525.53

Author: aritger

CHANGELOG.md

@@ -1,5 +1,22 @@
 # Changelog
 
+## Release 525 Entries
+
+### [525.53] 2022-11-10
+
+#### Changed
+
+- GSP firmware is now distributed as multiple firmware files: this release has `gsp_tu10x.bin` and `gsp_ad10x.bin` replacing `gsp.bin` from previous releases.
+    - Each file is named after a GPU architecture and supports GPUs from one or more architectures. This allows GSP firmware to better leverage each architecture's capabilities.
+    - The .run installer will continue to install firmware to `/lib/firmware/nvidia/<version>` and the `nvidia.ko` kernel module will load the appropriate firmware for each GPU at runtime.
+
+#### Fixed
+
+- Add support for IBT (indirect branch tracking) on supported platforms, [#256](https://github.com/NVIDIA/open-gpu-kernel-modules/issues/256) by @rnd-ash
+- Return EINVAL when [failing to] allocating memory, [#280](https://github.com/NVIDIA/open-gpu-kernel-modules/pull/280) by @YusufKhan-gamedev
+- Fix various typos in nvidia/src/kernel, [#16](https://github.com/NVIDIA/open-gpu-kernel-modules/pull/16) by @alexisgeoffrey
+- Added support for rotation in X11, Quadro Sync, Stereo, and YUV 4:2:0 on Turing.
+
 ## Release 520 Entries
 
 ### [520.56.06] 2022-10-12
@@ -29,6 +46,8 @@
 - Improved compatibility with new Linux kernel releases
 - Fixed possible excessive GPU power draw on an idle X11 or Wayland desktop when driving high resolutions or refresh rates
 
+### [515.65.07] 2022-10-19
+
 ### [515.65.01] 2022-08-02
 
 #### Fixed

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 520.56.06.
+version 525.53.
 
 
 ## How to Build
@@ -15,9 +15,9 @@ as root:
 
     make modules_install -j$(nproc)
 
-Note that the kernel modules built here must be used with gsp.bin
+Note that the kernel modules built here must be used with GSP
 firmware and user-space NVIDIA GPU driver components from a corresponding
-520.56.06 driver release.  This can be achieved by installing
+525.53 driver release.  This can be achieved by installing
 the NVIDIA GPU driver from the .run file using the `--no-kernel-modules`
 option.  E.g.,
 
@@ -167,15 +167,15 @@ for the target kernel.
 ## Compatible GPUs
 
 The open-gpu-kernel-modules can be used on any Turing or later GPU
-(see the table below). However, in the 520.56.06 release,
+(see the table below). However, in the 525.53 release,
 GeForce and Workstation support is still considered alpha-quality.
 
 To enable use of the open kernel modules on GeForce and Workstation GPUs,
 set the "NVreg_OpenRmEnableUnsupportedGpus" nvidia.ko kernel module
 parameter to 1. For more details, see the NVIDIA GPU driver end user
 README here:
 
-https://us.download.nvidia.com/XFree86/Linux-x86_64/520.56.06/README/kernel_open.html
+https://us.download.nvidia.com/XFree86/Linux-x86_64/525.53/README/kernel_open.html
 
