linux-brain/arch/arm64/kernel/cpufeature.c

1247 lines
40 KiB
C
Raw Normal View History

/*
* Contains CPU feature definitions
*
* Copyright (C) 2015 ARM Ltd.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#define pr_fmt(fmt) "CPU features: " fmt
#include <linux/bsearch.h>
#include <linux/cpumask.h>
#include <linux/sort.h>
#include <linux/stop_machine.h>
#include <linux/types.h>
#include <asm/cpu.h>
#include <asm/cpufeature.h>
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
#include <asm/cpu_ops.h>
#include <asm/mmu_context.h>
#include <asm/processor.h>
#include <asm/sysreg.h>
#include <asm/virt.h>
unsigned long elf_hwcap __read_mostly;
EXPORT_SYMBOL_GPL(elf_hwcap);
#ifdef CONFIG_COMPAT
#define COMPAT_ELF_HWCAP_DEFAULT \
(COMPAT_HWCAP_HALF|COMPAT_HWCAP_THUMB|\
COMPAT_HWCAP_FAST_MULT|COMPAT_HWCAP_EDSP|\
COMPAT_HWCAP_TLS|COMPAT_HWCAP_VFP|\
COMPAT_HWCAP_VFPv3|COMPAT_HWCAP_VFPv4|\
COMPAT_HWCAP_NEON|COMPAT_HWCAP_IDIV|\
COMPAT_HWCAP_LPAE)
unsigned int compat_elf_hwcap __read_mostly = COMPAT_ELF_HWCAP_DEFAULT;
unsigned int compat_elf_hwcap2 __read_mostly;
#endif
DECLARE_BITMAP(cpu_hwcaps, ARM64_NCAPS);
EXPORT_SYMBOL(cpu_hwcaps);
DEFINE_STATIC_KEY_ARRAY_FALSE(cpu_hwcap_keys, ARM64_NCAPS);
EXPORT_SYMBOL(cpu_hwcap_keys);
#define __ARM64_FTR_BITS(SIGNED, STRICT, TYPE, SHIFT, WIDTH, SAFE_VAL) \
{ \
.sign = SIGNED, \
.strict = STRICT, \
.type = TYPE, \
.shift = SHIFT, \
.width = WIDTH, \
.safe_val = SAFE_VAL, \
}
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
/* Define a feature with unsigned values */
#define ARM64_FTR_BITS(STRICT, TYPE, SHIFT, WIDTH, SAFE_VAL) \
__ARM64_FTR_BITS(FTR_UNSIGNED, STRICT, TYPE, SHIFT, WIDTH, SAFE_VAL)
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
/* Define a feature with a signed value */
#define S_ARM64_FTR_BITS(STRICT, TYPE, SHIFT, WIDTH, SAFE_VAL) \
__ARM64_FTR_BITS(FTR_SIGNED, STRICT, TYPE, SHIFT, WIDTH, SAFE_VAL)
#define ARM64_FTR_END \
{ \
.width = 0, \
}
/* meta feature for alternatives */
static bool __maybe_unused
cpufeature_pan_not_uao(const struct arm64_cpu_capabilities *entry, int __unused);
static const struct arm64_ftr_bits ftr_id_aa64isar0[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 32, 32, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64ISAR0_RDM_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 24, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64ISAR0_ATOMICS_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64ISAR0_CRC32_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64ISAR0_SHA2_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64ISAR0_SHA1_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64ISAR0_AES_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 0, 4, 0), /* RAZ */
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_aa64pfr0[] = {
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64PFR0_CSV3_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64PFR0_CSV2_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 32, 24, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 28, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64PFR0_GIC_SHIFT, 4, 0),
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
S_ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR0_ASIMD_SHIFT, 4, ID_AA64PFR0_ASIMD_NI),
S_ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR0_FP_SHIFT, 4, ID_AA64PFR0_FP_NI),
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64PFR0_CSV3_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64PFR0_CSV2_SHIFT, 4, 0),
/* Linux doesn't care about the EL3 */
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_EXACT, ID_AA64PFR0_EL3_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64PFR0_EL2_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64PFR0_EL1_SHIFT, 4, ID_AA64PFR0_EL1_64BIT_ONLY),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64PFR0_EL0_SHIFT, 4, ID_AA64PFR0_EL0_64BIT_ONLY),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_aa64mmfr0[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 32, 32, 0),
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
S_ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR0_TGRAN4_SHIFT, 4, ID_AA64MMFR0_TGRAN4_NI),
S_ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR0_TGRAN64_SHIFT, 4, ID_AA64MMFR0_TGRAN64_NI),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR0_TGRAN16_SHIFT, 4, ID_AA64MMFR0_TGRAN16_NI),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR0_BIGENDEL0_SHIFT, 4, 0),
/* Linux shouldn't care about secure memory */
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_EXACT, ID_AA64MMFR0_SNSMEM_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR0_BIGENDEL_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR0_ASID_SHIFT, 4, 0),
/*
* Differing PARange is fine as long as all peripherals and memory are mapped
* within the minimum PARange of all CPUs
*/
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_PARANGE_SHIFT, 4, 0),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_aa64mmfr1[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 32, 32, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64MMFR1_PAN_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR1_LOR_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR1_HPD_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR1_VHE_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR1_VMIDBITS_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR1_HADBS_SHIFT, 4, 0),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_aa64mmfr2[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR2_LVA_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR2_IESB_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR2_LSM_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR2_UAO_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64MMFR2_CNP_SHIFT, 4, 0),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_ctr[] = {
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 31, 1, 1), /* RAO */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 28, 3, 0),
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
ARM64_FTR_BITS(FTR_STRICT, FTR_HIGHER_SAFE, 24, 4, 0), /* CWG */
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 20, 4, 0), /* ERG */
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, CTR_DMINLINE_SHIFT, 4, 1),
/*
* Linux can handle differing I-cache policies. Userspace JITs will
* make use of *minLine.
* If we have differing I-cache policies, report it as the weakest - AIVIVT.
