diff options
Diffstat (limited to 'arch/arm64/kernel/fpsimd.c')
| -rw-r--r-- | arch/arm64/kernel/fpsimd.c | 1381 |
1 files changed, 1054 insertions, 327 deletions
diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 3eb338f14386..a5dc6f764195 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -12,8 +12,10 @@ #include <linux/bug.h> #include <linux/cache.h> #include <linux/compat.h> +#include <linux/compiler.h> #include <linux/cpu.h> #include <linux/cpu_pm.h> +#include <linux/ctype.h> #include <linux/kernel.h> #include <linux/linkage.h> #include <linux/irqflags.h> @@ -31,9 +33,11 @@ #include <linux/swab.h> #include <asm/esr.h> +#include <asm/exception.h> #include <asm/fpsimd.h> #include <asm/cpufeature.h> #include <asm/cputype.h> +#include <asm/neon.h> #include <asm/processor.h> #include <asm/simd.h> #include <asm/sigcontext.h> @@ -75,15 +79,19 @@ * indicate whether or not the userland FPSIMD state of the current task is * present in the registers. The flag is set unless the FPSIMD registers of this * CPU currently contain the most recent userland FPSIMD state of the current - * task. + * task. If the task is behaving as a VMM, then this is will be managed by + * KVM which will clear it to indicate that the vcpu FPSIMD state is currently + * loaded on the CPU, allowing the state to be saved if a FPSIMD-aware + * softirq kicks in. Upon vcpu_put(), KVM will save the vcpu FP state and + * flag the register state as invalid. * - * In order to allow softirq handlers to use FPSIMD, kernel_neon_begin() may - * save the task's FPSIMD context back to task_struct from softirq context. - * To prevent this from racing with the manipulation of the task's FPSIMD state - * from task context and thereby corrupting the state, it is necessary to - * protect any manipulation of a task's fpsimd_state or TIF_FOREIGN_FPSTATE - * flag with {, __}get_cpu_fpsimd_context(). This will still allow softirqs to - * run but prevent them to use FPSIMD. + * In order to allow softirq handlers to use FPSIMD, kernel_neon_begin() may be + * called from softirq context, which will save the task's FPSIMD context back + * to task_struct. To prevent this from racing with the manipulation of the + * task's FPSIMD state from task context and thereby corrupting the state, it + * is necessary to protect any manipulation of a task's fpsimd_state or + * TIF_FOREIGN_FPSTATE flag with get_cpu_fpsimd_context(), which will suspend + * softirq servicing entirely until put_cpu_fpsimd_context() is called. * * For a certain task, the sequence may look something like this: * - the task gets scheduled in; if both the task's fpsimd_cpu field @@ -110,72 +118,117 @@ * returned from the 2nd syscall yet, TIF_FOREIGN_FPSTATE is still set so * whatever is in the FPSIMD registers is not saved to memory, but discarded. */ -struct fpsimd_last_state_struct { - struct user_fpsimd_state *st; - void *sve_state; - unsigned int sve_vl; + +static DEFINE_PER_CPU(struct cpu_fp_state, fpsimd_last_state); + +__ro_after_init struct vl_info vl_info[ARM64_VEC_MAX] = { +#ifdef CONFIG_ARM64_SVE + [ARM64_VEC_SVE] = { + .type = ARM64_VEC_SVE, + .name = "SVE", + .min_vl = SVE_VL_MIN, + .max_vl = SVE_VL_MIN, + .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) +{ + 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; + } +} + +struct vl_config { + int __default_vl; /* Default VL for tasks */ }; -static DEFINE_PER_CPU(struct fpsimd_last_state_struct, fpsimd_last_state); +static struct vl_config vl_config[ARM64_VEC_MAX]; -/* Default VL for tasks that don't set it explicitly: */ -static int sve_default_vl = -1; +static inline int get_default_vl(enum vec_type type) +{ + return READ_ONCE(vl_config[type].__default_vl); +} #ifdef CONFIG_ARM64_SVE -/* Maximum supported vector length across all CPUs (initially poisoned) */ -int __ro_after_init sve_max_vl = SVE_VL_MIN; -int __ro_after_init sve_max_virtualisable_vl = SVE_VL_MIN; +static inline int get_sve_default_vl(void) +{ + return get_default_vl(ARM64_VEC_SVE); +} -/* - * Set of available vector lengths, - * where length vq encoded as bit __vq_to_bit(vq): - */ -__ro_after_init DECLARE_BITMAP(sve_vq_map, SVE_VQ_MAX); -/* Set of vector lengths present on at least one cpu: */ -static __ro_after_init DECLARE_BITMAP(sve_vq_partial_map, SVE_VQ_MAX); +static inline void set_default_vl(enum vec_type type, int val) +{ + WRITE_ONCE(vl_config[type].__default_vl, val); +} + +static inline void set_sve_default_vl(int val) +{ + set_default_vl(ARM64_VEC_SVE, val); +} static void __percpu *efi_sve_state; #else /* ! CONFIG_ARM64_SVE */ /* Dummy declaration for code that will be optimised out: */ -extern __ro_after_init DECLARE_BITMAP(sve_vq_map, SVE_VQ_MAX); -extern __ro_after_init DECLARE_BITMAP(sve_vq_partial_map, SVE_VQ_MAX); extern void __percpu *efi_sve_state; #endif /* ! CONFIG_ARM64_SVE */ -DEFINE_PER_CPU(bool, fpsimd_context_busy); -EXPORT_PER_CPU_SYMBOL(fpsimd_context_busy); +#ifdef CONFIG_ARM64_SME -static void __get_cpu_fpsimd_context(void) +static int get_sme_default_vl(void) { - bool busy = __this_cpu_xchg(fpsimd_context_busy, true); + return get_default_vl(ARM64_VEC_SME); +} - WARN_ON(busy); +static void set_sme_default_vl(int val) +{ + set_default_vl(ARM64_VEC_SME, val); } +static void sme_free(struct task_struct *); + +#else + +static inline void sme_free(struct task_struct *t) { } + +#endif + +static void fpsimd_bind_task_to_cpu(void); + /* * Claim ownership of the CPU FPSIMD context for use by the calling context. * * The caller may freely manipulate the FPSIMD context metadata until * put_cpu_fpsimd_context() is called. * - * The double-underscore version must only be called if you know the task - * can't be preempted. + * On RT kernels local_bh_disable() is not sufficient because it only + * serializes soft interrupt related sections via a local lock, but stays + * preemptible. Disabling preemption is the right choice here as bottom + * half processing is always in thread context on RT kernels so it + * implicitly prevents bottom half processing as well. */ static void get_cpu_fpsimd_context(void) { - preempt_disable(); - __get_cpu_fpsimd_context(); -} - -static void __put_cpu_fpsimd_context(void) -{ - bool busy = __this_cpu_xchg(fpsimd_context_busy, false); - - WARN_ON(!busy); /* No matching get_cpu_fpsimd_context()? */ + if (!IS_ENABLED(CONFIG_PREEMPT_RT)) + local_bh_disable(); + else + preempt_disable(); } /* @@ -187,55 +240,61 @@ static void __put_cpu_fpsimd_context(void) */ static void put_cpu_fpsimd_context(void) { - __put_cpu_fpsimd_context(); - preempt_enable(); + if (!IS_ENABLED(CONFIG_PREEMPT_RT)) + local_bh_enable(); + else + preempt_enable(); } -static bool have_cpu_fpsimd_context(void) +unsigned int task_get_vl(const struct task_struct *task, enum vec_type type) { - return !preemptible() && __this_cpu_read(fpsimd_context_busy); + return task->thread.vl[type]; } -/* - * Call __sve_free() directly only if you know task can't be scheduled - * or preempted. - */ -static void __sve_free(struct task_struct *task) +void task_set_vl(struct task_struct *task, enum vec_type type, + unsigned long