Merge patch series "RISC-V: Detect and report speed of unaligned vector accesses"

Charlie Jenkins <charlie@rivosinc.com> says:

Adds support for detecting and reporting the speed of unaligned vector
accesses on RISC-V CPUs. Adds vec_misaligned_speed key to the hwprobe
adds Zicclsm to cpufeature and fixes the check for scalar unaligned
emulated all CPUs. The vec_misaligned_speed key keeps the same format
as the scalar unaligned access speed key.

This set does not emulate unaligned vector accesses on CPUs that do not
support them. Only reports if userspace can run them and speed of
unaligned vector accesses if supported.

* b4-shazam-merge:
  RISC-V: hwprobe: Document unaligned vector perf key
  RISC-V: Report vector unaligned access speed hwprobe
  RISC-V: Detect unaligned vector accesses supported
  RISC-V: Replace RISCV_MISALIGNED with RISCV_SCALAR_MISALIGNED
  RISC-V: Scalar unaligned access emulated on hotplug CPUs
  RISC-V: Check scalar unaligned access on all CPUs

Link: https://lore.kernel.org/r/20241017-jesse_unaligned_vector-v10-0-5b33500160f8@rivosinc.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

Documentation/arch/riscv/hwprobe.rst

@@ -274,3 +274,19 @@ The following keys are defined:
   represent the highest userspace virtual address usable.
 
 * :c:macro:`RISCV_HWPROBE_KEY_TIME_CSR_FREQ`: Frequency (in Hz) of `time CSR`.
+
+* :c:macro:`RISCV_HWPROBE_KEY_MISALIGNED_VECTOR_PERF`: An enum value describing the
+     performance of misaligned vector accesses on the selected set of processors.
+
+  * :c:macro:`RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN`: The performance of misaligned
+    vector accesses is unknown.
+
+  * :c:macro:`RISCV_HWPROBE_MISALIGNED_VECTOR_SLOW`: 32-bit misaligned accesses using vector
+    registers are slower than the equivalent quantity of byte accesses via vector registers.
+    Misaligned accesses may be supported directly in hardware, or trapped and emulated by software.
+
+  * :c:macro:`RISCV_HWPROBE_MISALIGNED_VECTOR_FAST`: 32-bit misaligned accesses using vector
+    registers are faster than the equivalent quantity of byte accesses via vector registers.
+
+  * :c:macro:`RISCV_HWPROBE_MISALIGNED_VECTOR_UNSUPPORTED`: Misaligned vector accesses are
+    not supported at all and will generate a misaligned address fault.

arch/riscv/Kconfig

@@ -786,10 +786,24 @@ config THREAD_SIZE_ORDER
 
 config RISCV_MISALIGNED
 	bool
+	help
+	  Embed support for detecting and emulating misaligned
+	  scalar or vector loads and stores.
+
+config RISCV_SCALAR_MISALIGNED
+	bool
+	select RISCV_MISALIGNED
 	select SYSCTL_ARCH_UNALIGN_ALLOW
 	help
 	  Embed support for emulating misaligned loads and stores.
 
+config RISCV_VECTOR_MISALIGNED
+	bool
+	select RISCV_MISALIGNED
+	depends on RISCV_ISA_V
+	help
+	  Enable detecting support for vector misaligned loads and stores.
+
 choice
 	prompt "Unaligned Accesses Support"
 	default RISCV_PROBE_UNALIGNED_ACCESS
@@ -801,7 +815,7 @@ choice
 
 config RISCV_PROBE_UNALIGNED_ACCESS
 	bool "Probe for hardware unaligned access support"
-	select RISCV_MISALIGNED
+	select RISCV_SCALAR_MISALIGNED
 	help
 	  During boot, the kernel will run a series of tests to determine the
 	  speed of unaligned accesses. This probing will dynamically determine
@@ -812,7 +826,7 @@ config RISCV_PROBE_UNALIGNED_ACCESS
 
 config RISCV_EMULATED_UNALIGNED_ACCESS
 	bool "Emulate unaligned access where system support is missing"
-	select RISCV_MISALIGNED
+	select RISCV_SCALAR_MISALIGNED
 	help
 	  If unaligned memory accesses trap into the kernel as they are not
 	  supported by the system, the kernel will emulate the unaligned
@@ -841,6 +855,46 @@ config RISCV_EFFICIENT_UNALIGNED_ACCESS
 