*/
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_EXACT, 14, 2, ICACHE_POLICY_AIVIVT), /* L1Ip */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 4, 10, 0), /* RAZ */
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, CTR_IMINLINE_SHIFT, 4, 0),
ARM64_FTR_END,
};
struct arm64_ftr_reg arm64_ftr_reg_ctrel0 = {
.name = "SYS_CTR_EL0",
.ftr_bits = ftr_ctr
};
static const struct arm64_ftr_bits ftr_id_mmfr0[] = {
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
S_ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 28, 4, 0xf), /* InnerShr */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 24, 4, 0), /* FCSE */
ARM64_FTR_BITS(FTR_NONSTRICT, FTR_LOWER_SAFE, 20, 4, 0), /* AuxReg */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 16, 4, 0), /* TCM */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 12, 4, 0), /* ShareLvl */
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
S_ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 8, 4, 0xf), /* OuterShr */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 4, 4, 0), /* PMSA */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 0, 4, 0), /* VMSA */
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_aa64dfr0[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 32, 32, 0),
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR0_CTX_CMPS_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR0_WRPS_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, ID_AA64DFR0_BRPS_SHIFT, 4, 0),
S_ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64DFR0_PMUVER_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64DFR0_TRACEVER_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_AA64DFR0_DEBUGVER_SHIFT, 4, 0x6),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_mvfr2[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 8, 24, 0), /* RAZ */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 4, 4, 0), /* FPMisc */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 0, 4, 0), /* SIMDMisc */
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_dczid[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 5, 27, 0), /* RAZ */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 4, 1, 1), /* DZP */
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 0, 4, 0), /* BS */
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_isar5[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_ISAR5_RDM_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 20, 4, 0), /* RAZ */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_ISAR5_CRC32_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_ISAR5_SHA2_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_ISAR5_SHA1_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_ISAR5_AES_SHIFT, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, ID_ISAR5_SEVL_SHIFT, 4, 0),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_mmfr4[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 8, 24, 0), /* RAZ */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 4, 4, 0), /* ac2 */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 0, 4, 0), /* RAZ */
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_pfr0[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 16, 16, 0), /* RAZ */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 12, 4, 0), /* State3 */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 8, 4, 0), /* State2 */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 4, 4, 0), /* State1 */
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 0, 4, 0), /* State0 */
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_id_dfr0[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 28, 4, 0),
arm64: cpufeature: Fix the sign of feature bits There is a confusion on whether the values of a feature are signed or not in ARM. This is not clearly mentioned in the ARM ARM either. We have dealt most of the bits as signed so far, and marked the rest as unsigned explicitly. This fixed in ARM ARM and will be rolled out soon. Here is the criteria in a nutshell: 1) The fields, which are either signed or unsigned, use increasing numerical values to indicate an increase in functionality. Thus, if a value of 0x1 indicates the presence of some instructions, then the 0x2 value will indicate the presence of those instructions plus some additional instructions or functionality. 2) For ID field values where the value 0x0 defines that a feature is not present, the number is an unsigned value. 3) For some features where the feature was made optional or removed after the start of the definition of the architecture, the value 0x0 is used to indicate the presence of a feature, and 0xF indicates the absence of the feature. In these cases, the fields are, in effect, holding signed values. So with these rules applied, we have only the following fields which are signed and the rest are unsigned. a) ID_AA64PFR0_EL1: {FP, ASIMD} b) ID_AA64MMFR0_EL1: {TGran4K, TGran64K} c) ID_AA64DFR0_EL1: PMUVer (0xf - PMUv3 not implemented) d) ID_DFR0_EL1: PerfMon e) ID_MMFR0_EL1: {InnerShr, OuterShr} Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-01-26 19:58:14 +09:00
S_ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 24, 4, 0xf), /* PerfMon */
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 20, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 16, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 12, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 8, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 4, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 0, 4, 0),
ARM64_FTR_END,
};
/*
* Common ftr bits for a 32bit register with all hidden, strict
* attributes, with 4bit feature fields and a default safe value of
* 0. Covers the following 32bit registers:
* id_isar[0-4], id_mmfr[1-3], id_pfr1, mvfr[0-1]
*/
static const struct arm64_ftr_bits ftr_generic_32bits[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 28, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 24, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 20, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 16, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 12, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 8, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 4, 4, 0),
ARM64_FTR_BITS(FTR_STRICT, FTR_LOWER_SAFE, 0, 4, 0),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_generic[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 0, 64, 0),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_generic32[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 0, 32, 0),
ARM64_FTR_END,
};
static const struct arm64_ftr_bits ftr_aa64raz[] = {
ARM64_FTR_BITS(FTR_STRICT, FTR_EXACT, 0, 64, 0),
ARM64_FTR_END,
};
#define ARM64_FTR_REG(id, table) { \
.sys_id = id, \
.reg = &(struct arm64_ftr_reg){ \
.name = #id, \
.ftr_bits = &((table)[0]), \
}}
static const struct __ftr_reg_entry {
u32 sys_id;
struct arm64_ftr_reg *reg;
} arm64_ftr_regs[] = {
/* Op1 = 0, CRn = 0, CRm = 1 */
ARM64_FTR_REG(SYS_ID_PFR0_EL1, ftr_id_pfr0),
ARM64_FTR_REG(SYS_ID_PFR1_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_ID_DFR0_EL1, ftr_id_dfr0),
ARM64_FTR_REG(SYS_ID_MMFR0_EL1, ftr_id_mmfr0),
ARM64_FTR_REG(SYS_ID_MMFR1_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_ID_MMFR2_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_ID_MMFR3_EL1, ftr_generic_32bits),
/* Op1 = 0, CRn = 0, CRm = 2 */
ARM64_FTR_REG(SYS_ID_ISAR0_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_ID_ISAR1_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_ID_ISAR2_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_ID_ISAR3_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_ID_ISAR4_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_ID_ISAR5_EL1, ftr_id_isar5),
ARM64_FTR_REG(SYS_ID_MMFR4_EL1, ftr_id_mmfr4),
/* Op1 = 0, CRn = 0, CRm = 3 */
ARM64_FTR_REG(SYS_MVFR0_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_MVFR1_EL1, ftr_generic_32bits),
ARM64_FTR_REG(SYS_MVFR2_EL1, ftr_mvfr2),
/* Op1 = 0, CRn = 0, CRm = 4 */
ARM64_FTR_REG(SYS_ID_AA64PFR0_EL1, ftr_id_aa64pfr0),
ARM64_FTR_REG(SYS_ID_AA64PFR1_EL1, ftr_aa64raz),
/* Op1 = 0, CRn = 0, CRm = 5 */
ARM64_FTR_REG(SYS_ID_AA64DFR0_EL1, ftr_id_aa64dfr0),
ARM64_FTR_REG(SYS_ID_AA64DFR1_EL1, ftr_generic),
/* Op1 = 0, CRn = 0, CRm = 6 */
ARM64_FTR_REG(SYS_ID_AA64ISAR0_EL1, ftr_id_aa64isar0),
ARM64_FTR_REG(SYS_ID_AA64ISAR1_EL1, ftr_aa64raz),
/* Op1 = 0, CRn = 0, CRm = 7 */
ARM64_FTR_REG(SYS_ID_AA64MMFR0_EL1, ftr_id_aa64mmfr0),
ARM64_FTR_REG(SYS_ID_AA64MMFR1_EL1, ftr_id_aa64mmfr1),
ARM64_FTR_REG(SYS_ID_AA64MMFR2_EL1, ftr_id_aa64mmfr2),
/* Op1 = 3, CRn = 0, CRm = 0 */
{ SYS_CTR_EL0, &arm64_ftr_reg_ctrel0 },
ARM64_FTR_REG(SYS_DCZID_EL0, ftr_dczid),
/* Op1 = 3, CRn = 14, CRm = 0 */
ARM64_FTR_REG(SYS_CNTFRQ_EL0, ftr_generic32),
};
static int search_cmp_ftr_reg(const void *id, const void *regp)
{
return (int)(unsigned long)id - (int)((const struct __ftr_reg_entry *)regp)->sys_id;
}
/*
* get_arm64_ftr_reg - Lookup a feature register entry using its
* sys_reg() encoding. With the array arm64_ftr_regs sorted in the
* ascending order of sys_id , we use binary search to find a matching
* entry.
*
* returns - Upon success, matching ftr_reg entry for id.
* - NULL on failure. It is upto the caller to decide
* the impact of a failure.
*/
static struct arm64_ftr_reg *get_arm64_ftr_reg(u32 sys_id)
{
const struct __ftr_reg_entry *ret;
ret = bsearch((const void *)(unsigned long)sys_id,
arm64_ftr_regs,
ARRAY_SIZE(arm64_ftr_regs),
sizeof(arm64_ftr_regs[0]),
search_cmp_ftr_reg);
if (ret)
return ret->reg;
return NULL;
}
static u64 arm64_ftr_set_value(const struct arm64_ftr_bits *ftrp, s64 reg,
s64 ftr_val)
{
u64 mask = arm64_ftr_mask(ftrp);
reg &= ~mask;
reg |= (ftr_val << ftrp->shift) & mask;
return reg;
}
static s64 arm64_ftr_safe_value(const struct arm64_ftr_bits *ftrp, s64 new,
s64 cur)
{
s64 ret = 0;
switch (ftrp->type) {
case FTR_EXACT:
ret = ftrp->safe_val;
break;
case FTR_LOWER_SAFE:
ret = new < cur ? new : cur;
break;
case FTR_HIGHER_SAFE:
ret = new > cur ? new : cur;
break;
default:
BUG();
}
return ret;
}
static void __init sort_ftr_regs(void)
{
int i;
/* Check that the array is sorted so that we can do the binary search */
for (i = 1; i < ARRAY_SIZE(arm64_ftr_regs); i++)
BUG_ON(arm64_ftr_regs[i].sys_id < arm64_ftr_regs[i - 1].sys_id);
}
/*
* Initialise the CPU feature register from Boot CPU values.
* Also initiliases the strict_mask for the register.
