Open sourcing the checkpoint/restore driver API

[mattnappo]

The CUDA Checkpoint API has been present since driver version 550, and is continually updated. Is there any plan to open source this part of the driver? It would be helpful to know how checkpoint/restore is implemented.

Here's an example trace of the ioctls when running cuda-checkpoint --get-state

NV_ESC_CARD_INFO
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869758960, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 794326196, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_FREE
NV_ESC_CARD_INFO
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869760432, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 808424609, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869760432, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 808856178, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_FREE
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_REGISTER_FD fd: -425214640
NV_ESC_NUMA_INFO
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869752864, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 837164546, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_REGISTER_FD fd: -425214640
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869752928, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 854112037, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869752944, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 854643727, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_REGISTER_FD fd: -425214976
NV_ESC_ALLOC_OS_EVENT
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869752288, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 875690410, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_MAP_MEMORY: NVOS33_PARAMETERS { hClient: 3869752176, hDevice: 32496, hMemory: 4270797800, offset: 877088658, length: 2170, pLinearAddress: 0x1000100, status: 385673312, flags: 21938 }
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869752304, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 877618359, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869752432, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 878317543, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_MAP_MEMORY: NVOS33_PARAMETERS { hClient: 3869752256, hDevice: 32496, hMemory: 4270797800, offset: 878725661, length: 2170, pLinearAddress: 0x1000100, status: 385673312, flags: 21938 }
NV_ESC_RM_CONTROL
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869752416, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 881781093, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869752416, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 882663200, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_CONTROL
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869759840, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 897042979, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_ALLOC: NVOS02_PARAMETERS { hRoot: 3869760416, hObjectParent: 32496, hObjectNew: 4270797800, hClass: 442149, flags: 898108370, pMemory: 0x87a, limit: 16777472, status: 385673312 }
NV_ESC_RM_CONTROL