In preparation for adding SME support update the bulk of the implementation for the vector length configuration prctl() calls to be independent of vector type.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/fpsimd.h | 6 ++--- arch/arm64/kernel/fpsimd.c | 47 ++++++++++++++++++--------------- arch/arm64/kernel/ptrace.c | 4 +-- arch/arm64/kvm/reset.c | 8 +++--- 4 files changed, 34 insertions(+), 31 deletions(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 1d0b5fa253a0..577bfcc252d6 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -51,8 +51,8 @@ extern void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *state, extern void fpsimd_flush_task_state(struct task_struct *target); extern void fpsimd_save_and_flush_cpu_state(void);

-/* Maximum VL that SVE VL-agnostic software can transparently support */ -#define SVE_VL_ARCH_MAX 0x100 +/* Maximum VL that SVE/SME VL-agnostic software can transparently support */ +#define VL_ARCH_MAX 0x100

/* Offset of FFR in the SVE register dump */ static inline size_t sve_ffr_offset(int vl) @@ -122,7 +122,7 @@ extern void fpsimd_sync_to_sve(struct task_struct *task); extern void sve_sync_to_fpsimd(struct task_struct *task); extern void sve_sync_from_fpsimd_zeropad(struct task_struct *task);

-extern int sve_set_vector_length(struct task_struct *task, +extern int vec_set_vector_length(struct task_struct *task, enum vec_type type, unsigned long vl, unsigned long flags);

extern int sve_set_current_vl(unsigned long arg); diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index a81b3a6615f0..763891f14496 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -632,7 +632,7 @@ void sve_sync_from_fpsimd_zeropad(struct task_struct *task) __fpsimd_to_sve(sst, fst, vq); }

-int sve_set_vector_length(struct task_struct *task, +int vec_set_vector_length(struct task_struct *task, enum vec_type type, unsigned long vl, unsigned long flags) { if (flags & ~(unsigned long)(PR_SVE_VL_INHERIT | @@ -643,33 +643,35 @@ int sve_set_vector_length(struct task_struct *task, return -EINVAL;

/* - * Clamp to the maximum vector length that VL-agnostic SVE code can - * work with. A flag may be assigned in the future to allow setting - * of larger vector lengths without confusing older software. + * Clamp to the maximum vector length that VL-agnostic code + * can work with. A flag may be assigned in the future to + * allow setting of larger vector lengths without confusing + * older software. */ - if (vl > SVE_VL_ARCH_MAX) - vl = SVE_VL_ARCH_MAX; + if (vl > VL_ARCH_MAX) + vl = VL_ARCH_MAX;

- vl = find_supported_vector_length(ARM64_VEC_SVE, vl); + vl = find_supported_vector_length(type, vl);

if (flags & (PR_SVE_VL_INHERIT | PR_SVE_SET_VL_ONEXEC)) - task_set_sve_vl_onexec(task, vl); + task_set_vl_onexec(task, type, vl); else /* Reset VL to system default on next exec: */ - task_set_sve_vl_onexec(task, 0); + task_set_vl_onexec(task, type, 0);

/* Only actually set the VL if not deferred: */ if (flags & PR_SVE_SET_VL_ONEXEC) goto out;

- if (vl == task_get_sve_vl(task)) + if (vl == task_get_vl(task, type)) goto out;

/* * To ensure the FPSIMD bits of the SVE vector registers are preserved, * write any live register state back to task_struct, and convert to a - * non-SVE thread. + * regular FPSIMD thread. Since the vector length can only be changed + * with a syscall we can't be in streaming mode while reconfiguring. */ if (task == current) { get_cpu_fpsimd_context(); @@ -690,10 +692,10 @@ int sve_set_vector_length(struct task_struct *task, */ sve_free(task);

- task_set_sve_vl(task, vl); + task_set_vl(task, type, vl);

out: - update_tsk_thread_flag(task, TIF_SVE_VL_INHERIT, + update_tsk_thread_flag(task, vec_vl_inherit_flag(type), flags & PR_SVE_VL_INHERIT);

return 0; @@ -701,20 +703,21 @@ int sve_set_vector_length(struct task_struct *task,

/* * Encode the current vector length and flags for return. - * This is only required for prctl(): ptrace has separate fields + * This is only required for prctl(): ptrace has separate fields. + * SVE and SME use the same bits for _ONEXEC and _INHERIT. * - * flags are as for sve_set_vector_length(). + * flags are as for vec_set_vector_length(). */ -static int sve_prctl_status(unsigned long flags) +static int vec_prctl_status(enum vec_type type, unsigned long flags) { int ret;

if (flags & PR_SVE_SET_VL_ONEXEC) - ret = task_get_sve_vl_onexec(current); + ret = task_get_vl_onexec(current, type); else - ret = task_get_sve_vl(current); + ret = task_get_vl(current, type);

- if (test_thread_flag(TIF_SVE_VL_INHERIT)) + if (test_thread_flag(vec_vl_inherit_flag(type))) ret |= PR_SVE_VL_INHERIT;

return ret; @@ -732,11 +735,11 @@ int sve_set_current_vl(unsigned long arg) if (!system_supports_sve() || is_compat_task()) return -EINVAL;

- ret = sve_set_vector_length(current, vl, flags); + ret = vec_set_vector_length(current, ARM64_VEC_SVE, vl, flags); if (ret) return ret;

- return sve_prctl_status(flags); + return vec_prctl_status(ARM64_VEC_SVE, flags); }

/* PR_SVE_GET_VL */ @@ -745,7 +748,7 @@ int sve_get_current_vl(void) if (!system_supports_sve() || is_compat_task()) return -EINVAL;

- return sve_prctl_status(0); + return vec_prctl_status(ARM64_VEC_SVE, 0); }

static void vec_probe_vqs(struct vl_info *info, diff --git a/arch/arm64/kernel/ptrace.c b/arch/arm64/kernel/ptrace.c index 88a9034fb9b5..716dde289446 100644 --- a/arch/arm64/kernel/ptrace.c +++ b/arch/arm64/kernel/ptrace.c @@ -812,9 +812,9 @@ static int sve_set(struct task_struct *target,

/* * Apart from SVE_PT_REGS_MASK, all SVE_PT_* flags are consumed by - * sve_set_vector_length(), which will also validate them for us: + * vec_set_vector_length(), which will also validate them for us: */ - ret = sve_set_vector_length(target, header.vl, + ret = vec_set_vector_length(target, ARM64_VEC_SVE, header.vl, ((unsigned long)header.flags & ~SVE_PT_REGS_MASK) << 16); if (ret) goto out; diff --git a/arch/arm64/kvm/reset.c b/arch/arm64/kvm/reset.c index 09cd30a9aafb..0f6741c80226 100644 --- a/arch/arm64/kvm/reset.c +++ b/arch/arm64/kvm/reset.c @@ -52,10 +52,10 @@ int kvm_arm_init_sve(void) * The get_sve_reg()/set_sve_reg() ioctl interface will need * to be extended with multiple register slice support in * order to support vector lengths greater than - * SVE_VL_ARCH_MAX: + * VL_ARCH_MAX: */ - if (WARN_ON(kvm_sve_max_vl > SVE_VL_ARCH_MAX)) - kvm_sve_max_vl = SVE_VL_ARCH_MAX; + if (WARN_ON(kvm_sve_max_vl > VL_ARCH_MAX)) + kvm_sve_max_vl = VL_ARCH_MAX;

/* * Don't even try to make use of vector lengths that @@ -103,7 +103,7 @@ static int kvm_vcpu_finalize_sve(struct kvm_vcpu *vcpu) * set_sve_vls(). Double-check here just to be sure: */ if (WARN_ON(!sve_vl_valid(vl) || vl > sve_max_virtualisable_vl() || - vl > SVE_VL_ARCH_MAX)) + vl > VL_ARCH_MAX)) return -EIO;

buf = kzalloc(SVE_SIG_REGS_SIZE(sve_vq_from_vl(vl)), GFP_KERNEL);

Currently we have the facility to specify custom code to trigger a signal but none of the tests use it and for some reason the framework requires us to also specify a signal to send as a trigger in order to make use of a custom trigger. This doesn't seem to make much sense, instead allow the use of a custom trigger function without specifying a signal to inject.

Signed-off-by: Mark Brown broonie@kernel.org --- tools/testing/selftests/arm64/signal/test_signals_utils.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/testing/selftests/arm64/signal/test_signals_utils.c b/tools/testing/selftests/arm64/signal/test_signals_utils.c index 22722abc9dfa..8bb12be87a51 100644 --- a/tools/testing/selftests/arm64/signal/test_signals_utils.c +++ b/tools/testing/selftests/arm64/signal/test_signals_utils.c @@ -310,7 +310,7 @@ int test_setup(struct tdescr *td)

int test_run(struct tdescr *td) { - if (td->sig_trig) { + if (td->sig_trig || td->trigger) { if (td->trigger) return td->trigger(td); else

Provide ABI documentation for SME similar to that for SVE. Due to the very large overlap around streaming SVE mode in both implementation and interfaces documentation for streaming mode SVE is added to the SVE document rather than the SME one.

Signed-off-by: Mark Brown broonie@kernel.org --- Documentation/arm64/index.rst | 1 + Documentation/arm64/sme.rst | 428 ++++++++++++++++++++++++++++++++++ Documentation/arm64/sve.rst | 69 +++++- 3 files changed, 488 insertions(+), 10 deletions(-) create mode 100644 Documentation/arm64/sme.rst

diff --git a/Documentation/arm64/index.rst b/Documentation/arm64/index.rst index 4f840bac083e..ae21f8118830 100644 --- a/Documentation/arm64/index.rst +++ b/Documentation/arm64/index.rst @@ -21,6 +21,7 @@ ARM64 Architecture perf pointer-authentication silicon-errata + sme sve tagged-address-abi tagged-pointers diff --git a/Documentation/arm64/sme.rst b/Documentation/arm64/sme.rst new file mode 100644 index 000000000000..5a4294aca6dc --- /dev/null +++ b/Documentation/arm64/sme.rst @@ -0,0 +1,428 @@ +=================================================== +Scalable Matrix Extension support for AArch64 Linux +=================================================== + +This document outlines briefly the interface provided to userspace by Linux in +order to support use of the ARM Scalable Matrix Extension (SME). + +This is an outline of the most important features and issues only and not +intended to be exhaustive. It should be read in conjunction with the SVE +documentation in sve.rst which provides details on the Streaming SVE mode +included in SME. + +This document does not aim to describe the SME architecture or programmer's +model. To aid understanding, a minimal description of relevant programmer's +model features for SME is included in Appendix A. + + +1. General +----------- + +* PSTATE.SM and PSTATE.ZA, the streaming mode vector length, the ZA + register state and TPIDR2_EL0 are tracked per thread. + +* The presence of SVE is reported to userspace via HWCAP2_SME in the aux vector + AT_HWCAP2 entry. Presence of this flag implies the presence of the SME + instructions and registers, and the Linux-specific system interfaces + described in this document. SME is reported in /proc/cpuinfo as "sme". + +* Support for the execution of SME instructions in userspace can also be + detected by reading the CPU ID register ID_AA64PFR1_EL1 using an MRS + instruction, and checking that the value of the SME field is nonzero. [3] + + It does not guarantee the presence of the system interfaces described in the + following sections: software that needs to verify that those interfaces are + present must check for HWCAP2_SME instead. + +* There are a number of optional SME features, presence of these is reported + through AT_HWCAP2 through: + + HWCAP2_SME_I16I64 + HWCAP2_SME_F64F64 + HWCAP2_SME_I8I32 + HWCAP2_SME_F16F32 + HWCAP2_SME_B16F32 + HWCAP2_SME_F32F32 + + This list may be extended over time as the SME architecture evolves. + + These extensions are also reported via the CPU ID register ID_AA64SMFR0_EL1, + which userspace can read using an MRS instruction. See elf_hwcaps.txt and + cpu-feature-registers.txt for details. + +* Debuggers should restrict themselves to interacting with the target via the + NT_ARM_SVE, NT_ARM_SSVE and NT_ARM_ZA regsets. The recommended way + of detecting support for these regsets is to connect to a target process + first and then attempt a + + ptrace(PTRACE_GETREGSET, pid, NT_ARM_<regset>, &iov). + +* Whenever ZA register values are exchanged in memory between userspace and + the kernel, the register value is encoded in memory as a series of horizontal + vectors from 0 to VL/8-1 stored in the same endianness invariant format as is + used for SVE vectors. + +* On thread creation TPIDR2_EL0 is set to 0. + +2. Vector lengths +------------------ + +SME defines a second vector length similar to the SVE vector length which is +controls the size of the streaming mode SVE vectors and the ZA matrix array. +The ZA matrix is square with each side having as many bytes as a SVE vector. + + +3. Sharing of streaming and non-streaming mode SVE state +--------------------------------------------------------- + +It is implementation defined which if any parts of the SVE state are shared +between streaming and non-streaming modes. When switching between modes +via software interfaces such as ptrace if no register content is provided as +part of switching no state will be assumed to be shared and everything will +be zeroed. + + +4. System call behaviour +------------------------- + +* On syscall PSTATE.ZA is preserved, if PSTATE.ZA==1 then the contents of the + ZA matrix are preserved. + +* On syscall PSTATE.SM will be cleared and the SVE registers will be handled + as normal. + +* Neither the SVE registers nor ZA are used to pass arguments to or receive + results from any syscall. + +* On creation fork() or clone() the newly created process will have PSTATE.SM + and PSTATE.ZA cleared. + +* All other SME state of a thread, including the currently configured vector + length, the state of the PR_SME_VL_INHERIT flag, and the deferred vector + length (if any), is preserved across all syscalls, subject to the specific + exceptions for execve() described in section 6. + + +5. Signal handling +------------------- + +* A new signal frame record za_context encodes the ZA register contents on + signal delivery. [1] + +* The signal frame record for ZA always contains basic metadata, in particular + the thread's vector length (in za_context.vl). + +* The ZA matrix may or may not be included in the record, depending on + the value of PSTATE.ZA. The registers are present if and only if: + za_context.head.size >= ZA_SIG_CONTEXT_SIZE(sve_vq_from_vl(za_context.vl)) + in which case PSTATE.ZA == 1. + +* If matrix data is present, the remainder of the record has a vl-dependent + size and layout. Macros ZA_SIG_* are defined [1] to facilitate access to + them. + +* The matrix is stored as a series of horizontal vectors in the same format as + is used for SVE vectors. + +* If the ZA context is too big to fit in sigcontext.__reserved[], then extra + space is allocated on the stack, an extra_context record is written in + __reserved[] referencing this space. za_context is then written in the + extra space. Refer to [1] for further details about this mechanism. + + +5. Signal return +----------------- + +When returning from a signal handler: + +* If there is no za_context record in the signal frame, or if the record is + present but contains no register data as desribed in the previous section, + then ZA is disabled. + +* If za_context is present in the signal frame and contains matrix data then + PSTATE.ZA is set to 1 and ZA is populated with the specified data. + +* The vector length cannot be changed via signal return. If za_context.vl in + the signal frame does not match the current vector length, the signal return + attempt is treated as illegal, resulting in a forced SIGSEGV. + + +6. prctl extensions +-------------------- + +Some new prctl() calls are added to allow programs to manage the SME vector +length: + +prctl(PR_SME_SET_VL, unsigned long arg) + + Sets the vector length of the calling thread and related flags, where + arg == vl | flags. Other threads of the calling process are unaffected. + + vl is the desired vector length, where sve_vl_valid(vl) must be true. + + flags: + + PR_SME_VL_INHERIT + + Inherit the current vector length across execve(). Otherwise, the + vector length is reset to the system default at execve(). (See + Section 9.) + + PR_SME_SET_VL_ONEXEC + + Defer the requested vector length change until the next execve() + performed by this thread. + + The effect is equivalent to implicit exceution of the following + call immediately after the next execve() (if any) by the thread: + + prctl(PR_SME_SET_VL, arg & ~PR_SME_SET_VL_ONEXEC) + + This allows launching of a new program with a different vector + length, while avoiding runtime side effects in the caller. + + Without PR_SME_SET_VL_ONEXEC, the requested change takes effect + immediately. + + + Return value: a nonnegative on success, or a negative value on error: + EINVAL: SME not supported, invalid vector length requested, or + invalid flags. + + + On success: + + * Either the calling thread's vector length or the deferred vector length + to be applied at the next execve() by the thread (dependent on whether + PR_SME_SET_VL_ONEXEC is present in arg), is set to the largest value + supported by the system that is less than or equal to vl. If vl == + SVE_VL_MAX, the value set will be the largest value supported by the + system. + + * Any previously outstanding deferred vector length change in the calling + thread is cancelled. + + * The returned value describes the resulting configuration, encoded as for + PR_SME_GET_VL. The vector length reported in this value is the new + current vector length for this thread if PR_SME_SET_VL_ONEXEC was not + present in arg; otherwise, the reported vector length is the deferred + vector length that will be applied at the next execve() by the calling + thread. + + * Changing the vector length causes all of ZA, P0..P15, FFR and all bits of + Z0..Z31 except for Z0 bits [127:0] .. Z31 bits [127:0] to become + unspecified, including both streaming and non-streaming SVE state. + Calling PR_SME_SET_VL with vl equal to the thread's current vector + length, or calling PR_SME_SET_VL with the PR_SVE_SET_VL_ONEXEC flag, + does not constitute a change to the vector length for this purpose. + + * Changing the vector length causes PSTATE.ZA and PSTATE.SM to be cleared. + Calling PR_SME_SET_VL with vl equal to the thread's current vector + length, or calling PR_SME_SET_VL with the PR_SVE_SET_VL_ONEXEC flag, + does not constitute a change to the vector length for this purpose. + + +prctl(PR_SME_GET_VL) + + Gets the vector length of the calling thread. + + The following flag may be OR-ed into the result: + + PR_SME_VL_INHERIT + + Vector length will be inherited across execve(). + + There is no way to determine whether there is an outstanding deferred + vector length change (which would only normally be the case between a + fork() or vfork() and the corresponding execve() in typical use). + + To extract the vector length from the result, and it with + PR_SME_VL_LEN_MASK. + + Return value: a nonnegative value on success, or a negative value on error: + EINVAL: SME not supported. + + +7. ptrace extensions +--------------------- + +* A new regset NT_ARM_SSVE is defined for access to streaming mode SVE + state via PTRACE_GETREGSET and PTRACE_SETREGSET, this is documented in + sve.rst. + +* A new regset NT_ARM_ZA is defined for ZA state for access to ZA state via + PTRACE_GETREGSET and PTRACE_SETREGSET. + + Refer to [2] for definitions. + +The regset data starts with struct user_za_header, containing: + + size + + Size of the complete regset, in bytes. + This depends on vl and possibly on other things in the future. + + If a call to PTRACE_GETREGSET requests less data than the value of + size, the caller can allocate a larger buffer and retry in order to + read the complete regset. + + max_size + + Maximum size in bytes that the regset can grow to for the target + thread. The regset won't grow bigger than this even if the target + thread changes its vector length etc. + + vl + + Target thread's current streaming vector length, in bytes. + + max_vl + + Maximum possible streaming vector length for the target thread. + + flags + + Zero or more of the following flags, which have the same + meaning and behaviour as the corresponding PR_SET_VL_* flags: + + SME_PT_VL_INHERIT + + SME_PT_VL_ONEXEC (SETREGSET only). + +* The effects of changing the vector length and/or flags are equivalent to + those documented for PR_SME_SET_VL. + + The caller must make a further GETREGSET call if it needs to know what VL is + actually set by SETREGSET, unless is it known in advance that the requested + VL is supported. + +* The size and layout of the payload depends on the header fields. The + SME_PT_ZA_*() macros are provided to facilitate access to the data. + +* In either case, for SETREGSET it is permissible to omit the payload, in which + case the vector length and flags are changed and PSTATE.ZA is set to 0 + (along with any consequences of those changes). If a payload is provided + then PSTATE.ZA will be set to 1. + +* For SETREGSET, if the requested VL is not supported, the effect will be the + same as if the payload were omitted, except that an EIO error is reported. + No attempt is made to translate the payload data to the correct layout + for the vector length actually set. It is up to the caller to translate the + payload layout for the actual VL and retry. + +* The effect of writing a partial, incomplete payload is unspecified. + + +8. ELF coredump extensions +--------------------------- + +* NT_ARM_SSVE notes will be added to each coredump for + each thread of the dumped process. The contents will be equivalent to the + data that would have been read if a PTRACE_GETREGSET of the corresponding + type were executed for each thread when the coredump was generated. + +* A NT_ARM_ZA note will be added to each coredump for each thread of the + dumped process. The contents will be equivalent to the data that would have + been read if a PTRACE_GETREGSET of NT_ARM_ZA were executed for each thread + when the coredump was generated. + + +9. System runtime configuration +-------------------------------- + +* To mitigate the ABI impact of expansion of the signal frame, a policy + mechanism is provided for administrators, distro maintainers and developers + to set the default vector length for userspace processes: + +/proc/sys/abi/sme_default_vector_length + + Writing the text representation of an integer to this file sets the system + default vector length to the specified value, unless the value is greater + than the maximum vector length supported by the system in which case the + default vector length is set to that maximum. + + The result can be determined by reopening the file and reading its + contents. + + At boot, the default vector length is initially set to 32 or the maximum + supported vector length, whichever is smaller and supported. This + determines the initial vector length of the init process (PID 1). + + Reading this file returns the current system default vector length. + +* At every execve() call, the new vector length of the new process is set to + the system default vector length, unless + + * PR_SME_VL_INHERIT (or equivalently SME_PT_VL_INHERIT) is set for the + calling thread, or + + * a deferred vector length change is pending, established via the + PR_SME_SET_VL_ONEXEC flag (or SME_PT_VL_ONEXEC). + +* Modifying the system default vector length does not affect the vector length + of any existing process or thread that does not make an execve() call. + + +Appendix A. SME programmer's model (informative) +================================================= + +This section provides a minimal description of the additions made by SVE to the +ARMv8-A programmer's model that are relevant to this document. + +Note: This section is for information only and not intended to be complete or +to replace any architectural specification. + +A.1. Registers +--------------- + +In A64 state, SME adds the following: + +* A new mode, streaming mode, in which a subset of the normal FPSIMD and SVE + features are available. When supported EL0 software may enter and leave + streaming mode at any time. + + For best system performance it is strongly encourage for software to enable + streaming mode only when it is actively being used. + +* A new vector length controlling the size of ZA and the Z registers when in + streaming mode, separately to the vector length used for SVE when not in + streaming mode. There is no requirement that either the currently selected + vector length or the set of vector lengths supported for the two modes in + a given system have any relationship. The streaming mode vector length + is referred to as SVL. + +* A new ZA matrix register. This is a square matrix of SVLxSVL bits. Most + operations on ZA require that streaming mode be enabled but ZA can be + enabled without streaming mode in order to load, save and retain data. + + For best system performance it is strongly encourage for software to enable + ZA only when it is actively being used. + +* Two new 1 bit fields in PSTATE which may be controlled via the SMSTART and + SMSTOP instructions or by access to the SVCR system register: + + * PSTATE.ZA, if this is 1 then the ZA matrix is accessible and has valid + data while if it is 0 then ZA can not be accessed. When PSTATE.ZA is + changed from 0 to 1 all bits in ZA are cleared. + + * PSTATE.SM, if this is 1 then the PE is in streaming mode. When the value + of PSTATE.SM is changed then it is implementationd defined if the subset + of the floating point register bits valid in both modes may be retained. + Any other bits will be cleared. + + +References +========== + +[1] arch/arm64/include/uapi/asm/sigcontext.h + AArch64 Linux signal ABI definitions + +[2] arch/arm64/include/uapi/asm/ptrace.h + AArch64 Linux ptrace ABI definitions + +[3] Documentation/arm64/cpu-feature-registers.rst + +[4] ARM IHI0055C + http://infocenter.arm.com/help/topic/com.arm.doc.ihi0055c/IHI0055C_beta_aapc... + http://infocenter.arm.com/help/topic/com.arm.doc.subset.swdev.abi/index.html + Procedure Call Standard for the ARM 64-bit Architecture (AArch64) diff --git a/Documentation/arm64/sve.rst b/Documentation/arm64/sve.rst index 03137154299e..18b2bb1e8843 100644 --- a/Documentation/arm64/sve.rst +++ b/Documentation/arm64/sve.rst @@ -7,7 +7,9 @@ Author: Dave Martin Dave.Martin@arm.com Date: 4 August 2017

