* [PATCH v1 1/6] objtool/LoongArch: Enable objtool to be built
2023-07-25 8:15 [PATCH v1 0/6] Add objtool and orc support for LoongArch Tiezhu Yang
@ 2023-07-25 8:15 ` Tiezhu Yang
2023-07-25 8:15 ` [PATCH v1 2/6] objtool/LoongArch: Implement instruction decoder Tiezhu Yang
` (5 subsequent siblings)
6 siblings, 0 replies; 18+ messages in thread
From: Tiezhu Yang @ 2023-07-25 8:15 UTC (permalink / raw)
To: Josh Poimboeuf, Peter Zijlstra, Huacai Chen
Cc: loongarch, linux-kernel, loongson-kernel
Add the minimal changes to enable objtool build on LoongArch,
most of the functions are stubs to only fix the build errors
when make -C tools/objtool.
This is similar with commit e52ec98c5ab1 ("objtool/powerpc:
Enable objtool to be built on ppc").
Co-developed-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Co-developed-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
tools/objtool/arch/loongarch/Build | 2 +
tools/objtool/arch/loongarch/decode.c | 71 ++++++++++++++++++++++
.../objtool/arch/loongarch/include/arch/cfi_regs.h | 21 +++++++
tools/objtool/arch/loongarch/include/arch/elf.h | 30 +++++++++
.../objtool/arch/loongarch/include/arch/special.h | 33 ++++++++++
tools/objtool/arch/loongarch/special.c | 15 +++++
6 files changed, 172 insertions(+)
create mode 100644 tools/objtool/arch/loongarch/Build
create mode 100644 tools/objtool/arch/loongarch/decode.c
create mode 100644 tools/objtool/arch/loongarch/include/arch/cfi_regs.h
create mode 100644 tools/objtool/arch/loongarch/include/arch/elf.h
create mode 100644 tools/objtool/arch/loongarch/include/arch/special.h
create mode 100644 tools/objtool/arch/loongarch/special.c
diff --git a/tools/objtool/arch/loongarch/Build b/tools/objtool/arch/loongarch/Build
new file mode 100644
index 0000000..d24d563
--- /dev/null
+++ b/tools/objtool/arch/loongarch/Build
@@ -0,0 +1,2 @@
+objtool-y += decode.o
+objtool-y += special.o
diff --git a/tools/objtool/arch/loongarch/decode.c b/tools/objtool/arch/loongarch/decode.c
new file mode 100644
index 0000000..cc74ba4
--- /dev/null
+++ b/tools/objtool/arch/loongarch/decode.c
@@ -0,0 +1,71 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <string.h>
+#include <objtool/check.h>
+
+int arch_ftrace_match(char *name)
+{
+ return !strcmp(name, "_mcount");
+}
+
+unsigned long arch_jump_destination(struct instruction *insn)
+{
+ return insn->offset + (insn->immediate << 2);
+}
+
+unsigned long arch_dest_reloc_offset(int addend)
+{
+ return addend;
+}
+
+bool arch_pc_relative_reloc(struct reloc *reloc)
+{
+ return false;
+}
+
+bool arch_callee_saved_reg(unsigned char reg)
+{
+ switch (reg) {
+ case CFI_RA:
+ case CFI_FP:
+ case CFI_S0 ... CFI_S8:
+ return true;
+ default:
+ return false;
+ }
+}
+
+int arch_decode_hint_reg(u8 sp_reg, int *base)
+{
+ return 0;
+}
+
+int arch_decode_instruction(struct objtool_file *file, const struct section *sec,
+ unsigned long offset, unsigned int maxlen,
+ struct instruction *insn)
+{
+ return 0;
+}
+
+const char *arch_nop_insn(int len)
+{
+ return NULL;
+}
+
+const char *arch_ret_insn(int len)
+{
+ return NULL;
+}
+
+void arch_initial_func_cfi_state(struct cfi_init_state *state)
+{
+ int i;
+
+ for (i = 0; i < CFI_NUM_REGS; i++) {
+ state->regs[i].base = CFI_UNDEFINED;
+ state->regs[i].offset = 0;
+ }
+
+ /* initial CFA (call frame address) */
+ state->cfa.base = CFI_SP;
+ state->cfa.offset = 0;
+}
diff --git a/tools/objtool/arch/loongarch/include/arch/cfi_regs.h b/tools/objtool/arch/loongarch/include/arch/cfi_regs.h
new file mode 100644
index 0000000..c768d39
--- /dev/null
+++ b/tools/objtool/arch/loongarch/include/arch/cfi_regs.h
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef _OBJTOOL_ARCH_CFI_REGS_H
+#define _OBJTOOL_ARCH_CFI_REGS_H
+
+#define CFI_RA 1
+#define CFI_SP 3
+#define CFI_FP 22
+#define CFI_S0 23
+#define CFI_S1 24
+#define CFI_S2 25
+#define CFI_S3 26
+#define CFI_S4 27
+#define CFI_S5 28
+#define CFI_S6 29
+#define CFI_S7 30
+#define CFI_S8 31
+#define CFI_NUM_REGS 32
+
+#define CFI_BP CFI_FP
+
+#endif /* _OBJTOOL_ARCH_CFI_REGS_H */
diff --git a/tools/objtool/arch/loongarch/include/arch/elf.h b/tools/objtool/arch/loongarch/include/arch/elf.h
new file mode 100644
index 0000000..9623d66
--- /dev/null
+++ b/tools/objtool/arch/loongarch/include/arch/elf.h
@@ -0,0 +1,30 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef _OBJTOOL_ARCH_ELF_H
+#define _OBJTOOL_ARCH_ELF_H
+
+/*
+ * See the following link for more info about ELF Relocation types:
+ * https://loongson.github.io/LoongArch-Documentation/LoongArch-ELF-ABI-EN.html#_relocations
+ */
+#ifndef R_LARCH_NONE
+#define R_LARCH_NONE 0
+#endif
+#ifndef R_LARCH_32
+#define R_LARCH_32 1
+#endif
+#ifndef R_LARCH_64
+#define R_LARCH_64 2
+#endif
+#ifndef R_LARCH_32_PCREL
+#define R_LARCH_32_PCREL 99
+#endif
+
+#define R_NONE R_LARCH_NONE
+#define R_ABS32 R_LARCH_32
+#define R_ABS64 R_LARCH_64
+#define R_DATA32 R_LARCH_32_PCREL
+#define R_DATA64 R_LARCH_32_PCREL
+#define R_TEXT32 R_LARCH_32_PCREL
+#define R_TEXT64 R_LARCH_32_PCREL
+
+#endif /* _OBJTOOL_ARCH_ELF_H */
diff --git a/tools/objtool/arch/loongarch/include/arch/special.h b/tools/objtool/arch/loongarch/include/arch/special.h
new file mode 100644
index 0000000..1a8245c
--- /dev/null
+++ b/tools/objtool/arch/loongarch/include/arch/special.h
@@ -0,0 +1,33 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef _OBJTOOL_ARCH_SPECIAL_H
+#define _OBJTOOL_ARCH_SPECIAL_H
+
+/*
+ * See more info about struct exception_table_entry
+ * in arch/loongarch/include/asm/extable.h
+ */
+#define EX_ENTRY_SIZE 12
+#define EX_ORIG_OFFSET 0
+#define EX_NEW_OFFSET 4
+
+/*
+ * See more info about struct jump_entry
+ * in include/linux/jump_label.h
+ */
+#define JUMP_ENTRY_SIZE 16
+#define JUMP_ORIG_OFFSET 0
+#define JUMP_NEW_OFFSET 4
+#define JUMP_KEY_OFFSET 8
+
+/*
+ * See more info about struct alt_instr
+ * in arch/loongarch/include/asm/alternative.h
+ */
+#define ALT_ENTRY_SIZE 12
+#define ALT_ORIG_OFFSET 0
+#define ALT_NEW_OFFSET 4
+#define ALT_FEATURE_OFFSET 8
+#define ALT_ORIG_LEN_OFFSET 10
+#define ALT_NEW_LEN_OFFSET 11
+
+#endif /* _OBJTOOL_ARCH_SPECIAL_H */
diff --git a/tools/objtool/arch/loongarch/special.c b/tools/objtool/arch/loongarch/special.c
new file mode 100644
index 0000000..9bba1e9
--- /dev/null
+++ b/tools/objtool/arch/loongarch/special.c
@@ -0,0 +1,15 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <objtool/special.h>
+
+bool arch_support_alt_relocation(struct special_alt *special_alt,
+ struct instruction *insn,
+ struct reloc *reloc)
+{
+ return false;
+}
+
+struct reloc *arch_find_switch_table(struct objtool_file *file,
+ struct instruction *insn)
+{
+ return NULL;
+}
--
2.1.0
^ permalink raw reply [flat|nested] 18+ messages in thread* [PATCH v1 2/6] objtool/LoongArch: Implement instruction decoder
2023-07-25 8:15 [PATCH v1 0/6] Add objtool and orc support for LoongArch Tiezhu Yang
2023-07-25 8:15 ` [PATCH v1 1/6] objtool/LoongArch: Enable objtool to be built Tiezhu Yang
@ 2023-07-25 8:15 ` Tiezhu Yang
2023-07-25 8:15 ` [PATCH v1 3/6] objtool/x86: Separate arch-specific and generic parts Tiezhu Yang
` (4 subsequent siblings)
6 siblings, 0 replies; 18+ messages in thread
From: Tiezhu Yang @ 2023-07-25 8:15 UTC (permalink / raw)
To: Josh Poimboeuf, Peter Zijlstra, Huacai Chen
Cc: loongarch, linux-kernel, loongson-kernel
Only copy the minimal definitions of instruction opcodes and formats
in inst.h from arch/loongarch to tools/arch/loongarch, and also copy
the definition of sign_extend64() to tools/include/linux/bitops.h to
decode the following kinds of instructions:
(1) stack pointer related instructions
addi.d, ld.d, st.d, ldptr.d and stptr.d
(2) branch and jump related instructions
beq, bne, blt, bge, bltu, bgeu, beqz, bnez, b, bl and jirl
(3) other instructions
break, nop and ertn
See more info about instructions in LoongArch Reference Manual:
https://loongson.github.io/LoongArch-Documentation/LoongArch-Vol1-EN.html
Co-developed-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Co-developed-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
tools/arch/loongarch/include/asm/inst.h | 159 +++++++++++++++++++++
tools/include/linux/bitops.h | 11 ++
tools/objtool/arch/loongarch/decode.c | 245 +++++++++++++++++++++++++++++++-
3 files changed, 413 insertions(+), 2 deletions(-)
create mode 100644 tools/arch/loongarch/include/asm/inst.h
diff --git a/tools/arch/loongarch/include/asm/inst.h b/tools/arch/loongarch/include/asm/inst.h
new file mode 100644
index 0000000..407c49c
--- /dev/null
+++ b/tools/arch/loongarch/include/asm/inst.h
@@ -0,0 +1,159 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2020-2022 Loongson Technology Corporation Limited
+ */
+#ifndef _ASM_INST_H
+#define _ASM_INST_H
+
+#include <linux/bitops.h>
+
+#define LOONGARCH_INSN_NOP 0x03400000
+
+enum reg0i15_op {
+ break_op = 0x54,
+};
+
+enum reg0i26_op {
+ b_op = 0x14,
+ bl_op = 0x15,
+};
+
+enum reg1i21_op {
+ beqz_op = 0x10,
+ bnez_op = 0x11,
+};
+
+enum reg2_op {
+ ertn_op = 0x1920e,
+};
+
+enum reg2i12_op {
+ addid_op = 0x0b,
+ andi_op = 0x0d,
+ ldd_op = 0xa3,
+ std_op = 0xa7,
+};
+
+enum reg2i14_op {
+ ldptrd_op = 0x26,
+ stptrd_op = 0x27,
+};
+
+enum reg2i16_op {
+ jirl_op = 0x13,
+ beq_op = 0x16,
+ bne_op = 0x17,
+ blt_op = 0x18,
+ bge_op = 0x19,
+ bltu_op = 0x1a,
+ bgeu_op = 0x1b,
+};
+
+struct reg0i15_format {
+ unsigned int immediate : 15;
+ unsigned int opcode : 17;
+};
+
+struct reg0i26_format {
+ unsigned int immediate_h : 10;
+ unsigned int immediate_l : 16;
+ unsigned int opcode : 6;
+};
+
+struct reg1i21_format {
+ unsigned int immediate_h : 5;
+ unsigned int rj : 5;
+ unsigned int immediate_l : 16;
+ unsigned int opcode : 6;
+};
+
+struct reg2_format {
+ unsigned int rd : 5;
+ unsigned int rj : 5;
+ unsigned int opcode : 22;
+};
+
+struct reg2i12_format {
+ unsigned int rd : 5;
+ unsigned int rj : 5;
+ unsigned int immediate : 12;
+ unsigned int opcode : 10;
+};
+
+struct reg2i14_format {
+ unsigned int rd : 5;
+ unsigned int rj : 5;
+ unsigned int immediate : 14;
+ unsigned int opcode : 8;
+};
+
+struct reg2i16_format {
+ unsigned int rd : 5;
+ unsigned int rj : 5;
+ unsigned int immediate : 16;
+ unsigned int opcode : 6;
+};
+
+union loongarch_instruction {
+ unsigned int word;
+ struct reg0i15_format reg0i15_format;
+ struct reg0i26_format reg0i26_format;
+ struct reg1i21_format reg1i21_format;
+ struct reg2_format reg2_format;
+ struct reg2i12_format reg2i12_format;
+ struct reg2i14_format reg2i14_format;
+ struct reg2i16_format reg2i16_format;
+};
+
+#define LOONGARCH_INSN_SIZE sizeof(union loongarch_instruction)
+
+enum loongarch_gpr {
+ LOONGARCH_GPR_ZERO = 0,
+ LOONGARCH_GPR_RA = 1,
+ LOONGARCH_GPR_TP = 2,
+ LOONGARCH_GPR_SP = 3,
+ LOONGARCH_GPR_A0 = 4, /* Reused as V0 for return value */
+ LOONGARCH_GPR_A1, /* Reused as V1 for return value */
+ LOONGARCH_GPR_A2,
+ LOONGARCH_GPR_A3,
+ LOONGARCH_GPR_A4,
+ LOONGARCH_GPR_A5,
+ LOONGARCH_GPR_A6,
+ LOONGARCH_GPR_A7,
+ LOONGARCH_GPR_T0 = 12,
+ LOONGARCH_GPR_T1,
+ LOONGARCH_GPR_T2,
+ LOONGARCH_GPR_T3,
+ LOONGARCH_GPR_T4,
+ LOONGARCH_GPR_T5,
+ LOONGARCH_GPR_T6,
