From dc5dac748af9087e9240bd2ae6ae7db48d5360ae Mon Sep 17 00:00:00 2001 From: Nicholas Piggin Date: Sat, 8 Apr 2023 12:17:49 +1000 Subject: [PATCH] powerpc/64: Add support to build with prefixed instructions Add an option to build kernel and module with prefixed instructions if the CPU and toolchain support it. This is not related to kernel support for userspace execution of prefixed instructions. Building with prefixed instructions breaks some extended inline asm memory addressing, for example it will provide immediates that exceed the range of simple load/store displacement. Whether this is a toolchain or a kernel asm problem remains to be seen. For now, these are replaced with simpler and less efficient direct register addressing when compiling with prefixed. Signed-off-by: Nicholas Piggin Signed-off-by: Michael Ellerman Link: https://msgid.link/20230408021752.862660-4-npiggin@gmail.com --- arch/powerpc/Kconfig | 3 +++ arch/powerpc/Makefile | 4 ++++ arch/powerpc/include/asm/atomic.h | 24 ++++++++++++++++++---- arch/powerpc/include/asm/io.h | 37 ++++++++++++++++++++++++++++++++++ arch/powerpc/include/asm/uaccess.h | 28 +++++++++++++++++++++++-- arch/powerpc/kernel/trace/ftrace.c | 2 ++ arch/powerpc/platforms/Kconfig.cputype | 20 ++++++++++++++++++ 7 files changed, 112 insertions(+), 6 deletions(-) diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig index 3fb2c27..109c00b 100644 --- a/arch/powerpc/Kconfig +++ b/arch/powerpc/Kconfig @@ -4,6 +4,9 @@ source "arch/powerpc/platforms/Kconfig.cputype" config CC_HAS_ELFV2 def_bool PPC64 && $(cc-option, -mabi=elfv2) +config CC_HAS_PREFIXED + def_bool PPC64 && $(cc-option, -mcpu=power10 -mprefixed) + config 32BIT bool default y if PPC32 diff --git a/arch/powerpc/Makefile b/arch/powerpc/Makefile index 4343cca..9fb770d 100644 --- a/arch/powerpc/Makefile +++ b/arch/powerpc/Makefile @@ -180,7 +180,11 @@ ifdef CONFIG_476FPE_ERR46 endif # No prefix or pcrel +ifdef CONFIG_PPC_KERNEL_PREFIXED +KBUILD_CFLAGS += $(call cc-option,-mprefixed) +else KBUILD_CFLAGS += $(call cc-option,-mno-prefixed) +endif KBUILD_CFLAGS += $(call cc-option,-mno-pcrel) # No AltiVec or VSX or MMA instructions when building kernel diff --git a/arch/powerpc/include/asm/atomic.h b/arch/powerpc/include/asm/atomic.h index b3a5383..47228b1 100644 --- a/arch/powerpc/include/asm/atomic.h +++ b/arch/powerpc/include/asm/atomic.h @@ -27,14 +27,22 @@ static __inline__ int arch_atomic_read(const atomic_t *v) { int t; - __asm__ __volatile__("lwz%U1%X1 %0,%1" : "=r"(t) : "m<>"(v->counter)); + /* -mprefixed can generate offsets beyond range, fall back hack */ + if (IS_ENABLED(CONFIG_PPC_KERNEL_PREFIXED)) + __asm__ __volatile__("lwz %0,0(%1)" : "=r"(t) : "b"(&v->counter)); + else + __asm__ __volatile__("lwz%U1%X1 %0,%1" : "=r"(t) : "m<>"(v->counter)); return t; } static __inline__ void arch_atomic_set(atomic_t *v, int i) { - __asm__ __volatile__("stw%U0%X0 %1,%0" : "=m<>"(v->counter) : "r"(i)); + /* -mprefixed can generate offsets beyond range, fall back hack */ + if (IS_ENABLED(CONFIG_PPC_KERNEL_PREFIXED)) + __asm__ __volatile__("stw %1,0(%2)" : "=m"(v->counter) : "r"(i), "b"(&v->counter)); + else + __asm__ __volatile__("stw%U0%X0 %1,%0" : "=m<>"(v->counter) : "r"(i)); } #define ATOMIC_OP(op, asm_op, suffix, sign, ...) \ @@ -197,14 +205,22 @@ static __inline__ s64 arch_atomic64_read(const atomic64_t *v) { s64 t; - __asm__ __volatile__("ld%U1%X1 %0,%1" : "=r"(t) : "m<>"(v->counter)); + /* -mprefixed can generate offsets beyond range, fall back hack */ + if (IS_ENABLED(CONFIG_PPC_KERNEL_PREFIXED)) + __asm__ __volatile__("ld %0,0(%1)" : "=r"(t) : "b"(&v->counter)); + else + __asm__ __volatile__("ld%U1%X1 %0,%1" : "=r"(t) : "m<>"(v->counter)); return t; } static __inline__ void arch_atomic64_set(atomic64_t *v, s64 i) { - __asm__ __volatile__("std%U0%X0 %1,%0" : "=m<>"(v->counter) : "r"(i)); + /* -mprefixed can generate offsets beyond range, fall back hack */ + if (IS_ENABLED(CONFIG_PPC_KERNEL_PREFIXED)) + __asm__ __volatile__("std %1,0(%2)" : "=m"(v->counter) : "r"(i), "b"(&v->counter)); + else + __asm__ __volatile__("std%U0%X0 %1,%0" : "=m<>"(v->counter) : "r"(i)); } #define ATOMIC64_OP(op, asm_op) \ diff --git a/arch/powerpc/include/asm/io.h b/arch/powerpc/include/asm/io.h index fc112a9..f1e657c 100644 --- a/arch/powerpc/include/asm/io.h +++ b/arch/powerpc/include/asm/io.h @@ -97,6 +97,42 @@ extern bool isa_io_special; * */ +/* -mprefixed can generate offsets beyond range, fall back hack */ +#ifdef CONFIG_PPC_KERNEL_PREFIXED +#define DEF_MMIO_IN_X(name, size, insn) \ +static inline u##size name(const volatile u##size __iomem *addr) \ +{ \ + u##size ret; \ + __asm__ __volatile__("sync;"#insn" %0,0,%1;twi 0,%0,0;isync" \ + : "=r" (ret) : "r" (addr) : "memory"); \ + return ret; \ +} + +#define DEF_MMIO_OUT_X(name, size, insn) \ +static inline void name(volatile u##size __iomem *addr, u##size val) \ +{ \ + __asm__ __volatile__("sync;"#insn" %1,0,%0" \ + : : "r" (addr), "r" (val) : "memory"); \ + mmiowb_set_pending(); \ +} + +#define DEF_MMIO_IN_D(name, size, insn) \ +static inline u##size name(const volatile u##size __iomem *addr) \ +{ \ + u##size ret; \ + __asm__ __volatile__("sync;"#insn" %0,0(%1);twi 0,%0,0;isync"\ + : "=r" (ret) : "b" (addr) : "memory"); \ + return ret; \ +} + +#define DEF_MMIO_OUT_D(name, size, insn) \ +static inline void name(volatile u##size __iomem *addr, u##size val) \ +{ \ + __asm__ __volatile__("sync;"#insn" %1,0(%0)" \ + : : "b" (addr), "r" (val) : "memory"); \ + mmiowb_set_pending(); \ +} +#else #define DEF_MMIO_IN_X(name, size, insn) \ static inline u##size name(const volatile u##size __iomem *addr) \ { \ @@ -130,6 +166,7 @@ static inline void name(volatile u##size __iomem *addr, u##size val) \ : "=m<>" (*addr) : "r" (val) : "memory"); \ mmiowb_set_pending(); \ } +#endif DEF_MMIO_IN_D(in_8, 8, lbz); DEF_MMIO_OUT_D(out_8, 8, stb); diff --git a/arch/powerpc/include/asm/uaccess.h b/arch/powerpc/include/asm/uaccess.h index 52378e64..a2d255aa 100644 --- a/arch/powerpc/include/asm/uaccess.h +++ b/arch/powerpc/include/asm/uaccess.h @@ -71,14 +71,26 @@ __pu_failed: \ * because we do not write to any memory gcc knows about, so