 endchoice
 
+choice
+	prompt "Vector unaligned Accesses Support"
+	depends on RISCV_ISA_V
+	default RISCV_PROBE_VECTOR_UNALIGNED_ACCESS
+	help
+	  This determines the level of support for vector unaligned accesses. This
+	  information is used by the kernel to perform optimizations. It is also
+	  exposed to user space via the hwprobe syscall. The hardware will be
+	  probed at boot by default.
+
+config RISCV_PROBE_VECTOR_UNALIGNED_ACCESS
+	bool "Probe speed of vector unaligned accesses"
+	select RISCV_VECTOR_MISALIGNED
+	depends on RISCV_ISA_V
+	help
+	  During boot, the kernel will run a series of tests to determine the
+	  speed of vector unaligned accesses if they are supported. This probing
+	  will dynamically determine the speed of vector unaligned accesses on
+	  the underlying system if they are supported.
+
+config RISCV_SLOW_VECTOR_UNALIGNED_ACCESS
+	bool "Assume the system supports slow vector unaligned memory accesses"
+	depends on NONPORTABLE
+	help
+	  Assume that the system supports slow vector unaligned memory accesses. The
+	  kernel and userspace programs may not be able to run at all on systems
+	  that do not support unaligned memory accesses.
+
+config RISCV_EFFICIENT_VECTOR_UNALIGNED_ACCESS
+	bool "Assume the system supports fast vector unaligned memory accesses"
+	depends on NONPORTABLE
+	help
+	  Assume that the system supports fast vector unaligned memory accesses. When
+	  enabled, this option improves the performance of the kernel on such
+	  systems. However, the kernel and userspace programs will run much more
+	  slowly, or will not be able to run at all, on systems that do not
+	  support efficient unaligned memory accesses.
+
+endchoice
+
 source "arch/riscv/Kconfig.vendor"
 
 endmenu # "Platform type"

arch/riscv/include/asm/cpufeature.h

@@ -8,6 +8,7 @@
 
 #include <linux/bitmap.h>
 #include <linux/jump_label.h>
+#include <linux/workqueue.h>
 #include <asm/hwcap.h>
 #include <asm/alternative-macros.h>
 #include <asm/errno.h>
@@ -58,8 +59,9 @@ void __init riscv_user_isa_enable(void);
 #define __RISCV_ISA_EXT_SUPERSET_VALIDATE(_name, _id, _sub_exts, _validate) \
 	_RISCV_ISA_EXT_DATA(_name, _id, _sub_exts, ARRAY_SIZE(_sub_exts), _validate)
 
-#if defined(CONFIG_RISCV_MISALIGNED)
 bool check_unaligned_access_emulated_all_cpus(void);
+#if defined(CONFIG_RISCV_SCALAR_MISALIGNED)
+void check_unaligned_access_emulated(struct work_struct *work __always_unused);
 void unaligned_emulation_finish(void);
 bool unaligned_ctl_available(void);
 DECLARE_PER_CPU(long, misaligned_access_speed);
@@ -70,6 +72,12 @@ static inline bool unaligned_ctl_available(void)
 }
 #endif
 
+bool check_vector_unaligned_access_emulated_all_cpus(void);
+#if defined(CONFIG_RISCV_VECTOR_MISALIGNED)
+void check_vector_unaligned_access_emulated(struct work_struct *work __always_unused);
+DECLARE_PER_CPU(long, vector_misaligned_access);
+#endif
+
 #if defined(CONFIG_RISCV_PROBE_UNALIGNED_ACCESS)
 DECLARE_STATIC_KEY_FALSE(fast_unaligned_access_speed_key);
 

arch/riscv/include/asm/entry-common.h

@@ -25,18 +25,7 @@ static inline void arch_exit_to_user_mode_prepare(struct pt_regs *regs,
 void handle_page_fault(struct pt_regs *regs);
 void handle_break(struct pt_regs *regs);
 
-#ifdef CONFIG_RISCV_MISALIGNED
 int handle_misaligned_load(struct pt_regs *regs);
 int handle_misaligned_store(struct pt_regs *regs);
-#else
-static inline int handle_misaligned_load(struct pt_regs *regs)
-{
-	return -1;
-}
-static inline int handle_misaligned_store(struct pt_regs *regs)
-{
-	return -1;
-}
-#endif
 
