xref: /linux/drivers/cxl/Kconfig (revision c532de5a67a70f8533d495f8f2aaa9a0491c3ad0)
1# SPDX-License-Identifier: GPL-2.0-only
2menuconfig CXL_BUS
3	tristate "CXL (Compute Express Link) Devices Support"
4	depends on PCI
5	select FW_LOADER
6	select FW_UPLOAD
7	select PCI_DOE
8	select FIRMWARE_TABLE
9	select NUMA_KEEP_MEMINFO if NUMA_MEMBLKS
10	help
11	  CXL is a bus that is electrically compatible with PCI Express, but
12	  layers three protocols on that signalling (CXL.io, CXL.cache, and
13	  CXL.mem). The CXL.cache protocol allows devices to hold cachelines
14	  locally, the CXL.mem protocol allows devices to be fully coherent
15	  memory targets, the CXL.io protocol is equivalent to PCI Express.
16	  Say 'y' to enable support for the configuration and management of
17	  devices supporting these protocols.
18
19if CXL_BUS
20
21config CXL_PCI
22	tristate "PCI manageability"
23	default CXL_BUS
24	help
25	  The CXL specification defines a "CXL memory device" sub-class in the
26	  PCI "memory controller" base class of devices. Device's identified by
27	  this class code provide support for volatile and / or persistent
28	  memory to be mapped into the system address map (Host-managed Device
29	  Memory (HDM)).
30
31	  Say 'y/m' to enable a driver that will attach to CXL memory expander
32	  devices enumerated by the memory device class code for configuration
33	  and management primarily via the mailbox interface. See Chapter 2.3
34	  Type 3 CXL Device in the CXL 2.0 specification for more details.
35
36	  If unsure say 'm'.
37
38config CXL_MEM_RAW_COMMANDS
39	bool "RAW Command Interface for Memory Devices"
40	depends on CXL_PCI
41	help
42	  Enable CXL RAW command interface.
43
44	  The CXL driver ioctl interface may assign a kernel ioctl command
45	  number for each specification defined opcode. At any given point in
46	  time the number of opcodes that the specification defines and a device
47	  may implement may exceed the kernel's set of associated ioctl function
48	  numbers. The mismatch is either by omission, specification is too new,
49	  or by design. When prototyping new hardware, or developing / debugging
50	  the driver it is useful to be able to submit any possible command to
51	  the hardware, even commands that may crash the kernel due to their
52	  potential impact to memory currently in use by the kernel.
53
54	  If developing CXL hardware or the driver say Y, otherwise say N.
55
56config CXL_ACPI
57	tristate "CXL ACPI: Platform Support"
58	depends on ACPI
59	depends on ACPI_NUMA
60	default CXL_BUS
61	select ACPI_TABLE_LIB
62	select ACPI_HMAT
63	select CXL_PORT
64	help
65	  Enable support for host managed device memory (HDM) resources
66	  published by a platform's ACPI CXL memory layout description.  See
67	  Chapter 9.14.1 CXL Early Discovery Table (CEDT) in the CXL 2.0
68	  specification, and CXL Fixed Memory Window Structures (CEDT.CFMWS)
69	  (https://www.computeexpresslink.org/spec-landing). The CXL core
70	  consumes these resource to publish the root of a cxl_port decode
71	  hierarchy to map regions that represent System RAM, or Persistent
72	  Memory regions to be managed by LIBNVDIMM.
73
74	  If unsure say 'm'.
75
76config CXL_PMEM
77	tristate "CXL PMEM: Persistent Memory Support"
78	depends on LIBNVDIMM
79	default CXL_BUS
80	help
81	  In addition to typical memory resources a platform may also advertise
82	  support for persistent memory attached via CXL. This support is
83	  managed via a bridge driver from CXL to the LIBNVDIMM system
84	  subsystem. Say 'y/m' to enable support for enumerating and
85	  provisioning the persistent memory capacity of CXL memory expanders.
86
87	  If unsure say 'm'.
88
89config CXL_MEM
90	tristate "CXL: Memory Expansion"
91	depends on CXL_PCI
92	default CXL_BUS
93	help
94	  The CXL.mem protocol allows a device to act as a provider of "System
95	  RAM" and/or "Persistent Memory" that is fully coherent as if the
96	  memory were attached to the typical CPU memory controller. This is
97	  known as HDM "Host-managed Device Memory".
98
99	  Say 'y/m' to enable a driver that will attach to CXL.mem devices for
100	  memory expansion and control of HDM. See Chapter 9.13 in the CXL 2.0
101	  specification for a detailed description of HDM.
102
103	  If unsure say 'm'.
104
105config CXL_PORT
106	default CXL_BUS
107	tristate
108
109config CXL_SUSPEND
110	def_bool y
111	depends on SUSPEND && CXL_MEM
112
113config CXL_REGION
114	bool "CXL: Region Support"
115	default CXL_BUS
116	# For MAX_PHYSMEM_BITS
117	depends on SPARSEMEM
118	select MEMREGION
119	select GET_FREE_REGION
120	help
121	  Enable the CXL core to enumerate and provision CXL regions. A CXL
122	  region is defined by one or more CXL expanders that decode a given
123	  system-physical address range. For CXL regions established by
124	  platform-firmware this option enables memory error handling to
125	  identify the devices participating in a given interleaved memory
126	  range. Otherwise, platform-firmware managed CXL is enabled by being
127	  placed in the system address map and does not need a driver.
128
129	  If unsure say 'y'
130
131config CXL_REGION_INVALIDATION_TEST
132	bool "CXL: Region Cache Management Bypass (TEST)"
133	depends on CXL_REGION
134	help
135	  CXL Region management and security operations potentially invalidate
136	  the content of CPU caches without notifying those caches to
137	  invalidate the affected cachelines. The CXL Region driver attempts
138	  to invalidate caches when those events occur.  If that invalidation
139	  fails the region will fail to enable.  Reasons for cache
140	  invalidation failure are due to the CPU not providing a cache
141	  invalidation mechanism. For example usage of wbinvd is restricted to
142	  bare metal x86. However, for testing purposes toggling this option
143	  can disable that data integrity safety and proceed with enabling
144	  regions when there might be conflicting contents in the CPU cache.
145
146	  If unsure, or if this kernel is meant for production environments,
147	  say N.
148
149endif
150