1menu "Kernel hacking" 2 3config TRACE_IRQFLAGS_SUPPORT 4 def_bool y 5 6source "lib/Kconfig.debug" 7 8config X86_VERBOSE_BOOTUP 9 bool "Enable verbose x86 bootup info messages" 10 default y 11 ---help--- 12 Enables the informational output from the decompression stage 13 (e.g. bzImage) of the boot. If you disable this you will still 14 see errors. Disable this if you want silent bootup. 15 16config EARLY_PRINTK 17 bool "Early printk" if EXPERT 18 default y 19 ---help--- 20 Write kernel log output directly into the VGA buffer or to a serial 21 port. 22 23 This is useful for kernel debugging when your machine crashes very 24 early before the console code is initialized. For normal operation 25 it is not recommended because it looks ugly and doesn't cooperate 26 with klogd/syslogd or the X server. You should normally N here, 27 unless you want to debug such a crash. 28 29config EARLY_PRINTK_DBGP 30 bool "Early printk via EHCI debug port" 31 depends on EARLY_PRINTK && PCI 32 ---help--- 33 Write kernel log output directly into the EHCI debug port. 34 35 This is useful for kernel debugging when your machine crashes very 36 early before the console code is initialized. For normal operation 37 it is not recommended because it looks ugly and doesn't cooperate 38 with klogd/syslogd or the X server. You should normally N here, 39 unless you want to debug such a crash. You need usb debug device. 40 41config EARLY_PRINTK_EFI 42 bool "Early printk via the EFI framebuffer" 43 depends on EFI && EARLY_PRINTK 44 select FONT_SUPPORT 45 ---help--- 46 Write kernel log output directly into the EFI framebuffer. 47 48 This is useful for kernel debugging when your machine crashes very 49 early before the console code is initialized. 50 51config X86_PTDUMP_CORE 52 def_bool n 53 54config X86_PTDUMP 55 tristate "Export kernel pagetable layout to userspace via debugfs" 56 depends on DEBUG_KERNEL 57 select DEBUG_FS 58 select X86_PTDUMP_CORE 59 ---help--- 60 Say Y here if you want to show the kernel pagetable layout in a 61 debugfs file. This information is only useful for kernel developers 62 who are working in architecture specific areas of the kernel. 63 It is probably not a good idea to enable this feature in a production 64 kernel. 65 If in doubt, say "N" 66 67config EFI_PGT_DUMP 68 bool "Dump the EFI pagetable" 69 depends on EFI 70 select X86_PTDUMP_CORE 71 ---help--- 72 Enable this if you want to dump the EFI page table before 73 enabling virtual mode. This can be used to debug miscellaneous 74 issues with the mapping of the EFI runtime regions into that 75 table. 76 77config DEBUG_WX 78 bool "Warn on W+X mappings at boot" 79 select X86_PTDUMP_CORE 80 ---help--- 81 Generate a warning if any W+X mappings are found at boot. 82 83 This is useful for discovering cases where the kernel is leaving 84 W+X mappings after applying NX, as such mappings are a security risk. 85 86 Look for a message in dmesg output like this: 87 88 x86/mm: Checked W+X mappings: passed, no W+X pages found. 89 90 or like this, if the check failed: 91 92 x86/mm: Checked W+X mappings: FAILED, <N> W+X pages found. 93 94 Note that even if the check fails, your kernel is possibly 95 still fine, as W+X mappings are not a security hole in 96 themselves, what they do is that they make the exploitation 97 of other unfixed kernel bugs easier. 98 99 There is no runtime or memory usage effect of this option 100 once the kernel has booted up - it's a one time check. 101 102 If in doubt, say "Y". 