Merge branch 'for-next/sme-fixes' into for-next/core

* for-next/sme-fixes: (35 commits)
  arm64/fpsimd: Allow CONFIG_ARM64_SME to be selected
  arm64/fpsimd: ptrace: Gracefully handle errors
  arm64/fpsimd: ptrace: Mandate SVE payload for streaming-mode state
  arm64/fpsimd: ptrace: Do not present register data for inactive mode
  arm64/fpsimd: ptrace: Save task state before generating SVE header
  arm64/fpsimd: ptrace/prctl: Ensure VL changes leave task in a valid state
  arm64/fpsimd: ptrace/prctl: Ensure VL changes do not resurrect stale data
  arm64/fpsimd: Make clone() compatible with ZA lazy saving
  arm64/fpsimd: Clear PSTATE.SM during clone()
  arm64/fpsimd: Consistently preserve FPSIMD state during clone()
  arm64/fpsimd: Remove redundant task->mm check
  arm64/fpsimd: signal: Use SMSTOP behaviour in setup_return()
  arm64/fpsimd: Add task_smstop_sm()
  arm64/fpsimd: Factor out {sve,sme}_state_size() helpers
  arm64/fpsimd: Clarify sve_sync_*() functions
  arm64/fpsimd: ptrace: Consistently handle partial writes to NT_ARM_(S)SVE
  arm64/fpsimd: signal: Consistently read FPSIMD context
  arm64/fpsimd: signal: Mandate SVE payload for streaming-mode state
  arm64/fpsimd: signal: Clear PSTATE.SM when restoring FPSIMD frame only
  arm64/fpsimd: Do not discard modified SVE state
  ...

9 files changed, 431 insertions, 428 deletions

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index 4eafc6f2fff3..de9290d52fca 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -2285,7 +2285,6 @@ config ARM64_SME
 	bool "ARM Scalable Matrix Extension support"
 	default y
 	depends on ARM64_SVE
-	depends on BROKEN
 	help
 	  The Scalable Matrix Extension (SME) is an extension to the AArch64
 	  execution state which utilises a substantial subset of the SVE
diff --git a/arch/arm64/include/asm/esr.h b/arch/arm64/include/asm/esr.h
index d1b1a33f9a8b..63aed18a933f 100644
--- a/arch/arm64/include/asm/esr.h
+++ b/arch/arm64/include/asm/esr.h
@@ -371,12 +371,14 @@
 /*
  * 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
-#define ESR_ELx_SME_ISS_ZT_DISABLED	4
+#define ESR_ELx_SME_ISS_SMTC_MASK		GENMASK(2, 0)
+#define ESR_ELx_SME_ISS_SMTC(esr)		((esr) & ESR_ELx_SME_ISS_SMTC_MASK)
+
+#define ESR_ELx_SME_ISS_SMTC_SME_DISABLED	0
+#define ESR_ELx_SME_ISS_SMTC_ILL		1
+#define ESR_ELx_SME_ISS_SMTC_SM_DISABLED	2
+#define ESR_ELx_SME_ISS_SMTC_ZA_DISABLED	3
+#define ESR_ELx_SME_ISS_SMTC_ZT_DISABLED	4
 
 /* ISS field definitions for MOPS exceptions */
 #define ESR_ELx_MOPS_ISS_MEM_INST	(UL(1) << 24)
diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h
index 564bc09b3e06..b8cf0ea43cc0 100644
--- a/arch/arm64/include/asm/fpsimd.h
+++ b/arch/arm64/include/asm/fpsimd.h
@@ -6,6 +6,7 @@
 #define __ASM_FP_H
 
 #include <asm/errno.h>
+#include <asm/percpu.h>
 #include <asm/ptrace.h>
 #include <asm/processor.h>
 #include <asm/sigcontext.h>
@@ -76,7 +77,6 @@ extern void fpsimd_load_state(struct user_fpsimd_state *state);
 extern void fpsimd_thread_switch(struct task_struct *next);
 extern void fpsimd_flush_thread(void);
 
-extern void fpsimd_signal_preserve_current_state(void);
 extern void fpsimd_preserve_current_state(void);
 extern void fpsimd_restore_current_state(void);
 extern void fpsimd_update_current_state(struct user_fpsimd_state const *state);
@@ -93,9 +93,12 @@ struct cpu_fp_state {
 	enum fp_type to_save;
 };
 
+DECLARE_PER_CPU(struct cpu_fp_state, fpsimd_last_state);
+
 extern void fpsimd_bind_state_to_cpu(struct cpu_fp_state *fp_state);
 
 extern void fpsimd_flush_task_state(struct task_struct *target);
+extern void fpsimd_save_and_flush_current_state(void);
 extern void fpsimd_save_and_flush_cpu_state(void);
 
 static inline bool thread_sm_enabled(struct thread_struct *thread)
@@ -108,6 +111,8 @@ static inline bool thread_za_enabled(struct thread_struct *thread)
 	return system_supports_sme() && (thread->svcr & SVCR_ZA_MASK);
 }
 
+extern void task_smstop_sm(struct task_struct *task);
+
 /* Maximum VL that SVE/SME VL-agnostic software can transparently support */
 #define VL_ARCH_MAX 0x100
 
@@ -195,10 +200,8 @@ struct vl_info {
 
 extern void sve_alloc(struct task_struct *task, bool flush);
 extern void fpsimd_release_task(struct task_struct *task);
-extern void fpsimd_sync_to_sve(struct task_struct *task);
-extern void fpsimd_force_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 void fpsimd_sync_from_effective_state(struct task_struct *task);
+extern void fpsimd_sync_to_effective_state_zeropad(struct task_struct *task);
 
 extern int vec_set_vector_length(struct task_struct *task, enum vec_type type,
 				 unsigned long vl, unsigned long flags);
@@ -292,14 +295,29 @@ static inline bool sve_vq_available(unsigned int vq)
 	return vq_available(ARM64_VEC_SVE, vq);
 }
 
-size_t sve_state_size(struct task_struct const *task);
+static inline size_t __sve_state_size(unsigned int sve_vl, unsigned int sme_vl)
+{
+	unsigned int vl = max(sve_vl, sme_vl);
+	return SVE_SIG_REGS_SIZE(sve_vq_from_vl(vl));
+}
+
+/*
+ * Return how many bytes of memory are required to store the full SVE
+ * state for task, given task's currently configured vector length.
+ */
+static inline size_t sve_state_size(struct task_struct const *task)
+{
+	unsigned int sve_vl = task_get_sve_vl(task);
+	unsigned int sme_vl = task_get_sme_vl(task);
+	return __sve_state_size(sve_vl, sme_vl);
+}
 
 #else /* ! CONFIG_ARM64_SVE */
 
 static inline void sve_alloc(struct task_struct *task, bool flush) { }
 static inline void fpsimd_release_task(struct task_struct *task) { }
-static inline void sve_sync_to_fpsimd(struct task_struct *task) { }
-static inline void sve_sync_from_fpsimd_zeropad(struct task_struct *task) { }
+static inline void fpsimd_sync_from_effective_state(struct task_struct *task) { }
+static inline void fpsimd_sync_to_effective_state_zeropad(struct task_struct *task) { }
 
 static inline int sve_max_virtualisable_vl(void)
 {
@@ -333,6 +351,11 @@ static inline void vec_update_vq_map(enum vec_type t) { }
 static inline int vec_verify_vq_map(enum vec_type t) { return 0; }
 static inline void sve_setup(void) { }
 
