Device State Inspection

Enumerations

CUDBGBarrierScope: CUDA barrier scope.
CUDBGCbuThreadState: Thread state in the CBU (Convergence Barrier Unit).
CUDBGCoredumpGenerationFlags: Coredump generation flags.
CUDBGCudaLogLevel: CUDA Log severity level.
CUDBGCudaLogLevelFilter: CUDA Log level filter.
CUDBGCudaLogRuleAction: CUDA Log filtering rule action.
CUDBGDeviceInfoAttribute_t: Device-level attributes.
CUDBGDeviceInfoQueryType_t: Device info query type.
CUDBGLaneInfoAttribute_t: Lane-level (thread-level) attributes.
CUDBGSMInfoAttribute_t: SM-level attributes.
CUDBGTcgen05TmemAllocationPhase: Tensor Memory allocation phase.
CUDBGWarpInfoAttribute_t: Warp-level attributes.
ptxStorageKind: Memory segments for DWARF.

Structs

CUDBGCbuWarpState: Warp state in the CBU (Convergence Barrier Unit).
CUDBGCudaLogMessage: CUDA Log Message.
CUDBGCudaLogMessage129: CUDA Log Message for API version 12.9.
CUDBGCudaLogRule: CUDA Log filtering rule.
CUDBGDeviceInfo: Device-level information.
CUDBGDeviceInfoSizes: Sizes of the various structs returned by the batched device update APIs.
CUDBGLaneState: Lane state (state of a single thread).
CUDBGLoadedFunctionInfo: Information about a lazily loaded function.
CUDBGMemoryInfo: Memory information.
CUDBGSMInfo: SM-level information.
CUDBGTcgen05TmemInfo: tcgen05 Tensor Memory information for a warp's CTA.
CUDBGWarpInfo: Warp-level information.
CUDBGWarpResources: Warp resources.
CUDBGWarpState: Warp state information.
CUDBGWarpState120: Warp state for API version 12.0.
CUDBGWarpState127: Warp state for API version 12.7.
CUDBGWarpState60: Warp state for API version 6.0.