+ LOONGARCH_GPR_T7,
+ LOONGARCH_GPR_T8,
+ LOONGARCH_GPR_FP = 22,
+ LOONGARCH_GPR_S0 = 23,
+ LOONGARCH_GPR_S1,
+ LOONGARCH_GPR_S2,
+ LOONGARCH_GPR_S3,
+ LOONGARCH_GPR_S4,
+ LOONGARCH_GPR_S5,
+ LOONGARCH_GPR_S6,
+ LOONGARCH_GPR_S7,
+ LOONGARCH_GPR_S8,
+ LOONGARCH_GPR_MAX
+};
+
+#define DEF_EMIT_REG2I16_FORMAT(NAME, OP) \
+static inline void emit_##NAME(union loongarch_instruction *insn, \
+ enum loongarch_gpr rj, \
+ enum loongarch_gpr rd, \
+ int offset) \
+{ \
+ insn->reg2i16_format.opcode = OP; \
+ insn->reg2i16_format.immediate = offset; \
+ insn->reg2i16_format.rj = rj; \
+ insn->reg2i16_format.rd = rd; \
+}
+
+DEF_EMIT_REG2I16_FORMAT(jirl, jirl_op)
+
+#endif /* _ASM_INST_H */
diff --git a/tools/include/linux/bitops.h b/tools/include/linux/bitops.h
index f18683b9..7319f6c 100644
--- a/tools/include/linux/bitops.h
+++ b/tools/include/linux/bitops.h
@@ -87,4 +87,15 @@ static inline __u32 rol32(__u32 word, unsigned int shift)
return (word << shift) | (word >> ((-shift) & 31));
}
+/**
+ * sign_extend64 - sign extend a 64-bit value using specified bit as sign-bit
+ * @value: value to sign extend
+ * @index: 0 based bit index (0<=index<64) to sign bit
+ */
+static __always_inline __s64 sign_extend64(__u64 value, int index)
+{
+ __u8 shift = 63 - index;
+ return (__s64)(value << shift) >> shift;
+}
+
#endif
diff --git a/tools/objtool/arch/loongarch/decode.c b/tools/objtool/arch/loongarch/decode.c
index cc74ba4..2904fbc 100644
--- a/tools/objtool/arch/loongarch/decode.c
+++ b/tools/objtool/arch/loongarch/decode.c
@@ -1,6 +1,8 @@
// SPDX-License-Identifier: GPL-2.0-or-later
#include <string.h>
#include <objtool/check.h>
+#include <objtool/warn.h>
+#include <asm/inst.h>
int arch_ftrace_match(char *name)
{
@@ -39,21 +41,260 @@ int arch_decode_hint_reg(u8 sp_reg, int *base)
return 0;
}
+static bool is_loongarch(const struct elf *elf)
+{
+ if (elf->ehdr.e_machine == EM_LOONGARCH)
+ return true;
+
+ WARN("unexpected ELF machine type %d", elf->ehdr.e_machine);
+ return false;
+}
+
+#define ADD_OP(op) \
+ if (!(op = calloc(1, sizeof(*op)))) \
+ return -1; \
+ else for (*ops_list = op, ops_list = &op->next; op; op = NULL)
+
+static bool decode_insn_reg0i26_fomat(union loongarch_instruction inst,
+ struct instruction *insn)
+{
+ switch (inst.reg0i26_format.opcode) {
+ case b_op:
+ insn->type = INSN_JUMP_UNCONDITIONAL;
+ insn->immediate = sign_extend64(inst.reg0i26_format.immediate_h << 16 |
+ inst.reg0i26_format.immediate_l, 25);
+ break;
+ case bl_op:
+ insn->type = INSN_CALL;
+ insn->immediate = sign_extend64(inst.reg0i26_format.immediate_h << 16 |
+ inst.reg0i26_format.immediate_l, 25);
+ break;
+ default:
+ return false;
+ }
+
+ return true;
+}
+
+static bool decode_insn_reg1i21_fomat(union loongarch_instruction inst,
+ struct instruction *insn)
+{
+ switch (inst.reg1i21_format.opcode) {
+ case beqz_op:
+ case bnez_op:
+ insn->type = INSN_JUMP_CONDITIONAL;
+ insn->immediate = sign_extend64(inst.reg1i21_format.immediate_h << 16 |
+ inst.reg1i21_format.immediate_l, 20);
+ break;
+ default:
+ return false;
+ }
+
+ return true;
+}
+
+static bool decode_insn_reg2i12_fomat(union loongarch_instruction inst,
+ struct instruction *insn,
+ struct stack_op **ops_list,
+ struct stack_op *op)
+{
+ switch (inst.reg2i12_format.opcode) {
+ case addid_op:
+ if ((inst.reg2i12_format.rd == CFI_SP) || (inst.reg2i12_format.rj == CFI_SP)) {
+ /* addi.d sp,sp,si12 or addi.d fp,sp,si12 */
+ insn->immediate = sign_extend64(inst.reg2i12_format.immediate, 11);
+ ADD_OP(op) {
+ op->src.type = OP_SRC_ADD;
+ op->src.reg = inst.reg2i12_format.rj;
+ op->src.offset = insn->immediate;
+ op->dest.type = OP_DEST_REG;
+ op->dest.reg = inst.reg2i12_format.rd;
+ }
+ }
+ break;
+ case ldd_op:
+ if (inst.reg2i12_format.rj == CFI_SP) {
+ /* ld.d rd,sp,si12 */
+ insn->immediate = sign_extend64(inst.reg2i12_format.immediate, 11);
+ ADD_OP(op) {
+ op->src.type = OP_SRC_REG_INDIRECT;
+ op->src.reg = CFI_SP;
+ op->src.offset = insn->immediate;
+ op->dest.type = OP_DEST_REG;
+ op->dest.reg = inst.reg2i12_format.rd;
+ }
+ }
+ break;
+ case std_op:
+ if (inst.reg2i12_format.rj == CFI_SP) {
+ /* st.d rd,sp,si12 */
+ insn->immediate = sign_extend64(inst.reg2i12_format.immediate, 11);
+ ADD_OP(op) {
+ op->src.type = OP_SRC_REG;
+ op->src.reg = inst.reg2i12_format.rd;
+ op->dest.type = OP_DEST_REG_INDIRECT;
+ op->dest.reg = CFI_SP;
+ op->dest.offset = insn->immediate;
+ }
+ }
+ break;
+ case andi_op:
+ if (inst.reg2i12_format.rd == 0 &&
+ inst.reg2i12_format.rj == 0 &&
+ inst.reg2i12_format.immediate == 0)
+ /* andi r0,r0,0 */
+ insn->type = INSN_NOP;
+ break;
+ default:
+ return false;
+ }
+
+ return true;
+}
+
+static bool decode_insn_reg2i14_fomat(union loongarch_instruction inst,
+ struct instruction *insn,
+ struct stack_op **ops_list,
+ struct stack_op *op)
+{
+ switch (inst.reg2i14_format.opcode) {
+ case ldptrd_op:
+ if (inst.reg2i14_format.rj == CFI_SP) {
+ /* ldptr.d rd,sp,si14 */
+ insn->immediate = sign_extend64(inst.reg2i14_format.immediate, 13);
+ ADD_OP(op) {
+ op->src.type = OP_SRC_REG_INDIRECT;
+ op->src.reg = CFI_SP;
+ op->src.offset = insn->immediate;
+ op->dest.type = OP_DEST_REG;
+ op->dest.reg = inst.reg2i14_format.rd;
+ }
+ }
+ break;
+ case stptrd_op:
+ if (inst.reg2i14_format.rj == CFI_SP) {
+ /* stptr.d rd,sp,si14 */
+ insn->immediate = sign_extend64(inst.reg2i14_format.immediate, 13);
+ ADD_OP(op) {
+ op->src.type = OP_SRC_REG;
+ op->src.reg = inst.reg2i14_format.rd;
+ op->dest.type = OP_DEST_REG_INDIRECT;
+ op->dest.reg = CFI_SP;
+ op->dest.offset = insn->immediate;
+ }
+ }
+ break;
+ default:
+ return false;
+ }
+
+ return true;
+}
+
+static bool decode_insn_reg2i16_fomat(union loongarch_instruction inst,
+ struct instruction *insn)
+{
+ switch (inst.reg2i16_format.opcode) {
+ case jirl_op:
+ if (inst.reg2i16_format.rd == 0 &&
+ inst.reg2i16_format.rj == CFI_RA) {
+ /* jirl r0,ra,offs16 */
+ insn->type = INSN_RETURN;
+ } else if (inst.reg2i16_format.rd == CFI_RA) {
+ /* jirl ra,rj,offs16 */
+ insn->type = INSN_CALL_DYNAMIC;
+ } else if (inst.reg2i16_format.rd == 0) {
+ /* jirl r0,rj,offs16 */
+ insn->type = INSN_JUMP_DYNAMIC;
+ } else {
+ /* jirl rd,rj,offs16 */
+ insn->type = INSN_JUMP_UNCONDITIONAL;
+ }
+ insn->immediate = sign_extend64(inst.reg2i16_format.immediate, 15);
+ break;
+ case beq_op:
+ case bne_op:
+ case blt_op:
+ case bge_op:
+ case bltu_op:
+ case bgeu_op:
+ insn->type = INSN_JUMP_CONDITIONAL;
+ insn->immediate = sign_extend64(inst.reg2i16_format.immediate, 15);
+ break;
+ default:
+ return false;
+ }
+
+ return true;
+}
+
int arch_decode_instruction(struct objtool_file *file, const struct section *sec,
unsigned long offset, unsigned int maxlen,
struct instruction *insn)
{
+ struct stack_op **ops_list = &insn->stack_ops;
+ const struct elf *elf = file->elf;
+ struct stack_op *op = NULL;
+ union loongarch_instruction inst;
+
+ if (!is_loongarch(elf))
+ return -1;
+
+ if (maxlen < LOONGARCH_INSN_SIZE)
+ return 0;
+
+ insn->len = LOONGARCH_INSN_SIZE;
+ insn->type = INSN_OTHER;
+ insn->immediate = 0;
+
+ inst = *(union loongarch_instruction *)(sec->data->d_buf + offset);
+
+ if (decode_insn_reg0i26_fomat(inst, insn))
+ return 0;
+ if (decode_insn_reg1i21_fomat(inst, insn))
+ return 0;
+ if (decode_insn_reg2i12_fomat(inst, insn, ops_list, op))
+ return 0;
+ if (decode_insn_reg2i14_fomat(inst, insn, ops_list, op))
+ return 0;
+ if (decode_insn_reg2i16_fomat(inst, insn))
+ return 0;
+
+ if (inst.word == 0)
+ insn->type = INSN_NOP;
+ else if (inst.reg0i15_format.opcode == break_op) {
+ /* break */
+ insn->type = INSN_BUG;
+ } else if (inst.reg2_format.opcode == ertn_op) {
+ /* ertn */
+ insn->type = INSN_RETURN;
+ }
+
return 0;
}
const char *arch_nop_insn(int len)
{
- return NULL;
+ static u32 nop;
+
+ if (len != LOONGARCH_INSN_SIZE)
+ WARN("invalid NOP size: %d\n", len);
+
+ nop = LOONGARCH_INSN_NOP;
+
+ return (const char *)&nop;
}
const char *arch_ret_insn(int len)
{
- return NULL;
+ static u32 ret;
+
+ if (len != LOONGARCH_INSN_SIZE)
+ WARN("invalid RET size: %d\n", len);
+
+ emit_jirl((union loongarch_instruction *)&ret, LOONGARCH_GPR_RA, LOONGARCH_GPR_ZERO, 0);
+
+ return (const char *)&ret;
}
void arch_initial_func_cfi_state(struct cfi_init_state *state)
--
2.1.0
^ permalink raw reply [flat|nested] 18+ messages in thread* [PATCH v1 3/6] objtool/x86: Separate arch-specific and generic parts
2023-07-25 8:15 [PATCH v1 0/6] Add objtool and orc support for LoongArch Tiezhu Yang
2023-07-25 8:15 ` [PATCH v1 1/6] objtool/LoongArch: Enable objtool to be built Tiezhu Yang
2023-07-25 8:15 ` [PATCH v1 2/6] objtool/LoongArch: Implement instruction decoder Tiezhu Yang
@ 2023-07-25 8:15 ` Tiezhu Yang
2023-07-25 8:15 ` [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built Tiezhu Yang
` (3 subsequent siblings)
6 siblings, 0 replies; 18+ messages in thread
From: Tiezhu Yang @ 2023-07-25 8:15 UTC (permalink / raw)
To: Josh Poimboeuf, Peter Zijlstra, Huacai Chen
Cc: loongarch, linux-kernel, loongson-kernel
Move init_orc_entry(), reg_name(), orc_type_name() and print_reg() from
generic orc_gen.c and orc_dump.c to arch-specific orc.c, then introduce
a new function orc_print_dump() to print info.
Define update_cfi_state() as a weak function which may be overwritten
by the arch-specific implementation.
This is preparation for later patch, no functionality change.
Co-developed-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Co-developed-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
tools/objtool/arch/x86/Build | 1 +
tools/objtool/arch/x86/orc.c | 169 ++++++++++++++++++++++++++++++++++
tools/objtool/check.c | 12 +--
tools/objtool/include/objtool/check.h | 9 ++
tools/objtool/include/objtool/orc.h | 10 ++
tools/objtool/orc_dump.c | 69 +-------------
tools/objtool/orc_gen.c | 92 +-----------------
7 files changed, 198 insertions(+), 164 deletions(-)
create mode 100644 tools/objtool/arch/x86/orc.c
create mode 100644 tools/objtool/include/objtool/orc.h
diff --git a/tools/objtool/arch/x86/Build b/tools/objtool/arch/x86/Build
index 9f7869b..3dedb2f 100644
--- a/tools/objtool/arch/x86/Build
+++ b/tools/objtool/arch/x86/Build
@@ -1,5 +1,6 @@
objtool-y += special.o
objtool-y += decode.o
+objtool-y += orc.o
inat_tables_script = ../arch/x86/tools/gen-insn-attr-x86.awk
inat_tables_maps = ../arch/x86/lib/x86-opcode-map.txt
diff --git a/tools/objtool/arch/x86/orc.c b/tools/objtool/arch/x86/orc.c
new file mode 100644
index 0000000..a4365b8
--- /dev/null
+++ b/tools/objtool/arch/x86/orc.c
@@ -0,0 +1,169 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <linux/objtool_types.h>
+#include <asm/orc_types.h>
+
+#include <objtool/check.h>
+#include <objtool/orc.h>
+#include <objtool/warn.h>
+#include <objtool/endianness.h>
+
+int init_orc_entry(struct orc_entry *orc, struct cfi_state *cfi, struct instruction *insn)
+{
+ struct cfi_reg *bp = &cfi->regs[CFI_BP];
+
+ memset(orc, 0, sizeof(*orc));
+
+ if (!cfi) {
+ /*
+ * This is usually either unreachable nops/traps (which don't
+ * trigger unreachable instruction warnings), or
+ * STACK_FRAME_NON_STANDARD functions.