vl) { - kfree(task->thread.sve_state); - task->thread.sve_state = NULL; + task->thread.vl[type] = vl; } -static void sve_free(struct task_struct *task) +unsigned int task_get_vl_onexec(const struct task_struct *task, + enum vec_type type) { - WARN_ON(test_tsk_thread_flag(task, TIF_SVE)); + return task->thread.vl_onexec[type]; +} - __sve_free(task); +void task_set_vl_onexec(struct task_struct *task, enum vec_type type, + unsigned long vl) +{ + task->thread.vl_onexec[type] = vl; } /* + * TIF_SME controls whether a task can use SME without trapping while + * in userspace, when TIF_SME is set then we must have storage + * allocated in sve_state and sme_state to store the contents of both ZA + * and the SVE registers for both streaming and non-streaming modes. + * + * If both SVCR.ZA and SVCR.SM are disabled then at any point we + * may disable TIF_SME and reenable traps. + */ + + +/* * TIF_SVE controls whether a task can use SVE without trapping while - * in userspace, and also the way a task's FPSIMD/SVE state is stored - * in thread_struct. + * in userspace, and also (together with TIF_SME) the way a task's + * FPSIMD/SVE state is stored in thread_struct. * * The kernel uses this flag to track whether a user task is actively * using SVE, and therefore whether full SVE register state needs to * be tracked. If not, the cheaper FPSIMD context handling code can * be used instead of the more costly SVE equivalents. * - * * TIF_SVE set: + * * TIF_SVE or SVCR.SM set: * * The task can execute SVE instructions while in userspace without * trapping to the kernel. * - * When stored, Z0-Z31 (incorporating Vn in bits[127:0] or the - * corresponding Zn), P0-P15 and FFR are encoded in in - * task->thread.sve_state, formatted appropriately for vector - * length task->thread.sve_vl. - * - * task->thread.sve_state must point to a valid buffer at least - * sve_state_size(task) bytes in size. - * * During any syscall, the kernel may optionally clear TIF_SVE and * discard the vector state except for the FPSIMD subset. * @@ -245,7 +304,15 @@ static void sve_free(struct task_struct *task) * do_sve_acc() to be called, which does some preparation and then * sets TIF_SVE. * - * When stored, FPSIMD registers V0-V31 are encoded in + * During any syscall, the kernel may optionally clear TIF_SVE and + * discard the vector state except for the FPSIMD subset. + * + * The data will be stored in one of two formats: + * + * * FPSIMD only - FP_STATE_FPSIMD: + * + * When the FPSIMD only state stored task->thread.fp_type is set to + * FP_STATE_FPSIMD, the FPSIMD registers V0-V31 are encoded in * task->thread.uw.fpsimd_state; bits [max : 128] for each of Z0-Z31 are * logically zero but not stored anywhere; P0-P15 and FFR are not * stored and have unspecified values from userspace's point of @@ -253,7 +320,23 @@ static void sve_free(struct task_struct *task) * but userspace is discouraged from relying on this. * * task->thread.sve_state does not need to be non-NULL, valid or any - * particular size: it must not be dereferenced. + * particular size: it must not be dereferenced and any data stored + * there should be considered stale and not referenced. + * + * * SVE state - FP_STATE_SVE: + * + * When the full SVE state is stored task->thread.fp_type is set to + * FP_STATE_SVE and Z0-Z31 (incorporating Vn in bits[127:0] or the + * corresponding Zn), P0-P15 and FFR are encoded in in + * task->thread.sve_state, formatted appropriately for vector + * length task->thread.sve_vl or, if SVCR.SM is set, + * task->thread.sme_vl. The storage for the vector registers in + * task->thread.uw.fpsimd_state should be ignored. + * + * task->thread.sve_state must point to a valid buffer at least + * sve_state_size(task) bytes in size. The data stored in + * task->thread.uw.fpsimd_state.vregs should be considered stale + * and not referenced. * * * FPSR and FPCR are always stored in task->thread.uw.fpsimd_state * irrespective of whether TIF_SVE is clear or set, since these are @@ -269,45 +352,149 @@ static void sve_free(struct task_struct *task) */ static void task_fpsimd_load(void) { - WARN_ON(!have_cpu_fpsimd_context()); + bool restore_sve_regs = false; + bool restore_ffr; + + WARN_ON(!system_supports_fpsimd()); + WARN_ON(preemptible()); + WARN_ON(test_thread_flag(TIF_KERNEL_FPSTATE)); + + if (system_supports_sve() || system_supports_sme()) { + switch (current->thread.fp_type) { + case FP_STATE_FPSIMD: + /* Stop tracking SVE for this task until next use. */ + if (test_and_clear_thread_flag(TIF_SVE)) + sve_user_disable(); + break; + case FP_STATE_SVE: + if (!thread_sm_enabled(¤t->thread) && + !WARN_ON_ONCE(!test_and_set_thread_flag(TIF_SVE))) + sve_user_enable(); + + if (test_thread_flag(TIF_SVE)) + sve_set_vq(sve_vq_from_vl(task_get_sve_vl(current)) - 1); + + restore_sve_regs = true; + restore_ffr = true; + break; + default: + /* + * This indicates either a bug in + * fpsimd_save_user_state() or memory corruption, we + * should always record an explicit format + * when we save. We always at least have the + * memory allocated for FPSMID registers so + * try that and hope for the best. + */ + WARN_ON_ONCE(1); + clear_thread_flag(TIF_SVE); + break; + } + } - if (system_supports_sve() && test_thread_flag(TIF_SVE)) + /* Restore SME, override SVE register configuration if needed */ + 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); + + if (thread_za_enabled(¤t->thread)) + sme_load_state(current->thread.sme_state, + system_supports_sme2()); + + if (thread_sm_enabled(¤t->thread)) + restore_ffr = system_supports_fa64(); + } + + if (restore_sve_regs) { + WARN_ON_ONCE(current->thread.fp_type != FP_STATE_SVE); sve_load_state(sve_pffr(¤t->thread), ¤t->thread.uw.fpsimd_state.fpsr, - sve_vq_from_vl(current->thread.sve_vl) - 1); - else + restore_ffr); + } else { + WARN_ON_ONCE(current->thread.fp_type != FP_STATE_FPSIMD); fpsimd_load_state(¤t->thread.uw.fpsimd_state); + } } /* * Ensure FPSIMD/SVE storage in memory for the loaded context is up to - * date with respect to the CPU registers. + * date with respect to the CPU registers. Note carefully that the + * current context is the context last bound to the CPU stored in + * last, if KVM is involved this may be the guest VM context rather + * than the host thread for the VM pointed to by current. This means + * that we must always reference the state storage via last rather + * than via current, if we are saving KVM state then it will have + * ensured that the type of registers to save is set in last->to_save. */ -static void fpsimd_save(void) +static void fpsimd_save_user_state(void) { - struct fpsimd_last_state_struct const *last = + struct cpu_fp_state const *last = this_cpu_ptr(&fpsimd_last_state); /* set by fpsimd_bind_task_to_cpu() or fpsimd_bind_state_to_cpu() */ + bool save_sve_regs = false; + bool save_ffr; + unsigned int vl; - WARN_ON(!have_cpu_fpsimd_context()); + WARN_ON(!system_supports_fpsimd()); + WARN_ON(preemptible()); - if (!test_thread_flag(TIF_FOREIGN_FPSTATE)) { - if (system_supports_sve() && test_thread_flag(TIF_SVE)) { - if (WARN_ON(sve_get_vl() != last->sve_vl)) { - /* - * Can't save