Variables

CUDBGResult(* CUDBGAPI_st::consumeCudaLogs: Get CUDA error log entries.
CUDBGResult(* CUDBGAPI_st::consumeCudaLogs129: Get CUDA error log entries.
CUDBGResult(* CUDBGAPI_st::generateCoredump: Generate a coredump for the current GPU state.
CUDBGResult(* CUDBGAPI_st::getBindlessConstAddress: Get bindless constant GPU VA and size from its header.
CUDBGResult(* CUDBGAPI_st::getCbuWarpState: Get the CBU state of a given warp.
CUDBGResult(* CUDBGAPI_st::getClusterExceptionTargetBlock: Get the target block index and validity status for cluster exceptions.
CUDBGResult(* CUDBGAPI_st::getConstBankAddress: Get constant bank GPU VA and size.
CUDBGResult(* CUDBGAPI_st::getConstBankAddress123: Convert constant bank number and offset into GPU VA.
CUDBGResult(* CUDBGAPI_st::getCudaExceptionString: Get the error string for CUDA Exceptions.
CUDBGResult(* CUDBGAPI_st::getDeviceInfo: Read full device info for the device.
CUDBGResult(* CUDBGAPI_st::getDeviceInfoSizes: Return sizes for device info structs and defined attributes.
CUDBGResult(* CUDBGAPI_st::getDevicePCIBusInfo: Get PCI bus and device ids associated with device index.
CUDBGResult(* CUDBGAPI_st::getHardwareBarrierInfo: Get hardware barrier information.
CUDBGResult(* CUDBGAPI_st::getLoadedFunctionInfo: Get the section number and address of loaded functions for a given module.
CUDBGResult(* CUDBGAPI_st::getLoadedFunctionInfo118: Get the section number and address of loaded functions for a given module.
CUDBGResult(* CUDBGAPI_st::getManagedMemoryRegionInfo: Get a sorted list of managed memory regions.
CUDBGResult(* CUDBGAPI_st::memcheckReadErrorAddress: Get the address that memcheck detected an error on.
CUDBGResult(* CUDBGAPI_st::readActiveLanes: Read the lane bitmask of active threads on a valid warp.
CUDBGResult(* CUDBGAPI_st::readAllVirtualReturnAddresses: Read all the virtual return addresses for a thread (the full backtrace).
CUDBGResult(* CUDBGAPI_st::readBlockIdx: Read the CUDA block index running on a valid warp.
CUDBGResult(* CUDBGAPI_st::readBlockIdx32: Read the two-dimensional CUDA block index running on a valid warp.
CUDBGResult(* CUDBGAPI_st::readBrokenWarps: Read the bitmask of warps that are at a breakpoint on a given SM.
CUDBGResult(* CUDBGAPI_st::readCCRegister: Read the hardware CC register.
CUDBGResult(* CUDBGAPI_st::readCPUCallStack: Read the CPU call stack captured at the time of kernel launch.
CUDBGResult(* CUDBGAPI_st::readCallDepth: Read the call depth (number of calls) for a given thread.
CUDBGResult(* CUDBGAPI_st::readCallDepth32: Read the call depth (number of calls) for a given warp.
CUDBGResult(* CUDBGAPI_st::readClusterIdx: Read the CUDA cluster index running on a valid warp.
CUDBGResult(* CUDBGAPI_st::readCodeMemory: Read content at address in the code memory segment.
CUDBGResult(* CUDBGAPI_st::readConstMemory129: Read content at address in the constant memory segment.
CUDBGResult(* CUDBGAPI_st::readDeviceExceptionState: Get the exception state of the SMs on the device.
CUDBGResult(* CUDBGAPI_st::readDeviceExceptionState80: Get the exception state of the SMs on the device.
CUDBGResult(* CUDBGAPI_st::readErrorPC: Get the hardware reported error PC if it exists.
CUDBGResult(* CUDBGAPI_st::readGenericMemory: Read content at an address in any memory segment.
CUDBGResult(* CUDBGAPI_st::readGlobalMemory: Read content at an address in the global address space.
CUDBGResult(* CUDBGAPI_st::readGlobalMemory31: Read content at address in the global memory segment.
CUDBGResult(* CUDBGAPI_st::readGlobalMemory55: Read content at address in the global memory segment.
CUDBGResult(* CUDBGAPI_st::readGridId: Read the 64-bit CUDA grid index running on a valid warp.
CUDBGResult(* CUDBGAPI_st::readGridId50: Read the CUDA grid index running on a valid warp.
CUDBGResult(* CUDBGAPI_st::readLaneException: Read the exception type for a given thread.
CUDBGResult(* CUDBGAPI_st::readLaneStatus: Read the status of the given thread.
CUDBGResult(* CUDBGAPI_st::readLocalMemory: Read content at address in the local memory segment.
CUDBGResult(* CUDBGAPI_st::readPC: Read the PC offset on the given active thread.
CUDBGResult(* CUDBGAPI_st::readParamMemory: Read content at address in the param memory segment.
CUDBGResult(* CUDBGAPI_st::readPinnedMemory: Read content at pinned address in system memory.
CUDBGResult(* CUDBGAPI_st::readPredicates: Read content of hardware predicate registers.
CUDBGResult(* CUDBGAPI_st::readRegister: Read content of a hardware register.
CUDBGResult(* CUDBGAPI_st::readRegisterRange: Read content of a hardware range of hardware registers.
CUDBGResult(* CUDBGAPI_st::readRegisterRange60: Read content of a range of hardware registers.
CUDBGResult(* CUDBGAPI_st::readReturnAddress: Read the return address (offset) for a call level.
CUDBGResult(* CUDBGAPI_st::readReturnAddress32: Read the return address (offset) for a call level.
CUDBGResult(* CUDBGAPI_st::readRpcRegisters: Read the RPC.LO and/or RPC.HI hardware registers for a specific thread.
CUDBGResult(* CUDBGAPI_st::readSharedMemory: Read content at address in the shared memory segment.
CUDBGResult(* CUDBGAPI_st::readSmException: Get the SM exception status if it exists.
CUDBGResult(* CUDBGAPI_st::readSyscallCallDepth: Read the call depth of syscalls for a given thread.
CUDBGResult(* CUDBGAPI_st::readTcgen05TmemInfo: Read tcgen05 Tensor Memory metadata for the CTA containing a warp.
CUDBGResult(* CUDBGAPI_st::readTcgen05TmemMemory: Read tcgen05 Tensor Memory data for the CTA containing a warp.
CUDBGResult(* CUDBGAPI_st::readTextureMemory: This method is no longer supported since CUDA 12.0.
CUDBGResult(* CUDBGAPI_st::readTextureMemoryBindless: This method is no longer supported since CUDA 12.0.
CUDBGResult(* CUDBGAPI_st::readThreadIdx: Read the CUDA thread index running on valid thread.
CUDBGResult(* CUDBGAPI_st::readUniformPredicates: Read contents of uniform predicate registers.
CUDBGResult(* CUDBGAPI_st::readUniformRegisterRange: Read a range of uniform registers.
CUDBGResult(* CUDBGAPI_st::readValidLanes: Read the lane bitmask of valid threads on a given warp.
CUDBGResult(* CUDBGAPI_st::readValidWarps: Read the bitmask of valid warps on a given SM.
CUDBGResult(* CUDBGAPI_st::readVirtualPC: Read the PC (virtual address) on the given active thread.
CUDBGResult(* CUDBGAPI_st::readVirtualReturnAddress: Read the virtual return address for a call level.
CUDBGResult(* CUDBGAPI_st::readVirtualReturnAddress32: Read the virtual return address for a call level.
CUDBGResult(* CUDBGAPI_st::readWarpResources: Get the resources assigned to a given warp.
CUDBGResult(* CUDBGAPI_st::readWarpState: Read the state of a given warp.
CUDBGResult(* CUDBGAPI_st::readWarpState120: Read the state of a given warp.
CUDBGResult(* CUDBGAPI_st::readWarpState127: Read the state of a given warp.
CUDBGResult(* CUDBGAPI_st::readWarpState60: Read the state of a given warp.
CUDBGResult(* CUDBGAPI_st::setCudaLogRules: Set CUDA log filtering rules.

