xref: /linux/arch/Kconfig (revision ec83db0f78cd44c3b586ec1c3a348d1a8a389797)
1fb32e03fSMathieu Desnoyers#
2fb32e03fSMathieu Desnoyers# General architecture dependent options
3fb32e03fSMathieu Desnoyers#
4125e5645SMathieu Desnoyers
5125e5645SMathieu Desnoyersconfig OPROFILE
6b309a294SRobert Richter	tristate "OProfile system profiling"
7125e5645SMathieu Desnoyers	depends on PROFILING
8125e5645SMathieu Desnoyers	depends on HAVE_OPROFILE
9d69d59f4SIngo Molnar	select RING_BUFFER
109a5963ebSChristian Borntraeger	select RING_BUFFER_ALLOW_SWAP
11125e5645SMathieu Desnoyers	help
12125e5645SMathieu Desnoyers	  OProfile is a profiling system capable of profiling the
13125e5645SMathieu Desnoyers	  whole system, include the kernel, kernel modules, libraries,
14125e5645SMathieu Desnoyers	  and applications.
15125e5645SMathieu Desnoyers
16125e5645SMathieu Desnoyers	  If unsure, say N.
17125e5645SMathieu Desnoyers
184d4036e0SJason Yehconfig OPROFILE_EVENT_MULTIPLEX
194d4036e0SJason Yeh	bool "OProfile multiplexing support (EXPERIMENTAL)"
204d4036e0SJason Yeh	default n
214d4036e0SJason Yeh	depends on OPROFILE && X86
224d4036e0SJason Yeh	help
234d4036e0SJason Yeh	  The number of hardware counters is limited. The multiplexing
244d4036e0SJason Yeh	  feature enables OProfile to gather more events than counters
254d4036e0SJason Yeh	  are provided by the hardware. This is realized by switching
264d4036e0SJason Yeh	  between events at an user specified time interval.
274d4036e0SJason Yeh
284d4036e0SJason Yeh	  If unsure, say N.
294d4036e0SJason Yeh
30125e5645SMathieu Desnoyersconfig HAVE_OPROFILE
319ba16087SJan Beulich	bool
32125e5645SMathieu Desnoyers
33dcfce4a0SRobert Richterconfig OPROFILE_NMI_TIMER
34dcfce4a0SRobert Richter	def_bool y
35dcfce4a0SRobert Richter	depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI
36dcfce4a0SRobert Richter
37125e5645SMathieu Desnoyersconfig KPROBES
38125e5645SMathieu Desnoyers	bool "Kprobes"
3905ed160eSMasami Hiramatsu	depends on MODULES
40125e5645SMathieu Desnoyers	depends on HAVE_KPROBES
4105ed160eSMasami Hiramatsu	select KALLSYMS
42125e5645SMathieu Desnoyers	help
43125e5645SMathieu Desnoyers	  Kprobes allows you to trap at almost any kernel address and
44125e5645SMathieu Desnoyers	  execute a callback function.  register_kprobe() establishes
45125e5645SMathieu Desnoyers	  a probepoint and specifies the callback.  Kprobes is useful
46125e5645SMathieu Desnoyers	  for kernel debugging, non-intrusive instrumentation and testing.
47125e5645SMathieu Desnoyers	  If in doubt, say "N".
48125e5645SMathieu Desnoyers
4945f81b1cSSteven Rostedtconfig JUMP_LABEL
50c5905afbSIngo Molnar       bool "Optimize very unlikely/likely branches"
5145f81b1cSSteven Rostedt       depends on HAVE_ARCH_JUMP_LABEL
5245f81b1cSSteven Rostedt       help
53c5905afbSIngo Molnar         This option enables a transparent branch optimization that
54c5905afbSIngo Molnar	 makes certain almost-always-true or almost-always-false branch
55c5905afbSIngo Molnar	 conditions even cheaper to execute within the kernel.
5645f81b1cSSteven Rostedt
57c5905afbSIngo Molnar	 Certain performance-sensitive kernel code, such as trace points,
58c5905afbSIngo Molnar	 scheduler functionality, networking code and KVM have such
59c5905afbSIngo Molnar	 branches and include support for this optimization technique.
