search for: nvkm_gsp

Hi folks: Here is the leftover of the previous spin of NVKM GSP RPC fixes, which is handling the return of large GSP message. PATCH 1 and 2 in the previous spin were merged [1], and this spin is based on top of PATCH 1 and PATCH 2 in the previous spin. Besides the support of the large GSP message, kernel doc and many cleanups are introduced according to the comments in the previous spin [2].

Store the struct device pointer used to allocate the DMA buffer in the nvkm_gsp_mem object. This allows nvkm_gsp_mem_dtor() to release the buffer without needing the nvkm_gsp. This is needed so that we can retain DMA buffers even after the nvkm_gsp object is deleted. Signed-off-by: Timur Tabi <ttabi at nvidia.com> --- v2: rebased to drm-misc-next .../gpu/drm/nouveau...

Store the struct device pointer used to allocate the DMA buffer in the nvkm_gsp_mem object. This allows nvkm_gsp_mem_dtor() to release the buffer without needing the nvkm_gsp. This is needed so that we can retain DMA buffers even after the nvkm_gsp object is deleted. Signed-off-by: Timur Tabi <ttabi at nvidia.com> v2: added get/put_device calls --- .../gpu/drm/nouve...

Store the struct device pointer used to allocate the DMA buffer in the nvkm_gsp_mem object. This allows nvkm_gsp_mem_dtor() to release the buffer without needing the nvkm_gsp. This is needed so that we can retain DMA buffers even after the nvkm_gsp object is deleted. Signed-off-by: Timur Tabi <ttabi at nvidia.com> --- .../gpu/drm/nouveau/include/nvkm/subdev/gsp.h |...

Store the struct device pointer used to allocate the DMA buffer in the nvkm_gsp_mem object. This allows nvkm_gsp_mem_dtor() to release the buffer without needing the nvkm_gsp. This is needed so that we can retain DMA buffers even after the nvkm_gsp object is deleted. Signed-off-by: Timur Tabi <ttabi at nvidia.com> --- Notes: v2: added get/put_device calls ....

...device) <path> is the PCI ID of the GPU (e.g. 0000:65:00.0). Since LOGPMU is not needed for normal GSP-RM operation, it is only created if debugfs is available. Otherwise, the NV_VGPU_MSG_EVENT_UCODE_LIBOS_PRINT RPCs are ignored. A simple way to test the buffer migration feature is to have nvkm_gsp_init() return an error code. Signed-off-by: Timur Tabi <ttabi at nvidia.com> --- v9: - rebased to drm-misc-next - move nvkm_gsp logging dentries into debugfs struct - add gsp_logging_debugfs_root and always create the root dentry - remove root struct in r535.c, along with mutex code - rena...

...(-) diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h index 6f5d376d8fcc..3fbc57b16a05 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h @@ -211,6 +211,12 @@ struct nvkm_gsp { struct mutex mutex;; struct idr idr; } client_id; + + /* A linked list of registry items. The registry RPC will be built from it. */ + struct list_head registry_list; + + /* The size of the registry RPC */ + size_t registry_rpc_size; }; static inline bool diff --git a/drivers/gpu/drm/...

...u where <path> is the PCI ID of the GPU (e.g. 0000:65:00.0). Since LOGPMU is not needed for normal GSP-RM operation, it is only created if debugfs is available. Otherwise, the NV_VGPU_MSG_EVENT_UCODE_LIBOS_PRINT RPCs are ignored. A simple way to test the buffer migration feature is to have nvkm_gsp_init() return an error code. Signed-off-by: Timur Tabi <ttabi at nvidia.com> v6: rebased onto drm-misc-next renamed nvif_log.head to entry only one space after . added NVIF_LOGS_DECLARE added nvif_log_shutdown documented NV_GSP_MSG_EVENT_UCODE_LIBOS_CLASS_PMU removed the "See xxx"...

On Mon, 2024-07-29 at 18:07 -0500, Timur Tabi wrote: Store the struct device pointer used to allocate the DMA buffer in the nvkm_gsp_mem object. This allows nvkm_gsp_mem_dtor() to release the buffer without needing the nvkm_gsp. This is needed so that we can retain DMA buffers even after the nvkm_gsp object is deleted. Signed-off-by: Timur Tabi <ttabi at nvidia.com<mailto:ttabi at nvidia.com>> v2: added get/...

...(-) diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h index d1437c08645f..5c6b8536e31c 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h @@ -211,6 +211,12 @@ struct nvkm_gsp { struct mutex mutex;; struct idr idr; } client_id; + + /* A linked list of registry items. The registry RPC will be built from it. */ + struct list_head registry_list; + + /* The size of the registry RPC */ + size_t registry_rpc_size; }; static inline bool diff --git a/drivers/gpu/drm/...

...he GPU (e.g. > 0000:65:00.0). > > Since LOGPMU is not needed for normal GSP-RM operation, it is only > created if debugfs is available. Otherwise, the > NV_VGPU_MSG_EVENT_UCODE_LIBOS_PRINT RPCs are ignored. > > A simple way to test the buffer migration feature is to have > nvkm_gsp_init() return an error code. > > Signed-off-by: Timur Tabi <ttabi at nvidia.com> > --- > v8: > Rebased onto drm-next 6.11.0-rc5 > Using /sys/kernel/debug/nouveau instead of debug/dri > Note: this is an extensive change from v7, so please review thoroughly. > >...