*/
static void __init init_cpu_ftr_reg(u32 sys_reg, u64 new)
{
u64 val = 0;
u64 strict_mask = ~0x0ULL;
const struct arm64_ftr_bits *ftrp;
struct arm64_ftr_reg *reg = get_arm64_ftr_reg(sys_reg);
BUG_ON(!reg);
for (ftrp = reg->ftr_bits; ftrp->width; ftrp++) {
s64 ftr_new = arm64_ftr_value(ftrp, new);
val = arm64_ftr_set_value(ftrp, val, ftr_new);
if (!ftrp->strict)
strict_mask &= ~arm64_ftr_mask(ftrp);
}
reg->sys_val = val;
reg->strict_mask = strict_mask;
}
void __init init_cpu_features(struct cpuinfo_arm64 *info)
{
/* Before we start using the tables, make sure it is sorted */
sort_ftr_regs();
init_cpu_ftr_reg(SYS_CTR_EL0, info->reg_ctr);
init_cpu_ftr_reg(SYS_DCZID_EL0, info->reg_dczid);
init_cpu_ftr_reg(SYS_CNTFRQ_EL0, info->reg_cntfrq);
init_cpu_ftr_reg(SYS_ID_AA64DFR0_EL1, info->reg_id_aa64dfr0);
init_cpu_ftr_reg(SYS_ID_AA64DFR1_EL1, info->reg_id_aa64dfr1);
init_cpu_ftr_reg(SYS_ID_AA64ISAR0_EL1, info->reg_id_aa64isar0);
init_cpu_ftr_reg(SYS_ID_AA64ISAR1_EL1, info->reg_id_aa64isar1);
init_cpu_ftr_reg(SYS_ID_AA64MMFR0_EL1, info->reg_id_aa64mmfr0);
init_cpu_ftr_reg(SYS_ID_AA64MMFR1_EL1, info->reg_id_aa64mmfr1);
init_cpu_ftr_reg(SYS_ID_AA64MMFR2_EL1, info->reg_id_aa64mmfr2);
init_cpu_ftr_reg(SYS_ID_AA64PFR0_EL1, info->reg_id_aa64pfr0);
init_cpu_ftr_reg(SYS_ID_AA64PFR1_EL1, info->reg_id_aa64pfr1);
if (id_aa64pfr0_32bit_el0(info->reg_id_aa64pfr0)) {
init_cpu_ftr_reg(SYS_ID_DFR0_EL1, info->reg_id_dfr0);
init_cpu_ftr_reg(SYS_ID_ISAR0_EL1, info->reg_id_isar0);
init_cpu_ftr_reg(SYS_ID_ISAR1_EL1, info->reg_id_isar1);
init_cpu_ftr_reg(SYS_ID_ISAR2_EL1, info->reg_id_isar2);
init_cpu_ftr_reg(SYS_ID_ISAR3_EL1, info->reg_id_isar3);
init_cpu_ftr_reg(SYS_ID_ISAR4_EL1, info->reg_id_isar4);
init_cpu_ftr_reg(SYS_ID_ISAR5_EL1, info->reg_id_isar5);
init_cpu_ftr_reg(SYS_ID_MMFR0_EL1, info->reg_id_mmfr0);
init_cpu_ftr_reg(SYS_ID_MMFR1_EL1, info->reg_id_mmfr1);
init_cpu_ftr_reg(SYS_ID_MMFR2_EL1, info->reg_id_mmfr2);
init_cpu_ftr_reg(SYS_ID_MMFR3_EL1, info->reg_id_mmfr3);
init_cpu_ftr_reg(SYS_ID_PFR0_EL1, info->reg_id_pfr0);
init_cpu_ftr_reg(SYS_ID_PFR1_EL1, info->reg_id_pfr1);
init_cpu_ftr_reg(SYS_MVFR0_EL1, info->reg_mvfr0);
init_cpu_ftr_reg(SYS_MVFR1_EL1, info->reg_mvfr1);
init_cpu_ftr_reg(SYS_MVFR2_EL1, info->reg_mvfr2);
}
}
static void update_cpu_ftr_reg(struct arm64_ftr_reg *reg, u64 new)
{
const struct arm64_ftr_bits *ftrp;
for (ftrp = reg->ftr_bits; ftrp->width; ftrp++) {
s64 ftr_cur = arm64_ftr_value(ftrp, reg->sys_val);
s64 ftr_new = arm64_ftr_value(ftrp, new);
if (ftr_cur == ftr_new)
continue;
/* Find a safe value */
ftr_new = arm64_ftr_safe_value(ftrp, ftr_new, ftr_cur);
reg->sys_val = arm64_ftr_set_value(ftrp, reg->sys_val, ftr_new);
}
}
static int check_update_ftr_reg(u32 sys_id, int cpu, u64 val, u64 boot)
{
struct arm64_ftr_reg *regp = get_arm64_ftr_reg(sys_id);
BUG_ON(!regp);
update_cpu_ftr_reg(regp, val);
if ((boot & regp->strict_mask) == (val & regp->strict_mask))
return 0;
pr_warn("SANITY CHECK: Unexpected variation in %s. Boot CPU: %#016llx, CPU%d: %#016llx\n",
regp->name, boot, cpu, val);
return 1;
}
/*
* Update system wide CPU feature registers with the values from a
* non-boot CPU. Also performs SANITY checks to make sure that there
* aren't any insane variations from that of the boot CPU.
*/
void update_cpu_features(int cpu,
struct cpuinfo_arm64 *info,
struct cpuinfo_arm64 *boot)
{
int taint = 0;
/*
* The kernel can handle differing I-cache policies, but otherwise
* caches should look identical. Userspace JITs will make use of
* *minLine.
*/
taint |= check_update_ftr_reg(SYS_CTR_EL0, cpu,
info->reg_ctr, boot->reg_ctr);
/*
* Userspace may perform DC ZVA instructions. Mismatched block sizes
* could result in too much or too little memory being zeroed if a
* process is preempted and migrated between CPUs.
*/
taint |= check_update_ftr_reg(SYS_DCZID_EL0, cpu,
info->reg_dczid, boot->reg_dczid);
/* If different, timekeeping will be broken (especially with KVM) */
taint |= check_update_ftr_reg(SYS_CNTFRQ_EL0, cpu,
info->reg_cntfrq, boot->reg_cntfrq);
/*
* The kernel uses self-hosted debug features and expects CPUs to
* support identical debug features. We presently need CTX_CMPs, WRPs,
* and BRPs to be identical.
* ID_AA64DFR1 is currently RES0.
*/
taint |= check_update_ftr_reg(SYS_ID_AA64DFR0_EL1, cpu,
info->reg_id_aa64dfr0, boot->reg_id_aa64dfr0);
taint |= check_update_ftr_reg(SYS_ID_AA64DFR1_EL1, cpu,
info->reg_id_aa64dfr1, boot->reg_id_aa64dfr1);
/*
* Even in big.LITTLE, processors should be identical instruction-set
* wise.
*/
taint |= check_update_ftr_reg(SYS_ID_AA64ISAR0_EL1, cpu,
info->reg_id_aa64isar0, boot->reg_id_aa64isar0);
taint |= check_update_ftr_reg(SYS_ID_AA64ISAR1_EL1, cpu,
info->reg_id_aa64isar1, boot->reg_id_aa64isar1);
/*
* Differing PARange support is fine as long as all peripherals and
* memory are mapped within the minimum PARange of all CPUs.
* Linux should not care about secure memory.
*/
taint |= check_update_ftr_reg(SYS_ID_AA64MMFR0_EL1, cpu,
info->reg_id_aa64mmfr0, boot->reg_id_aa64mmfr0);
taint |= check_update_ftr_reg(SYS_ID_AA64MMFR1_EL1, cpu,
info->reg_id_aa64mmfr1, boot->reg_id_aa64mmfr1);
taint |= check_update_ftr_reg(SYS_ID_AA64MMFR2_EL1, cpu,
info->reg_id_aa64mmfr2, boot->reg_id_aa64mmfr2);
/*
* EL3 is not our concern.
* ID_AA64PFR1 is currently RES0.
*/
taint |= check_update_ftr_reg(SYS_ID_AA64PFR0_EL1, cpu,
info->reg_id_aa64pfr0, boot->reg_id_aa64pfr0);
taint |= check_update_ftr_reg(SYS_ID_AA64PFR1_EL1, cpu,
info->reg_id_aa64pfr1, boot->reg_id_aa64pfr1);
/*
* If we have AArch32, we care about 32-bit features for compat.
* If the system doesn't support AArch32, don't update them.