 #endif /* _ASM_RISCV_ENTRY_COMMON_H */

arch/riscv/include/asm/hwprobe.h

@@ -8,7 +8,7 @@
 
 #include <uapi/asm/hwprobe.h>
 
-#define RISCV_HWPROBE_MAX_KEY 9
+#define RISCV_HWPROBE_MAX_KEY 10
 
 static inline bool riscv_hwprobe_key_is_valid(__s64 key)
 {

arch/riscv/include/asm/vector.h

@@ -21,6 +21,7 @@
 
 extern unsigned long riscv_v_vsize;
 int riscv_v_setup_vsize(void);
+bool insn_is_vector(u32 insn_buf);
 bool riscv_v_first_use_handler(struct pt_regs *regs);
 void kernel_vector_begin(void);
 void kernel_vector_end(void);
@@ -268,6 +269,7 @@ struct pt_regs;
 
 static inline int riscv_v_setup_vsize(void) { return -EOPNOTSUPP; }
 static __always_inline bool has_vector(void) { return false; }
+static __always_inline bool insn_is_vector(u32 insn_buf) { return false; }
 static inline bool riscv_v_first_use_handler(struct pt_regs *regs) { return false; }
 static inline bool riscv_v_vstate_query(struct pt_regs *regs) { return false; }
 static inline bool riscv_v_vstate_ctrl_user_allowed(void) { return false; }

arch/riscv/include/uapi/asm/hwprobe.h

@@ -88,6 +88,11 @@ struct riscv_hwprobe {
 #define		RISCV_HWPROBE_MISALIGNED_SCALAR_SLOW		2
 #define		RISCV_HWPROBE_MISALIGNED_SCALAR_FAST		3
 #define		RISCV_HWPROBE_MISALIGNED_SCALAR_UNSUPPORTED	4
+#define RISCV_HWPROBE_KEY_MISALIGNED_VECTOR_PERF	10
+#define		RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN		0
+#define		RISCV_HWPROBE_MISALIGNED_VECTOR_SLOW		2
+#define		RISCV_HWPROBE_MISALIGNED_VECTOR_FAST		3
+#define		RISCV_HWPROBE_MISALIGNED_VECTOR_UNSUPPORTED	4
 /* Increase RISCV_HWPROBE_MAX_KEY when adding items. */
 
 /* Flags */

arch/riscv/kernel/Makefile

@@ -70,7 +70,8 @@ obj-$(CONFIG_MMU) += vdso.o vdso/
 
 obj-$(CONFIG_RISCV_MISALIGNED)	+= traps_misaligned.o
 obj-$(CONFIG_RISCV_MISALIGNED)	+= unaligned_access_speed.o
-obj-$(CONFIG_RISCV_PROBE_UNALIGNED_ACCESS)	+= copy-unaligned.o
+obj-$(CONFIG_RISCV_PROBE_UNALIGNED_ACCESS)		+= copy-unaligned.o
+obj-$(CONFIG_RISCV_PROBE_VECTOR_UNALIGNED_ACCESS)	+= vec-copy-unaligned.o
 
 obj-$(CONFIG_FPU)		+= fpu.o
 obj-$(CONFIG_FPU)		+= kernel_mode_fpu.o

arch/riscv/kernel/copy-unaligned.h

@@ -10,4 +10,9 @@
 void __riscv_copy_words_unaligned(void *dst, const void *src, size_t size);
 void __riscv_copy_bytes_unaligned(void *dst, const void *src, size_t size);
 
+#ifdef CONFIG_RISCV_PROBE_VECTOR_UNALIGNED_ACCESS
+void __riscv_copy_vec_words_unaligned(void *dst, const void *src, size_t size);
+void __riscv_copy_vec_bytes_unaligned(void *dst, const void *src, size_t size);
+#endif
+
 #endif /* __RISCV_KERNEL_COPY_UNALIGNED_H */

arch/riscv/kernel/fpu.S

@@ -170,7 +170,7 @@ SYM_FUNC_END(__fstate_restore)
 	__access_func(f31)
 