the user regs, so current would - * re-enter user with corrupt state. - * There's no way to recover, so kill it: - */ - force_signal_inject(SIGKILL, SI_KERNEL, 0); - return; - } + if (test_thread_flag(TIF_FOREIGN_FPSTATE)) + return; + + /* + * If a task is in a syscall the ABI allows us to only + * preserve the state shared with FPSIMD so don't bother + * saving the full SVE state in that case. + */ + if ((last->to_save == FP_STATE_CURRENT && test_thread_flag(TIF_SVE) && + !in_syscall(current_pt_regs())) || + last->to_save == FP_STATE_SVE) { + save_sve_regs = true; + save_ffr = true; + vl = last->sve_vl; + } - sve_save_state((char *)last->sve_state + - sve_ffr_offset(last->sve_vl), - &last->st->fpsr); - } else - fpsimd_save_state(last->st); + if (system_supports_sme()) { + u64 *svcr = last->svcr; + + *svcr = read_sysreg_s(SYS_SVCR); + + if (*svcr & SVCR_ZA_MASK) + sme_save_state(last->sme_state, + system_supports_sme2()); + + /* If we are in streaming mode override regular SVE. */ + if (*svcr & SVCR_SM_MASK) { + save_sve_regs = true; + save_ffr = system_supports_fa64(); + vl = last->sme_vl; + } + } + + if (IS_ENABLED(CONFIG_ARM64_SVE) && save_sve_regs) { + /* Get the configured VL from RDVL, will account for SM */ + if (WARN_ON(sve_get_vl() != vl)) { + /* + * Can't save the user regs, so current would + * re-enter user with corrupt state. + * There's no way to recover, so kill it: + */ + force_signal_inject(SIGKILL, SI_KERNEL, 0, 0); + return; + } + + sve_save_state((char *)last->sve_state + + sve_ffr_offset(vl), + &last->st->fpsr, save_ffr); + *last->fp_type = FP_STATE_SVE; + } else { + fpsimd_save_state(last->st); + *last->fp_type = FP_STATE_FPSIMD; } } @@ -317,33 +504,38 @@ static void fpsimd_save(void) * If things go wrong there's a bug somewhere, but try to fall back to a * safe choice. */ -static unsigned int find_supported_vector_length(unsigned int vl) +static unsigned int find_supported_vector_length(enum vec_type type, + unsigned int vl) { + struct vl_info *info = &vl_info[type]; int bit; - int max_vl = sve_max_vl; + int max_vl = info->max_vl; if (WARN_ON(!sve_vl_valid(vl))) - vl = SVE_VL_MIN; + vl = info->min_vl; if (WARN_ON(!sve_vl_valid(max_vl))) - max_vl = SVE_VL_MIN; + max_vl = info->min_vl; if (vl > max_vl) vl = max_vl; + if (vl < info->min_vl) + vl = info->min_vl; - bit = find_next_bit(sve_vq_map, SVE_VQ_MAX, + bit = find_next_bit(info->vq_map, SVE_VQ_MAX, __vq_to_bit(sve_vq_from_vl(vl))); return sve_vl_from_vq(__bit_to_vq(bit)); } -#ifdef CONFIG_SYSCTL +#if defined(CONFIG_ARM64_SVE) && defined(CONFIG_SYSCTL) -static int sve_proc_do_default_vl(struct ctl_table *table, int write, - void __user *buffer, size_t *lenp, - loff_t *ppos) +static int vec_proc_do_default_vl(struct ctl_table *table, int write, + void *buffer, size_t *lenp, loff_t *ppos) { + struct vl_info *info = table->extra1; + enum vec_type type = info->type; int ret; - int vl = sve_default_vl; + int vl = get_default_vl(type); struct ctl_table tmp_table = { .data = &vl, .maxlen = sizeof(vl), @@ -355,12 +547,12 @@ static int sve_proc_do_default_vl(struct ctl_table *table, int write, /* Writing -1 has the special meaning "set to max": */ if (vl == -1) - vl = sve_max_vl; + vl = info->max_vl; if (!sve_vl_valid(vl)) return -EINVAL; - sve_default_vl = find_supported_vector_length(vl); + set_default_vl(type, find_supported_vector_length(type, vl)); return 0; } @@ -368,9 +560,9 @@ static struct ctl_table sve_default_vl_table[] = { { .procname = "sve_default_vector_length", .mode = 0644, - .proc_handler = sve_proc_do_default_vl, + .proc_handler = vec_proc_do_default_vl, + .extra1 = &vl_info[ARM64_VEC_SVE], }, - { } }; static int __init sve_sysctl_init(void) @@ -382,9 +574,32 @@ static int __init sve_sysctl_init(void) return 0; } -#else /* ! CONFIG_SYSCTL */ +#else /* ! (CONFIG_ARM64_SVE && CONFIG_SYSCTL) */ static int __init sve_sysctl_init(void) { return 0; } -#endif /* ! CONFIG_SYSCTL */ +#endif /* ! (CONFIG_ARM64_SVE && CONFIG_SYSCTL) */ + +#if defined(CONFIG_ARM64_SME) && defined(CONFIG_SYSCTL) +static struct ctl_table sme_default_vl_table[] = { + { + .procname = "sme_default_vector_length", + .mode = 0644, + .proc_handler = vec_proc_do_default_vl, + .extra1 = &vl_info[ARM64_VEC_SME], + }, +}; + +static int __init sme_sysctl_init(void) +{ + if (system_supports_sme()) + if (!register_sysctl("abi", sme_default_vl_table)) + return -EINVAL; + + return 0; +} + +#else /* ! (CONFIG_ARM64_SME && CONFIG_SYSCTL) */ +static int __init sme_sysctl_init(void) { return 0; } +#endif /* ! (CONFIG_ARM64_SME && CONFIG_SYSCTL) */ #define ZREG(sve_state, vq, n) ((char *)(sve_state) + \ (SVE_SIG_ZREG_OFFSET(vq, n) - SVE_SIG_REGS_OFFSET)) @@ -436,10 +651,10 @@ static void fpsimd_to_sve(struct task_struct *task) void *sst = task->thread.sve_state; struct user_fpsimd_state const *fst = &task->thread.uw.fpsimd_state; - if (!system_supports_sve()) + if (!system_supports_sve() && !system_supports_sme()) return; - vq = sve_vq_from_vl(task->thread.sve_vl); + vq = sve_vq_from_vl(thread_get_cur_vl(&task->thread)); __fpsimd_to_sve(sst, fst, vq); } @@ -456,16 +671,17 @@ 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; __uint128_t const *p; - if (!system_supports_sve()) + if (!system_supports_sve() && !system_supports_sme()) return; - vq = sve_vq_from_vl(task->thread.sve_vl); + vl = thread_get_cur_vl(&task->thread); + 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); @@ -473,6 +689,22 @@ static void sve_to_fpsimd(struct task_struct *task) } #ifdef CONFIG_ARM64_SVE +/* + * Call __sve_free() directly only if you know task can't be scheduled + * or preempted. + */ +static void __sve_free(struct task_struct *task) +{ + kfree(task->thread.sve_state); + task->thread.sve_state = NULL; +} + +static void sve_free(struct task_struct *task) +{ + WARN_ON(test_tsk_thread_flag(task, TIF_SVE)); + + __sve_free(task); +} /* * Return how many bytes of memory are required to store the full SVE @@ -480,7 +712,14 @@ static void sve_to_fpsimd(struct task_struct *task) */ size_t sve_state_size(struct task_struct const *task) { - return SVE_SIG_REGS_SIZE(sve_vq_from_vl(task->thread.sve_vl)); + unsigned int vl = 0; + + if (system_supports_sve()) + vl = task_get_sve_vl(task); + if (system_supports_sme()) + vl = max(vl, task_get_sme_vl(task)); + + return SVE_SIG_REGS_SIZE(sve_vq_from_vl(vl)); } /* @@ -493,26 +732,35 @@ size_t sve_state_size(struct task_struct const *task) * do_sve_acc() case, there is no ABI requirement to hide stale data * written previously be task. */ -void sve_alloc(struct task_struct *task) +void sve_alloc(struct task_struct *task, bool flush) { if (task->thread.sve_state) { - memset(task->thread.sve_state, 0, sve_state_size(current)); + if (flush) + memset(task->thread.sve_state, 0, + sve_state_size(task)); return; } /* This is a small allocation (maximum ~8KB) and Should Not Fail. */ task->thread.sve_state = kzalloc(sve_state_size(task), GFP_KERNEL); - - /* - * If future SVE revisions can have larger vectors though, - * this may cease to be true: - */ - BUG_ON(!task->thread.sve_state); } /* + * Force the FPSIMD state shared with SVE