1 # SPDX-License-Identifier: GPL-2.0
3 config TRACE_IRQFLAGS_SUPPORT
6 config EARLY_PRINTK_USB
9 config X86_VERBOSE_BOOTUP
10 bool "Enable verbose x86 bootup info messages"
13 Enables the informational output from the decompression stage
14 (e.g. bzImage) of the boot. If you disable this you will still
15 see errors. Disable this if you want silent bootup.
18 bool "Early printk" if EXPERT
21 Write kernel log output directly into the VGA buffer or to a serial
24 This is useful for kernel debugging when your machine crashes very
25 early before the console code is initialized. For normal operation
26 it is not recommended because it looks ugly and doesn't cooperate
27 with klogd/syslogd or the X server. You should normally say N here,
28 unless you want to debug such a crash.
30 config EARLY_PRINTK_DBGP
31 bool "Early printk via EHCI debug port"
32 depends on EARLY_PRINTK && PCI
33 select EARLY_PRINTK_USB
35 Write kernel log output directly into the EHCI debug port.
37 This is useful for kernel debugging when your machine crashes very
38 early before the console code is initialized. For normal operation
39 it is not recommended because it looks ugly and doesn't cooperate
40 with klogd/syslogd or the X server. You should normally say N here,
41 unless you want to debug such a crash. You need usb debug device.
43 config EARLY_PRINTK_USB_XDBC
44 bool "Early printk via the xHCI debug port"
45 depends on EARLY_PRINTK && PCI
46 select EARLY_PRINTK_USB
48 Write kernel log output directly into the xHCI debug port.
50 One use for this feature is kernel debugging, for example when your
51 machine crashes very early before the regular console code is
52 initialized. Other uses include simpler, lockless logging instead of
53 a full-blown printk console driver + klogd.
55 For normal production environments this is normally not recommended,
56 because it doesn't feed events into klogd/syslogd and doesn't try to
57 print anything on the screen.
59 You should normally say N here, unless you want to debug early
60 crashes or need a very simple printk logging facility.
65 config X86_PTDUMP_CORE
69 tristate "Export kernel pagetable layout to userspace via debugfs"
70 depends on DEBUG_KERNEL
72 select X86_PTDUMP_CORE
74 Say Y here if you want to show the kernel pagetable layout in a
75 debugfs file. This information is only useful for kernel developers
76 who are working in architecture specific areas of the kernel.
77 It is probably not a good idea to enable this feature in a production
82 bool "Dump the EFI pagetable"
84 select X86_PTDUMP_CORE
86 Enable this if you want to dump the EFI page table before
87 enabling virtual mode. This can be used to debug miscellaneous
88 issues with the mapping of the EFI runtime regions into that
92 bool "Warn on W+X mappings at boot"
93 select X86_PTDUMP_CORE
95 Generate a warning if any W+X mappings are found at boot.
97 This is useful for discovering cases where the kernel is leaving
98 W+X mappings after applying NX, as such mappings are a security risk.
100 Look for a message in dmesg output like this:
102 x86/mm: Checked W+X mappings: passed, no W+X pages found.
104 or like this, if the check failed:
106 x86/mm: Checked W+X mappings: FAILED, <N> W+X pages found.
108 Note that even if the check fails, your kernel is possibly
109 still fine, as W+X mappings are not a security hole in
110 themselves, what they do is that they make the exploitation
111 of other unfixed kernel bugs easier.
113 There is no runtime or memory usage effect of this option
114 once the kernel has booted up - it's a one time check.
116 If in doubt, say "Y".
120 bool "Enable doublefault exception handler" if EXPERT
122 This option allows trapping of rare doublefault exceptions that
123 would otherwise cause a system to silently reboot. Disabling this
124 option saves about 4k and might cause you much additional grey
127 config DEBUG_TLBFLUSH
128 bool "Set upper limit of TLB entries to flush one-by-one"
129 depends on DEBUG_KERNEL
134 This option allows the user to tune the amount of TLB entries the
135 kernel flushes one-by-one instead of doing a full TLB flush. In
136 certain situations, the former is cheaper. This is controlled by the
137 tlb_flushall_shift knob under /sys/kernel/debug/x86. If you set it
138 to -1, the code flushes the whole TLB unconditionally. Otherwise,
139 for positive values of it, the kernel will use single TLB entry
140 invalidating instructions according to the following formula:
142 flush_entries <= active_tlb_entries / 2^tlb_flushall_shift
144 If in doubt, say "N".
147 bool "Enable IOMMU debugging"
148 depends on GART_IOMMU && DEBUG_KERNEL
151 Force the IOMMU to on even when you have less than 4GB of
152 memory and add debugging code. On overflow always panic. And
153 allow to enable IOMMU leak tracing. Can be disabled at boot
154 time with iommu=noforce. This will also enable scatter gather
155 list merging. Currently not recommended for production
156 code. When you use it make sure you have a big enough
157 IOMMU/AGP aperture. Most of the options enabled by this can
158 be set more finegrained using the iommu= command line
159 options. See Documentation/x86/x86_64/boot-options.rst for more
163 bool "IOMMU leak tracing"
164 depends on IOMMU_DEBUG && DMA_API_DEBUG
166 Add a simple leak tracer to the IOMMU code. This is useful when you
167 are debugging a buggy device driver that leaks IOMMU mappings.