Enumerations

enum CUDBGBarrierScope

CUDA barrier scope.

Values:

enumerator CUDBG_BARRIER_SCOPE_INVALID = 0xFFFFFFFFU: Invalid barrier scope.

enumerator CUDBG_BARRIER_SCOPE_NONE = 0: No barrier.

enumerator CUDBG_BARRIER_SCOPE_WARP = 1: Warp-wide barrier.

enumerator CUDBG_BARRIER_SCOPE_WARP_GROUP = 2: Warp group-wide barrier.

enumerator CUDBG_BARRIER_SCOPE_BLOCK = 3: Block-wide barrier.

enumerator CUDBG_BARRIER_SCOPE_CLUSTER = 4: Cluster-wide barrier.

enumerator CUDBG_BARRIER_SCOPE_KERNEL = 5: Kernel-wide barrier.

enum CUDBGCbuThreadState

Thread state in the CBU (Convergence Barrier Unit).

Values:

enumerator CUDBG_CBU_THREAD_STATE_INVALID = 0xFFFFFFFFU

enumerator CUDBG_CBU_THREAD_STATE_EXITED = 0

enumerator CUDBG_CBU_THREAD_STATE_READY

enumerator CUDBG_CBU_THREAD_STATE_YIELDED

enumerator CUDBG_CBU_THREAD_STATE_SLEEP

enumerator CUDBG_CBU_THREAD_STATE_SLEEPYIELD

enumerator CUDBG_CBU_THREAD_STATE_READYATNEXT

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDPLUS

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDALL

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDCOLLECTIVE

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB0

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB1

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB2

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB3

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB4

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB5

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB6

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB7

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB8

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB9

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB10

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB11

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB12

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB13

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB14

enumerator CUDBG_CBU_THREAD_STATE_BLOCKEDB15

enum CUDBGCoredumpGenerationFlags

Coredump generation flags.