This document outlines briefly the interface provided to userspace by Linux in -order to support use of the ARM Scalable Vector Extension (SVE). +order to support use of the ARM Scalable Vector Extension (SVE), including +interactions with Streaming SVE mode added by the Scalable Matrix Extension +(SME).

This is an outline of the most important features and issues only and not intended to be exhaustive. @@ -23,6 +25,9 @@ model features for SVE is included in Appendix A. * SVE registers Z0..Z31, P0..P15 and FFR and the current vector length VL, are tracked per-thread.

+* In streaming mode FFR is not accessible, when these interfaces are used to + access streaming mode FFR is read and written as zero. + * The presence of SVE is reported to userspace via HWCAP_SVE in the aux vector AT_HWCAP entry. Presence of this flag implies the presence of the SVE instructions and registers, and the Linux-specific system interfaces @@ -53,10 +58,19 @@ model features for SVE is included in Appendix A. which userspace can read using an MRS instruction. See elf_hwcaps.txt and cpu-feature-registers.txt for details.

+* On hardware that supports the SME extensions, HWCAP2_SME will also be + reported in the AT_HWCAP2 aux vector entry. Among other things SME adds + streaming mode which provides a subset of the SVE feature set using a + separate SME vector length and the same Z/V registers. See sme.rst + for more details. + * Debuggers should restrict themselves to interacting with the target via the NT_ARM_SVE regset. The recommended way of detecting support for this regset is to connect to a target process first and then attempt a - ptrace(PTRACE_GETREGSET, pid, NT_ARM_SVE, &iov). + ptrace(PTRACE_GETREGSET, pid, NT_ARM_SVE, &iov). Note that when SME is + present and streaming SVE mode is in use the FPSIMD subset of registers + will be read via NT_ARM_SVE and NT_ARM_SVE writes will exit streaming mode + in the target.

* Whenever SVE scalable register values (Zn, Pn, FFR) are exchanged in memory between userspace and the kernel, the register value is encoded in memory in @@ -126,6 +140,11 @@ the SVE instruction set architecture. are only present in fpsimd_context. For convenience, the content of V0..V31 is duplicated between sve_context and fpsimd_context.

+* The record contains a flag field which includes a flag SVE_SIG_FLAG_SM which + if set indicates that the thread is in streaming mode and the vector length + and register data (if present) describe the streaming SVE data and vector + length. + * The signal frame record for SVE always contains basic metadata, in particular the thread's vector length (in sve_context.vl).

@@ -170,6 +189,11 @@ When returning from a signal handler: the signal frame does not match the current vector length, the signal return attempt is treated as illegal, resulting in a forced SIGSEGV.

+* It is permitted to enter or leave streaming mode by setting or clearing + the SVE_SIG_FLAG_SM flag but applications should take care to ensure that + when doing so sve_context.vl and any register data are appropriate for the + vector length in the new mode. +

6. prctl extensions -------------------- @@ -265,8 +289,14 @@ prctl(PR_SVE_GET_VL) 7. ptrace extensions ---------------------

-* A new regset NT_ARM_SVE is defined for use with PTRACE_GETREGSET and - PTRACE_SETREGSET. +* New regsets NT_ARM_SVE and NT_ARM_SSVE are defined for use with + PTRACE_GETREGSET and PTRACE_SETREGSET. NT_ARM_SSVE describes the + streaming mode SVE registers and NT_ARM_SVE describes the + non-streaming mode SVE registers. + + In this description a register set is referred to as being "live" when + the target is in the appropriate streaming or non-streaming mode and is + using data beyond the subset shared with the FPSIMD Vn registers.

Refer to [2] for definitions.

@@ -297,7 +327,7 @@ The regset data starts with struct user_sve_header, containing:

flags

- either + at most one of

SVE_PT_REGS_FPSIMD

@@ -331,6 +361,10 @@ The regset data starts with struct user_sve_header, containing:

SVE_PT_VL_ONEXEC (SETREGSET only).

+ If neither FPSIMD nor SVE flags are provided then no register + payload is available, this is only possible when SME is implemented. + + * The effects of changing the vector length and/or flags are equivalent to those documented for PR_SVE_SET_VL.

@@ -346,6 +380,13 @@ The regset data starts with struct user_sve_header, containing: case only the vector length and flags are changed (along with any consequences of those changes).

+* In systems supporting SME when in streaming mode a GETREGSET for + NT_REG_SVE will return only the user_sve_header with no register data, + similarly a GETREGSET for NT_REG_SSVE will not return any register data + when not in streaming mode. + +* A GETREGSET for NT_ARM_SSVE will never return SVE_PT_REGS_FPSIMD. + * For SETREGSET, if an SVE_PT_REGS_SVE payload is present and the requested VL is not supported, the effect will be the same as if the payload were omitted, except that an EIO error is reported. No @@ -355,17 +396,25 @@ The regset data starts with struct user_sve_header, containing: unspecified. It is up to the caller to translate the payload layout for the actual VL and retry.

+* Where SME is implemented it is not possible to GETREGSET the register + state for normal SVE when in streaming mode, nor the streaming mode + register state when in normal mode, regardless of the implementation defined + behaviour of the hardware for sharing data between the two modes. + +* Any SETREGSET of NT_ARM_SVE will exit streaming mode if the target was in + streaming mode and any SETREGSET of NT_ARM_SSVE will enter streaming mode + if the target was not in streaming mode. + * The effect of writing a partial, incomplete payload is unspecified.

8. ELF coredump extensions ---------------------------

-* A NT_ARM_SVE note will be added to each coredump for each thread of the - dumped process. The contents will be equivalent to the data that would have - been read if a PTRACE_GETREGSET of NT_ARM_SVE were executed for each thread - when the coredump was generated. - +* NT_ARM_SVE and NT_ARM_SSVE notes will be added to each coredump for + each thread of the dumped process. The contents will be equivalent to the + data that would have been read if a PTRACE_GETREGSET of the corresponding + type were executed for each thread when the coredump was generated.

9. System runtime configuration --------------------------------

The arm64 Scalable Matrix Extension (SME) adds some new system registers, fields in existing system registers and exception syndromes. This patch adds definitions for these for use in future patches implementing support for this extension.

Since SME will be the first user of FEAT_HCX in the kernel also include the definitions for enumerating it and the HCRX system register it adds.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/esr.h | 12 +++++++- arch/arm64/include/asm/kvm_arm.h | 1 + arch/arm64/include/asm/sysreg.h | 53 ++++++++++++++++++++++++++++++++ arch/arm64/kernel/traps.c | 1 + 4 files changed, 66 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/esr.h b/arch/arm64/include/asm/esr.h index 29f97eb3dad4..31418bc7644a 100644 --- a/arch/arm64/include/asm/esr.h +++ b/arch/arm64/include/asm/esr.h @@ -37,7 +37,8 @@ #define ESR_ELx_EC_ERET (0x1a) /* EL2 only */ /* Unallocated EC: 0x1B */ #define ESR_ELx_EC_FPAC (0x1C) /* EL1 and above */ -/* Unallocated EC: 0x1D - 0x1E */ +#define ESR_ELx_EC_SME (0x1D) +/* Unallocated EC: 0x1E */ #define ESR_ELx_EC_IMP_DEF (0x1f) /* EL3 only */ #define ESR_ELx_EC_IABT_LOW (0x20) #define ESR_ELx_EC_IABT_CUR (0x21) @@ -320,6 +321,15 @@ #define ESR_ELx_CP15_32_ISS_SYS_CNTFRQ (ESR_ELx_CP15_32_ISS_SYS_VAL(0, 0, 14, 0) |\ ESR_ELx_CP15_32_ISS_DIR_READ)

+/* + * ISS values for SME traps + */ + +#define ESR_ELx_SME_ISS_SME_DISABLED 0 +#define ESR_ELx_SME_ISS_ILL 1 +#define ESR_ELx_SME_ISS_SM_DISABLED 2 +#define ESR_ELx_SME_ISS_ZA_DISABLED 3 + #ifndef __ASSEMBLY__ #include <asm/types.h>

diff --git a/arch/arm64/include/asm/kvm_arm.h b/arch/arm64/include/asm/kvm_arm.h index 327120c0089f..438de250ab39 100644 --- a/arch/arm64/include/asm/kvm_arm.h +++ b/arch/arm64/include/asm/kvm_arm.h @@ -279,6 +279,7 @@ #define CPTR_EL2_TCPAC (1 << 31) #define CPTR_EL2_TAM (1 << 30) #define CPTR_EL2_TTA (1 << 20) +#define CPTR_EL2_TSM (1 << 12) #define CPTR_EL2_TFP (1 << CPTR_EL2_TFP_SHIFT) #define CPTR_EL2_TZ (1 << 8) #define CPTR_NVHE_EL2_RES1 0x000032ff /* known RES1 bits in CPTR_EL2 (nVHE) */ diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h index b268082d67ed..86f442a3769e 100644 --- a/arch/arm64/include/asm/sysreg.h +++ b/arch/arm64/include/asm/sysreg.h @@ -171,6 +171,7 @@ #define SYS_ID_AA64PFR0_EL1 sys_reg(3, 0, 0, 4, 0) #define SYS_ID_AA64PFR1_EL1 sys_reg(3, 0, 0, 4, 1) #define SYS_ID_AA64ZFR0_EL1 sys_reg(3, 0, 0, 4, 4) +#define SYS_ID_AA64SMFR0_EL1 sys_reg(3, 0, 0, 4, 5)

#define SYS_ID_AA64DFR0_EL1 sys_reg(3, 0, 0, 5, 0) #define SYS_ID_AA64DFR1_EL1 sys_reg(3, 0, 0, 5, 1) @@ -193,6 +194,8 @@

#define SYS_ZCR_EL1 sys_reg(3, 0, 1, 2, 0) #define SYS_TRFCR_EL1 sys_reg(3, 0, 1, 2, 1) +#define SYS_SMPRI_EL1 sys_reg(3, 0, 1, 2, 4) +#define SYS_SMCR_EL1 sys_reg(3, 0, 1, 2, 6)

#define SYS_TTBR0_EL1 sys_reg(3, 0, 2, 0, 0) #define SYS_TTBR1_EL1 sys_reg(3, 0, 2, 0, 1) @@ -385,6 +388,8 @@ #define TRBIDR_ALIGN_MASK GENMASK(3, 0) #define TRBIDR_ALIGN_SHIFT 0

+#define SMPRI_EL1_PRIORITY_MASK 0xf + #define SYS_PMINTENSET_EL1 sys_reg(3, 0, 9, 14, 1) #define SYS_PMINTENCLR_EL1 sys_reg(3, 0, 9, 14, 2)

@@ -440,8 +445,13 @@ #define SYS_CCSIDR_EL1 sys_reg(3, 1, 0, 0, 0) #define SYS_CLIDR_EL1 sys_reg(3, 1, 0, 0, 1) #define SYS_GMID_EL1 sys_reg(3, 1, 0, 0, 4) +#define SYS_SMIDR_EL1 sys_reg(3, 1, 0, 0, 6) #define SYS_AIDR_EL1 sys_reg(3, 1, 0, 0, 7)

+#define SYS_SMIDR_EL1_IMPLEMENTER_SHIFT 24 +#define SYS_SMIDR_EL1_SMPS_SHIFT 15 +#define SYS_SMIDR_EL1_AFFINITY_SHIFT 0 + #define SYS_CSSELR_EL1 sys_reg(3, 2, 0, 0, 0)