*/
if (id_aa64pfr0_32bit_el0(read_system_reg(SYS_ID_AA64PFR0_EL1)) &&
id_aa64pfr0_32bit_el0(info->reg_id_aa64pfr0)) {
taint |= check_update_ftr_reg(SYS_ID_DFR0_EL1, cpu,
info->reg_id_dfr0, boot->reg_id_dfr0);
taint |= check_update_ftr_reg(SYS_ID_ISAR0_EL1, cpu,
info->reg_id_isar0, boot->reg_id_isar0);
taint |= check_update_ftr_reg(SYS_ID_ISAR1_EL1, cpu,
info->reg_id_isar1, boot->reg_id_isar1);
taint |= check_update_ftr_reg(SYS_ID_ISAR2_EL1, cpu,
info->reg_id_isar2, boot->reg_id_isar2);
taint |= check_update_ftr_reg(SYS_ID_ISAR3_EL1, cpu,
info->reg_id_isar3, boot->reg_id_isar3);
taint |= check_update_ftr_reg(SYS_ID_ISAR4_EL1, cpu,
info->reg_id_isar4, boot->reg_id_isar4);
taint |= check_update_ftr_reg(SYS_ID_ISAR5_EL1, cpu,
info->reg_id_isar5, boot->reg_id_isar5);
/*
* Regardless of the value of the AuxReg field, the AIFSR, ADFSR, and
* ACTLR formats could differ across CPUs and therefore would have to
* be trapped for virtualization anyway.
*/
taint |= check_update_ftr_reg(SYS_ID_MMFR0_EL1, cpu,
info->reg_id_mmfr0, boot->reg_id_mmfr0);
taint |= check_update_ftr_reg(SYS_ID_MMFR1_EL1, cpu,
info->reg_id_mmfr1, boot->reg_id_mmfr1);
taint |= check_update_ftr_reg(SYS_ID_MMFR2_EL1, cpu,
info->reg_id_mmfr2, boot->reg_id_mmfr2);
taint |= check_update_ftr_reg(SYS_ID_MMFR3_EL1, cpu,
info->reg_id_mmfr3, boot->reg_id_mmfr3);
taint |= check_update_ftr_reg(SYS_ID_PFR0_EL1, cpu,
info->reg_id_pfr0, boot->reg_id_pfr0);
taint |= check_update_ftr_reg(SYS_ID_PFR1_EL1, cpu,
info->reg_id_pfr1, boot->reg_id_pfr1);
taint |= check_update_ftr_reg(SYS_MVFR0_EL1, cpu,
info->reg_mvfr0, boot->reg_mvfr0);
taint |= check_update_ftr_reg(SYS_MVFR1_EL1, cpu,
info->reg_mvfr1, boot->reg_mvfr1);
taint |= check_update_ftr_reg(SYS_MVFR2_EL1, cpu,
info->reg_mvfr2, boot->reg_mvfr2);
}
/*
* Mismatched CPU features are a recipe for disaster. Don't even
* pretend to support them.
*/
WARN_TAINT_ONCE(taint, TAINT_CPU_OUT_OF_SPEC,
"Unsupported CPU feature variation.\n");
}
u64 read_system_reg(u32 id)
{
struct arm64_ftr_reg *regp = get_arm64_ftr_reg(id);
/* We shouldn't get a request for an unsupported register */
BUG_ON(!regp);
return regp->sys_val;
}
/*
* __raw_read_system_reg() - Used by a STARTING cpu before cpuinfo is populated.
* Read the system register on the current CPU
*/
static u64 __raw_read_system_reg(u32 sys_id)
{
switch (sys_id) {
case SYS_ID_PFR0_EL1: return read_cpuid(ID_PFR0_EL1);
case SYS_ID_PFR1_EL1: return read_cpuid(ID_PFR1_EL1);
case SYS_ID_DFR0_EL1: return read_cpuid(ID_DFR0_EL1);
case SYS_ID_MMFR0_EL1: return read_cpuid(ID_MMFR0_EL1);
case SYS_ID_MMFR1_EL1: return read_cpuid(ID_MMFR1_EL1);
case SYS_ID_MMFR2_EL1: return read_cpuid(ID_MMFR2_EL1);
case SYS_ID_MMFR3_EL1: return read_cpuid(ID_MMFR3_EL1);
case SYS_ID_ISAR0_EL1: return read_cpuid(ID_ISAR0_EL1);
case SYS_ID_ISAR1_EL1: return read_cpuid(ID_ISAR1_EL1);
case SYS_ID_ISAR2_EL1: return read_cpuid(ID_ISAR2_EL1);
case SYS_ID_ISAR3_EL1: return read_cpuid(ID_ISAR3_EL1);
case SYS_ID_ISAR4_EL1: return read_cpuid(ID_ISAR4_EL1);
case SYS_ID_ISAR5_EL1: return read_cpuid(ID_ISAR5_EL1);
case SYS_MVFR0_EL1: return read_cpuid(MVFR0_EL1);
case SYS_MVFR1_EL1: return read_cpuid(MVFR1_EL1);
case SYS_MVFR2_EL1: return read_cpuid(MVFR2_EL1);
case SYS_ID_AA64PFR0_EL1: return read_cpuid(ID_AA64PFR0_EL1);
case SYS_ID_AA64PFR1_EL1: return read_cpuid(ID_AA64PFR1_EL1);
case SYS_ID_AA64DFR0_EL1: return read_cpuid(ID_AA64DFR0_EL1);
case SYS_ID_AA64DFR1_EL1: return read_cpuid(ID_AA64DFR1_EL1);
case SYS_ID_AA64MMFR0_EL1: return read_cpuid(ID_AA64MMFR0_EL1);
case SYS_ID_AA64MMFR1_EL1: return read_cpuid(ID_AA64MMFR1_EL1);
case SYS_ID_AA64MMFR2_EL1: return read_cpuid(ID_AA64MMFR2_EL1);
case SYS_ID_AA64ISAR0_EL1: return read_cpuid(ID_AA64ISAR0_EL1);
case SYS_ID_AA64ISAR1_EL1: return read_cpuid(ID_AA64ISAR1_EL1);
case SYS_CNTFRQ_EL0: return read_cpuid(CNTFRQ_EL0);
case SYS_CTR_EL0: return read_cpuid(CTR_EL0);
case SYS_DCZID_EL0: return read_cpuid(DCZID_EL0);
default:
BUG();
return 0;
}
}
#include <linux/irqchip/arm-gic-v3.h>
static bool
feature_matches(u64 reg, const struct arm64_cpu_capabilities *entry)
{
int val = cpuid_feature_extract_field(reg, entry->field_pos, entry->sign);
return val >= entry->min_field_value;
}
static bool
has_cpuid_feature(const struct arm64_cpu_capabilities *entry, int scope)
{
u64 val;
WARN_ON(scope == SCOPE_LOCAL_CPU && preemptible());
if (scope == SCOPE_SYSTEM)
val = read_system_reg(entry->sys_reg);
else
val = __raw_read_system_reg(entry->sys_reg);
return feature_matches(val, entry);
}
static bool has_useable_gicv3_cpuif(const struct arm64_cpu_capabilities *entry, int scope)
{
bool has_sre;
if (!has_cpuid_feature(entry, scope))
return false;
has_sre = gic_enable_sre();
if (!has_sre)
pr_warn_once("%s present but disabled by higher exception level\n",
entry->desc);
return has_sre;
}
static bool has_no_hw_prefetch(const struct arm64_cpu_capabilities *entry, int __unused)
{
u32 midr = read_cpuid_id();
u32 rv_min, rv_max;
/* Cavium ThunderX pass 1.x and 2.x */
rv_min = 0;
rv_max = (1 << MIDR_VARIANT_SHIFT) | MIDR_REVISION_MASK;
return MIDR_IS_CPU_MODEL_RANGE(midr, MIDR_THUNDERX, rv_min, rv_max);
}
static bool runs_at_el2(const struct arm64_cpu_capabilities *entry, int __unused)
{
return is_kernel_in_hyp_mode();
}
static bool hyp_offset_low(const struct arm64_cpu_capabilities *entry,
int __unused)
{
phys_addr_t idmap_addr = virt_to_phys(__hyp_idmap_text_start);
/*
* Activate the lower HYP offset only if:
* - the idmap doesn't clash with it,
* - the kernel is not running at EL2.
