw_bin_hdr = NULL; + struct flcn_acr_desc_v1 *acr_dmem_desc; + struct wpr_carveout_info wpr_inf; + u32 *acr_ucode_header = NULL; + u32 *acr_ucode_data = NULL; + + nvgpu_log_fn(g, " "); + + acr_fw_bin_hdr = (struct bin_hdr *)acr_fw->data; + acr_fw_hdr = (struct acr_fw_header *) + (acr_fw->data + acr_fw_bin_hdr->header_offset); + + acr_ucode_data = (u32 *)(acr_fw->data + acr_fw_bin_hdr->data_offset); + acr_ucode_header = (u32 *)(acr_fw->data + acr_fw_hdr->hdr_offset); + + acr->get_wpr_info(g, &wpr_inf); + + acr_dmem_desc = (struct flcn_acr_desc_v1 *) + &(((u8 *)acr_ucode_data)[acr_ucode_header[2U]]); + + acr_dmem_desc->nonwpr_ucode_blob_start = wpr_inf.nonwpr_base; + acr_dmem_desc->nonwpr_ucode_blob_size = wpr_inf.size; + acr_dmem_desc->regions.no_regions = 1U; + acr_dmem_desc->wpr_offset = 0U; + + acr_dmem_desc->wpr_region_id = 1U; + acr_dmem_desc->regions.region_props[0U].region_id = 1U; + acr_dmem_desc->regions.region_props[0U].start_addr = + (wpr_inf.wpr_base) >> 8U; + acr_dmem_desc->regions.region_props[0U].end_addr = + ((wpr_inf.wpr_base) + wpr_inf.size) >> 8U; + acr_dmem_desc->regions.region_props[0U].shadowmMem_startaddress = + wpr_inf.nonwpr_base >> 8U; + + return 0; +} + +int gp106_acr_fill_bl_dmem_desc(struct gk20a *g, + struct nvgpu_acr *acr, struct hs_acr *acr_desc, + u32 *acr_ucode_header) +{ + struct nvgpu_mem *acr_ucode_mem = &acr_desc->acr_ucode; + struct flcn_bl_dmem_desc_v1 *bl_dmem_desc = + &acr_desc->bl_dmem_desc_v1; + + nvgpu_log_fn(g, " "); + + memset(bl_dmem_desc, 0U, sizeof(struct flcn_bl_dmem_desc_v1)); + + bl_dmem_desc->signature[0] = 0U; + bl_dmem_desc->signature[1] = 0U; + bl_dmem_desc->signature[2] = 0U; + bl_dmem_desc->signature[3] = 0U; + bl_dmem_desc->ctx_dma = GK20A_PMU_DMAIDX_VIRT; + + flcn64_set_dma(&bl_dmem_desc->code_dma_base, + acr_ucode_mem->gpu_va); + + bl_dmem_desc->non_sec_code_off = acr_ucode_header[0U]; + bl_dmem_desc->non_sec_code_size = acr_ucode_header[1U]; + bl_dmem_desc->sec_code_off = acr_ucode_header[5U]; + bl_dmem_desc->sec_code_size = acr_ucode_header[6U]; + bl_dmem_desc->code_entry_point = 0U; + + flcn64_set_dma(&bl_dmem_desc->data_dma_base, + acr_ucode_mem->gpu_va + acr_ucode_header[2U]); + + bl_dmem_desc->data_size = acr_ucode_header[3U]; + + return 0; +} + +static void nvgpu_gp106_acr_default_sw_init(struct gk20a *g, struct hs_acr *hs_acr) +{ + struct hs_flcn_bl *hs_bl = &hs_acr->acr_hs_bl; + + nvgpu_log_fn(g, " "); + + hs_bl->bl_fw_name = HSBIN_ACR_BL_UCODE_IMAGE; + + hs_acr->acr_type = ACR_DEFAULT; + hs_acr->acr_fw_name = HSBIN_ACR_UCODE_IMAGE; + + hs_acr->ptr_bl_dmem_desc = &hs_acr->bl_dmem_desc_v1; + hs_acr->bl_dmem_desc_size = sizeof(struct flcn_bl_dmem_desc_v1); + + hs_acr->acr_flcn = &g->sec2_flcn; + hs_acr->acr_flcn_setup_hw_and_bl_bootstrap = + gp106_sec2_setup_hw_and_bl_bootstrap; +} + +void nvgpu_gp106_acr_sw_init(struct gk20a *g, struct nvgpu_acr *acr) +{ + nvgpu_log_fn(g, " "); + + acr->g = g; + + acr->bootstrap_owner = LSF_FALCON_ID_SEC2; + acr->max_supported_lsfm = MAX_SUPPORTED_LSFM; + + nvgpu_gp106_acr_default_sw_init(g, &acr->acr); + + acr->get_wpr_info = gp106_wpr_info; + acr->alloc_blob_space = gp106_alloc_blob_space; + acr->bootstrap_hs_acr = gm20b_bootstrap_hs_acr; + acr->patch_wpr_info_to_ucode = + gp106_acr_patch_wpr_info_to_ucode; + acr->acr_fill_bl_dmem_desc = + gp106_acr_fill_bl_dmem_desc; + + acr->remove_support = gm20b_remove_acr_support; +} -- cgit v1.2.2