 
-#ifdef CONFIG_RISCV_MISALIGNED
+#ifdef CONFIG_RISCV_SCALAR_MISALIGNED
 
 /*
  * Disable compressed instructions set to keep a constant offset between FP
@@ -224,4 +224,4 @@ SYM_FUNC_START(get_f64_reg)
 	fp_access_epilogue
 SYM_FUNC_END(get_f64_reg)
 
-#endif /* CONFIG_RISCV_MISALIGNED */
+#endif /* CONFIG_RISCV_SCALAR_MISALIGNED */

arch/riscv/kernel/sys_hwprobe.c

@@ -201,6 +201,43 @@ static u64 hwprobe_misaligned(const struct cpumask *cpus)
 }
 #endif
 
+#ifdef CONFIG_RISCV_VECTOR_MISALIGNED
+static u64 hwprobe_vec_misaligned(const struct cpumask *cpus)
+{
+	int cpu;
+	u64 perf = -1ULL;
+
+	/* Return if supported or not even if speed wasn't probed */
+	for_each_cpu(cpu, cpus) {
+		int this_perf = per_cpu(vector_misaligned_access, cpu);
+
+		if (perf == -1ULL)
+			perf = this_perf;
+
+		if (perf != this_perf) {
+			perf = RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN;
+			break;
+		}
+	}
+
+	if (perf == -1ULL)
+		return RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN;
+
+	return perf;
+}
+#else
+static u64 hwprobe_vec_misaligned(const struct cpumask *cpus)
+{
+	if (IS_ENABLED(CONFIG_RISCV_EFFICIENT_VECTOR_UNALIGNED_ACCESS))
+		return RISCV_HWPROBE_MISALIGNED_VECTOR_FAST;
+
+	if (IS_ENABLED(CONFIG_RISCV_SLOW_VECTOR_UNALIGNED_ACCESS))
+		return RISCV_HWPROBE_MISALIGNED_VECTOR_SLOW;
+
+	return RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN;
+}
+#endif
+
 static void hwprobe_one_pair(struct riscv_hwprobe *pair,
 			     const struct cpumask *cpus)
 {
@@ -229,6 +266,10 @@ static void hwprobe_one_pair(struct riscv_hwprobe *pair,
 		pair->value = hwprobe_misaligned(cpus);
 		break;
 
+	case RISCV_HWPROBE_KEY_MISALIGNED_VECTOR_PERF:
+		pair->value = hwprobe_vec_misaligned(cpus);
+		break;
+
 	case RISCV_HWPROBE_KEY_ZICBOZ_BLOCK_SIZE:
 		pair->value = 0;
 		if (hwprobe_ext0_has(cpus, RISCV_HWPROBE_EXT_ZICBOZ))

arch/riscv/kernel/traps_misaligned.c

@@ -16,6 +16,7 @@
 #include <asm/entry-common.h>
 #include <asm/hwprobe.h>
 #include <asm/cpufeature.h>
+#include <asm/vector.h>
 
 #define INSN_MATCH_LB			0x3
 #define INSN_MASK_LB			0x707f
@@ -322,12 +323,37 @@ union reg_data {
 	u64 data_u64;
 };
 
-static bool unaligned_ctl __read_mostly;
-
 /* sysctl hooks */
 int unaligned_enabled __read_mostly = 1;	/* Enabled by default */
 
-int handle_misaligned_load(struct pt_regs *regs)
+#ifdef CONFIG_RISCV_VECTOR_MISALIGNED
+static int handle_vector_misaligned_load(struct pt_regs *regs)
+{
+	unsigned long epc = regs->epc;
+	unsigned long insn;
+
+	if (get_insn(regs, epc, &insn))
+		return -1;
+
+	/* Only return 0 when in check_vector_unaligned_access_emulated */
+	if (*this_cpu_ptr(&vector_misaligned_access) == RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN) {
+		*this_cpu_ptr(&vector_misaligned_access) = RISCV_HWPROBE_MISALIGNED_VECTOR_UNSUPPORTED;
+		regs->epc = epc + INSN_LEN(insn);
+		return 0;
+	}
+
+	/* If vector instruction we don't emulate it yet */
+	regs->epc = epc;
+	return -1;
+}
+#else
+static int handle_vector_misaligned_load(struct pt_regs *regs)
+{
+	return -1;
+}
+#endif
+
+static int handle_scalar_misaligned_load(struct pt_regs *regs)
 {
 	union reg_data val;
 	unsigned long epc = regs->epc;
@@ -435,7 +461,7 @@ int handle_misaligned_load(struct pt_regs *regs)
 	return 0;
 }
 