*/
return idmap_addr > GENMASK(VA_BITS - 2, 0) && !is_kernel_in_hyp_mode();
}
#ifdef CONFIG_UNMAP_KERNEL_AT_EL0
static int __kpti_forced; /* 0: not forced, >0: forced on, <0: forced off */
static bool unmap_kernel_at_el0(const struct arm64_cpu_capabilities *entry,
int __unused)
{
char const *str = "command line option";
u64 pfr0 = read_system_reg(SYS_ID_AA64PFR0_EL1);
/*
* For reasons that aren't entirely clear, enabling KPTI on Cavium
* ThunderX leads to apparent I-cache corruption of kernel text, which
* ends as well as you might imagine. Don't even try.
*/
if (cpus_have_const_cap(ARM64_WORKAROUND_CAVIUM_27456)) {
str = "ARM64_WORKAROUND_CAVIUM_27456";
__kpti_forced = -1;
}
/* Forced? */
if (__kpti_forced) {
pr_info_once("kernel page table isolation forced %s by %s\n",
__kpti_forced > 0 ? "ON" : "OFF", str);
return __kpti_forced > 0;
}
/* Useful for KASLR robustness */
if (IS_ENABLED(CONFIG_RANDOMIZE_BASE))
return true;
/* Don't force KPTI for CPUs that are not vulnerable */
switch (read_cpuid_id() & MIDR_CPU_MODEL_MASK) {
case MIDR_CAVIUM_THUNDERX2:
case MIDR_BRCM_VULCAN:
return false;
}
/* Defer to CPU feature registers */
return !cpuid_feature_extract_unsigned_field(pfr0,
ID_AA64PFR0_CSV3_SHIFT);
}
static int kpti_install_ng_mappings(void *__unused)
{
typedef void (kpti_remap_fn)(int, int, phys_addr_t);
extern kpti_remap_fn idmap_kpti_install_ng_mappings;
kpti_remap_fn *remap_fn;
static bool kpti_applied = false;
int cpu = smp_processor_id();
if (kpti_applied)
return 0;
remap_fn = (void *)__pa_symbol(idmap_kpti_install_ng_mappings);
cpu_install_idmap();
remap_fn(cpu, num_online_cpus(), __pa_symbol(swapper_pg_dir));
cpu_uninstall_idmap();
if (!cpu)
kpti_applied = true;
return 0;
}
static int __init parse_kpti(char *str)
{
bool enabled;
int ret = strtobool(str, &enabled);
if (ret)
return ret;
__kpti_forced = enabled ? 1 : -1;
return 0;
}
early_param("kpti", parse_kpti);
#endif /* CONFIG_UNMAP_KERNEL_AT_EL0 */
static int cpu_copy_el2regs(void *__unused)
{
/*
* Copy register values that aren't redirected by hardware.
*
* Before code patching, we only set tpidr_el1, all CPUs need to copy
* this value to tpidr_el2 before we patch the code. Once we've done
* that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
* do anything here.
*/
if (!alternatives_applied)
write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
return 0;
}
static const struct arm64_cpu_capabilities arm64_features[] = {
{
.desc = "GIC system register CPU interface",
.capability = ARM64_HAS_SYSREG_GIC_CPUIF,
.def_scope = SCOPE_SYSTEM,
.matches = has_useable_gicv3_cpuif,
.sys_reg = SYS_ID_AA64PFR0_EL1,
.field_pos = ID_AA64PFR0_GIC_SHIFT,
.sign = FTR_UNSIGNED,
.min_field_value = 1,
},
#ifdef CONFIG_ARM64_PAN
{
.desc = "Privileged Access Never",
.capability = ARM64_HAS_PAN,
.def_scope = SCOPE_SYSTEM,
.matches = has_cpuid_feature,
.sys_reg = SYS_ID_AA64MMFR1_EL1,
.field_pos = ID_AA64MMFR1_PAN_SHIFT,
.sign = FTR_UNSIGNED,
.min_field_value = 1,
.enable = cpu_enable_pan,
},
#endif /* CONFIG_ARM64_PAN */
#if defined(CONFIG_AS_LSE) && defined(CONFIG_ARM64_LSE_ATOMICS)
{
.desc = "LSE atomic instructions",
.capability = ARM64_HAS_LSE_ATOMICS,
.def_scope = SCOPE_SYSTEM,
.matches = has_cpuid_feature,
.sys_reg = SYS_ID_AA64ISAR0_EL1,
.field_pos = ID_AA64ISAR0_ATOMICS_SHIFT,
.sign = FTR_UNSIGNED,
.min_field_value = 2,
},
#endif /* CONFIG_AS_LSE && CONFIG_ARM64_LSE_ATOMICS */
{
.desc = "Software prefetching using PRFM",
.capability = ARM64_HAS_NO_HW_PREFETCH,
.def_scope = SCOPE_SYSTEM,
.matches = has_no_hw_prefetch,
},
#ifdef CONFIG_ARM64_UAO
{
.desc = "User Access Override",
.capability = ARM64_HAS_UAO,
.def_scope = SCOPE_SYSTEM,
.matches = has_cpuid_feature,
.sys_reg = SYS_ID_AA64MMFR2_EL1,
.field_pos = ID_AA64MMFR2_UAO_SHIFT,
.min_field_value = 1,
.enable = cpu_enable_uao,
},
#endif /* CONFIG_ARM64_UAO */
#ifdef CONFIG_ARM64_PAN
{
.capability = ARM64_ALT_PAN_NOT_UAO,
.def_scope = SCOPE_SYSTEM,
.matches = cpufeature_pan_not_uao,
},
#endif /* CONFIG_ARM64_PAN */
{
.desc = "Virtualization Host Extensions",
.capability = ARM64_HAS_VIRT_HOST_EXTN,
.def_scope = SCOPE_SYSTEM,
.matches = runs_at_el2,
.enable = cpu_copy_el2regs,
},
{
.desc = "32-bit EL0 Support",
.capability = ARM64_HAS_32BIT_EL0,
.def_scope = SCOPE_SYSTEM,
.matches = has_cpuid_feature,
.sys_reg = SYS_ID_AA64PFR0_EL1,
.sign = FTR_UNSIGNED,
.field_pos = ID_AA64PFR0_EL0_SHIFT,
.min_field_value = ID_AA64PFR0_EL0_32BIT_64BIT,
},
{
.desc = "Reduced HYP mapping offset",
.capability = ARM64_HYP_OFFSET_LOW,
.def_scope = SCOPE_SYSTEM,
.matches = hyp_offset_low,
},
#ifdef CONFIG_UNMAP_KERNEL_AT_EL0
{
.desc = "Kernel page table isolation (KPTI)",
.capability = ARM64_UNMAP_KERNEL_AT_EL0,
.def_scope = SCOPE_SYSTEM,
.matches = unmap_kernel_at_el0,
.enable = kpti_install_ng_mappings,
},
#endif
{},
};
#define HWCAP_CAP(reg, field, s, min_value, type, cap) \
{ \
.desc = #cap, \
.def_scope = SCOPE_SYSTEM, \
.matches = has_cpuid_feature, \
.sys_reg = reg, \
.field_pos = field, \
.sign = s, \
.min_field_value = min_value, \
.hwcap_type = type, \
.hwcap = cap, \
}
static const struct arm64_cpu_capabilities arm64_elf_hwcaps[] = {
HWCAP_CAP(SYS_ID_AA64ISAR0_EL1, ID_AA64ISAR0_AES_SHIFT, FTR_UNSIGNED, 2, CAP_HWCAP, HWCAP_PMULL),
HWCAP_CAP(SYS_ID_AA64ISAR0_EL1, ID_AA64ISAR0_AES_SHIFT, FTR_UNSIGNED, 1, CAP_HWCAP, HWCAP_AES),
HWCAP_CAP(SYS_ID_AA64ISAR0_EL1, ID_AA64ISAR0_SHA1_SHIFT, FTR_UNSIGNED, 1, CAP_HWCAP, HWCAP_SHA1),
HWCAP_CAP(SYS_ID_AA64ISAR0_EL1, ID_AA64ISAR0_SHA2_SHIFT, FTR_UNSIGNED, 1, CAP_HWCAP, HWCAP_SHA2),
HWCAP_CAP(SYS_ID_AA64ISAR0_EL1, ID_AA64ISAR0_CRC32_SHIFT, FTR_UNSIGNED, 1, CAP_HWCAP, HWCAP_CRC32),
HWCAP_CAP(SYS_ID_AA64ISAR0_EL1, ID_AA64ISAR0_ATOMICS_SHIFT, FTR_UNSIGNED, 2, CAP_HWCAP, HWCAP_ATOMICS),
HWCAP_CAP(SYS_ID_AA64PFR0_EL1, ID_AA64PFR0_FP_SHIFT, FTR_SIGNED, 0, CAP_HWCAP, HWCAP_FP),
HWCAP_CAP(SYS_ID_AA64PFR0_EL1, ID_AA64PFR0_FP_SHIFT, FTR_SIGNED, 1, CAP_HWCAP, HWCAP_FPHP),
HWCAP_CAP(SYS_ID_AA64PFR0_EL1, ID_AA64PFR0_ASIMD_SHIFT, FTR_SIGNED, 0, CAP_HWCAP, HWCAP_ASIMD),
HWCAP_CAP(SYS_ID_AA64PFR0_EL1, ID_AA64PFR0_ASIMD_SHIFT, FTR_SIGNED, 1, CAP_HWCAP, HWCAP_ASIMDHP),
{},
};
static const struct arm64_cpu_capabilities compat_elf_hwcaps[] = {
#ifdef CONFIG_COMPAT
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_AES_SHIFT, FTR_UNSIGNED, 2, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_PMULL),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_AES_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_AES),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_SHA1_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_SHA1),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_SHA2_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_SHA2),
HWCAP_CAP(SYS_ID_ISAR5_EL1, ID_ISAR5_CRC32_SHIFT, FTR_UNSIGNED, 1, CAP_COMPAT_HWCAP2, COMPAT_HWCAP2_CRC32),
#endif
{},
};