+ */
+ orc->type = ORC_TYPE_UNDEFINED;
+ return 0;
+ }
+
+ switch (cfi->type) {
+ case UNWIND_HINT_TYPE_UNDEFINED:
+ orc->type = ORC_TYPE_UNDEFINED;
+ return 0;
+ case UNWIND_HINT_TYPE_END_OF_STACK:
+ orc->type = ORC_TYPE_END_OF_STACK;
+ return 0;
+ case UNWIND_HINT_TYPE_CALL:
+ orc->type = ORC_TYPE_CALL;
+ break;
+ case UNWIND_HINT_TYPE_REGS:
+ orc->type = ORC_TYPE_REGS;
+ break;
+ case UNWIND_HINT_TYPE_REGS_PARTIAL:
+ orc->type = ORC_TYPE_REGS_PARTIAL;
+ break;
+ default:
+ WARN_INSN(insn, "unknown unwind hint type %d", cfi->type);
+ return -1;
+ }
+
+ orc->signal = cfi->signal;
+
+ switch (cfi->cfa.base) {
+ case CFI_SP:
+ orc->sp_reg = ORC_REG_SP;
+ break;
+ case CFI_SP_INDIRECT:
+ orc->sp_reg = ORC_REG_SP_INDIRECT;
+ break;
+ case CFI_BP:
+ orc->sp_reg = ORC_REG_BP;
+ break;
+ case CFI_BP_INDIRECT:
+ orc->sp_reg = ORC_REG_BP_INDIRECT;
+ break;
+ case CFI_R10:
+ orc->sp_reg = ORC_REG_R10;
+ break;
+ case CFI_R13:
+ orc->sp_reg = ORC_REG_R13;
+ break;
+ case CFI_DI:
+ orc->sp_reg = ORC_REG_DI;
+ break;
+ case CFI_DX:
+ orc->sp_reg = ORC_REG_DX;
+ break;
+ default:
+ WARN_INSN(insn, "unknown CFA base reg %d", cfi->cfa.base);
+ return -1;
+ }
+
+ switch (bp->base) {
+ case CFI_UNDEFINED:
+ orc->bp_reg = ORC_REG_UNDEFINED;
+ break;
+ case CFI_CFA:
+ orc->bp_reg = ORC_REG_PREV_SP;
+ break;
+ case CFI_BP:
+ orc->bp_reg = ORC_REG_BP;
+ break;
+ default:
+ WARN_INSN(insn, "unknown BP base reg %d", bp->base);
+ return -1;
+ }
+
+ orc->sp_offset = cfi->cfa.offset;
+ orc->bp_offset = bp->offset;
+
+ return 0;
+}
+
+static const char *reg_name(unsigned int reg)
+{
+ switch (reg) {
+ case ORC_REG_PREV_SP:
+ return "prevsp";
+ case ORC_REG_DX:
+ return "dx";
+ case ORC_REG_DI:
+ return "di";
+ case ORC_REG_BP:
+ return "bp";
+ case ORC_REG_SP:
+ return "sp";
+ case ORC_REG_R10:
+ return "r10";
+ case ORC_REG_R13:
+ return "r13";
+ case ORC_REG_BP_INDIRECT:
+ return "bp(ind)";
+ case ORC_REG_SP_INDIRECT:
+ return "sp(ind)";
+ default:
+ return "?";
+ }
+}
+
+static const char *orc_type_name(unsigned int type)
+{
+ switch (type) {
+ case ORC_TYPE_UNDEFINED:
+ return "(und)";
+ case ORC_TYPE_END_OF_STACK:
+ return "end";
+ case ORC_TYPE_CALL:
+ return "call";
+ case ORC_TYPE_REGS:
+ return "regs";
+ case ORC_TYPE_REGS_PARTIAL:
+ return "regs (partial)";
+ default:
+ return "?";
+ }
+}
+
+static void print_reg(unsigned int reg, int offset)
+{
+ if (reg == ORC_REG_BP_INDIRECT)
+ printf("(bp%+d)", offset);
+ else if (reg == ORC_REG_SP_INDIRECT)
+ printf("(sp)%+d", offset);
+ else if (reg == ORC_REG_UNDEFINED)
+ printf("(und)");
+ else
+ printf("%s%+d", reg_name(reg), offset);
+}
+
+void orc_print_dump(struct elf *dummy_elf, struct orc_entry *orc, int i)
+{
+ printf("type:%s", orc_type_name(orc[i].type));
+
+ printf(" sp:");
+
+ print_reg(orc[i].sp_reg, bswap_if_needed(dummy_elf, orc[i].sp_offset));
+
+ printf(" bp:");
+
+ print_reg(orc[i].bp_reg, bswap_if_needed(dummy_elf, orc[i].bp_offset));
+
+ printf(" signal:%d\n", orc[i].signal);
+}
diff --git a/tools/objtool/check.c b/tools/objtool/check.c
index 8936a05..9e5e462 100644
--- a/tools/objtool/check.c
+++ b/tools/objtool/check.c
@@ -2685,9 +2685,9 @@ static bool has_valid_stack_frame(struct insn_state *state)
return false;
}
-static int update_cfi_state_regs(struct instruction *insn,
- struct cfi_state *cfi,
- struct stack_op *op)
+int update_cfi_state_regs(struct instruction *insn,
+ struct cfi_state *cfi,
+ struct stack_op *op)
{
struct cfi_reg *cfa = &cfi->cfa;
@@ -2710,7 +2710,7 @@ static int update_cfi_state_regs(struct instruction *insn,
return 0;
}
-static void save_reg(struct cfi_state *cfi, unsigned char reg, int base, int offset)
+void save_reg(struct cfi_state *cfi, unsigned char reg, int base, int offset)
{
if (arch_callee_saved_reg(reg) &&
cfi->regs[reg].base == CFI_UNDEFINED) {
@@ -2719,7 +2719,7 @@ static void save_reg(struct cfi_state *cfi, unsigned char reg, int base, int off
}
}
-static void restore_reg(struct cfi_state *cfi, unsigned char reg)
+void restore_reg(struct cfi_state *cfi, unsigned char reg)
{
cfi->regs[reg].base = initial_func_cfi.regs[reg].base;
cfi->regs[reg].offset = initial_func_cfi.regs[reg].offset;
@@ -2778,7 +2778,7 @@ static void restore_reg(struct cfi_state *cfi, unsigned char reg)
* 41 5d pop %r13
* c3 retq
*/
-static int update_cfi_state(struct instruction *insn,
+int __weak update_cfi_state(struct instruction *insn,
struct instruction *next_insn,
struct cfi_state *cfi, struct stack_op *op)
{
diff --git a/tools/objtool/include/objtool/check.h b/tools/objtool/include/objtool/check.h
index daa46f1..620c9c0 100644
--- a/tools/objtool/include/objtool/check.h
+++ b/tools/objtool/include/objtool/check.h
@@ -121,4 +121,13 @@ struct instruction *next_insn_same_sec(struct objtool_file *file, struct instruc
insn && insn->sec == _sec; \
insn = next_insn_same_sec(file, insn))
+int update_cfi_state_regs(struct instruction *insn,
+ struct cfi_state *cfi,
+ struct stack_op *op);
+void save_reg(struct cfi_state *cfi, unsigned char reg, int base, int offset);
+void restore_reg(struct cfi_state *cfi, unsigned char reg);
+int update_cfi_state(struct instruction *insn,
+ struct instruction *next_insn,
+ struct cfi_state *cfi, struct stack_op *op);
+
#endif /* _CHECK_H */
diff --git a/tools/objtool/include/objtool/orc.h b/tools/objtool/include/objtool/orc.h
new file mode 100644
index 0000000..4c9f946
--- /dev/null
+++ b/tools/objtool/include/objtool/orc.h
@@ -0,0 +1,10 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef _OBJTOOL_ORC_H
+#define _OBJTOOL_ORC_H
+
+#include <objtool/check.h>
+
+int init_orc_entry(struct orc_entry *orc, struct cfi_state *cfi, struct instruction *insn);
+void orc_print_dump(struct elf *dummy_elf, struct orc_entry *orc, int i);
+
+#endif /* _OBJTOOL_ORC_H */
diff --git a/tools/objtool/orc_dump.c b/tools/objtool/orc_dump.c
index 0e183bb..a62247ef 100644
--- a/tools/objtool/orc_dump.c
+++ b/tools/objtool/orc_dump.c
@@ -6,65 +6,10 @@
#include <unistd.h>
#include <asm/orc_types.h>
#include <objtool/objtool.h>
+#include <objtool/orc.h>
#include <objtool/warn.h>
#include <objtool/endianness.h>
-static const char *reg_name(unsigned int reg)
-{
- switch (reg) {
- case ORC_REG_PREV_SP:
- return "prevsp";
- case ORC_REG_DX:
- return "dx";
- case ORC_REG_DI:
- return "di";
- case ORC_REG_BP:
- return "bp";
- case ORC_REG_SP:
- return "sp";
- case ORC_REG_R10:
- return "r10";
- case ORC_REG_R13:
- return "r13";
- case ORC_REG_BP_INDIRECT:
- return "bp(ind)";
- case ORC_REG_SP_INDIRECT:
- return "sp(ind)";
- default:
- return "?";
- }
-}
-
-static const char *orc_type_name(unsigned int type)
-{
- switch (type) {
- case ORC_TYPE_UNDEFINED:
- return "(und)";
- case ORC_TYPE_END_OF_STACK:
- return "end";
- case ORC_TYPE_CALL:
- return "call";
- case ORC_TYPE_REGS:
- return "regs";
- case ORC_TYPE_REGS_PARTIAL:
- return "regs (partial)";
- default:
- return "?";
- }
-}
-
-static void print_reg(unsigned int reg, int offset)
-{
- if (reg == ORC_REG_BP_INDIRECT)
- printf("(bp%+d)", offset);
- else if (reg == ORC_REG_SP_INDIRECT)
- printf("(sp)%+d", offset);
- else if (reg == ORC_REG_UNDEFINED)
- printf("(und)");
- else
- printf("%s%+d", reg_name(reg), offset);
-}
-
int orc_dump(const char *_objname)
{
int fd, nr_entries, i, *orc_ip = NULL, orc_size = 0;
@@ -205,17 +150,7 @@ int orc_dump(const char *_objname)
printf("%llx:", (unsigned long long)(orc_ip_addr + (i * sizeof(int)) + orc_ip[i]));
}
- printf("type:%s", orc_type_name(orc[i].type));
-
- printf(" sp:");
-
- print_reg(orc[i].sp_reg, bswap_if_needed(&dummy_elf, orc[i].sp_offset));
-
- printf(" bp:");
-
- print_reg(orc[i].bp_reg, bswap_if_needed(&dummy_elf, orc[i].bp_offset));
-
- printf(" signal:%d\n", orc[i].signal);
+ orc_print_dump(&dummy_elf, orc, i);
}
elf_end(elf);
diff --git a/tools/objtool/orc_gen.c b/tools/objtool/orc_gen.c
index bae3439..1eff7e0a 100644
--- a/tools/objtool/orc_gen.c
+++ b/tools/objtool/orc_gen.c
@@ -10,100 +10,10 @@
#include <asm/orc_types.h>
#include <objtool/check.h>
+#include <objtool/orc.h>
#include <objtool/warn.h>
#include <objtool/endianness.h>
-static int init_orc_entry(struct orc_entry *orc, struct cfi_state *cfi,
- struct instruction *insn)
-{
- struct cfi_reg *bp = &cfi->regs[CFI_BP];
-
- memset(orc, 0, sizeof(*orc));
-
- if (!cfi) {
- /*
- * This is usually either unreachable nops/traps (which don't
- * trigger unreachable instruction warnings), or
- * STACK_FRAME_NON_STANDARD functions.
- */
- orc->type = ORC_TYPE_UNDEFINED;
- return 0;
- }
-
- switch (cfi->type) {
- case UNWIND_HINT_TYPE_UNDEFINED:
- orc->type = ORC_TYPE_UNDEFINED;
- return 0;
- case UNWIND_HINT_TYPE_END_OF_STACK:
- orc->type = ORC_TYPE_END_OF_STACK;
- return 0;
- case UNWIND_HINT_TYPE_CALL:
- orc->type = ORC_TYPE_CALL;
- break;
- case UNWIND_HINT_TYPE_REGS:
- orc->type = ORC_TYPE_REGS;
- break;
- case UNWIND_HINT_TYPE_REGS_PARTIAL:
- orc->type = ORC_TYPE_REGS_PARTIAL;
- break;
- default:
- WARN_INSN(insn, "unknown unwind hint type %d", cfi->type);
- return -1;
- }
-
- orc->signal = cfi->signal;
-
- switch (cfi->cfa.base) {
- case CFI_SP:
- orc->sp_reg = ORC_REG_SP;
- break;
- case CFI_SP_INDIRECT:
- orc->sp_reg = ORC_REG_SP_INDIRECT;
- break;
- case CFI_BP:
- orc->sp_reg = ORC_REG_BP;
- break;
- case CFI_BP_INDIRECT:
- orc->sp_reg = ORC_REG_BP_INDIRECT;
- break;
- case CFI_R10:
- orc->sp_reg = ORC_REG_R10;
- break;
- case CFI_R13:
- orc->sp_reg = ORC_REG_R13;
- break;
- case CFI_DI:
- orc->sp_reg = ORC_REG_DI;
- break;
- case CFI_DX:
- orc->sp_reg = ORC_REG_DX;
- break;
- default:
- WARN_INSN(insn, "unknown CFA base reg %d", cfi->cfa.base);
- return -1;
- }
-
- switch (bp->base) {
- case CFI_UNDEFINED:
- orc->bp_reg = ORC_REG_UNDEFINED;
- break;
- case CFI_CFA:
- orc->bp_reg = ORC_REG_PREV_SP;
- break;
- case CFI_BP:
- orc->bp_reg = ORC_REG_BP;
- break;
- default:
- WARN_INSN(insn, "unknown BP base reg %d", bp->base);
- return -1;
- }
-
- orc->sp_offset = cfi->cfa.offset;
- orc->bp_offset = bp->offset;
-
- return 0;
-}
-
static int write_orc_entry(struct elf *elf, struct section *orc_sec,
struct section *ip_sec, unsigned int idx,
struct section *insn_sec, unsigned long insn_off,
--
2.1.0
^ permalink raw reply [flat|nested] 18+ messages in thread* [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built
2023-07-25 8:15 [PATCH v1 0/6] Add objtool and orc support for LoongArch Tiezhu Yang
` (2 preceding siblings ...)
2023-07-25 8:15 ` [PATCH v1 3/6] objtool/x86: Separate arch-specific and generic parts Tiezhu Yang
@ 2023-07-25 8:15 ` Tiezhu Yang
2023-07-25 11:51 ` Peter Zijlstra
2023-07-25 11:56 ` Peter Zijlstra
2023-07-25 8:15 ` [PATCH v1 5/6] objtool: Add skipped member in struct reloc Tiezhu Yang
` (2 subsequent siblings)
6 siblings, 2 replies; 18+ messages in thread
From: Tiezhu Yang @ 2023-07-25 8:15 UTC (permalink / raw)
To: Josh Poimboeuf, Peter Zijlstra, Huacai Chen
Cc: loongarch, linux-kernel, loongson-kernel
Implement init_orc_entry(), reg_name(), orc_type_name(), print_reg(),
orc_print_dump(), arch_write_orc() and update_cfi_state(), and then
set BUILD_ORC as y to build the orc related files.