to be updated in the SVE state + * even if the SVE state is the current active state. + * + * This should only be called by ptrace. task must be non-runnable. + * task->thread.sve_state must point to at least sve_state_size(task) + * bytes of allocated kernel memory. + */ +void fpsimd_force_sync_to_sve(struct task_struct *task) +{ + fpsimd_to_sve(task); +} + +/* * Ensure that task->thread.sve_state is up to date with respect to * the user task, irrespective of when SVE is in use or not. * @@ -522,7 +770,8 @@ void sve_alloc(struct task_struct *task) */ void fpsimd_sync_to_sve(struct task_struct *task) { - if (!test_tsk_thread_flag(task, TIF_SVE)) + if (!test_tsk_thread_flag(task, TIF_SVE) && + !thread_sm_enabled(&task->thread)) fpsimd_to_sve(task); } @@ -536,7 +785,7 @@ void fpsimd_sync_to_sve(struct task_struct *task) */ void sve_sync_to_fpsimd(struct task_struct *task) { - if (test_tsk_thread_flag(task, TIF_SVE)) + if (task->thread.fp_type == FP_STATE_SVE) sve_to_fpsimd(task); } @@ -558,18 +807,21 @@ void sve_sync_from_fpsimd_zeropad(struct task_struct *task) void *sst = task->thread.sve_state; struct user_fpsimd_state const *fst = &task->thread.uw.fpsimd_state; - if (!test_tsk_thread_flag(task, TIF_SVE)) + if (!test_tsk_thread_flag(task, TIF_SVE) && + !thread_sm_enabled(&task->thread)) return; - vq = sve_vq_from_vl(task->thread.sve_vl); + vq = sve_vq_from_vl(thread_get_cur_vl(&task->thread)); memset(sst, 0, SVE_SIG_REGS_SIZE(vq)); __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) { + bool free_sme = false; + if (flags & ~(unsigned long)(PR_SVE_VL_INHERIT | PR_SVE_SET_VL_ONEXEC)) return -EINVAL; @@ -578,57 +830,82 @@ 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(vl); + vl = find_supported_vector_length(type, vl); if (flags & (PR_SVE_VL_INHERIT | PR_SVE_SET_VL_ONEXEC)) - task->thread.sve_vl_onexec = vl; + task_set_vl_onexec(task, type, vl); else /* Reset VL to system default on next exec: */ - task->thread.sve_vl_onexec = 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->thread.sve_vl) + 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. */ if (task == current) { get_cpu_fpsimd_context(); - fpsimd_save(); + fpsimd_save_user_state(); } fpsimd_flush_task_state(task); - if (test_and_clear_tsk_thread_flag(task, TIF_SVE)) + if (test_and_clear_tsk_thread_flag(task, TIF_SVE) || + thread_sm_enabled(&task->thread)) { sve_to_fpsimd(task); + task->thread.fp_type = FP_STATE_FPSIMD; + } + + if (system_supports_sme()) { + if (type == ARM64_VEC_SME || + !(task->thread.svcr & (SVCR_SM_MASK | SVCR_ZA_MASK))) { + /* + * We are changing the SME VL or weren't using + * SME anyway, discard the state and force a + * reallocation. + */ + task->thread.svcr &= ~(SVCR_SM_MASK | + SVCR_ZA_MASK); + clear_tsk_thread_flag(task, TIF_SME); + free_sme = true; + } + } if (task == current) put_cpu_fpsimd_context(); + task_set_vl(task, type, vl); + /* - * Force reallocation of task SVE state to the correct size - * on next use: + * Free the changed states if they are not in use, SME will be + * reallocated to the correct size on next use and we just + * allocate SVE now in case it is needed for use in streaming + * mode. */ sve_free(task); + sve_alloc(task, true); - task->thread.sve_vl = vl; + if (free_sme) + sme_free(task); 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; @@ -636,20 +913,21 @@ out: /* * 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 = current->thread.sve_vl_onexec; + ret = task_get_vl_onexec(current, type); else - ret = current->thread.sve_vl; + 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; @@ -664,38 +942,81 @@ int sve_set_current_vl(unsigned long arg) vl = arg & PR_SVE_VL_LEN_MASK; flags = arg & ~vl; - if (!system_supports_sve()) + 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 */ int sve_get_current_vl(void) { - if (!system_supports_sve()) + if (!system_supports_sve() || is_compat_task()) + return -EINVAL; + + 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 sve_prctl_status(0); + return vec_prctl_status(ARM64_VEC_SME, 0); } +#endif /* CONFIG_ARM64_SME */ -static void sve_probe_vqs(DECLARE_BITMAP(map, SVE_VQ_MAX)) +static void vec_probe_vqs(struct vl_info *info, + DECLARE_BITMAP(map, SVE_VQ_MAX)) { unsigned int vq, vl; - unsigned long zcr; bitmap_zero(map, SVE_VQ_MAX); - zcr = ZCR_ELx_LEN_MASK; - zcr = read_sysreg_s(SYS_ZCR_EL1) & ~zcr; - for (vq = SVE_VQ_MAX; vq >= SVE_VQ_MIN; --vq) { - write_sysreg_s(zcr | (vq - 1), SYS_ZCR_EL1); /* self-syncing */ - vl = sve_get_vl(); + write_vl(info->type, vq - 1); /* self-syncing */ + + switch (info->type) { + case ARM64_VEC_SVE: + vl = sve_get_vl(); + break; + case ARM64_VEC_SME: + vl = sme_get_vl(); + break; + default: + vl = 0; + break; + } + + /* 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); } @@ -705,10 +1026,11 @@ static void sve_probe_vqs(DECLARE_BITMAP(map, SVE_VQ_MAX)) * Initialise the set of known supported VQs for the boot CPU. * This is called during kernel boot, before secondary CPUs are brought up. */ -void __init sve_init_vq_map(void) +void __init vec_init_vq_map(enum vec_type type) { - sve_probe_vqs(sve_vq_map); - bitmap_copy(sve_vq_partial_map, sve_vq_map, SVE_VQ_MAX); + struct vl_info *info = &vl_info[type]; + vec_probe_vqs(info, info->vq_map); + bitmap_copy(info->vq_partial_map, info->vq_map, SVE_VQ_MAX); } /* @@ -716,30 +1038,33 @@ void __init sve_init_vq_map(void) * those not supported by the current CPU. * This function is called during the bring-up of early secondary CPUs only. */ -void sve_update_vq_map(void) +void vec_update_vq_map(enum vec_type type) { + struct vl_info *info = &vl_info[type]; DECLARE_BITMAP(tmp_map, SVE_VQ_MAX); - sve_probe_vqs(tmp_map); - bitmap_and(sve_vq_map, sve_vq_map, tmp_map, SVE_VQ_MAX); - bitmap_or(sve_vq_partial_map, sve_vq_partial_map, tmp_map, SVE_VQ_MAX); + vec_probe_vqs(info, tmp_map); + bitmap_and(info->vq_map, info->vq_map, tmp_map, SVE_VQ_MAX); + bitmap_or(info->vq_partial_map, info->vq_partial_map, tmp_map, + SVE_VQ_MAX); } /* * Check whether the current CPU supports all VQs in the committed set. * This function is called during the bring-up of late secondary CPUs only. */ -int sve_verify_vq_map(void) +int vec_verify_vq_map(enum vec_type type) { + struct vl_info *info = &vl_info[type]; DECLARE_BITMAP(tmp_map, SVE_VQ_MAX); unsigned long b; - sve_probe_vqs(tmp_map); + vec_probe_vqs(info, tmp_map); bitmap_complement(tmp_map, tmp_map, SVE_VQ_MAX); - if (bitmap_intersects(tmp_map, sve_vq_map, SVE_VQ_MAX)) { - pr_warn("SVE: cpu%d: Required vector length(s) missing\n", - smp_processor_id()); + if (bitmap_intersects(tmp_map, info->vq_map, SVE_VQ_MAX)) { + pr_warn("%s: cpu%d: Required vector length(s) missing\n", + info->name, smp_processor_id()); return -EINVAL; } @@ -755,7 +1080,7 @@ int sve_verify_vq_map(void) /* Recover the set of supported VQs: */ bitmap_complement(tmp_map, tmp_map, SVE_VQ_MAX); /* Find VQs supported that are not globally supported: */ - bitmap_andnot(tmp_map, tmp_map, sve_vq_map, SVE_VQ_MAX); + bitmap_andnot(tmp_map, tmp_map, info->vq_map, SVE_VQ_MAX); /* Find the lowest such VQ, if any: */ b = find_last_bit(tmp_map, SVE_VQ_MAX); @@ -766,9 +1091,9 @@ int sve_verify_vq_map(void) * Mismatches above sve_max_virtualisable_vl are fine, since * no guest is allowed to configure ZCR_EL2.LEN to exceed this: */ - if (sve_vl_from_vq(__bit_to_vq(b)) <= sve_max_virtualisable_vl) { - pr_warn("SVE: cpu%d: Unsupported vector length(s) present\n", - smp_processor_id()); + if (sve_vl_from_vq(__bit_to_vq(b)) <= info->max_virtualisable_vl) { + pr_warn("%s: cpu%d: Unsupported vector length(s) present\n", + info->name, smp_processor_id()); return -EINVAL; } @@ -777,19 +1102,25 @@ int sve_verify_vq_map(void) static void __init sve_efi_setup(void) { + int max_vl = 0; + int i; + if (!IS_ENABLED(CONFIG_EFI)) return; + for (i = 0; i < ARRAY_SIZE(vl_info); i++) + max_vl = max(vl_info[i].max_vl, max_vl); + /* * alloc_percpu() warns and prints a backtrace if this goes wrong. * This is evidence of a crippled system and we are returning void, * so no attempt is made to handle this situation here. */ - if (!sve_vl_valid(sve_max_vl)) + if (!sve_vl_valid(max_vl)) goto fail; efi_sve_state = __alloc_percpu( - SVE_SIG_REGS_SIZE(sve_vq_from_vl(sve_max_vl)), SVE_VQ_BYTES); + SVE_SIG_REGS_SIZE(sve_vq_from_vl(max_vl)), SVE_VQ_BYTES); if (!efi_sve_state) goto fail; @@ -799,48 +1130,18 @@ fail: panic("Cannot allocate percpu memory for EFI SVE save/restore"); } -/* - * Enable SVE for EL1. - * Intended for use by the cpufeatures code during CPU boot. - */ -void sve_kernel_enable(const struct arm64_cpu_capabilities *__always_unused p) +void cpu_enable_sve(const struct arm64_cpu_capabilities *__always_unused p) { write_sysreg(read_sysreg(CPACR_EL1) | CPACR_EL1_ZEN_EL1EN, CPACR_EL1); isb(); } -/* - * Read the pseudo-ZCR used by cpufeatures to identify the supported SVE - * vector length. - * - * Use only if SVE is present. - * This function clobbers the SVE vector length. - */ -u64 read_zcr_features(void) -{ - u64 zcr; - unsigned int vq_max; - - /* - * Set the maximum possible VL, and write zeroes to all other - * bits to see if they stick. - */ - sve_kernel_enable(NULL); - write_sysreg_s(ZCR_ELx_LEN_MASK, SYS_ZCR_EL1); - - zcr = read_sysreg_s(SYS_ZCR_EL1); - zcr &= ~(u64)ZCR_ELx_LEN_MASK; /* find sticky 1s outside LEN field */ - vq_max = sve_vq_from_vl(sve_get_vl()); - zcr |= vq_max - 1; /* set LEN field to maximum effective value */ - - return zcr; -} - void __init sve_setup(void) { - u64 zcr; + struct vl_info *info = &vl_info[ARM64_VEC_SVE]; DECLARE_BITMAP(tmp_map, SVE_VQ_MAX); unsigned long b; + int max_bit; if (!system_supports_sve()) return; @@ -850,49 +1151,43 @@ void __init sve_setup(void) * so sve_vq_map must have at least SVE_VQ_MIN set. * If something went wrong, at least try to patch it up: */ - if (WARN_ON(!test_bit(__vq_to_bit(SVE_VQ_MIN), sve_vq_map))) - set_bit(__vq_to_bit(SVE_VQ_MIN), sve_vq_map); - - zcr = read_sanitised_ftr_reg(SYS_ZCR_EL1); - sve_max_vl = sve_vl_from_vq((zcr & ZCR_ELx_LEN_MASK) + 1); + if (WARN_ON(!test_bit(__vq_to_bit(SVE_VQ_MIN), info->vq_map))) + set_bit(__vq_to_bit(SVE_VQ_MIN), info->vq_map); - /* - * Sanity-check that the max VL we determined through CPU features - * corresponds properly to sve_vq_map. If not, do our best: - */ - if (WARN_ON(sve_max_vl != find_supported_vector_length(sve_max_vl))) - sve_max_vl = find_supported_vector_length(sve_max_vl); + max_bit = find_first_bit(info->vq_map, SVE_VQ_MAX); + info->max_vl = sve_vl_from_vq(__bit_to_vq(max_bit)); /* * For the default VL, pick the maximum supported value <= 64. * VL == 64 is guaranteed not to grow the signal frame. */ - sve_default_vl = find_supported_vector_length(64); + set_sve_default_vl(find_supported_vector_length(ARM64_VEC_SVE, 64)); - bitmap_andnot(tmp_map, sve_vq_partial_map, sve_vq_map, + bitmap_andnot(tmp_map, info->vq_partial_map, info->vq_map, SVE_VQ_MAX); b = find_last_bit(tmp_map, SVE_VQ_MAX); if (b >= SVE_VQ_MAX) /* No non-virtualisable VLs found */ - sve_max_virtualisable_vl = SVE_VQ_MAX; + info->max_virtualisable_vl = SVE_VQ_MAX; else if (WARN_ON(b == SVE_VQ_MAX - 1)) /* No virtualisable VLs? This is architecturally forbidden. */ - sve_max_virtualisable_vl = SVE_VQ_MIN; + info->max_virtualisable_vl = SVE_VQ_MIN; else /* b + 1 < SVE_VQ_MAX */ - sve_max_virtualisable_vl = sve_vl_from_vq(__bit_to_vq(b + 1)); + info->max_virtualisable_vl = sve_vl_from_vq(__bit_to_vq(b + 1)); - if (sve_max_virtualisable_vl > sve_max_vl) - sve_max_virtualisable_vl = sve_max_vl; + if (info->max_virtualisable_vl > info->max_vl) + info->max_virtualisable_vl = info->max_vl; - pr_info("SVE: maximum available vector length %u bytes per vector\n", - sve_max_vl); - pr_info("SVE: default vector length %u bytes per vector\n", - sve_default_vl); + pr_info("%s: maximum available vector length %u bytes per vector\n", + info->name, info->max_vl); + pr_info("%s: default vector length %u bytes per vector\n", + info->name, get_sve_default_vl()); /* KVM decides whether to support mismatched systems. Just warn here: */ - if (sve_max_virtualisable_vl < sve_max_vl) - pr_warn("SVE: unvirtualisable vector lengths present\n"); + if (sve_max_virtualisable_vl() < sve_max_vl()) + pr_warn("%s: unvirtualisable vector lengths present\n", + info->name); sve_efi_setup(); } @@ -904,59 +1199,260 @@ void __init sve_setup(void) void fpsimd_release_task(struct task_struct *dead_task) { __sve_free(dead_task); + sme_free(dead_task); } #endif /* CONFIG_ARM64_SVE */ +#ifdef CONFIG_ARM64_SME + +/* + * Ensure that task->thread.sme_state is allocated and sufficiently large. + * + * This function should be used only in preparation for replacing + * task->thread.sme_state with new data. The memory is always zeroed + * here to prevent stale data from showing through: this is done in + * the interest of testability and predictability, the architecture + * guarantees that when ZA is enabled it will be zeroed. + */ +void sme_alloc(struct task_struct *task, bool flush) +{ + if (task->thread.sme_state) { + if (flush) + memset(task->thread.sme_state, 0, + sme_state_size(task)); + return; + } + + /* This could potentially be up to 64K. */ + task->thread.sme_state = + kzalloc(sme_state_size(task), GFP_KERNEL); +} + +static void sme_free(struct task_struct *task) +{ + kfree(task->thread.sme_state); + task->thread.sme_state = NULL; +} + +void cpu_enable_sme(const struct arm64_cpu_capabilities *__always_unused p) +{ + /* Set priority for all PEs to architecturally defined minimum */ + write_sysreg_s(read_sysreg_s(SYS_SMPRI_EL1) & ~SMPRI_EL1_PRIORITY_MASK, + SYS_SMPRI_EL1); + + /* Allow SME in kernel */ + write_sysreg(read_sysreg(CPACR_EL1) | CPACR_EL1_SMEN_EL1EN, CPACR_EL1); + isb(); + + /* Allow EL0 to access TPIDR2 */ + write_sysreg(read_sysreg(SCTLR_EL1) | SCTLR_ELx_ENTP2, SCTLR_EL1); + isb(); +} + +void cpu_enable_sme2(const struct arm64_cpu_capabilities *__always_unused p) +{ + /* This must be enabled after SME */ + BUILD_BUG_ON(ARM64_SME2 <= ARM64_SME); + + /* Allow use of ZT0 */ + write_sysreg_s(read_sysreg_s(SYS_SMCR_EL1) | SMCR_ELx_EZT0_MASK, + SYS_SMCR_EL1); +} + +void cpu_enable_fa64(const struct arm64_cpu_capabilities *__always_unused p) +{ + /* This must be enabled after SME */ + BUILD_BUG_ON(ARM64_SME_FA64 <= ARM64_SME); + + /* Allow use of FA64 */ + write_sysreg_s(read_sysreg_s(SYS_SMCR_EL1) | SMCR_ELx_FA64_MASK, + SYS_SMCR_EL1); +} + +void __init sme_setup(void) +{ + struct vl_info *info = &vl_info[ARM64_VEC_SME]; + int min_bit, max_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. The bitmap is SVE_VQ_MAP sized for + * sharing with SVE. + */ + 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)); + + max_bit = find_first_bit(info->vq_map, SVE_VQ_MAX); + info->max_vl = sve_vl_from_vq(__bit_to_vq(max_bit)); + + 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 */ + +static void sve_init_regs(void) +{ + /* + * Convert the FPSIMD state to SVE, zeroing all the state that + * is not shared with FPSIMD. If (as is likely) the current + * state is live in the registers then do this there and + * update our metadata for the current task including + * disabling the trap, otherwise update our in-memory copy. + * We are guaranteed to not be in streaming mode, we can only + * take a SVE trap when not in streaming mode and we can't be + * in streaming mode when taking a SME trap. + */ + if (!test_thread_flag(TIF_FOREIGN_FPSTATE)) { + unsigned long vq_minus_one = + sve_vq_from_vl(task_get_sve_vl(current)) - 1; + sve_set_vq(vq_minus_one); + sve_flush_live(true, vq_minus_one); + fpsimd_bind_task_to_cpu(); + } else { + fpsimd_to_sve(current); + current->thread.fp_type = FP_STATE_SVE; + } +} + /* * Trapped SVE access * * Storage is allocated for the full SVE state, the current FPSIMD - * register contents are migrated across, and TIF_SVE is set so that - * the SVE access trap will be disabled the next time this task - * reaches ret_to_user. + * register contents are migrated across, and the access trap is + * disabled. * - * TIF_SVE should be clear on entry: otherwise, task_fpsimd_load() + * TIF_SVE should be clear on entry: otherwise, fpsimd_restore_current_state() * would have disabled the SVE access trap for userspace during * ret_to_user, making an SVE access trap impossible in that case. */ -void do_sve_acc(unsigned int esr, struct pt_regs *regs) +void do_sve_acc(unsigned long esr, struct pt_regs *regs) { /* Even if we chose not to use SVE, the hardware could still trap: */ if (unlikely(!system_supports_sve()) || WARN_ON(is_compat_task())) { - force_signal_inject(SIGILL, ILL_ILLOPC, regs->pc); + force_signal_inject(SIGILL, ILL_ILLOPC, regs->pc, 0); return; } - sve_alloc(current); + sve_alloc(current, true); + if (!current->thread.sve_state) { + force_sig(SIGKILL); + return; + } get_cpu_fpsimd_context(); - fpsimd_save(); - - /* Force ret_to_user to reload the registers: */ - fpsimd_flush_task_state(current); - - fpsimd_to_sve(current); if (test_and_set_thread_flag(TIF_SVE)) WARN_ON(1); /* SVE access shouldn't have trapped */ + /* + * Even if the task can have used streaming mode we can only + * generate SVE access traps in normal SVE mode and + * transitioning out of streaming mode may discard any + * streaming mode state. Always clear the high bits to avoid + * any potential errors tracking what is properly initialised. + */ + sve_init_regs(); + + put_cpu_fpsimd_context(); +} + +/* + * Trapped SME access + * + * Storage is allocated for the full SVE and SME state, the current + * FPSIMD register contents are migrated to SVE if SVE is not already + * active, and the access trap is disabled. + * + * TIF_SME should be clear on entry: otherwise, fpsimd_restore_current_state() + * would have disabled the SME access trap for userspace during + * ret_to_user, making an SME access trap impossible in that case. + */ +void do_sme_acc(unsigned long esr, struct pt_regs *regs) +{ + /* Even if we chose not to use SME, the hardware could still trap: */ + if (unlikely(!system_supports_sme()) || WARN_ON(is_compat_task())) { + force_signal_inject(SIGILL, ILL_ILLOPC, regs->pc, 0); + return; + } + + /* + * If this not a trap due to SME being disabled then something + * is being used in the wrong mode, report as SIGILL. + */ + if (ESR_ELx_ISS(esr) != ESR_ELx_SME_ISS_SME_DISABLED) { + force_signal_inject(SIGILL, ILL_ILLOPC, regs->pc, 0); + return; + } + + sve_alloc(current, false); + sme_alloc(current, true); + if (!current->thread.sve_state || !current->thread.sme_state) { + force_sig(SIGKILL); + return; + } + + get_cpu_fpsimd_context(); + + /* With TIF_SME userspace shouldn't generate any traps */ + if (test_and_set_thread_flag(TIF_SME)) + WARN_ON(1); + + if (!test_thread_flag(TIF_FOREIGN_FPSTATE)) { + unsigned long vq_minus_one = + sve_vq_from_vl(task_get_sme_vl(current)) - 1; + sme_set_vq(vq_minus_one); + + fpsimd_bind_task_to_cpu(); + } + put_cpu_fpsimd_context(); } /* * Trapped FP/ASIMD access. */ -void do_fpsimd_acc(unsigned int esr, struct pt_regs *regs) +void do_fpsimd_acc(unsigned long esr, struct pt_regs *regs) { - /* TODO: implement lazy context saving/restoring */ - WARN_ON(1); + /* Even if we chose not to use FPSIMD, the hardware could still trap: */ + if (!system_supports_fpsimd()) { + force_signal_inject(SIGILL, ILL_ILLOPC, regs->pc, 0); + return; + } + + /* + * When FPSIMD is enabled, we should never take a trap unless something + * has gone very wrong. + */ + BUG(); } /* * Raise a SIGFPE for the current process. */ -void do_fpsimd_exc(unsigned int esr, struct pt_regs *regs) +void do_fpsimd_exc(unsigned long esr, struct pt_regs *regs) { unsigned int si_code = FPE_FLTUNK; @@ -978,6 +1474,34 @@ void do_fpsimd_exc(unsigned int esr, struct pt_regs *regs) current); } +static void fpsimd_load_kernel_state(struct task_struct *task) +{ + struct cpu_fp_state *last = this_cpu_ptr(&fpsimd_last_state); + + /* + * Elide the load if this CPU holds the most recent kernel mode + * FPSIMD context of the current task. + */ + if (last->st == &task->thread.kernel_fpsimd_state && + task->thread.kernel_fpsimd_cpu == smp_processor_id()) + return; + + fpsimd_load_state(&task->thread.kernel_fpsimd_state); +} + +static void fpsimd_save_kernel_state(struct task_struct *task) +{ + struct cpu_fp_state cpu_fp_state = { + .st = &task->thread.kernel_fpsimd_state, + .to_save = FP_STATE_FPSIMD, + }; + + fpsimd_save_state(&task->thread.kernel_fpsimd_state); + fpsimd_bind_state_to_cpu(&cpu_fp_state); + + task->thread.kernel_fpsimd_cpu = smp_processor_id(); +} + void fpsimd_thread_switch(struct task_struct *next) { bool wrong_task, wrong_cpu; @@ -985,29 +1509,72 @@ void fpsimd_thread_switch(struct task_struct *next) if (!system_supports_fpsimd()) return; - __get_cpu_fpsimd_context(); + WARN_ON_ONCE(!irqs_disabled()); /* Save unsaved fpsimd state, if any: */ - fpsimd_save(); + if (test_thread_flag(TIF_KERNEL_FPSTATE)) + fpsimd_save_kernel_state(current); + else + fpsimd_save_user_state(); + + if (test_tsk_thread_flag(next, TIF_KERNEL_FPSTATE)) { + fpsimd_load_kernel_state(next); + set_tsk_thread_flag(next, TIF_FOREIGN_FPSTATE); + } else { + /* + * Fix up TIF_FOREIGN_FPSTATE to correctly describe next's + * state. For kernel threads, FPSIMD registers are never + * loaded with user mode FPSIMD state and so wrong_task and + * wrong_cpu will always be true. + */ + wrong_task = __this_cpu_read(fpsimd_last_state.st) != + &next->thread.uw.fpsimd_state; + wrong_cpu = next->thread.fpsimd_cpu != smp_processor_id(); + + update_tsk_thread_flag(next, TIF_FOREIGN_FPSTATE, + wrong_task || wrong_cpu); + } +} + +static void fpsimd_flush_thread_vl(enum vec_type type) +{ + int vl, supported_vl; /* - * Fix up TIF_FOREIGN_FPSTATE to correctly describe next's - * state. For kernel threads, FPSIMD registers are never loaded - * and wrong_task and wrong_cpu will always be true. + * Reset the task vector length as required. This is where we + * ensure that all user tasks have a valid vector length + * configured: no kernel task can become a user task without + * an exec and hence a call to this function. By the time the + * first call to this function is made, all early hardware + * probing is complete, so __sve_default_vl should be valid. + * If a bug causes this to go wrong, we make some noise and + * try to fudge thread.sve_vl to a safe value here. */ - wrong_task = __this_cpu_read(fpsimd_last_state.st) != - &next->thread.uw.fpsimd_state; - wrong_cpu = next->thread.fpsimd_cpu != smp_processor_id(); + vl = task_get_vl_onexec(current, type); + if (!vl) + vl = get_default_vl(type); + + if (WARN_ON(!sve_vl_valid(vl))) + vl = vl_info[type].min_vl; + + supported_vl = find_supported_vector_length(type, vl); + if (WARN_ON(supported_vl != vl)) + vl = supported_vl; - update_tsk_thread_flag(next, TIF_FOREIGN_FPSTATE, - wrong_task || wrong_cpu); + task_set_vl(current, type, vl); - __put_cpu_fpsimd_context(); + /* + * If the task is not set to inherit, ensure that the vector + * length will be reset by a subsequent exec: + */ + if (!test_thread_flag(vec_vl_inherit_flag(type))) + task_set_vl_onexec(current, type, 0); } void fpsimd_flush_thread(void) { - int vl, supported_vl; + void *sve_state = NULL; + void *sme_state = NULL; if (!system_supports_fpsimd()) return; @@ -1020,40 +1587,30 @@ void fpsimd_flush_thread(void) if (system_supports_sve()) { clear_thread_flag(TIF_SVE); - sve_free(current); - /* - * Reset the task vector length as required. - * This is where we ensure that all user tasks have a valid - * vector length configured: no kernel task can become a user - * task without an exec and hence a call to this function. - * By the time the first call to this function is made, all - * early hardware probing is complete, so sve_default_vl - * should be valid. - * If a bug causes this to go wrong, we make some noise and - * try to fudge thread.sve_vl to a safe value here. - */ - vl = current->thread.sve_vl_onexec ? - current->thread.sve_vl_onexec : sve_default_vl; + /* Defer kfree() while in atomic context */ + sve_state = current->thread.sve_state; + current->thread.sve_state = NULL; - if (WARN_ON(!sve_vl_valid(vl))) - vl = SVE_VL_MIN; + fpsimd_flush_thread_vl(ARM64_VEC_SVE); + } - supported_vl = find_supported_vector_length(vl); - if (WARN_ON(supported_vl != vl)) - vl = supported_vl; + if (system_supports_sme()) { + clear_thread_flag(TIF_SME); - current->thread.sve_vl = vl; + /* Defer kfree() while in atomic context */ + sme_state = current->thread.sme_state; + current->thread.sme_state = NULL; - /* - * If the task is not set to inherit, ensure that the vector - * length will be reset by a subsequent exec: - */ - if (!test_thread_flag(TIF_SVE_VL_INHERIT)) - current->thread.sve_vl_onexec = 0; + fpsimd_flush_thread_vl(ARM64_VEC_SME); + current->thread.svcr = 0; } + current->thread.fp_type = FP_STATE_FPSIMD; + put_cpu_fpsimd_context(); + kfree(sve_state); + kfree(sme_state); } /* @@ -1066,7 +1623,7 @@ void fpsimd_preserve_current_state(void) return; get_cpu_fpsimd_context(); - fpsimd_save(); + fpsimd_save_user_state(); put_cpu_fpsimd_context(); } @@ -1078,58 +1635,115 @@ void fpsimd_preserve_current_state(void) void fpsimd_signal_preserve_current_state(void) { fpsimd_preserve_current_state(); - if (system_supports_sve() && test_thread_flag(TIF_SVE)) + if (test_thread_flag(TIF_SVE)) sve_to_fpsimd(current); } /* + * Called by KVM when entering the guest. + */ +void fpsimd_kvm_prepare(void) +{ + if (!system_supports_sve()) + return; + + /* + * KVM does not save host SVE state since we can only enter + * the guest from a syscall so the ABI means that only the + * non-saved SVE state needs to be saved. If we have left + * SVE enabled for performance reasons then update the task + * state to be FPSIMD only. + */ + get_cpu_fpsimd_context(); + + if (test_and_clear_thread_flag(TIF_SVE)) { + sve_to_fpsimd(current); + current->thread.fp_type = FP_STATE_FPSIMD; + } + + put_cpu_fpsimd_context(); +} + +/* * Associate current's FPSIMD context with this cpu * The caller must have ownership of the cpu FPSIMD context before calling * this function. */ -void fpsimd_bind_task_to_cpu(void) +static void fpsimd_bind_task_to_cpu(void) { - struct fpsimd_last_state_struct *last = - this_cpu_ptr(&fpsimd_last_state); + struct cpu_fp_state *last = this_cpu_ptr(&fpsimd_last_state); + WARN_ON(!system_supports_fpsimd()); last->st = ¤t->thread.uw.fpsimd_state; last->sve_state = current->thread.sve_state; - last->sve_vl = current->thread.sve_vl; + last->sme_state = current->thread.sme_state; + last->sve_vl = task_get_sve_vl(current); + last->sme_vl = task_get_sme_vl(current); + last->svcr = ¤t->thread.svcr; + last->fp_type = ¤t->thread.fp_type; + last->to_save = FP_STATE_CURRENT; current->thread.fpsimd_cpu = smp_processor_id(); + /* + * Toggle SVE and SME trapping for userspace if needed, these + * are serialsied by ret_to_user(). + */ + if (system_supports_sme()) { + if (test_thread_flag(TIF_SME)) + sme_user_enable(); + else + sme_user_disable(); + } + if (system_supports_sve()) { - /* Toggle SVE trapping for userspace if needed */ if (test_thread_flag(TIF_SVE)) sve_user_enable(); else sve_user_disable(); - - /* Serialised by exception return to user */ } } -void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, - unsigned int sve_vl) +void fpsimd_bind_state_to_cpu(struct cpu_fp_state *state) { - struct fpsimd_last_state_struct *last = - this_cpu_ptr(&fpsimd_last_state); + struct cpu_fp_state *last = this_cpu_ptr(&fpsimd_last_state); + WARN_ON(!system_supports_fpsimd()); WARN_ON(!in_softirq() && !irqs_disabled()); - last->st = st; - last->sve_state = sve_state; - last->sve_vl = sve_vl; + *last = *state; } /* * Load the userland FPSIMD state of 'current' from memory, but only if the * FPSIMD state already held in the registers is /not/ the most recent FPSIMD - * state of 'current' + * state of 'current'. This is called when we are preparing to return to + * userspace to ensure that userspace sees a good register state. */ void fpsimd_restore_current_state(void) { - if (!system_supports_fpsimd()) + /* + * TIF_FOREIGN_FPSTATE is set on the init task and copied by + * arch_dup_task_struct() regardless of whether FP/SIMD is detected. + * Thus user threads can have this set even when FP/SIMD hasn't been + * detected. + * + * When FP/SIMD is detected, begin_new_exec() will set + * TIF_FOREIGN_FPSTATE via flush_thread() -> fpsimd_flush_thread(), + * and fpsimd_thread_switch() will set TIF_FOREIGN_FPSTATE when + * switching tasks. We detect FP/SIMD