570.195.03

Author: mmaneetsingh

README.md

@@ -1,7 +1,7 @@
 # NVIDIA Linux Open GPU Kernel Module Source
 
 This is the source release of the NVIDIA Linux open GPU kernel modules,
-version 570.190.
+version 570.195.03.
 
 
 ## How to Build
@@ -17,7 +17,7 @@ as root:
 
 Note that the kernel modules built here must be used with GSP
 firmware and user-space NVIDIA GPU driver components from a corresponding
-570.190 driver release.  This can be achieved by installing
+570.195.03 driver release.  This can be achieved by installing
 the NVIDIA GPU driver from the .run file using the `--no-kernel-modules`
 option.  E.g.,
 
@@ -185,7 +185,7 @@ table below).
 For details on feature support and limitations, see the NVIDIA GPU driver
 end user README here:
 
-https://us.download.nvidia.com/XFree86/Linux-x86_64/570.190/README/kernel_open.html
+https://us.download.nvidia.com/XFree86/Linux-x86_64/570.195.03/README/kernel_open.html
 
 For vGPU support, please refer to the README.vgpu packaged in the vGPU Host
 Package for more details.

kernel-open/Kbuild

@@ -79,7 +79,7 @@ ccflags-y += -I$(src)/common/inc
 ccflags-y += -I$(src)
 ccflags-y += -Wall $(DEFINES) $(INCLUDES) -Wno-cast-qual -Wno-format-extra-args
 ccflags-y += -D__KERNEL__ -DMODULE -DNVRM
-ccflags-y += -DNV_VERSION_STRING=\"570.190\"
+ccflags-y += -DNV_VERSION_STRING=\"570.195.03\"
 
 ifneq ($(SYSSRCHOST1X),)
  ccflags-y += -I$(SYSSRCHOST1X)

kernel-open/nvidia/nv.c

@@ -2423,6 +2423,12 @@ nvidia_ioctl(
     {
         nv_ioctl_wait_open_complete_t *params = arg_copy;
 
+        if (arg_size != sizeof(nv_ioctl_wait_open_complete_t))
+        {
+            status = -EINVAL;
+            goto done_early;
+        }
+
         params->rc = nvlfp->open_rc;
         params->adapterStatus = nvlfp->adapter_status;
         goto done_early;
@@ -2503,21 +2509,28 @@ nvidia_ioctl(
                 goto done;
             }
 
+            /* atomically check and alloc attached_gpus */
+            down(&nvl->ldata_lock);
+
             if (nvlfp->num_attached_gpus != 0)
             {
+                up(&nvl->ldata_lock);
                 status = -EINVAL;
                 goto done;
             }
 
             NV_KMALLOC(nvlfp->attached_gpus, arg_size);
             if (nvlfp->attached_gpus == NULL)
             {
+                up(&nvl->ldata_lock);
                 status = -ENOMEM;
                 goto done;
             }
             memcpy(nvlfp->attached_gpus, arg_copy, arg_size);
             nvlfp->num_attached_gpus = num_arg_gpus;
 
+            up(&nvl->ldata_lock);
+
             for (i = 0; i < nvlfp->num_attached_gpus; i++)
             {
                 if (nvlfp->attached_gpus[i] == 0)
@@ -2533,9 +2546,14 @@ nvidia_ioctl(
                             nvidia_dev_put(nvlfp->attached_gpus[i], sp);
                     }
 
+                    /* atomically free attached_gpus */
+                    down(&nvl->ldata_lock);
+
                     NV_KFREE(nvlfp->attached_gpus, arg_size);
                     nvlfp->num_attached_gpus = 0;
 
+                    up(&nvl->ldata_lock);
+
                     status = -EINVAL;
                     break;
                 }

src/common/inc/nvBldVer.h

@@ -36,25 +36,25 @@
 // and then checked back in. You cannot make changes to these sections without
 // corresponding changes to the buildmeister script
 #ifndef NV_BUILD_BRANCH
-    #define NV_BUILD_BRANCH             r570_00
+    #define NV_BUILD_BRANCH             r573_76
 #endif
 #ifndef NV_PUBLIC_BRANCH
-    #define NV_PUBLIC_BRANCH             r570_00
+    #define NV_PUBLIC_BRANCH             r573_76
 #endif
 