static void __init cap_set_elf_hwcap(const struct arm64_cpu_capabilities *cap)
{
switch (cap->hwcap_type) {
case CAP_HWCAP:
elf_hwcap |= cap->hwcap;
break;
#ifdef CONFIG_COMPAT
case CAP_COMPAT_HWCAP:
compat_elf_hwcap |= (u32)cap->hwcap;
break;
case CAP_COMPAT_HWCAP2:
compat_elf_hwcap2 |= (u32)cap->hwcap;
break;
#endif
default:
WARN_ON(1);
break;
}
}
/* Check if we have a particular HWCAP enabled */
static bool cpus_have_elf_hwcap(const struct arm64_cpu_capabilities *cap)
{
bool rc;
switch (cap->hwcap_type) {
case CAP_HWCAP:
rc = (elf_hwcap & cap->hwcap) != 0;
break;
#ifdef CONFIG_COMPAT
case CAP_COMPAT_HWCAP:
rc = (compat_elf_hwcap & (u32)cap->hwcap) != 0;
break;
case CAP_COMPAT_HWCAP2:
rc = (compat_elf_hwcap2 & (u32)cap->hwcap) != 0;
break;
#endif
default:
WARN_ON(1);
rc = false;
}
return rc;
}
static void __init setup_elf_hwcaps(const struct arm64_cpu_capabilities *hwcaps)
{
for (; hwcaps->matches; hwcaps++)
if (hwcaps->matches(hwcaps, hwcaps->def_scope))
cap_set_elf_hwcap(hwcaps);
}
/*
* Check if the current CPU has a given feature capability.
* Should be called from non-preemptible context.
*/
static bool __this_cpu_has_cap(const struct arm64_cpu_capabilities *cap_array,
unsigned int cap)
{
const struct arm64_cpu_capabilities *caps;
if (WARN_ON(preemptible()))
return false;
for (caps = cap_array; caps->matches; caps++)
if (caps->capability == cap &&
caps->matches(caps, SCOPE_LOCAL_CPU))
return true;
return false;
}
void update_cpu_capabilities(const struct arm64_cpu_capabilities *caps,
const char *info)
{
for (; caps->matches; caps++) {
if (!caps->matches(caps, caps->def_scope))
continue;
if (!cpus_have_cap(caps->capability) && caps->desc)
pr_info("%s %s\n", info, caps->desc);
cpus_set_cap(caps->capability);
}
}
/*
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
* Run through the enabled capabilities and enable() it on all active
* CPUs
*/
void __init enable_cpu_capabilities(const struct arm64_cpu_capabilities *caps)
{
arm64/cpufeature: don't use mutex in bringup path commit 63a1e1c95e60e798fa09ab3c536fb555aa5bbf2b upstream. Currently, cpus_set_cap() calls static_branch_enable_cpuslocked(), which must take the jump_label mutex. We call cpus_set_cap() in the secondary bringup path, from the idle thread where interrupts are disabled. Taking a mutex in this path "is a NONO" regardless of whether it's contended, and something we must avoid. We didn't spot this until recently, as ___might_sleep() won't warn for this case until all CPUs have been brought up. This patch avoids taking the mutex in the secondary bringup path. The poking of static keys is deferred until enable_cpu_capabilities(), which runs in a suitable context on the boot CPU. To account for the static keys being set later, cpus_have_const_cap() is updated to use another static key to check whether the const cap keys have been initialised, falling back to the caps bitmap until this is the case. This means that users of cpus_have_const_cap() gain should only gain a single additional NOP in the fast path once the const caps are initialised, but should always see the current cap value. The hyp code should never dereference the caps array, since the caps are initialized before we run the module initcall to initialise hyp. A check is added to the hyp init code to document this requirement. This change will sidestep a number of issues when the upcoming hotplug locking rework is merged. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Marc Zyniger <marc.zyngier@arm.com> Reviewed-by: Suzuki Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Cc: Christoffer Dall <christoffer.dall@linaro.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Sebastian Sewior <bigeasy@linutronix.de> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> [4.9: this avoids an IPI before GICv3 is up, preventing a boot time crash] Signed-off-by: Mark Rutland <mark.rutland@arm.com> [v4.9 backport] Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-05-16 23:18:05 +09:00
for (; caps->matches; caps++) {
unsigned int num = caps->capability;
if (!cpus_have_cap(num))
continue;
/* Ensure cpus_have_const_cap(num) works */
static_branch_enable(&cpu_hwcap_keys[num]);
if (caps->enable) {
/*
* Use stop_machine() as it schedules the work allowing
* us to modify PSTATE, instead of on_each_cpu() which
* uses an IPI, giving us a PSTATE that disappears when
* we return.
*/
stop_machine(caps->enable, (void *)caps, cpu_online_mask);
arm64/cpufeature: don't use mutex in bringup path commit 63a1e1c95e60e798fa09ab3c536fb555aa5bbf2b upstream. Currently, cpus_set_cap() calls static_branch_enable_cpuslocked(), which must take the jump_label mutex. We call cpus_set_cap() in the secondary bringup path, from the idle thread where interrupts are disabled. Taking a mutex in this path "is a NONO" regardless of whether it's contended, and something we must avoid. We didn't spot this until recently, as ___might_sleep() won't warn for this case until all CPUs have been brought up. This patch avoids taking the mutex in the secondary bringup path. The poking of static keys is deferred until enable_cpu_capabilities(), which runs in a suitable context on the boot CPU. To account for the static keys being set later, cpus_have_const_cap() is updated to use another static key to check whether the const cap keys have been initialised, falling back to the caps bitmap until this is the case. This means that users of cpus_have_const_cap() gain should only gain a single additional NOP in the fast path once the const caps are initialised, but should always see the current cap value. The hyp code should never dereference the caps array, since the caps are initialized before we run the module initcall to initialise hyp. A check is added to the hyp init code to document this requirement. This change will sidestep a number of issues when the upcoming hotplug locking rework is merged. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Marc Zyniger <marc.zyngier@arm.com> Reviewed-by: Suzuki Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Cc: Christoffer Dall <christoffer.dall@linaro.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Sebastian Sewior <bigeasy@linutronix.de> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> [4.9: this avoids an IPI before GICv3 is up, preventing a boot time crash] Signed-off-by: Mark Rutland <mark.rutland@arm.com> [v4.9 backport] Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-05-16 23:18:05 +09:00
}
}
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
}
/*
* Flag to indicate if we have computed the system wide
* capabilities based on the boot time active CPUs. This
* will be used to determine if a new booting CPU should
* go through the verification process to make sure that it
* supports the system capabilities, without using a hotplug
* notifier.