there * are no aliasing issues. */ +/* -mprefixed can generate offsets beyond range, fall back hack */ +#ifdef CONFIG_PPC_KERNEL_PREFIXED +#define __put_user_asm_goto(x, addr, label, op) \ + asm_volatile_goto( \ + "1: " op " %0,0(%1) # put_user\n" \ + EX_TABLE(1b, %l2) \ + : \ + : "r" (x), "b" (addr) \ + : \ + : label) +#else #define __put_user_asm_goto(x, addr, label, op) \ asm_volatile_goto( \ "1: " op "%U1%X1 %0,%1 # put_user\n" \ EX_TABLE(1b, %l2) \ : \ - : "r" (x), "m<>" (*addr) \ + : "r" (x), "m<>" (*addr) \ : \ : label) +#endif #ifdef __powerpc64__ #define __put_user_asm2_goto(x, ptr, label) \ @@ -131,14 +143,26 @@ do { \ #ifdef CONFIG_CC_HAS_ASM_GOTO_OUTPUT +/* -mprefixed can generate offsets beyond range, fall back hack */ +#ifdef CONFIG_PPC_KERNEL_PREFIXED +#define __get_user_asm_goto(x, addr, label, op) \ + asm_volatile_goto( \ + "1: "op" %0,0(%1) # get_user\n" \ + EX_TABLE(1b, %l2) \ + : "=r" (x) \ + : "b" (addr) \ + : \ + : label) +#else #define __get_user_asm_goto(x, addr, label, op) \ asm_volatile_goto( \ "1: "op"%U1%X1 %0, %1 # get_user\n" \ EX_TABLE(1b, %l2) \ : "=r" (x) \ - : "m<>" (*addr) \ + : "m<>" (*addr) \ : \ : label) +#endif #ifdef __powerpc64__ #define __get_user_asm2_goto(x, addr, label) \ diff --git a/arch/powerpc/kernel/trace/ftrace.c b/arch/powerpc/kernel/trace/ftrace.c index 7b85c3b..72864fb 100644 --- a/arch/powerpc/kernel/trace/ftrace.c +++ b/arch/powerpc/kernel/trace/ftrace.c @@ -194,6 +194,8 @@ __ftrace_make_nop(struct module *mod, * get corrupted. * * Use a b +8 to jump over the load. + * XXX: could make PCREL depend on MPROFILE_KERNEL + * XXX: check PCREL && MPROFILE_KERNEL calling sequence */ if (IS_ENABLED(CONFIG_MPROFILE_KERNEL) || IS_ENABLED(CONFIG_PPC32)) pop = ppc_inst(PPC_RAW_NOP()); diff --git a/arch/powerpc/platforms/Kconfig.cputype b/arch/powerpc/platforms/Kconfig.cputype index 046b5714..1ff0d28 100644 --- a/arch/powerpc/platforms/Kconfig.cputype +++ b/arch/powerpc/platforms/Kconfig.cputype @@ -180,6 +180,7 @@ config POWER10_CPU bool "POWER10" depends on PPC_BOOK3S_64 select ARCH_HAS_FAST_MULTIPLIER + select PPC_HAVE_PREFIXED_SUPPORT config E5500_CPU bool "Freescale e5500" @@ -454,6 +455,22 @@ config PPC_RADIX_MMU_DEFAULT If you're unsure, say Y. +config PPC_KERNEL_PREFIXED + depends on PPC_HAVE_PREFIXED_SUPPORT + depends on CC_HAS_PREFIXED + default n + bool "Build Kernel with Prefixed Instructions" + help + POWER10 and later CPUs support prefixed instructions, 8 byte + instructions that include large immediate, pc relative addressing, + and various floating point, vector, MMA. + + This option builds the kernel with prefixed instructions, and + allows a pc relative addressing option to be selected. + + Kernel support for prefixed instructions in applications and guests + is not affected by this option. + config PPC_KUEP bool "Kernel Userspace Execution Prevention" if !40x default y if !40x @@ -490,6 +507,9 @@ config PPC_MMU_NOHASH config PPC_HAVE_PMU_SUPPORT bool +config PPC_HAVE_PREFIXED_SUPPORT + bool + config PMU_SYSFS bool "Create PMU SPRs sysfs file" default n -- 2.7.4