+static inline size_t __sve_state_size(unsigned int sve_vl, unsigned int sme_vl)
+{
+	return 0;
+}
+
 static inline size_t sve_state_size(struct task_struct const *task)
 {
 	return 0;
@@ -385,6 +408,16 @@ extern int sme_set_current_vl(unsigned long arg);
 extern int sme_get_current_vl(void);
 extern void sme_suspend_exit(void);
 
+static inline size_t __sme_state_size(unsigned int sme_vl)
+{
+	size_t size = ZA_SIG_REGS_SIZE(sve_vq_from_vl(sme_vl));
+
+	if (system_supports_sme2())
+		size += ZT_SIG_REG_SIZE;
+
+	return size;
+}
+
 /*
  * Return how many bytes of memory are required to store the full SME
  * specific state for task, given task's currently configured vector
@@ -392,15 +425,7 @@ extern void sme_suspend_exit(void);
  */
 static inline size_t sme_state_size(struct task_struct const *task)
 {
-	unsigned int vl = task_get_sme_vl(task);
-	size_t size;
-
-	size = ZA_SIG_REGS_SIZE(sve_vq_from_vl(vl));
-
-	if (system_supports_sme2())
-		size += ZT_SIG_REG_SIZE;
-
-	return size;
+	return __sme_state_size(task_get_sme_vl(task));
 }
 
 #else
@@ -421,6 +446,11 @@ static inline int sme_set_current_vl(unsigned long arg) { return -EINVAL; }
 static inline int sme_get_current_vl(void) { return -EINVAL; }
 static inline void sme_suspend_exit(void) { }
 
+static inline size_t __sme_state_size(unsigned int sme_vl)
+{
+	return 0;
+}
+
 static inline size_t sme_state_size(struct task_struct const *task)
 {
 	return 0;
diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c
index 7993fab0cab4..7c1970b341b8 100644
--- a/arch/arm64/kernel/entry-common.c
+++ b/arch/arm64/kernel/entry-common.c
@@ -393,20 +393,16 @@ static bool cortex_a76_erratum_1463225_debug_handler(struct pt_regs *regs)
  * As per the ABI exit SME streaming mode and clear the SVE state not
  * shared with FPSIMD on syscall entry.
  */
-static inline void fp_user_discard(void)
+static inline void fpsimd_syscall_enter(void)
 {
-	/*
-	 * If SME is active then exit streaming mode.  If ZA is active
-	 * then flush the SVE registers but leave userspace access to
-	 * both SVE and SME enabled, otherwise disable SME for the
-	 * task and fall through to disabling SVE too.  This means
-	 * that after a syscall we never have any streaming mode
-	 * register state to track, if this changes the KVM code will
-	 * need updating.
-	 */
+	/* Ensure PSTATE.SM is clear, but leave PSTATE.ZA as-is. */
 	if (system_supports_sme())
 		sme_smstop_sm();
 
+	/*
+	 * The CPU is not in streaming mode. If non-streaming SVE is not
+	 * supported, there is no SVE state that needs to be discarded.
+	 */
 	if (!system_supports_sve())
 		return;
 
@@ -416,6 +412,33 @@ static inline void fp_user_discard(void)
 		sve_vq_minus_one = sve_vq_from_vl(task_get_sve_vl(current)) - 1;
 		sve_flush_live(true, sve_vq_minus_one);
 	}
+
+	/*
+	 * Any live non-FPSIMD SVE state has been zeroed. Allow
+	 * fpsimd_save_user_state() to lazily discard SVE state until either
+	 * the live state is unbound or fpsimd_syscall_exit() is called.
+	 */
+	__this_cpu_write(fpsimd_last_state.to_save, FP_STATE_FPSIMD);
+}
+
+static __always_inline void fpsimd_syscall_exit(void)
+{
+	if (!system_supports_sve())
+		return;
+
+	/*
+	 * The current task's user FPSIMD/SVE/SME state is now bound to this
+	 * CPU. The fpsimd_last_state.to_save value is either:
+	 *
+	 * - FP_STATE_FPSIMD, if the state has not been reloaded on this CPU
+	 *   since fpsimd_syscall_enter().
+	 *
+	 * - FP_STATE_CURRENT, if the state has been reloaded on this CPU at
+	 *   any point.
+	 *
+	 * Reset this to FP_STATE_CURRENT to stop lazy discarding.
+	 */
+	__this_cpu_write(fpsimd_last_state.to_save, FP_STATE_CURRENT);
 }
 
 UNHANDLED(el1t, 64, sync)
@@ -739,10 +762,11 @@ static void noinstr el0_svc(struct pt_regs *regs)
 {
 	enter_from_user_mode(regs);
 	cortex_a76_erratum_1463225_svc_handler();
-	fp_user_discard();
+	fpsimd_syscall_enter();
 	local_daif_restore(DAIF_PROCCTX);
 	do_el0_svc(regs);
 	exit_to_user_mode(regs);
+	fpsimd_syscall_exit();
 }
 
 static void noinstr el0_fpac(struct pt_regs *regs, unsigned long esr)
diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c
index ae2ea0196030..c37f02d7194e 100644
--- a/arch/arm64/kernel/fpsimd.c
+++ b/arch/arm64/kernel/fpsimd.c
@@ -119,7 +119,7 @@
  *   whatever is in the FPSIMD registers is not saved to memory, but discarded.
  */
 
-static DEFINE_PER_CPU(struct cpu_fp_state, fpsimd_last_state);
+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
@@ -359,20 +359,15 @@ static void task_fpsimd_load(void)
 	WARN_ON(preemptible());
 	WARN_ON(test_thread_flag(TIF_KERNEL_FPSTATE));
 
-	if (system_supports_fpmr())
-		write_sysreg_s(current->thread.uw.fpmr, SYS_FPMR);
-
 	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();
+			clear_thread_flag(TIF_SVE);
 			break;
 		case FP_STATE_SVE:
-			if (!thread_sm_enabled(&current->thread) &&
-			    !WARN_ON_ONCE(!test_and_set_thread_flag(TIF_SVE)))
-				sve_user_enable();
+			if (!thread_sm_enabled(&current->thread))
+				WARN_ON_ONCE(!test_and_set_thread_flag(TIF_SVE));
 
 			if (test_thread_flag(TIF_SVE))
 				sve_set_vq(sve_vq_from_vl(task_get_sve_vl(current)) - 1);
@@ -413,6 +408,9 @@ static void task_fpsimd_load(void)
 			restore_ffr = system_supports_fa64();
 	}
 
+	if (system_supports_fpmr())
+		write_sysreg_s(current->thread.uw.fpmr, SYS_FPMR);
+
 	if (restore_sve_regs) {
 		WARN_ON_ONCE(current->thread.fp_type != FP_STATE_SVE);
 		sve_load_state(sve_pffr(&current->thread),
@@ -453,12 +451,15 @@ static void fpsimd_save_user_state(void)
 		*(last->fpmr) = read_sysreg_s(SYS_FPMR);
 
