drm/nouveau/acr: implement new subdev to replace "secure boot"

ACR is responsible for managing the firmware for LS (Low Secure) falcons,
this was previously handled in the driver by SECBOOT.

This rewrite started from some test code that attempted to replicate the
procedure RM uses in order to debug early Turing ACR firmwares that were
provided by NVIDIA for development.

Compared with SECBOOT, the code is structured into more individual steps,
with the aim of making the process easier to follow/debug, whilst making
it possible to support newer firmware versions that may have a different
binary format or API interface.

The HS (High Secure) binary(s) are now booted earlier in device init, to
match the behaviour of RM, whereas SECBOOT would delay this until we try
to boot the first LS falcon.

There's also additional debugging features available, with the intention
of making it easier to solve issues during FW/HW bring-up in the future.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

6 files changed, 149 insertions, 12 deletions

diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c
index 8314f10c359a..6c1a1e074721 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c
@@ -26,9 +26,9 @@
 #include "fuc/os.h"
 
 #include <core/client.h>
-#include <core/option.h>
 #include <core/firmware.h>
-#include <subdev/secboot.h>
+#include <core/option.h>
+#include <subdev/acr.h>
 #include <subdev/fb.h>
 #include <subdev/mc.h>
 #include <subdev/pmu.h>
@@ -1690,28 +1690,30 @@ gf100_gr_init_ctxctl_ext(struct gf100_gr *gr)
 {
 	struct nvkm_subdev *subdev = &gr->base.engine.subdev;
 	struct nvkm_device *device = subdev->device;
-	struct nvkm_secboot *sb = device->secboot;
-	u32 secboot_mask = 0;
+	u32 lsf_mask = 0;
 	int ret;
 
 	/* load fuc microcode */
 	nvkm_mc_unk260(device, 0);
 
 	/* securely-managed falcons must be reset using secure boot */
-	if (nvkm_secboot_is_managed(sb, NVKM_SECBOOT_FALCON_FECS))
-		secboot_mask |= BIT(NVKM_SECBOOT_FALCON_FECS);
-	else
+
+	if (!nvkm_acr_managed_falcon(device, NVKM_ACR_LSF_FECS)) {
 		gf100_gr_init_fw(&gr->fecs.falcon, &gr->fecs.inst,
 						   &gr->fecs.data);
+	} else {
+		lsf_mask |= BIT(NVKM_ACR_LSF_FECS);
+	}
 
-	if (nvkm_secboot_is_managed(sb, NVKM_SECBOOT_FALCON_GPCCS))
-		secboot_mask |= BIT(NVKM_SECBOOT_FALCON_GPCCS);
-	else
+	if (!nvkm_acr_managed_falcon(device, NVKM_ACR_LSF_GPCCS)) {
 		gf100_gr_init_fw(&gr->gpccs.falcon, &gr->gpccs.inst,
 						    &gr->gpccs.data);
+	} else {
+		lsf_mask |= BIT(NVKM_ACR_LSF_GPCCS);
+	}
 
-	if (secboot_mask != 0) {
-		int ret = nvkm_secboot_reset(sb, secboot_mask);
+	if (lsf_mask) {
+		ret = nvkm_acr_bootstrap_falcons(device, lsf_mask);
 		if (ret)
 			return ret;
 	}
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h
index aa5c9ddfd93c..67286bb57e55 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h
@@ -31,6 +31,8 @@
 #include <subdev/mmu.h>
 #include <engine/falcon.h>
 
+struct nvkm_acr_lsfw;
+
 #define GPC_MAX 32
 #define TPC_MAX_PER_GPC 8
 #define TPC_MAX (GPC_MAX * TPC_MAX_PER_GPC)
@@ -400,6 +402,8 @@ extern const struct nvkm_acr_lsf_func gm200_gr_gpccs_acr;
 extern const struct nvkm_acr_lsf_func gm200_gr_fecs_acr;
 
 extern const struct nvkm_acr_lsf_func gm20b_gr_fecs_acr;
+void gm20b_gr_acr_bld_write(struct nvkm_acr *, u32, struct nvkm_acr_lsfw *);
+void gm20b_gr_acr_bld_patch(struct nvkm_acr *, u32, s64);
 
 extern const struct nvkm_acr_lsf_func gp108_gr_gpccs_acr;
 extern const struct nvkm_acr_lsf_func gp108_gr_fecs_acr;
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c
index 3ad94d791fcf..3d67cfb08395 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c
@@ -28,18 +28,58 @@
 #include <subdev/acr.h>
 #include <subdev/secboot.h>
 
