550.127.05

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 550.120.
+version 550.127.05.
 
 
 ## 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
-550.120 driver release.  This can be achieved by installing
+550.127.05 driver release.  This can be achieved by installing
 the NVIDIA GPU driver from the .run file using the `--no-kernel-modules`
 option.  E.g.,
 
@@ -188,7 +188,7 @@ encountered specific to them.
 For details on feature support and limitations, see the NVIDIA GPU driver
 end user README here:
 
-https://us.download.nvidia.com/XFree86/Linux-x86_64/550.120/README/kernel_open.html
+https://us.download.nvidia.com/XFree86/Linux-x86_64/550.127.05/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

@@ -72,7 +72,7 @@ EXTRA_CFLAGS += -I$(src)/common/inc
 EXTRA_CFLAGS += -I$(src)
 EXTRA_CFLAGS += -Wall $(DEFINES) $(INCLUDES) -Wno-cast-qual -Wno-format-extra-args
 EXTRA_CFLAGS += -D__KERNEL__ -DMODULE -DNVRM
-EXTRA_CFLAGS += -DNV_VERSION_STRING=\"550.120\"
+EXTRA_CFLAGS += -DNV_VERSION_STRING=\"550.127.05\"
 
 ifneq ($(SYSSRCHOST1X),)
  EXTRA_CFLAGS += -I$(SYSSRCHOST1X)

kernel-open/nvidia-uvm/uvm.h

@@ -379,6 +379,17 @@ NV_STATUS UvmIsPageableMemoryAccessSupportedOnGpu(const NvProcessorUuid *gpuUuid
 //         OS state required to register the GPU is malformed, or the partition
 //         identified by the user handles or its configuration changed.
 //
+//     NV_ERR_NVSWITCH_FABRIC_NOT_READY:
+//         (On NvSwitch-connected system) Indicates that the fabric has not been
+//         configured yet. Caller must retry GPU registration.
+//
+//     NV_ERR_NVSWITCH_FABRIC_FAILURE:
+//         (On NvSwitch-connected systems) Indicates that the NvLink fabric
+//         failed to be configured.
+//
+//     NV_ERR_GPU_MEMORY_ONLINING_FAULURE:
+//         (On coherent systems) The GPU's memory onlining failed.
+//
 //     NV_ERR_GENERIC:
 //         Unexpected error. We try hard to avoid returning this error code,
 //         because it is not very informative.

kernel-open/nvidia-uvm/uvm_gpu.c

@@ -138,6 +138,7 @@ static NV_STATUS get_gpu_caps(uvm_gpu_t *gpu)
 
     if (gpu_caps.numaEnabled) {
         UVM_ASSERT(uvm_parent_gpu_is_coherent(gpu->parent));
+
         gpu->mem_info.numa.enabled = true;
         gpu->mem_info.numa.node_id = gpu_caps.numaNodeId;
     }
@@ -1280,7 +1281,8 @@ static NV_STATUS init_gpu(uvm_gpu_t *gpu, const UvmGpuInfo *gpu_info)
 
     status = get_gpu_caps(gpu);
     if (status != NV_OK) {
-        UVM_ERR_PRINT("Failed to get GPU caps: %s, GPU %s\n", nvstatusToString(status), uvm_gpu_name(gpu));
+        if (status != NV_ERR_NVSWITCH_FABRIC_NOT_READY)
+            UVM_ERR_PRINT("Failed to get GPU caps: %s, GPU %s\n", nvstatusToString(status), uvm_gpu_name(gpu));
         return status;
     }
 

kernel-open/nvidia/nv-pci.c

@@ -40,6 +40,9 @@
 #if !defined(NV_BUS_TYPE_HAS_IOMMU_OPS)
 #include <linux/iommu.h>
 #endif
+#if NV_IS_EXPORT_SYMBOL_GPL_pci_ats_supported
+#include <linux/pci-ats.h>
+#endif
 
 static void
 nv_check_and_exclude_gpu(
@@ -781,10 +784,15 @@ next_bar:
     // PPC64LE platform where ATS is currently supported (IBM P9).
     nv_ats_supported &= nv_platform_supports_numa(nvl);
 #else
-#if defined(NV_PCI_DEV_HAS_ATS_ENABLED)
+#if NV_IS_EXPORT_SYMBOL_GPL_pci_ats_supported
+    nv_ats_supported &= pci_ats_supported(pci_dev);
+#elif defined(NV_PCI_DEV_HAS_ATS_ENABLED)
     nv_ats_supported &= pci_dev->ats_enabled;
+#else
+    nv_ats_supported = NV_FALSE;
 #endif
 #endif
+
     if (nv_ats_supported)
     {
         NV_DEV_PRINTF(NV_DBG_INFO, nv, "ATS supported by this GPU!\n");

kernel-open/nvidia/nv.c

@@ -188,11 +188,7 @@ struct semaphore nv_linux_devices_lock;
 