 	/*
-	 * 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.
+	 * Save SVE state if it is live.
+	 *
+	 * The syscall ABI discards live SVE state at syscall entry. When
+	 * entering a syscall, fpsimd_syscall_enter() sets to_save to
+	 * FP_STATE_FPSIMD to allow the SVE state to be lazily discarded until
+	 * either new SVE state is loaded+bound or fpsimd_syscall_exit() is
+	 * called prior to a return to userspace.
 	 */
-	if ((last->to_save == FP_STATE_CURRENT && test_thread_flag(TIF_SVE) &&
-	     !in_syscall(current_pt_regs())) ||
+	if ((last->to_save == FP_STATE_CURRENT && test_thread_flag(TIF_SVE)) ||
 	    last->to_save == FP_STATE_SVE) {
 		save_sve_regs = true;
 		save_ffr = true;
@@ -651,7 +652,7 @@ static void __fpsimd_to_sve(void *sst, struct user_fpsimd_state const *fst,
  * task->thread.uw.fpsimd_state must be up to date before calling this
  * function.
  */
-static void fpsimd_to_sve(struct task_struct *task)
+static inline void fpsimd_to_sve(struct task_struct *task)
 {
 	unsigned int vq;
 	void *sst = task->thread.sve_state;
@@ -675,7 +676,7 @@ static void fpsimd_to_sve(struct task_struct *task)
  * bytes of allocated kernel memory.
  * task->thread.sve_state must be up to date before calling this function.
  */
-static void sve_to_fpsimd(struct task_struct *task)
+static inline void sve_to_fpsimd(struct task_struct *task)
 {
 	unsigned int vq, vl;
 	void const *sst = task->thread.sve_state;
@@ -694,44 +695,39 @@ static void sve_to_fpsimd(struct task_struct *task)
 	}
 }
 
-void cpu_enable_fpmr(const struct arm64_cpu_capabilities *__always_unused p)
+static inline void __fpsimd_zero_vregs(struct user_fpsimd_state *fpsimd)
 {
-	write_sysreg_s(read_sysreg_s(SYS_SCTLR_EL1) | SCTLR_EL1_EnFPM_MASK,
-		       SYS_SCTLR_EL1);
+	memset(&fpsimd->vregs, 0, sizeof(fpsimd->vregs));
 }
 
-#ifdef CONFIG_ARM64_SVE
 /*
- * Call __sve_free() directly only if you know task can't be scheduled
- * or preempted.
+ * Simulate the effects of an SMSTOP SM instruction.
  */
-static void __sve_free(struct task_struct *task)
+void task_smstop_sm(struct task_struct *task)
 {
-	kfree(task->thread.sve_state);
-	task->thread.sve_state = NULL;
-}
+	if (!thread_sm_enabled(&task->thread))
+		return;
 
-static void sve_free(struct task_struct *task)
-{
-	WARN_ON(test_tsk_thread_flag(task, TIF_SVE));
+	__fpsimd_zero_vregs(&task->thread.uw.fpsimd_state);
+	task->thread.uw.fpsimd_state.fpsr = 0x0800009f;
+	if (system_supports_fpmr())
+		task->thread.uw.fpmr = 0;
 
-	__sve_free(task);
+	task->thread.svcr &= ~SVCR_SM_MASK;
+	task->thread.fp_type = FP_STATE_FPSIMD;
 }
 
-/*
- * Return how many bytes of memory are required to store the full SVE
- * state for task, given task's currently configured vector length.
- */
-size_t sve_state_size(struct task_struct const *task)
+void cpu_enable_fpmr(const struct arm64_cpu_capabilities *__always_unused p)
 {
-	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));
+	write_sysreg_s(read_sysreg_s(SYS_SCTLR_EL1) | SCTLR_EL1_EnFPM_MASK,
+		       SYS_SCTLR_EL1);
+}
 
-	return SVE_SIG_REGS_SIZE(sve_vq_from_vl(vl));
+#ifdef CONFIG_ARM64_SVE
+static void sve_free(struct task_struct *task)
+{
+	kfree(task->thread.sve_state);
+	task->thread.sve_state = NULL;
 }
 
 /*
@@ -758,69 +754,34 @@ void sve_alloc(struct task_struct *task, bool flush)
 		kzalloc(sve_state_size(task), GFP_KERNEL);
 }
 
-
 /*
- * Force the FPSIMD state shared with SVE to be updated in the SVE state
- * even if the SVE state is the current active state.
+ * Ensure that task->thread.uw.fpsimd_state is up to date with respect to the
+ * task's currently effective FPSIMD/SVE 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.
- *
- * 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_sync_to_sve(struct task_struct *task)
-{
-	if (!test_tsk_thread_flag(task, TIF_SVE) &&
-	    !thread_sm_enabled(&task->thread))
-		fpsimd_to_sve(task);
-}
-
-/*
- * Ensure that task->thread.uw.fpsimd_state is up to date with respect to
- * the user task, irrespective of whether SVE is in use or not.
- *
- * 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.
+ * The task's FPSIMD/SVE/SME state must not be subject to concurrent
+ * manipulation.
  */
-void sve_sync_to_fpsimd(struct task_struct *task)
+void fpsimd_sync_from_effective_state(struct task_struct *task)
 {
 	if (task->thread.fp_type == FP_STATE_SVE)
 		sve_to_fpsimd(task);
 }
 
 /*
- * Ensure that task->thread.sve_state is up to date with respect to
- * the task->thread.uw.fpsimd_state.
+ * Ensure that the task's currently effective FPSIMD/SVE state is up to date
+ * with respect to task->thread.uw.fpsimd_state, zeroing any effective
+ * non-FPSIMD (S)SVE state.
  *
- * This should only be called by ptrace to merge new FPSIMD register
- * values into a task for which SVE is currently active.
- * task must be non-runnable.
- * task->thread.sve_state must point to at least sve_state_size(task)
- * bytes of allocated kernel memory.
- * task->thread.uw.fpsimd_state must already have been initialised with
- * the new FPSIMD register values to be merged in.
+ * The task's FPSIMD/SVE/SME state must not be subject to concurrent
+ * manipulation.
  */
-void sve_sync_from_fpsimd_zeropad(struct task_struct *task)
+void fpsimd_sync_to_effective_state_zeropad(struct task_struct *task)
 {
 	unsigned int vq;
 	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) &&
-	    !thread_sm_enabled(&task->thread))
+	if (task->thread.fp_type != FP_STATE_SVE)
 		return;
 
 	vq = sve_vq_from_vl(thread_get_cur_vl(&task->thread));
@@ -829,10 +790,73 @@ void sve_sync_from_fpsimd_zeropad(struct task_struct *task)
 	__fpsimd_to_sve(sst, fst, vq);
 }
 