 In the below table, if three IDs are listed, the first is the PCI Device 
 ID, the second is the PCI Subsystem Vendor ID, and the third is the PCI
@@ -652,6 +652,17 @@ Subsystem Device ID.
 | NVIDIA PG506-232                                | 20B6 10DE 1492 |
 | NVIDIA A30                                      | 20B7 10DE 1532 |
 | NVIDIA A100-PCIE-40GB                           | 20F1 10DE 145F |
+| NVIDIA A800-SXM4-80GB                           | 20F3 10DE 179B |
+| NVIDIA A800-SXM4-80GB                           | 20F3 10DE 179C |
+| NVIDIA A800-SXM4-80GB                           | 20F3 10DE 179D |
+| NVIDIA A800-SXM4-80GB                           | 20F3 10DE 179E |
+| NVIDIA A800-SXM4-80GB                           | 20F3 10DE 179F |
+| NVIDIA A800-SXM4-80GB                           | 20F3 10DE 17A0 |
+| NVIDIA A800-SXM4-80GB                           | 20F3 10DE 17A1 |
+| NVIDIA A800-SXM4-80GB                           | 20F3 10DE 17A2 |
+| NVIDIA A800 80GB PCIe                           | 20F5 10DE 1799 |
+| NVIDIA A800 80GB PCIe LC                        | 20F5 10DE 179A |
+| NVIDIA A800 40GB PCIe                           | 20F6 10DE 17A3 |
 | NVIDIA GeForce GTX 1660 Ti                      | 2182           |
 | NVIDIA GeForce GTX 1660                         | 2184           |
 | NVIDIA GeForce GTX 1650 SUPER                   | 2187           |

kernel-open/Kbuild

@@ -72,7 +72,7 @@ EXTRA_CFLAGS += -I$(src)/common/inc
 EXTRA_CFLAGS += -I$(src)
 EXTRA_CFLAGS += -Wall -MD $(DEFINES) $(INCLUDES) -Wno-cast-qual -Wno-error -Wno-format-extra-args
 EXTRA_CFLAGS += -D__KERNEL__ -DMODULE -DNVRM
-EXTRA_CFLAGS += -DNV_VERSION_STRING=\"520.56.06\"
+EXTRA_CFLAGS += -DNV_VERSION_STRING=\"525.53\"
 
 EXTRA_CFLAGS += -Wno-unused-function
 
@@ -229,6 +229,7 @@ NV_HEADER_PRESENCE_TESTS = \
  drm/drm_ioctl.h \
  drm/drm_device.h \
  drm/drm_mode_config.h \
+ drm/drm_modeset_lock.h \
  dt-bindings/interconnect/tegra_icc_id.h \
  generated/autoconf.h \
  generated/compile.h \
@@ -243,6 +244,8 @@ NV_HEADER_PRESENCE_TESTS = \
  linux/log2.h \
  linux/of.h \
  linux/bug.h \
+ linux/sched.h \
+ linux/sched/mm.h \
  linux/sched/signal.h \
  linux/sched/task.h \
  linux/sched/task_stack.h \
@@ -286,7 +289,10 @@ NV_HEADER_PRESENCE_TESTS = \
  linux/ioasid.h \
  linux/stdarg.h \
  linux/iosys-map.h \
- asm/coco.h
+ asm/coco.h \
+ linux/vfio_pci_core.h \
+ soc/tegra/bpmp-abi.h \
+ soc/tegra/bpmp.h
 
 # Filename to store the define for the header in $(1); this is only consumed by
 # the rule below that concatenates all of these together.

kernel-open/common/inc/cpuopsys.h

@@ -242,7 +242,7 @@
 #endif
 
 /* For verification-only features not intended to be included in normal drivers */
-#if (defined(NV_MODS) || defined(NV_GSP_MODS)) && defined(DEBUG) && !defined(DISABLE_VERIF_FEATURES)
+#if defined(ENABLE_VERIF_FEATURES)
 #define NV_VERIF_FEATURES
 #endif
 
@@ -276,12 +276,6 @@
 #define NV_IS_MODS 0
 #endif
 
-#if defined(NV_GSP_MODS)
-#define NV_IS_GSP_MODS 1
-#else
-#define NV_IS_GSP_MODS 0
-#endif
-
 #if defined(NV_WINDOWS)
 #define NVOS_IS_WINDOWS 1
 #else