 // True if all the successfully probed devices support ATS
 // Assigned at device probe (module init) time
-NvBool nv_ats_supported = NVCPU_IS_PPC64LE
+NvBool nv_ats_supported = NV_TRUE;
-#if defined(NV_PCI_DEV_HAS_ATS_ENABLED)
-                          || NV_TRUE
-#endif
-;
 
 // allow an easy way to convert all debug printfs related to events
 // back and forth between 'info' and 'errors'

kernel-open/nvidia/nvidia.Kbuild

@@ -231,6 +231,7 @@ NV_CONFTEST_SYMBOL_COMPILE_TESTS += is_export_symbol_present_tsec_comms_free_gsc
 NV_CONFTEST_SYMBOL_COMPILE_TESTS += is_export_symbol_present_memory_block_size_bytes
 NV_CONFTEST_SYMBOL_COMPILE_TESTS += crypto
 NV_CONFTEST_SYMBOL_COMPILE_TESTS += is_export_symbol_present_follow_pte
+NV_CONFTEST_SYMBOL_COMPILE_TESTS += is_export_symbol_gpl_pci_ats_supported
 
 NV_CONFTEST_TYPE_COMPILE_TESTS += dma_ops
 NV_CONFTEST_TYPE_COMPILE_TESTS += swiotlb_dma_ops

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             r550_00
+    #define NV_BUILD_BRANCH             r553_17
 #endif
 #ifndef NV_PUBLIC_BRANCH
-    #define NV_PUBLIC_BRANCH             r550_00
+    #define NV_PUBLIC_BRANCH             r553_17
 #endif
 
 #if defined(NV_LINUX) || defined(NV_BSD) || defined(NV_SUNOS)
-#define NV_BUILD_BRANCH_VERSION         "rel/gpu_drv/r550/r550_00-410"
+#define NV_BUILD_BRANCH_VERSION         "rel/gpu_drv/r550/r553_17-429"
-#define NV_BUILD_CHANGELIST_NUM         (34843164)
+#define NV_BUILD_CHANGELIST_NUM         (34957518)
 #define NV_BUILD_TYPE                   "Official"
-#define NV_BUILD_NAME                   "rel/gpu_drv/r550/r550_00-410"
+#define NV_BUILD_NAME                   "rel/gpu_drv/r550/r553_17-429"
-#define NV_LAST_OFFICIAL_CHANGELIST_NUM (34843164)
+#define NV_LAST_OFFICIAL_CHANGELIST_NUM (34957518)
 
 #else     /* Windows builds */
-#define NV_BUILD_BRANCH_VERSION         "r550_00-390"
+#define NV_BUILD_BRANCH_VERSION         "r553_17-2"
-#define NV_BUILD_CHANGELIST_NUM         (34843164)
+#define NV_BUILD_CHANGELIST_NUM         (34902203)
 #define NV_BUILD_TYPE                   "Official"
-#define NV_BUILD_NAME                   "553.09"
+#define NV_BUILD_NAME                   "553.20"
-#define NV_LAST_OFFICIAL_CHANGELIST_NUM (34843164)
+#define NV_LAST_OFFICIAL_CHANGELIST_NUM (34902203)
 #define NV_BUILD_BRANCH_BASE_VERSION    R550
 #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               "550.120"
+#define NV_VERSION_STRING               "550.127.05"
 
 #else
 

src/common/inc/swref/published/nvswitch/ls10/dev_ctrl_ip_addendum.h

@@ -62,4 +62,14 @@
 #define NV_CTRL_CPU_INTR_UNITS_PRIV_RING                      15:15
 #define NV_CTRL_CPU_INTR_UNITS_FSP                            16:16
 