+static int change_live_vector_length(struct task_struct *task,
+				     enum vec_type type,
+				     unsigned long vl)
+{
+	unsigned int sve_vl = task_get_sve_vl(task);
+	unsigned int sme_vl = task_get_sme_vl(task);
+	void *sve_state = NULL, *sme_state = NULL;
+
+	if (type == ARM64_VEC_SME)
+		sme_vl = vl;
+	else
+		sve_vl = vl;
+
+	/*
+	 * Allocate the new sve_state and sme_state before freeing the old
+	 * copies so that allocation failure can be handled without needing to
+	 * mutate the task's state in any way.
+	 *
+	 * Changes to the SVE vector length must not discard live ZA state or
+	 * clear PSTATE.ZA, as userspace code which is unaware of the AAPCS64
+	 * ZA lazy saving scheme may attempt to change the SVE vector length
+	 * while unsaved/dormant ZA state exists.
+	 */
+	sve_state = kzalloc(__sve_state_size(sve_vl, sme_vl), GFP_KERNEL);
+	if (!sve_state)
+		goto out_mem;
+
+	if (type == ARM64_VEC_SME) {
+		sme_state = kzalloc(__sme_state_size(sme_vl), GFP_KERNEL);
+		if (!sme_state)
+			goto out_mem;
+	}
+
+	if (task == current)
+		fpsimd_save_and_flush_current_state();
+	else
+		fpsimd_flush_task_state(task);
+
+	/*
+	 * Always preserve PSTATE.SM and the effective FPSIMD state, zeroing
+	 * other SVE state.
+	 */
+	fpsimd_sync_from_effective_state(task);
+	task_set_vl(task, type, vl);
+	kfree(task->thread.sve_state);
+	task->thread.sve_state = sve_state;
+	fpsimd_sync_to_effective_state_zeropad(task);
+
+	if (type == ARM64_VEC_SME) {
+		task->thread.svcr &= ~SVCR_ZA_MASK;
+		kfree(task->thread.sme_state);
+		task->thread.sme_state = sme_state;
+	}
+
+	return 0;
+
+out_mem:
+	kfree(sve_state);
+	kfree(sme_state);
+	return -ENOMEM;
+}
+
 int vec_set_vector_length(struct task_struct *task, enum vec_type type,
 			  unsigned long vl, unsigned long flags)
 {
-	bool free_sme = false;
+	bool onexec = flags & PR_SVE_SET_VL_ONEXEC;
+	bool inherit = flags & PR_SVE_VL_INHERIT;
 
 	if (flags & ~(unsigned long)(PR_SVE_VL_INHERIT |
 				     PR_SVE_SET_VL_ONEXEC))
@@ -852,71 +876,17 @@ int vec_set_vector_length(struct task_struct *task, enum vec_type type,
 
 	vl = find_supported_vector_length(type, vl);
 
-	if (flags & (PR_SVE_VL_INHERIT |
-		     PR_SVE_SET_VL_ONEXEC))
+	if (!onexec && vl != task_get_vl(task, type)) {
+		if (change_live_vector_length(task, type, vl))
+			return -ENOMEM;
+	}
+
+	if (onexec || inherit)
 		task_set_vl_onexec(task, type, vl);
 	else
 		/* Reset VL to system default on next exec: */
 		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_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
-	 * regular FPSIMD thread.
-	 */
-	if (task == current) {
-		get_cpu_fpsimd_context();
-
-		fpsimd_save_user_state();
-	}
-
-	fpsimd_flush_task_state(task);
-	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);
-
-	/*
-	 * 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);
-
-	if (free_sme)
-		sme_free(task);
-
-out:
 	update_tsk_thread_flag(task, vec_vl_inherit_flag(type),
 			       flags & PR_SVE_VL_INHERIT);
 
@@ -1212,7 +1182,7 @@ void __init sve_setup(void)
  */
 void fpsimd_release_task(struct task_struct *dead_task)
 {
-	__sve_free(dead_task);
+	sve_free(dead_task);
 	sme_free(dead_task);
 }
 
@@ -1436,7 +1406,7 @@ void do_sme_acc(unsigned long esr, struct pt_regs *regs)
 	 * 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) {
+	if (ESR_ELx_SME_ISS_SMTC(esr) != ESR_ELx_SME_ISS_SMTC_SME_DISABLED) {
 		force_signal_inject(SIGILL, ILL_ILLOPC, regs->pc, 0);
 		return;
 	}
@@ -1460,6 +1430,8 @@ void do_sme_acc(unsigned long esr, struct pt_regs *regs)
 		sme_set_vq(vq_minus_one);
 
 		fpsimd_bind_task_to_cpu();
+	} else {
+		fpsimd_flush_task_state(current);
 	}
 
 	put_cpu_fpsimd_context();
@@ -1573,8 +1545,8 @@ void fpsimd_thread_switch(struct task_struct *next)
 		fpsimd_save_user_state();
 
 	if (test_tsk_thread_flag(next, TIF_KERNEL_FPSTATE)) {
-		fpsimd_load_kernel_state(next);
 		fpsimd_flush_cpu_state();
+		fpsimd_load_kernel_state(next);
 	} else {
 		/*
 		 * Fix up TIF_FOREIGN_FPSTATE to correctly describe next's
@@ -1661,6 +1633,9 @@ void fpsimd_flush_thread(void)
 		current->thread.svcr = 0;
 	}
 
+	if (system_supports_fpmr())
+		current->thread.uw.fpmr = 0;
+
 	current->thread.fp_type = FP_STATE_FPSIMD;
 
 	put_cpu_fpsimd_context();
@@ -1683,18 +1658,6 @@ void fpsimd_preserve_current_state(void)
 }
 
 /*
- * Like fpsimd_preserve_current_state(), but ensure that
- * current->thread.uw.fpsimd_state is updated so that it can be copied to
- * the signal frame.
- */
-void fpsimd_signal_preserve_current_state(void)
-{
-	fpsimd_preserve_current_state();
-	if (current->thread.fp_type == FP_STATE_SVE)
-		sve_to_fpsimd(current);
-}
-
-/*
  * Associate current's FPSIMD context with this cpu
  * The caller must have ownership of the cpu FPSIMD context before calling
  * this function.
@@ -1786,30 +1749,14 @@ void fpsimd_restore_current_state(void)
 	put_cpu_fpsimd_context();
 }
 
-/*
- * 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'. 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 (WARN_ON(!system_supports_fpsimd()))
 		return;
 
-	get_cpu_fpsimd_context();
-
 	current->thread.uw.fpsimd_state = *state;
-	if (test_thread_flag(TIF_SVE))
+	if (current->thread.fp_type == FP_STATE_SVE)
 		fpsimd_to_sve(current);
-
-	task_fpsimd_load();
-	fpsimd_bind_task_to_cpu();
-
-	clear_thread_flag(TIF_FOREIGN_FPSTATE);
-
-	put_cpu_fpsimd_context();
 }
 
 /*
@@ -1839,6 +1786,17 @@ void fpsimd_flush_task_state(struct task_struct *t)
 	barrier();
 }
 
+void fpsimd_save_and_flush_current_state(void)
+{
+	if (!system_supports_fpsimd())
+		return;
+
+	get_cpu_fpsimd_context();
+	fpsimd_save_user_state();
+	fpsimd_flush_task_state(current);
+	put_cpu_fpsimd_context();
+}
+
 /*
  * Save the FPSIMD state to memory and invalidate cpu view.
  * This function must be called with preemption disabled.
diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
index 45a55fe81788..a5ca15daeb8a 100644
--- a/arch/arm64/kernel/process.c
+++ b/arch/arm64/kernel/process.c
@@ -344,50 +344,34 @@ void arch_release_task_struct(struct task_struct *tsk)
 
 int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
 {
-	if (current->mm)
-		fpsimd_preserve_current_state();
+	/*
+	 * The current/src task's FPSIMD state may or may not be live, and may
+	 * have been altered by ptrace after entry to the kernel. Save the
+	 * effective FPSIMD state so that this will be copied into dst.
+	 */
+	fpsimd_save_and_flush_current_state();
+	fpsimd_sync_from_effective_state(src);
+
 	*dst = *src;
 