*/
static bool sys_caps_initialised;
static inline void set_sys_caps_initialised(void)
{
sys_caps_initialised = true;
}
/*
* Check for CPU features that are used in early boot
* based on the Boot CPU value.
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
*/
static void check_early_cpu_features(void)
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
{
verify_cpu_run_el();
verify_cpu_asid_bits();
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
}
static void
verify_local_elf_hwcaps(const struct arm64_cpu_capabilities *caps)
{
for (; caps->matches; caps++)
if (cpus_have_elf_hwcap(caps) && !caps->matches(caps, SCOPE_LOCAL_CPU)) {
pr_crit("CPU%d: missing HWCAP: %s\n",
smp_processor_id(), caps->desc);
cpu_die_early();
}
}
static void
verify_local_cpu_features(const struct arm64_cpu_capabilities *caps_list)
{
const struct arm64_cpu_capabilities *caps = caps_list;
for (; caps->matches; caps++) {
if (!cpus_have_cap(caps->capability))
continue;
/*
* If the new CPU misses an advertised feature, we cannot proceed
* further, park the cpu.
*/
if (!__this_cpu_has_cap(caps_list, caps->capability)) {
pr_crit("CPU%d: missing feature: %s\n",
smp_processor_id(), caps->desc);
cpu_die_early();
}
if (caps->enable)
caps->enable((void *)caps);
}
}
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
/*
* Run through the enabled system capabilities and enable() it on this CPU.
* The capabilities were decided based on the available CPUs at the boot time.
* Any new CPU should match the system wide status of the capability. If the
* new CPU doesn't have a capability which the system now has enabled, we
* cannot do anything to fix it up and could cause unexpected failures. So
* we park the CPU.
*/
arm64: Rearrange CPU errata workaround checks Right now we run through the work around checks on a CPU from __cpuinfo_store_cpu. There are some problems with that: 1) We initialise the system wide CPU feature registers only after the Boot CPU updates its cpuinfo. Now, if a work around depends on the variance of a CPU ID feature (e.g, check for Cache Line size mismatch), we have no way of performing it cleanly for the boot CPU. 2) It is out of place, invoked from __cpuinfo_store_cpu() in cpuinfo.c. It is not an obvious place for that. This patch rearranges the CPU specific capability(aka work around) checks. 1) At the moment we use verify_local_cpu_capabilities() to check if a new CPU has all the system advertised features. Use this for the secondary CPUs to perform the work around check. For that we rename verify_local_cpu_capabilities() => check_local_cpu_capabilities() which: If the system wide capabilities haven't been initialised (i.e, the CPU is activated at the boot), update the system wide detected work arounds. Otherwise (i.e a CPU hotplugged in later) verify that this CPU conforms to the system wide capabilities. 2) Boot CPU updates the work arounds from smp_prepare_boot_cpu() after we have initialised the system wide CPU feature values. Cc: Mark Rutland <mark.rutland@arm.com> Cc: Andre Przywara <andre.przywara@arm.com> Cc: Will Deacon <will.deacon@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2016-09-09 22:07:10 +09:00
static void verify_local_cpu_capabilities(void)
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
{
arm64: Rearrange CPU errata workaround checks Right now we run through the work around checks on a CPU from __cpuinfo_store_cpu. There are some problems with that: 1) We initialise the system wide CPU feature registers only after the Boot CPU updates its cpuinfo. Now, if a work around depends on the variance of a CPU ID feature (e.g, check for Cache Line size mismatch), we have no way of performing it cleanly for the boot CPU. 2) It is out of place, invoked from __cpuinfo_store_cpu() in cpuinfo.c. It is not an obvious place for that. This patch rearranges the CPU specific capability(aka work around) checks. 1) At the moment we use verify_local_cpu_capabilities() to check if a new CPU has all the system advertised features. Use this for the secondary CPUs to perform the work around check. For that we rename verify_local_cpu_capabilities() => check_local_cpu_capabilities() which: If the system wide capabilities haven't been initialised (i.e, the CPU is activated at the boot), update the system wide detected work arounds. Otherwise (i.e a CPU hotplugged in later) verify that this CPU conforms to the system wide capabilities. 2) Boot CPU updates the work arounds from smp_prepare_boot_cpu() after we have initialised the system wide CPU feature values. Cc: Mark Rutland <mark.rutland@arm.com> Cc: Andre Przywara <andre.przywara@arm.com> Cc: Will Deacon <will.deacon@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2016-09-09 22:07:10 +09:00
verify_local_cpu_errata_workarounds();
verify_local_cpu_features(arm64_features);
verify_local_elf_hwcaps(arm64_elf_hwcaps);
if (system_supports_32bit_el0())
verify_local_elf_hwcaps(compat_elf_hwcaps);
}
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
arm64: Rearrange CPU errata workaround checks Right now we run through the work around checks on a CPU from __cpuinfo_store_cpu. There are some problems with that: 1) We initialise the system wide CPU feature registers only after the Boot CPU updates its cpuinfo. Now, if a work around depends on the variance of a CPU ID feature (e.g, check for Cache Line size mismatch), we have no way of performing it cleanly for the boot CPU. 2) It is out of place, invoked from __cpuinfo_store_cpu() in cpuinfo.c. It is not an obvious place for that. This patch rearranges the CPU specific capability(aka work around) checks. 1) At the moment we use verify_local_cpu_capabilities() to check if a new CPU has all the system advertised features. Use this for the secondary CPUs to perform the work around check. For that we rename verify_local_cpu_capabilities() => check_local_cpu_capabilities() which: If the system wide capabilities haven't been initialised (i.e, the CPU is activated at the boot), update the system wide detected work arounds. Otherwise (i.e a CPU hotplugged in later) verify that this CPU conforms to the system wide capabilities. 2) Boot CPU updates the work arounds from smp_prepare_boot_cpu() after we have initialised the system wide CPU feature values. Cc: Mark Rutland <mark.rutland@arm.com> Cc: Andre Przywara <andre.przywara@arm.com> Cc: Will Deacon <will.deacon@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2016-09-09 22:07:10 +09:00
void check_local_cpu_capabilities(void)
{
/*
* All secondary CPUs should conform to the early CPU features
* in use by the kernel based on boot CPU.
*/
check_early_cpu_features();
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
/*
arm64: Rearrange CPU errata workaround checks Right now we run through the work around checks on a CPU from __cpuinfo_store_cpu. There are some problems with that: 1) We initialise the system wide CPU feature registers only after the Boot CPU updates its cpuinfo. Now, if a work around depends on the variance of a CPU ID feature (e.g, check for Cache Line size mismatch), we have no way of performing it cleanly for the boot CPU. 2) It is out of place, invoked from __cpuinfo_store_cpu() in cpuinfo.c. It is not an obvious place for that. This patch rearranges the CPU specific capability(aka work around) checks. 1) At the moment we use verify_local_cpu_capabilities() to check if a new CPU has all the system advertised features. Use this for the secondary CPUs to perform the work around check. For that we rename verify_local_cpu_capabilities() => check_local_cpu_capabilities() which: If the system wide capabilities haven't been initialised (i.e, the CPU is activated at the boot), update the system wide detected work arounds. Otherwise (i.e a CPU hotplugged in later) verify that this CPU conforms to the system wide capabilities. 2) Boot CPU updates the work arounds from smp_prepare_boot_cpu() after we have initialised the system wide CPU feature values. Cc: Mark Rutland <mark.rutland@arm.com> Cc: Andre Przywara <andre.przywara@arm.com> Cc: Will Deacon <will.deacon@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2016-09-09 22:07:10 +09:00
* If we haven't finalised the system capabilities, this CPU gets
* a chance to update the errata work arounds.
* Otherwise, this CPU should verify that it has all the system
* advertised capabilities.
