Merge branch 'for-next/kernel-mode-neon' into for-next/core

* for-next/kernel-mode-neon:
  arm64: neon/efi: Make EFI fpsimd save/restore variables static
  arm64: neon: Forbid when irqs are disabled
  arm64: neon: Export kernel_neon_busy to loadable modules
  arm64: neon: Temporarily add a kernel_mode_begin_partial() definition
  arm64: neon: Remove support for nested or hardirq kernel-mode NEON
  arm64: neon: Allow EFI runtime services to use FPSIMD in irq context
  arm64: fpsimd: Consistently use __this_cpu_ ops where appropriate
  arm64: neon: Add missing header guard in <asm/neon.h>
  arm64: neon: replace generic definition of may_use_simd()

6 files changed, 75 insertions, 75 deletions

diff --git a/arch/arm64/include/asm/Kbuild b/arch/arm64/include/asm/Kbuild
index f81c7b685fc6..2326e39d5892 100644
--- a/arch/arm64/include/asm/Kbuild
+++ b/arch/arm64/include/asm/Kbuild
@@ -20,7 +20,6 @@ generic-y += rwsem.h
 generic-y += segment.h
 generic-y += serial.h
 generic-y += set_memory.h
-generic-y += simd.h
 generic-y += sizes.h
 generic-y += switch_to.h
 generic-y += trace_clock.h
diff --git a/arch/arm64/include/asm/efi.h b/arch/arm64/include/asm/efi.h
index 2b1e5def2e49..0cad5a5894b9 100644
--- a/arch/arm64/include/asm/efi.h
+++ b/arch/arm64/include/asm/efi.h
@@ -3,6 +3,7 @@
 
 #include <asm/boot.h>
 #include <asm/cpufeature.h>
+#include <asm/fpsimd.h>
 #include <asm/io.h>
 #include <asm/memory.h>
 #include <asm/mmu_context.h>
@@ -21,8 +22,8 @@ int efi_set_mapping_permissions(struct mm_struct *mm, efi_memory_desc_t *md);
 
 #define arch_efi_call_virt_setup()					\
 ({									\
-	kernel_neon_begin();						\
 	efi_virtmap_load();						\
+	__efi_fpsimd_begin();						\
 })
 
 #define arch_efi_call_virt(p, f, args...)				\
@@ -34,8 +35,8 @@ int efi_set_mapping_permissions(struct mm_struct *mm, efi_memory_desc_t *md);
 
 #define arch_efi_call_virt_teardown()					\
 ({									\
+	__efi_fpsimd_end();						\
 	efi_virtmap_unload();						\
-	kernel_neon_end();						\
 })
 
 #define ARCH_EFI_IRQ_FLAGS_MASK (PSR_D_BIT | PSR_A_BIT | PSR_I_BIT | PSR_F_BIT)
diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h
index 50f559f574fe..410c48163c6a 100644
--- a/arch/arm64/include/asm/fpsimd.h
+++ b/arch/arm64/include/asm/fpsimd.h
@@ -41,16 +41,6 @@ struct fpsimd_state {
 	unsigned int cpu;
 };
 
-/*
- * Struct for stacking the bottom 'n' FP/SIMD registers.
- */
-struct fpsimd_partial_state {
-	u32		fpsr;
-	u32		fpcr;
-	u32		num_regs;
-	__uint128_t	vregs[32];
-};
-
 
 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
 /* Masks for extracting the FPSR and FPCR from the FPSCR */
@@ -77,9 +67,9 @@ extern void fpsimd_update_current_state(struct fpsimd_state *state);
 
 extern void fpsimd_flush_task_state(struct task_struct *target);
 
-extern void fpsimd_save_partial_state(struct fpsimd_partial_state *state,
-				      u32 num_regs);
-extern void fpsimd_load_partial_state(struct fpsimd_partial_state *state);
+/* For use by EFI runtime services calls only */
+extern void __efi_fpsimd_begin(void);
+extern void __efi_fpsimd_end(void);
 