 	/*
-	 * Detach src's sve_state (if any) from dst so that it does not
-	 * get erroneously used or freed prematurely.  dst's copies
-	 * will be allocated on demand later on if dst uses SVE.
-	 * For consistency, also clear TIF_SVE here: this could be done
-	 * later in copy_process(), but to avoid tripping up future
-	 * maintainers it is best not to leave TIF flags and buffers in
-	 * an inconsistent state, even temporarily.
+	 * Drop stale reference to src's sve_state and convert dst to
+	 * non-streaming FPSIMD mode.
 	 */
+	dst->thread.fp_type = FP_STATE_FPSIMD;
 	dst->thread.sve_state = NULL;
 	clear_tsk_thread_flag(dst, TIF_SVE);
+	task_smstop_sm(dst);
 
 	/*
-	 * In the unlikely event that we create a new thread with ZA
-	 * enabled we should retain the ZA and ZT state so duplicate
-	 * it here.  This may be shortly freed if we exec() or if
-	 * CLONE_SETTLS but it's simpler to do it here. To avoid
-	 * confusing the rest of the code ensure that we have a
-	 * sve_state allocated whenever sme_state is allocated.
+	 * Drop stale reference to src's sme_state and ensure dst has ZA
+	 * disabled.
+	 *
+	 * When necessary, ZA will be inherited later in copy_thread_za().
 	 */
-	if (thread_za_enabled(&src->thread)) {
-		dst->thread.sve_state = kzalloc(sve_state_size(src),
-						GFP_KERNEL);
-		if (!dst->thread.sve_state)
-			return -ENOMEM;
-
-		dst->thread.sme_state = kmemdup(src->thread.sme_state,
-						sme_state_size(src),
-						GFP_KERNEL);
-		if (!dst->thread.sme_state) {
-			kfree(dst->thread.sve_state);
-			dst->thread.sve_state = NULL;
-			return -ENOMEM;
-		}
-	} else {
-		dst->thread.sme_state = NULL;
-		clear_tsk_thread_flag(dst, TIF_SME);
-	}
-
-	dst->thread.fp_type = FP_STATE_FPSIMD;
+	dst->thread.sme_state = NULL;
+	clear_tsk_thread_flag(dst, TIF_SME);
+	dst->thread.svcr &= ~SVCR_ZA_MASK;
 
 	/* clear any pending asynchronous tag fault raised by the parent */
 	clear_tsk_thread_flag(dst, TIF_MTE_ASYNC_FAULT);
@@ -395,6 +379,31 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
 	return 0;
 }
 
+static int copy_thread_za(struct task_struct *dst, struct task_struct *src)
+{
+	if (!thread_za_enabled(&src->thread))
+		return 0;
+
+	dst->thread.sve_state = kzalloc(sve_state_size(src),
+					GFP_KERNEL);
+	if (!dst->thread.sve_state)
+		return -ENOMEM;
+
+	dst->thread.sme_state = kmemdup(src->thread.sme_state,
+					sme_state_size(src),
+					GFP_KERNEL);
+	if (!dst->thread.sme_state) {
+		kfree(dst->thread.sve_state);
+		dst->thread.sve_state = NULL;
+		return -ENOMEM;
+	}
+
+	set_tsk_thread_flag(dst, TIF_SME);
+	dst->thread.svcr |= SVCR_ZA_MASK;
+
+	return 0;
+}
+
 asmlinkage void ret_from_fork(void) asm("ret_from_fork");
 
 int copy_thread(struct task_struct *p, const struct kernel_clone_args *args)
@@ -427,8 +436,6 @@ int copy_thread(struct task_struct *p, const struct kernel_clone_args *args)
 		 * out-of-sync with the saved value.
 		 */
 		*task_user_tls(p) = read_sysreg(tpidr_el0);
-		if (system_supports_tpidr2())
-			p->thread.tpidr2_el0 = read_sysreg_s(SYS_TPIDR2_EL0);
 
 		if (system_supports_poe())
 			p->thread.por_el0 = read_sysreg_s(SYS_POR_EL0);
@@ -441,13 +448,39 @@ int copy_thread(struct task_struct *p, const struct kernel_clone_args *args)
 		}
 
 		/*
+		 * Due to the AAPCS64 "ZA lazy saving scheme", PSTATE.ZA and
+		 * TPIDR2 need to be manipulated as a pair, and either both
+		 * need to be inherited or both need to be reset.
+		 *
+		 * Within a process, child threads must not inherit their
+		 * parent's TPIDR2 value or they may clobber their parent's
+		 * stack at some later point.
+		 *
+		 * When a process is fork()'d, the child must inherit ZA and
+		 * TPIDR2 from its parent in case there was dormant ZA state.
+		 *
+		 * Use CLONE_VM to determine when the child will share the
+		 * address space with the parent, and cannot safely inherit the
+		 * state.
+		 */
+		if (system_supports_sme()) {
+			if (!(clone_flags & CLONE_VM)) {
+				p->thread.tpidr2_el0 = read_sysreg_s(SYS_TPIDR2_EL0);
+				ret = copy_thread_za(p, current);
+				if (ret)
+					return ret;
+			} else {
+				p->thread.tpidr2_el0 = 0;
+				WARN_ON_ONCE(p->thread.svcr & SVCR_ZA_MASK);
+			}
+		}
+
+		/*
 		 * If a TLS pointer was passed to clone, use it for the new
-		 * thread.  We also reset TPIDR2 if it's in use.
+		 * thread.
 		 */
-		if (clone_flags & CLONE_SETTLS) {
+		if (clone_flags & CLONE_SETTLS)
 			p->thread.uw.tp_value = tls;
-			p->thread.tpidr2_el0 = 0;
-		}
 
 		ret = copy_thread_gcs(p, args);
 		if (ret != 0)
diff --git a/arch/arm64/kernel/ptrace.c b/arch/arm64/kernel/ptrace.c
index f79b0d5f71ac..a360e52db02f 100644
--- a/arch/arm64/kernel/ptrace.c
+++ b/arch/arm64/kernel/ptrace.c
@@ -594,7 +594,7 @@ static int __fpr_get(struct task_struct *target,
 {
 	struct user_fpsimd_state *uregs;
 
-	sve_sync_to_fpsimd(target);
+	fpsimd_sync_from_effective_state(target);
 
 	uregs = &target->thread.uw.fpsimd_state;
 
@@ -626,7 +626,7 @@ static int __fpr_set(struct task_struct *target,
 	 * Ensure target->thread.uw.fpsimd_state is up to date, so that a
 	 * short copyin can't resurrect stale data.
 	 */
-	sve_sync_to_fpsimd(target);
+	fpsimd_sync_from_effective_state(target);
 
 	newstate = target->thread.uw.fpsimd_state;
 
@@ -653,7 +653,7 @@ static int fpr_set(struct task_struct *target, const struct user_regset *regset,
 	if (ret)
 		return ret;
 
-	sve_sync_from_fpsimd_zeropad(target);
+	fpsimd_sync_to_effective_state_zeropad(target);
 	fpsimd_flush_task_state(target);
 
 	return ret;
@@ -775,6 +775,11 @@ static void sve_init_header_from_task(struct user_sve_header *header,
 		task_type = ARM64_VEC_SVE;
 	active = (task_type == type);
 