#define SYS_CTR_EL0 sys_reg(3, 3, 0, 0, 1) @@ -450,6 +460,10 @@ #define SYS_RNDR_EL0 sys_reg(3, 3, 2, 4, 0) #define SYS_RNDRRS_EL0 sys_reg(3, 3, 2, 4, 1)

+#define SYS_SVCR_EL0 sys_reg(3, 3, 4, 2, 2) +#define SYS_SVCR_EL0_ZA_MASK 2 +#define SYS_SVCR_EL0_SM_MASK 1 + #define SYS_PMCR_EL0 sys_reg(3, 3, 9, 12, 0) #define SYS_PMCNTENSET_EL0 sys_reg(3, 3, 9, 12, 1) #define SYS_PMCNTENCLR_EL0 sys_reg(3, 3, 9, 12, 2) @@ -466,6 +480,7 @@

#define SYS_TPIDR_EL0 sys_reg(3, 3, 13, 0, 2) #define SYS_TPIDRRO_EL0 sys_reg(3, 3, 13, 0, 3) +#define SYS_TPIDR2_EL0 sys_reg(3, 3, 13, 0, 5)

#define SYS_SCXTNUM_EL0 sys_reg(3, 3, 13, 0, 7)

@@ -532,6 +547,9 @@ #define SYS_HFGITR_EL2 sys_reg(3, 4, 1, 1, 6) #define SYS_ZCR_EL2 sys_reg(3, 4, 1, 2, 0) #define SYS_TRFCR_EL2 sys_reg(3, 4, 1, 2, 1) +#define SYS_HCRX_EL2 sys_reg(3, 4, 1, 2, 2) +#define SYS_SMPRIMAP_EL2 sys_reg(3, 4, 1, 2, 5) +#define SYS_SMCR_EL2 sys_reg(3, 4, 1, 2, 6) #define SYS_DACR32_EL2 sys_reg(3, 4, 3, 0, 0) #define SYS_HDFGRTR_EL2 sys_reg(3, 4, 3, 1, 4) #define SYS_HDFGWTR_EL2 sys_reg(3, 4, 3, 1, 5) @@ -591,6 +609,7 @@ #define SYS_SCTLR_EL12 sys_reg(3, 5, 1, 0, 0) #define SYS_CPACR_EL12 sys_reg(3, 5, 1, 0, 2) #define SYS_ZCR_EL12 sys_reg(3, 5, 1, 2, 0) +#define SYS_SMCR_EL12 sys_reg(3, 5, 1, 2, 3) #define SYS_TTBR0_EL12 sys_reg(3, 5, 2, 0, 0) #define SYS_TTBR1_EL12 sys_reg(3, 5, 2, 0, 1) #define SYS_TCR_EL12 sys_reg(3, 5, 2, 0, 2) @@ -614,6 +633,7 @@ #define SYS_CNTV_CVAL_EL02 sys_reg(3, 5, 14, 3, 2)

/* Common SCTLR_ELx flags. */ +#define SCTLR_ELx_ENTP2 (BIT(60)) #define SCTLR_ELx_DSSBS (BIT(44)) #define SCTLR_ELx_ATA (BIT(43))

@@ -793,6 +813,7 @@ #define ID_AA64PFR0_ELx_32BIT_64BIT 0x2

/* id_aa64pfr1 */ +#define ID_AA64PFR1_SME_SHIFT 24 #define ID_AA64PFR1_MPAMFRAC_SHIFT 16 #define ID_AA64PFR1_RASFRAC_SHIFT 12 #define ID_AA64PFR1_MTE_SHIFT 8 @@ -803,6 +824,7 @@ #define ID_AA64PFR1_SSBS_PSTATE_ONLY 1 #define ID_AA64PFR1_SSBS_PSTATE_INSNS 2 #define ID_AA64PFR1_BT_BTI 0x1 +#define ID_AA64PFR1_SME 1

#define ID_AA64PFR1_MTE_NI 0x0 #define ID_AA64PFR1_MTE_EL0 0x1 @@ -830,6 +852,21 @@ #define ID_AA64ZFR0_AES_PMULL 0x2 #define ID_AA64ZFR0_SVEVER_SVE2 0x1

+/* id_aa64smfr0 */ +#define ID_AA64SMFR0_I16I64_SHIFT 52 +#define ID_AA64SMFR0_F64F64_SHIFT 48 +#define ID_AA64SMFR0_I8I32_SHIFT 36 +#define ID_AA64SMFR0_F16F32_SHIFT 35 +#define ID_AA64SMFR0_B16F32_SHIFT 34 +#define ID_AA64SMFR0_F32F32_SHIFT 32 + +#define ID_AA64SMFR0_I16I64 0x4 +#define ID_AA64SMFR0_F64F64 0x1 +#define ID_AA64SMFR0_I8I32 0x4 +#define ID_AA64SMFR0_F16F32 0x1 +#define ID_AA64SMFR0_B16F32 0x1 +#define ID_AA64SMFR0_F32F32 0x1 + /* id_aa64mmfr0 */ #define ID_AA64MMFR0_ECV_SHIFT 60 #define ID_AA64MMFR0_FGT_SHIFT 56 @@ -881,6 +918,7 @@ #endif

/* id_aa64mmfr1 */ +#define ID_AA64MMFR1_HCX_SHIFT 40 #define ID_AA64MMFR1_ETS_SHIFT 36 #define ID_AA64MMFR1_TWED_SHIFT 32 #define ID_AA64MMFR1_XNX_SHIFT 28 @@ -1072,6 +1110,19 @@ #define ZCR_ELx_LEN_SIZE 9 #define ZCR_ELx_LEN_MASK 0x1ff

+/* + * The SMCR_ELx_LEN_* definitions intentionally include bits [8:4] which + * are reserved by the SME architecture for future expansion of the LEN + * field, with compatible semantics. + */ +#define SMCR_ELx_LEN_SHIFT 0 +#define SMCR_ELx_LEN_SIZE 9 +#define SMCR_ELx_LEN_MASK 0x1ff + +#define CPACR_EL1_SMEN_EL1EN (BIT(24)) /* enable EL1 access */ +#define CPACR_EL1_SMEN_EL0EN (BIT(25)) /* enable EL0 access, if EL1EN set */ +#define CPACR_EL1_SMEN (CPACR_EL1_SMEN_EL1EN | CPACR_EL1_SMEN_EL0EN) + #define CPACR_EL1_ZEN_EL1EN (BIT(16)) /* enable EL1 access */ #define CPACR_EL1_ZEN_EL0EN (BIT(17)) /* enable EL0 access, if EL1EN set */ #define CPACR_EL1_ZEN (CPACR_EL1_ZEN_EL1EN | CPACR_EL1_ZEN_EL0EN) @@ -1125,6 +1176,8 @@ #define TRFCR_ELx_ExTRE BIT(1) #define TRFCR_ELx_E0TRE BIT(0)

+/* HCRX_EL2 definitions */ +#define HCRX_EL2_SMPME_MASK (1 << 5)

/* GIC Hypervisor interface registers */ /* ICH_MISR_EL2 bit definitions */ diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c index b03e383d944a..cb8934e75654 100644 --- a/arch/arm64/kernel/traps.c +++ b/arch/arm64/kernel/traps.c @@ -816,6 +816,7 @@ static const char *esr_class_str[] = { [ESR_ELx_EC_SVE] = "SVE", [ESR_ELx_EC_ERET] = "ERET/ERETAA/ERETAB", [ESR_ELx_EC_FPAC] = "FPAC", + [ESR_ELx_EC_SME] = "SME", [ESR_ELx_EC_IMP_DEF] = "EL3 IMP DEF", [ESR_ELx_EC_IABT_LOW] = "IABT (lower EL)", [ESR_ELx_EC_IABT_CUR] = "IABT (current EL)",

SME requires similar setup to that for SVE: disable traps to EL2 and make sure that the maximum vector length is available to EL1, for SME we have two traps - one for SME itself and one for TPIDR2.

In addition since we currently make no active use of priority control for SCMUs we map all SME priorities lower ELs may configure to 0, the architecture specified minimum priority, to ensure that nothing we manage is able to configure itself to consume excessive resources. This will need to be revisited should there be a need to manage SME priorities at runtime.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/el2_setup.h | 36 ++++++++++++++++++++++++++++++ 1 file changed, 36 insertions(+)

diff --git a/arch/arm64/include/asm/el2_setup.h b/arch/arm64/include/asm/el2_setup.h index 3198acb2aad8..895a27a1dcb5 100644 --- a/arch/arm64/include/asm/el2_setup.h +++ b/arch/arm64/include/asm/el2_setup.h @@ -143,6 +143,41 @@ .Lskip_sve_@: .endm

+/* SME register access and priority mapping */ +.macro __init_el2_nvhe_sme + mrs x1, id_aa64pfr1_el1 + ubfx x1, x1, #ID_AA64PFR1_SME_SHIFT, #4 + cbz x1, .Lskip_sme_@ + + bic x0, x0, #CPTR_EL2_TSM // Also disable SME traps + msr cptr_el2, x0 // Disable copro. traps to EL2 + isb + + mrs x1, sctlr_el2 + orr x1, x1, #SCTLR_ELx_ENTP2 // Disable TPIDR2 traps + msr sctlr_el2, x1 + isb + + mov x1, #SMCR_ELx_LEN_MASK // Enable full SME vector + msr_s SYS_SMCR_EL2, x1 // length for EL1. + + mrs_s x1, SYS_SMIDR_EL1 // Priority mapping supported? + ubfx x1, x1, #SYS_SMIDR_EL1_SMPS_SHIFT, #1 + cbz x1, .Lskip_sme_@ + + msr_s SYS_SMPRIMAP_EL2, xzr // Make all priorities equal + + mrs x1, id_aa64mmfr1_el1 // HCRX_EL2 present? + ubfx x1, x1, #ID_AA64MMFR1_HCX_SHIFT, #4 + cbz x1, .Lskip_sme_@ + + mrs_s x1, SYS_HCRX_EL2 + orr x1, x1, #HCRX_EL2_SMPME_MASK // Enable priority mapping + msr_s SYS_HCRX_EL2, x1 + +.Lskip_sme_@: +.endm + /* Disable any fine grained traps */ .macro __init_el2_fgt mrs x1, id_aa64mmfr0_el1 @@ -196,6 +231,7 @@ __init_el2_nvhe_idregs __init_el2_nvhe_cptr __init_el2_nvhe_sve + __init_el2_nvhe_sme __init_el2_fgt __init_el2_nvhe_prepare_eret .endm

The vector lengths used for SME are controlled through a similar set of registers to those for SVE and enumerated using a similar algorithm with some slight differences due to the fact that unlike SVE there are no restrictions on which combinations of vector lengths can be supported nor any mandatory vector lengths which must be implemented. Add a new vector type and implement support for enumerating it.

One slightly awkward feature is that we need to read the current vector length using the SVE RVDL instruction while in streaming mode. Rather than add an ops structure we add special cases directly in the otherwise generic vec_probe_vqs() function, this is a bit inelegant but it's the only place where this is an issue.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/cpu.h | 3 + arch/arm64/include/asm/cpufeature.h | 7 ++ arch/arm64/include/asm/fpsimd.h | 44 ++++++++++ arch/arm64/include/asm/processor.h | 1 + arch/arm64/kernel/cpufeature.c | 49 +++++++++++ arch/arm64/kernel/cpuinfo.c | 4 + arch/arm64/kernel/fpsimd.c | 131 +++++++++++++++++++++++++++- 7 files changed, 238 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/cpu.h b/arch/arm64/include/asm/cpu.h index 667b66fe1a53..707f30dccbf1 100644 --- a/arch/arm64/include/asm/cpu.h +++ b/arch/arm64/include/asm/cpu.h @@ -63,6 +63,9 @@ struct cpuinfo_arm64 {

/* pseudo-ZCR for recording maximum ZCR_EL1 LEN value: */ u64 reg_zcr; + + /* pseudo-SMCR for recording maximum ZCR_EL1 LEN value: */ + u64 reg_smcr; };

DECLARE_PER_CPU(struct cpuinfo_arm64, cpu_data); diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h index e1b745bf5fbe..9a183267b341 100644 --- a/arch/arm64/include/asm/cpufeature.h +++ b/arch/arm64/include/asm/cpufeature.h @@ -619,6 +619,13 @@ static inline bool id_aa64pfr0_sve(u64 pfr0) return val > 0; }

+static inline bool id_aa64pfr1_sme(u64 pfr1) +{ + u32 val = cpuid_feature_extract_unsigned_field(pfr1, ID_AA64PFR1_SME_SHIFT); + + return val > 0; +} + static inline bool id_aa64pfr1_mte(u64 pfr1) { u32 val = cpuid_feature_extract_unsigned_field(pfr1, ID_AA64PFR1_MTE_SHIFT); diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index e28e09f30ad4..a65779fe6d56 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -77,6 +77,7 @@ extern void sve_kernel_enable(const struct arm64_cpu_capabilities *__unused); extern void sme_kernel_enable(const struct arm64_cpu_capabilities *__unused);

extern u64 read_zcr_features(void); +extern u64 read_smcr_features(void);

/* * Helpers to translate bit indices in sve_vq_map to VQ values (and @@ -171,6 +172,12 @@ static inline void write_vl(enum vec_type type, u64 val) tmp = read_sysreg_s(SYS_ZCR_EL1) & ~ZCR_ELx_LEN_MASK; write_sysreg_s(tmp | val, SYS_ZCR_EL1); break; +#endif +#ifdef CONFIG_ARM64_SME + case ARM64_VEC_SME: + tmp = read_sysreg_s(SYS_SMCR_EL1) & ~SMCR_ELx_LEN_MASK; + write_sysreg_s(tmp | val, SYS_SMCR_EL1); + break; #endif default: WARN_ON_ONCE(1); @@ -245,6 +252,43 @@ static inline void sve_setup(void) { }

#endif /* ! CONFIG_ARM64_SVE */

+#ifdef CONFIG_ARM64_SME + +extern void __init sme_setup(void); + +static inline void sme_smstart_sm(void) +{ + /* SMSTART SM is an alias for MSR SVCRSM, #1 */ + asm volatile(".inst 0xd503437f"); +} + +static inline void sme_smstop_sm(void) +{ + /* SMSTOP SM is an alias for MSR SVCRSM, #0 */ + asm volatile(".inst 0xd503427f"); +} + +static inline int sme_max_vl(void) +{ + return vec_max_vl(ARM64_VEC_SME); +} + +static inline int sme_max_virtualisable_vl(void) +{ + return vec_max_virtualisable_vl(ARM64_VEC_SME); +} + +#else + +static inline void sme_setup(void) { } +static inline int sme_max_vl(void) { return 0; } +static inline int sme_max_virtualisable_vl(void) { return 0; } + +static inline void sme_smstart_sm(void) { } +static inline void sme_smstop_sm(void) { } + +#endif /* ! CONFIG_ARM64_SME */ + /* For use by EFI runtime services calls only */ extern void __efi_fpsimd_begin(void); extern void __efi_fpsimd_end(void); diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 9b854e8196df..575a1fe719b7 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -117,6 +117,7 @@ struct debug_info {

enum vec_type { ARM64_VEC_SVE = 0, + ARM64_VEC_SME, ARM64_VEC_MAX, };

diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index 9d3e87ba5d5a..84aec4704885 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -562,6 +562,12 @@ static const struct arm64_ftr_bits ftr_zcr[] = { ARM64_FTR_END, };

+static const struct arm64_ftr_bits ftr_smcr[] = { + ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, + SMCR_ELx_LEN_SHIFT, SMCR_ELx_LEN_SIZE, 0), /* LEN */ + 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 @@ -660,6 +666,7 @@ static const struct __ftr_reg_entry {

/* Op1 = 0, CRn = 1, CRm = 2 */ ARM64_FTR_REG(SYS_ZCR_EL1, ftr_zcr), + ARM64_FTR_REG(SYS_SMCR_EL1, ftr_smcr),

/* Op1 = 1, CRn = 0, CRm = 0 */ ARM64_FTR_REG(SYS_GMID_EL1, ftr_gmid), @@ -963,6 +970,14 @@ void __init init_cpu_features(struct cpuinfo_arm64 *info) vec_init_vq_map(ARM64_VEC_SVE); }

+ if (id_aa64pfr1_sme(info->reg_id_aa64pfr1)) { + init_cpu_ftr_reg(SYS_SMCR_EL1, info->reg_smcr); + if (IS_ENABLED(CONFIG_ARM64_SME)) { + sme_kernel_enable(NULL); + vec_init_vq_map(ARM64_VEC_SME); + } + } + if (id_aa64pfr1_mte(info->reg_id_aa64pfr1)) init_cpu_ftr_reg(SYS_GMID_EL1, info->reg_gmid);

@@ -1187,6 +1202,9 @@ void update_cpu_features(int cpu, taint |= check_update_ftr_reg(SYS_ID_AA64ZFR0_EL1, cpu, info->reg_id_aa64zfr0, boot->reg_id_aa64zfr0);

+ taint |= check_update_ftr_reg(SYS_ID_AA64SMFR0_EL1, cpu, + info->reg_id_aa64smfr0, boot->reg_id_aa64smfr0); + if (id_aa64pfr0_sve(info->reg_id_aa64pfr0)) { taint |= check_update_ftr_reg(SYS_ZCR_EL1, cpu, info->reg_zcr, boot->reg_zcr); @@ -1197,6 +1215,16 @@ void update_cpu_features(int cpu, vec_update_vq_map(ARM64_VEC_SVE); }

+ if (id_aa64pfr1_sme(info->reg_id_aa64pfr1)) { + taint |= check_update_ftr_reg(SYS_SMCR_EL1, cpu, + info->reg_smcr, boot->reg_smcr); + + /* Probe vector lengths, unless we already gave up on SME */ + if (id_aa64pfr1_sme(read_sanitised_ftr_reg(SYS_ID_AA64PFR1_EL1)) && + !system_capabilities_finalized()) + vec_update_vq_map(ARM64_VEC_SME); + } + /* * The kernel uses the LDGM/STGM instructions and the number of tags * they read/write depends on the GMID_EL1.BS field. Check that the @@ -2789,6 +2817,23 @@ static void verify_sve_features(void) /* Add checks on other ZCR bits here if necessary */ }