kernel-open/common/inc/nv-firmware.h

@@ -0,0 +1,132 @@
+/*
+ * SPDX-FileCopyrightText: Copyright (c) 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef NV_FIRMWARE_H
+#define NV_FIRMWARE_H
+
+
+
+#include <nvtypes.h>
+#include <nvmisc.h>
+
+typedef enum
+{
+    NV_FIRMWARE_TYPE_GSP,
+    NV_FIRMWARE_TYPE_GSP_LOG
+} nv_firmware_type_t;
+
+typedef enum
+{
+    NV_FIRMWARE_CHIP_FAMILY_NULL = 0,
+    NV_FIRMWARE_CHIP_FAMILY_TU10X = 1,
+    NV_FIRMWARE_CHIP_FAMILY_TU11X = 2,
+    NV_FIRMWARE_CHIP_FAMILY_GA100 = 3,
+    NV_FIRMWARE_CHIP_FAMILY_GA10X = 4,
+    NV_FIRMWARE_CHIP_FAMILY_AD10X = 5,
+    NV_FIRMWARE_CHIP_FAMILY_GH100 = 6,
+    NV_FIRMWARE_CHIP_FAMILY_END,
+} nv_firmware_chip_family_t;
+
+static inline const char *nv_firmware_chip_family_to_string(
+    nv_firmware_chip_family_t fw_chip_family
+)
+{
+    switch (fw_chip_family) {
+        case NV_FIRMWARE_CHIP_FAMILY_GH100: return "gh100";
+        case NV_FIRMWARE_CHIP_FAMILY_AD10X: return "ad10x";
+        case NV_FIRMWARE_CHIP_FAMILY_GA10X: return "ga10x";
+        case NV_FIRMWARE_CHIP_FAMILY_GA100: return "ga100";
+        case NV_FIRMWARE_CHIP_FAMILY_TU11X: return "tu11x";
+        case NV_FIRMWARE_CHIP_FAMILY_TU10X: return "tu10x";
+
+        case NV_FIRMWARE_CHIP_FAMILY_END:  // fall through
+        case NV_FIRMWARE_CHIP_FAMILY_NULL:
+            return NULL;
+    }
+    return NULL;
+}
+
+// The includer (presumably nv.c) may optionally define
+// NV_FIRMWARE_PATH_FOR_FILENAME(filename)
+// to return a string "path" given a gsp_*.bin or gsp_log_*.bin filename.
+//
+// The function nv_firmware_path will then be available.
+#if defined(NV_FIRMWARE_PATH_FOR_FILENAME)
+static inline const char *nv_firmware_path(
+    nv_firmware_type_t fw_type,
+    nv_firmware_chip_family_t fw_chip_family
+)
+{
+    if (fw_type == NV_FIRMWARE_TYPE_GSP)
+    {
+        switch (fw_chip_family)
+        {
+            case NV_FIRMWARE_CHIP_FAMILY_AD10X:
+                return NV_FIRMWARE_PATH_FOR_FILENAME("gsp_ad10x.bin");
+
+            case NV_FIRMWARE_CHIP_FAMILY_GH100:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_GA100:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_GA10X:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_TU11X:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_TU10X:
+                return NV_FIRMWARE_PATH_FOR_FILENAME("gsp_tu10x.bin");
+
+            case NV_FIRMWARE_CHIP_FAMILY_END:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_NULL:
+                return "";
+        }
+    }
+    else if (fw_type == NV_FIRMWARE_TYPE_GSP_LOG)
+    {
+        switch (fw_chip_family)
+        {
+            case NV_FIRMWARE_CHIP_FAMILY_AD10X:
+                return NV_FIRMWARE_PATH_FOR_FILENAME("gsp_log_ad10x.bin");
+
+            case NV_FIRMWARE_CHIP_FAMILY_GH100:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_GA100:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_GA10X:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_TU11X:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_TU10X:
+                return NV_FIRMWARE_PATH_FOR_FILENAME("gsp_log_tu10x.bin");
+
+            case NV_FIRMWARE_CHIP_FAMILY_END:  // fall through
+            case NV_FIRMWARE_CHIP_FAMILY_NULL:
+                return "";
+        }
+    }
+
+    return "";
+}
+#endif  // defined(NV_FIRMWARE_PATH_FOR_FILENAME)
+
+// The includer (presumably nv.c) may optionally define
+// NV_FIRMWARE_DECLARE_GSP_FILENAME(filename)
+// which will then be invoked (at the top-level) for each
+// gsp_*.bin (but not gsp_log_*.bin)
+#if defined(NV_FIRMWARE_DECLARE_GSP_FILENAME)
+NV_FIRMWARE_DECLARE_GSP_FILENAME("gsp_ad10x.bin")
+NV_FIRMWARE_DECLARE_GSP_FILENAME("gsp_tu10x.bin")
+#endif  // defined(NV_FIRMWARE_DECLARE_GSP_FILENAME)
+
+#endif  // NV_FIRMWARE_DECLARE_GSP_FILENAME