Values:

enumerator CUDBG_COREDUMP_DEFAULT_FLAGS = 0: Default flags.

enumerator CUDBG_COREDUMP_SKIP_NONRELOCATED_ELF_IMAGES = (1 << 0): Skip dumping non-relocated ELF images.

enumerator CUDBG_COREDUMP_SKIP_GLOBAL_MEMORY = (1 << 1): Skip dumping global memory.

enumerator CUDBG_COREDUMP_SKIP_SHARED_MEMORY = (1 << 2): Skip dumping shared memory.

enumerator CUDBG_COREDUMP_SKIP_LOCAL_MEMORY = (1 << 3): Skip dumping local memory.

enumerator CUDBG_COREDUMP_SKIP_CONSTBANK_MEMORY = (1 << 5): Skip dumping constant bank memory.

enumerator CUDBG_COREDUMP_GZIP_COMPRESS = (1 << 6): Compress the coredump with gzip.

enumerator CUDBG_COREDUMP_FAULTED_CONTEXTS_ONLY = (1 << 7)

Only include data for contexts which have encountered an exception.

If this flag is used and there are no faulted contexts on any device, then generateCoredump() method will return CUDBG_ERROR_INVALID_CONTEXT error code. Contexts that have warps at breakpoints count as faulted.

enumerator CUDBG_COREDUMP_SKIP_TENSOR_MEMORY = (1 << 8): Skip dumping tensor memory.

enumerator CUDBG_COREDUMP_LIGHTWEIGHT_FLAGS = CUDBG_COREDUMP_SKIP_NONRELOCATED_ELF_IMAGES | CUDBG_COREDUMP_SKIP_GLOBAL_MEMORY | CUDBG_COREDUMP_SKIP_SHARED_MEMORY | CUDBG_COREDUMP_SKIP_LOCAL_MEMORY | CUDBG_COREDUMP_SKIP_CONSTBANK_MEMORY | CUDBG_COREDUMP_SKIP_TENSOR_MEMORY : Lightweight flags.

enum CUDBGCudaLogLevel

CUDA Log severity level.

Values:

enumerator CUDBG_CUDA_LOG_LEVEL_INVALID = 0xFFFFFFFFU: Invalid log level.

enumerator CUDBG_CUDA_LOG_LEVEL_ERROR = 0: Error log level, matches CU_LOG_LEVEL_ERROR.

enumerator CUDBG_CUDA_LOG_LEVEL_WARNING = 1: Warning log level, matches CU_LOG_LEVEL_WARNING.

enum CUDBGCudaLogLevelFilter

CUDA Log level filter.

Values:

enumerator CUDBG_CUDA_LOG_LEVEL_FILTER_ANY = 0xFFFFFFFFU: Any log level.

enumerator CUDBG_CUDA_LOG_LEVEL_FILTER_ERROR = 0: Error log level.

enumerator CUDBG_CUDA_LOG_LEVEL_FILTER_WARNING = 1: Warning log level.

enum CUDBGCudaLogRuleAction

CUDA Log filtering rule action.

Values:

enumerator CUDBG_CUDA_LOG_RULE_ACTION_EXCLUDE = 0xFFFFFFFFU: Exclude the log message.

enumerator CUDBG_CUDA_LOG_RULE_ACTION_SEND_ASYNC = 0: Send the log message asynchronously.

enumerator CUDBG_CUDA_LOG_RULE_ACTION_SEND_SYNC = 1: Send the log message synchronously.

enum CUDBGDeviceInfoAttribute_t

Device-level attributes.

Values:

enumerator CUDBG_DEVICE_ATTRIBUTE_SM_UPDATE_MASK = 0

Mask of updated SMs reported by this response.

