1# SPDX-License-Identifier: GPL-2.0-only 2menuconfig MODULES 3 bool "Enable loadable module support" 4 modules 5 help 6 Kernel modules are small pieces of compiled code which can 7 be inserted in the running kernel, rather than being 8 permanently built into the kernel. You use the "modprobe" 9 tool to add (and sometimes remove) them. If you say Y here, 10 many parts of the kernel can be built as modules (by 11 answering M instead of Y where indicated): this is most 12 useful for infrequently used options which are not required 13 for booting. For more information, see the man pages for 14 modprobe, lsmod, modinfo, insmod and rmmod. 15 16 If you say Y here, you will need to run "make 17 modules_install" to put the modules under /lib/modules/ 18 where modprobe can find them (you may need to be root to do 19 this). 20 21 If unsure, say Y. 22 23if MODULES 24 25config MODULE_DEBUGFS 26 bool 27 28config MODULE_DEBUG 29 bool "Module debugging" 30 depends on DEBUG_FS 31 help 32 Allows you to enable / disable features which can help you debug 33 modules. You don't need these options on production systems. 34 35if MODULE_DEBUG 36 37config MODULE_STATS 38 bool "Module statistics" 39 depends on DEBUG_FS 40 select MODULE_DEBUGFS 41 help 42 This option allows you to maintain a record of module statistics. 43 For example, size of all modules, average size, text size, a list 44 of failed modules and the size for each of those. For failed 45 modules we keep track of modules which failed due to either the 46 existing module taking too long to load or that module was already 47 loaded. 48 49 You should enable this if you are debugging production loads 50 and want to see if userspace or the kernel is doing stupid things 51 with loading modules when it shouldn't or if you want to help 52 optimize userspace / kernel space module autoloading schemes. 53 You might want to do this because failed modules tend to use 54 up significant amount of memory, and so you'd be doing everyone a 55 favor in avoiding these failures proactively. 56 57 This functionality is also useful for those experimenting with 58 module .text ELF section optimization. 59 60 If unsure, say N. 61 62config MODULE_DEBUG_AUTOLOAD_DUPS 63 bool "Debug duplicate modules with auto-loading" 64 help 65 Module autoloading allows in-kernel code to request modules through 66 the *request_module*() API calls. This in turn just calls userspace 67 modprobe. Although modprobe checks to see if a module is already 68 loaded before trying to load a module there is a small time window in 69 which multiple duplicate requests can end up in userspace and multiple 70 modprobe calls race calling finit_module() around the same time for 71 duplicate modules. The finit_module() system call can consume in the 72 worst case more than twice the respective module size in virtual 73 memory for each duplicate module requests. Although duplicate module 74 requests are non-fatal virtual memory is a limited resource and each 75 duplicate module request ends up just unnecessarily straining virtual 76 memory. 77 78 This debugging facility will create pr_warn() splats for duplicate 79 module requests to help identify if module auto-loading may be the 80 culprit to your early boot virtual memory pressure. Since virtual 81 memory abuse caused by duplicate module requests could render a 82 system unusable this functionality will also converge races in 83 requests for the same module to a single request. You can boot with 84 the module.enable_dups_trace=1 kernel parameter to use WARN_ON() 85 instead of the pr_warn(). 86 87 If the first module request used request_module_nowait() we cannot 88 use that as the anchor to wait for duplicate module requests, since 89 users of request_module() do want a proper return value. If a call 90 for the same module happened earlier with request_module() though, 91 then a duplicate request_module_nowait() would be detected. The 92 non-wait request_module() call is synchronous and waits until modprobe 93 completes. Subsequent auto-loading requests for the same module do 94 not trigger a new finit_module() calls and do not strain virtual 95 memory, and so as soon as modprobe successfully completes we remove 96 tracking for duplicates for that module. 97 98 Enable this functionality to try to debug virtual memory abuse during 99 boot on systems which are failing to boot or if you suspect you may be 100 straining virtual memory during boot, and you want to identify if the 101 abuse was due to module auto-loading. These issues are currently only 102 known to occur on systems with many CPUs (over 400) and is likely the 103 result of udev issuing duplicate module requests for each CPU, and so 104 module auto-loading is not the culprit. There may very well still be 105 many duplicate module auto-loading requests which could be optimized 106 for and this debugging facility can be used to help identify them. 107 108 Only enable this for debugging system functionality, never have it 109 enabled on real systems. 