Co-developed-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Co-developed-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
tools/arch/loongarch/include/asm/orc_types.h | 58 ++++++++++
tools/objtool/Makefile | 4 +
tools/objtool/arch/loongarch/Build | 1 +
tools/objtool/arch/loongarch/decode.c | 94 ++++++++++++++++
tools/objtool/arch/loongarch/orc.c | 158 +++++++++++++++++++++++++++
tools/objtool/include/objtool/orc.h | 1 +
tools/objtool/orc_gen.c | 3 +
7 files changed, 319 insertions(+)
create mode 100644 tools/arch/loongarch/include/asm/orc_types.h
create mode 100644 tools/objtool/arch/loongarch/orc.c
diff --git a/tools/arch/loongarch/include/asm/orc_types.h b/tools/arch/loongarch/include/asm/orc_types.h
new file mode 100644
index 0000000..1d37e62
--- /dev/null
+++ b/tools/arch/loongarch/include/asm/orc_types.h
@@ -0,0 +1,58 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef _ORC_TYPES_H
+#define _ORC_TYPES_H
+
+#include <linux/types.h>
+
+/*
+ * The ORC_REG_* registers are base registers which are used to find other
+ * registers on the stack.
+ *
+ * ORC_REG_PREV_SP, also known as DWARF Call Frame Address (CFA), is the
+ * address of the previous frame: the caller's SP before it called the current
+ * function.
+ *
+ * ORC_REG_UNDEFINED means the corresponding register's value didn't change in
+ * the current frame.
+ *
+ * The most commonly used base registers are SP and BP -- which the previous SP
+ * is usually based on -- and PREV_SP and UNDEFINED -- which the previous BP is
+ * usually based on.
+ *
+ * The rest of the base registers are needed for special cases like entry code
+ * and GCC realigned stacks.
+ */
+#define ORC_REG_UNDEFINED 0
+#define ORC_REG_PREV_SP 1
+#define ORC_REG_SP 2
+#define ORC_REG_BP 3
+#define ORC_REG_MAX 4
+
+#define ORC_TYPE_UNDEFINED 0
+#define ORC_TYPE_END_OF_STACK 1
+#define ORC_TYPE_CALL 2
+#define ORC_TYPE_REGS 3
+#define ORC_TYPE_REGS_PARTIAL 4
+
+#ifndef __ASSEMBLY__
+/*
+ * This struct is more or less a vastly simplified version of the DWARF Call
+ * Frame Information standard. It contains only the necessary parts of DWARF
+ * CFI, simplified for ease of access by the in-kernel unwinder. It tells the
+ * unwinder how to find the previous SP and BP (and sometimes entry regs) on
+ * the stack for a given code address. Each instance of the struct corresponds
+ * to one or more code locations.
+ */
+struct orc_entry {
+ s16 sp_offset;
+ s16 bp_offset;
+ s16 ra_offset;
+ unsigned int sp_reg:4;
+ unsigned int bp_reg:4;
+ unsigned int ra_reg:4;
+ unsigned int type:3;
+ unsigned int signal:1;
+};
+#endif /* __ASSEMBLY__ */
+
+#endif /* _ORC_TYPES_H */
diff --git a/tools/objtool/Makefile b/tools/objtool/Makefile
index 83b100c..bf7f7f8 100644
--- a/tools/objtool/Makefile
+++ b/tools/objtool/Makefile
@@ -57,6 +57,10 @@ ifeq ($(SRCARCH),x86)
BUILD_ORC := y
endif
+ifeq ($(SRCARCH),loongarch)
+ BUILD_ORC := y
+endif
+
export BUILD_ORC
export srctree OUTPUT CFLAGS SRCARCH AWK
include $(srctree)/tools/build/Makefile.include
diff --git a/tools/objtool/arch/loongarch/Build b/tools/objtool/arch/loongarch/Build
index d24d563..1d4b784 100644
--- a/tools/objtool/arch/loongarch/Build
+++ b/tools/objtool/arch/loongarch/Build
@@ -1,2 +1,3 @@
objtool-y += decode.o
objtool-y += special.o
+objtool-y += orc.o
diff --git a/tools/objtool/arch/loongarch/decode.c b/tools/objtool/arch/loongarch/decode.c
index 2904fbc..8489509 100644
--- a/tools/objtool/arch/loongarch/decode.c
+++ b/tools/objtool/arch/loongarch/decode.c
@@ -3,6 +3,8 @@
#include <objtool/check.h>
#include <objtool/warn.h>
#include <asm/inst.h>
+#include <asm/orc_types.h>
+#include <linux/objtool_types.h>
int arch_ftrace_match(char *name)
{
@@ -38,6 +40,20 @@ bool arch_callee_saved_reg(unsigned char reg)
int arch_decode_hint_reg(u8 sp_reg, int *base)
{
+ switch (sp_reg) {
+ case ORC_REG_UNDEFINED:
+ *base = CFI_UNDEFINED;
+ break;
+ case ORC_REG_SP:
+ *base = CFI_SP;
+ break;
+ case ORC_REG_BP:
+ *base = CFI_FP;
+ break;
+ default:
+ return -1;
+ }
+
return 0;
}
@@ -310,3 +326,81 @@ void arch_initial_func_cfi_state(struct cfi_init_state *state)
state->cfa.base = CFI_SP;
state->cfa.offset = 0;
}
+
+int update_cfi_state(struct instruction *insn,
+ struct instruction *next_insn,
+ struct cfi_state *cfi, struct stack_op *op)
+{
+ struct cfi_reg *cfa = &cfi->cfa;
+ struct cfi_reg *regs = cfi->regs;
+
+ /* ignore UNWIND_HINT_UNDEFINED regions */
+ if (cfi->force_undefined)
+ return 0;
+
+ /* stack operations don't make sense with an undefined CFA */
+ if (cfa->base == CFI_UNDEFINED) {
+ if (insn_func(insn)) {
+ WARN_INSN(insn, "undefined stack state");
+ return -1;
+ }
+ return 0;
+ }
+
+ if (cfi->type == UNWIND_HINT_TYPE_REGS ||
+ cfi->type == UNWIND_HINT_TYPE_REGS_PARTIAL)
+ return update_cfi_state_regs(insn, cfi, op);
+
+ switch (op->dest.type) {
+ case OP_DEST_REG:
+ switch (op->src.type) {
+ case OP_SRC_ADD:
+ if (op->dest.reg == CFI_SP && op->src.reg == CFI_SP) {
+ /* addi.d sp,sp,si12 */
+ cfi->stack_size -= op->src.offset;
+ if (cfa->base == CFI_SP)
+ cfa->offset -= op->src.offset;
+ } else if (op->dest.reg == CFI_FP && op->src.reg == CFI_SP) {
+ /* addi.d fp,sp,si12 */
+ if (cfa->base == CFI_SP && cfa->offset == op->src.offset) {
+ cfa->base = CFI_FP;
+ cfa->offset = 0;
+ }
+ } else if (op->dest.reg == CFI_SP && op->src.reg == CFI_FP) {
+ /* addi.d sp,fp,si12 */
+ if (cfa->base == CFI_FP && cfa->offset == 0) {
+ cfa->base = CFI_SP;
+ cfa->offset = -op->src.offset;
+ }
+ }
+ break;
+ case OP_SRC_REG_INDIRECT:
+ /* ld.d rd,sp,si12 */
+ if (op->src.reg == CFI_SP &&
+ op->src.offset == (regs[op->dest.reg].offset + cfi->stack_size)) {
+ restore_reg(cfi, op->dest.reg);
+ /* GCC may not restore sp, we adjust it directly. */
+ if (cfa->base == CFI_FP && cfa->offset == 0) {
+ cfa->base = CFI_SP;
+ cfa->offset = cfi->stack_size;
+ }
+ }
+ break;
+ default:
+ break;
+ }
+ break;
+ case OP_DEST_REG_INDIRECT:
+ if (op->src.type == OP_SRC_REG)
+ /* st.d rd,sp,si12 */
+ if (op->dest.offset)
+ save_reg(cfi, op->src.reg, CFI_CFA,
+ op->dest.offset - cfi->stack_size);
+ break;
+ default:
+ WARN_FUNC("unknown stack-related instruction", insn->sec, insn->offset);
+ return -1;
+ }
+
+ return 0;
+}
diff --git a/tools/objtool/arch/loongarch/orc.c b/tools/objtool/arch/loongarch/orc.c
new file mode 100644
index 0000000..f59ab10
--- /dev/null
+++ b/tools/objtool/arch/loongarch/orc.c
@@ -0,0 +1,158 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <linux/objtool_types.h>
+#include <asm/orc_types.h>
+
+#include <objtool/check.h>
+#include <objtool/orc.h>
+#include <objtool/warn.h>
+#include <objtool/endianness.h>
+
+int init_orc_entry(struct orc_entry *orc, struct cfi_state *cfi, struct instruction *insn)
+{
+ struct cfi_reg *bp = &cfi->regs[CFI_BP];
+ struct cfi_reg *ra = &cfi->regs[CFI_RA];
+
+ memset(orc, 0, sizeof(*orc));
+
+ if (!cfi) {
+ /*
+ * This is usually either unreachable nops/traps (which don't
+ * trigger unreachable instruction warnings), or
+ * STACK_FRAME_NON_STANDARD functions.
+ */
+ orc->type = ORC_TYPE_UNDEFINED;
+ return 0;
+ }
+
+ switch (cfi->type) {
+ case UNWIND_HINT_TYPE_UNDEFINED:
+ orc->type = ORC_TYPE_UNDEFINED;
+ return 0;
+ case UNWIND_HINT_TYPE_END_OF_STACK:
+ orc->type = ORC_TYPE_END_OF_STACK;
+ return 0;
+ case UNWIND_HINT_TYPE_CALL:
+ orc->type = ORC_TYPE_CALL;
+ break;
+ case UNWIND_HINT_TYPE_REGS:
+ orc->type = ORC_TYPE_REGS;
+ break;
+ case UNWIND_HINT_TYPE_REGS_PARTIAL:
+ orc->type = ORC_TYPE_REGS_PARTIAL;
+ break;
+ default:
+ WARN_INSN(insn, "unknown unwind hint type %d", cfi->type);
+ return -1;
+ }
+
+ orc->signal = cfi->signal;
+
+ switch (cfi->cfa.base) {
+ case CFI_SP:
+ orc->sp_reg = ORC_REG_SP;
+ break;
+ case CFI_BP:
+ orc->sp_reg = ORC_REG_BP;
+ break;
+ default:
+ WARN_INSN(insn, "unknown CFA base reg %d", cfi->cfa.base);
+ return -1;
+ }
+
+ switch (bp->base) {
+ case CFI_UNDEFINED:
+ orc->bp_reg = ORC_REG_UNDEFINED;
+ break;
+ case CFI_CFA:
+ orc->bp_reg = ORC_REG_PREV_SP;
+ break;
+ case CFI_BP:
+ orc->bp_reg = ORC_REG_BP;
+ break;
+ default:
+ WARN_INSN(insn, "unknown BP base reg %d", bp->base);
+ return -1;
+ }
+
+ switch (ra->base) {
+ case CFI_UNDEFINED:
+ orc->ra_reg = ORC_REG_UNDEFINED;
+ break;
+ case CFI_CFA:
+ orc->ra_reg = ORC_REG_PREV_SP;
+ break;
+ case CFI_BP:
+ orc->ra_reg = ORC_REG_BP;
+ break;
+ default:
+ WARN_INSN(insn, "unknown RA base reg %d", ra->base);
+ return -1;
+ }
+
+ orc->sp_offset = cfi->cfa.offset;
+ orc->bp_offset = bp->offset;
+ orc->ra_offset = ra->offset;
+
+ return 0;
+}
+
+static const char *reg_name(unsigned int reg)
+{
+ switch (reg) {
+ case ORC_REG_SP:
+ return "sp";
+ case ORC_REG_BP:
+ return "fp";
+ case ORC_REG_PREV_SP:
+ return "prevsp";
+ default:
+ return "?";
+ }
+}
+
+static const char *orc_type_name(unsigned int type)
+{
+ switch (type) {
+ case UNWIND_HINT_TYPE_CALL:
+ return "call";
+ case UNWIND_HINT_TYPE_REGS:
+ return "regs";
+ case UNWIND_HINT_TYPE_REGS_PARTIAL:
+ return "regs (partial)";
+ default:
+ return "?";
+ }
+}
+
+static void print_reg(unsigned int reg, int offset)
+{
+ if (reg == ORC_REG_UNDEFINED)
+ printf(" (und) ");
+ else
+ printf("%s + %3d", reg_name(reg), offset);
+
+}
+
+void orc_print_dump(struct elf *dummy_elf, struct orc_entry *orc, int i)
+{
+ printf("type:%s", orc_type_name(orc[i].type));
+
+ printf(" sp:");
+
+ print_reg(orc[i].sp_reg, bswap_if_needed(dummy_elf, orc[i].sp_offset));
+
+ printf(" bp:");
+
+ print_reg(orc[i].bp_reg, bswap_if_needed(dummy_elf, orc[i].bp_offset));
+
+ printf(" ra:");
+
+ print_reg(orc[i].ra_reg, bswap_if_needed(dummy_elf, orc[i].ra_offset));
+
+ printf(" signal:%d\n", orc[i].signal);
+}
+
+void arch_write_orc(struct elf *elf, struct orc_entry *orc)
+{
+ orc->ra_offset = bswap_if_needed(elf, orc->ra_offset);
+}
diff --git a/tools/objtool/include/objtool/orc.h b/tools/objtool/include/objtool/orc.h
index 4c9f946..85ce2bf 100644
--- a/tools/objtool/include/objtool/orc.h
+++ b/tools/objtool/include/objtool/orc.h
@@ -6,5 +6,6 @@
int init_orc_entry(struct orc_entry *orc, struct cfi_state *cfi, struct instruction *insn);
void orc_print_dump(struct elf *dummy_elf, struct orc_entry *orc, int i);
+void arch_write_orc(struct elf *elf, struct orc_entry *orc);
#endif /* _OBJTOOL_ORC_H */
diff --git a/tools/objtool/orc_gen.c b/tools/objtool/orc_gen.c
index 1eff7e0a..6975056 100644
--- a/tools/objtool/orc_gen.c
+++ b/tools/objtool/orc_gen.c
@@ -14,6 +14,8 @@
#include <objtool/warn.h>
#include <objtool/endianness.h>
+void __weak arch_write_orc(struct elf *elf, struct orc_entry *orc) {}
+
static int write_orc_entry(struct elf *elf, struct section *orc_sec,
struct section *ip_sec, unsigned int idx,
struct section *insn_sec, unsigned long insn_off,
@@ -26,6 +28,7 @@ static int write_orc_entry(struct elf *elf, struct section *orc_sec,
memcpy(orc, o, sizeof(*orc));
orc->sp_offset = bswap_if_needed(elf, orc->sp_offset);
orc->bp_offset = bswap_if_needed(elf, orc->bp_offset);
+ arch_write_orc(elf, orc);
/* populate reloc for ip */
if (!elf_init_reloc_text_sym(elf, ip_sec, idx * sizeof(int), idx,
--
2.1.0
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built
2023-07-25 8:15 ` [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built Tiezhu Yang
@ 2023-07-25 11:51 ` Peter Zijlstra
2023-07-27 10:58 ` Tiezhu Yang
2023-07-25 11:56 ` Peter Zijlstra
1 sibling, 1 reply; 18+ messages in thread
From: Peter Zijlstra @ 2023-07-25 11:51 UTC (permalink / raw)
To: Tiezhu Yang
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On Tue, Jul 25, 2023 at 04:15:08PM +0800, Tiezhu Yang wrote:
Previous Changelog had:
> Define update_cfi_state() as a weak function which may be overwritten
> by the arch-specific implementation.