+#include <nvfw/flcn.h>
+
 #include <nvif/class.h>
 
 /*******************************************************************************
  * PGRAPH engine/subdev functions
  ******************************************************************************/
 
+static void
+gm200_gr_acr_bld_patch(struct nvkm_acr *acr, u32 bld, s64 adjust)
+{
+	struct flcn_bl_dmem_desc_v1 hdr;
+	nvkm_robj(acr->wpr, bld, &hdr, sizeof(hdr));
+	hdr.code_dma_base = hdr.code_dma_base + adjust;
+	hdr.data_dma_base = hdr.data_dma_base + adjust;
+	nvkm_wobj(acr->wpr, bld, &hdr, sizeof(hdr));
+	flcn_bl_dmem_desc_v1_dump(&acr->subdev, &hdr);
+}
+
+static void
+gm200_gr_acr_bld_write(struct nvkm_acr *acr, u32 bld,
+		       struct nvkm_acr_lsfw *lsfw)
+{
+	const u64 base = lsfw->offset.img + lsfw->app_start_offset;
+	const u64 code = base + lsfw->app_resident_code_offset;
+	const u64 data = base + lsfw->app_resident_data_offset;
+	const struct flcn_bl_dmem_desc_v1 hdr = {
+		.ctx_dma = FALCON_DMAIDX_UCODE,
+		.code_dma_base = code,
+		.non_sec_code_off = lsfw->app_resident_code_offset,
+		.non_sec_code_size = lsfw->app_resident_code_size,
+		.code_entry_point = lsfw->app_imem_entry,
+		.data_dma_base = data,
+		.data_size = lsfw->app_resident_data_size,
+	};
+
+	nvkm_wobj(acr->wpr, bld, &hdr, sizeof(hdr));
+}
+
 const struct nvkm_acr_lsf_func
 gm200_gr_gpccs_acr = {
+	.flags = NVKM_ACR_LSF_FORCE_PRIV_LOAD,
+	.bld_size = sizeof(struct flcn_bl_dmem_desc_v1),
+	.bld_write = gm200_gr_acr_bld_write,
+	.bld_patch = gm200_gr_acr_bld_patch,
 };
 
 const struct nvkm_acr_lsf_func
 gm200_gr_fecs_acr = {
+	.bld_size = sizeof(struct flcn_bl_dmem_desc_v1),
+	.bld_write = gm200_gr_acr_bld_write,
+	.bld_patch = gm200_gr_acr_bld_patch,
 };
 
 int
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c
index aaf5aff036f2..b45e8f10ec73 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c
@@ -26,10 +26,55 @@
 #include <subdev/acr.h>
 #include <subdev/timer.h>
 
+#include <nvfw/flcn.h>
+
 #include <nvif/class.h>
 
+void
+gm20b_gr_acr_bld_patch(struct nvkm_acr *acr, u32 bld, s64 adjust)
+{
+	struct flcn_bl_dmem_desc hdr;
+	u64 addr;
+
+	nvkm_robj(acr->wpr, bld, &hdr, sizeof(hdr));
+	addr = ((u64)hdr.code_dma_base1 << 40 | hdr.code_dma_base << 8);
+	hdr.code_dma_base  = lower_32_bits((addr + adjust) >> 8);
+	hdr.code_dma_base1 = upper_32_bits((addr + adjust) >> 8);
+	addr = ((u64)hdr.data_dma_base1 << 40 | hdr.data_dma_base << 8);
+	hdr.data_dma_base  = lower_32_bits((addr + adjust) >> 8);
+	hdr.data_dma_base1 = upper_32_bits((addr + adjust) >> 8);
+	nvkm_wobj(acr->wpr, bld, &hdr, sizeof(hdr));
+
+	flcn_bl_dmem_desc_dump(&acr->subdev, &hdr);
+}
+
+void
+gm20b_gr_acr_bld_write(struct nvkm_acr *acr, u32 bld,
+		       struct nvkm_acr_lsfw *lsfw)
+{
+	const u64 base = lsfw->offset.img + lsfw->app_start_offset;
+	const u64 code = (base + lsfw->app_resident_code_offset) >> 8;
+	const u64 data = (base + lsfw->app_resident_data_offset) >> 8;
+	const struct flcn_bl_dmem_desc hdr = {
+		.ctx_dma = FALCON_DMAIDX_UCODE,
+		.code_dma_base = lower_32_bits(code),
+		.non_sec_code_off = lsfw->app_resident_code_offset,
+		.non_sec_code_size = lsfw->app_resident_code_size,
+		.code_entry_point = lsfw->app_imem_entry,
+		.data_dma_base = lower_32_bits(data),
+		.data_size = lsfw->app_resident_data_size,
+		.code_dma_base1 = upper_32_bits(code),
+		.data_dma_base1 = upper_32_bits(data),
+	};
+
+	nvkm_wobj(acr->wpr, bld, &hdr, sizeof(hdr));
+}
+
 const struct nvkm_acr_lsf_func
 gm20b_gr_fecs_acr = {
+	.bld_size = sizeof(struct flcn_bl_dmem_desc),
+	.bld_write = gm20b_gr_acr_bld_write,
+	.bld_patch = gm20b_gr_acr_bld_patch,
 };
 