+static void verify_sme_features(void) +{ + u64 safe_smcr = read_sanitised_ftr_reg(SYS_SMCR_EL1); + u64 smcr = read_smcr_features(); + + unsigned int safe_len = safe_smcr & SMCR_ELx_LEN_MASK; + unsigned int len = smcr & SMCR_ELx_LEN_MASK; + + if (len < safe_len || vec_verify_vq_map(ARM64_VEC_SME)) { + pr_crit("CPU%d: SME: vector length support mismatch\n", + smp_processor_id()); + cpu_die_early(); + } + + /* Add checks on other SMCR bits here if necessary */ +} + static void verify_hyp_capabilities(void) { u64 safe_mmfr1, mmfr0, mmfr1; @@ -2841,6 +2886,9 @@ static void verify_local_cpu_capabilities(void) if (system_supports_sve()) verify_sve_features();

+ if (system_supports_sme()) + verify_sme_features(); + if (is_hyp_mode_available()) verify_hyp_capabilities(); } @@ -2957,6 +3005,7 @@ void __init setup_cpu_features(void) pr_info("emulated: Privileged Access Never (PAN) using TTBR0_EL1 switching\n");

sve_setup(); + sme_setup(); minsigstksz_setup();

/* Advertise that we have computed the system capabilities */ diff --git a/arch/arm64/kernel/cpuinfo.c b/arch/arm64/kernel/cpuinfo.c index 9830fa0c7647..572da7d1f194 100644 --- a/arch/arm64/kernel/cpuinfo.c +++ b/arch/arm64/kernel/cpuinfo.c @@ -415,6 +415,10 @@ static void __cpuinfo_store_cpu(struct cpuinfo_arm64 *info) id_aa64pfr0_sve(info->reg_id_aa64pfr0)) info->reg_zcr = read_zcr_features();

+ if (IS_ENABLED(CONFIG_ARM64_SME) && + id_aa64pfr1_sme(info->reg_id_aa64pfr1)) + info->reg_smcr = read_smcr_features(); + cpuinfo_detect_icache_policy(info); }

diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 8416e4889596..0b2c4dfdf9fb 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -132,6 +132,12 @@ __ro_after_init struct vl_info vl_info[ARM64_VEC_MAX] = { .max_virtualisable_vl = SVE_VL_MIN, }, #endif +#ifdef CONFIG_ARM64_SME + [ARM64_VEC_SME] = { + .type = ARM64_VEC_SME, + .name = "SME", + }, +#endif };

static unsigned int vec_vl_inherit_flag(enum vec_type type) @@ -182,6 +188,20 @@ extern void __percpu *efi_sve_state;

#endif /* ! CONFIG_ARM64_SVE */

+#ifdef CONFIG_ARM64_SME + +static int get_sme_default_vl(void) +{ + return get_default_vl(ARM64_VEC_SME); +} + +static void set_sme_default_vl(int val) +{ + set_default_vl(ARM64_VEC_SME, val); +} + +#endif + DEFINE_PER_CPU(bool, fpsimd_context_busy); EXPORT_PER_CPU_SYMBOL(fpsimd_context_busy);

@@ -399,6 +419,8 @@ static unsigned int find_supported_vector_length(enum vec_type type,

if (vl > max_vl) vl = max_vl; + if (vl < info->min_vl) + vl = info->min_vl;

bit = find_next_bit(info->vq_map, SVE_VQ_MAX, __vq_to_bit(sve_vq_from_vl(vl))); @@ -758,12 +780,38 @@ static void vec_probe_vqs(struct vl_info *info,

bitmap_zero(map, SVE_VQ_MAX);

+ /* + * Enter streaming mode for SME; we don't use an op as the + * vector length info is used from KVM. + */ + switch (info->type) { + case ARM64_VEC_SME: + sme_smstart_sm(); + break; + default: + break; + } + for (vq = SVE_VQ_MAX; vq >= SVE_VQ_MIN; --vq) { write_vl(info->type, vq - 1); /* self-syncing */ + vl = sve_get_vl(); + + /* Minimum VL identified? */ + if (sve_vq_from_vl(vl) > vq) + break; + vq = sve_vq_from_vl(vl); /* skip intervening lengths */ set_bit(__vq_to_bit(vq), map); } + + switch (info->type) { + case ARM64_VEC_SME: + sme_smstop_sm(); + break; + default: + break; + } }

/* @@ -996,7 +1044,88 @@ void sme_kernel_enable(const struct arm64_cpu_capabilities *__always_unused p) isb(); }

-#endif /* CONFIG_ARM64_SVE */ +/* + * Read the pseudo-SMCR used by cpufeatures to identify the supported + * vector length. + * + * Use only if SME is present. + * This function clobbers the SME vector length. + */ +u64 read_smcr_features(void) +{ + u64 smcr; + unsigned int vq_max; + + sme_kernel_enable(NULL); + sme_smstart_sm(); + + /* + * Set the maximum possible VL, and write zeroes to all other + * bits to see if they stick. + */ + write_sysreg_s(SMCR_ELx_LEN_MASK, SYS_SMCR_EL1); + + smcr = read_sysreg_s(SYS_SMCR_EL1); + smcr &= ~(u64)SMCR_ELx_LEN_MASK; /* find sticky 1s outside LEN field */ + vq_max = sve_vq_from_vl(sve_get_vl()); + smcr |= vq_max - 1; /* set LEN field to maximum effective value */ + + sme_smstop_sm(); + + return smcr; +} + +void __init sme_setup(void) +{ + struct vl_info *info = &vl_info[ARM64_VEC_SME]; + u64 smcr; + int min_bit; + + if (!system_supports_sme()) + return; + + /* + * SME doesn't require any particular vector length be + * supported but it does require at least one. We should have + * disabled the feature entirely while bringing up CPUs but + * let's double check here. + */ + WARN_ON(bitmap_empty(info->vq_map, SVE_VQ_MAX)); + + min_bit = find_last_bit(info->vq_map, SVE_VQ_MAX); + info->min_vl = sve_vl_from_vq(__bit_to_vq(min_bit)); + + smcr = read_sanitised_ftr_reg(SYS_SMCR_EL1); + info->max_vl = sve_vl_from_vq((smcr & SMCR_ELx_LEN_MASK) + 1); + + /* + * Sanity-check that the max VL we determined through CPU features + * corresponds properly to sme_vq_map. If not, do our best: + */ + if (WARN_ON(info->max_vl != find_supported_vector_length(ARM64_VEC_SME, + info->max_vl))) + info->max_vl = find_supported_vector_length(ARM64_VEC_SME, + info->max_vl); + + WARN_ON(info->min_vl > info->max_vl); + + /* + * For the default VL, pick the maximum supported value <= 32 + * (256 bits) if there is one since this is guaranteed not to + * grow the signal frame when in streaming mode, otherwise the + * minimum available VL will be used. + */ + set_sme_default_vl(find_supported_vector_length(ARM64_VEC_SME, 32)); + + pr_info("SME: minimum available vector length %u bytes per vector\n", + info->min_vl); + pr_info("SME: maximum available vector length %u bytes per vector\n", + info->max_vl); + pr_info("SME: default vector length %u bytes per vector\n", + get_sme_default_vl()); +} + +#endif /* CONFIG_ARM64_SME */

/* * Trapped SVE access

As for SVE provide a prctl() interface which allows processes to configure their SME vector length.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/fpsimd.h | 13 +++++++++++ arch/arm64/include/asm/processor.h | 4 +++- arch/arm64/include/asm/thread_info.h | 1 + arch/arm64/kernel/fpsimd.c | 32 ++++++++++++++++++++++++++++ include/uapi/linux/prctl.h | 9 ++++++++ kernel/sys.c | 12 +++++++++++ 6 files changed, 70 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index a65779fe6d56..3550317cbb3d 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -278,6 +278,9 @@ static inline int sme_max_virtualisable_vl(void) return vec_max_virtualisable_vl(ARM64_VEC_SME); }

+extern int sme_set_current_vl(unsigned long arg); +extern int sme_get_current_vl(void); + #else

static inline void sme_setup(void) { } @@ -287,6 +290,16 @@ static inline int sme_max_virtualisable_vl(void) { return 0; } static inline void sme_smstart_sm(void) { } static inline void sme_smstop_sm(void) { }

+static inline int sme_set_current_vl(unsigned long arg) +{ + return -EINVAL; +} + +static inline int sme_get_current_vl(void) +{ + return -EINVAL; +} + #endif /* ! CONFIG_ARM64_SME */

/* For use by EFI runtime services calls only */ diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 575a1fe719b7..a62d2f8045bf 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -354,9 +354,11 @@ extern void __init minsigstksz_setup(void); */ #include <asm/fpsimd.h>

-/* Userspace interface for PR_SVE_{SET,GET}_VL prctl()s: */ +/* Userspace interface for PR_S[MV]E_{SET,GET}_VL prctl()s: */ #define SVE_SET_VL(arg) sve_set_current_vl(arg) #define SVE_GET_VL() sve_get_current_vl() +#define SME_SET_VL(arg) sme_set_current_vl(arg) +#define SME_GET_VL() sme_get_current_vl()

/* PR_PAC_RESET_KEYS prctl */ #define PAC_RESET_KEYS(tsk, arg) ptrauth_prctl_reset_keys(tsk, arg) diff --git a/arch/arm64/include/asm/thread_info.h b/arch/arm64/include/asm/thread_info.h index d5c8ac81ce11..5c4355204f4a 100644 --- a/arch/arm64/include/asm/thread_info.h +++ b/arch/arm64/include/asm/thread_info.h @@ -81,6 +81,7 @@ int arch_dup_task_struct(struct task_struct *dst, #define TIF_SVE_VL_INHERIT 24 /* Inherit SVE vl_onexec across exec */ #define TIF_SSBD 25 /* Wants SSB mitigation */ #define TIF_TAGGED_ADDR 26 /* Allow tagged user addresses */ +#define TIF_SME_VL_INHERIT 28 /* Inherit SME vl_onexec across exec */

#define _TIF_SIGPENDING (1 << TIF_SIGPENDING) #define _TIF_NEED_RESCHED (1 << TIF_NEED_RESCHED) diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 65b6ccc3f8de..a31b47d78291 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -145,6 +145,8 @@ static unsigned int vec_vl_inherit_flag(enum vec_type type) switch (type) { case ARM64_VEC_SVE: return TIF_SVE_VL_INHERIT; + case ARM64_VEC_SME: + return TIF_SME_VL_INHERIT; default: WARN_ON_ONCE(1); return 0; @@ -797,6 +799,36 @@ int sve_get_current_vl(void) return vec_prctl_status(ARM64_VEC_SVE, 0); }

+#ifdef CONFIG_ARM64_SME +/* PR_SME_SET_VL */ +int sme_set_current_vl(unsigned long arg) +{ + unsigned long vl, flags; + int ret; + + vl = arg & PR_SME_VL_LEN_MASK; + flags = arg & ~vl; + + if (!system_supports_sme() || is_compat_task()) + return -EINVAL; + + ret = vec_set_vector_length(current, ARM64_VEC_SME, vl, flags); + if (ret) + return ret; + + return vec_prctl_status(ARM64_VEC_SME, flags); +} + +/* PR_SME_GET_VL */ +int sme_get_current_vl(void) +{ + if (!system_supports_sme() || is_compat_task()) + return -EINVAL; + + return vec_prctl_status(ARM64_VEC_SME, 0); +} +#endif /* CONFIG_ARM64_SME */ + static void vec_probe_vqs(struct vl_info *info, DECLARE_BITMAP(map, SVE_VQ_MAX)) { diff --git a/include/uapi/linux/prctl.h b/include/uapi/linux/prctl.h index 43bd7f713c39..b3212d73c198 100644 --- a/include/uapi/linux/prctl.h +++ b/include/uapi/linux/prctl.h @@ -269,4 +269,13 @@ struct prctl_mm_map { # define PR_SCHED_CORE_SHARE_FROM 3 /* pull core_sched cookie to pid */ # define PR_SCHED_CORE_MAX 4

+/* arm64 Scalable Matrix Extension controls */ +/* Flag values must be in sync with SVE versions */ +#define PR_SME_SET_VL 63 /* set task vector length */ +# define PR_SME_SET_VL_ONEXEC (1 << 18) /* defer effect until exec */ +#define PR_SME_GET_VL 64 /* get task vector length */ +/* Bits common to PR_SME_SET_VL and PR_SME_GET_VL */ +# define PR_SME_VL_LEN_MASK 0xffff +# define PR_SME_VL_INHERIT (1 << 17) /* inherit across exec */ + #endif /* _LINUX_PRCTL_H */ diff --git a/kernel/sys.c b/kernel/sys.c index 8fdac0d90504..59b1ce29b6c9 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -116,6 +116,12 @@ #ifndef SVE_GET_VL # define SVE_GET_VL() (-EINVAL) #endif +#ifndef SME_SET_VL +# define SME_SET_VL(a) (-EINVAL) +#endif +#ifndef SME_GET_VL +# define SME_GET_VL() (-EINVAL) +#endif #ifndef PAC_RESET_KEYS # define PAC_RESET_KEYS(a, b) (-EINVAL) #endif @@ -2463,6 +2469,12 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3, case PR_SVE_GET_VL: error = SVE_GET_VL(); break; + case PR_SME_SET_VL: + error = SME_SET_VL(arg2); + break; + case PR_SME_GET_VL: + error = SME_GET_VL(); + break; case PR_GET_SPECULATION_CTRL: if (arg3 || arg4 || arg5) return -EINVAL;

In SME the use of both streaming SVE mode and ZA are tracked through PSTATE.SM and PSTATE.ZA, visible through the system register SVCR. In order to context switch the floating point state for SME we need to context switch the contents of this register as part of context switching the floating point state.

Since changing the vector length exits streaming SVE mode and disables ZA we also make sure we update SVCR appropriately when setting vector length, and similarly ensure that new threads have streaming SVE mode and ZA disabled.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/processor.h | 1 + arch/arm64/include/asm/thread_info.h | 1 + arch/arm64/kernel/fpsimd.c | 11 +++++++++++ arch/arm64/kernel/process.c | 2 ++ 4 files changed, 15 insertions(+)

diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 51eca2513cb5..3c235e165725 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -168,6 +168,7 @@ struct thread_struct { u64 mte_ctrl; #endif u64 sctlr_user; + u64 svcr; u64 tpidr2_el0; };

diff --git a/arch/arm64/include/asm/thread_info.h b/arch/arm64/include/asm/thread_info.h index 5c4355204f4a..03cb88f63317 100644 --- a/arch/arm64/include/asm/thread_info.h +++ b/arch/arm64/include/asm/thread_info.h @@ -81,6 +81,7 @@ int arch_dup_task_struct(struct task_struct *dst, #define TIF_SVE_VL_INHERIT 24 /* Inherit SVE vl_onexec across exec */ #define TIF_SSBD 25 /* Wants SSB mitigation */ #define TIF_TAGGED_ADDR 26 /* Allow tagged user addresses */ +#define TIF_SME 27 /* SME in use */ #define TIF_SME_VL_INHERIT 28 /* Inherit SME vl_onexec across exec */

#define _TIF_SIGPENDING (1 << TIF_SIGPENDING) diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index cb246ba4ca98..0064fa45ef38 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -355,6 +355,9 @@ static void task_fpsimd_load(void) WARN_ON(!system_supports_fpsimd()); WARN_ON(!have_cpu_fpsimd_context());

+ if (IS_ENABLED(CONFIG_ARM64_SME) && test_thread_flag(TIF_SME)) + write_sysreg_s(current->thread.svcr, SYS_SVCR_EL0); + if (IS_ENABLED(CONFIG_ARM64_SVE) && test_thread_flag(TIF_SVE)) { sve_set_vq(sve_vq_from_vl(task_get_sve_vl(current)) - 1); sve_load_state(sve_pffr(&current->thread), @@ -380,6 +383,10 @@ static void fpsimd_save(void) if (test_thread_flag(TIF_FOREIGN_FPSTATE)) return;

+ if (IS_ENABLED(CONFIG_ARM64_SME) && + test_thread_flag(TIF_SME)) + current->thread.svcr = read_sysreg_s(SYS_SVCR_EL0); + if (IS_ENABLED(CONFIG_ARM64_SVE) && test_thread_flag(TIF_SVE)) { if (WARN_ON(sve_get_vl() != last->sve_vl)) { @@ -731,6 +738,10 @@ int vec_set_vector_length(struct task_struct *task, enum vec_type type, if (test_and_clear_tsk_thread_flag(task, TIF_SVE)) sve_to_fpsimd(task);

+ if (system_supports_sme() && type == ARM64_VEC_SME) + task->thread.svcr &= ~(SYS_SVCR_EL0_SM_MASK | + SYS_SVCR_EL0_ZA_MASK); + if (task == current) put_cpu_fpsimd_context();

diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c index 3f4279ad68bc..4dd35619ece6 100644 --- a/arch/arm64/kernel/process.c +++ b/arch/arm64/kernel/process.c @@ -309,6 +309,8 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src) dst->thread.sve_state = NULL; clear_tsk_thread_flag(dst, TIF_SVE);

+ dst->thread.svcr = 0; + /* clear any pending asynchronous tag fault raised by the parent */ clear_tsk_thread_flag(dst, TIF_MTE_ASYNC_FAULT);

Allocate space for storing ZA on first access to SME and use that to save and restore ZA state when context switching. We do this by using the vector form of the LDR and STR ZA instructions, these do not require streaming mode and have implementation recommendations that they avoid contention issues in shared SMCU implementations.