before we exec the first user + * process, ensuring this has TIF_FOREIGN_FPSTATE set and + * do_notify_resume() will call fpsimd_restore_current_state() to + * install the user FP/SIMD context. + * + * When FP/SIMD is not detected, nothing else will clear or set + * TIF_FOREIGN_FPSTATE prior to the first return to userspace, and + * we must clear TIF_FOREIGN_FPSTATE to avoid do_notify_resume() + * looping forever calling fpsimd_restore_current_state(). + */ + if (!system_supports_fpsimd()) { + clear_thread_flag(TIF_FOREIGN_FPSTATE); return; + } get_cpu_fpsimd_context(); @@ -1144,17 +1758,19 @@ void fpsimd_restore_current_state(void) /* * Load an updated userland FPSIMD state for 'current' from memory and set the * flag that indicates that the FPSIMD register contents are the most recent - * FPSIMD state of 'current' + * FPSIMD state of 'current'. This is used by the signal code to restore the + * register state when returning from a signal handler in FPSIMD only cases, + * any SVE context will be discarded. */ void fpsimd_update_current_state(struct user_fpsimd_state const *state) { - if (!system_supports_fpsimd()) + if (WARN_ON(!system_supports_fpsimd())) return; get_cpu_fpsimd_context(); current->thread.uw.fpsimd_state = *state; - if (system_supports_sve() && test_thread_flag(TIF_SVE)) + if (test_thread_flag(TIF_SVE)) fpsimd_to_sve(current); task_fpsimd_load(); @@ -1179,7 +1795,13 @@ void fpsimd_update_current_state(struct user_fpsimd_state const *state) void fpsimd_flush_task_state(struct task_struct *t) { t->thread.fpsimd_cpu = NR_CPUS; - + /* + * If we don't support fpsimd, bail out after we have + * reset the fpsimd_cpu for this task and clear the + * FPSTATE. + */ + if (!system_supports_fpsimd()) + return; barrier(); set_tsk_thread_flag(t, TIF_FOREIGN_FPSTATE); @@ -1193,7 +1815,17 @@ void fpsimd_flush_task_state(struct task_struct *t) */ 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 increase power + * consumption. + */ + if (system_supports_sme()) + sme_smstop(); + set_thread_flag(TIF_FOREIGN_FPSTATE); } @@ -1203,11 +1835,15 @@ static void fpsimd_flush_cpu_state(void) */ void fpsimd_save_and_flush_cpu_state(void) { + unsigned long flags; + + if (!system_supports_fpsimd()) + return; WARN_ON(preemptible()); - __get_cpu_fpsimd_context(); - fpsimd_save(); + local_irq_save(flags); + fpsimd_save_user_state(); fpsimd_flush_cpu_state(); - __put_cpu_fpsimd_context(); + local_irq_restore(flags); } #ifdef CONFIG_KERNEL_MODE_NEON @@ -1239,12 +1875,39 @@ void kernel_neon_begin(void) get_cpu_fpsimd_context(); /* Save unsaved fpsimd state, if any: */ - fpsimd_save(); + if (test_thread_flag(TIF_KERNEL_FPSTATE)) { + BUG_ON(IS_ENABLED(CONFIG_PREEMPT_RT) || !in_serving_softirq()); + fpsimd_save_kernel_state(current); + } else { + fpsimd_save_user_state(); + + /* + * Set the thread flag so that the kernel mode FPSIMD state + * will be context switched along with the rest of the task + * state. + * + * On non-PREEMPT_RT, softirqs may interrupt task level kernel + * mode FPSIMD, but the task will not be preemptible so setting + * TIF_KERNEL_FPSTATE for those would be both wrong (as it + * would mark the task context FPSIMD state as requiring a + * context switch) and unnecessary. + * + * On PREEMPT_RT, softirqs are serviced from a separate thread, + * which is scheduled as usual, and this guarantees that these + * softirqs are not interrupting use of the FPSIMD in kernel + * mode in task context. So in this case, setting the flag here + * is always appropriate. + */ + if (IS_ENABLED(CONFIG_PREEMPT_RT) || !in_serving_softirq()) + set_thread_flag(TIF_KERNEL_FPSTATE); + } /* Invalidate any task state remaining in the fpsimd regs: */ fpsimd_flush_cpu_state(); + + put_cpu_fpsimd_context(); } -EXPORT_SYMBOL(kernel_neon_begin); +EXPORT_SYMBOL_GPL(kernel_neon_begin); /* * kernel_neon_end(): give the CPU FPSIMD registers back to the current task @@ -1260,15 +1923,25 @@ void kernel_neon_end(void) if (!system_supports_fpsimd()) return; - put_cpu_fpsimd_context(); + /* + * If we are returning from a nested use of kernel mode FPSIMD, restore + * the task context kernel mode FPSIMD state. This can only happen when + * running in softirq context on non-PREEMPT_RT. + */ + if (!IS_ENABLED(CONFIG_PREEMPT_RT) && in_serving_softirq() && + test_thread_flag(TIF_KERNEL_FPSTATE)) + fpsimd_load_kernel_state(current); + else + clear_thread_flag(TIF_KERNEL_FPSTATE); } -EXPORT_SYMBOL(kernel_neon_end); +EXPORT_SYMBOL_GPL(kernel_neon_end); #ifdef CONFIG_EFI static DEFINE_PER_CPU(struct user_fpsimd_state, efi_fpsimd_state); static DEFINE_PER_CPU(bool, efi_fpsimd_state_used); static DEFINE_PER_CPU(bool, efi_sve_state_used); +static DEFINE_PER_CPU(bool, efi_sm_state); /* * EFI runtime services support functions @@ -1303,11 +1976,33 @@ void __efi_fpsimd_begin(void) */ if (system_supports_sve() && likely(efi_sve_state)) { char *sve_state = this_cpu_ptr(efi_sve_state); + bool ffr = true; + u64 svcr; __this_cpu_write(efi_sve_state_used, true); - sve_save_state(sve_state + sve_ffr_offset(sve_max_vl), - &this_cpu_ptr(&efi_fpsimd_state)->fpsr); + if (system_supports_sme()) { + svcr = read_sysreg_s(SYS_SVCR); + + __this_cpu_write(efi_sm_state, + svcr & SVCR_SM_MASK); + + /* + * Unless we have FA64 FFR does not + * exist in streaming mode. + */ + if (!system_supports_fa64()) + ffr = !(svcr & SVCR_SM_MASK); + } + + sve_save_state(sve_state + sve_ffr_offset(sve_max_vl()), + &this_cpu_ptr(&efi_fpsimd_state)->fpsr, + ffr); + + if (system_supports_sme()) + sysreg_clear_set_s(SYS_SVCR, + SVCR_SM_MASK, 0); + } else { fpsimd_save_state(this_cpu_ptr(&efi_fpsimd_state)); } @@ -1330,10 +2025,31 @@ void __efi_fpsimd_end(void) if (system_supports_sve() && likely(__this_cpu_read(efi_sve_state_used))) { char const *sve_state = this_cpu_ptr(efi_sve_state); + bool ffr = true; + + /* + * Restore streaming mode; EFI calls are + * normal function calls so should not return in + * streaming mode. + */ + if (system_supports_sme()) { + if (__this_cpu_read(efi_sm_state)) { + sysreg_clear_set_s(SYS_SVCR, + 0, + SVCR_SM_MASK); + + /* + * Unless we have FA64 FFR does not + * exist in streaming mode. + */ + if (!system_supports_fa64()) + ffr = false; + } + } - sve_load_state(sve_state + sve_ffr_offset(sve_max_vl), + sve_load_state(sve_state + sve_ffr_offset(sve_max_vl()), &this_cpu_ptr(&efi_fpsimd_state)->fpsr, - sve_vq_from_vl(sve_get_vl()) - 1); + ffr); __this_cpu_write(efi_sve_state_used, false); } else { @@ -1393,6 +2109,13 @@ static inline void fpsimd_hotplug_init(void) static inline void fpsimd_hotplug_init(void) { } #endif +void cpu_enable_fpsimd(const struct arm64_cpu_capabilities *__always_unused p) +{ + unsigned long enable = CPACR_EL1_FPEN_EL1EN | CPACR_EL1_FPEN_EL0EN; + write_sysreg(read_sysreg(CPACR_EL1) | enable, CPACR_EL1); + isb(); +} + /* * FP/SIMD support code initialisation. */ @@ -1408,6 +2131,10 @@ static int __init fpsimd_init(void) if (!cpu_have_named_feature(ASIMD)) pr_notice("Advanced SIMD is not implemented\n"); - return sve_sysctl_init(); + + sve_sysctl_init(); + sme_sysctl_init(); + + return 0; } core_initcall(fpsimd_init); |