Optional: Yes, assume all 1’s if absent. Size: Number of SMs-sized bitmask, rounded up to be divisible by 8.

enumerator CUDBG_DEVICE_ATTRIBUTE_SM_ACTIVE_MASK = 1

Mask of SMs with any valid warp.

Optional: No, always returned by the API. Size: Number of SMs-sized bitmask, rounded up to be divisible by 8.

enumerator CUDBG_DEVICE_ATTRIBUTE_SM_EXCEPTION_MASK = 2

Mask of SMs with any warps with exceptions.

Optional: Yes, assume all 0’s if absent. Size: Number of SMs-sized bitmask, rounded up to be divisible by 8.

enumerator CUDBG_DEVICE_ATTRIBUTE_COUNT: Device attributes count.

enum CUDBGDeviceInfoQueryType_t

Device info query type.

Values:

enumerator CUDBG_RESPONSE_TYPE_FULL: Request state information for all valid SMs/Warps/Lanes.

enumerator CUDBG_RESPONSE_TYPE_UPDATE

Request state information for all changed SMs/Warps/Lanes since the last call.

It’s safe to always use this type, the API will respond with the full information when necessary.

enumerator CUDBG_RESPONSE_TYPE_UNKNOWN = 0xFFFFFFFFU: Unknown device info query type.

enum CUDBGLaneInfoAttribute_t

Lane-level (thread-level) attributes.

Values:

enumerator CUDBG_LANE_ATTRIBUTE_COUNT: Lane (thread) attributes count.

enum CUDBGSMInfoAttribute_t

SM-level attributes.

Values:

enumerator CUDBG_SM_ATTRIBUTE_WARP_UPDATE_MASK = 0

Mask of updated warps reported by this response.

Optional: Yes, assume all 1’s if absent. Size: uint64_t.

enumerator CUDBG_SM_ATTRIBUTE_COUNT: SM attributes count.

enum CUDBGTcgen05TmemAllocationPhase

Tensor Memory allocation phase.

Values:

enumerator CUDBG_TCGEN05_TMEM_ALLOCATION_PHASE_UNKNOWN = 0: Tensor Memory allocation phase is unknown to this API version.

enumerator CUDBG_TCGEN05_TMEM_ALLOCATION_PHASE_ALLOC_PERMIT_AVAILABLE = 1: Tensor Memory allocation permit is available.

enumerator CUDBG_TCGEN05_TMEM_ALLOCATION_PHASE_ALLOC_PERMIT_RELINQUISHED = 2: Tensor Memory allocation permit has been relinquished.

enumerator CUDBG_TCGEN05_TMEM_ALLOCATION_PHASE_EXCLUSIVE_ALLOC_ACTIVE = 3: Tensor Memory allocation permit has been claimed by an exclusive allocation.

enum CUDBGWarpInfoAttribute_t

Warp-level attributes.

Values:

enumerator CUDBG_WARP_ATTRIBUTE_LANE_UPDATE_MASK = 0

Mask of updated lanes reported by this response.

Optional: Yes, assume all 1’s if absent. Size: uint32_t.

enumerator CUDBG_WARP_ATTRIBUTE_LANE_ATTRIBUTES = 1

Signals whether the attribute flags field is present on the lane level for this warp.

Optional: Yes, assume no lane attributes for this warp if absent. Size: 0 (doesn’t have an associated warp-level field).

enumerator CUDBG_WARP_ATTRIBUTE_EXCEPTION = 2

CUDBGException_t for this warp.

Optional: Yes, assume CUDBG_EXCEPTION_NONE if absent. Size: uint32_t.

enumerator CUDBG_WARP_ATTRIBUTE_ERRORPC = 3

Error PC for this warp.

Optional: Yes, assume no error PC is available if absent. Size: uint64_t.

enumerator CUDBG_WARP_ATTRIBUTE_CLUSTERIDX = 4

Cluster index for this warp.