Since ZA is architecturally guaranteed to be zeroed when enabled we do not need to explicitly zero ZA, either we will be restoring from a saved copy or trapping on first use of SME so we know that ZA must be disabled.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/fpsimd.h | 4 +++- arch/arm64/include/asm/fpsimdmacros.h | 22 ++++++++++++++++++++++ arch/arm64/include/asm/processor.h | 1 + arch/arm64/kernel/entry-fpsimd.S | 22 ++++++++++++++++++++++ arch/arm64/kernel/fpsimd.c | 17 +++++++++++------ arch/arm64/kvm/fpsimd.c | 2 +- 6 files changed, 60 insertions(+), 8 deletions(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 391db07566aa..c9cefb17d534 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -47,7 +47,7 @@ extern void fpsimd_update_current_state(struct user_fpsimd_state const *state);

extern void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *state, void *sve_state, unsigned int sve_vl, - unsigned int sme_vl); + void *za_state, unsigned int sme_vl);

extern void fpsimd_flush_task_state(struct task_struct *target); extern void fpsimd_save_and_flush_cpu_state(void); @@ -90,6 +90,8 @@ extern void sve_flush_live(bool flush_ffr, unsigned long vq_minus_1); extern unsigned int sve_get_vl(void); extern void sve_set_vq(unsigned long vq_minus_1); extern void sme_set_vq(unsigned long vq_minus_1); +extern void sme_save_state(void *state, unsigned int vq_minus_1); +extern void sme_load_state(void const *state, unsigned int vq_minus_1);

struct arm64_cpu_capabilities; extern void sve_kernel_enable(const struct arm64_cpu_capabilities *__unused); diff --git a/arch/arm64/include/asm/fpsimdmacros.h b/arch/arm64/include/asm/fpsimdmacros.h index 57bcd9ec4cb9..113c37b1a8e9 100644 --- a/arch/arm64/include/asm/fpsimdmacros.h +++ b/arch/arm64/include/asm/fpsimdmacros.h @@ -309,3 +309,25 @@ ldr w\nxtmp, [\xpfpsr, #4] msr fpcr, x\nxtmp .endm + +.macro sme_save_za nxbase, xvl, nw + mov w\nw, #0 + +423: + _sme_str_zav \nw, \nxbase + add x\nxbase, x\nxbase, \xvl + add x\nw, x\nw, #1 + cmp \xvl, x\nw + bne 423b +.endm + +.macro sme_load_za nxbase, xvl, nw + mov w\nw, #0 + +423: + _sme_ldr_zav \nw, \nxbase + add x\nxbase, x\nxbase, \xvl + add x\nw, x\nw, #1 + cmp \xvl, x\nw + bne 423b +.endm diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 338cb03811bd..e4688a58f365 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -153,6 +153,7 @@ struct thread_struct {

unsigned int fpsimd_cpu; void *sve_state; /* SVE registers, if any */ + void *za_state; /* ZA register, if any */ unsigned int vl[ARM64_VEC_MAX]; /* vector length */ unsigned int vl_onexec[ARM64_VEC_MAX]; /* vl after next exec */ unsigned long fault_address; /* fault info */ diff --git a/arch/arm64/kernel/entry-fpsimd.S b/arch/arm64/kernel/entry-fpsimd.S index b06bf42e7856..6a224cfcfd50 100644 --- a/arch/arm64/kernel/entry-fpsimd.S +++ b/arch/arm64/kernel/entry-fpsimd.S @@ -94,4 +94,26 @@ SYM_FUNC_START(sme_set_vq) ret SYM_FUNC_END(sme_set_vq)

+/* + * Save the SME state + * + * x0 - pointer to buffer for state + * x1 - Bytes per vector + */ +SYM_FUNC_START(sme_save_state) + sme_save_za 0, x1, 12 + ret +SYM_FUNC_END(sme_save_state) + +/* + * Load the SME state + * + * x0 - pointer to buffer for state + * x1 - bytes per vector + */ +SYM_FUNC_START(sme_load_state) + sme_load_za 0, x1, 12 + ret +SYM_FUNC_END(sme_load_state) + #endif /* CONFIG_ARM64_SME */ diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 0aad81d60818..348364980878 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -117,6 +117,7 @@ struct fpsimd_last_state_struct { struct user_fpsimd_state *st; void *sve_state; + void *za_state; unsigned int sve_vl; unsigned int sme_vl; }; @@ -382,11 +383,15 @@ static void task_fpsimd_load(void) if (system_supports_sme()) { unsigned long sme_vl = task_get_sme_vl(current);

+ /* Ensure VL is set up for restoring data */ if (test_thread_flag(TIF_SME)) sme_set_vq(sve_vq_from_vl(sme_vl) - 1);

write_sysreg_s(current->thread.svcr, SYS_SVCR_EL0);

+ if (thread_za_enabled(&current->thread)) + sme_load_state(current->thread.za_state, sme_vl); + if (thread_sm_enabled(&current->thread)) { restore_sve_regs = true; restore_ffr = false; @@ -429,11 +434,8 @@ static void fpsimd_save(void) if (system_supports_sme()) { current->thread.svcr = read_sysreg_s(SYS_SVCR_EL0);

- if (thread_za_enabled(&current->thread)) { - /* ZA state managment is not implemented yet */ - force_signal_inject(SIGKILL, SI_KERNEL, 0, 0); - return; - } + if (thread_za_enabled(&current->thread)) + sme_save_state(last->za_state, last->sme_vl);

/* If we are in streaming mode override regular SVE. */ if (thread_sm_enabled(&current->thread)) { @@ -1470,6 +1472,7 @@ static void fpsimd_bind_task_to_cpu(void) WARN_ON(!system_supports_fpsimd()); last->st = &current->thread.uw.fpsimd_state; last->sve_state = current->thread.sve_state; + last->za_state = current->thread.za_state; last->sve_vl = task_get_sve_vl(current); last->sme_vl = task_get_sme_vl(current); current->thread.fpsimd_cpu = smp_processor_id(); @@ -1486,7 +1489,8 @@ static void fpsimd_bind_task_to_cpu(void) }

void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, - unsigned int sve_vl, unsigned int sme_vl) + unsigned int sve_vl, void *za_state, + unsigned int sme_vl) { struct fpsimd_last_state_struct *last = this_cpu_ptr(&fpsimd_last_state); @@ -1496,6 +1500,7 @@ void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state,

last->st = st; last->sve_state = sve_state; + last->za_state = za_state; last->sve_vl = sve_vl; last->sme_vl = sme_vl; } diff --git a/arch/arm64/kvm/fpsimd.c b/arch/arm64/kvm/fpsimd.c index d96871002081..007b2e8b9ae9 100644 --- a/arch/arm64/kvm/fpsimd.c +++ b/arch/arm64/kvm/fpsimd.c @@ -100,7 +100,7 @@ void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu) fpsimd_bind_state_to_cpu(&vcpu->arch.ctxt.fp_regs, vcpu->arch.sve_state, vcpu->arch.sve_max_vl, - 0); + NULL, 0);

clear_thread_flag(TIF_FOREIGN_FPSTATE); update_thread_flag(TIF_SVE, vcpu_has_sve(vcpu));

When in streaming mode we have the same set of SVE registers as we do in regular SVE mode with the exception of FFR and the use of the SME vector length. Provide signal handling for these registers by taking one of the reserved words in the SVE signal context as a flags field and defining a flag with a flag which is set for streaming mode. When the flag is set the vector length is set to the streaming mode vector length and we save and restore streaming mode data. We support entering or leaving streaming mode based on the value of the flag but do not support changing the vector length, this is not currently supported SVE signal handling.

We could instead allocate a separate record in the signal frame for the streaming mode SVE context but this inflates the size of the maximal signal frame required and adds complication when validating signal frames from userspace, especially given the current structure of the code.

Any implementation of support for streaming mode vectors in signals will have some potential for causing issues for applications that attempt to handle SVE vectors in signals, use streaming mode but do not understand streaming mode in their signal handling code, it is hard to identify a case that is clearly better than any other - they all have cases where they could cause unexpected register corruption or faults.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/uapi/asm/sigcontext.h | 16 ++++++-- arch/arm64/kernel/signal.c | 48 ++++++++++++++++++------ 2 files changed, 50 insertions(+), 14 deletions(-)

diff --git a/arch/arm64/include/uapi/asm/sigcontext.h b/arch/arm64/include/uapi/asm/sigcontext.h index 0c796c795dbe..3a3366d4fbc2 100644 --- a/arch/arm64/include/uapi/asm/sigcontext.h +++ b/arch/arm64/include/uapi/asm/sigcontext.h @@ -134,9 +134,12 @@ struct extra_context { struct sve_context { struct _aarch64_ctx head; __u16 vl; - __u16 __reserved[3]; + __u16 flags; + __u16 __reserved[2]; };

+#define SVE_SIG_FLAG_SM 0x1 /* Context describes streaming mode */ + #endif /* !__ASSEMBLY__ */

#include <asm/sve_context.h> @@ -186,9 +189,16 @@ struct sve_context { * sve_context.vl must equal the thread's current vector length when * doing a sigreturn. * + * On systems with support for SME the SVE register state may reflect either + * streaming or non-streaming mode. In streaming mode the streaming mode + * vector length will be used and the flag SVE_SIG_FLAG_SM will be set in + * the flags field. It is permitted to enter or leave streaming mode in + * a signal return, applications should take care to ensure that any difference + * in vector length between the two modes is handled, including any resixing + * and movement of context blocks. * - * Note: for all these macros, the "vq" argument denotes the SVE - * vector length in quadwords (i.e., units of 128 bits). + * Note: for all these macros, the "vq" argument denotes the vector length + * in quadwords (i.e., units of 128 bits). * * The correct way to obtain vq is to use sve_vq_from_vl(vl). The * result is valid if and only if sve_vl_valid(vl) is true. This is diff --git a/arch/arm64/kernel/signal.c b/arch/arm64/kernel/signal.c index 8f6372b44b65..fea0e1d30449 100644 --- a/arch/arm64/kernel/signal.c +++ b/arch/arm64/kernel/signal.c @@ -227,11 +227,17 @@ static int preserve_sve_context(struct sve_context __user *ctx) { int err = 0; u16 reserved[ARRAY_SIZE(ctx->__reserved)]; + u16 flags = 0; unsigned int vl = task_get_sve_vl(current); unsigned int vq = 0;

- if (test_thread_flag(TIF_SVE)) + if (thread_sm_enabled(&current->thread)) { + vl = task_get_sme_vl(current); vq = sve_vq_from_vl(vl); + flags |= SVE_SIG_FLAG_SM; + } else if (test_thread_flag(TIF_SVE)) { + vq = sve_vq_from_vl(vl); + }

memset(reserved, 0, sizeof(reserved));

@@ -239,6 +245,7 @@ static int preserve_sve_context(struct sve_context __user *ctx) __put_user_error(round_up(SVE_SIG_CONTEXT_SIZE(vq), 16), &ctx->head.size, err); __put_user_error(vl, &ctx->vl, err); + __put_user_error(flags, &ctx->flags, err); BUILD_BUG_ON(sizeof(ctx->__reserved) != sizeof(reserved)); err |= __copy_to_user(&ctx->__reserved, reserved, sizeof(reserved));

@@ -259,18 +266,28 @@ static int preserve_sve_context(struct sve_context __user *ctx) static int restore_sve_fpsimd_context(struct user_ctxs *user) { int err; - unsigned int vq; + unsigned int vl, vq; struct user_fpsimd_state fpsimd; struct sve_context sve;

if (__copy_from_user(&sve, user->sve, sizeof(sve))) return -EFAULT;

- if (sve.vl != task_get_sve_vl(current)) + if (sve.flags & SVE_SIG_FLAG_SM) { + if (!system_supports_sme()) + return -EINVAL; + + vl = task_get_sme_vl(current); + } else { + vl = task_get_sve_vl(current); + } + + if (sve.vl != vl) return -EINVAL;

if (sve.head.size <= sizeof(*user->sve)) { clear_thread_flag(TIF_SVE); + current->thread.svcr &= ~SYS_SVCR_EL0_SM_MASK; goto fpsimd_only; }

@@ -302,7 +319,10 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user) if (err) return -EFAULT;

- set_thread_flag(TIF_SVE); + if (sve.flags & SVE_SIG_FLAG_SM) + current->thread.svcr |= SYS_SVCR_EL0_SM_MASK; + else + set_thread_flag(TIF_SVE);

fpsimd_only: /* copy the FP and status/control registers */ @@ -394,7 +414,7 @@ static int parse_user_sigframe(struct user_ctxs *user, break;

case SVE_MAGIC: - if (!system_supports_sve()) + if (!system_supports_sve() && !system_supports_sme()) goto invalid;

if (user->sve) @@ -593,11 +613,16 @@ static int setup_sigframe_layout(struct rt_sigframe_user_layout *user, if (system_supports_sve()) { unsigned int vq = 0;

- if (add_all || test_thread_flag(TIF_SVE)) { - int vl = sve_max_vl(); + if (add_all || test_thread_flag(TIF_SVE) || + thread_sm_enabled(&current->thread)) { + int vl = max(sve_max_vl(), sme_max_vl());

- if (!add_all) - vl = task_get_sve_vl(current); + if (!add_all) { + if (thread_sm_enabled(&current->thread)) + vl = task_get_sme_vl(current); + else + vl = task_get_sve_vl(current); + }

vq = sve_vq_from_vl(vl); } @@ -648,8 +673,9 @@ static int setup_sigframe(struct rt_sigframe_user_layout *user, __put_user_error(current->thread.fault_code, &esr_ctx->esr, err); }

- /* Scalable Vector Extension state, if present */ - if (system_supports_sve() && err == 0 && user->sve_offset) { + /* Scalable Vector Extension state (including streaming), if present */ + if ((system_supports_sve() || system_supports_sme()) && + err == 0 && user->sve_offset) { struct sve_context __user *sve_ctx = apply_user_offset(user, user->sve_offset); err |= preserve_sve_context(sve_ctx);

The streaming mode SVE registers are represented using the same data structures as for SVE but since the vector lengths supported and in use may not be the same as SVE we represent them with a new type NT_ARM_SSVE. Unfortunately we only have a single 16 bit reserved field available in the header so there is no space to fit the current and maximum vector length for both standard and streaming SVE mode without redefining the structure in a way the creates a complicatd and fragile ABI. Since FFR is not present in streaming mode it is read and written as zero.

Setting NT_ARM_SSVE registers will put the task into streaming mode, similarly setting NT_ARM_SVE registers will exit it. Reads that do not correspond to the current mode of the task will return the header with no register data. For compatibility reasons on write setting no flag for the register type will be interpreted as setting SVE registers, though users can provide no register data as an alternative mechanism for doing so.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/uapi/asm/ptrace.h | 13 +- arch/arm64/kernel/fpsimd.c | 21 ++- arch/arm64/kernel/ptrace.c | 212 +++++++++++++++++++++------ include/uapi/linux/elf.h | 1 + 4 files changed, 190 insertions(+), 57 deletions(-)

diff --git a/arch/arm64/include/uapi/asm/ptrace.h b/arch/arm64/include/uapi/asm/ptrace.h index 758ae984ff97..522b925a78c1 100644 --- a/arch/arm64/include/uapi/asm/ptrace.h +++ b/arch/arm64/include/uapi/asm/ptrace.h @@ -109,7 +109,7 @@ struct user_hwdebug_state { } dbg_regs[16]; };

-/* SVE/FP/SIMD state (NT_ARM_SVE) */ +/* SVE/FP/SIMD state (NT_ARM_SVE & NT_ARM_SSVE) */

struct user_sve_header { __u32 size; /* total meaningful regset content in bytes */ @@ -220,6 +220,7 @@ struct user_sve_header { (SVE_PT_SVE_PREG_OFFSET(vq, __SVE_NUM_PREGS) - \ SVE_PT_SVE_PREGS_OFFSET(vq))

+/* For streaming mode SVE (SSVE) FFR must be read and written as zero */ #define SVE_PT_SVE_FFR_OFFSET(vq) \ (SVE_PT_REGS_OFFSET + __SVE_FFR_OFFSET(vq))

@@ -240,10 +241,12 @@ struct user_sve_header { - SVE_PT_SVE_OFFSET + (__SVE_VQ_BYTES - 1)) \ / __SVE_VQ_BYTES * __SVE_VQ_BYTES)

-#define SVE_PT_SIZE(vq, flags) \ - (((flags) & SVE_PT_REGS_MASK) == SVE_PT_REGS_SVE ? \ - SVE_PT_SVE_OFFSET + SVE_PT_SVE_SIZE(vq, flags) \ - : SVE_PT_FPSIMD_OFFSET + SVE_PT_FPSIMD_SIZE(vq, flags)) +#define SVE_PT_SIZE(vq, flags) \ + (((flags) & SVE_PT_REGS_MASK) == SVE_PT_REGS_SVE ? \ + SVE_PT_SVE_OFFSET + SVE_PT_SVE_SIZE(vq, flags) \ + : ((((flags) & SVE_PT_REGS_MASK) == SVE_PT_REGS_FPSIMD ? \ + SVE_PT_FPSIMD_OFFSET + SVE_PT_FPSIMD_SIZE(vq, flags) \ + : SVE_PT_REGS_OFFSET)))

/* pointer authentication masks (NT_ARM_PAC_MASK) */

diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 95bffab1c5ea..af46db99ec82 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -622,14 +622,19 @@ static void __fpsimd_to_sve(void *sst, struct user_fpsimd_state const *fst, */ static void fpsimd_to_sve(struct task_struct *task) { - unsigned int vq; + unsigned int vq, vl; void *sst = task->thread.sve_state; struct user_fpsimd_state const *fst = &task->thread.uw.fpsimd_state;

if (!system_supports_sve()) return;

- vq = sve_vq_from_vl(task_get_sve_vl(task)); + if (thread_sm_enabled(&task->thread)) + vl = task_get_sme_vl(task); + else + vl = task_get_sve_vl(task); + + vq = sve_vq_from_vl(vl); __fpsimd_to_sve(sst, fst, vq); }