169 config HAVE_MMIOTRACE_SUPPORT
172 config X86_DECODER_SELFTEST
173 bool "x86 instruction decoder selftest"
174 depends on DEBUG_KERNEL && INSTRUCTION_DECODER
175 depends on !COMPILE_TEST
177 Perform x86 instruction decoder selftests at build time.
178 This option is useful for checking the sanity of x86 instruction
183 prompt "IO delay type"
184 default IO_DELAY_0X80
187 bool "port 0x80 based port-IO delay [recommended]"
189 This is the traditional Linux IO delay used for in/out_p.
190 It is the most tested hence safest selection here.
193 bool "port 0xed based port-IO delay"
195 Use port 0xed as the IO delay. This frees up port 0x80 which is
196 often used as a hardware-debug port.
198 config IO_DELAY_UDELAY
199 bool "udelay based port-IO delay"
201 Use udelay(2) as the IO delay method. This provides the delay
202 while not having any side-effect on the IO port space.
205 bool "no port-IO delay"
207 No port-IO delay. Will break on old boxes that require port-IO
208 delay for certain operations. Should work on most new machines.
212 config DEBUG_BOOT_PARAMS
213 bool "Debug boot parameters"
214 depends on DEBUG_KERNEL
217 This option will cause struct boot_params to be exported via debugfs.
220 bool "CPA self-test code"
221 depends on DEBUG_KERNEL
223 Do change_page_attr() self-tests every 30 seconds.
226 bool "Debug low-level entry code"
227 depends on DEBUG_KERNEL
229 This option enables sanity checks in x86's low-level entry code.
230 Some of these sanity checks may slow down kernel entries and
231 exits or otherwise impact performance.
235 config DEBUG_NMI_SELFTEST
237 depends on DEBUG_KERNEL && X86_LOCAL_APIC
239 Enabling this option turns on a quick NMI selftest to verify
240 that the NMI behaves correctly.
242 This might help diagnose strange hangs that rely on NMI to
247 config DEBUG_IMR_SELFTEST
248 bool "Isolated Memory Region self test"
251 This option enables automated sanity testing of the IMR code.
252 Some simple tests are run to verify IMR bounds checking, alignment
253 and overlapping. This option is really only useful if you are
254 debugging an IMR memory map or are modifying the IMR code and want to
257 If unsure say N here.
260 bool "Debug the x86 FPU code"
261 depends on DEBUG_KERNEL
264 If this option is enabled then there will be extra sanity
265 checks and (boot time) debug printouts added to the kernel.
266 This debugging adds some small amount of runtime overhead
271 config PUNIT_ATOM_DEBUG
272 tristate "ATOM Punit debug driver"
277 This is a debug driver, which gets the power states
278 of all Punit North Complex devices. The power states of
279 each device is exposed as part of the debugfs interface.
280 The current power state can be read from
281 /sys/kernel/debug/punit_atom/dev_power_state
284 prompt "Choose kernel unwinder"
285 default UNWINDER_ORC if X86_64
286 default UNWINDER_FRAME_POINTER if X86_32
288 This determines which method will be used for unwinding kernel stack
289 traces for panics, oopses, bugs, warnings, perf, /proc/<pid>/stack,
290 livepatch, lockdep, and more.
295 select STACK_VALIDATION
297 This option enables the ORC (Oops Rewind Capability) unwinder for
298 unwinding kernel stack traces. It uses a custom data format which is
299 a simplified version of the DWARF Call Frame Information standard.
301 This unwinder is more accurate across interrupt entry frames than the
302 frame pointer unwinder. It also enables a 5-10% performance
303 improvement across the entire kernel compared to frame pointers.
305 Enabling this option will increase the kernel's runtime memory usage
306 by roughly 2-4MB, depending on your kernel config.
308 config UNWINDER_FRAME_POINTER
309 bool "Frame pointer unwinder"
312 This option enables the frame pointer unwinder for unwinding kernel
315 The unwinder itself is fast and it uses less RAM than the ORC
316 unwinder, but the kernel text size will grow by ~3% and the kernel's
317 overall performance will degrade by roughly 5-10%.
319 This option is recommended if you want to use the livepatch
320 consistency model, as this is currently the only way to get a
321 reliable stack trace (CONFIG_HAVE_RELIABLE_STACKTRACE).
323 config UNWINDER_GUESS
324 bool "Guess unwinder"
326 depends on !STACKDEPOT
328 This option enables the "guess" unwinder for unwinding kernel stack
329 traces. It scans the stack and reports every kernel text address it
330 finds. Some of the addresses it reports may be incorrect.
332 While this option often produces false positives, it can still be
333 useful in many cases. Unlike the other unwinders, it has no runtime
339 depends on !UNWINDER_ORC && !UNWINDER_GUESS