-int handle_misaligned_store(struct pt_regs *regs)
+static int handle_scalar_misaligned_store(struct pt_regs *regs)
 {
 	union reg_data val;
 	unsigned long epc = regs->epc;
@@ -526,11 +552,96 @@ int handle_misaligned_store(struct pt_regs *regs)
 	return 0;
 }
 
-static bool check_unaligned_access_emulated(int cpu)
+int handle_misaligned_load(struct pt_regs *regs)
 {
+	unsigned long epc = regs->epc;
+	unsigned long insn;
+
+	if (IS_ENABLED(CONFIG_RISCV_VECTOR_MISALIGNED)) {
+		if (get_insn(regs, epc, &insn))
+			return -1;
+
+		if (insn_is_vector(insn))
+			return handle_vector_misaligned_load(regs);
+	}
+
+	if (IS_ENABLED(CONFIG_RISCV_SCALAR_MISALIGNED))
+		return handle_scalar_misaligned_load(regs);
+
+	return -1;
+}
+
+int handle_misaligned_store(struct pt_regs *regs)
+{
+	if (IS_ENABLED(CONFIG_RISCV_SCALAR_MISALIGNED))
+		return handle_scalar_misaligned_store(regs);
+
+	return -1;
+}
+
+#ifdef CONFIG_RISCV_VECTOR_MISALIGNED
+void check_vector_unaligned_access_emulated(struct work_struct *work __always_unused)
+{
+	long *mas_ptr = this_cpu_ptr(&vector_misaligned_access);
+	unsigned long tmp_var;
+
+	*mas_ptr = RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN;
+
+	kernel_vector_begin();
+	/*
+	 * In pre-13.0.0 versions of GCC, vector registers cannot appear in
+	 * the clobber list. This inline asm clobbers v0, but since we do not
+	 * currently build the kernel with V enabled, the v0 clobber arg is not
+	 * needed (as the compiler will not emit vector code itself). If the kernel
+	 * is changed to build with V enabled, the clobber arg will need to be
+	 * added here.
+	 */
+	__asm__ __volatile__ (
+		".balign 4\n\t"
+		".option push\n\t"
+		".option arch, +zve32x\n\t"
+		"       vsetivli zero, 1, e16, m1, ta, ma\n\t"	// Vectors of 16b
+		"       vle16.v v0, (%[ptr])\n\t"		// Load bytes
+		".option pop\n\t"
+		: : [ptr] "r" ((u8 *)&tmp_var + 1));
+	kernel_vector_end();
+}
+
+bool check_vector_unaligned_access_emulated_all_cpus(void)
+{
+	int cpu;
+
+	if (!has_vector()) {
+		for_each_online_cpu(cpu)
+			per_cpu(vector_misaligned_access, cpu) = RISCV_HWPROBE_MISALIGNED_VECTOR_UNSUPPORTED;
+		return false;
+	}
+
+	schedule_on_each_cpu(check_vector_unaligned_access_emulated);
+
+	for_each_online_cpu(cpu)
+		if (per_cpu(vector_misaligned_access, cpu)
+		    == RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN)
+			return false;
+
+	return true;
+}
+#else
+bool check_vector_unaligned_access_emulated_all_cpus(void)
+{
+	return false;
+}
+#endif
+
+#ifdef CONFIG_RISCV_SCALAR_MISALIGNED
+
+static bool unaligned_ctl __read_mostly;
+
+void check_unaligned_access_emulated(struct work_struct *work __always_unused)
+{
+	int cpu = smp_processor_id();
 	long *mas_ptr = per_cpu_ptr(&misaligned_access_speed, cpu);
 	unsigned long tmp_var, tmp_val;
-	bool misaligned_emu_detected;
 
 	*mas_ptr = RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN;
 
@@ -538,19 +649,16 @@ static bool check_unaligned_access_emulated(int cpu)
 		"       "REG_L" %[tmp], 1(%[ptr])\n"
 		: [tmp] "=r" (tmp_val) : [ptr] "r" (&tmp_var) : "memory");
 