60c5905afbSIngo Molnar
61c5905afbSIngo Molnar         If it is detected that the compiler has support for "asm goto",
62c5905afbSIngo Molnar	 the kernel will compile such branches with just a nop
63c5905afbSIngo Molnar	 instruction. When the condition flag is toggled to true, the
64c5905afbSIngo Molnar	 nop will be converted to a jump instruction to execute the
65c5905afbSIngo Molnar	 conditional block of instructions.
66c5905afbSIngo Molnar
67c5905afbSIngo Molnar	 This technique lowers overhead and stress on the branch prediction
68c5905afbSIngo Molnar	 of the processor and generally makes the kernel faster. The update
69c5905afbSIngo Molnar	 of the condition is slower, but those are always very rare.
70c5905afbSIngo Molnar
71c5905afbSIngo Molnar	 ( On 32-bit x86, the necessary options added to the compiler
72c5905afbSIngo Molnar	   flags may increase the size of the kernel slightly. )
7345f81b1cSSteven Rostedt
74afd66255SMasami Hiramatsuconfig OPTPROBES
755cc718b9SMasami Hiramatsu	def_bool y
765cc718b9SMasami Hiramatsu	depends on KPROBES && HAVE_OPTPROBES
77afd66255SMasami Hiramatsu	depends on !PREEMPT
78afd66255SMasami Hiramatsu
792b144498SSrikar Dronamrajuconfig UPROBES
807b2d81d4SIngo Molnar	bool "Transparent user-space probes (EXPERIMENTAL)"
81*ec83db0fSSrikar Dronamraju	depends on UPROBE_EVENT && PERF_EVENTS
822b144498SSrikar Dronamraju	default n
832b144498SSrikar Dronamraju	help
847b2d81d4SIngo Molnar	  Uprobes is the user-space counterpart to kprobes: they
857b2d81d4SIngo Molnar	  enable instrumentation applications (such as 'perf probe')
867b2d81d4SIngo Molnar	  to establish unintrusive probes in user-space binaries and
877b2d81d4SIngo Molnar	  libraries, by executing handler functions when the probes
887b2d81d4SIngo Molnar	  are hit by user-space applications.
897b2d81d4SIngo Molnar
907b2d81d4SIngo Molnar	  ( These probes come in the form of single-byte breakpoints,
917b2d81d4SIngo Molnar	    managed by the kernel and kept transparent to the probed
927b2d81d4SIngo Molnar	    application. )
932b144498SSrikar Dronamraju
942b144498SSrikar Dronamraju	  If in doubt, say "N".
952b144498SSrikar Dronamraju
9658340a07SJohannes Bergconfig HAVE_EFFICIENT_UNALIGNED_ACCESS
979ba16087SJan Beulich	bool
9858340a07SJohannes Berg	help
9958340a07SJohannes Berg	  Some architectures are unable to perform unaligned accesses
10058340a07SJohannes Berg	  without the use of get_unaligned/put_unaligned. Others are
10158340a07SJohannes Berg	  unable to perform such accesses efficiently (e.g. trap on
10258340a07SJohannes Berg	  unaligned access and require fixing it up in the exception
10358340a07SJohannes Berg	  handler.)
10458340a07SJohannes Berg
10558340a07SJohannes Berg	  This symbol should be selected by an architecture if it can
10658340a07SJohannes Berg	  perform unaligned accesses efficiently to allow different
10758340a07SJohannes Berg	  code paths to be selected for these cases. Some network
10858340a07SJohannes Berg	  drivers, for example, could opt to not fix up alignment
10958340a07SJohannes Berg	  problems with received packets if doing so would not help
11058340a07SJohannes Berg	  much.
11158340a07SJohannes Berg
11258340a07SJohannes Berg	  See Documentation/unaligned-memory-access.txt for more
11358340a07SJohannes Berg	  information on the topic of unaligned memory accesses.
