xref: /openbmc/linux/drivers/cxl/Kconfig (revision e11ec2b8)
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 PCI_DOE
6	help
7	  CXL is a bus that is electrically compatible with PCI Express, but
8	  layers three protocols on that signalling (CXL.io, CXL.cache, and
9	  CXL.mem). The CXL.cache protocol allows devices to hold cachelines
10	  locally, the CXL.mem protocol allows devices to be fully coherent
11	  memory targets, the CXL.io protocol is equivalent to PCI Express.
12	  Say 'y' to enable support for the configuration and management of
13	  devices supporting these protocols.
14
15if CXL_BUS
16
17config CXL_PCI
18	tristate "PCI manageability"
19	default CXL_BUS
20	help
21	  The CXL specification defines a "CXL memory device" sub-class in the
22	  PCI "memory controller" base class of devices. Device's identified by
23	  this class code provide support for volatile and / or persistent
24	  memory to be mapped into the system address map (Host-managed Device
25	  Memory (HDM)).
26
27	  Say 'y/m' to enable a driver that will attach to CXL memory expander
28	  devices enumerated by the memory device class code for configuration
29	  and management primarily via the mailbox interface. See Chapter 2.3
30	  Type 3 CXL Device in the CXL 2.0 specification for more details.
31
32	  If unsure say 'm'.
33
34config CXL_MEM_RAW_COMMANDS
35	bool "RAW Command Interface for Memory Devices"
36	depends on CXL_PCI
37	help
38	  Enable CXL RAW command interface.
39
40	  The CXL driver ioctl interface may assign a kernel ioctl command
41	  number for each specification defined opcode. At any given point in
42	  time the number of opcodes that the specification defines and a device
43	  may implement may exceed the kernel's set of associated ioctl function
44	  numbers. The mismatch is either by omission, specification is too new,
45	  or by design. When prototyping new hardware, or developing / debugging
46	  the driver it is useful to be able to submit any possible command to
47	  the hardware, even commands that may crash the kernel due to their
48	  potential impact to memory currently in use by the kernel.
49
50	  If developing CXL hardware or the driver say Y, otherwise say N.
51
52config CXL_ACPI
53	tristate "CXL ACPI: Platform Support"
54	depends on ACPI
55	default CXL_BUS
56	select ACPI_TABLE_LIB
57	help
58	  Enable support for host managed device memory (HDM) resources
59	  published by a platform's ACPI CXL memory layout description.  See
60	  Chapter 9.14.1 CXL Early Discovery Table (CEDT) in the CXL 2.0
61	  specification, and CXL Fixed Memory Window Structures (CEDT.CFMWS)
62	  (https://www.computeexpresslink.org/spec-landing). The CXL core
63	  consumes these resource to publish the root of a cxl_port decode
64	  hierarchy to map regions that represent System RAM, or Persistent
65	  Memory regions to be managed by LIBNVDIMM.
66
67	  If unsure say 'm'.
68
69config CXL_PMEM
70	tristate "CXL PMEM: Persistent Memory Support"
71	depends on LIBNVDIMM
72	default CXL_BUS
73	help
74	  In addition to typical memory resources a platform may also advertise
75	  support for persistent memory attached via CXL. This support is
76	  managed via a bridge driver from CXL to the LIBNVDIMM system
77	  subsystem. Say 'y/m' to enable support for enumerating and
78	  provisioning the persistent memory capacity of CXL memory expanders.
79
80	  If unsure say 'm'.
81
82config CXL_MEM
83	tristate "CXL: Memory Expansion"
84	depends on CXL_PCI
85	select FW_UPLOAD
86	default CXL_BUS
87	help
88	  The CXL.mem protocol allows a device to act as a provider of "System
89	  RAM" and/or "Persistent Memory" that is fully coherent as if the
90	  memory were attached to the typical CPU memory controller. This is
91	  known as HDM "Host-managed Device Memory".
92
93	  Say 'y/m' to enable a driver that will attach to CXL.mem devices for
94	  memory expansion and control of HDM. See Chapter 9.13 in the CXL 2.0
95	  specification for a detailed description of HDM.
96
97	  If unsure say 'm'.
98
99config CXL_PORT
100	default CXL_BUS
101	tristate
102
103config CXL_SUSPEND
104	def_bool y
105	depends on SUSPEND && CXL_MEM
106
107config CXL_REGION
108	bool "CXL: Region Support"
109	default CXL_BUS
110	# For MAX_PHYSMEM_BITS
111	depends on SPARSEMEM
112	select MEMREGION
113	select GET_FREE_REGION
114	help
115	  Enable the CXL core to enumerate and provision CXL regions. A CXL
116	  region is defined by one or more CXL expanders that decode a given
117	  system-physical address range. For CXL regions established by
118	  platform-firmware this option enables memory error handling to
119	  identify the devices participating in a given interleaved memory
120	  range. Otherwise, platform-firmware managed CXL is enabled by being
121	  placed in the system address map and does not need a driver.
122
123	  If unsure say 'y'
124
125config CXL_REGION_INVALIDATION_TEST
126	bool "CXL: Region Cache Management Bypass (TEST)"
127	depends on CXL_REGION
128	help
129	  CXL Region management and security operations potentially invalidate
130	  the content of CPU caches without notifying those caches to
131	  invalidate the affected cachelines. The CXL Region driver attempts
132	  to invalidate caches when those events occur.  If that invalidation
133	  fails the region will fail to enable.  Reasons for cache
134	  invalidation failure are due to the CPU not providing a cache
135	  invalidation mechanism. For example usage of wbinvd is restricted to
136	  bare metal x86. However, for testing purposes toggling this option
137	  can disable that data integrity safety and proceed with enabling
138	  regions when there might be conflicting contents in the CPU cache.
139
140	  If unsure, or if this kernel is meant for production environments,
141	  say N.
142
143config CXL_PMU
144	tristate "CXL Performance Monitoring Unit"
145	default CXL_BUS
146	depends on PERF_EVENTS
147	help
148	  Support performance monitoring as defined in CXL rev 3.0
149	  section 13.2: Performance Monitoring. CXL components may have
150	  one or more CXL Performance Monitoring Units (CPMUs).
151
152	  Say 'y/m' to enable a driver that will attach to performance
153	  monitoring units and provide standard perf based interfaces.
154
155	  If unsure say 'm'.
156endif
157