-	misaligned_emu_detected = (*mas_ptr == RISCV_HWPROBE_MISALIGNED_SCALAR_EMULATED);
 	/*
 	 * If unaligned_ctl is already set, this means that we detected that all
 	 * CPUS uses emulated misaligned access at boot time. If that changed
 	 * when hotplugging the new cpu, this is something we don't handle.
 	 */
-	if (unlikely(unaligned_ctl && !misaligned_emu_detected)) {
+	if (unlikely(unaligned_ctl && (*mas_ptr != RISCV_HWPROBE_MISALIGNED_SCALAR_EMULATED))) {
 		pr_crit("CPU misaligned accesses non homogeneous (expected all emulated)\n");
 		while (true)
 			cpu_relax();
 	}
-
-	return misaligned_emu_detected;
 }
 
 bool check_unaligned_access_emulated_all_cpus(void)
@@ -562,8 +670,11 @@ bool check_unaligned_access_emulated_all_cpus(void)
 	 * accesses emulated since tasks requesting such control can run on any
 	 * CPU.
 	 */
+	schedule_on_each_cpu(check_unaligned_access_emulated);
+
 	for_each_online_cpu(cpu)
-		if (!check_unaligned_access_emulated(cpu))
+		if (per_cpu(misaligned_access_speed, cpu)
+		    != RISCV_HWPROBE_MISALIGNED_SCALAR_EMULATED)
 			return false;
 
 	unaligned_ctl = true;
@@ -574,3 +685,9 @@ bool unaligned_ctl_available(void)
 {
 	return unaligned_ctl;
 }
+#else
+bool check_unaligned_access_emulated_all_cpus(void)
+{
+	return false;
+}
+#endif

arch/riscv/kernel/unaligned_access_speed.c

@@ -6,11 +6,13 @@
 #include <linux/cpu.h>
 #include <linux/cpumask.h>
 #include <linux/jump_label.h>
+#include <linux/kthread.h>
 #include <linux/mm.h>
 #include <linux/smp.h>
 #include <linux/types.h>
 #include <asm/cpufeature.h>
 #include <asm/hwprobe.h>
+#include <asm/vector.h>
 
 #include "copy-unaligned.h"
 
@@ -19,7 +21,8 @@
 #define MISALIGNED_BUFFER_ORDER get_order(MISALIGNED_BUFFER_SIZE)
 #define MISALIGNED_COPY_SIZE ((MISALIGNED_BUFFER_SIZE / 2) - 0x80)
 
-DEFINE_PER_CPU(long, misaligned_access_speed);
+DEFINE_PER_CPU(long, misaligned_access_speed) = RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN;
+DEFINE_PER_CPU(long, vector_misaligned_access) = RISCV_HWPROBE_MISALIGNED_VECTOR_UNSUPPORTED;
 
 #ifdef CONFIG_RISCV_PROBE_UNALIGNED_ACCESS
 static cpumask_t fast_misaligned_access;
@@ -191,6 +194,7 @@ static int riscv_online_cpu(unsigned int cpu)
 	if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN)
 		goto exit;
 