+#define NV_CTRL_CPU_INTR_TOP_LEAF_BIT(i)                      (i/2):(i/2)
+#define NV_CTRL_CPU_INTR_TOP_LEAF_INTR_UNITS                  NV_CTRL_CPU_INTR_TOP_LEAF_BIT(NV_CTRL_CPU_INTR_UNITS_IDX)
+#define NV_CTRL_CPU_INTR_TOP_LEAF_INTR_NPG_FATAL              NV_CTRL_CPU_INTR_TOP_LEAF_BIT(NV_CTRL_CPU_INTR_NPG_FATAL_IDX)
+#define NV_CTRL_CPU_INTR_TOP_LEAF_INTR_NPG_NON_FATAL          NV_CTRL_CPU_INTR_TOP_LEAF_BIT(NV_CTRL_CPU_INTR_NPG_NON_FATAL_IDX)
+#define NV_CTRL_CPU_INTR_TOP_LEAF_INTR_NPG_CORRECTABLE        NV_CTRL_CPU_INTR_TOP_LEAF_BIT(NV_CTRL_CPU_INTR_NPG_CORRECTABLE_IDX)
+#define NV_CTRL_CPU_INTR_TOP_LEAF_INTR_NVLW_FATAL             NV_CTRL_CPU_INTR_TOP_LEAF_BIT(NV_CTRL_CPU_INTR_NVLW_FATAL_IDX)
+#define NV_CTRL_CPU_INTR_TOP_LEAF_INTR_NVLW_NON_FATAL         NV_CTRL_CPU_INTR_TOP_LEAF_BIT(NV_CTRL_CPU_INTR_NVLW_NON_FATAL_IDX)
+#define NV_CTRL_CPU_INTR_TOP_LEAF_INTR_NVLW_CORRECTABLE       NV_CTRL_CPU_INTR_TOP_LEAF_BIT(NV_CTRL_CPU_INTR_NVLW_CORRECTABLE_IDX)
+#define NV_CTRL_CPU_INTR_TOP_LEAF_INTR_NXBAR_FATAL            NV_CTRL_CPU_INTR_TOP_LEAF_BIT(NV_CTRL_CPU_INTR_NXBAR_FATAL_IDX)
+
 #endif // __ls10_dev_ctrl_ip_addendum_h__

src/common/nvswitch/kernel/nvswitch.c

@@ -6138,26 +6138,66 @@ nvswitch_lib_ctrl_tnvl_lock_only
                 _nvswitch_ctrl_get_inforom_bbx_temp_samples,
                 NVSWITCH_GET_TEMP_SAMPLES_PARAMS,
                 osPrivate, flags);
-        NVSWITCH_DEV_CMD_DISPATCH_PRIVILEGED(
+        NVSWITCH_DEV_CMD_DISPATCH(
                 CTRL_NVSWITCH_GET_ATTESTATION_CERTIFICATE_CHAIN,
                 _nvswitch_ctrl_get_attestation_certificate_chain,
-                NVSWITCH_GET_ATTESTATION_CERTIFICATE_CHAIN_PARAMS,
+                NVSWITCH_GET_ATTESTATION_CERTIFICATE_CHAIN_PARAMS);
-                osPrivate, flags);
+        NVSWITCH_DEV_CMD_DISPATCH(
-        NVSWITCH_DEV_CMD_DISPATCH_PRIVILEGED(
                 CTRL_NVSWITCH_GET_ATTESTATION_REPORT,
                 _nvswitch_ctrl_get_attestation_report,
-                NVSWITCH_GET_ATTESTATION_REPORT_PARAMS,
+                NVSWITCH_GET_ATTESTATION_REPORT_PARAMS);
-                osPrivate, flags);
+        NVSWITCH_DEV_CMD_DISPATCH(
-        NVSWITCH_DEV_CMD_DISPATCH_PRIVILEGED(
                 CTRL_NVSWITCH_GET_TNVL_STATUS,
                 _nvswitch_ctrl_get_tnvl_status,
-                NVSWITCH_GET_TNVL_STATUS_PARAMS,
+                NVSWITCH_GET_TNVL_STATUS_PARAMS);
-                osPrivate, flags);
         NVSWITCH_DEV_CMD_DISPATCH_PRIVILEGED(
                 CTRL_NVSWITCH_SET_FM_DRIVER_STATE,
                 nvswitch_ctrl_set_fm_driver_state,
                 NVSWITCH_SET_FM_DRIVER_STATE_PARAMS,
                 osPrivate, flags);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_ERRORS,
+                nvswitch_ctrl_get_errors,
+                NVSWITCH_GET_ERRORS_PARAMS);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_BIOS_INFO,
+                _nvswitch_ctrl_get_bios_info,
+                NVSWITCH_GET_BIOS_INFO_PARAMS);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_TEMPERATURE,
+                _nvswitch_ctrl_therm_read_temperature,
+                NVSWITCH_CTRL_GET_TEMPERATURE_PARAMS);
+        NVSWITCH_DEV_CMD_DISPATCH(
+                CTRL_NVSWITCH_GET_TEMPERATURE_LIMIT,
+                _nvswitch_ctrl_therm_get_temperature_limit,
+                NVSWITCH_CTRL_GET_TEMPERATURE_LIMIT_PARAMS);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_FATAL_ERROR_SCOPE,
+                _nvswitch_ctrl_get_fatal_error_scope,
+                NVSWITCH_GET_FATAL_ERROR_SCOPE_PARAMS);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_INFO,
+                _nvswitch_ctrl_get_info,
+                NVSWITCH_GET_INFO);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_VOLTAGE,
+                _nvswitch_ctrl_therm_read_voltage,
+                NVSWITCH_CTRL_GET_VOLTAGE_PARAMS);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_POWER,
+                _nvswitch_ctrl_therm_read_power,
+                NVSWITCH_GET_POWER_PARAMS);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_NVLINK_STATUS,
+                _nvswitch_ctrl_get_nvlink_status,
+                NVSWITCH_GET_NVLINK_STATUS_PARAMS);
+        NVSWITCH_DEV_CMD_DISPATCH_PRIVILEGED(
+                CTRL_NVSWITCH_GET_NVLINK_ECC_ERRORS,
+                _nvswitch_ctrl_get_nvlink_ecc_errors,
+                NVSWITCH_GET_NVLINK_ECC_ERRORS_PARAMS,
+                osPrivate, flags);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_INTERNAL_LATENCY,
+                _nvswitch_ctrl_get_internal_latency,
+                NVSWITCH_GET_INTERNAL_LATENCY);
+        NVSWITCH_DEV_CMD_DISPATCH_PRIVILEGED(CTRL_NVSWITCH_SET_NVLINK_ERROR_THRESHOLD,
+                _nvswitch_ctrl_set_nvlink_error_threshold,
+                NVSWITCH_SET_NVLINK_ERROR_THRESHOLD_PARAMS,
+                osPrivate, flags);
+        NVSWITCH_DEV_CMD_DISPATCH(CTRL_NVSWITCH_GET_NVLINK_ERROR_THRESHOLD,
+                _nvswitch_ctrl_get_nvlink_error_threshold,
+                NVSWITCH_GET_NVLINK_ERROR_THRESHOLD_PARAMS);
         default:
             nvswitch_os_print(NVSWITCH_DBG_LEVEL_INFO, "ioctl %x is not permitted when TNVL is locked\n", cmd);
             return -NVL_ERR_INSUFFICIENT_PERMISSIONS;