 #endif
 
diff --git a/arch/arm64/include/asm/fpsimdmacros.h b/arch/arm64/include/asm/fpsimdmacros.h
index a2daf1293028..0f5fdd388b0d 100644
--- a/arch/arm64/include/asm/fpsimdmacros.h
+++ b/arch/arm64/include/asm/fpsimdmacros.h
@@ -75,59 +75,3 @@
 	ldr	w\tmpnr, [\state, #16 * 2 + 4]
 	fpsimd_restore_fpcr x\tmpnr, \state
 .endm
-
-.macro fpsimd_save_partial state, numnr, tmpnr1, tmpnr2
-	mrs	x\tmpnr1, fpsr
-	str	w\numnr, [\state, #8]
-	mrs	x\tmpnr2, fpcr
-	stp	w\tmpnr1, w\tmpnr2, [\state]
-	adr	x\tmpnr1, 0f
-	add	\state, \state, x\numnr, lsl #4
-	sub	x\tmpnr1, x\tmpnr1, x\numnr, lsl #1
-	br	x\tmpnr1
-	stp	q30, q31, [\state, #-16 * 30 - 16]
-	stp	q28, q29, [\state, #-16 * 28 - 16]
-	stp	q26, q27, [\state, #-16 * 26 - 16]
-	stp	q24, q25, [\state, #-16 * 24 - 16]
-	stp	q22, q23, [\state, #-16 * 22 - 16]
-	stp	q20, q21, [\state, #-16 * 20 - 16]
-	stp	q18, q19, [\state, #-16 * 18 - 16]
-	stp	q16, q17, [\state, #-16 * 16 - 16]
-	stp	q14, q15, [\state, #-16 * 14 - 16]
-	stp	q12, q13, [\state, #-16 * 12 - 16]
-	stp	q10, q11, [\state, #-16 * 10 - 16]
-	stp	q8, q9, [\state, #-16 * 8 - 16]
-	stp	q6, q7, [\state, #-16 * 6 - 16]
-	stp	q4, q5, [\state, #-16 * 4 - 16]
-	stp	q2, q3, [\state, #-16 * 2 - 16]
-	stp	q0, q1, [\state, #-16 * 0 - 16]
-0:
-.endm
-
-.macro fpsimd_restore_partial state, tmpnr1, tmpnr2
-	ldp	w\tmpnr1, w\tmpnr2, [\state]
-	msr	fpsr, x\tmpnr1
-	fpsimd_restore_fpcr x\tmpnr2, x\tmpnr1
-	adr	x\tmpnr1, 0f
-	ldr	w\tmpnr2, [\state, #8]
-	add	\state, \state, x\tmpnr2, lsl #4
-	sub	x\tmpnr1, x\tmpnr1, x\tmpnr2, lsl #1
-	br	x\tmpnr1
-	ldp	q30, q31, [\state, #-16 * 30 - 16]
-	ldp	q28, q29, [\state, #-16 * 28 - 16]
-	ldp	q26, q27, [\state, #-16 * 26 - 16]
-	ldp	q24, q25, [\state, #-16 * 24 - 16]
-	ldp	q22, q23, [\state, #-16 * 22 - 16]
-	ldp	q20, q21, [\state, #-16 * 20 - 16]
-	ldp	q18, q19, [\state, #-16 * 18 - 16]
-	ldp	q16, q17, [\state, #-16 * 16 - 16]
-	ldp	q14, q15, [\state, #-16 * 14 - 16]
-	ldp	q12, q13, [\state, #-16 * 12 - 16]
-	ldp	q10, q11, [\state, #-16 * 10 - 16]
-	ldp	q8, q9, [\state, #-16 * 8 - 16]
-	ldp	q6, q7, [\state, #-16 * 6 - 16]
-	ldp	q4, q5, [\state, #-16 * 4 - 16]
-	ldp	q2, q3, [\state, #-16 * 2 - 16]
-	ldp	q0, q1, [\state, #-16 * 0 - 16]
-0:
-.endm
diff --git a/arch/arm64/include/asm/neon.h b/arch/arm64/include/asm/neon.h
index ad4cdc966c0f..f922eaf780f9 100644
--- a/arch/arm64/include/asm/neon.h
+++ b/arch/arm64/include/asm/neon.h
@@ -8,12 +8,22 @@
  * published by the Free Software Foundation.
  */
 
+#ifndef __ASM_NEON_H
+#define __ASM_NEON_H
+
 #include <linux/types.h>
 #include <asm/fpsimd.h>
 
 #define cpu_has_neon()		system_supports_fpsimd()
 
-#define kernel_neon_begin()	kernel_neon_begin_partial(32)
-
-void kernel_neon_begin_partial(u32 num_regs);
+void kernel_neon_begin(void);
 void kernel_neon_end(void);
+
+/*
+ * Temporary macro to allow the crypto code to compile. Note that the
+ * semantics of kernel_neon_begin_partial() are now different from the
+ * original as it does not allow being called in an interrupt context.
+ */
+#define kernel_neon_begin_partial(num_regs)	kernel_neon_begin()
+
+#endif /* ! __ASM_NEON_H */
diff --git a/arch/arm64/include/asm/simd.h b/arch/arm64/include/asm/simd.h
new file mode 100644
index 000000000000..fa8b3fe932e6
--- /dev/null
+++ b/arch/arm64/include/asm/simd.h
@@ -0,0 +1,56 @@
+/*
+ * Copyright (C) 2017 Linaro Ltd. <ard.biesheuvel@linaro.org>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published
+ * by the Free Software Foundation.
+ */
+
+#ifndef __ASM_SIMD_H
+#define __ASM_SIMD_H
+
+#include <linux/compiler.h>
+#include <linux/irqflags.h>
+#include <linux/percpu.h>
+#include <linux/preempt.h>
+#include <linux/types.h>
+
+#ifdef CONFIG_KERNEL_MODE_NEON
+
+DECLARE_PER_CPU(bool, kernel_neon_busy);
+
+/*
+ * may_use_simd - whether it is allowable at this time to issue SIMD
+ *                instructions or access the SIMD register file
+ *
+ * Callers must not assume that the result remains true beyond the next
+ * preempt_enable() or return from softirq context.
+ */
+static __must_check inline bool may_use_simd(void)
+{
+	/*
+	 * The raw_cpu_read() is racy if called with preemption enabled.
+	 * This is not a bug: kernel_neon_busy is only set when
+	 * preemption is disabled, so we cannot migrate to another CPU
+	 * while it is set, nor can we migrate to a CPU where it is set.
+	 * So, if we find it clear on some CPU then we're guaranteed to
+	 * find it clear on any CPU we could migrate to.
+	 *
+	 * If we are in between kernel_neon_begin()...kernel_neon_end(),
+	 * the flag will be set, but preemption is also disabled, so we
+	 * can't migrate to another CPU and spuriously see it become
+	 * false.
+	 */
+	return !in_irq() && !irqs_disabled() && !in_nmi() &&
+		!raw_cpu_read(kernel_neon_busy);
+}
+
+#else /* ! CONFIG_KERNEL_MODE_NEON */
+
+static __must_check inline bool may_use_simd(void) {
+	return false;
+}
+
+#endif /* ! CONFIG_KERNEL_MODE_NEON */
+
+#endif