+	check_unaligned_access_emulated(NULL);
 	buf = alloc_pages(GFP_KERNEL, MISALIGNED_BUFFER_ORDER);
 	if (!buf) {
 		pr_warn("Allocation failure, not measuring misaligned performance\n");
@@ -259,23 +263,159 @@ static int check_unaligned_access_speed_all_cpus(void)
 	kfree(bufs);
 	return 0;
 }
+#else /* CONFIG_RISCV_PROBE_UNALIGNED_ACCESS */
+static int check_unaligned_access_speed_all_cpus(void)
+{
+	return 0;
+}
+#endif
+
+#ifdef CONFIG_RISCV_PROBE_VECTOR_UNALIGNED_ACCESS
+static void check_vector_unaligned_access(struct work_struct *work __always_unused)
+{
+	int cpu = smp_processor_id();
+	u64 start_cycles, end_cycles;
+	u64 word_cycles;
+	u64 byte_cycles;
+	int ratio;
+	unsigned long start_jiffies, now;
+	struct page *page;
+	void *dst;
+	void *src;
+	long speed = RISCV_HWPROBE_MISALIGNED_VECTOR_SLOW;
+
+	if (per_cpu(vector_misaligned_access, cpu) != RISCV_HWPROBE_MISALIGNED_VECTOR_UNKNOWN)
+		return;
+
+	page = alloc_pages(GFP_KERNEL, MISALIGNED_BUFFER_ORDER);
+	if (!page) {
+		pr_warn("Allocation failure, not measuring vector misaligned performance\n");
+		return;
+	}
+
+	/* Make an unaligned destination buffer. */
+	dst = (void *)((unsigned long)page_address(page) | 0x1);
+	/* Unalign src as well, but differently (off by 1 + 2 = 3). */
+	src = dst + (MISALIGNED_BUFFER_SIZE / 2);
+	src += 2;
+	word_cycles = -1ULL;
+
+	/* Do a warmup. */
+	kernel_vector_begin();
+	__riscv_copy_vec_words_unaligned(dst, src, MISALIGNED_COPY_SIZE);
+
+	start_jiffies = jiffies;
+	while ((now = jiffies) == start_jiffies)
+		cpu_relax();
+
+	/*
+	 * For a fixed amount of time, repeatedly try the function, and take
+	 * the best time in cycles as the measurement.
+	 */
+	while (time_before(jiffies, now + (1 << MISALIGNED_ACCESS_JIFFIES_LG2))) {
+		start_cycles = get_cycles64();
+		/* Ensure the CSR read can't reorder WRT to the copy. */
+		mb();
+		__riscv_copy_vec_words_unaligned(dst, src, MISALIGNED_COPY_SIZE);
+		/* Ensure the copy ends before the end time is snapped. */
+		mb();
+		end_cycles = get_cycles64();
+		if ((end_cycles - start_cycles) < word_cycles)
+			word_cycles = end_cycles - start_cycles;
+	}
+
+	byte_cycles = -1ULL;
+	__riscv_copy_vec_bytes_unaligned(dst, src, MISALIGNED_COPY_SIZE);
+	start_jiffies = jiffies;
+	while ((now = jiffies) == start_jiffies)
+		cpu_relax();
+
+	while (time_before(jiffies, now + (1 << MISALIGNED_ACCESS_JIFFIES_LG2))) {
+		start_cycles = get_cycles64();
+		/* Ensure the CSR read can't reorder WRT to the copy. */
+		mb();
+		__riscv_copy_vec_bytes_unaligned(dst, src, MISALIGNED_COPY_SIZE);
+		/* Ensure the copy ends before the end time is snapped. */
+		mb();
+		end_cycles = get_cycles64();
+		if ((end_cycles - start_cycles) < byte_cycles)
+			byte_cycles = end_cycles - start_cycles;
+	}
+
+	kernel_vector_end();
+
+	/* Don't divide by zero. */
+	if (!word_cycles || !byte_cycles) {
+		pr_warn("cpu%d: rdtime lacks granularity needed to measure unaligned vector access speed\n",
+			cpu);
+
+		return;
+	}
+
+	if (word_cycles < byte_cycles)
+		speed = RISCV_HWPROBE_MISALIGNED_VECTOR_FAST;
+
+	ratio = div_u64((byte_cycles * 100), word_cycles);
+	pr_info("cpu%d: Ratio of vector byte access time to vector unaligned word access is %d.%02d, unaligned accesses are %s\n",
+		cpu,
+		ratio / 100,
+		ratio % 100,
+		(speed ==  RISCV_HWPROBE_MISALIGNED_VECTOR_FAST) ? "fast" : "slow");
+
+	per_cpu(vector_misaligned_access, cpu) = speed;
+}
+
+static int riscv_online_cpu_vec(unsigned int cpu)
+{
+	if (!has_vector())
+		return 0;
+
+	if (per_cpu(vector_misaligned_access, cpu) != RISCV_HWPROBE_MISALIGNED_VECTOR_UNSUPPORTED)
+		return 0;
+
+	check_vector_unaligned_access_emulated(NULL);
+	check_vector_unaligned_access(NULL);
+	return 0;
+}
+
+/* Measure unaligned access speed on all CPUs present at boot in parallel. */
+static int vec_check_unaligned_access_speed_all_cpus(void *unused __always_unused)
+{
+	schedule_on_each_cpu(check_vector_unaligned_access);
+
+	/*
+	 * Setup hotplug callbacks for any new CPUs that come online or go
+	 * offline.
+	 */
+	cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN, "riscv:online",
+				  riscv_online_cpu_vec, NULL);
+
+	return 0;
+}
+#else /* CONFIG_RISCV_PROBE_VECTOR_UNALIGNED_ACCESS */
+static int vec_check_unaligned_access_speed_all_cpus(void *unused __always_unused)
+{
+	return 0;
+}
+#endif
 