And then this patch does:
> +int update_cfi_state(struct instruction *insn,
> + struct instruction *next_insn,
> + struct cfi_state *cfi, struct stack_op *op)
> +{
> + struct cfi_reg *cfa = &cfi->cfa;
> + struct cfi_reg *regs = cfi->regs;
> +
> + /* ignore UNWIND_HINT_UNDEFINED regions */
> + if (cfi->force_undefined)
> + return 0;
> +
> + /* stack operations don't make sense with an undefined CFA */
> + if (cfa->base == CFI_UNDEFINED) {
> + if (insn_func(insn)) {
> + WARN_INSN(insn, "undefined stack state");
> + return -1;
> + }
> + return 0;
> + }
> +
> + if (cfi->type == UNWIND_HINT_TYPE_REGS ||
> + cfi->type == UNWIND_HINT_TYPE_REGS_PARTIAL)
> + return update_cfi_state_regs(insn, cfi, op);
> +
> + switch (op->dest.type) {
> + case OP_DEST_REG:
> + switch (op->src.type) {
> + case OP_SRC_ADD:
> + if (op->dest.reg == CFI_SP && op->src.reg == CFI_SP) {
> + /* addi.d sp,sp,si12 */
> + cfi->stack_size -= op->src.offset;
> + if (cfa->base == CFI_SP)
> + cfa->offset -= op->src.offset;
> + } else if (op->dest.reg == CFI_FP && op->src.reg == CFI_SP) {
> + /* addi.d fp,sp,si12 */
> + if (cfa->base == CFI_SP && cfa->offset == op->src.offset) {
> + cfa->base = CFI_FP;
> + cfa->offset = 0;
> + }
> + } else if (op->dest.reg == CFI_SP && op->src.reg == CFI_FP) {
> + /* addi.d sp,fp,si12 */
> + if (cfa->base == CFI_FP && cfa->offset == 0) {
> + cfa->base = CFI_SP;
> + cfa->offset = -op->src.offset;
> + }
> + }
> + break;
> + case OP_SRC_REG_INDIRECT:
> + /* ld.d rd,sp,si12 */
> + if (op->src.reg == CFI_SP &&
> + op->src.offset == (regs[op->dest.reg].offset + cfi->stack_size)) {
> + restore_reg(cfi, op->dest.reg);
> + /* GCC may not restore sp, we adjust it directly. */
> + if (cfa->base == CFI_FP && cfa->offset == 0) {
> + cfa->base = CFI_SP;
> + cfa->offset = cfi->stack_size;
> + }
> + }
> + break;
> + default:
> + break;
> + }
> + break;
> + case OP_DEST_REG_INDIRECT:
> + if (op->src.type == OP_SRC_REG)
> + /* st.d rd,sp,si12 */
> + if (op->dest.offset)
> + save_reg(cfi, op->src.reg, CFI_CFA,
> + op->dest.offset - cfi->stack_size);
> + break;
> + default:
> + WARN_FUNC("unknown stack-related instruction", insn->sec, insn->offset);
> + return -1;
> + }
> +
> + return 0;
> +}
Why ?!? what is the actual irreconcilable difference?
If you want us to review this, you'll have to explain things.
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built
2023-07-25 11:51 ` Peter Zijlstra
@ 2023-07-27 10:58 ` Tiezhu Yang
0 siblings, 0 replies; 18+ messages in thread
From: Tiezhu Yang @ 2023-07-27 10:58 UTC (permalink / raw)
To: Peter Zijlstra
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On 07/25/2023 07:51 PM, Peter Zijlstra wrote:
> On Tue, Jul 25, 2023 at 04:15:08PM +0800, Tiezhu Yang wrote:
>
> Previous Changelog had:
>
>> Define update_cfi_state() as a weak function which may be overwritten
>> by the arch-specific implementation.
>
> And then this patch does:
>
>> +int update_cfi_state(struct instruction *insn,
>> + struct instruction *next_insn,
>> + struct cfi_state *cfi, struct stack_op *op)
>> +{
>> + struct cfi_reg *cfa = &cfi->cfa;
>> + struct cfi_reg *regs = cfi->regs;
...
>> + default:
>> + WARN_FUNC("unknown stack-related instruction", insn->sec, insn->offset);
>> + return -1;
>> + }
>> +
>> + return 0;
>> +}
>
> Why ?!? what is the actual irreconcilable difference?
I went in deeper to analysis the code and did more test,
the conclusion is that there is no need to define the
arch-specific update_cfi_state(), although it seems simpler
than the generic update_cfi_state(), the generic function
can handle all the cases and works well.
> If you want us to review this, you'll have to explain things.
I will drop the related changes in the next version, thank you.
Thanks,
Tiezhu
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built
2023-07-25 8:15 ` [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built Tiezhu Yang
2023-07-25 11:51 ` Peter Zijlstra
@ 2023-07-25 11:56 ` Peter Zijlstra
2023-07-28 1:31 ` Tiezhu Yang
1 sibling, 1 reply; 18+ messages in thread
From: Peter Zijlstra @ 2023-07-25 11:56 UTC (permalink / raw)
To: Tiezhu Yang
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On Tue, Jul 25, 2023 at 04:15:08PM +0800, Tiezhu Yang wrote:
> +void arch_write_orc(struct elf *elf, struct orc_entry *orc)
> +{
> + orc->ra_offset = bswap_if_needed(elf, orc->ra_offset);
> +}
This hardly 'writes' anything, all it does is a bswap.
> diff --git a/tools/objtool/orc_gen.c b/tools/objtool/orc_gen.c
> index 1eff7e0a..6975056 100644
> --- a/tools/objtool/orc_gen.c
> +++ b/tools/objtool/orc_gen.c
> @@ -14,6 +14,8 @@
> #include <objtool/warn.h>
> #include <objtool/endianness.h>
>
> +void __weak arch_write_orc(struct elf *elf, struct orc_entry *orc) {}
And for that you need a weak function? What's wrong with an inline if
anything?
> static int write_orc_entry(struct elf *elf, struct section *orc_sec,
> struct section *ip_sec, unsigned int idx,
> struct section *insn_sec, unsigned long insn_off,
> @@ -26,6 +28,7 @@ static int write_orc_entry(struct elf *elf, struct section *orc_sec,
> memcpy(orc, o, sizeof(*orc));
> orc->sp_offset = bswap_if_needed(elf, orc->sp_offset);
> orc->bp_offset = bswap_if_needed(elf, orc->bp_offset);
> + arch_write_orc(elf, orc);
Why can't this simply be:
orc->ra_offset = bswap_if_needed(elf, orc->ra_offset);
AFAICT this won't actually do anything for x86 and I don't think Power
got around to doing ORC yet.
>
> /* populate reloc for ip */
> if (!elf_init_reloc_text_sym(elf, ip_sec, idx * sizeof(int), idx,
Again, not much explaination for why you did things
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built
2023-07-25 11:56 ` Peter Zijlstra
@ 2023-07-28 1:31 ` Tiezhu Yang
0 siblings, 0 replies; 18+ messages in thread
From: Tiezhu Yang @ 2023-07-28 1:31 UTC (permalink / raw)
To: Peter Zijlstra
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On 07/25/2023 07:56 PM, Peter Zijlstra wrote:
> On Tue, Jul 25, 2023 at 04:15:08PM +0800, Tiezhu Yang wrote:
>
>> +void arch_write_orc(struct elf *elf, struct orc_entry *orc)
>> +{
>> + orc->ra_offset = bswap_if_needed(elf, orc->ra_offset);
>> +}
>
> This hardly 'writes' anything, all it does is a bswap.
>
>> diff --git a/tools/objtool/orc_gen.c b/tools/objtool/orc_gen.c
>> index 1eff7e0a..6975056 100644
>> --- a/tools/objtool/orc_gen.c
>> +++ b/tools/objtool/orc_gen.c
>> @@ -14,6 +14,8 @@
>> #include <objtool/warn.h>
>> #include <objtool/endianness.h>
>>
>> +void __weak arch_write_orc(struct elf *elf, struct orc_entry *orc) {}
>
> And for that you need a weak function? What's wrong with an inline if
> anything?
>
>> static int write_orc_entry(struct elf *elf, struct section *orc_sec,
>> struct section *ip_sec, unsigned int idx,
>> struct section *insn_sec, unsigned long insn_off,
>> @@ -26,6 +28,7 @@ static int write_orc_entry(struct elf *elf, struct section *orc_sec,
>> memcpy(orc, o, sizeof(*orc));
>> orc->sp_offset = bswap_if_needed(elf, orc->sp_offset);
>> orc->bp_offset = bswap_if_needed(elf, orc->bp_offset);
>> + arch_write_orc(elf, orc);
>
> Why can't this simply be:
>
> orc->ra_offset = bswap_if_needed(elf, orc->ra_offset);
>
> AFAICT this won't actually do anything for x86 and I don't think Power
> got around to doing ORC yet.
>
>>
>> /* populate reloc for ip */
>> if (!elf_init_reloc_text_sym(elf, ip_sec, idx * sizeof(int), idx,
>
> Again, not much explaination for why you did things
>
The initial aim is to keep consistent with the above two lines,
but for now, I think there is no need to call bswap_if_needed()
for LoongArch, the related changes are not necessary, I will
drop them too.
LoongArch is little-endian, the __BYTE_ORDER is defined as
__LITTLE_ENDIAN, and elf->ehdr.e_ident[EI_DATA] is 1 which
is equal with ELFDATA2LSB, so need_bswap() returns false.
static inline bool need_bswap(struct elf *elf)
{
return (__BYTE_ORDER == __LITTLE_ENDIAN) ^
(elf->ehdr.e_ident[EI_DATA] == ELFDATA2LSB);
}
Thank you very much for your review, sorry for that.
Thanks,
Tiezhu
^ permalink raw reply [flat|nested] 18+ messages in thread
* [PATCH v1 5/6] objtool: Add skipped member in struct reloc
2023-07-25 8:15 [PATCH v1 0/6] Add objtool and orc support for LoongArch Tiezhu Yang
` (3 preceding siblings ...)
2023-07-25 8:15 ` [PATCH v1 4/6] objtool/LoongArch: Enable orc to be built Tiezhu Yang
@ 2023-07-25 8:15 ` Tiezhu Yang
2023-07-25 11:59 ` Peter Zijlstra
2023-07-25 8:15 ` [PATCH v1 6/6] LoongArch: Add ORC unwinder support Tiezhu Yang
2023-07-25 12:01 ` [PATCH v1 0/6] Add objtool and orc support for LoongArch Peter Zijlstra
6 siblings, 1 reply; 18+ messages in thread
From: Tiezhu Yang @ 2023-07-25 8:15 UTC (permalink / raw)
To: Josh Poimboeuf, Peter Zijlstra, Huacai Chen
Cc: loongarch, linux-kernel, loongson-kernel
There exist multiple relocation types in one location, such as a pair of
R_LARCH_ADD32 and R_LARCH_SUB32 in section .rela.discard.unreachable and
.rela.discard.reachable on LoongArch.
Here is an example:
$ readelf -rW init/main.o
Relocation section '.rela.discard.unreachable' at offset 0x3e20 contains 2 entries:
Offset Info Type Symbol's Value Symbol's Name + Addend
0000000000000000 0000000a00000032 R_LARCH_ADD32 0000000000000000 .init.text + 230
0000000000000000 0000001a00000037 R_LARCH_SUB32 0000000000000000 L0^A + 0
Relocation section '.rela.discard.reachable' at offset 0x4a00 contains 6 entries:
Offset Info Type Symbol's Value Symbol's Name + Addend
0000000000000000 0000000a00000032 R_LARCH_ADD32 0000000000000000 .init.text + ae0
0000000000000000 0000002800000037 R_LARCH_SUB32 0000000000000000 L0^A + 0
0000000000000004 0000000a00000032 R_LARCH_ADD32 0000000000000000 .init.text + b88
0000000000000004 0000002b00000037 R_LARCH_SUB32 0000000000000004 L0^A + 0
0000000000000008 0000000200000032 R_LARCH_ADD32 0000000000000000 .text + 23c
0000000000000008 0000002e00000037 R_LARCH_SUB32 0000000000000008 L0^A + 0
In order to silence the related objtool warnings, add skipped member
in struct reloc to record and skip the latter relocation.
Co-developed-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Co-developed-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
tools/objtool/check.c | 4 ++++
tools/objtool/elf.c | 6 ++++++
tools/objtool/include/objtool/elf.h | 1 +
3 files changed, 11 insertions(+)
diff --git a/tools/objtool/check.c b/tools/objtool/check.c
index 9e5e462..602318e 100644
--- a/tools/objtool/check.c
+++ b/tools/objtool/check.c
@@ -594,6 +594,8 @@ static int add_dead_ends(struct objtool_file *file)
goto reachable;
for_each_reloc(rsec, reloc) {
+ if (reloc->skipped)
+ continue;
if (reloc->sym->type != STT_SECTION) {
WARN("unexpected relocation symbol type in %s", rsec->name);
@@ -633,6 +635,8 @@ static int add_dead_ends(struct objtool_file *file)
return 0;
for_each_reloc(rsec, reloc) {
+ if (reloc->skipped)
+ continue;
if (reloc->sym->type != STT_SECTION) {
WARN("unexpected relocation symbol type in %s", rsec->name);
diff --git a/tools/objtool/elf.c b/tools/objtool/elf.c
index d420b5d..bd950d4 100644
--- a/tools/objtool/elf.c
+++ b/tools/objtool/elf.c
@@ -975,6 +975,12 @@ static int read_relocs(struct elf *elf)
return -1;
}
+ if (!(strcmp(rsec->name, ".rela.discard.unreachable")) ||
+ !(strcmp(rsec->name, ".rela.discard.reachable"))) {
+ if (reloc->sym->type != STT_SECTION)
+ reloc->skipped = true;
+ }
+
elf_hash_add(reloc, &reloc->hash, reloc_hash(reloc));
reloc->sym_next_reloc = sym->relocs;
sym->relocs = reloc;
diff --git a/tools/objtool/include/objtool/elf.h b/tools/objtool/include/objtool/elf.h
index c532d70..4141837 100644
--- a/tools/objtool/include/objtool/elf.h
+++ b/tools/objtool/include/objtool/elf.h
@@ -75,6 +75,7 @@ struct reloc {
struct section *sec;
struct symbol *sym;
struct reloc *sym_next_reloc;
+ bool skipped;
};
struct elf {
--
2.1.0
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH v1 5/6] objtool: Add skipped member in struct reloc
2023-07-25 8:15 ` [PATCH v1 5/6] objtool: Add skipped member in struct reloc Tiezhu Yang
@ 2023-07-25 11:59 ` Peter Zijlstra
2023-08-03 11:36 ` Tiezhu Yang
0 siblings, 1 reply; 18+ messages in thread
From: Peter Zijlstra @ 2023-07-25 11:59 UTC (permalink / raw)
To: Tiezhu Yang
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On Tue, Jul 25, 2023 at 04:15:09PM +0800, Tiezhu Yang wrote:
> There exist multiple relocation types in one location, such as a pair of
> R_LARCH_ADD32 and R_LARCH_SUB32 in section .rela.discard.unreachable and
> .rela.discard.reachable on LoongArch.