+	if (active && target->thread.fp_type == FP_STATE_SVE)
+		header->flags = SVE_PT_REGS_SVE;
+	else
+		header->flags = SVE_PT_REGS_FPSIMD;
+
 	switch (type) {
 	case ARM64_VEC_SVE:
 		if (test_tsk_thread_flag(target, TIF_SVE_VL_INHERIT))
@@ -789,19 +794,14 @@ static void sve_init_header_from_task(struct user_sve_header *header,
 		return;
 	}
 
-	if (active) {
-		if (target->thread.fp_type == FP_STATE_FPSIMD) {
-			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 = vec_max_vl(type);
-	header->size = SVE_PT_SIZE(vq, header->flags);
+	if (active)
+		header->size = SVE_PT_SIZE(vq, header->flags);
+	else
+		header->size = sizeof(header);
 	header->max_size = SVE_PT_SIZE(sve_vq_from_vl(header->max_vl),
 				      SVE_PT_REGS_SVE);
 }
@@ -820,18 +820,25 @@ static int sve_get_common(struct task_struct *target,
 	unsigned int vq;
 	unsigned long start, end;
 
+	if (target == current)
+		fpsimd_preserve_current_state();
+
 	/* Header */
 	sve_init_header_from_task(&header, target, type);
 	vq = sve_vq_from_vl(header.vl);
 
 	membuf_write(&to, &header, sizeof(header));
 
-	if (target == current)
-		fpsimd_preserve_current_state();
-
 	BUILD_BUG_ON(SVE_PT_FPSIMD_OFFSET != sizeof(header));
 	BUILD_BUG_ON(SVE_PT_SVE_OFFSET != sizeof(header));
 
+	/*
+	 * When the requested vector type is not active, do not present data
+	 * from the other mode to userspace.
+	 */
+	if (header.size == sizeof(header))
+		return 0;
+
 	switch ((header.flags & SVE_PT_REGS_MASK)) {
 	case SVE_PT_REGS_FPSIMD:
 		return __fpr_get(target, regset, to);
@@ -859,7 +866,7 @@ static int sve_get_common(struct task_struct *target,
 		return membuf_zero(&to, end - start);
 
 	default:
-		return 0;
+		BUILD_BUG();
 	}
 }
 
@@ -883,6 +890,9 @@ static int sve_set_common(struct task_struct *target,
 	struct user_sve_header header;
 	unsigned int vq;
 	unsigned long start, end;
+	bool fpsimd;
+
+	fpsimd_flush_task_state(target);
 
 	/* Header */
 	if (count < sizeof(header))
@@ -890,7 +900,16 @@ static int sve_set_common(struct task_struct *target,
 	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &header,
 				 0, sizeof(header));
 	if (ret)
-		goto out;
+		return ret;
+
+	/*
+	 * Streaming SVE data is always stored and presented in SVE format.
+	 * Require the user to provide SVE formatted data for consistency, and
+	 * to avoid the risk that we configure the task into an invalid state.
+	 */
+	fpsimd = (header.flags & SVE_PT_REGS_MASK) == SVE_PT_REGS_FPSIMD;
+	if (fpsimd && type == ARM64_VEC_SME)
+		return -EINVAL;
 
 	/*
 	 * Apart from SVE_PT_REGS_MASK, all SVE_PT_* flags are consumed by
@@ -899,7 +918,21 @@ static int sve_set_common(struct task_struct *target,
 	ret = vec_set_vector_length(target, type, header.vl,
 		((unsigned long)header.flags & ~SVE_PT_REGS_MASK) << 16);
 	if (ret)
-		goto out;
+		return ret;
+
+	/* Allocate SME storage if necessary, preserving any existing ZA/ZT state */
+	if (type == ARM64_VEC_SME) {
+		sme_alloc(target, false);
+		if (!target->thread.sme_state)
+			return -ENOMEM;
+	}
+
+	/* Allocate SVE storage if necessary, zeroing any existing SVE state */
+	if (!fpsimd) {
+		sve_alloc(target, true);
+		if (!target->thread.sve_state)
+			return -ENOMEM;
+	}
 
 	/*
 	 * Actual VL set may be different from what the user asked
@@ -910,81 +943,47 @@ static int sve_set_common(struct task_struct *target,
 
 	/* Enter/exit streaming mode */
 	if (system_supports_sme()) {
-		u64 old_svcr = target->thread.svcr;
-
 		switch (type) {
 		case ARM64_VEC_SVE:
 			target->thread.svcr &= ~SVCR_SM_MASK;
+			set_tsk_thread_flag(target, TIF_SVE);
 			break;
 		case ARM64_VEC_SME:
 			target->thread.svcr |= SVCR_SM_MASK;
-
-			/*
-			 * Disable traps and ensure there is SME storage but
-			 * preserve any currently set values in ZA/ZT.
-			 */
-			sme_alloc(target, false);
 			set_tsk_thread_flag(target, TIF_SME);
 			break;
 		default:
 			WARN_ON_ONCE(1);
-			ret = -EINVAL;
-			goto out;
+			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, true);
 	}
 
+	/* Always zero V regs, FPSR, and FPCR */
+	memset(&current->thread.uw.fpsimd_state, 0,
+	       sizeof(current->thread.uw.fpsimd_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) {
-		ret = __fpr_set(target, regset, pos, count, kbuf, ubuf,
-				SVE_PT_FPSIMD_OFFSET);
+	if (fpsimd) {
 		clear_tsk_thread_flag(target, TIF_SVE);
 		target->thread.fp_type = FP_STATE_FPSIMD;
-		goto out;
+		ret = __fpr_set(target, regset, pos, count, kbuf, ubuf,
+				SVE_PT_FPSIMD_OFFSET);
+		return ret;
 	}
 
-	/*
-	 * Otherwise: no registers or full SVE case.  For backwards
-	 * compatibility reasons we treat empty flags as SVE registers.
-	 */
+	/* Otherwise: no registers or full SVE case. */
+
+	target->thread.fp_type = FP_STATE_SVE;
 
 	/*
 	 * If setting a different VL from the requested VL and there is
 	 * register data, the data layout will be wrong: don't even
 	 * try to set the registers in this case.
 	 */
-	if (count && vq != sve_vq_from_vl(header.vl)) {
-		ret = -EIO;
-		goto out;
-	}
-
-	sve_alloc(target, true);
-	if (!target->thread.sve_state) {
-		ret = -ENOMEM;
-		clear_tsk_thread_flag(target, TIF_SVE);
-		target->thread.fp_type = FP_STATE_FPSIMD;
-		goto out;
-	}
-
-	/*
-	 * Ensure target->thread.sve_state is up to date with target's
-	 * FPSIMD regs, so that a short copyin leaves trailing
-	 * registers unmodified.  Only enable SVE if we are
-	 * configuring normal SVE, a system with streaming SVE may not
-	 * have normal SVE.
-	 */
-	fpsimd_sync_to_sve(target);
-	if (type == ARM64_VEC_SVE)
-		set_tsk_thread_flag(target, TIF_SVE);
-	target->thread.fp_type = FP_STATE_SVE;
+	if (count && vq != sve_vq_from_vl(header.vl))
+		return -EIO;
 