 static int check_unaligned_access_all_cpus(void)
 {
-	bool all_cpus_emulated = check_unaligned_access_emulated_all_cpus();
+	bool all_cpus_emulated, all_cpus_vec_unsupported;
+
+	all_cpus_emulated = check_unaligned_access_emulated_all_cpus();
+	all_cpus_vec_unsupported = check_vector_unaligned_access_emulated_all_cpus();
+
+	if (!all_cpus_vec_unsupported &&
+	    IS_ENABLED(CONFIG_RISCV_PROBE_VECTOR_UNALIGNED_ACCESS)) {
+		kthread_run(vec_check_unaligned_access_speed_all_cpus,
+			    NULL, "vec_check_unaligned_access_speed_all_cpus");
+	}
 
 	if (!all_cpus_emulated)
 		return check_unaligned_access_speed_all_cpus();
 
 	return 0;
 }
-#else /* CONFIG_RISCV_PROBE_UNALIGNED_ACCESS */
-static int check_unaligned_access_all_cpus(void)
-{
-	check_unaligned_access_emulated_all_cpus();
-
-	return 0;
-}
-#endif
 
 arch_initcall(check_unaligned_access_all_cpus);

arch/riscv/kernel/vec-copy-unaligned.S

@@ -0,0 +1,58 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (C) 2024 Rivos Inc. */
+
+#include <linux/args.h>
+#include <linux/linkage.h>
+#include <asm/asm.h>
+
+	.text
+
+#define WORD_EEW 32
+
+#define WORD_SEW CONCATENATE(e, WORD_EEW)
+#define VEC_L CONCATENATE(vle, WORD_EEW).v
+#define VEC_S CONCATENATE(vle, WORD_EEW).v
+
+/* void __riscv_copy_vec_words_unaligned(void *, const void *, size_t) */
+/* Performs a memcpy without aligning buffers, using word loads and stores. */
+/* Note: The size is truncated to a multiple of WORD_EEW */
+SYM_FUNC_START(__riscv_copy_vec_words_unaligned)
+	andi  a4, a2, ~(WORD_EEW-1)
+	beqz  a4, 2f
+	add   a3, a1, a4
+	.option push
+	.option arch, +zve32x
+1:
+	vsetivli t0, 8, WORD_SEW, m8, ta, ma
+	VEC_L v0, (a1)
+	VEC_S v0, (a0)
+	addi  a0, a0, WORD_EEW
+	addi  a1, a1, WORD_EEW
+	bltu  a1, a3, 1b
+
+2:
+	.option pop
+	ret
+SYM_FUNC_END(__riscv_copy_vec_words_unaligned)
+
+/* void __riscv_copy_vec_bytes_unaligned(void *, const void *, size_t) */
+/* Performs a memcpy without aligning buffers, using only byte accesses. */
+/* Note: The size is truncated to a multiple of 8 */
+SYM_FUNC_START(__riscv_copy_vec_bytes_unaligned)
+	andi a4, a2, ~(8-1)
+	beqz a4, 2f
+	add  a3, a1, a4
+	.option push
+	.option arch, +zve32x
+1:
+	vsetivli t0, 8, e8, m8, ta, ma
+	vle8.v v0, (a1)
+	vse8.v v0, (a0)
+	addi a0, a0, 8
+	addi a1, a1, 8
+	bltu a1, a3, 1b
+
+2:
+	.option pop
+	ret
+SYM_FUNC_END(__riscv_copy_vec_bytes_unaligned)

arch/riscv/kernel/vector.c

@@ -66,7 +66,7 @@ void __init riscv_v_setup_ctx_cache(void)
 #endif
 }
 
-static bool insn_is_vector(u32 insn_buf)
+bool insn_is_vector(u32 insn_buf)
 {
 	u32 opcode = insn_buf & __INSN_OPCODE_MASK;
 	u32 width, csr;