>
> Here is an example:
>
> $ readelf -rW init/main.o
>
> Relocation section '.rela.discard.unreachable' at offset 0x3e20 contains 2 entries:
> Offset Info Type Symbol's Value Symbol's Name + Addend
> 0000000000000000 0000000a00000032 R_LARCH_ADD32 0000000000000000 .init.text + 230
> 0000000000000000 0000001a00000037 R_LARCH_SUB32 0000000000000000 L0^A + 0
>
Please explain; why is this?
How does:
#define __annotate_unreachable(c) ({ \
asm volatile(__stringify_label(c) ":\n\t" \
".pushsection .discard.unreachable\n\t" \
".long " __stringify_label(c) "b - .\n\t" \
".popsection\n\t" : : "i" (c)); \
})
#define annotate_unreachable() __annotate_unreachable(__COUNTER__)
Manage to generate this..
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH v1 5/6] objtool: Add skipped member in struct reloc
2023-07-25 11:59 ` Peter Zijlstra
@ 2023-08-03 11:36 ` Tiezhu Yang
2023-08-03 12:11 ` Peter Zijlstra
0 siblings, 1 reply; 18+ messages in thread
From: Tiezhu Yang @ 2023-08-03 11:36 UTC (permalink / raw)
To: Peter Zijlstra
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On 07/25/2023 07:59 PM, Peter Zijlstra wrote:
> On Tue, Jul 25, 2023 at 04:15:09PM +0800, Tiezhu Yang wrote:
>> There exist multiple relocation types in one location, such as a pair of
>> R_LARCH_ADD32 and R_LARCH_SUB32 in section .rela.discard.unreachable and
>> .rela.discard.reachable on LoongArch.
>>
>> Here is an example:
>>
>> $ readelf -rW init/main.o
>>
>> Relocation section '.rela.discard.unreachable' at offset 0x3e20 contains 2 entries:
>> Offset Info Type Symbol's Value Symbol's Name + Addend
>> 0000000000000000 0000000a00000032 R_LARCH_ADD32 0000000000000000 .init.text + 230
>> 0000000000000000 0000001a00000037 R_LARCH_SUB32 0000000000000000 L0^A + 0
>>
>
> Please explain; why is this?
>
> How does:
>
> #define __annotate_unreachable(c) ({ \
> asm volatile(__stringify_label(c) ":\n\t" \
> ".pushsection .discard.unreachable\n\t" \
> ".long " __stringify_label(c) "b - .\n\t" \
> ".popsection\n\t" : : "i" (c)); \
> })
> #define annotate_unreachable() __annotate_unreachable(__COUNTER__)
>
> Manage to generate this..
>
Sorry for the late reply, let me try to explain it.
R_LARCH_ADD32 relocation type is 32-bit label addition,
R_LARCH_SUB32 relocation type is 32-bit label subtraction,
they are intended for local labels, the label difference
will be calculated as a constant before linking, preserve
a pair of R_LARCH_ADD32 and R_LARCH_SUB32 to fix the label
difference.
Here is a simple example:
-------------------------------
.L0
// do something
.L1
.data
.word .L0 - .L1
-------------------------------
NewDiff = 0
R_LARCH_ADD32:
NewDiff = NewDiff + .L0_Addr
R_LARCH_SUB32:
NewDiff = NewDiff - .L1_Addr
I discussed offline with the engineers who are familiar with gnu
assembler of LoongArch, maybe it can modify the gnu assembler
to use R_LARCH_32_PCREL to replace a pair of R_LARCH_ADD32 and
R_LARCH_SUB32, then I will test it again and drop this change if
possible.
Here is LoongArch ELF Relocations spec:
https://github.com/loongson/la-abi-specs/blob/release/laelf.adoc
R_LARCH_ADD32
32-bit in-place addition
*(int32_t *) PC += S + A
R_LARCH_SUB32
32-bit in-place subtraction
*(int32_t *) PC -= S + A
R_LARCH_32_PCREL
32-bit PC relative
(*(uint32_t *) PC) = (S+A-PC) [31 ... 0]
Thanks very much for your review comments, it is helpful to
make progress in the right direction.
Thanks,
Tiezhu
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH v1 5/6] objtool: Add skipped member in struct reloc
2023-08-03 11:36 ` Tiezhu Yang
@ 2023-08-03 12:11 ` Peter Zijlstra
2023-08-09 11:51 ` Xi Ruoyao
0 siblings, 1 reply; 18+ messages in thread
From: Peter Zijlstra @ 2023-08-03 12:11 UTC (permalink / raw)
To: Tiezhu Yang
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On Thu, Aug 03, 2023 at 07:36:24PM +0800, Tiezhu Yang wrote:
> > ".long " __stringify_label(c) "b - .\n\t" \
> I discussed offline with the engineers who are familiar with gnu
> assembler of LoongArch, maybe it can modify the gnu assembler
> to use R_LARCH_32_PCREL to replace a pair of R_LARCH_ADD32 and
> R_LARCH_SUB32, then I will test it again and drop this change if
> possible.
Yes, this is exactly what PC relative relocations are for.
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH v1 5/6] objtool: Add skipped member in struct reloc
2023-08-03 12:11 ` Peter Zijlstra
@ 2023-08-09 11:51 ` Xi Ruoyao
2023-08-09 11:58 ` Xi Ruoyao
0 siblings, 1 reply; 18+ messages in thread
From: Xi Ruoyao @ 2023-08-09 11:51 UTC (permalink / raw)
To: Peter Zijlstra, Tiezhu Yang
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On Thu, 2023-08-03 at 14:11 +0200, Peter Zijlstra wrote:
> On Thu, Aug 03, 2023 at 07:36:24PM +0800, Tiezhu Yang wrote:
>
> > > ".long " __stringify_label(c) "b - .\n\t" \
>
> > I discussed offline with the engineers who are familiar with gnu
> > assembler of LoongArch, maybe it can modify the gnu assembler
> > to use R_LARCH_32_PCREL to replace a pair of R_LARCH_ADD32 and
> > R_LARCH_SUB32, then I will test it again and drop this change if
> > possible.
Hmm, but I don't like the idea to disallow users having GAS <= 2.41 from
using ORC unwinder.
Can we emit the R_LARCH_32_PCREL reloc with an explicit .reloc
directive?
> Yes, this is exactly what PC relative relocations are for.
--
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH v1 5/6] objtool: Add skipped member in struct reloc
2023-08-09 11:51 ` Xi Ruoyao
@ 2023-08-09 11:58 ` Xi Ruoyao
0 siblings, 0 replies; 18+ messages in thread
From: Xi Ruoyao @ 2023-08-09 11:58 UTC (permalink / raw)
To: Peter Zijlstra, Tiezhu Yang
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On Wed, 2023-08-09 at 19:51 +0800, Xi Ruoyao wrote:
> On Thu, 2023-08-03 at 14:11 +0200, Peter Zijlstra wrote:
> > On Thu, Aug 03, 2023 at 07:36:24PM +0800, Tiezhu Yang wrote:
> >
> > > > ".long " __stringify_label(c) "b - .\n\t" \
> >
> > > I discussed offline with the engineers who are familiar with gnu
> > > assembler of LoongArch, maybe it can modify the gnu assembler
> > > to use R_LARCH_32_PCREL to replace a pair of R_LARCH_ADD32 and
> > > R_LARCH_SUB32, then I will test it again and drop this change if
> > > possible.
>
> Hmm, but I don't like the idea to disallow users having GAS <= 2.41 from
> using ORC unwinder.
>
> Can we emit the R_LARCH_32_PCREL reloc with an explicit .reloc
> directive?
Answering myself: "maybe", I've written a simple PoC:
$ cat t.s
.global main
main:
la.pcrel $t0, f
ld.w $t1, $t0, 0
ldx.d $a0, $t1, $t0
jr $ra
t:
.dword 42
.zero 64
f:
.reloc f, R_LARCH_32_PCREL, t
.word 0
$ ./a.out
$ echo $?
42
--
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University
^ permalink raw reply [flat|nested] 18+ messages in thread
* [PATCH v1 6/6] LoongArch: Add ORC unwinder support
2023-07-25 8:15 [PATCH v1 0/6] Add objtool and orc support for LoongArch Tiezhu Yang
` (4 preceding siblings ...)
2023-07-25 8:15 ` [PATCH v1 5/6] objtool: Add skipped member in struct reloc Tiezhu Yang
@ 2023-07-25 8:15 ` Tiezhu Yang
2023-07-25 12:01 ` [PATCH v1 0/6] Add objtool and orc support for LoongArch Peter Zijlstra
6 siblings, 0 replies; 18+ messages in thread
From: Tiezhu Yang @ 2023-07-25 8:15 UTC (permalink / raw)
To: Josh Poimboeuf, Peter Zijlstra, Huacai Chen
Cc: loongarch, linux-kernel, loongson-kernel
The kernel CONFIG_UNWINDER_ORC option enables the ORC unwinder, which is
similar in concept to a DWARF unwinder. The difference is that the format
of the ORC data is much simpler than DWARF, which in turn allows the ORC
unwinder to be much simpler and faster.
The ORC data consists of unwind tables which are generated by objtool.
They contain out-of-band data which is used by the in-kernel ORC unwinder.
Objtool generates the ORC data by first doing compile-time stack metadata
validation (CONFIG_STACK_VALIDATION). After analyzing all the code paths
of a .o file, it determines information about the stack state at each
instruction address in the file and outputs that information to the
.orc_unwind and .orc_unwind_ip sections.
The per-object ORC sections are combined at link time and are sorted and
post-processed at boot time. The unwinder uses the resulting data to
correlate instruction addresses with their stack states at run time.
Most of the logic are similar with x86, in order to get ra info before ra
is saved into stack, add ra_reg and ra_offset into orc_entry. At the same
time, modify some arch-specific code to silence the objtool warnings.
Co-developed-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Co-developed-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Youling Tang <tangyouling@loongson.cn>
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
arch/loongarch/Kconfig | 2 +
arch/loongarch/Kconfig.debug | 11 +
arch/loongarch/Makefile | 16 +
arch/loongarch/include/asm/Kbuild | 1 +
arch/loongarch/include/asm/bug.h | 1 +
arch/loongarch/include/asm/module.h | 7 +
| 19 +
arch/loongarch/include/asm/orc_lookup.h | 34 ++
arch/loongarch/include/asm/orc_types.h | 58 +++
arch/loongarch/include/asm/stackframe.h | 3 +
arch/loongarch/include/asm/unwind.h | 22 +-
arch/loongarch/include/asm/unwind_hints.h | 23 ++
arch/loongarch/kernel/Makefile | 3 +
arch/loongarch/kernel/entry.S | 2 +
arch/loongarch/kernel/genex.S | 2 +
arch/loongarch/kernel/head.S | 1 +
arch/loongarch/kernel/module.c | 21 +-
arch/loongarch/kernel/relocate_kernel.S | 11 +-
arch/loongarch/kernel/setup.c | 2 +
arch/loongarch/kernel/stacktrace.c | 1 +
arch/loongarch/kernel/unwind_orc.c | 586 ++++++++++++++++++++++++++++++
arch/loongarch/kernel/vmlinux.lds.S | 3 +
arch/loongarch/power/Makefile | 2 +
arch/loongarch/vdso/Makefile | 2 +
include/linux/compiler.h | 9 +
scripts/Makefile | 5 +-
26 files changed, 839 insertions(+), 8 deletions(-)
create mode 100644 arch/loongarch/include/asm/orc_header.h
create mode 100644 arch/loongarch/include/asm/orc_lookup.h
create mode 100644 arch/loongarch/include/asm/orc_types.h
create mode 100644 arch/loongarch/include/asm/unwind_hints.h
create mode 100644 arch/loongarch/kernel/unwind_orc.c
diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig
index e55511a..ae1d269 100644
--- a/arch/loongarch/Kconfig
+++ b/arch/loongarch/Kconfig
@@ -124,6 +124,7 @@ config LOONGARCH
select HAVE_KRETPROBES
select HAVE_MOD_ARCH_SPECIFIC
select HAVE_NMI
+ select HAVE_OBJTOOL if AS_HAS_EXPLICIT_RELOCS
select HAVE_PCI
select HAVE_PERF_EVENTS
select HAVE_PERF_REGS
@@ -134,6 +135,7 @@ config LOONGARCH
select HAVE_SAMPLE_FTRACE_DIRECT
select HAVE_SAMPLE_FTRACE_DIRECT_MULTI
select HAVE_SETUP_PER_CPU_AREA if NUMA
+ select HAVE_STACK_VALIDATION if HAVE_OBJTOOL
select HAVE_STACKPROTECTOR
select HAVE_SYSCALL_TRACEPOINTS
select HAVE_TIF_NOHZ
diff --git a/arch/loongarch/Kconfig.debug b/arch/loongarch/Kconfig.debug
index 8d36aab..98d6063 100644
--- a/arch/loongarch/Kconfig.debug
+++ b/arch/loongarch/Kconfig.debug
@@ -26,4 +26,15 @@ config UNWINDER_PROLOGUE
Some of the addresses it reports may be incorrect (but better than the
Guess unwinder).
+config UNWINDER_ORC
+ bool "ORC unwinder"
+ select OBJTOOL
+ help
+ This option enables the ORC (Oops Rewind Capability) unwinder for
+ unwinding kernel stack traces. It uses a custom data format which is
+ a simplified version of the DWARF Call Frame Information standard.
+
+ Enabling this option will increase the kernel's runtime memory usage
+ by roughly 2-4MB, depending on your kernel config.