@@ -646,7 +651,7 @@ static void fpsimd_to_sve(struct task_struct *task) */ static void sve_to_fpsimd(struct task_struct *task) { - unsigned int vq; + unsigned int vq, vl; void const *sst = task->thread.sve_state; struct user_fpsimd_state *fst = &task->thread.uw.fpsimd_state; unsigned int i; @@ -655,7 +660,12 @@ static void sve_to_fpsimd(struct task_struct *task) if (!system_supports_sve()) return;

- vq = sve_vq_from_vl(task_get_sve_vl(task)); + if (thread_sm_enabled(&task->thread)) + vl = task_get_sme_vl(task); + else + vl = task_get_sve_vl(task); + + vq = sve_vq_from_vl(vl); for (i = 0; i < SVE_NUM_ZREGS; ++i) { p = (__uint128_t const *)ZREG(sst, vq, i); fst->vregs[i] = arm64_le128_to_cpu(*p); @@ -796,8 +806,7 @@ int vec_set_vector_length(struct task_struct *task, enum vec_type type, /* * To ensure the FPSIMD bits of the SVE vector registers are preserved, * write any live register state back to task_struct, and convert to a - * regular FPSIMD thread. Since the vector length can only be changed - * with a syscall we can't be in streaming mode while reconfiguring. + * regular FPSIMD thread. */ if (task == current) { get_cpu_fpsimd_context(); diff --git a/arch/arm64/kernel/ptrace.c b/arch/arm64/kernel/ptrace.c index 716dde289446..414126ce5897 100644 --- a/arch/arm64/kernel/ptrace.c +++ b/arch/arm64/kernel/ptrace.c @@ -714,21 +714,51 @@ static int system_call_set(struct task_struct *target, #ifdef CONFIG_ARM64_SVE

static void sve_init_header_from_task(struct user_sve_header *header, - struct task_struct *target) + struct task_struct *target, + enum vec_type type) { unsigned int vq; + bool active; + bool fpsimd_only; + enum vec_type task_type;

memset(header, 0, sizeof(*header));

- header->flags = test_tsk_thread_flag(target, TIF_SVE) ? - SVE_PT_REGS_SVE : SVE_PT_REGS_FPSIMD; - if (test_tsk_thread_flag(target, TIF_SVE_VL_INHERIT)) - header->flags |= SVE_PT_VL_INHERIT; + /* Check if the requested registers are active for the task */ + if (thread_sm_enabled(&target->thread)) + task_type = ARM64_VEC_SME; + else + task_type = ARM64_VEC_SVE; + active = (task_type == type); + + switch (type) { + case ARM64_VEC_SVE: + if (test_tsk_thread_flag(target, TIF_SVE_VL_INHERIT)) + header->flags |= SVE_PT_VL_INHERIT; + fpsimd_only = !test_tsk_thread_flag(target, TIF_SVE); + break; + case ARM64_VEC_SME: + if (test_tsk_thread_flag(target, TIF_SME_VL_INHERIT)) + header->flags |= SVE_PT_VL_INHERIT; + fpsimd_only = false; + break; + default: + WARN_ON_ONCE(1); + return; + }

- header->vl = task_get_sve_vl(target); + if (active) { + if (fpsimd_only) { + header->flags |= SVE_PT_REGS_FPSIMD; + } else { + header->flags |= SVE_PT_REGS_SVE; + } + } + + header->vl = task_get_vl(target, type); vq = sve_vq_from_vl(header->vl);

- header->max_vl = sve_max_vl(); + header->max_vl = vec_max_vl(type); header->size = SVE_PT_SIZE(vq, header->flags); header->max_size = SVE_PT_SIZE(sve_vq_from_vl(header->max_vl), SVE_PT_REGS_SVE); @@ -739,19 +769,17 @@ static unsigned int sve_size_from_header(struct user_sve_header const *header) return ALIGN(header->size, SVE_VQ_BYTES); }

-static int sve_get(struct task_struct *target, - const struct user_regset *regset, - struct membuf to) +static int sve_get_common(struct task_struct *target, + const struct user_regset *regset, + struct membuf to, + enum vec_type type) { struct user_sve_header header; unsigned int vq; unsigned long start, end;

- if (!system_supports_sve()) - return -EINVAL; - /* Header */ - sve_init_header_from_task(&header, target); + sve_init_header_from_task(&header, target, type); vq = sve_vq_from_vl(header.vl);

membuf_write(&to, &header, sizeof(header)); @@ -759,49 +787,61 @@ static int sve_get(struct task_struct *target, if (target == current) fpsimd_preserve_current_state();

- /* Registers: FPSIMD-only case */ - BUILD_BUG_ON(SVE_PT_FPSIMD_OFFSET != sizeof(header)); - if ((header.flags & SVE_PT_REGS_MASK) == SVE_PT_REGS_FPSIMD) + BUILD_BUG_ON(SVE_PT_SVE_OFFSET != sizeof(header)); + + switch ((header.flags & SVE_PT_REGS_MASK)) { + case SVE_PT_REGS_FPSIMD: return __fpr_get(target, regset, to);

- /* Otherwise: full SVE case */ + case SVE_PT_REGS_SVE: + start = SVE_PT_SVE_OFFSET; + end = SVE_PT_SVE_FFR_OFFSET(vq) + SVE_PT_SVE_FFR_SIZE(vq); + membuf_write(&to, target->thread.sve_state, end - start);

- BUILD_BUG_ON(SVE_PT_SVE_OFFSET != sizeof(header)); - start = SVE_PT_SVE_OFFSET; - end = SVE_PT_SVE_FFR_OFFSET(vq) + SVE_PT_SVE_FFR_SIZE(vq); - membuf_write(&to, target->thread.sve_state, end - start); + start = end; + end = SVE_PT_SVE_FPSR_OFFSET(vq); + membuf_zero(&to, end - start);

- start = end; - end = SVE_PT_SVE_FPSR_OFFSET(vq); - membuf_zero(&to, end - start); + /* + * Copy fpsr, and fpcr which must follow contiguously in + * struct fpsimd_state: + */ + start = end; + end = SVE_PT_SVE_FPCR_OFFSET(vq) + SVE_PT_SVE_FPCR_SIZE; + membuf_write(&to, &target->thread.uw.fpsimd_state.fpsr, + end - start);

- /* - * Copy fpsr, and fpcr which must follow contiguously in - * struct fpsimd_state: - */ - start = end; - end = SVE_PT_SVE_FPCR_OFFSET(vq) + SVE_PT_SVE_FPCR_SIZE; - membuf_write(&to, &target->thread.uw.fpsimd_state.fpsr, end - start); + start = end; + end = sve_size_from_header(&header); + return membuf_zero(&to, end - start);

- start = end; - end = sve_size_from_header(&header); - return membuf_zero(&to, end - start); + default: + return 0; + } }

-static int sve_set(struct task_struct *target, +static int sve_get(struct task_struct *target, const struct user_regset *regset, - unsigned int pos, unsigned int count, - const void *kbuf, const void __user *ubuf) + struct membuf to) +{ + if (!system_supports_sve()) + return -EINVAL; + + return sve_get_common(target, regset, to, ARM64_VEC_SVE); +} + +static int sve_set_common(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf, + enum vec_type type) { int ret; struct user_sve_header header; unsigned int vq; unsigned long start, end;

- if (!system_supports_sve()) - return -EINVAL; - /* Header */ if (count < sizeof(header)) return -EINVAL; @@ -814,13 +854,37 @@ static int sve_set(struct task_struct *target, * Apart from SVE_PT_REGS_MASK, all SVE_PT_* flags are consumed by * vec_set_vector_length(), which will also validate them for us: */ - ret = vec_set_vector_length(target, ARM64_VEC_SVE, header.vl, + ret = vec_set_vector_length(target, type, header.vl, ((unsigned long)header.flags & ~SVE_PT_REGS_MASK) << 16); if (ret) goto out;

/* Actual VL set may be less than the user asked for: */ - vq = sve_vq_from_vl(task_get_sve_vl(target)); + vq = sve_vq_from_vl(task_get_vl(target, type)); + + /* Enter/exit streaming mode */ + if (system_supports_sme()) { + u64 old_svcr = target->thread.svcr; + + switch (type) { + case ARM64_VEC_SVE: + target->thread.svcr &= ~SYS_SVCR_EL0_SM_MASK; + break; + case ARM64_VEC_SME: + target->thread.svcr |= SYS_SVCR_EL0_SM_MASK; + break; + default: + WARN_ON_ONCE(1); + return -EINVAL; + } + + /* + * If we switched then invalidate any existing SVE + * state and ensure there's storage. + */ + if (target->thread.svcr != old_svcr) + sve_alloc(target); + }

/* Registers: FPSIMD-only case */

@@ -832,7 +896,10 @@ static int sve_set(struct task_struct *target, goto out; }

- /* Otherwise: full SVE case */ + /* + * Otherwise: no registers or full SVE case. For backwards + * compatibility reasons we treat empty flags as SVE registers. + */

/* * If setting a different VL from the requested VL and there is @@ -853,8 +920,9 @@ static int sve_set(struct task_struct *target,

/* * Ensure target->thread.sve_state is up to date with target's - * FPSIMD regs, so that a short copyin leaves trailing registers - * unmodified. + * FPSIMD regs, so that a short copyin leaves trailing + * registers unmodified. Always enable SVE even if going into + * streaming mode. */ fpsimd_sync_to_sve(target); set_tsk_thread_flag(target, TIF_SVE); @@ -890,8 +958,46 @@ static int sve_set(struct task_struct *target, return ret; }

+static int sve_set(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf) +{ + if (!system_supports_sve()) + return -EINVAL; + + return sve_set_common(target, regset, pos, count, kbuf, ubuf, + ARM64_VEC_SVE); +} + #endif /* CONFIG_ARM64_SVE */

+#ifdef CONFIG_ARM64_SME + +static int ssve_get(struct task_struct *target, + const struct user_regset *regset, + struct membuf to) +{ + if (!system_supports_sme()) + return -EINVAL; + + return sve_get_common(target, regset, to, ARM64_VEC_SME); +} + +static int ssve_set(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf) +{ + if (!system_supports_sme()) + return -EINVAL; + + return sve_set_common(target, regset, pos, count, kbuf, ubuf, + ARM64_VEC_SME); +} + +#endif /* CONFIG_ARM64_SME */ + #ifdef CONFIG_ARM64_PTR_AUTH static int pac_mask_get(struct task_struct *target, const struct user_regset *regset, @@ -1109,6 +1215,9 @@ enum aarch64_regset { #ifdef CONFIG_ARM64_SVE REGSET_SVE, #endif +#ifdef CONFIG_ARM64_SVE + REGSET_SSVE, +#endif #ifdef CONFIG_ARM64_PTR_AUTH REGSET_PAC_MASK, REGSET_PAC_ENABLED_KEYS, @@ -1189,6 +1298,17 @@ static const struct user_regset aarch64_regsets[] = { .set = sve_set, }, #endif +#ifdef CONFIG_ARM64_SME + [REGSET_SSVE] = { /* Streaming mode SVE */ + .core_note_type = NT_ARM_SSVE, + .n = DIV_ROUND_UP(SVE_PT_SIZE(SVE_VQ_MAX, SVE_PT_REGS_SVE), + SVE_VQ_BYTES), + .size = SVE_VQ_BYTES, + .align = SVE_VQ_BYTES, + .regset_get = ssve_get, + .set = ssve_set, + }, +#endif #ifdef CONFIG_ARM64_PTR_AUTH [REGSET_PAC_MASK] = { .core_note_type = NT_ARM_PAC_MASK, diff --git a/include/uapi/linux/elf.h b/include/uapi/linux/elf.h index 61bf4774b8f2..61502388683f 100644 --- a/include/uapi/linux/elf.h +++ b/include/uapi/linux/elf.h @@ -427,6 +427,7 @@ typedef struct elf64_shdr { #define NT_ARM_PACG_KEYS 0x408 /* ARM pointer authentication generic key */ #define NT_ARM_TAGGED_ADDR_CTRL 0x409 /* arm64 tagged address control (prctl()) */ #define NT_ARM_PAC_ENABLED_KEYS 0x40a /* arm64 ptr auth enabled keys (prctl()) */ +#define NT_ARM_SSVE 0x40b /* ARM Streaming SVE registers */ #define NT_ARC_V2 0x600 /* ARCv2 accumulator/extra registers */ #define NT_VMCOREDD 0x700 /* Vmcore Device Dump Note */ #define NT_MIPS_DSP 0x800 /* MIPS DSP ASE registers */

Both streaming mode and ZA may increase power consumption when they are enabled and streaming mode makes many FPSIMD and SVE instructions undefined which will cause problems for any kernel mode floating point so disable both when we flush the CPU state. This covers both kernel_neon_begin() and idle and after flushing the state a reload is always required anyway.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/fpsimd.h | 7 +++++++ arch/arm64/kernel/fpsimd.c | 9 +++++++++ 2 files changed, 16 insertions(+)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 180548c13940..f1cd552ebb81 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -300,6 +300,12 @@ static inline void sme_smstop_sm(void) asm volatile(".inst 0xd503427f"); }

+static inline void sme_smstop(void) +{ + /* SMSTOP SM is an alias for MSR SVCRSMZA, #0 */ + asm volatile(".inst 0x7f4603d5"); +} + static inline int sme_max_vl(void) { return vec_max_vl(ARM64_VEC_SME); @@ -322,6 +328,7 @@ static inline void sme_alloc(struct task_struct *task) { }

static inline void sme_smstart_sm(void) { } static inline void sme_smstop_sm(void) { } +static inline void sme_smstop(void) { }

static inline int sme_set_current_vl(unsigned long arg) { diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index af46db99ec82..cdbc1fd8fde2 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -1744,6 +1744,15 @@ static void fpsimd_flush_cpu_state(void) { WARN_ON(!system_supports_fpsimd()); __this_cpu_write(fpsimd_last_state.st, NULL); + + /* + * Leaving streaming mode enabled will cause issues for any kernel + * NEON and leaving streaming mode or ZA enabled may incrase power + * consumption. + */ + if (system_supports_sme()) + sme_smstop(); + set_thread_flag(TIF_FOREIGN_FPSTATE); }

Now that basline support for the Scalable Matrix Extension (SME) is present introduce the Kconfig option allowing it to be built. While there is no requirement for a system with SME to support SVE at runtime the support for streaming mode SVE is mostly shared with normal SVE so depend on SVE.

Since there is currently no support for KVM and no handling of either streaming mode or ZA with KVM the option the feature is marked as being incompatible with KVM and not enabled by default.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/Kconfig | 11 +++++++++++ 1 file changed, 11 insertions(+)

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig index 5c7ae4c3954b..f7004dd0a1b4 100644 --- a/arch/arm64/Kconfig +++ b/arch/arm64/Kconfig @@ -1719,6 +1719,17 @@ config ARM64_SVE booting the kernel. If unsure and you are not observing these symptoms, you should assume that it is safe to say Y.

+config ARM64_SME + bool "ARM Scalable Matrix Extension support" + depends on ARM64_SVE + depends on !KVM + help + The Scalable Matrix Extension (SME) is an extension to the AArch64 + execution state which utilises a substantial subset of the SVE + instruction set, together with the addition of new architectural + register state capable of holding two dimensional matrix tiles to + enable various matrix operations. + config ARM64_MODULE_PLTS bool "Use PLTs to allow module memory to spill over into vmalloc area" depends on MODULES

The Scalable Matrix Extension adds a new system register TPIDR2 intended to be used by libc for its own thread specific use, add some kselftests which exercise the ABI for it.

Since this test should with some adjustment work for TPIDR and any other similar registers added in future add tests for it in a separate directory rather than placing it with the other floating point tests, nothing existing looked suitable so I created a new test directory called "abi".

Since this feature is intended to be used by libc the test is built as freestanding code using nolibc so we don't end up with the test program and libc both trying to manage the register simultaneously and distrupting each other. As a result of being written using nolibc rather than using hwcaps to identify if SME is available in the system we check for the default SME vector length configuration in proc, adding hwcap support to nolibc seems like disproportionate effort and didn't feel entirely idiomatic for what nolibc is trying to do.