Optional: Yes if warp is not in a cluster. Size: CuDim3.

enumerator CUDBG_WARP_ATTRIBUTE_CLUSTERDIM = 5

Cluster dimensions for this warp.

Optional: Yes if warp is not in a cluster. Size: CuDim3.

enumerator CUDBG_WARP_ATTRIBUTE_CLUSTER_EXCEPTION_TARGET_BLOCK_IDX = 6

For cluster exceptions, this represents the target block index handling cluster requests.

Optional: Yes, assume no block index is available if absent. Size: CuDim3.

enumerator CUDBG_WARP_ATTRIBUTE_IN_SYSCALL_LANES = 7

Lane mask showing threads that are in a syscall.

Optional: Yes, use readSyscallCallDepth() if this attribute is not present. Size: uint32_t.

enumerator CUDBG_WARP_ATTRIBUTE_HIT_BREAKPOINT_HANDLE = 8: Breakpoint handle of a broken warp (of type CUDBGBreakpointHandle) Optional: Yes, only present for broken warps Size: CUDBGBreakpointHandle.

enumerator CUDBG_WARP_ATTRIBUTE_COUNT: Warp attributes count.

enum ptxStorageKind

Memory segments for DWARF.

Values:

enumerator ptxUNSPECIFIEDStorage = 0x0

enumerator ptxCodeStorage = 0x1

enumerator ptxRegStorage = 0x2

enumerator ptxSregStorage = 0x3

enumerator ptxConstStorage = 0x4

enumerator ptxGlobalStorage = 0x5

enumerator ptxLocalStorage = 0x6

enumerator ptxParamStorage = 0x7

enumerator ptxSharedStorage = 0x8

enumerator ptxSurfStorage = 0x9

enumerator ptxTexStorage = 0xA

enumerator ptxTexSamplerStorage = 0xB

enumerator ptxGenericStorage = 0xC

enumerator ptxIParamStorage = 0xD

enumerator ptxOParamStorage = 0xE

enumerator ptxFrameStorage = 0xF

enumerator ptxURegStorage = 0x10

enumerator ptxBoundConstStorage0 = 0x20

enumerator ptxBoundConstStorage1 = 0x21

enumerator ptxBoundConstStorage2 = 0x22

enumerator ptxBoundConstStorage3 = 0x23

enumerator ptxBoundConstStorage4 = 0x24

enumerator ptxBoundConstStorage5 = 0x25

enumerator ptxBoundConstStorage6 = 0x26

enumerator ptxBoundConstStorage7 = 0x27

enumerator ptxBoundConstStorage8 = 0x28

enumerator ptxBoundConstStorage9 = 0x29

enumerator ptxBoundConstStorage10 = 0x2A

enumerator ptxBoundConstStorage11 = 0x2B

enumerator ptxBoundConstStorage12 = 0x2C

enumerator ptxBoundConstStorage13 = 0x2D

enumerator ptxBoundConstStorage14 = 0x2E

enumerator ptxBoundConstStorage15 = 0x2F

enumerator ptxBoundConstStorage16 = 0x30

enumerator ptxBoundConstStorage17 = 0x31

enumerator ptxBoundConstStorage18 = 0x32

enumerator ptxBoundConstStorage19 = 0x33

enumerator ptxBoundConstStorage20 = 0x34

enumerator ptxBoundConstStorage21 = 0x35

enumerator ptxBoundConstStorage22 = 0x36

enumerator ptxBoundConstStorage23 = 0x37

enumerator ptxBoundConstStorage24 = 0x38

enumerator ptxBoundConstStorage25 = 0x39

enumerator ptxBoundConstStorage26 = 0x3A

enumerator ptxBoundConstStorage27 = 0x3B

enumerator ptxBoundConstStorage28 = 0x3C

enumerator ptxBoundConstStorage29 = 0x3D

enumerator ptxBoundConstStorage30 = 0x3E

enumerator ptxBoundConstStorage31 = 0x3F

enumerator ptxMAXStorage