+
endchoice
diff --git a/arch/loongarch/Makefile b/arch/loongarch/Makefile
index 09ba338..8977956 100644
--- a/arch/loongarch/Makefile
+++ b/arch/loongarch/Makefile
@@ -25,6 +25,10 @@ endif
32bit-emul = elf32loongarch
64bit-emul = elf64loongarch
+ifeq ($(CONFIG_OBJTOOL),y)
+KBUILD_CFLAGS += -fno-jump-tables
+endif
+
ifdef CONFIG_DYNAMIC_FTRACE
KBUILD_CPPFLAGS += -DCC_USING_PATCHABLE_FUNCTION_ENTRY
CC_FLAGS_FTRACE := -fpatchable-function-entry=2
@@ -125,6 +129,18 @@ drivers-y += arch/loongarch/crypto/
# suspend and hibernation support
drivers-$(CONFIG_PM) += arch/loongarch/power/
+ifdef CONFIG_UNWINDER_ORC
+orc_hash_h := arch/$(SRCARCH)/include/generated/asm/orc_hash.h
+orc_hash_sh := $(srctree)/scripts/orc_hash.sh
+targets += $(orc_hash_h)
+quiet_cmd_orc_hash = GEN $@
+ cmd_orc_hash = mkdir -p $(dir $@); \
+ $(CONFIG_SHELL) $(orc_hash_sh) < $< > $@
+$(orc_hash_h): $(srctree)/arch/loongarch/include/asm/orc_types.h $(orc_hash_sh) FORCE
+ $(call if_changed,orc_hash)
+archprepare: $(orc_hash_h)
+endif
+
ifeq ($(KBUILD_EXTMOD),)
prepare: vdso_prepare
vdso_prepare: prepare0
diff --git a/arch/loongarch/include/asm/Kbuild b/arch/loongarch/include/asm/Kbuild
index 6b222f2..f74159d 100644
--- a/arch/loongarch/include/asm/Kbuild
+++ b/arch/loongarch/include/asm/Kbuild
@@ -1,4 +1,5 @@
# SPDX-License-Identifier: GPL-2.0
+generated-y += orc_hash.h
generic-y += dma-contiguous.h
generic-y += export.h
generic-y += mcs_spinlock.h
diff --git a/arch/loongarch/include/asm/bug.h b/arch/loongarch/include/asm/bug.h
index d4ca3ba..0838887 100644
--- a/arch/loongarch/include/asm/bug.h
+++ b/arch/loongarch/include/asm/bug.h
@@ -44,6 +44,7 @@
do { \
instrumentation_begin(); \
__BUG_FLAGS(BUGFLAG_WARNING|(flags)); \
+ annotate_reachable(); \
instrumentation_end(); \
} while (0)
diff --git a/arch/loongarch/include/asm/module.h b/arch/loongarch/include/asm/module.h
index 2ecd82b..96af0ba 100644
--- a/arch/loongarch/include/asm/module.h
+++ b/arch/loongarch/include/asm/module.h
@@ -6,6 +6,7 @@
#define _ASM_MODULE_H
#include <asm/inst.h>
+#include <asm/orc_types.h>
#include <asm-generic/module.h>
#define RELA_STACK_DEPTH 16
@@ -23,6 +24,12 @@ struct mod_arch_specific {
/* For CONFIG_DYNAMIC_FTRACE */
struct plt_entry *ftrace_trampolines;
+
+#ifdef CONFIG_UNWINDER_ORC
+ unsigned int num_orcs;
+ int *orc_unwind_ip;
+ struct orc_entry *orc_unwind;
+#endif
};
struct got_entry {
--git a/arch/loongarch/include/asm/orc_header.h b/arch/loongarch/include/asm/orc_header.h
new file mode 100644
index 0000000..07bacf3
--- /dev/null
+++ b/arch/loongarch/include/asm/orc_header.h
@@ -0,0 +1,19 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/* Copyright (c) Meta Platforms, Inc. and affiliates. */
+
+#ifndef _ORC_HEADER_H
+#define _ORC_HEADER_H
+
+#include <linux/types.h>
+#include <linux/compiler.h>
+#include <asm/orc_hash.h>
+
+/*
+ * The header is currently a 20-byte hash of the ORC entry definition; see
+ * scripts/orc_hash.sh.
+ */
+#define ORC_HEADER \
+ __used __section(".orc_header") __aligned(4) \
+ static const u8 orc_header[] = { ORC_HASH }
+
+#endif /* _ORC_HEADER_H */
diff --git a/arch/loongarch/include/asm/orc_lookup.h b/arch/loongarch/include/asm/orc_lookup.h
new file mode 100644
index 0000000..2416312
--- /dev/null
+++ b/arch/loongarch/include/asm/orc_lookup.h
@@ -0,0 +1,34 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (C) 2017 Josh Poimboeuf <jpoimboe@redhat.com>
+ */
+#ifndef _ORC_LOOKUP_H
+#define _ORC_LOOKUP_H
+
+/*
+ * This is a lookup table for speeding up access to the .orc_unwind table.
+ * Given an input address offset, the corresponding lookup table entry
+ * specifies a subset of the .orc_unwind table to search.
+ *
+ * Each block represents the end of the previous range and the start of the
+ * next range. An extra block is added to give the last range an end.
+ *
+ * The block size should be a power of 2 to avoid a costly 'div' instruction.
+ *
+ * A block size of 256 was chosen because it roughly doubles unwinder
+ * performance while only adding ~5% to the ORC data footprint.
+ */
+#define LOOKUP_BLOCK_ORDER 8
+#define LOOKUP_BLOCK_SIZE (1 << LOOKUP_BLOCK_ORDER)
+
+#ifndef LINKER_SCRIPT
+
+extern unsigned int orc_lookup[];
+extern unsigned int orc_lookup_end[];
+
+#define LOOKUP_START_IP (unsigned long)_stext
+#define LOOKUP_STOP_IP (unsigned long)_etext
+
+#endif /* LINKER_SCRIPT */
+
+#endif /* _ORC_LOOKUP_H */
diff --git a/arch/loongarch/include/asm/orc_types.h b/arch/loongarch/include/asm/orc_types.h
new file mode 100644
index 0000000..1d37e62
--- /dev/null
+++ b/arch/loongarch/include/asm/orc_types.h
@@ -0,0 +1,58 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef _ORC_TYPES_H
+#define _ORC_TYPES_H
+
+#include <linux/types.h>
+
+/*
+ * The ORC_REG_* registers are base registers which are used to find other
+ * registers on the stack.
+ *
+ * ORC_REG_PREV_SP, also known as DWARF Call Frame Address (CFA), is the
+ * address of the previous frame: the caller's SP before it called the current
+ * function.
+ *
+ * ORC_REG_UNDEFINED means the corresponding register's value didn't change in
+ * the current frame.
+ *
+ * The most commonly used base registers are SP and BP -- which the previous SP
+ * is usually based on -- and PREV_SP and UNDEFINED -- which the previous BP is
+ * usually based on.
+ *
+ * The rest of the base registers are needed for special cases like entry code
+ * and GCC realigned stacks.
+ */
+#define ORC_REG_UNDEFINED 0
+#define ORC_REG_PREV_SP 1
+#define ORC_REG_SP 2
+#define ORC_REG_BP 3
+#define ORC_REG_MAX 4
+
+#define ORC_TYPE_UNDEFINED 0
+#define ORC_TYPE_END_OF_STACK 1
+#define ORC_TYPE_CALL 2
+#define ORC_TYPE_REGS 3
+#define ORC_TYPE_REGS_PARTIAL 4
+
+#ifndef __ASSEMBLY__
+/*
+ * This struct is more or less a vastly simplified version of the DWARF Call
+ * Frame Information standard. It contains only the necessary parts of DWARF
+ * CFI, simplified for ease of access by the in-kernel unwinder. It tells the
+ * unwinder how to find the previous SP and BP (and sometimes entry regs) on
+ * the stack for a given code address. Each instance of the struct corresponds
+ * to one or more code locations.
+ */
+struct orc_entry {
+ s16 sp_offset;
+ s16 bp_offset;
+ s16 ra_offset;
+ unsigned int sp_reg:4;
+ unsigned int bp_reg:4;
+ unsigned int ra_reg:4;
+ unsigned int type:3;
+ unsigned int signal:1;
+};
+#endif /* __ASSEMBLY__ */
+
+#endif /* _ORC_TYPES_H */
diff --git a/arch/loongarch/include/asm/stackframe.h b/arch/loongarch/include/asm/stackframe.h
index 7df80e6..ab16f2d 100644
--- a/arch/loongarch/include/asm/stackframe.h
+++ b/arch/loongarch/include/asm/stackframe.h
@@ -13,6 +13,7 @@
#include <asm/asm-offsets.h>
#include <asm/loongarch.h>
#include <asm/thread_info.h>
+#include <asm/unwind_hints.h>
/* Make the addition of cfi info a little easier. */
.macro cfi_rel_offset reg offset=0 docfi=0
@@ -158,6 +159,7 @@
cfi_st u0, PT_R21, \docfi
csrrd u0, PERCPU_BASE_KS
9:
+ UNWIND_HINT_REGS
.endm
.macro SAVE_ALL docfi=0
@@ -215,6 +217,7 @@
.macro RESTORE_SP_AND_RET docfi=0
cfi_ld sp, PT_R3, \docfi
+ UNWIND_HINT_FUNC
ertn
.endm
diff --git a/arch/loongarch/include/asm/unwind.h b/arch/loongarch/include/asm/unwind.h
index b9dce87..d36e04e 100644
--- a/arch/loongarch/include/asm/unwind.h
+++ b/arch/loongarch/include/asm/unwind.h
@@ -16,6 +16,7 @@
enum unwinder_type {
UNWINDER_GUESS,
UNWINDER_PROLOGUE,
+ UNWINDER_ORC,
};
struct unwind_state {
@@ -24,7 +25,7 @@ struct unwind_state {
struct task_struct *task;
bool first, error, reset;
int graph_idx;
- unsigned long sp, pc, ra;
+ unsigned long sp, pc, ra, fp;
};
bool default_next_frame(struct unwind_state *state);
@@ -34,6 +35,17 @@ void unwind_start(struct unwind_state *state,
bool unwind_next_frame(struct unwind_state *state);
unsigned long unwind_get_return_address(struct unwind_state *state);
+#ifdef CONFIG_UNWINDER_ORC
+void unwind_init(void);
+void unwind_module_init(struct module *mod, void *orc_ip, size_t orc_ip_size,
+ void *orc, size_t orc_size);
+#else
+static inline void unwind_init(void) {}
+static inline
+void unwind_module_init(struct module *mod, void *orc_ip, size_t orc_ip_size,
+ void *orc, size_t orc_size) {}
+#endif
+
static inline bool unwind_done(struct unwind_state *state)
{
return state->stack_info.type == STACK_TYPE_UNKNOWN;
@@ -61,14 +73,17 @@ static __always_inline void __unwind_start(struct unwind_state *state,
state->sp = regs->regs[3];
state->pc = regs->csr_era;
state->ra = regs->regs[1];
+ state->fp = regs->regs[22];
} else if (task && task != current) {
state->sp = thread_saved_fp(task);
state->pc = thread_saved_ra(task);
state->ra = 0;
+ state->fp = 0;
} else {
state->sp = (unsigned long)__builtin_frame_address(0);
state->pc = (unsigned long)__builtin_return_address(0);
state->ra = 0;
+ state->fp = 0;
}
state->task = task;
get_stack_info(state->sp, state->task, &state->stack_info);
@@ -77,6 +92,9 @@ static __always_inline void __unwind_start(struct unwind_state *state,
static __always_inline unsigned long __unwind_get_return_address(struct unwind_state *state)
{
- return unwind_done(state) ? 0 : state->pc;
+ if (unwind_done(state))
+ return 0;
+
+ return __kernel_text_address(state->pc) ? state->pc : 0;
}
#endif /* _ASM_UNWIND_H */
diff --git a/arch/loongarch/include/asm/unwind_hints.h b/arch/loongarch/include/asm/unwind_hints.h
new file mode 100644
index 0000000..836b4b6
--- /dev/null
+++ b/arch/loongarch/include/asm/unwind_hints.h
@@ -0,0 +1,23 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_LOONGARCH_UNWIND_HINTS_H
+#define _ASM_LOONGARCH_UNWIND_HINTS_H
+
+#include <linux/objtool.h>
+
+#ifdef __ASSEMBLY__
+
+.macro UNWIND_HINT_EMPTY
+ UNWIND_HINT sp_reg=ORC_REG_UNDEFINED type=UNWIND_HINT_TYPE_CALL
+.endm
+
+.macro UNWIND_HINT_REGS base=ORC_REG_SP offset=0
+ UNWIND_HINT sp_reg=\base sp_offset=\offset type=UNWIND_HINT_TYPE_REGS
+.endm
+
+.macro UNWIND_HINT_FUNC sp_offset=0
+ UNWIND_HINT sp_reg=ORC_REG_SP sp_offset=\sp_offset type=UNWIND_HINT_TYPE_CALL
+.endm
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* _ASM_LOONGARCH_UNWIND_HINTS_H */