11458340a07SJohannes Berg
1151a94bc34SHeiko Carstensconfig HAVE_SYSCALL_WRAPPERS
1161a94bc34SHeiko Carstens	bool
1171a94bc34SHeiko Carstens
1189edddaa2SAnanth N Mavinakayanahalliconfig KRETPROBES
1199edddaa2SAnanth N Mavinakayanahalli	def_bool y
1209edddaa2SAnanth N Mavinakayanahalli	depends on KPROBES && HAVE_KRETPROBES
1219edddaa2SAnanth N Mavinakayanahalli
1227c68af6eSAvi Kivityconfig USER_RETURN_NOTIFIER
1237c68af6eSAvi Kivity	bool
1247c68af6eSAvi Kivity	depends on HAVE_USER_RETURN_NOTIFIER
1257c68af6eSAvi Kivity	help
1267c68af6eSAvi Kivity	  Provide a kernel-internal notification when a cpu is about to
1277c68af6eSAvi Kivity	  switch to user mode.
1287c68af6eSAvi Kivity
12928b2ee20SRik van Rielconfig HAVE_IOREMAP_PROT
1309ba16087SJan Beulich	bool
13128b2ee20SRik van Riel
132125e5645SMathieu Desnoyersconfig HAVE_KPROBES
1339ba16087SJan Beulich	bool
1349edddaa2SAnanth N Mavinakayanahalli
1359edddaa2SAnanth N Mavinakayanahalliconfig HAVE_KRETPROBES
1369ba16087SJan Beulich	bool
13774bc7ceeSArthur Kepner
138afd66255SMasami Hiramatsuconfig HAVE_OPTPROBES
139afd66255SMasami Hiramatsu	bool
140d314d74cSCong Wang
141d314d74cSCong Wangconfig HAVE_NMI_WATCHDOG
142d314d74cSCong Wang	bool
1431f5a4ad9SRoland McGrath#
1441f5a4ad9SRoland McGrath# An arch should select this if it provides all these things:
1451f5a4ad9SRoland McGrath#
1461f5a4ad9SRoland McGrath#	task_pt_regs()		in asm/processor.h or asm/ptrace.h
1471f5a4ad9SRoland McGrath#	arch_has_single_step()	if there is hardware single-step support
1481f5a4ad9SRoland McGrath#	arch_has_block_step()	if there is hardware block-step support
1491f5a4ad9SRoland McGrath#	asm/syscall.h		supplying asm-generic/syscall.h interface
1501f5a4ad9SRoland McGrath#	linux/regset.h		user_regset interfaces
1511f5a4ad9SRoland McGrath#	CORE_DUMP_USE_REGSET	#define'd in linux/elf.h
1521f5a4ad9SRoland McGrath#	TIF_SYSCALL_TRACE	calls tracehook_report_syscall_{entry,exit}
1531f5a4ad9SRoland McGrath#	TIF_NOTIFY_RESUME	calls tracehook_notify_resume()
1541f5a4ad9SRoland McGrath#	signal delivery		calls tracehook_signal_handler()
1551f5a4ad9SRoland McGrath#
1561f5a4ad9SRoland McGrathconfig HAVE_ARCH_TRACEHOOK
1579ba16087SJan Beulich	bool
1581f5a4ad9SRoland McGrath
15974bc7ceeSArthur Kepnerconfig HAVE_DMA_ATTRS
1609ba16087SJan Beulich	bool
1613d442233SJens Axboe
1623d442233SJens Axboeconfig USE_GENERIC_SMP_HELPERS
1639ba16087SJan Beulich	bool
1649483a578SDavid Brownell
165f850c30cSHeiko Carstensconfig HAVE_REGS_AND_STACK_ACCESS_API
166f850c30cSHeiko Carstens	bool
167e01292b1SHeiko Carstens	help
168e01292b1SHeiko Carstens	  This symbol should be selected by an architecure if it supports
169e01292b1SHeiko Carstens	  the API needed to access registers and stack entries from pt_regs,
170e01292b1SHeiko Carstens	  declared in asm/ptrace.h
171e01292b1SHeiko Carstens	  For example the kprobes-based event tracer needs this API.
172f850c30cSHeiko Carstens
1739483a578SDavid Brownellconfig HAVE_CLK
1749ba16087SJan Beulich	bool
1759483a578SDavid Brownell	help
1769483a578SDavid Brownell	  The <linux/clk.h> calls support software clock gating and
1779483a578SDavid Brownell	  thus are a key power management tool on many systems.