 #if defined(NV_LINUX) || defined(NV_BSD) || defined(NV_SUNOS)
-#define NV_BUILD_BRANCH_VERSION         "rel/gpu_drv/r570/r570_00-575"
-#define NV_BUILD_CHANGELIST_NUM         (36467544)
+#define NV_BUILD_BRANCH_VERSION         "rel/gpu_drv/r570/r573_76-590"
+#define NV_BUILD_CHANGELIST_NUM         (36569223)
 #define NV_BUILD_TYPE                   "Official"
-#define NV_BUILD_NAME                   "rel/gpu_drv/r570/r570_00-575"
-#define NV_LAST_OFFICIAL_CHANGELIST_NUM (36467544)
+#define NV_BUILD_NAME                   "rel/gpu_drv/r570/r573_76-590"
+#define NV_LAST_OFFICIAL_CHANGELIST_NUM (36569223)
 
 #else     /* Windows builds */
-#define NV_BUILD_BRANCH_VERSION         "r570_00-569"
-#define NV_BUILD_CHANGELIST_NUM         (36467544)
-#define NV_BUILD_TYPE                   "Official"
-#define NV_BUILD_NAME                   "573.73"
-#define NV_LAST_OFFICIAL_CHANGELIST_NUM (36467544)
+#define NV_BUILD_BRANCH_VERSION         "r573_76-1"
+#define NV_BUILD_CHANGELIST_NUM         (36518415)
+#define NV_BUILD_TYPE                   "Nightly"
+#define NV_BUILD_NAME                   "r573_76-250909"
+#define NV_LAST_OFFICIAL_CHANGELIST_NUM (36506718)
 #define NV_BUILD_BRANCH_BASE_VERSION    R570
 #endif
 // End buildmeister python edited section

src/common/inc/nvUnixVersion.h

@@ -4,7 +4,7 @@
 #if defined(NV_LINUX) || defined(NV_BSD) || defined(NV_SUNOS) || defined(NV_VMWARE) || defined(NV_QNX) || defined(NV_INTEGRITY) || \
     (defined(RMCFG_FEATURE_PLATFORM_GSP) && RMCFG_FEATURE_PLATFORM_GSP == 1)
 
-#define NV_VERSION_STRING               "570.190"
+#define NV_VERSION_STRING               "570.195.03"
 
 #else
 

src/common/sdk/nvidia/inc/nverror.h

@@ -109,7 +109,7 @@
 #define ROBUST_CHANNEL_NVJPG5_ERROR                     (103)
 #define ROBUST_CHANNEL_NVJPG6_ERROR                     (104)
 #define ROBUST_CHANNEL_NVJPG7_ERROR                     (105)
-#define DESTINATION_FLA_TRANSLATION_ERROR               (108)
+#define NVLINK_REMOTE_TRANSLATION_ERROR                 (108)
 #define SEC_FAULT_ERROR                                 (110)
 #define GSP_RPC_TIMEOUT                                 (119)
 #define GSP_ERROR                                       (120)
@@ -129,7 +129,7 @@
 #define ROBUST_CHANNEL_CE18_ERROR                       (134)
 #define ROBUST_CHANNEL_CE19_ERROR                       (135)
 #define ALI_TRAINING_FAIL                               (136)
-#define NVLINK_FLA_PRIV_ERR                             (137)
+#define NVLINK_PRIV_ERR                                 (137)
 #define ROBUST_CHANNEL_DLA_ERROR                        (138)
 #define ROBUST_CHANNEL_OFA1_ERROR                       (139)
 #define UNRECOVERABLE_ECC_ERROR_ESCAPE                  (140)

src/nvidia/generated/g_kern_gmmu_nvoc.h

@@ -500,7 +500,6 @@ struct KernelGmmu {
     NvBool PDB_PROP_KGMMU_REDUCE_NR_FAULT_BUFFER_SIZE;
 