src/nvidia/src/kernel/gpu/arch/hopper/kern_gpu_gh100.c

@@ -493,9 +493,8 @@ gpuIsProtectedPcieEnabledInHw_GH100
     OBJGPU *pGpu
 )
 {
-    NvU32 val = GPU_REG_RD32(pGpu, NV_PGC6_AON_SECURE_SCRATCH_GROUP_20_CC);
+    // Bug 4870925: Disabled PPCIE
-    return FLD_TEST_DRF(_PGC6, _AON_SECURE_SCRATCH_GROUP_20_CC, _MULTI_GPU_MODE,
+    return NV_FALSE;
-                        _PROTECTED_PCIE, val);
 }
 
 /*!

src/nvidia/src/kernel/gpu/conf_compute/conf_compute_api.c

@@ -134,6 +134,8 @@ confComputeApiCtrlCmdSystemGetCapabilities_IMPL
 
     if (pCcCaps->bMultiGpuProtectedPcieModeEnabled)
     {
+        // Do not advertise HCC as ON to callers when PPCIe is ON
+        pParams->ccFeature = NV_CONF_COMPUTE_SYSTEM_FEATURE_DISABLED;
         pParams->multiGpuMode = NV_CONF_COMPUTE_SYSTEM_MULTI_GPU_MODE_PROTECTED_PCIE;
     }
 

src/nvidia/src/kernel/virtualization/kernel_vgpu_mgr.c

@@ -2121,8 +2121,8 @@ kvgpumgrCreateRequestVgpu(NvU32 gpuPciId, const NvU8 *pMdevUuid,
 
     if (pGpu == NULL)
     {
-        NV_PRINTF(LEVEL_ERROR, "GPU handle is not valid \n");
+        NV_PRINTF(LEVEL_ERROR, "GPU %u is not initialized yet \n", gpuPciBdf);
-        return NV_ERR_INVALID_STATE;
+        return NV_ERR_TIMEOUT_RETRY;
     }
 
     if ((status = kvgpumgrGetPgpuIndex(pKernelVgpuMgr, gpuPciId, &pgpuIndex)) != NV_OK)

version.mk

@@ -1,4 +1,4 @@
-NVIDIA_VERSION = 550.120
+NVIDIA_VERSION = 550.127.05
 
 # This file.
 VERSION_MK_FILE := $(lastword $(MAKEFILE_LIST))