kernel-open/common/inc/nv-hash.h

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2020 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-FileCopyrightText: Copyright (c) 2020-22 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
  * SPDX-License-Identifier: MIT
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -91,6 +91,6 @@ static inline void _nv_hash_init(struct hlist_head *ht, unsigned int sz)
  * @key: the key of the objects to iterate over
  */
 #define nv_hash_for_each_possible(name, obj, member, key) \
-    nv_hlist_for_each_entry(obj, &name[NV_HASH_MIN(key, NV_HASH_BITS(name))], member)
+    hlist_for_each_entry(obj, &name[NV_HASH_MIN(key, NV_HASH_BITS(name))], member)
 
 #endif // __NV_HASH_H__

kernel-open/common/inc/nv-hypervisor.h

@@ -27,15 +27,13 @@
 #include <nv-kernel-interface-api.h>
 
 // Enums for supported hypervisor types.
-// New hypervisor type should be added before OS_HYPERVISOR_CUSTOM_FORCED
+// New hypervisor type should be added before OS_HYPERVISOR_UNKNOWN
 typedef enum _HYPERVISOR_TYPE
 {
     OS_HYPERVISOR_XEN = 0,
     OS_HYPERVISOR_VMWARE,
     OS_HYPERVISOR_HYPERV,
     OS_HYPERVISOR_KVM,
-    OS_HYPERVISOR_PARALLELS,
-    OS_HYPERVISOR_CUSTOM_FORCED,
     OS_HYPERVISOR_UNKNOWN
 } HYPERVISOR_TYPE;
 

kernel-open/common/inc/nv-kthread-q.h

@@ -115,11 +115,6 @@ struct nv_kthread_q_item
     void *function_args;
 };
 
-#if defined(NV_KTHREAD_CREATE_ON_NODE_PRESENT)
-    #define NV_KTHREAD_Q_SUPPORTS_AFFINITY() 1
-#else
-    #define NV_KTHREAD_Q_SUPPORTS_AFFINITY() 0
-#endif
 
 #ifndef NUMA_NO_NODE
 #define NUMA_NO_NODE (-1)
@@ -142,18 +137,12 @@ struct nv_kthread_q_item
 //
 // A short prefix of the qname arg will show up in []'s, via the ps(1) utility.
 //
-// The kernel thread stack is preferably allocated on the specified NUMA node if
-// NUMA-affinity (NV_KTHREAD_Q_SUPPORTS_AFFINITY() == 1) is supported, but
-// fallback to another node is possible because kernel allocators do not
+// The kernel thread stack is preferably allocated on the specified NUMA node, 
+// but fallback to another node is possible because kernel allocators do not
 // guarantee affinity. Note that NUMA-affinity applies only to
 // the kthread stack. This API does not do anything about limiting the CPU
 // affinity of the kthread. That is left to the caller.
 //
-// On kernels, which do not support NUMA-aware kthread stack allocations
-// (NV_KTHTREAD_Q_SUPPORTS_AFFINITY() == 0), the API will return -ENOTSUPP
-// if the value supplied for 'preferred_node' is anything other than
-// NV_KTHREAD_NO_NODE.
-//
 // Reusing a queue: once a queue is initialized, it must be safely shut down
 // (see "Stopping the queue(s)", below), before it can be reused. So, for
 // a simple queue use case, the following will work: