1 # SPDX-License-Identifier: GPL-2.0
2 comment "Processor Type"
4 # Select CPU types depending on the architecture selected. This selects
5 # which CPUs we support in the kernel image, and the compiler instruction
15 select CPU_PABRT_LEGACY
17 A 32-bit RISC microprocessor based on the ARM7 processor core
18 which has no memory control unit and cache.
20 Say Y if you want support for the ARM7TDMI processor.
30 select CPU_COPY_V4WT if MMU
32 select CPU_PABRT_LEGACY
33 select CPU_THUMB_CAPABLE
34 select CPU_TLB_V4WT if MMU
36 A 32-bit RISC processor with 8kByte Cache, Write Buffer and
37 MMU built around an ARM7TDMI core.
39 Say Y if you want support for the ARM720T processor.
50 select CPU_PABRT_LEGACY
51 select CPU_THUMB_CAPABLE
53 A 32-bit RISC processor with 8KB cache or 4KB variants,
54 write buffer and MPU(Protection Unit) built around
57 Say Y if you want support for the ARM740T processor.
67 select CPU_PABRT_LEGACY
69 A 32-bit RISC microprocessor based on the ARM9 processor core
70 which has no memory control unit and cache.
72 Say Y if you want support for the ARM9TDMI processor.
82 select CPU_COPY_V4WB if MMU
84 select CPU_PABRT_LEGACY
85 select CPU_THUMB_CAPABLE
86 select CPU_TLB_V4WBI if MMU
88 The ARM920T is licensed to be produced by numerous vendors,
89 and is used in the Cirrus EP93xx and the Samsung S3C2410.
91 Say Y if you want support for the ARM920T processor.
100 select CPU_CACHE_VIVT
101 select CPU_COPY_V4WB if MMU
103 select CPU_PABRT_LEGACY
104 select CPU_THUMB_CAPABLE
105 select CPU_TLB_V4WBI if MMU
107 The ARM922T is a version of the ARM920T, but with smaller
108 instruction and data caches. It is used in Altera's
109 Excalibur XA device family and the ARM Integrator.
111 Say Y if you want support for the ARM922T processor.
119 select CPU_CACHE_V4WT
120 select CPU_CACHE_VIVT
121 select CPU_COPY_V4WB if MMU
123 select CPU_PABRT_LEGACY
124 select CPU_THUMB_CAPABLE
125 select CPU_TLB_V4WBI if MMU
127 The ARM925T is a mix between the ARM920T and ARM926T, but with
128 different instruction and data caches. It is used in TI's OMAP
131 Say Y if you want support for the ARM925T processor.
138 select CPU_ABRT_EV5TJ
139 select CPU_CACHE_VIVT
140 select CPU_COPY_V4WB if MMU
142 select CPU_PABRT_LEGACY
143 select CPU_THUMB_CAPABLE
144 select CPU_TLB_V4WBI if MMU
146 This is a variant of the ARM920. It has slightly different
147 instruction sequences for cache and TLB operations. Curiously,
148 there is no documentation on it at the ARM corporate website.
150 Say Y if you want support for the ARM926T processor.
159 select CPU_CACHE_VIVT
160 select CPU_COPY_FA if MMU
162 select CPU_PABRT_LEGACY
163 select CPU_TLB_FA if MMU
165 The FA526 is a version of the ARMv4 compatible processor with
166 Branch Target Buffer, Unified TLB and cache line size 16.
168 Say Y if you want support for the FA526 processor.
176 select CPU_ABRT_NOMMU
177 select CPU_CACHE_VIVT
179 select CPU_PABRT_LEGACY
180 select CPU_THUMB_CAPABLE
182 ARM940T is a member of the ARM9TDMI family of general-
183 purpose microprocessors with MPU and separate 4KB
184 instruction and 4KB data cases, each with a 4-word line
187 Say Y if you want support for the ARM940T processor.
195 select CPU_ABRT_NOMMU
196 select CPU_CACHE_VIVT
198 select CPU_PABRT_LEGACY
199 select CPU_THUMB_CAPABLE
201 ARM946E-S is a member of the ARM9E-S family of high-
202 performance, 32-bit system-on-chip processor solutions.
203 The TCM and ARMv5TE 32-bit instruction set is supported.
205 Say Y if you want support for the ARM946E-S processor.
208 # ARM1020 - needs validating
213 select CPU_CACHE_V4WT
214 select CPU_CACHE_VIVT
215 select CPU_COPY_V4WB if MMU
217 select CPU_PABRT_LEGACY
218 select CPU_THUMB_CAPABLE
219 select CPU_TLB_V4WBI if MMU
221 The ARM1020 is the 32K cached version of the ARM10 processor,
222 with an addition of a floating-point unit.
224 Say Y if you want support for the ARM1020 processor.
227 # ARM1020E - needs validating
233 select CPU_CACHE_V4WT
234 select CPU_CACHE_VIVT
235 select CPU_COPY_V4WB if MMU
237 select CPU_PABRT_LEGACY
238 select CPU_THUMB_CAPABLE
239 select CPU_TLB_V4WBI if MMU
246 select CPU_CACHE_VIVT
247 select CPU_COPY_V4WB if MMU # can probably do better
249 select CPU_PABRT_LEGACY
250 select CPU_THUMB_CAPABLE
251 select CPU_TLB_V4WBI if MMU
253 The ARM1022E is an implementation of the ARMv5TE architecture
254 based upon the ARM10 integer core with a 16KiB L1 Harvard cache,
255 embedded trace macrocell, and a floating-point unit.
257 Say Y if you want support for the ARM1022E processor.
264 select CPU_ABRT_EV5T # But need Jazelle, but EV5TJ ignores bit 10
265 select CPU_CACHE_VIVT
266 select CPU_COPY_V4WB if MMU # can probably do better
268 select CPU_PABRT_LEGACY
269 select CPU_THUMB_CAPABLE
270 select CPU_TLB_V4WBI if MMU
272 The ARM1026EJ-S is an implementation of the ARMv5TEJ architecture
273 based upon the ARM10 integer core.
275 Say Y if you want support for the ARM1026EJ-S processor.
281 select CPU_32v3 if ARCH_RPC
282 select CPU_32v4 if !ARCH_RPC
284 select CPU_CACHE_V4WB
285 select CPU_CACHE_VIVT
286 select CPU_COPY_V4WB if MMU
288 select CPU_PABRT_LEGACY
289 select CPU_TLB_V4WB if MMU
291 The Intel StrongARM(R) SA-110 is a 32-bit microprocessor and
292 is available at five speeds ranging from 100 MHz to 233 MHz.
293 More information is available at
294 <http://developer.intel.com/design/strong/sa110.htm>.
296 Say Y if you want support for the SA-110 processor.
304 select CPU_CACHE_V4WB
305 select CPU_CACHE_VIVT
307 select CPU_PABRT_LEGACY
308 select CPU_TLB_V4WB if MMU
315 select CPU_CACHE_VIVT
317 select CPU_PABRT_LEGACY
318 select CPU_THUMB_CAPABLE
319 select CPU_TLB_V4WBI if MMU
321 # XScale Core Version 3
326 select CPU_CACHE_VIVT
328 select CPU_PABRT_LEGACY
329 select CPU_THUMB_CAPABLE
330 select CPU_TLB_V4WBI if MMU
333 # Marvell PJ1 (Mohawk)
338 select CPU_CACHE_VIVT
339 select CPU_COPY_V4WB if MMU
341 select CPU_PABRT_LEGACY
342 select CPU_THUMB_CAPABLE
343 select CPU_TLB_V4WBI if MMU
350 select CPU_CACHE_VIVT
351 select CPU_COPY_FEROCEON if MMU
353 select CPU_PABRT_LEGACY
354 select CPU_THUMB_CAPABLE
355 select CPU_TLB_FEROCEON if MMU
357 config CPU_FEROCEON_OLD_ID
358 bool "Accept early Feroceon cores with an ARM926 ID"
359 depends on CPU_FEROCEON && !CPU_ARM926T
362 This enables the usage of some old Feroceon cores
363 for which the CPU ID is equal to the ARM926 ID.
364 Relevant for Feroceon-1850 and early Feroceon-2850.
382 select CPU_CACHE_VIPT
383 select CPU_COPY_V6 if MMU
385 select CPU_HAS_ASID if MMU
387 select CPU_THUMB_CAPABLE
388 select CPU_TLB_V6 if MMU
389 select SMP_ON_UP if SMP
398 select CPU_CACHE_VIPT
399 select CPU_COPY_V6 if MMU
401 select CPU_HAS_ASID if MMU
403 select CPU_THUMB_CAPABLE
404 select CPU_TLB_V6 if MMU
406 # ARMv7 and ARMv8 architectures
413 select CPU_CACHE_VIPT
414 select CPU_COPY_V6 if MMU
415 select CPU_CP15_MMU if MMU
416 select CPU_CP15_MPU if !MMU
417 select CPU_HAS_ASID if MMU
419 select CPU_SPECTRE if MMU
420 select CPU_THUMB_CAPABLE
421 select CPU_TLB_V7 if MMU
427 select CPU_ABRT_NOMMU
430 select CPU_PABRT_LEGACY
435 select CPU_THUMB_CAPABLE
436 # There are no CPUs available with MMU that don't implement an ARM ISA:
439 Select this if your CPU doesn't support the 32 bit ARM instructions.
441 config CPU_THUMB_CAPABLE
444 Select this if your CPU can support Thumb mode.
446 # Figure out what processor architecture version we should be using.
447 # This defines the compiler instruction set which depends on the machine type.
450 select CPU_USE_DOMAINS if MMU
451 select NEED_KUSER_HELPERS
452 select TLS_REG_EMUL if SMP || !MMU
453 select CPU_NO_EFFICIENT_FFS
457 select CPU_USE_DOMAINS if MMU
458 select NEED_KUSER_HELPERS
459 select TLS_REG_EMUL if SMP || !MMU
460 select CPU_NO_EFFICIENT_FFS
464 select CPU_USE_DOMAINS if MMU
465 select NEED_KUSER_HELPERS
466 select TLS_REG_EMUL if SMP || !MMU
467 select CPU_NO_EFFICIENT_FFS
471 select CPU_USE_DOMAINS if MMU
472 select NEED_KUSER_HELPERS
473 select TLS_REG_EMUL if SMP || !MMU
477 select TLS_REG_EMUL if !CPU_32v6K && !MMU
489 config CPU_ABRT_NOMMU
504 config CPU_ABRT_EV5TJ
513 config CPU_PABRT_LEGACY
526 config CPU_CACHE_V4WT
529 config CPU_CACHE_V4WB
541 config CPU_CACHE_VIVT
544 config CPU_CACHE_VIPT
554 # The copy-page model
561 config CPU_COPY_FEROCEON
570 # This selects the TLB model
574 ARM Architecture Version 4 TLB with writethrough cache.
579 ARM Architecture Version 4 TLB with writeback cache.
584 ARM Architecture Version 4 TLB with writeback cache and invalidate
585 instruction cache entry.
587 config CPU_TLB_FEROCEON
590 Feroceon TLB (v4wbi with non-outer-cachable page table walks).
595 Faraday ARM FA526 architecture, unified TLB with writeback cache
596 and invalidate instruction cache entry. Branch target buffer is
610 This indicates whether the CPU has the ASID register; used to
611 tag TLB and possibly cache entries.
616 Processor has the CP15 register.
622 Processor has the CP15 register, which has MMU related registers.
628 Processor has the CP15 register, which has MPU related registers.
630 config CPU_USE_DOMAINS
633 This option enables or disables the use of domain switching
634 using the DACR (domain access control register) to protect memory
635 domains from each other. In Linux we use three domains: kernel, user
636 and IO. The domains are used to protect userspace from kernelspace
637 and to handle IO-space as a special type of memory by assigning
638 manager or client roles to running code (such as a process).
640 config CPU_V7M_NUM_IRQ
641 int "Number of external interrupts connected to the NVIC"
643 default 90 if ARCH_STM32
644 default 112 if SOC_VF610
647 This option indicates the number of interrupts connected to the NVIC.
648 The value can be larger than the real number of interrupts supported
649 by the system, but must not be lower.
650 The default value is 240, corresponding to the maximum number of
651 interrupts supported by the NVIC on Cortex-M family.
653 If unsure, keep default value.
656 # CPU supports 36-bit I/O
661 comment "Processor Features"
664 bool "Support for the Large Physical Address Extension"
665 depends on MMU && CPU_32v7 && !CPU_32v6 && !CPU_32v5 && \
666 !CPU_32v4 && !CPU_32v3
667 select PHYS_ADDR_T_64BIT
670 Say Y if you have an ARMv7 processor supporting the LPAE page
671 table format and you would like to access memory beyond the
672 4GB limit. The resulting kernel image will not run on
673 processors without the LPA extension.
679 depends on ARM_LPAE && ARM_PATCH_PHYS_VIRT && ARCH_KEYSTONE
682 bool "Support Thumb user binaries" if !CPU_THUMBONLY && EXPERT
683 depends on CPU_THUMB_CAPABLE && !CPU_32v4
686 Say Y if you want to include kernel support for running user space
689 The Thumb instruction set is a compressed form of the standard ARM
690 instruction set resulting in smaller binaries at the expense of
691 slightly less efficient code.
693 If this option is disabled, and you run userspace that switches to
694 Thumb mode, signal handling will not work correctly, resulting in
695 segmentation faults or illegal instruction aborts.
697 If you don't know what this all is, saying Y is a safe choice.
700 bool "Enable ThumbEE CPU extension"
703 Say Y here if you have a CPU with the ThumbEE extension and code to
704 make use of it. Say N for code that can run on CPUs without ThumbEE.
710 Enable the kernel to make use of the ARM Virtualization
711 Extensions to install hypervisors without run-time firmware
714 A compliant bootloader is required in order to make maximum
715 use of this feature. Refer to Documentation/arch/arm/booting.rst for
719 bool "Emulate SWP/SWPB instructions" if !SMP
722 select HAVE_PROC_CPU if PROC_FS
724 ARMv6 architecture deprecates use of the SWP/SWPB instructions.
725 ARMv7 multiprocessing extensions introduce the ability to disable
726 these instructions, triggering an undefined instruction exception
727 when executed. Say Y here to enable software emulation of these
728 instructions for userspace (not kernel) using LDREX/STREX.
729 Also creates /proc/cpu/swp_emulation for statistics.
731 In some older versions of glibc [<=2.8] SWP is used during futex
732 trylock() operations with the assumption that the code will not
733 be preempted. This invalid assumption may be more likely to fail
734 with SWP emulation enabled, leading to deadlock of the user
737 NOTE: when accessing uncached shared regions, LDREX/STREX rely
738 on an external transaction monitoring block called a global
739 monitor to maintain update atomicity. If your system does not
740 implement a global monitor, this option can cause programs that
741 perform SWP operations to uncached memory to deadlock.
746 prompt "CPU Endianness"
747 default CPU_LITTLE_ENDIAN
749 config CPU_LITTLE_ENDIAN
750 bool "Built little-endian kernel"
752 Say Y if you plan on running a kernel in little-endian mode.
753 This is the default and is used in practically all modern user
756 config CPU_BIG_ENDIAN
757 bool "Build big-endian kernel"
758 depends on !LD_IS_LLD
760 Say Y if you plan on running a kernel in big-endian mode.
761 This works on many machines using ARMv6 or newer processors
762 but requires big-endian user space.
764 The only ARMv5 platform with big-endian support is
769 config CPU_ENDIAN_BE8
771 depends on CPU_BIG_ENDIAN
772 default CPU_V6 || CPU_V6K || CPU_V7 || CPU_V7M
774 Support for the BE-8 (big-endian) mode on ARMv6 and ARMv7 processors.
776 config CPU_ENDIAN_BE32
778 depends on CPU_BIG_ENDIAN
779 default !CPU_ENDIAN_BE8
781 Support for the BE-32 (big-endian) mode on pre-ARMv6 processors.
783 config CPU_HIGH_VECTOR
784 depends on !MMU && CPU_CP15 && !CPU_ARM740T
785 bool "Select the High exception vector"
787 Say Y here to select high exception vector(0xFFFF0000~).
788 The exception vector can vary depending on the platform
789 design in nommu mode. If your platform needs to select
790 high exception vector, say Y.
791 Otherwise or if you are unsure, say N, and the low exception
792 vector (0x00000000~) will be used.
794 config CPU_ICACHE_DISABLE
795 bool "Disable I-Cache (I-bit)"
796 depends on (CPU_CP15 && !(CPU_ARM720T || CPU_ARM740T || CPU_XSCALE || CPU_XSC3)) || CPU_V7M
798 Say Y here to disable the processor instruction cache. Unless
799 you have a reason not to or are unsure, say N.
801 config CPU_ICACHE_MISMATCH_WORKAROUND
802 bool "Workaround for I-Cache line size mismatch between CPU cores"
803 depends on SMP && CPU_V7
805 Some big.LITTLE systems have I-Cache line size mismatch between
806 LITTLE and big cores. Say Y here to enable a workaround for
807 proper I-Cache support on such systems. If unsure, say N.
809 config CPU_DCACHE_DISABLE
810 bool "Disable D-Cache (C-bit)"
811 depends on (CPU_CP15 && !SMP) || CPU_V7M
813 Say Y here to disable the processor data cache. Unless
814 you have a reason not to or are unsure, say N.
816 config CPU_DCACHE_SIZE
818 depends on CPU_ARM740T || CPU_ARM946E
819 default 0x00001000 if CPU_ARM740T
820 default 0x00002000 # default size for ARM946E-S
822 Some cores are synthesizable to have various sized cache. For
823 ARM946E-S case, it can vary from 0KB to 1MB.
824 To support such cache operations, it is efficient to know the size
826 If your SoC is configured to have a different size, define the value
827 here with proper conditions.
829 config CPU_DCACHE_WRITETHROUGH
830 bool "Force write through D-cache"
831 depends on (CPU_ARM740T || CPU_ARM920T || CPU_ARM922T || CPU_ARM925T || CPU_ARM926T || CPU_ARM940T || CPU_ARM946E || CPU_ARM1020 || CPU_FA526) && !CPU_DCACHE_DISABLE
832 default y if CPU_ARM925T
834 Say Y here to use the data cache in writethrough mode. Unless you
835 specifically require this or are unsure, say N.
837 config CPU_CACHE_ROUND_ROBIN
838 bool "Round robin I and D cache replacement algorithm"
839 depends on (CPU_ARM926T || CPU_ARM946E || CPU_ARM1020) && (!CPU_ICACHE_DISABLE || !CPU_DCACHE_DISABLE)
841 Say Y here to use the predictable round-robin cache replacement
842 policy. Unless you specifically require this or are unsure, say N.
844 config CPU_BPREDICT_DISABLE
845 bool "Disable branch prediction"
846 depends on CPU_ARM1020 || CPU_V6 || CPU_V6K || CPU_MOHAWK || CPU_XSC3 || CPU_V7 || CPU_FA526 || CPU_V7M
848 Say Y here to disable branch prediction. If unsure, say N.
852 select GENERIC_CPU_VULNERABILITIES
854 config HARDEN_BRANCH_PREDICTOR
855 bool "Harden the branch predictor against aliasing attacks" if EXPERT
856 depends on CPU_SPECTRE
859 Speculation attacks against some high-performance processors rely
860 on being able to manipulate the branch predictor for a victim
861 context by executing aliasing branches in the attacker context.
862 Such attacks can be partially mitigated against by clearing
863 internal branch predictor state and limiting the prediction
864 logic in some situations.
866 This config option will take CPU-specific actions to harden
867 the branch predictor against aliasing attacks and may rely on
868 specific instruction sequences or control bits being set by
873 config HARDEN_BRANCH_HISTORY
874 bool "Harden Spectre style attacks against branch history" if EXPERT
875 depends on CPU_SPECTRE
878 Speculation attacks against some high-performance processors can
879 make use of branch history to influence future speculation. When
880 taking an exception, a sequence of branches overwrites the branch
881 history, or branch history is invalidated.
885 select NEED_KUSER_HELPERS
887 An SMP system using a pre-ARMv6 processor (there are apparently
888 a few prototypes like that in existence) and therefore access to
889 that required register must be emulated.
891 config NEED_KUSER_HELPERS
895 bool "Enable kuser helpers in vector page" if !NEED_KUSER_HELPERS
899 Warning: disabling this option may break user programs.
901 Provide kuser helpers in the vector page. The kernel provides
902 helper code to userspace in read only form at a fixed location
903 in the high vector page to allow userspace to be independent of
904 the CPU type fitted to the system. This permits binaries to be
905 run on ARMv4 through to ARMv7 without modification.
907 See Documentation/arch/arm/kernel_user_helpers.rst for details.
909 However, the fixed address nature of these helpers can be used
910 by ROP (return orientated programming) authors when creating
913 If all of the binaries and libraries which run on your platform
914 are built specifically for your platform, and make no use of
915 these helpers, then you can turn this option off to hinder
916 such exploits. However, in that case, if a binary or library
917 relying on those helpers is run, it will receive a SIGILL signal,
918 which will terminate the program.
920 Say N here only if you are absolutely certain that you do not
921 need these helpers; otherwise, the safe option is to say Y.
924 bool "Enable VDSO for acceleration of some system calls"
925 depends on AEABI && MMU && CPU_V7
926 default y if ARM_ARCH_TIMER
927 select HAVE_GENERIC_VDSO
928 select GENERIC_TIME_VSYSCALL
929 select GENERIC_VDSO_32
930 select GENERIC_GETTIMEOFDAY
932 Place in the process address space an ELF shared object
933 providing fast implementations of gettimeofday and
934 clock_gettime. Systems that implement the ARM architected
935 timer will receive maximum benefit.
937 You must have glibc 2.22 or later for programs to seamlessly
938 take advantage of this.
940 config DMA_CACHE_RWFO
941 bool "Enable read/write for ownership DMA cache maintenance"
942 depends on CPU_V6K && SMP
945 The Snoop Control Unit on ARM11MPCore does not detect the
946 cache maintenance operations and the dma_{map,unmap}_area()
947 functions may leave stale cache entries on other CPUs. By
948 enabling this option, Read or Write For Ownership in the ARMv6
949 DMA cache maintenance functions is performed. These LDR/STR
950 instructions change the cache line state to shared or modified
951 so that the cache operation has the desired effect.
953 Note that the workaround is only valid on processors that do
954 not perform speculative loads into the D-cache. For such
955 processors, if cache maintenance operations are not broadcast
956 in hardware, other workarounds are needed (e.g. cache
957 maintenance broadcasting in software via FIQ).
962 config OUTER_CACHE_SYNC
966 The outer cache has a outer_cache_fns.sync function pointer
967 that can be used to drain the write buffer of the outer cache.
970 bool "Enable the Broadcom Brahma-B15 read-ahead cache controller"
971 depends on ARCH_BRCMSTB
974 This option enables the Broadcom Brahma-B15 read-ahead cache
975 controller. If disabled, the read-ahead cache remains off.
977 config CACHE_FEROCEON_L2
978 bool "Enable the Feroceon L2 cache controller"
979 depends on ARCH_MV78XX0 || ARCH_MVEBU
983 This option enables the Feroceon L2 cache controller.
985 config CACHE_FEROCEON_L2_WRITETHROUGH
986 bool "Force Feroceon L2 cache write through"
987 depends on CACHE_FEROCEON_L2
989 Say Y here to use the Feroceon L2 cache in writethrough mode.
990 Unless you specifically require this, say N for writeback mode.
992 config MIGHT_HAVE_CACHE_L2X0
995 This option should be selected by machines which have a L2x0
996 or PL310 cache controller, but where its use is optional.
998 The only effect of this option is to make CACHE_L2X0 and
999 related options available to the user for configuration.
1001 Boards or SoCs which always require the cache controller
1002 support to be present should select CACHE_L2X0 directly
1003 instead of this option, thus preventing the user from
1004 inadvertently configuring a broken kernel.
1007 bool "Enable the L2x0 outer cache controller" if MIGHT_HAVE_CACHE_L2X0
1008 default MIGHT_HAVE_CACHE_L2X0
1010 select OUTER_CACHE_SYNC
1012 This option enables the L2x0 PrimeCell.
1014 config CACHE_L2X0_PMU
1015 bool "L2x0 performance monitor support" if CACHE_L2X0
1016 depends on PERF_EVENTS
1018 This option enables support for the performance monitoring features
1019 of the L220 and PL310 outer cache controllers.
1023 config PL310_ERRATA_588369
1024 bool "PL310 errata: Clean & Invalidate maintenance operations do not invalidate clean lines"
1026 The PL310 L2 cache controller implements three types of Clean &
1027 Invalidate maintenance operations: by Physical Address
1028 (offset 0x7F0), by Index/Way (0x7F8) and by Way (0x7FC).
1029 They are architecturally defined to behave as the execution of a
1030 clean operation followed immediately by an invalidate operation,
1031 both performing to the same memory location. This functionality
1032 is not correctly implemented in PL310 prior to r2p0 (fixed in r2p0)
1033 as clean lines are not invalidated as a result of these operations.
1035 config PL310_ERRATA_727915
1036 bool "PL310 errata: Background Clean & Invalidate by Way operation can cause data corruption"
1038 PL310 implements the Clean & Invalidate by Way L2 cache maintenance
1039 operation (offset 0x7FC). This operation runs in background so that
1040 PL310 can handle normal accesses while it is in progress. Under very
1041 rare circumstances, due to this erratum, write data can be lost when
1042 PL310 treats a cacheable write transaction during a Clean &
1043 Invalidate by Way operation. Revisions prior to r3p1 are affected by
1044 this errata (fixed in r3p1).
1046 config PL310_ERRATA_753970
1047 bool "PL310 errata: cache sync operation may be faulty"
1049 This option enables the workaround for the 753970 PL310 (r3p0) erratum.
1051 Under some condition the effect of cache sync operation on
1052 the store buffer still remains when the operation completes.
1053 This means that the store buffer is always asked to drain and
1054 this prevents it from merging any further writes. The workaround
1055 is to replace the normal offset of cache sync operation (0x730)
1056 by another offset targeting an unmapped PL310 register 0x740.
1057 This has the same effect as the cache sync operation: store buffer
1058 drain and waiting for all buffers empty.
1060 config PL310_ERRATA_769419
1061 bool "PL310 errata: no automatic Store Buffer drain"
1063 On revisions of the PL310 prior to r3p2, the Store Buffer does
1064 not automatically drain. This can cause normal, non-cacheable
1065 writes to be retained when the memory system is idle, leading
1066 to suboptimal I/O performance for drivers using coherent DMA.
1067 This option adds a write barrier to the cpu_idle loop so that,
1068 on systems with an outer cache, the store buffer is drained
1073 config CACHE_TAUROS2
1074 bool "Enable the Tauros2 L2 cache controller"
1075 depends on (CPU_MOHAWK || CPU_PJ4)
1079 This option enables the Tauros2 L2 cache controller (as
1082 config CACHE_UNIPHIER
1083 bool "Enable the UniPhier outer cache controller"
1084 depends on ARCH_UNIPHIER
1085 select ARM_L1_CACHE_SHIFT_7
1087 select OUTER_CACHE_SYNC
1089 This option enables the UniPhier outer cache (system cache)
1093 bool "Enable the L2 cache on XScale3"
1098 This option enables the L2 cache on XScale3.
1100 config ARM_L1_CACHE_SHIFT_6
1104 Setting ARM L1 cache line size to 64 Bytes.
1106 config ARM_L1_CACHE_SHIFT_7
1109 Setting ARM L1 cache line size to 128 Bytes.
1111 config ARM_L1_CACHE_SHIFT
1113 default 7 if ARM_L1_CACHE_SHIFT_7
1114 default 6 if ARM_L1_CACHE_SHIFT_6
1117 config ARM_DMA_MEM_BUFFERABLE
1118 bool "Use non-cacheable memory for DMA" if (CPU_V6 || CPU_V6K || CPU_V7M) && !CPU_V7
1119 default y if CPU_V6 || CPU_V6K || CPU_V7 || CPU_V7M
1121 Historically, the kernel has used strongly ordered mappings to
1122 provide DMA coherent memory. With the advent of ARMv7, mapping
1123 memory with differing types results in unpredictable behaviour,
1124 so on these CPUs, this option is forced on.
1126 Multiple mappings with differing attributes is also unpredictable
1127 on ARMv6 CPUs, but since they do not have aggressive speculative
1128 prefetch, no harm appears to occur.
1130 However, drivers may be missing the necessary barriers for ARMv6,
1131 and therefore turning this on may result in unpredictable driver
1132 behaviour. Therefore, we offer this as an option.
1134 On some of the beefier ARMv7-M machines (with DMA and write
1135 buffers) you likely want this enabled, while those that
1136 didn't need it until now also won't need it in the future.
1138 You are recommended say 'Y' here and debug any affected drivers.
1143 config DEBUG_ALIGN_RODATA
1144 bool "Make rodata strictly non-executable"
1145 depends on STRICT_KERNEL_RWX
1148 If this is set, rodata will be made explicitly non-executable. This
1149 provides protection on the rare chance that attackers might find and
1150 use ROP gadgets that exist in the rodata section. This adds an
1151 additional section-aligned split of rodata from kernel text so it
1152 can be made explicitly non-executable. This padding may waste memory
1153 space to gain the additional protection.