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
*/
if (!sys_caps_initialised)
arm64: Rearrange CPU errata workaround checks Right now we run through the work around checks on a CPU from __cpuinfo_store_cpu. There are some problems with that: 1) We initialise the system wide CPU feature registers only after the Boot CPU updates its cpuinfo. Now, if a work around depends on the variance of a CPU ID feature (e.g, check for Cache Line size mismatch), we have no way of performing it cleanly for the boot CPU. 2) It is out of place, invoked from __cpuinfo_store_cpu() in cpuinfo.c. It is not an obvious place for that. This patch rearranges the CPU specific capability(aka work around) checks. 1) At the moment we use verify_local_cpu_capabilities() to check if a new CPU has all the system advertised features. Use this for the secondary CPUs to perform the work around check. For that we rename verify_local_cpu_capabilities() => check_local_cpu_capabilities() which: If the system wide capabilities haven't been initialised (i.e, the CPU is activated at the boot), update the system wide detected work arounds. Otherwise (i.e a CPU hotplugged in later) verify that this CPU conforms to the system wide capabilities. 2) Boot CPU updates the work arounds from smp_prepare_boot_cpu() after we have initialised the system wide CPU feature values. Cc: Mark Rutland <mark.rutland@arm.com> Cc: Andre Przywara <andre.przywara@arm.com> Cc: Will Deacon <will.deacon@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2016-09-09 22:07:10 +09:00
update_cpu_errata_workarounds();
else
verify_local_cpu_capabilities();
}
static void __init setup_feature_capabilities(void)
{
update_cpu_capabilities(arm64_features, "detected feature:");
enable_cpu_capabilities(arm64_features);
}
arm64/cpufeature: don't use mutex in bringup path commit 63a1e1c95e60e798fa09ab3c536fb555aa5bbf2b upstream. Currently, cpus_set_cap() calls static_branch_enable_cpuslocked(), which must take the jump_label mutex. We call cpus_set_cap() in the secondary bringup path, from the idle thread where interrupts are disabled. Taking a mutex in this path "is a NONO" regardless of whether it's contended, and something we must avoid. We didn't spot this until recently, as ___might_sleep() won't warn for this case until all CPUs have been brought up. This patch avoids taking the mutex in the secondary bringup path. The poking of static keys is deferred until enable_cpu_capabilities(), which runs in a suitable context on the boot CPU. To account for the static keys being set later, cpus_have_const_cap() is updated to use another static key to check whether the const cap keys have been initialised, falling back to the caps bitmap until this is the case. This means that users of cpus_have_const_cap() gain should only gain a single additional NOP in the fast path once the const caps are initialised, but should always see the current cap value. The hyp code should never dereference the caps array, since the caps are initialized before we run the module initcall to initialise hyp. A check is added to the hyp init code to document this requirement. This change will sidestep a number of issues when the upcoming hotplug locking rework is merged. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Marc Zyniger <marc.zyngier@arm.com> Reviewed-by: Suzuki Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Cc: Christoffer Dall <christoffer.dall@linaro.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Sebastian Sewior <bigeasy@linutronix.de> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> [4.9: this avoids an IPI before GICv3 is up, preventing a boot time crash] Signed-off-by: Mark Rutland <mark.rutland@arm.com> [v4.9 backport] Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-05-16 23:18:05 +09:00
DEFINE_STATIC_KEY_FALSE(arm64_const_caps_ready);
EXPORT_SYMBOL(arm64_const_caps_ready);
static void __init mark_const_caps_ready(void)
{
static_branch_enable(&arm64_const_caps_ready);
}
extern const struct arm64_cpu_capabilities arm64_errata[];
bool this_cpu_has_cap(unsigned int cap)
{
return (__this_cpu_has_cap(arm64_features, cap) ||
__this_cpu_has_cap(arm64_errata, cap));
}
void __init setup_cpu_features(void)
{
u32 cwg;
int cls;
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
/* Set the CPU feature capabilies */
setup_feature_capabilities();
enable_errata_workarounds();
arm64/cpufeature: don't use mutex in bringup path commit 63a1e1c95e60e798fa09ab3c536fb555aa5bbf2b upstream. Currently, cpus_set_cap() calls static_branch_enable_cpuslocked(), which must take the jump_label mutex. We call cpus_set_cap() in the secondary bringup path, from the idle thread where interrupts are disabled. Taking a mutex in this path "is a NONO" regardless of whether it's contended, and something we must avoid. We didn't spot this until recently, as ___might_sleep() won't warn for this case until all CPUs have been brought up. This patch avoids taking the mutex in the secondary bringup path. The poking of static keys is deferred until enable_cpu_capabilities(), which runs in a suitable context on the boot CPU. To account for the static keys being set later, cpus_have_const_cap() is updated to use another static key to check whether the const cap keys have been initialised, falling back to the caps bitmap until this is the case. This means that users of cpus_have_const_cap() gain should only gain a single additional NOP in the fast path once the const caps are initialised, but should always see the current cap value. The hyp code should never dereference the caps array, since the caps are initialized before we run the module initcall to initialise hyp. A check is added to the hyp init code to document this requirement. This change will sidestep a number of issues when the upcoming hotplug locking rework is merged. Signed-off-by: Mark Rutland <mark.rutland@arm.com> Reviewed-by: Marc Zyniger <marc.zyngier@arm.com> Reviewed-by: Suzuki Poulose <suzuki.poulose@arm.com> Acked-by: Will Deacon <will.deacon@arm.com> Cc: Christoffer Dall <christoffer.dall@linaro.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Sebastian Sewior <bigeasy@linutronix.de> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> [4.9: this avoids an IPI before GICv3 is up, preventing a boot time crash] Signed-off-by: Mark Rutland <mark.rutland@arm.com> [v4.9 backport] Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-05-16 23:18:05 +09:00
mark_const_caps_ready();
setup_elf_hwcaps(arm64_elf_hwcaps);
if (system_supports_32bit_el0())
setup_elf_hwcaps(compat_elf_hwcaps);
arm64: Delay cpu feature capability checks At the moment we run through the arm64_features capability list for each CPU and set the capability if one of the CPU supports it. This could be problematic in a heterogeneous system with differing capabilities. Delay the CPU feature checks until all the enabled CPUs are up(i.e, smp_cpus_done(), so that we can make better decisions based on the overall system capability. Once we decide and advertise the capabilities the alternatives can be applied. From this state, we cannot roll back a feature to disabled based on the values from a new hotplugged CPU, due to the runtime patching and other reasons. So, for all new CPUs, we need to make sure that they have the established system capabilities. Failing which, we bring the CPU down, preventing it from turning online. Once the capabilities are decided, any new CPU booting up goes through verification to ensure that it has all the enabled capabilities and also invokes the respective enable() method on the CPU. The CPU errata checks are not delayed and is still executed per-CPU to detect the respective capabilities. If we ever come across a non-errata capability that needs to be checked on each-CPU, we could introduce them via a new capability table(or introduce a flag), which can be processed per CPU. The next patch will make the feature checks use the system wide safe value of a feature register. NOTE: The enable() methods associated with the capability is scheduled on all the CPUs (which is the only use case at the moment). If we need a different type of 'enable()' which only needs to be run once on any CPU, we should be able to handle that when needed. Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Tested-by: Dave Martin <Dave.Martin@arm.com> [catalin.marinas@arm.com: static variable and coding style fixes] Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2015-10-19 22:24:50 +09:00
/* Advertise that we have computed the system capabilities */
set_sys_caps_initialised();
/*
* Check for sane CTR_EL0.CWG value.
*/
cwg = cache_type_cwg();
cls = cache_line_size();
if (!cwg)
pr_warn("No Cache Writeback Granule information, assuming cache line size %d\n",
cls);
if (L1_CACHE_BYTES < cls)
pr_warn("L1_CACHE_BYTES smaller than the Cache Writeback Granule (%d < %d)\n",
L1_CACHE_BYTES, cls);
}
static bool __maybe_unused
cpufeature_pan_not_uao(const struct arm64_cpu_capabilities *entry, int __unused)
{
return (cpus_have_const_cap(ARM64_HAS_PAN) && !cpus_have_const_cap(ARM64_HAS_UAO));
}