lkml.org 
[lkml]   [2013]   [Oct]   [16]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v3 3/7] arm64: move encode_insn_immediate() from module.c to insn.c
On Wed, Oct 16, 2013 at 04:18:08AM +0100, Jiang Liu wrote:
> From: Jiang Liu <jiang.liu@huawei.com>
>
> Function encode_insn_immediate() will be used by other instruction
> manipulate related functions, so move it into insn.c and rename it
> as aarch64_insn_encode_immediate().
>
> Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
> Cc: Jiang Liu <liuj97@gmail.com>
> ---
> arch/arm64/include/asm/insn.h | 14 ++++
> arch/arm64/kernel/insn.c | 77 +++++++++++++++++++++
> arch/arm64/kernel/module.c | 151 +++++++++---------------------------------
> 3 files changed, 123 insertions(+), 119 deletions(-)
>
> diff --git a/arch/arm64/include/asm/insn.h b/arch/arm64/include/asm/insn.h
> index 2dfcdb4..8dc0a91 100644
> --- a/arch/arm64/include/asm/insn.h
> +++ b/arch/arm64/include/asm/insn.h
> @@ -28,6 +28,18 @@ enum aarch64_insn_class {
> * system instructions */
> };
>
> +enum aarch64_insn_imm_type {
> + AARCH64_INSN_IMM_MOVNZ,
> + AARCH64_INSN_IMM_MOVK,
> + AARCH64_INSN_IMM_ADR,
> + AARCH64_INSN_IMM_26,
> + AARCH64_INSN_IMM_19,
> + AARCH64_INSN_IMM_16,
> + AARCH64_INSN_IMM_14,
> + AARCH64_INSN_IMM_12,
> + AARCH64_INSN_IMM_9,
> +};
> +
> #define __AARCH64_INSN_FUNCS(abbr, mask, val) \
> static __always_inline bool aarch64_insn_is_##abbr(u32 code) \
> { return (code & (mask)) == (val); } \
> @@ -47,6 +59,8 @@ __AARCH64_INSN_FUNCS(nop, 0xFFFFFFFF, 0xD503201F)
> #undef __AARCH64_INSN_FUNCS
>
> enum aarch64_insn_class aarch64_get_insn_class(u32 insn);
> +u32 aarch64_insn_encode_immediate(enum aarch64_insn_imm_type type,
> + u32 insn, u64 imm);
> u32 aarch64_insn_read(void *addr);
> void aarch64_insn_write(void *addr, u32 insn);
> bool aarch64_insn_hotpatch_safe(u32 old_insn, u32 new_insn);
> diff --git a/arch/arm64/kernel/insn.c b/arch/arm64/kernel/insn.c
> index ad4185f..90cc312 100644
> --- a/arch/arm64/kernel/insn.c
> +++ b/arch/arm64/kernel/insn.c
> @@ -179,3 +179,80 @@ int __kprobes aarch64_insn_patch_text(void *addrs[], u32 insns[], int cnt)
> else
> return aarch64_insn_patch_text_sync(addrs, insns, cnt);
> }
> +
> +u32 aarch64_insn_encode_immediate(enum aarch64_insn_imm_type type,
> + u32 insn, u64 imm)
> +{
> + u32 immlo, immhi, lomask, himask, mask;
> + int shift;
> +
> + switch (type) {
> + case AARCH64_INSN_IMM_MOVNZ:
> + /*
> + * For signed MOVW relocations, we have to manipulate the
> + * instruction encoding depending on whether or not the
> + * immediate is less than zero.
> + */
> + insn &= ~(3 << 29);
> + if ((s64)imm >= 0) {
> + /* >=0: Set the instruction to MOVZ (opcode 10b). */
> + insn |= 2 << 29;
> + } else {
> + /*
> + * <0: Set the instruction to MOVN (opcode 00b).
> + * Since we've masked the opcode already, we
> + * don't need to do anything other than
> + * inverting the new immediate field.
> + */
> + imm = ~imm;
> + }

I'm really not comfortable with this. This code is performing static
relocations and re-encoding instructions as required by the AArch64 ELF
spec. That's not really what you'd expect from a generic instruction
encoder!

Will


\
 
 \ /
  Last update: 2013-10-16 14:21    [W:0.134 / U:10.056 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site