...OGPMU is not needed for normal GSP-RM operation, it is only created if debugfs is available. Otherwise, the NV_VGPU_MSG_EVENT_UCODE_LIBOS_PRINT RPCs are ignored. Signed-off-by: Timur Tabi <ttabi at nvidia.com<mailto:ttabi at nvidia.com>> --- v5: rebased to drm-misc-next repaced nvkm_gsp_log with nvif_log minor rearrangement of some code drivers/gpu/drm/nouveau/include/nvif/log.h | 32 ++ .../gpu/drm/nouveau/include/nvkm/subdev/gsp.h | 13 + drivers/gpu/drm/nouveau/nouveau_drm.c | 19 + .../gpu/drm/nouveau/nvkm/subdev/gsp/r535.c | 360 +++++++++++++++++- 4...

...GPU (e.g. 0000:65:00.0)." > > Since LOGPMU is not needed for normal GSP-RM operation, it is only > created if debugfs is available. Otherwise, the > NV_VGPU_MSG_EVENT_UCODE_LIBOS_PRINT RPCs are ignored. > > A simple way to test the buffer migration feature is to have > nvkm_gsp_init() return an error code. > > Signed-off-by: Timur Tabi <ttabi at nvidia.com> > > v6: > rebased onto drm-misc-next > renamed nvif_log.head to entry > only one space after . > added NVIF_LOGS_DECLARE > added nvif_log_shutdown > documented NV_GSP_MSG_EVENT_UCO...

...the following Smatch static checker warning: drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c:1016 r535_gsp_rpc_unloading_guest_driver() warn: 'rpc' isn't an ERR_PTR drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c 1010 static int 1011 r535_gsp_rpc_unloading_guest_driver(struct nvkm_gsp *gsp, bool suspend) 1012 { 1013 rpc_unloading_guest_driver_v1F_07 *rpc; 1014 1015 rpc = nvkm_gsp_rpc_get(gsp, NV_VGPU_MSG_FUNCTION_UNLOADING_GUEST_DRIVER, sizeof(*rpc)); nvkm_gsp_rpc_get() returns NULL on error. --> 1016 if (IS_ERR(rpc)) 1017...

...warning: > > drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c:1016 r535_gsp_rpc_unloading_guest_driver() > warn: 'rpc' isn't an ERR_PTR > > drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c > 1010 static int > 1011 r535_gsp_rpc_unloading_guest_driver(struct nvkm_gsp *gsp, bool suspend) > 1012 { > 1013 rpc_unloading_guest_driver_v1F_07 *rpc; > 1014 > 1015 rpc = nvkm_gsp_rpc_get(gsp, NV_VGPU_MSG_FUNCTION_UNLOADING_GUEST_DRIVER, sizeof(*rpc)); > > nvkm_gsp_rpc_get() returns NULL on error. There are also code...

...diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/ga102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/ga102.c index 525267412c3e..a3996ceca995 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/ga102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/ga102.c @@ -45,7 +45,7 @@ ga102_gsp_nofw(struct nvkm_gsp *gsp, int ver, const struct nvkm_gsp_fwif *fwif) return 0; } -struct nvkm_gsp_fwif +static struct nvkm_gsp_fwif ga102_gsps[] = { { -1, ga102_gsp_nofw, &ga102_gsp }, {} -- 2.27.0

...print(void *priv, u32 fn, void *repv, u32 repc) +{ + /* work out what we should do here. */ + return 0; +} + +static int +r535_gsp_msg_gsp_send_user_shared_data(void *priv, u32 fn, void *repv, u32 repc) +{ + /* this seems to send some sort of assert counts from gsp */ + return 0; +} + static void nvkm_gsp_mem_dtor(struct nvkm_gsp *gsp, struct nvkm_gsp_mem *mem) { @@ -2104,7 +2124,9 @@ r535_gsp_oneinit(struct nvkm_gsp *gsp) r535_gsp_msg_ntfy_add(gsp, NV_VGPU_MSG_EVENT_MMU_FAULT_QUEUED, r535_gsp_msg_mmu_fault_queued, gsp); r535_gsp_msg_ntfy_add(gsp, NV_VGPU_MSG_EVENT_OS_ERROR_LOG, r535_...

...diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c index f6725a5f5bfb..6c0a8fbf0061 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c @@ -298,7 +298,8 @@ r535_gsp_msg_recv(struct nvkm_gsp *gsp, int fn, u32 repc) struct nvkm_gsp_msgq_ntfy *ntfy = &gsp->msgq.ntfy[i]; if (ntfy->fn == msg->function) { - ntfy->func(ntfy->priv, ntfy->fn, msg->data, msg->length - sizeof(*msg)); + if (ntfy->func) + ntfy->func(ntfy->priv, ntfy->fn, msg-&...

...a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c index 9ee58e2a0eb2..5e1fa176aac4 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c @@ -1078,7 +1078,6 @@ r535_gsp_rpc_set_registry(struct nvkm_gsp *gsp) if (IS_ERR(rpc)) return PTR_ERR(rpc); - rpc->size = sizeof(*rpc); rpc->numEntries = NV_GSP_REG_NUM_ENTRIES; str_offset = offsetof(typeof(*rpc), entries[NV_GSP_REG_NUM_ENTRIES]); @@ -1094,6 +1093,7 @@ r535_gsp_rpc_set_registry(struct nvkm_gsp *gsp) strings += name_len;...

...n(+), 1 deletion(-) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/fwsec.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/fwsec.c index 330d72b1a4af..52412965fac1 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/fwsec.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/fwsec.c @@ -324,7 +324,7 @@ nvkm_gsp_fwsec_sb(struct nvkm_gsp *gsp) return ret; /* Verify. */ - err = nvkm_rd32(device, 0x001400 + (0xf * 4)) & 0x0000ffff; + err = nvkm_rd32(device, 0x001400 + (0x15 * 4)) & 0x0000ffff; if (err) { nvkm_error(subdev, "fwsec-sb: 0x%04x\n", err); return -EIO; -- 2.45.2