1789483a578SDavid Brownell
1795ee00bd4SJoerg Roedelconfig HAVE_DMA_API_DEBUG
1805ee00bd4SJoerg Roedel	bool
18136cd3c9fSHeiko Carstens
18262a038d3SK.Prasadconfig HAVE_HW_BREAKPOINT
18362a038d3SK.Prasad	bool
18499e8c5a3SFrederic Weisbecker	depends on PERF_EVENTS
18562a038d3SK.Prasad
1860102752eSFrederic Weisbeckerconfig HAVE_MIXED_BREAKPOINTS_REGS
1870102752eSFrederic Weisbecker	bool
1880102752eSFrederic Weisbecker	depends on HAVE_HW_BREAKPOINT
1890102752eSFrederic Weisbecker	help
1900102752eSFrederic Weisbecker	  Depending on the arch implementation of hardware breakpoints,
1910102752eSFrederic Weisbecker	  some of them have separate registers for data and instruction
1920102752eSFrederic Weisbecker	  breakpoints addresses, others have mixed registers to store
1930102752eSFrederic Weisbecker	  them but define the access type in a control register.
1940102752eSFrederic Weisbecker	  Select this option if your arch implements breakpoints under the
1950102752eSFrederic Weisbecker	  latter fashion.
1960102752eSFrederic Weisbecker
1977c68af6eSAvi Kivityconfig HAVE_USER_RETURN_NOTIFIER
1987c68af6eSAvi Kivity	bool
199a1922ed6SIngo Molnar
200c01d4323SFrederic Weisbeckerconfig HAVE_PERF_EVENTS_NMI
201c01d4323SFrederic Weisbecker	bool
20223637d47SFrederic Weisbecker	help
20323637d47SFrederic Weisbecker	  System hardware can generate an NMI using the perf event
20423637d47SFrederic Weisbecker	  subsystem.  Also has support for calculating CPU cycle events
20523637d47SFrederic Weisbecker	  to determine how many clock cycles in a given period.
206c01d4323SFrederic Weisbecker
207bf5438fcSJason Baronconfig HAVE_ARCH_JUMP_LABEL
208bf5438fcSJason Baron	bool
209bf5438fcSJason Baron
210335d7afbSGerald Schaeferconfig HAVE_ARCH_MUTEX_CPU_RELAX
211335d7afbSGerald Schaefer	bool
212335d7afbSGerald Schaefer
21326723911SPeter Zijlstraconfig HAVE_RCU_TABLE_FREE
21426723911SPeter Zijlstra	bool
21526723911SPeter Zijlstra
216df013ffbSHuang Yingconfig ARCH_HAVE_NMI_SAFE_CMPXCHG
217df013ffbSHuang Ying	bool
218df013ffbSHuang Ying
21943570fd2SHeiko Carstensconfig HAVE_ALIGNED_STRUCT_PAGE
22043570fd2SHeiko Carstens	bool
22143570fd2SHeiko Carstens	help
22243570fd2SHeiko Carstens	  This makes sure that struct pages are double word aligned and that
22343570fd2SHeiko Carstens	  e.g. the SLUB allocator can perform double word atomic operations
22443570fd2SHeiko Carstens	  on a struct page for better performance. However selecting this
22543570fd2SHeiko Carstens	  might increase the size of a struct page by a word.
22643570fd2SHeiko Carstens
2274156153cSHeiko Carstensconfig HAVE_CMPXCHG_LOCAL
2284156153cSHeiko Carstens	bool
2294156153cSHeiko Carstens
2302565409fSHeiko Carstensconfig HAVE_CMPXCHG_DOUBLE
2312565409fSHeiko Carstens	bool
2322565409fSHeiko Carstens
23348b25c43SChris Metcalfconfig ARCH_WANT_OLD_COMPAT_IPC
23448b25c43SChris Metcalf	bool
23548b25c43SChris Metcalf
2362521f2c2SPeter Oberparleitersource "kernel/gcov/Kconfig"
237