110 111config MODULE_DEBUG_AUTOLOAD_DUPS_TRACE 112 bool "Force full stack trace when duplicates are found" 113 depends on MODULE_DEBUG_AUTOLOAD_DUPS 114 help 115 Enabling this will force a full stack trace for duplicate module 116 auto-loading requests using WARN_ON() instead of pr_warn(). You 117 should keep this disabled at all times unless you are a developer 118 and are doing a manual inspection and want to debug exactly why 119 these duplicates occur. 120 121endif # MODULE_DEBUG 122 123config MODULE_FORCE_LOAD 124 bool "Forced module loading" 125 default n 126 help 127 Allow loading of modules without version information (ie. modprobe 128 --force). Forced module loading sets the 'F' (forced) taint flag and 129 is usually a really bad idea. 130 131config MODULE_UNLOAD 132 bool "Module unloading" 133 help 134 Without this option you will not be able to unload any 135 modules (note that some modules may not be unloadable 136 anyway), which makes your kernel smaller, faster 137 and simpler. If unsure, say Y. 138 139config MODULE_FORCE_UNLOAD 140 bool "Forced module unloading" 141 depends on MODULE_UNLOAD 142 help 143 This option allows you to force a module to unload, even if the 144 kernel believes it is unsafe: the kernel will remove the module 145 without waiting for anyone to stop using it (using the -f option to 146 rmmod). This is mainly for kernel developers and desperate users. 147 If unsure, say N. 148 149config MODULE_UNLOAD_TAINT_TRACKING 150 bool "Tainted module unload tracking" 151 depends on MODULE_UNLOAD 152 select MODULE_DEBUGFS 153 help 154 This option allows you to maintain a record of each unloaded 155 module that tainted the kernel. In addition to displaying a 156 list of linked (or loaded) modules e.g. on detection of a bad 157 page (see bad_page()), the aforementioned details are also 158 shown. If unsure, say N. 159 160config MODVERSIONS 161 bool "Module versioning support" 162 help 163 Usually, you have to use modules compiled with your kernel. 164 Saying Y here makes it sometimes possible to use modules 165 compiled for different kernels, by adding enough information 166 to the modules to (hopefully) spot any changes which would 167 make them incompatible with the kernel you are running. If 168 unsure, say N. 169 170config ASM_MODVERSIONS 171 bool 172 default HAVE_ASM_MODVERSIONS && MODVERSIONS 173 help 174 This enables module versioning for exported symbols also from 175 assembly. This can be enabled only when the target architecture 176 supports it. 177 178config MODULE_SRCVERSION_ALL 179 bool "Source checksum for all modules" 180 help 181 Modules which contain a MODULE_VERSION get an extra "srcversion" 182 field inserted into their modinfo section, which contains a 183 sum of the source files which made it. This helps maintainers 184 see exactly which source was used to build a module (since 185 others sometimes change the module source without updating 186 the version). With this option, such a "srcversion" field 187 will be created for all modules. If unsure, say N. 188 189config MODULE_SIG 190 bool "Module signature verification" 191 select MODULE_SIG_FORMAT 192 help 193 Check modules for valid signatures upon load: the signature 194 is simply appended to the module. For more information see 195 <file:Documentation/admin-guide/module-signing.rst>. 196 197 Note that this option adds the OpenSSL development packages as a 198 kernel build dependency so that the signing tool can use its crypto 199 library. 200 201 You should enable this option if you wish to use either 202 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via 203 another LSM - otherwise unsigned modules will be loadable regardless 204 of the lockdown policy. 205 206 !!!WARNING!!! If you enable this option, you MUST make sure that the 207 module DOES NOT get stripped after being signed. This includes the 208 debuginfo strip done by some packagers (such as rpmbuild) and 209 inclusion into an initramfs that wants the module size reduced. 210 211config MODULE_SIG_FORCE 212 bool "Require modules to be validly signed" 213 depends on MODULE_SIG 214 help 215 Reject unsigned modules or signed modules for which we don't have a 216 key. Without this, such modules will simply taint the kernel. 217 218config MODULE_SIG_ALL 219 bool "Automatically sign all modules" 220 default y 221 depends on MODULE_SIG || IMA_APPRAISE_MODSIG 222 help 223 Sign all modules during make modules_install. Without this option, 224 modules must be signed manually, using the scripts/sign-file tool. 225 226comment "Do not forget to sign required modules with scripts/sign-file" 227 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL 228 229choice 230 prompt "Which hash algorithm should modules be signed with?" 231 depends on MODULE_SIG || IMA_APPRAISE_MODSIG 232 default MODULE_SIG_SHA512 233 help 234 This determines which sort of hashing algorithm will be used during 235 signature generation. This algorithm _must_ be built into the kernel 236 directly so that signature verification can take place. It is not 237 possible to load a signed module containing the algorithm to check 238 the signature on that module. 