 	BUILD_BUG_ON(SVE_PT_SVE_OFFSET != sizeof(header));
 	start = SVE_PT_SVE_OFFSET;
@@ -993,7 +992,7 @@ static int sve_set_common(struct task_struct *target,
 				 target->thread.sve_state,
 				 start, end);
 	if (ret)
-		goto out;
+		return ret;
 
 	start = end;
 	end = SVE_PT_SVE_FPSR_OFFSET(vq);
@@ -1009,8 +1008,6 @@ static int sve_set_common(struct task_struct *target,
 				 &target->thread.uw.fpsimd_state.fpsr,
 				 start, end);
 
-out:
-	fpsimd_flush_task_state(target);
 	return ret;
 }
 
diff --git a/arch/arm64/kernel/signal.c b/arch/arm64/kernel/signal.c
index a7c37afb4ebe..417140cd399b 100644
--- a/arch/arm64/kernel/signal.c
+++ b/arch/arm64/kernel/signal.c
@@ -250,6 +250,8 @@ static int preserve_fpsimd_context(struct fpsimd_context __user *ctx)
 		&current->thread.uw.fpsimd_state;
 	int err;
 
+	fpsimd_sync_from_effective_state(current);
+
 	/* copy the FP and status/control registers */
 	err = __copy_to_user(ctx->vregs, fpsimd->vregs, sizeof(fpsimd->vregs));
 	__put_user_error(fpsimd->fpsr, &ctx->fpsr, err);
@@ -262,37 +264,46 @@ static int preserve_fpsimd_context(struct fpsimd_context __user *ctx)
 	return err ? -EFAULT : 0;
 }
 
-static int restore_fpsimd_context(struct user_ctxs *user)
+static int read_fpsimd_context(struct user_fpsimd_state *fpsimd,
+			       struct user_ctxs *user)
 {
-	struct user_fpsimd_state fpsimd;
-	int err = 0;
+	int err;
 
 	/* check the size information */
 	if (user->fpsimd_size != sizeof(struct fpsimd_context))
 		return -EINVAL;
 
 	/* copy the FP and status/control registers */
-	err = __copy_from_user(fpsimd.vregs, &(user->fpsimd->vregs),
-			       sizeof(fpsimd.vregs));
-	__get_user_error(fpsimd.fpsr, &(user->fpsimd->fpsr), err);
-	__get_user_error(fpsimd.fpcr, &(user->fpsimd->fpcr), err);
+	err = __copy_from_user(fpsimd->vregs, &(user->fpsimd->vregs),
+			       sizeof(fpsimd->vregs));
+	__get_user_error(fpsimd->fpsr, &(user->fpsimd->fpsr), err);
+	__get_user_error(fpsimd->fpcr, &(user->fpsimd->fpcr), err);
+
+	return err ? -EFAULT : 0;
+}
+
+static int restore_fpsimd_context(struct user_ctxs *user)
+{
+	struct user_fpsimd_state fpsimd;
+	int err;
+
+	err = read_fpsimd_context(&fpsimd, user);
+	if (err)
+		return err;
 
 	clear_thread_flag(TIF_SVE);
+	current->thread.svcr &= ~SVCR_SM_MASK;
 	current->thread.fp_type = FP_STATE_FPSIMD;
 
 	/* load the hardware registers from the fpsimd_state structure */
-	if (!err)
-		fpsimd_update_current_state(&fpsimd);
-
-	return err ? -EFAULT : 0;
+	fpsimd_update_current_state(&fpsimd);
+	return 0;
 }
 
 static int preserve_fpmr_context(struct fpmr_context __user *ctx)
 {
 	int err = 0;
 
-	current->thread.uw.fpmr = read_sysreg_s(SYS_FPMR);
-
 	__put_user_error(FPMR_MAGIC, &ctx->head.magic, err);
 	__put_user_error(sizeof(*ctx), &ctx->head.size, err);
 	__put_user_error(current->thread.uw.fpmr, &ctx->fpmr, err);
@@ -310,7 +321,7 @@ static int restore_fpmr_context(struct user_ctxs *user)
 
 	__get_user_error(fpmr, &user->fpmr->fpmr, err);
 	if (!err)
-		write_sysreg_s(fpmr, SYS_FPMR);
+		current->thread.uw.fpmr = fpmr;
 
 	return err;
 }
@@ -372,11 +383,6 @@ static int preserve_sve_context(struct sve_context __user *ctx)
 	err |= __copy_to_user(&ctx->__reserved, reserved, sizeof(reserved));
 
 	if (vq) {
-		/*
-		 * This assumes that the SVE state has already been saved to
-		 * the task struct by calling the function
-		 * fpsimd_signal_preserve_current_state().
-		 */
 		err |= __copy_to_user((char __user *)ctx + SVE_SIG_REGS_OFFSET,
 				      current->thread.sve_state,
 				      SVE_SIG_REGS_SIZE(vq));
@@ -391,6 +397,7 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user)
 	unsigned int vl, vq;
 	struct user_fpsimd_state fpsimd;
 	u16 user_vl, flags;
+	bool sm;
 
 	if (user->sve_size < sizeof(*user->sve))
 		return -EINVAL;
@@ -400,7 +407,8 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user)
 	if (err)
 		return err;
 
-	if (flags & SVE_SIG_FLAG_SM) {
+	sm = flags & SVE_SIG_FLAG_SM;
+	if (sm) {
 		if (!system_supports_sme())
 			return -EINVAL;
 
@@ -420,28 +428,23 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user)
 	if (user_vl != vl)
 		return -EINVAL;
 
-	if (user->sve_size == sizeof(*user->sve)) {
-		clear_thread_flag(TIF_SVE);
-		current->thread.svcr &= ~SVCR_SM_MASK;
-		current->thread.fp_type = FP_STATE_FPSIMD;
-		goto fpsimd_only;
-	}
+	/*
+	 * Non-streaming SVE state may be preserved without an SVE payload, in
+	 * which case the SVE context only has a header with VL==0, and all
+	 * state can be restored from the FPSIMD context.
+	 *
+	 * Streaming SVE state is always preserved with an SVE payload. For
+	 * consistency and robustness, reject restoring streaming SVE state
+	 * without an SVE payload.
+	 */
+	if (!sm && user->sve_size == sizeof(*user->sve))
+		return restore_fpsimd_context(user);
 
 	vq = sve_vq_from_vl(vl);
 
 	if (user->sve_size < SVE_SIG_CONTEXT_SIZE(vq))
 		return -EINVAL;
 
-	/*
-	 * Careful: we are about __copy_from_user() directly into
-	 * thread.sve_state with preemption enabled, so protection is
-	 * needed to prevent a racing context switch from writing stale
-	 * registers back over the new data.
-	 */
-
-	fpsimd_flush_task_state(current);
-	/* From now, fpsimd_thread_switch() won't touch thread.sve_state */
-
 	sve_alloc(current, true);
 	if (!current->thread.sve_state) {
 		clear_thread_flag(TIF_SVE);
@@ -461,19 +464,14 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user)
 		set_thread_flag(TIF_SVE);
 	current->thread.fp_type = FP_STATE_SVE;
 
-fpsimd_only:
-	/* copy the FP and status/control registers */
-	/* restore_sigframe() already checked that user->fpsimd != NULL. */
-	err = __copy_from_user(fpsimd.vregs, user->fpsimd->vregs,
-			       sizeof(fpsimd.vregs));
-	__get_user_error(fpsimd.fpsr, &user->fpsimd->fpsr, err);
-	__get_user_error(fpsimd.fpcr, &user->fpsimd->fpcr, err);
+	err = read_fpsimd_context(&fpsimd, user);
+	if (err)
+		return err;
 