Signed-off-by: Mark Brown broonie@kernel.org --- tools/testing/selftests/arm64/Makefile | 2 +- tools/testing/selftests/arm64/abi/.gitignore | 1 + tools/testing/selftests/arm64/abi/Makefile | 13 + tools/testing/selftests/arm64/abi/tpidr2.c | 298 +++++++++++++++++++ 4 files changed, 313 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/arm64/abi/.gitignore create mode 100644 tools/testing/selftests/arm64/abi/Makefile create mode 100644 tools/testing/selftests/arm64/abi/tpidr2.c

diff --git a/tools/testing/selftests/arm64/Makefile b/tools/testing/selftests/arm64/Makefile index ced910fb4019..1e8d9a8f59df 100644 --- a/tools/testing/selftests/arm64/Makefile +++ b/tools/testing/selftests/arm64/Makefile @@ -4,7 +4,7 @@ ARCH ?= $(shell uname -m 2>/dev/null || echo not)

ifneq (,$(filter $(ARCH),aarch64 arm64)) -ARM64_SUBTARGETS ?= tags signal pauth fp mte bti +ARM64_SUBTARGETS ?= tags signal pauth fp mte bti abi else ARM64_SUBTARGETS := endif diff --git a/tools/testing/selftests/arm64/abi/.gitignore b/tools/testing/selftests/arm64/abi/.gitignore new file mode 100644 index 000000000000..4b04670993ea --- /dev/null +++ b/tools/testing/selftests/arm64/abi/.gitignore @@ -0,0 +1 @@ +tpidr2 diff --git a/tools/testing/selftests/arm64/abi/Makefile b/tools/testing/selftests/arm64/abi/Makefile new file mode 100644 index 000000000000..59773689c4dd --- /dev/null +++ b/tools/testing/selftests/arm64/abi/Makefile @@ -0,0 +1,13 @@ +# SPDX-License-Identifier: GPL-2.0 +# Copyright (C) 2021 ARM Limited + +TEST_GEN_PROGS := tpidr2 + +include ../../lib.mk + +# Build with nolibc since TPIDR2 is intended to be actively managed by +# libc and we're trying to test the functionality that it depends on here. +$(OUTPUT)/tpidr2: tpidr2.c + $(CC) -fno-asynchronous-unwind-tables -fno-ident -s -Os -nostdlib \ + -static -include ../../../../include/nolibc/nolibc.h \ + -ffreestanding -Wall $^ -o $@ -lgcc diff --git a/tools/testing/selftests/arm64/abi/tpidr2.c b/tools/testing/selftests/arm64/abi/tpidr2.c new file mode 100644 index 000000000000..351a098b503a --- /dev/null +++ b/tools/testing/selftests/arm64/abi/tpidr2.c @@ -0,0 +1,298 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include <linux/sched.h> +#include <linux/wait.h> + +#define SYS_TPIDR2 "S3_3_C13_C0_5" + +#define EXPECTED_TESTS 5 + +static void putstr(const char *str) +{ + write(1, str, strlen(str)); +} + +static void putnum(unsigned int num) +{ + char c; + + if (num / 10) + putnum(num / 10); + + c = '0' + (num % 10); + write(1, &c, 1); +} + +static int tests_run; +static int tests_passed; +static int tests_failed; +static int tests_skipped; + +static void set_tpidr2(uint64_t val) +{ + asm volatile ( + "msr " SYS_TPIDR2 ", %0\n" + : + : "r"(val) + : "cc"); +} + +static uint64_t get_tpidr2(void) +{ + uint64_t val; + + asm volatile ( + "mrs %0, " SYS_TPIDR2 "\n" + : "=r"(val) + : + : "cc"); + + return val; +} + +static void print_summary(void) +{ + if (tests_passed + tests_failed + tests_skipped != EXPECTED_TESTS) + putstr("# UNEXPECTED TEST COUNT: "); + + putstr("# Totals: pass:"); + putnum(tests_passed); + putstr(" fail:"); + putnum(tests_failed); + putstr(" xfail:0 xpass:0 skip:"); + putnum(tests_skipped); + putstr(" error:0\n"); +} + +/* Processes should start with TPIDR2 == 0 */ +static int default_value(void) +{ + return get_tpidr2() == 0; +} + +/* If we set TPIDR2 we should read that value */ +static int write_read(void) +{ + set_tpidr2(getpid()); + + return getpid() == get_tpidr2(); +} + +/* If we set a value we should read the same value after scheduling out */ +static int write_sleep_read(void) +{ + set_tpidr2(getpid()); + + msleep(100); + + return getpid() == get_tpidr2(); +} + +/* + * If we fork the value in the parent should be unchanged and the + * child should start with the same value and be able to set its own + * value. + */ +static int write_fork_read(void) +{ + pid_t newpid, waiting, oldpid; + int status; + + set_tpidr2(getpid()); + + oldpid = getpid(); + newpid = fork(); + if (newpid == 0) { + /* In child */ + if (get_tpidr2() != oldpid) { + putstr("# TPIDR2 changed in child: "); + putnum(get_tpidr2()); + putstr("\n"); + exit(0); + } + + set_tpidr2(getpid()); + if (get_tpidr2() == getpid()) { + exit(1); + } else { + putstr("# Failed to set TPIDR2 in child\n"); + exit(0); + } + } + if (newpid < 0) { + putstr("# fork() failed: -"); + putnum(-newpid); + putstr("\n"); + return 0; + } + + for (;;) { + waiting = waitpid(newpid, &status, 0); + + if (waiting < 0) { + if (errno == EINTR) + continue; + putstr("# waitpid() failed: "); + putnum(errno); + putstr("\n"); + return 0; + } + if (waiting != newpid) { + putstr("# waitpid() returned wrong PID\n"); + return 0; + } + + if (!WIFEXITED(status)) { + putstr("# child did not exit\n"); + return 0; + } + + if (getpid() != get_tpidr2()) { + putstr("# TPIDR2 corrupted in parent\n"); + return 0; + } + + return WEXITSTATUS(status); + } +} + +/* + * sys_clone() has a lot of per architecture variation so just define + * it here rather than adding it to nolibc, plus the raw API is a + * little more convenient for this test. + */ +static int sys_clone(unsigned long clone_flags, unsigned long newsp, + int *parent_tidptr, unsigned long tls, + int *child_tidptr) +{ + return my_syscall5(__NR_clone, clone_flags, newsp, parent_tidptr, tls, + child_tidptr); +} + +/* + * If we clone with CLONE_SETTLS then the value in the parent should + * be unchanged and the child should start with zero and be able to + * set its own value. + */ +static int write_clone_read(void) +{ + int parent_tid, child_tid; + pid_t parent, waiting; + int ret, status; + + parent = getpid(); + set_tpidr2(parent); + + ret = sys_clone(CLONE_SETTLS, 0, &parent_tid, 0, &child_tid); + if (ret == -1) { + putstr("# clone() failed\n"); + putnum(errno); + putstr("\n"); + return 0; + } + + if (ret == 0) { + /* In child */ + if (get_tpidr2() != 0) { + putstr("# TPIDR2 non-zero in child: "); + putnum(get_tpidr2()); + putstr("\n"); + exit(0); + } + + if (gettid() == 0) + putstr("# Child TID==0\n"); + set_tpidr2(gettid()); + if (get_tpidr2() == gettid()) { + exit(1); + } else { + putstr("# Failed to set TPIDR2 in child\n"); + exit(0); + } + } + + for (;;) { + waiting = wait4(ret, &status, __WCLONE, NULL); + + if (waiting < 0) { + if (errno == EINTR) + continue; + putstr("# wait4() failed: "); + putnum(errno); + putstr("\n"); + return 0; + } + if (waiting != ret) { + putstr("# wait4() returned wrong PID "); + putnum(waiting); + putstr("\n"); + return 0; + } + + if (!WIFEXITED(status)) { + putstr("# child did not exit\n"); + return 0; + } + + if (parent != get_tpidr2()) { + putstr("# TPIDR2 corrupted in parent\n"); + return 0; + } + + return WEXITSTATUS(status); + } +} + +#define run_test(name) \ + if (name()) { \ + tests_passed++; \ + } else { \ + tests_failed++; \ + putstr("not "); \ + } \ + putstr("ok "); \ + putnum(++tests_run); \ + putstr(" " #name "\n"); + +int main(int argc, char **argv) +{ + int ret, i; + + putstr("TAP version 13\n"); + putstr("1.."); + putnum(EXPECTED_TESTS); + putstr("\n"); + + putstr("# PID: "); + putnum(getpid()); + putstr("\n"); + + /* + * This test is run with nolibc which doesn't support hwcap and + * it's probably disproportionate to implement so instead check + * for the default vector length configuration in /proc. + */ + ret = open("/proc/sys/abi/sme_default_vector_length", O_RDONLY, 0); + if (ret >= 0) { + run_test(default_value); + run_test(write_read); + run_test(write_sleep_read); + run_test(write_fork_read); + run_test(write_clone_read); + + } else { + putstr("# SME support not present\n"); + + for (i = 0; i < EXPECTED_TESTS; i++) { + putstr("ok "); + putnum(i); + putstr(" skipped, TPIDR2 not supported\n"); + } + + tests_skipped += EXPECTED_TESTS; + } + + print_summary(); + + return 0; +}

One of the features of SME is the addition of streaming mode, in which we have access to a set of streaming mode SVE registers at the SME vector length. Since these are accessed using the SVE instructions let's reuse the existing SVE stress test for testing with a compile time option for controlling the few small differences needed:

- Enter streaming mode immediately on starting the program. - In streaming mode FFR is removed so skip reading and writing FFR.

In order to avoid requiring a cutting edge toolchain with SME support use the op/CR form for specifying SVCR.

Signed-off-by: Mark Brown broonie@kernel.org --- tools/testing/selftests/arm64/fp/.gitignore | 1 + tools/testing/selftests/arm64/fp/Makefile | 3 + tools/testing/selftests/arm64/fp/ssve-stress | 59 ++++++++++++++++++++ tools/testing/selftests/arm64/fp/sve-test.S | 30 ++++++++++ 4 files changed, 93 insertions(+) create mode 100644 tools/testing/selftests/arm64/fp/ssve-stress

diff --git a/tools/testing/selftests/arm64/fp/.gitignore b/tools/testing/selftests/arm64/fp/.gitignore index 885dd592807b..73c600e1ab81 100644 --- a/tools/testing/selftests/arm64/fp/.gitignore +++ b/tools/testing/selftests/arm64/fp/.gitignore @@ -4,5 +4,6 @@ rdvl-sve sve-probe-vls sve-ptrace sve-test +ssve-test vec-syscfg vlset diff --git a/tools/testing/selftests/arm64/fp/Makefile b/tools/testing/selftests/arm64/fp/Makefile index 11d4fc3b7115..6d9e4d1922e4 100644 --- a/tools/testing/selftests/arm64/fp/Makefile +++ b/tools/testing/selftests/arm64/fp/Makefile @@ -5,6 +5,7 @@ TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg TEST_PROGS_EXTENDED := fpsimd-test fpsimd-stress \ rdvl-sme rdvl-sve \ sve-test sve-stress \ + ssve-test ssve-stress \ vlset

all: $(TEST_GEN_PROGS) $(TEST_PROGS_EXTENDED) @@ -17,6 +18,8 @@ sve-ptrace: sve-ptrace.o sve-probe-vls: sve-probe-vls.o rdvl.o sve-test: sve-test.o asm-utils.o $(CC) -nostdlib $^ -o $@ +ssve-test: sve-test.S asm-utils.o + $(CC) -DSSVE -nostdlib $^ -o $@ vec-syscfg: vec-syscfg.o rdvl.o vlset: vlset.o

diff --git a/tools/testing/selftests/arm64/fp/ssve-stress b/tools/testing/selftests/arm64/fp/ssve-stress new file mode 100644 index 000000000000..e2bd2cc184ad --- /dev/null +++ b/tools/testing/selftests/arm64/fp/ssve-stress @@ -0,0 +1,59 @@ +#!/bin/bash +# SPDX-License-Identifier: GPL-2.0-only +# Copyright (C) 2015-2019 ARM Limited. +# Original author: Dave Martin Dave.Martin@arm.com + +set -ue + +NR_CPUS=`nproc` + +pids= +logs= + +cleanup () { + trap - INT TERM CHLD + set +e + + if [ -n "$pids" ]; then + kill $pids + wait $pids + pids= + fi + + if [ -n "$logs" ]; then + cat $logs + rm $logs + logs= + fi +} + +interrupt () { + cleanup + exit 0 +} + +child_died () { + cleanup + exit 1 +} + +trap interrupt INT TERM EXIT + +for x in `seq 0 $((NR_CPUS * 4))`; do + log=`mktemp` + logs=$logs\ $log + ./ssve-test >$log & + pids=$pids\ $! +done + +# Wait for all child processes to be created: +sleep 10 + +while :; do + kill -USR1 $pids +done & +pids=$pids\ $! + +wait + +exit 1 diff --git a/tools/testing/selftests/arm64/fp/sve-test.S b/tools/testing/selftests/arm64/fp/sve-test.S index f5b1b48ffff2..31764e8370db 100644 --- a/tools/testing/selftests/arm64/fp/sve-test.S +++ b/tools/testing/selftests/arm64/fp/sve-test.S @@ -156,6 +156,7 @@ endfunction // We fill the upper lanes of FFR with zeros. // Beware: corrupts P0. function setup_ffr +#ifndef SSVE mov x4, x30

and w0, w0, #0x3 @@ -178,6 +179,9 @@ function setup_ffr wrffr p0.b

ret x4 +#else + ret +#endif endfunction

// Trivial memory compare: compare x2 bytes starting at address x0 with @@ -260,6 +264,7 @@ endfunction // Beware -- corrupts P0. // Clobbers x0-x5. function check_ffr +#ifndef SSVE mov x3, x30

ldr x4, =scratch @@ -280,6 +285,9 @@ function check_ffr mov x2, x5 mov x30, x3 b memcmp +#else + ret +#endif endfunction

// Any SVE register modified here can cause corruption in the main @@ -295,13 +303,26 @@ function irritator_handler movi v0.8b, #1 movi v9.16b, #2 movi v31.8b, #3 +#ifndef SSVE // And P0 rdffr p0.b // And FFR wrffr p15.b +#endif + + ret +endfunction + +#ifdef SSVE +function enable_sm + // Set SVCR.SM to 1, equivalent to SMSTART SM but doesn't need a + // SME capable toolchain. + mov x0, #1 + msr S3_3_C4_C2_2, x0

ret endfunction +#endif

function terminate_handler mov w21, w0 @@ -359,6 +380,11 @@ endfunction .globl _start function _start _start: +#ifdef SSVE + puts "Streaming mode " + bl enable_sm +#endif + // Sanity-check and report the vector length

rdvl x19, #8 @@ -407,6 +433,10 @@ _start: orr w2, w2, #SA_NODEFER bl setsignal

+#ifdef SSVE + bl enable_sm // syscalls will have exited streaming mode +#endif + mov x22, #0 // generation number, increments per iteration .Ltest_loop: rdvl x0, #8

Add test cases for the SME signal handing ABI patterned off the SVE tests. Due to the small size of the tests and the differences in ABI (especially around needing to account for both streaming SVE and ZA) there is some code duplication here.

We currently cover: - Reporting of the vector length. - Lack of support for changing vector length. - Presence and size of register state for streaming SVE and ZA.

As with the SVE tests we do not yet have any validation of register contents.

Signed-off-by: Mark Brown broonie@kernel.org --- .../testing/selftests/arm64/signal/.gitignore | 2 + .../selftests/arm64/signal/test_signals.h | 2 + .../arm64/signal/test_signals_utils.c | 3 + .../testcases/fake_sigreturn_sme_change_vl.c | 92 +++++++++++++ .../arm64/signal/testcases/sme_trap_za.c | 36 +++++ .../selftests/arm64/signal/testcases/sme_vl.c | 70 ++++++++++ .../arm64/signal/testcases/ssve_regs.c | 129 ++++++++++++++++++ 7 files changed, 334 insertions(+) create mode 100644 tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_sme_change_vl.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/sme_trap_za.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/sme_vl.c create mode 100644 tools/testing/selftests/arm64/signal/testcases/ssve_regs.c

diff --git a/tools/testing/selftests/arm64/signal/.gitignore b/tools/testing/selftests/arm64/signal/.gitignore index c1742755abb9..4de8eb26d4de 100644 --- a/tools/testing/selftests/arm64/signal/.gitignore +++ b/tools/testing/selftests/arm64/signal/.gitignore @@ -1,5 +1,7 @@ # SPDX-License-Identifier: GPL-2.0-only mangle_* fake_sigreturn_* +sme_* +ssve_* sve_* !*.[ch] diff --git a/tools/testing/selftests/arm64/signal/test_signals.h b/tools/testing/selftests/arm64/signal/test_signals.h index ebe8694dbef0..d0523a50ee78 100644 --- a/tools/testing/selftests/arm64/signal/test_signals.h +++ b/tools/testing/selftests/arm64/signal/test_signals.h @@ -34,11 +34,13 @@ enum { FSSBS_BIT, FSVE_BIT, + FSME_BIT, FMAX_END };

#define FEAT_SSBS (1UL << FSSBS_BIT) #define FEAT_SVE (1UL << FSVE_BIT) +#define FEAT_SME (1UL << FSME_BIT)

/* * A descriptor used to describe and configure a test case. diff --git a/tools/testing/selftests/arm64/signal/test_signals_utils.c b/tools/testing/selftests/arm64/signal/test_signals_utils.c index 8bb12be87a51..cfb95010791a 100644 --- a/tools/testing/selftests/arm64/signal/test_signals_utils.c +++ b/tools/testing/selftests/arm64/signal/test_signals_utils.c @@ -27,6 +27,7 @@ static int sig_copyctx = SIGTRAP; static char const *const feats_names[FMAX_END] = { " SSBS ", " SVE ", + " SME ", };