103 104config DOUBLEFAULT 105 default y 106 bool "Enable doublefault exception handler" if EXPERT 107 ---help--- 108 This option allows trapping of rare doublefault exceptions that 109 would otherwise cause a system to silently reboot. Disabling this 110 option saves about 4k and might cause you much additional grey 111 hair. 112 113config DEBUG_TLBFLUSH 114 bool "Set upper limit of TLB entries to flush one-by-one" 115 depends on DEBUG_KERNEL 116 ---help--- 117 118 X86-only for now. 119 120 This option allows the user to tune the amount of TLB entries the 121 kernel flushes one-by-one instead of doing a full TLB flush. In 122 certain situations, the former is cheaper. This is controlled by the 123 tlb_flushall_shift knob under /sys/kernel/debug/x86. If you set it 124 to -1, the code flushes the whole TLB unconditionally. Otherwise, 125 for positive values of it, the kernel will use single TLB entry 126 invalidating instructions according to the following formula: 127 128 flush_entries <= active_tlb_entries / 2^tlb_flushall_shift 129 130 If in doubt, say "N". 131 132config IOMMU_DEBUG 133 bool "Enable IOMMU debugging" 134 depends on GART_IOMMU && DEBUG_KERNEL 135 depends on X86_64 136 ---help--- 137 Force the IOMMU to on even when you have less than 4GB of 138 memory and add debugging code. On overflow always panic. And 139 allow to enable IOMMU leak tracing. Can be disabled at boot 140 time with iommu=noforce. This will also enable scatter gather 141 list merging. Currently not recommended for production 142 code. When you use it make sure you have a big enough 143 IOMMU/AGP aperture. Most of the options enabled by this can 144 be set more finegrained using the iommu= command line 145 options. See Documentation/x86/x86_64/boot-options.txt for more 146 details. 147 148config IOMMU_STRESS 149 bool "Enable IOMMU stress-test mode" 150 ---help--- 151 This option disables various optimizations in IOMMU related 152 code to do real stress testing of the IOMMU code. This option 153 will cause a performance drop and should only be enabled for 154 testing. 155 156config IOMMU_LEAK 157 bool "IOMMU leak tracing" 158 depends on IOMMU_DEBUG && DMA_API_DEBUG 159 ---help--- 160 Add a simple leak tracer to the IOMMU code. This is useful when you 161 are debugging a buggy device driver that leaks IOMMU mappings. 162 163config HAVE_MMIOTRACE_SUPPORT 164 def_bool y 165 166config X86_DECODER_SELFTEST 167 bool "x86 instruction decoder selftest" 168 depends on DEBUG_KERNEL && KPROBES 169 depends on !COMPILE_TEST 170 ---help--- 171 Perform x86 instruction decoder selftests at build time. 172 This option is useful for checking the sanity of x86 instruction 173 decoder code. 174 If unsure, say "N". 175 176# 177# IO delay types: 178# 179 180config IO_DELAY_TYPE_0X80 181 int 182 default "0" 183 184config IO_DELAY_TYPE_0XED 185 int 186 default "1" 187 188config IO_DELAY_TYPE_UDELAY 189 int 190 default "2" 191 192config IO_DELAY_TYPE_NONE 193 int 194 default "3" 195 196choice 197 prompt "IO delay type" 198 default IO_DELAY_0X80 199 200config IO_DELAY_0X80 201 bool "port 0x80 based port-IO delay [recommended]" 202 ---help--- 203 This is the traditional Linux IO delay used for in/out_p. 204 It is the most tested hence safest selection here. 205 206config IO_DELAY_0XED 207 bool "port 0xed based port-IO delay" 208 ---help--- 209 Use port 0xed as the IO delay. This frees up port 0x80 which is 210 often used as a hardware-debug port. 211 212config IO_DELAY_UDELAY 213 bool "udelay based port-IO delay" 214 ---help--- 215 Use udelay(2) as the IO delay method. This provides the delay 216 while not having any side-effect on the IO port space. 