     // Data members
-    NvBool bReportFlaTranslationXid;
     MEMORY_DESCRIPTOR *pFakeSparseBuffer;
     NvU64 fakeSparseEntry[3];
     NV2080_CTRL_INTERNAL_GMMU_GET_STATIC_INFO_PARAMS *pStaticInfo;
@@ -636,7 +635,6 @@ struct KernelGmmu_PRIVATE {
     NvBool PDB_PROP_KGMMU_REDUCE_NR_FAULT_BUFFER_SIZE;
 
     // Data members
-    NvBool bReportFlaTranslationXid;
     MEMORY_DESCRIPTOR *pFakeSparseBuffer;
     NvU64 fakeSparseEntry[3];
     NV2080_CTRL_INTERNAL_GMMU_GET_STATIC_INFO_PARAMS *pStaticInfo;

src/nvidia/inc/kernel/core/thread_state.h

@@ -76,6 +76,7 @@ struct THREAD_STATE_NODE
      */
     NvU32                threadSeqId;
     NvBool               bValid;
+    NvBool               bUsingHeap;
     THREAD_TIMEOUT_STATE timeout;
     NvU32                cpuNum;
     NvU32                flags;
@@ -208,6 +209,7 @@ void        threadStateOnlyProcessWorkISRAndDeferredIntHandler(THREAD_STATE_NODE
 void        threadStateOnlyFreeISRAndDeferredIntHandler(THREAD_STATE_NODE *, OBJGPU*, NvU32);
 void        threadStateFreeISRAndDeferredIntHandler(THREAD_STATE_NODE *, OBJGPU*, NvU32);
 void        threadStateInit(THREAD_STATE_NODE *pThreadNode, NvU32 flags);
+THREAD_STATE_NODE* threadStateAlloc(NvU32 flags);
 void        threadStateFree(THREAD_STATE_NODE *pThreadNode, NvU32 flags);
 
 NV_STATUS   threadStateGetCurrent(THREAD_STATE_NODE **ppThreadNode, OBJGPU *pGpu);

src/nvidia/inc/kernel/gpu/rpc/objrpc.h

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2004-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-FileCopyrightText: Copyright (c) 2004-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
  * SPDX-License-Identifier: MIT
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -50,8 +50,8 @@ TYPEDEF_BITVECTOR(MC_ENGINE_BITVECTOR);
 #include "g_rpc_hal.h" // For RPC_HAL_IFACES
 #include "g_rpc_odb.h" // For RPC_HAL_IFACES
 
-#define RPC_TIMEOUT_LIMIT_PRINT_RATE_THRESH 3  // rate limit after 3 prints
-#define RPC_TIMEOUT_LIMIT_PRINT_RATE_SKIP   29 // skip 29 of 30 prints
+#define RPC_TIMEOUT_GPU_RESET_THRESHOLD 3  // Reset GPU after 3 back to back GSP RPC timeout
+#define RPC_TIMEOUT_PRINT_RATE_SKIP   29 // skip 29 of 30 prints
 
 #define RPC_HISTORY_DEPTH 128
 

src/nvidia/inc/kernel/mem_mgr/ctx_buf_pool.h

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2016-2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-FileCopyrightText: Copyright (c) 2016-2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
  * SPDX-License-Identifier: MIT
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -44,10 +44,9 @@ struct CTX_BUF_POOL_INFO
 {
     //
     // Each array index corresponds to a pointer to memory pool with
-    // page size corresponding to RM_ATTR_PAGE_SIZE_*
-    // Pool corresponding to RM_ATTR_PAGE_SIZE_DEFAULT will be left unused
+    // page size corresponding to POOL_CONFIG_MODE
     //
-    RM_POOL_ALLOC_MEM_RESERVE_INFO *pMemPool[RM_ATTR_PAGE_SIZE_INVALID];
+    RM_POOL_ALLOC_MEM_RESERVE_INFO *pMemPool[POOL_CONFIG_MAX_SUPPORTED];
 };
 
 // List of all context buffers supported by memory pools