 static void
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp108.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp108.c
index 1fe58461095a..113e4c1ba9e8 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp108.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp108.c
@@ -23,12 +23,52 @@
 
 #include <subdev/acr.h>
 
+#include <nvfw/flcn.h>
+
+static void
+gp108_gr_acr_bld_patch(struct nvkm_acr *acr, u32 bld, s64 adjust)
+{
+	struct flcn_bl_dmem_desc_v2 hdr;
+	nvkm_robj(acr->wpr, bld, &hdr, sizeof(hdr));
+	hdr.code_dma_base = hdr.code_dma_base + adjust;
+	hdr.data_dma_base = hdr.data_dma_base + adjust;
+	nvkm_wobj(acr->wpr, bld, &hdr, sizeof(hdr));
+	flcn_bl_dmem_desc_v2_dump(&acr->subdev, &hdr);
+}
+
+static void
+gp108_gr_acr_bld_write(struct nvkm_acr *acr, u32 bld,
+		       struct nvkm_acr_lsfw *lsfw)
+{
+	const u64 base = lsfw->offset.img + lsfw->app_start_offset;
+	const u64 code = base + lsfw->app_resident_code_offset;
+	const u64 data = base + lsfw->app_resident_data_offset;
+	const struct flcn_bl_dmem_desc_v2 hdr = {
+		.ctx_dma = FALCON_DMAIDX_UCODE,
+		.code_dma_base = code,
+		.non_sec_code_off = lsfw->app_resident_code_offset,
+		.non_sec_code_size = lsfw->app_resident_code_size,
+		.code_entry_point = lsfw->app_imem_entry,
+		.data_dma_base = data,
+		.data_size = lsfw->app_resident_data_size,
+	};
+
+	nvkm_wobj(acr->wpr, bld, &hdr, sizeof(hdr));
+}
+
 const struct nvkm_acr_lsf_func
 gp108_gr_gpccs_acr = {
+	.flags = NVKM_ACR_LSF_FORCE_PRIV_LOAD,
+	.bld_size = sizeof(struct flcn_bl_dmem_desc_v2),
+	.bld_write = gp108_gr_acr_bld_write,
+	.bld_patch = gp108_gr_acr_bld_patch,
 };
 
 const struct nvkm_acr_lsf_func
 gp108_gr_fecs_acr = {
+	.bld_size = sizeof(struct flcn_bl_dmem_desc_v2),
+	.bld_write = gp108_gr_acr_bld_write,
+	.bld_patch = gp108_gr_acr_bld_patch,
 };
 
 MODULE_FIRMWARE("nvidia/gp108/gr/fecs_bl.bin");
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c
index e22211906b42..a3db2a95ff9a 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c
@@ -27,8 +27,14 @@
 
 #include <nvif/class.h>
 
+#include <nvfw/flcn.h>
+
 static const struct nvkm_acr_lsf_func
 gp10b_gr_gpccs_acr = {
+	.flags = NVKM_ACR_LSF_FORCE_PRIV_LOAD,
+	.bld_size = sizeof(struct flcn_bl_dmem_desc),
+	.bld_write = gm20b_gr_acr_bld_write,
+	.bld_patch = gm20b_gr_acr_bld_patch,
 };
 
 static const struct gf100_gr_func