239 240config MODULE_SIG_SHA1 241 bool "Sign modules with SHA-1" 242 select CRYPTO_SHA1 243 244config MODULE_SIG_SHA224 245 bool "Sign modules with SHA-224" 246 select CRYPTO_SHA256 247 248config MODULE_SIG_SHA256 249 bool "Sign modules with SHA-256" 250 select CRYPTO_SHA256 251 252config MODULE_SIG_SHA384 253 bool "Sign modules with SHA-384" 254 select CRYPTO_SHA512 255 256config MODULE_SIG_SHA512 257 bool "Sign modules with SHA-512" 258 select CRYPTO_SHA512 259 260endchoice 261 262config MODULE_SIG_HASH 263 string 264 depends on MODULE_SIG || IMA_APPRAISE_MODSIG 265 default "sha1" if MODULE_SIG_SHA1 266 default "sha224" if MODULE_SIG_SHA224 267 default "sha256" if MODULE_SIG_SHA256 268 default "sha384" if MODULE_SIG_SHA384 269 default "sha512" if MODULE_SIG_SHA512 270 271choice 272 prompt "Module compression mode" 273 help 274 This option allows you to choose the algorithm which will be used to 275 compress modules when 'make modules_install' is run. (or, you can 276 choose to not compress modules at all.) 277 278 External modules will also be compressed in the same way during the 279 installation. 280 281 For modules inside an initrd or initramfs, it's more efficient to 282 compress the whole initrd or initramfs instead. 283 284 This is fully compatible with signed modules. 285 286 Please note that the tool used to load modules needs to support the 287 corresponding algorithm. module-init-tools MAY support gzip, and kmod 288 MAY support gzip, xz and zstd. 289 290 Your build system needs to provide the appropriate compression tool 291 to compress the modules. 292 293 If in doubt, select 'None'. 294 295config MODULE_COMPRESS_NONE 296 bool "None" 297 help 298 Do not compress modules. The installed modules are suffixed 299 with .ko. 300 301config MODULE_COMPRESS_GZIP 302 bool "GZIP" 303 help 304 Compress modules with GZIP. The installed modules are suffixed 305 with .ko.gz. 306 307config MODULE_COMPRESS_XZ 308 bool "XZ" 309 help 310 Compress modules with XZ. The installed modules are suffixed 311 with .ko.xz. 312 313config MODULE_COMPRESS_ZSTD 314 bool "ZSTD" 315 help 316 Compress modules with ZSTD. The installed modules are suffixed 317 with .ko.zst. 318 319endchoice 320 321config MODULE_DECOMPRESS 322 bool "Support in-kernel module decompression" 323 depends on MODULE_COMPRESS_GZIP || MODULE_COMPRESS_XZ || MODULE_COMPRESS_ZSTD 324 select ZLIB_INFLATE if MODULE_COMPRESS_GZIP 325 select XZ_DEC if MODULE_COMPRESS_XZ 326 select ZSTD_DECOMPRESS if MODULE_COMPRESS_ZSTD 327 help 328 329 Support for decompressing kernel modules by the kernel itself 330 instead of relying on userspace to perform this task. Useful when 331 load pinning security policy is enabled. 332 333 If unsure, say N. 334 335config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS 336 bool "Allow loading of modules with missing namespace imports" 337 help 338 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in 339 a namespace. A module that makes use of a symbol exported with such a 340 namespace is required to import the namespace via MODULE_IMPORT_NS(). 341 There is no technical reason to enforce correct namespace imports, 342 but it creates consistency between symbols defining namespaces and 343 users importing namespaces they make use of. This option relaxes this 344 requirement and lifts the enforcement when loading a module. 345 346 If unsure, say N. 347 348config MODPROBE_PATH 349 string "Path to modprobe binary" 350 default "/sbin/modprobe" 351 help 352 When kernel code requests a module, it does so by calling 353 the "modprobe" userspace utility. This option allows you to 354 set the path where that binary is found. This can be changed 355 at runtime via the sysctl file 356 /proc/sys/kernel/modprobe. Setting this to the empty string 357 removes the kernel's ability to request modules (but 358 userspace can still load modules explicitly). 359 360config TRIM_UNUSED_KSYMS 361 bool "Trim unused exported kernel symbols" if EXPERT 362 depends on !COMPILE_TEST 363 help 364 The kernel and some modules make many symbols available for 365 other modules to use via EXPORT_SYMBOL() and variants. Depending 366 on the set of modules being selected in your kernel configuration, 367 many of those exported symbols might never be used. 368 369 This option allows for unused exported symbols to be dropped from 370 the build. In turn, this provides the compiler more opportunities 371 (especially when using LTO) for optimizing the code and reducing 372 binary size. This might have some security advantages as well. 373 374 If unsure, or if you need to build out-of-tree modules, say N. 375 376config UNUSED_KSYMS_WHITELIST 377 string "Whitelist of symbols to keep in ksymtab" 378 depends on TRIM_UNUSED_KSYMS 379 help 380 By default, all unused exported symbols will be un-exported from the 381 build when TRIM_UNUSED_KSYMS is selected. 382 383 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept 384 exported at all times, even in absence of in-tree users. The value to 385 set here is the path to a text file containing the list of symbols, 386 one per line. The path can be absolute, or relative to the kernel 387 source tree. 388 389config MODULES_TREE_LOOKUP 390 def_bool y 391 depends on PERF_EVENTS || TRACING || CFI_CLANG 392 393endif # MODULES 394