217 218config IO_DELAY_NONE 219 bool "no port-IO delay" 220 ---help--- 221 No port-IO delay. Will break on old boxes that require port-IO 222 delay for certain operations. Should work on most new machines. 223 224endchoice 225 226if IO_DELAY_0X80 227config DEFAULT_IO_DELAY_TYPE 228 int 229 default IO_DELAY_TYPE_0X80 230endif 231 232if IO_DELAY_0XED 233config DEFAULT_IO_DELAY_TYPE 234 int 235 default IO_DELAY_TYPE_0XED 236endif 237 238if IO_DELAY_UDELAY 239config DEFAULT_IO_DELAY_TYPE 240 int 241 default IO_DELAY_TYPE_UDELAY 242endif 243 244if IO_DELAY_NONE 245config DEFAULT_IO_DELAY_TYPE 246 int 247 default IO_DELAY_TYPE_NONE 248endif 249 250config DEBUG_BOOT_PARAMS 251 bool "Debug boot parameters" 252 depends on DEBUG_KERNEL 253 depends on DEBUG_FS 254 ---help--- 255 This option will cause struct boot_params to be exported via debugfs. 256 257config CPA_DEBUG 258 bool "CPA self-test code" 259 depends on DEBUG_KERNEL 260 ---help--- 261 Do change_page_attr() self-tests every 30 seconds. 262 263config OPTIMIZE_INLINING 264 bool "Allow gcc to uninline functions marked 'inline'" 265 ---help--- 266 This option determines if the kernel forces gcc to inline the functions 267 developers have marked 'inline'. Doing so takes away freedom from gcc to 268 do what it thinks is best, which is desirable for the gcc 3.x series of 269 compilers. The gcc 4.x series have a rewritten inlining algorithm and 270 enabling this option will generate a smaller kernel there. Hopefully 271 this algorithm is so good that allowing gcc 4.x and above to make the 272 decision will become the default in the future. Until then this option 273 is there to test gcc for this. 274 275 If unsure, say N. 276 277config DEBUG_ENTRY 278 bool "Debug low-level entry code" 279 depends on DEBUG_KERNEL 280 ---help--- 281 This option enables sanity checks in x86's low-level entry code. 282 Some of these sanity checks may slow down kernel entries and 283 exits or otherwise impact performance. 284 285 This is currently used to help test NMI code. 286 287 If unsure, say N. 288 289config DEBUG_NMI_SELFTEST 290 bool "NMI Selftest" 291 depends on DEBUG_KERNEL && X86_LOCAL_APIC 292 ---help--- 293 Enabling this option turns on a quick NMI selftest to verify 294 that the NMI behaves correctly. 295 296 This might help diagnose strange hangs that rely on NMI to 297 function properly. 298 299 If unsure, say N. 300 301config DEBUG_IMR_SELFTEST 302 bool "Isolated Memory Region self test" 303 default n 304 depends on INTEL_IMR 305 ---help--- 306 This option enables automated sanity testing of the IMR code. 307 Some simple tests are run to verify IMR bounds checking, alignment 308 and overlapping. This option is really only useful if you are 309 debugging an IMR memory map or are modifying the IMR code and want to 310 test your changes. 311 312 If unsure say N here. 313 314config X86_DEBUG_FPU 315 bool "Debug the x86 FPU code" 316 depends on DEBUG_KERNEL 317 default y 318 ---help--- 319 If this option is enabled then there will be extra sanity 320 checks and (boot time) debug printouts added to the kernel. 321 This debugging adds some small amount of runtime overhead 322 to the kernel. 323 324 If unsure, say N. 325 326config PUNIT_ATOM_DEBUG 327 tristate "ATOM Punit debug driver" 328 select DEBUG_FS 329 select IOSF_MBI 330 ---help--- 331 This is a debug driver, which gets the power states 332 of all Punit North Complex devices. The power states of 333 each device is exposed as part of the debugfs interface. 334 The current power state can be read from 335 /sys/kernel/debug/punit_atom/dev_power_state 336 337endmenu 338