-	/* load the hardware registers from the fpsimd_state structure */
-	if (!err)
-		fpsimd_update_current_state(&fpsimd);
+	/* Merge the FPSIMD registers into the SVE state */
+	fpsimd_update_current_state(&fpsimd);
 
-	return err ? -EFAULT : 0;
+	return 0;
 }
 
 #else /* ! CONFIG_ARM64_SVE */
@@ -493,13 +491,12 @@ extern int preserve_sve_context(void __user *ctx);
 
 static int preserve_tpidr2_context(struct tpidr2_context __user *ctx)
 {
+	u64 tpidr2_el0 = read_sysreg_s(SYS_TPIDR2_EL0);
 	int err = 0;
 
-	current->thread.tpidr2_el0 = read_sysreg_s(SYS_TPIDR2_EL0);
-
 	__put_user_error(TPIDR2_MAGIC, &ctx->head.magic, err);
 	__put_user_error(sizeof(*ctx), &ctx->head.size, err);
-	__put_user_error(current->thread.tpidr2_el0, &ctx->tpidr2, err);
+	__put_user_error(tpidr2_el0, &ctx->tpidr2, err);
 
 	return err;
 }
@@ -541,11 +538,6 @@ static int preserve_za_context(struct za_context __user *ctx)
 	err |= __copy_to_user(&ctx->__reserved, reserved, sizeof(reserved));
 
 	if (vq) {
-		/*
-		 * This assumes that the ZA state has already been saved to
-		 * the task struct by calling the function
-		 * fpsimd_signal_preserve_current_state().
-		 */
 		err |= __copy_to_user((char __user *)ctx + ZA_SIG_REGS_OFFSET,
 				      current->thread.sme_state,
 				      ZA_SIG_REGS_SIZE(vq));
@@ -580,16 +572,6 @@ static int restore_za_context(struct user_ctxs *user)
 	if (user->za_size < ZA_SIG_CONTEXT_SIZE(vq))
 		return -EINVAL;
 
-	/*
-	 * Careful: we are about __copy_from_user() directly into
-	 * thread.sme_state with preemption enabled, so protection is
-	 * needed to prevent a racing context switch from writing stale
-	 * registers back over the new data.
-	 */
-
-	fpsimd_flush_task_state(current);
-	/* From now, fpsimd_thread_switch() won't touch thread.sve_state */
-
 	sme_alloc(current, true);
 	if (!current->thread.sme_state) {
 		current->thread.svcr &= ~SVCR_ZA_MASK;
@@ -627,11 +609,6 @@ static int preserve_zt_context(struct zt_context __user *ctx)
 	BUILD_BUG_ON(sizeof(ctx->__reserved) != sizeof(reserved));
 	err |= __copy_to_user(&ctx->__reserved, reserved, sizeof(reserved));
 
-	/*
-	 * This assumes that the ZT state has already been saved to
-	 * the task struct by calling the function
-	 * fpsimd_signal_preserve_current_state().
-	 */
 	err |= __copy_to_user((char __user *)ctx + ZT_SIG_REGS_OFFSET,
 			      thread_zt_state(&current->thread),
 			      ZT_SIG_REGS_SIZE(1));
@@ -657,16 +634,6 @@ static int restore_zt_context(struct user_ctxs *user)
 	if (nregs != 1)
 		return -EINVAL;
 
-	/*
-	 * Careful: we are about __copy_from_user() directly into
-	 * thread.zt_state with preemption enabled, so protection is
-	 * needed to prevent a racing context switch from writing stale
-	 * registers back over the new data.
-	 */
-
-	fpsimd_flush_task_state(current);
-	/* From now, fpsimd_thread_switch() won't touch ZT in thread state */
-
 	err = __copy_from_user(thread_zt_state(&current->thread),
 			       (char __user const *)user->zt +
 					ZT_SIG_REGS_OFFSET,
@@ -1017,6 +984,8 @@ static int restore_sigframe(struct pt_regs *regs,
 	 */
 	forget_syscall(regs);
 
+	fpsimd_save_and_flush_current_state();
+
 	err |= !valid_user_regs(&regs->user_regs, current);
 	if (err == 0)
 		err = parse_user_sigframe(&user, sf);
@@ -1507,21 +1476,9 @@ static int setup_return(struct pt_regs *regs, struct ksignal *ksig,
 
 	/* Signal handlers are invoked with ZA and streaming mode disabled */
 	if (system_supports_sme()) {
-		/*
-		 * If we were in streaming mode the saved register
-		 * state was SVE but we will exit SM and use the
-		 * FPSIMD register state - flush the saved FPSIMD
-		 * register state in case it gets loaded.
-		 */
-		if (current->thread.svcr & SVCR_SM_MASK) {
-			memset(&current->thread.uw.fpsimd_state, 0,
-			       sizeof(current->thread.uw.fpsimd_state));
-			current->thread.fp_type = FP_STATE_FPSIMD;
-		}
-
-		current->thread.svcr &= ~(SVCR_ZA_MASK |
-					  SVCR_SM_MASK);
-		sme_smstop();
+		task_smstop_sm(current);
+		current->thread.svcr &= ~SVCR_ZA_MASK;
+		write_sysreg_s(0, SYS_TPIDR2_EL0);
 	}
 
 	return 0;
@@ -1535,7 +1492,7 @@ static int setup_rt_frame(int usig, struct ksignal *ksig, sigset_t *set,
 	struct user_access_state ua_state;
 	int err = 0;
 
-	fpsimd_signal_preserve_current_state();
+	fpsimd_save_and_flush_current_state();
 
 	if (get_sigframe(&user, ksig, regs))
 		return 1;
diff --git a/arch/arm64/kernel/signal32.c b/arch/arm64/kernel/signal32.c
index 81e798b6dada..bb3b526ff43f 100644
--- a/arch/arm64/kernel/signal32.c
+++ b/arch/arm64/kernel/signal32.c
@@ -103,7 +103,7 @@ static int compat_preserve_vfp_context(struct compat_vfp_sigframe __user *frame)
 	 * Note that this also saves V16-31, which aren't visible
 	 * in AArch32.
 	 */
-	fpsimd_signal_preserve_current_state();
+	fpsimd_save_and_flush_current_state();
 
 	/* Place structure header on the stack */
 	__put_user_error(magic, &frame->magic, err);
@@ -169,14 +169,17 @@ static int compat_restore_vfp_context(struct compat_vfp_sigframe __user *frame)
 	fpsimd.fpsr = fpscr & VFP_FPSCR_STAT_MASK;
 	fpsimd.fpcr = fpscr & VFP_FPSCR_CTRL_MASK;
 
+	if (err)
+		return -EFAULT;
+
 	/*
 	 * We don't need to touch the exception register, so
 	 * reload the hardware state.
 	 */
-	if (!err)
-		fpsimd_update_current_state(&fpsimd);
+	fpsimd_save_and_flush_current_state();
+	current->thread.uw.fpsimd_state = fpsimd;
 
-	return err ? -EFAULT : 0;
+	return 0;
 }
 
 static int compat_restore_sigframe(struct pt_regs *regs,