Merge v6.12.34linux-rolling-lts

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

2 files changed, 57 insertions, 25 deletions

diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c
index 3fcd9d080bf2..d23315ef7b67 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)
@@ -707,10 +730,11 @@ 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 f38d22dac140..8854bce5cfe2 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
@@ -359,9 +359,6 @@ 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:
@@ -413,6 +410,9 @@ 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 +453,15 @@ 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 +654,7 @@ 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 +678,7 @@ 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;
@@ -1436,7 +1439,7 @@ 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 +1463,8 @@ 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 +1578,8 @@ 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 +1666,9 @@ 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();
@@ -1801,7 +1809,7 @@ void fpsimd_update_current_state(struct user_fpsimd_state const *state)
 	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();