#define MAX_FEATS_SZ 128 @@ -266,6 +267,8 @@ int test_init(struct tdescr *td) td->feats_supported |= FEAT_SSBS; if (getauxval(AT_HWCAP) & HWCAP_SVE) td->feats_supported |= FEAT_SVE; + if (getauxval(AT_HWCAP2) & HWCAP2_SME) + td->feats_supported |= FEAT_SME; if (feats_ok(td)) { fprintf(stderr, "Required Features: [%s] supported\n", diff --git a/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_sme_change_vl.c b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_sme_change_vl.c new file mode 100644 index 000000000000..7ed762b7202f --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/fake_sigreturn_sme_change_vl.c @@ -0,0 +1,92 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Attempt to change the streaming SVE vector length in a signal + * handler, this is not supported and is expected to segfault. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +struct fake_sigframe sf; +static unsigned int vls[SVE_VQ_MAX]; +unsigned int nvls = 0; + +static bool sme_get_vls(struct tdescr *td) +{ + int vq, vl; + + /* + * Enumerate up to SVE_VQ_MAX vector lengths + */ + for (vq = SVE_VQ_MAX; vq > 0; --vq) { + vl = prctl(PR_SVE_SET_VL, vq * 16); + if (vl == -1) + return false; + + vl &= PR_SME_VL_LEN_MASK; + + /* Skip missing VLs */ + vq = sve_vq_from_vl(vl); + + vls[nvls++] = vl; + } + + /* We need at least two VLs */ + if (nvls < 2) { + fprintf(stderr, "Only %d VL supported\n", nvls); + return false; + } + + return true; +} + +static int fake_sigreturn_ssve_change_vl(struct tdescr *td, + siginfo_t *si, ucontext_t *uc) +{ + size_t resv_sz, offset; + struct _aarch64_ctx *head = GET_SF_RESV_HEAD(sf); + struct sve_context *sve; + + /* Get a signal context with a SME ZA frame in it */ + if (!get_current_context(td, &sf.uc)) + return 1; + + resv_sz = GET_SF_RESV_SIZE(sf); + head = get_header(head, SVE_MAGIC, resv_sz, &offset); + if (!head) { + fprintf(stderr, "No SVE context\n"); + return 1; + } + + if (head->size != sizeof(struct sve_context)) { + fprintf(stderr, "Register data present, aborting\n"); + return 1; + } + + sve = (struct sve_context *)head; + + /* No changes are supported; init left us at minimum VL so go to max */ + fprintf(stderr, "Attempting to change VL from %d to %d\n", + sve->vl, vls[0]); + sve->vl = vls[0]; + + fake_sigreturn(&sf, sizeof(sf), 0); + + return 1; +} + +struct tdescr tde = { + .name = "FAKE_SIGRETURN_SSVE_CHANGE", + .descr = "Attempt to change Streaming SVE VL", + .feats_required = FEAT_SME, + .sig_ok = SIGSEGV, + .timeout = 3, + .init = sme_get_vls, + .run = fake_sigreturn_ssve_change_vl, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/sme_trap_za.c b/tools/testing/selftests/arm64/signal/testcases/sme_trap_za.c new file mode 100644 index 000000000000..3a7747af4715 --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/sme_trap_za.c @@ -0,0 +1,36 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Verify that accessing ZA without enabling it generates a SIGILL. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +int sme_trap_za_trigger(struct tdescr *td) +{ + /* ZERO ZA */ + asm volatile(".inst 0xc00800ff"); + + return 0; +} + +int sme_trap_za_run(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + return 1; +} + +struct tdescr tde = { + .name = "SME ZA trap", + .descr = "Check that we get a SIGILL if we access ZA without enabling", + .timeout = 3, + .sanity_disabled = true, + .trigger = sme_trap_za_trigger, + .run = sme_trap_za_run, + .sig_ok = SIGILL, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/sme_vl.c b/tools/testing/selftests/arm64/signal/testcases/sme_vl.c new file mode 100644 index 000000000000..c40e339a1bc3 --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/sme_vl.c @@ -0,0 +1,70 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Check that the SME vector length reported in signal contexts is the + * expected one. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +struct fake_sigframe sf; +unsigned int vl; + +static bool get_sme_vl(struct tdescr *td) +{ + int ret = prctl(PR_SME_GET_VL); + if (ret == -1) + return false; + + vl = ret; + + return true; +} + +static int sme_vl(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + size_t resv_sz, offset; + struct _aarch64_ctx *head = GET_SF_RESV_HEAD(sf); + struct sve_context *sve; + + /* Get a signal context which should have a SVE frame in it */ + if (!get_current_context(td, &sf.uc)) + return 1; + + resv_sz = GET_SF_RESV_SIZE(sf); + head = get_header(head, SVE_MAGIC, resv_sz, &offset); + if (!head) { + fprintf(stderr, "No SVE context\n"); + return 1; + } + sve = (struct sve_context *)head; + + if (sve->vl != vl) { + fprintf(stderr, "SSVE sigframe VL %u, expected %u\n", + sve->vl, vl); + return 1; + } else { + fprintf(stderr, "got SSVE expected VL %u\n", vl); + } + + /* Also check ZA VL */ + + td->pass = 1; + + return 0; +} + +struct tdescr tde = { + .name = "SME VL", + .descr = "Check that we get the right SME VL reported", + .feats_required = FEAT_SME, + .timeout = 3, + .init = get_sme_vl, + .run = sme_vl, +}; diff --git a/tools/testing/selftests/arm64/signal/testcases/ssve_regs.c b/tools/testing/selftests/arm64/signal/testcases/ssve_regs.c new file mode 100644 index 000000000000..44a08d43cd50 --- /dev/null +++ b/tools/testing/selftests/arm64/signal/testcases/ssve_regs.c @@ -0,0 +1,129 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2021 ARM Limited + * + * Verify that the streaming SVE register context in signal frames is + * set up as expected. + */ + +#include <signal.h> +#include <ucontext.h> +#include <sys/prctl.h> + +#include "test_signals_utils.h" +#include "testcases.h" + +struct fake_sigframe sf; +static unsigned int vls[SVE_VQ_MAX]; +unsigned int nvls = 0; + +static bool sme_get_vls(struct tdescr *td) +{ + int vq, vl; + + /* + * Enumerate up to SVE_VQ_MAX vector lengths + */ + for (vq = SVE_VQ_MAX; vq > 0; --vq) { + vl = prctl(PR_SVE_SET_VL, vq * 16); + if (vl == -1) + return false; + + vl &= PR_SME_VL_LEN_MASK; + + /* Skip missing VLs */ + vq = sve_vq_from_vl(vl); + + vls[nvls++] = vl; + } + + /* We need at least one VL */ + if (nvls < 1) { + fprintf(stderr, "Only %d VL supported\n", nvls); + return false; + } + + return true; +} + +static void setup_ssve_regs(void) +{ + /* SMSTART SM */ + asm volatile(".inst 0x7f4303d5"); + + /* RDVL x16, #1 so we should have SVE regs; real data is TODO */ + asm volatile(".inst 0x04bf5030" : : : "x16" ); +} + +static int do_one_sme_vl(struct tdescr *td, siginfo_t *si, ucontext_t *uc, + unsigned int vl) +{ + size_t resv_sz, offset; + struct _aarch64_ctx *head = GET_SF_RESV_HEAD(sf); + struct sve_context *ssve; + + fprintf(stderr, "Testing VL %d\n", vl); + + if (prctl(PR_SME_SET_VL, vl) == -1) { + fprintf(stderr, "Failed to set VL\n"); + return 1; + } + + /* + * Get a signal context which should have a SVE frame and registers + * in it. + */ + setup_ssve_regs(); + if (!get_current_context(td, &sf.uc)) + return 1; + + resv_sz = GET_SF_RESV_SIZE(sf); + head = get_header(head, SVE_MAGIC, resv_sz, &offset); + if (!head) { + fprintf(stderr, "No SVE context\n"); + return 1; + } + + ssve = (struct sve_context *)head; + if (ssve->vl != vl) { + fprintf(stderr, "Got VL %d, expected %d\n", ssve->vl, vl); + return 1; + } + + /* The actual size validation is done in get_current_context() */ + fprintf(stderr, "Got expected size %u and VL %d\n", + head->size, ssve->vl); + + return 0; +} + +static int sme_regs(struct tdescr *td, siginfo_t *si, ucontext_t *uc) +{ + int i; + + for (i = 0; i < nvls; i++) { + /* + * TODO: the signal test helpers can't currently cope + * with signal frames bigger than struct sigcontext, + * skip VLs that will trigger that. + */ + if (vls[i] > 64) + continue; + + if (do_one_sme_vl(td, si, uc, vls[i])) + return 1; + } + + td->pass = 1; + + return 0; +} + +struct tdescr tde = { + .name = "Streaming SVE registers", + .descr = "Check that we get the right Streaming SVE registers reported", + .feats_required = FEAT_SME, + .timeout = 3, + .init = sme_get_vls, + .run = sme_regs, +};

Add some basic coverage for the ZA ptrace interface, including walking through all the vector lengths supported in the system. As with SVE we do not currently validate the data all the way through to the running process.

Signed-off-by: Mark Brown broonie@kernel.org --- tools/testing/selftests/arm64/fp/.gitignore | 1 + tools/testing/selftests/arm64/fp/Makefile | 3 +- tools/testing/selftests/arm64/fp/za-ptrace.c | 353 +++++++++++++++++++ 3 files changed, 356 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/arm64/fp/za-ptrace.c

diff --git a/tools/testing/selftests/arm64/fp/.gitignore b/tools/testing/selftests/arm64/fp/.gitignore index 1178fecc7aa1..59afc01f2019 100644 --- a/tools/testing/selftests/arm64/fp/.gitignore +++ b/tools/testing/selftests/arm64/fp/.gitignore @@ -7,4 +7,5 @@ sve-test ssve-test vec-syscfg vlset +za-ptrace za-test diff --git a/tools/testing/selftests/arm64/fp/Makefile b/tools/testing/selftests/arm64/fp/Makefile index d77e9903116b..d7fb200d0794 100644 --- a/tools/testing/selftests/arm64/fp/Makefile +++ b/tools/testing/selftests/arm64/fp/Makefile @@ -1,7 +1,7 @@ # SPDX-License-Identifier: GPL-2.0

CFLAGS += -I../../../../../usr/include/ -TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg +TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg za-ptrace TEST_PROGS_EXTENDED := fpsimd-test fpsimd-stress \ rdvl-sme rdvl-sve \ sve-test sve-stress \ @@ -25,5 +25,6 @@ vec-syscfg: vec-syscfg.o rdvl.o vlset: vlset.o za-test: za-test.o asm-utils.o $(CC) -nostdlib $^ -o $@ +za-ptrace: za-ptrace.o

include ../../lib.mk diff --git a/tools/testing/selftests/arm64/fp/za-ptrace.c b/tools/testing/selftests/arm64/fp/za-ptrace.c new file mode 100644 index 000000000000..6c172a5629dc --- /dev/null +++ b/tools/testing/selftests/arm64/fp/za-ptrace.c @@ -0,0 +1,353 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Copyright (C) 2021 ARM Limited. + */ +#include <errno.h> +#include <stdbool.h> +#include <stddef.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <unistd.h> +#include <sys/auxv.h> +#include <sys/prctl.h> +#include <sys/ptrace.h> +#include <sys/types.h> +#include <sys/uio.h> +#include <sys/wait.h> +#include <asm/sigcontext.h> +#include <asm/ptrace.h> + +#include "../../kselftest.h" + +#define ARRAY_SIZE(a) (sizeof(a) / sizeof(a[0])) + +/* <linux/elf.h> and <sys/auxv.h> don't like each other, so: */ +#ifndef NT_ARM_ZA +#define NT_ARM_ZA 0x40c +#endif + +#define EXPECTED_TESTS (((SVE_VQ_MAX - SVE_VQ_MIN) + 1) * 3) + +static void fill_buf(char *buf, size_t size) +{ + int i; + + for (i = 0; i < size; i++) + buf[i] = random(); +} + +static int do_child(void) +{ + if (ptrace(PTRACE_TRACEME, -1, NULL, NULL)) + ksft_exit_fail_msg("PTRACE_TRACEME", strerror(errno)); + + if (raise(SIGSTOP)) + ksft_exit_fail_msg("raise(SIGSTOP)", strerror(errno)); + + return EXIT_SUCCESS; +} + +static struct user_za_header *get_za(pid_t pid, void **buf, size_t *size) +{ + struct user_za_header *za; + void *p; + size_t sz = sizeof(*za); + struct iovec iov; + + while (1) { + if (*size < sz) { + p = realloc(*buf, sz); + if (!p) { + errno = ENOMEM; + goto error; + } + + *buf = p; + *size = sz; + } + + iov.iov_base = *buf; + iov.iov_len = sz; + if (ptrace(PTRACE_GETREGSET, pid, NT_ARM_ZA, &iov)) + goto error; + + za = *buf; + if (za->size <= sz) + break; + + sz = za->size; + } + + return za; + +error: + return NULL; +} + +static int set_za(pid_t pid, const struct user_za_header *za) +{ + struct iovec iov; + + iov.iov_base = (void *)za; + iov.iov_len = za->size; + return ptrace(PTRACE_SETREGSET, pid, NT_ARM_ZA, &iov); +} + +/* Validate attempting to set the specfied VL via ptrace */ +static void ptrace_set_get_vl(pid_t child, unsigned int vl, bool *supported) +{ + struct user_za_header za; + struct user_za_header *new_za = NULL; + size_t new_za_size = 0; + int ret, prctl_vl; + + *supported = false; + + /* Check if the VL is supported in this process */ + prctl_vl = prctl(PR_SME_SET_VL, vl); + if (prctl_vl == -1) + ksft_exit_fail_msg("prctl(PR_SME_SET_VL) failed: %s (%d)\n", + strerror(errno), errno); + + /* If the VL is not supported then a supported VL will be returned */ + *supported = (prctl_vl == vl); + + /* Set the VL by doing a set with no register payload */ + memset(&za, 0, sizeof(za)); + za.size = sizeof(za); + za.vl = vl; + ret = set_za(child, &za); + if (ret != 0) { + ksft_test_result_fail("Failed to set VL %u\n", vl); + return; + } + + /* + * Read back the new register state and verify that we have the + * same VL that we got from prctl() on ourselves. + */ + if (!get_za(child, (void **)&new_za, &new_za_size)) { + ksft_test_result_fail("Failed to read VL %u\n", vl); + return; + } + + ksft_test_result(new_za->vl = prctl_vl, "Set VL %u\n", vl); + + free(new_za); +} + +/* Validate attempting to set no ZA data and read it back */ +static void ptrace_set_no_data(pid_t child, unsigned int vl) +{ + void *read_buf = NULL; + struct user_za_header write_za; + struct user_za_header *read_za; + size_t read_za_size = 0; + int ret; + + /* Set up some data and write it out */ + memset(&write_za, 0, sizeof(write_za)); + write_za.size = ZA_PT_ZA_OFFSET; + write_za.vl = vl; + + ret = set_za(child, &write_za); + if (ret != 0) { + ksft_test_result_fail("Failed to set VL %u no data\n", vl); + return; + } + + /* Read the data back */ + if (!get_za(child, (void **)&read_buf, &read_za_size)) { + ksft_test_result_fail("Failed to read VL %u no data\n", vl); + return; + } + read_za = read_buf; + + /* We might read more data if there's extensions we don't know */ + if (read_za->size < write_za.size) { + ksft_test_result_fail("VL %u wrote %d bytes, only read %d\n", + vl, write_za.size, read_za->size); + goto out_read; + } + + ksft_test_result(read_za->size == write_za.size, + "Disabled ZA for VL %u\n", vl); + +out_read: + free(read_buf); +} + +/* Validate attempting to set data and read it back */ +static void ptrace_set_get_data(pid_t child, unsigned int vl) +{ + void *write_buf; + void *read_buf = NULL; + struct user_za_header *write_za; + struct user_za_header *read_za; + size_t read_za_size = 0; + unsigned int vq = sve_vq_from_vl(vl); + int ret; + size_t data_size; + + data_size = ZA_PT_SIZE(vq); + write_buf = malloc(data_size); + if (!write_buf) { + ksft_test_result_fail("Error allocating %d byte buffer for VL %u\n", + data_size, vl); + return; + } + write_za = write_buf; + + /* Set up some data and write it out */ + memset(write_za, 0, data_size); + write_za->size = data_size; + write_za->vl = vl; + + fill_buf(write_buf + ZA_PT_ZA_OFFSET, ZA_PT_ZA_SIZE(vq)); + + ret = set_za(child, write_za); + if (ret != 0) { + ksft_test_result_fail("Failed to set VL %u data\n", vl); + goto out; + } + + /* Read the data back */ + if (!get_za(child, (void **)&read_buf, &read_za_size)) { + ksft_test_result_fail("Failed to read VL %u data\n", vl); + goto out; + } + read_za = read_buf; + + /* We might read more data if there's extensions we don't know */ + if (read_za->size < write_za->size) { + ksft_test_result_fail("VL %u wrote %d bytes, only read %d\n", + vl, write_za->size, read_za->size); + goto out_read; + } + + ksft_test_result(memcmp(write_buf + ZA_PT_ZA_OFFSET, + read_buf + ZA_PT_ZA_OFFSET, + ZA_PT_ZA_SIZE(vq)) == 0, + "Data match for VL %u\n", vl); + +out_read: + free(read_buf); +out: + free(write_buf); +} + +static int do_parent(pid_t child) +{ + int ret = EXIT_FAILURE; + pid_t pid; + int status; + siginfo_t si; + unsigned int vq, vl; + bool vl_supported; + + /* Attach to the child */ + while (1) { + int sig; + + pid = wait(&status); + if (pid == -1) { + perror("wait"); + goto error; + } + + /* + * This should never happen but it's hard to flag in + * the framework. + */ + if (pid != child) + continue; + + if (WIFEXITED(status) || WIFSIGNALED(status)) + ksft_exit_fail_msg("Child died unexpectedly\n"); + + if (!WIFSTOPPED(status)) + goto error; + + sig = WSTOPSIG(status); + + if (ptrace(PTRACE_GETSIGINFO, pid, NULL, &si)) { + if (errno == ESRCH) + goto disappeared; + + if (errno == EINVAL) { + sig = 0; /* bust group-stop */ + goto cont; + } + + ksft_test_result_fail("PTRACE_GETSIGINFO: %s\n", + strerror(errno)); + goto error; + } + + if (sig == SIGSTOP && si.si_code == SI_TKILL && + si.si_pid == pid) + break; + + cont: + if (ptrace(PTRACE_CONT, pid, NULL, sig)) { + if (errno == ESRCH) + goto disappeared; + + ksft_test_result_fail("PTRACE_CONT: %s\n", + strerror(errno)); + goto error; + } + } + + /* Step through every possible VQ */ + for (vq = SVE_VQ_MIN; vq <= SVE_VQ_MAX; vq++) { + vl = sve_vl_from_vq(vq); + + /* First, try to set this vector length */ + ptrace_set_get_vl(child, vl, &vl_supported); + + /* If the VL is supported validate data set/get */ + if (vl_supported) { + ptrace_set_no_data(child, vl); + ptrace_set_get_data(child, vl); + } else { + ksft_test_result_skip("Disabled ZA for VL %u\n", vl); + ksft_test_result_skip("Get and set data for VL %u\n", + vl); + } + } + + ret = EXIT_SUCCESS; + +error: + kill(child, SIGKILL); + +disappeared: + return ret; +} + +int main(void) +{ + int ret = EXIT_SUCCESS; + pid_t child; + + srandom(getpid()); + + ksft_print_header(); + ksft_set_plan(EXPECTED_TESTS); + + if (!(getauxval(AT_HWCAP2) & HWCAP2_SME)) + ksft_exit_skip("SME not available\n"); + + child = fork(); + if (!child) + return do_child(); + + if (do_parent(child)) + ret = EXIT_FAILURE; + + ksft_print_cnts(); + + return ret; +}