diff --git a/arch/loongarch/kernel/Makefile b/arch/loongarch/kernel/Makefile
index 8e279f0..2d03c5a 100644
--- a/arch/loongarch/kernel/Makefile
+++ b/arch/loongarch/kernel/Makefile
@@ -3,6 +3,8 @@
# Makefile for the Linux/LoongArch kernel.
#
+OBJECT_FILES_NON_STANDARD_head.o := y
+
extra-y := vmlinux.lds
obj-y += head.o cpu-probe.o cacheinfo.o env.o setup.o entry.o genex.o \
@@ -50,6 +52,7 @@ obj-$(CONFIG_CRASH_DUMP) += crash_dump.o
obj-$(CONFIG_UNWINDER_GUESS) += unwind_guess.o
obj-$(CONFIG_UNWINDER_PROLOGUE) += unwind_prologue.o
+obj-$(CONFIG_UNWINDER_ORC) += unwind_orc.o
obj-$(CONFIG_PERF_EVENTS) += perf_event.o perf_regs.o
obj-$(CONFIG_HAVE_HW_BREAKPOINT) += hw_breakpoint.o
diff --git a/arch/loongarch/kernel/entry.S b/arch/loongarch/kernel/entry.S
index d737e3c..458d4e8 100644
--- a/arch/loongarch/kernel/entry.S
+++ b/arch/loongarch/kernel/entry.S
@@ -70,6 +70,7 @@ SYM_FUNC_END(handle_syscall)
_ASM_NOKPROBE(handle_syscall)
SYM_CODE_START(ret_from_fork)
+ UNWIND_HINT_REGS
bl schedule_tail # a0 = struct task_struct *prev
move a0, sp
bl syscall_exit_to_user_mode
@@ -79,6 +80,7 @@ SYM_CODE_START(ret_from_fork)
SYM_CODE_END(ret_from_fork)
SYM_CODE_START(ret_from_kernel_thread)
+ UNWIND_HINT_REGS
bl schedule_tail # a0 = struct task_struct *prev
move a0, s1
jirl ra, s0, 0
diff --git a/arch/loongarch/kernel/genex.S b/arch/loongarch/kernel/genex.S
index 78f0663..d103a96 100644
--- a/arch/loongarch/kernel/genex.S
+++ b/arch/loongarch/kernel/genex.S
@@ -77,7 +77,9 @@ SYM_FUNC_END(except_vec_cex)
668:
RESTORE_ALL_AND_RET
SYM_FUNC_END(handle_\exception)
+ .pushsection ".data", "aw", %progbits
SYM_DATA(unwind_hint_\exception, .word 668b - 666b)
+ .popsection
.endm
BUILD_HANDLER ade ade badv
diff --git a/arch/loongarch/kernel/head.S b/arch/loongarch/kernel/head.S
index 5e828a8..f979bb5 100644
--- a/arch/loongarch/kernel/head.S
+++ b/arch/loongarch/kernel/head.S
@@ -43,6 +43,7 @@ SYM_DATA(kernel_offset, .long _kernel_offset);
.align 12
SYM_CODE_START(kernel_entry) # kernel entry point
+ UNWIND_HINT_EMPTY
/* Config direct window and set PG */
li.d t0, CSR_DMW0_INIT # UC, PLV0, 0x8000 xxxx xxxx xxxx
diff --git a/arch/loongarch/kernel/module.c b/arch/loongarch/kernel/module.c
index b8b8608..9e66a9a 100644
--- a/arch/loongarch/kernel/module.c
+++ b/arch/loongarch/kernel/module.c
@@ -20,6 +20,7 @@
#include <linux/kernel.h>
#include <asm/alternative.h>
#include <asm/inst.h>
+#include <asm/unwind.h>
static int rela_stack_push(s64 stack_value, s64 *rela_stack, size_t *rela_stack_top)
{
@@ -367,6 +368,15 @@ static int apply_r_larch_got_pc(struct module *mod,
return apply_r_larch_pcala(mod, location, got, rela_stack, rela_stack_top, type);
}
+static int apply_r_larch_32_pcrel(struct module *mod, u32 *location, Elf_Addr v,
+ s64 *rela_stack, size_t *rela_stack_top, unsigned int type)
+{
+ ptrdiff_t offset = (void *)v - (void *)location;
+
+ *(u32 *)location = offset;
+ return 0;
+}
+
/*
* reloc_handlers_rela() - Apply a particular relocation to a module
* @mod: the module to apply the reloc to
@@ -396,6 +406,7 @@ static reloc_rela_handler reloc_rela_handlers[] = {
[R_LARCH_SOP_POP_32_S_10_5 ... R_LARCH_SOP_POP_32_U] = apply_r_larch_sop_imm_field,
[R_LARCH_ADD32 ... R_LARCH_SUB64] = apply_r_larch_add_sub,
[R_LARCH_PCALA_HI20...R_LARCH_PCALA64_HI12] = apply_r_larch_pcala,
+ [R_LARCH_32_PCREL] = apply_r_larch_32_pcrel,
};
int apply_relocate_add(Elf_Shdr *sechdrs, const char *strtab,
@@ -495,7 +506,7 @@ static void module_init_ftrace_plt(const Elf_Ehdr *hdr,
int module_finalize(const Elf_Ehdr *hdr,
const Elf_Shdr *sechdrs, struct module *mod)
{
- const Elf_Shdr *s, *se;
+ const Elf_Shdr *s, *se, *orc = NULL, *orc_ip = NULL;
const char *secstrs = (void *)hdr + sechdrs[hdr->e_shstrndx].sh_offset;
for (s = sechdrs, se = sechdrs + hdr->e_shnum; s < se; s++) {
@@ -503,7 +514,15 @@ int module_finalize(const Elf_Ehdr *hdr,
apply_alternatives((void *)s->sh_addr, (void *)s->sh_addr + s->sh_size);
if (!strcmp(".ftrace_trampoline", secstrs + s->sh_name))
module_init_ftrace_plt(hdr, s, mod);
+ if (!strcmp(".orc_unwind", secstrs + s->sh_name))
+ orc = s;
+ if (!strcmp(".orc_unwind_ip", secstrs + s->sh_name))
+ orc_ip = s;
}
+ if (orc && orc_ip)
+ unwind_module_init(mod, (void *)orc_ip->sh_addr, orc_ip->sh_size,
+ (void *)orc->sh_addr, orc->sh_size);
+
return 0;
}
diff --git a/arch/loongarch/kernel/relocate_kernel.S b/arch/loongarch/kernel/relocate_kernel.S
index d132525..9699be4 100644
--- a/arch/loongarch/kernel/relocate_kernel.S
+++ b/arch/loongarch/kernel/relocate_kernel.S
@@ -13,8 +13,10 @@
#include <asm/loongarch.h>
#include <asm/stackframe.h>
#include <asm/addrspace.h>
+#include <asm/unwind_hints.h>
SYM_CODE_START(relocate_new_kernel)
+ UNWIND_HINT_EMPTY
/*
* a0: EFI boot flag for the new kernel
* a1: Command line pointer for the new kernel
@@ -72,7 +74,6 @@ copy_word:
LONG_ADDI s5, s5, -1
beqz s5, process_entry
b copy_word
- b process_entry
done:
ibar 0
@@ -91,6 +92,8 @@ SYM_CODE_END(relocate_new_kernel)
* then start at the entry point from LOONGARCH_IOCSR_MBUF0.
*/
SYM_CODE_START(kexec_smp_wait)
+ UNWIND_HINT_EMPTY
+
1: li.w t0, 0x100 /* wait for init loop */
2: addi.w t0, t0, -1 /* limit mailbox access */
bnez t0, 2b
@@ -107,6 +110,6 @@ SYM_CODE_END(kexec_smp_wait)
relocate_new_kernel_end:
-SYM_DATA_START(relocate_new_kernel_size)
- PTR relocate_new_kernel_end - relocate_new_kernel
-SYM_DATA_END(relocate_new_kernel_size)
+.pushsection ".data", "aw", %progbits
+SYM_DATA(relocate_new_kernel_size, .long relocate_new_kernel_end - relocate_new_kernel)
+.popsection
diff --git a/arch/loongarch/kernel/setup.c b/arch/loongarch/kernel/setup.c
index 78a0035..3107c3c 100644
--- a/arch/loongarch/kernel/setup.c
+++ b/arch/loongarch/kernel/setup.c
@@ -48,6 +48,7 @@
#include <asm/sections.h>
#include <asm/setup.h>
#include <asm/time.h>
+#include <asm/unwind.h>
#define SMBIOS_BIOSSIZE_OFFSET 0x09
#define SMBIOS_BIOSEXTERN_OFFSET 0x13
@@ -590,6 +591,7 @@ static void __init prefill_possible_map(void)
void __init setup_arch(char **cmdline_p)
{
cpu_probe();
+ unwind_init();
init_environ();
efi_init();
diff --git a/arch/loongarch/kernel/stacktrace.c b/arch/loongarch/kernel/stacktrace.c
index 2463d2f..f7030fb 100644
--- a/arch/loongarch/kernel/stacktrace.c
+++ b/arch/loongarch/kernel/stacktrace.c
@@ -29,6 +29,7 @@ void arch_stack_walk(stack_trace_consume_fn consume_entry, void *cookie,
}
regs->regs[1] = 0;
+ regs->regs[22] = 0;
for (unwind_start(&state, task, regs);
!unwind_done(&state) && !unwind_error(&state); unwind_next_frame(&state)) {
addr = unwind_get_return_address(&state);
diff --git a/arch/loongarch/kernel/unwind_orc.c b/arch/loongarch/kernel/unwind_orc.c
new file mode 100644
index 0000000..0234762
--- /dev/null
+++ b/arch/loongarch/kernel/unwind_orc.c
@@ -0,0 +1,586 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include <linux/objtool.h>
+#include <linux/module.h>
+#include <linux/sort.h>
+#include <asm/ptrace.h>
+#include <asm/setup.h>
+#include <asm/stacktrace.h>
+#include <asm/unwind.h>
+#include <asm/orc_types.h>
+#include <asm/orc_lookup.h>
+#include <asm/orc_header.h>
+
+ORC_HEADER;
+
+#define orc_warn(fmt, ...) \
+ printk_deferred_once(KERN_WARNING "WARNING: " fmt, ##__VA_ARGS__)
+
+extern int __start_orc_unwind_ip[];
+extern int __stop_orc_unwind_ip[];
+extern struct orc_entry __start_orc_unwind[];
+extern struct orc_entry __stop_orc_unwind[];
+
+static bool orc_init __ro_after_init;
+static unsigned int lookup_num_blocks __ro_after_init;
+
+extern asmlinkage void handle_ade(void);
+extern asmlinkage void handle_ale(void);
+extern asmlinkage void handle_sys(void);
+extern asmlinkage void handle_bp(void);
+extern asmlinkage void handle_ri(void);
+extern asmlinkage void handle_fpu(void);
+extern asmlinkage void handle_fpe(void);
+extern asmlinkage void handle_lbt(void);
+extern asmlinkage void handle_lsx(void);
+extern asmlinkage void handle_lasx(void);
+extern asmlinkage void handle_reserved(void);
+extern asmlinkage void handle_watch(void);
+extern asmlinkage void handle_vint(void);
+extern asmlinkage void handle_tlb_load(void);
+extern asmlinkage void handle_tlb_store(void);
+extern asmlinkage void handle_tlb_modify(void);
+extern asmlinkage void handle_tlb_protect(void);
+
+/* Fake frame pointer entry -- used as a fallback for generated code */
+static struct orc_entry orc_fp_entry = {
+ .type = UNWIND_HINT_TYPE_CALL,
+ .sp_reg = ORC_REG_BP,
+ .sp_offset = 16,
+ .bp_reg = ORC_REG_PREV_SP,
+ .bp_offset = -16,
+ .ra_reg = ORC_REG_PREV_SP,
+ .ra_offset = -8,
+};
+
+static inline unsigned long orc_ip(const int *ip)
+{
+ return (unsigned long)ip + *ip;
+}
+
+static struct orc_entry *__orc_find(int *ip_table, struct orc_entry *u_table,
+ unsigned int num_entries, unsigned long ip)
+{
+ int *first = ip_table;
+ int *last = ip_table + num_entries - 1;
+ int *mid = first, *found = first;
+
+ if (!num_entries)
+ return NULL;
+
+ /*
+ * Do a binary range search to find the rightmost duplicate of a given
+ * starting address. Some entries are section terminators which are
+ * "weak" entries for ensuring there are no gaps. They should be
+ * ignored when they conflict with a real entry.
+ */
+ while (first <= last) {
+ mid = first + ((last - first) / 2);
+
+ if (orc_ip(mid) <= ip) {
+ found = mid;
+ first = mid + 1;
+ } else
+ last = mid - 1;
+ }
+
+ return u_table + (found - ip_table);
+}
+
+#ifdef CONFIG_MODULES
+static struct orc_entry *orc_module_find(unsigned long ip)
+{
+ struct module *mod;
+
+ mod = __module_address(ip);
+ if (!mod || !mod->arch.orc_unwind || !mod->arch.orc_unwind_ip)
+ return NULL;
+ return __orc_find(mod->arch.orc_unwind_ip, mod->arch.orc_unwind,
+ mod->arch.num_orcs, ip);
+}
+#else
+static struct orc_entry *orc_module_find(unsigned long ip)
+{
+ return NULL;
+}
+#endif
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+static struct orc_entry *orc_find(unsigned long ip);
+
+/*
+ * Ftrace dynamic trampolines do not have orc entries of their own.
+ * But they are copies of the ftrace entries that are static and
+ * defined in ftrace_*.S, which do have orc entries.
+ *
+ * If the unwinder comes across a ftrace trampoline, then find the
+ * ftrace function that was used to create it, and use that ftrace
+ * function's orc entry, as the placement of the return code in
+ * the stack will be identical.
+ */
+static struct orc_entry *orc_ftrace_find(unsigned long ip)
+{
+ struct ftrace_ops *ops;
+ unsigned long tramp_addr, offset;
+
+ ops = ftrace_ops_trampoline(ip);
+ if (!ops)
+ return NULL;
+
+ /* Set tramp_addr to the start of the code copied by the trampoline */
+ if (ops->flags & FTRACE_OPS_FL_SAVE_REGS)
+ tramp_addr = (unsigned long)ftrace_regs_caller;
+ else
+ tramp_addr = (unsigned long)ftrace_caller;
+
+ /* Now place tramp_addr to the location within the trampoline ip is at */
+ offset = ip - ops->trampoline;
+ tramp_addr += offset;
+
+ /* Prevent unlikely recursion */
+ if (ip == tramp_addr)
+ return NULL;
+
+ return orc_find(tramp_addr);
+}
+#else
+static struct orc_entry *orc_ftrace_find(unsigned long ip)
+{
+ return NULL;
+}
+#endif
+
+/*
+ * If we crash with IP==0, the last successfully executed instruction
+ * was probably an indirect function call with a NULL function pointer,
+ * and we don't have unwind information for NULL.
+ * This hardcoded ORC entry for IP==0 allows us to unwind from a NULL function
+ * pointer into its parent and then continue normally from there.
+ */
+static struct orc_entry null_orc_entry = {
+ .sp_offset = sizeof(long),
+ .sp_reg = ORC_REG_SP,
+ .bp_reg = ORC_REG_UNDEFINED,
+ .type = ORC_TYPE_CALL
+};
+
+static struct orc_entry *orc_find(unsigned long ip)
+{
+ static struct orc_entry *orc;
+
+ if (ip == 0)
+ return &null_orc_entry;
+
+ /* For non-init vmlinux addresses, use the fast lookup table: */
+ if (ip >= LOOKUP_START_IP && ip < LOOKUP_STOP_IP) {
+ unsigned int idx, start, stop;
+
+ idx = (ip - LOOKUP_START_IP) / LOOKUP_BLOCK_SIZE;
+
+ if (unlikely((idx >= lookup_num_blocks-1))) {
+ orc_warn("WARNING: bad lookup idx: idx=%u num=%u ip=%pB\n",
+ idx, lookup_num_blocks, (void *)ip);
+ return NULL;
+ }
+
+ start = orc_lookup[idx];
+ stop = orc_lookup[idx + 1] + 1;
+
+ if (unlikely((__start_orc_unwind + start >= __stop_orc_unwind) ||
+ (__start_orc_unwind + stop > __stop_orc_unwind))) {
+ orc_warn("WARNING: bad lookup value: idx=%u num=%u start=%u stop=%u ip=%pB\n",
+ idx, lookup_num_blocks, start, stop, (void *)ip);
+ return NULL;
+ }
+
+ return __orc_find(__start_orc_unwind_ip + start,
+ __start_orc_unwind + start, stop - start, ip);
+ }
+
+ /* vmlinux .init slow lookup: */
+ if (is_kernel_inittext(ip))
+ return __orc_find(__start_orc_unwind_ip, __start_orc_unwind,
+ __stop_orc_unwind_ip - __start_orc_unwind_ip, ip);
+
+ /* Module lookup: */
+ orc = orc_module_find(ip);
+ if (orc)
+ return orc;
+
+ return orc_ftrace_find(ip);
+}
+
+#ifdef CONFIG_MODULES
+
+static DEFINE_MUTEX(sort_mutex);
+static int *cur_orc_ip_table = __start_orc_unwind_ip;
+static struct orc_entry *cur_orc_table = __start_orc_unwind;
+
+static void orc_sort_swap(void *_a, void *_b, int size)
+{
+ struct orc_entry *orc_a, *orc_b;
+ int *a = _a, *b = _b, tmp;
+ int delta = _b - _a;
+
+ /* Swap the .orc_unwind_ip entries: */
+ tmp = *a;
+ *a = *b + delta;
+ *b = tmp - delta;
+
+ /* Swap the corresponding .orc_unwind entries: */
+ orc_a = cur_orc_table + (a - cur_orc_ip_table);
+ orc_b = cur_orc_table + (b - cur_orc_ip_table);
+ swap(*orc_a, *orc_b);
+}
+
+static int orc_sort_cmp(const void *_a, const void *_b)
+{
+ struct orc_entry *orc_a;
+ const int *a = _a, *b = _b;
+ unsigned long a_val = orc_ip(a);
+ unsigned long b_val = orc_ip(b);
+
+ if (a_val > b_val)
+ return 1;
+ if (a_val < b_val)
+ return -1;
+
+ /*
+ * The "weak" section terminator entries need to always be first
+ * to ensure the lookup code skips them in favor of real entries.
+ * These terminator entries exist to handle any gaps created by
+ * whitelisted .o files which didn't get objtool generation.
+ */
+ orc_a = cur_orc_table + (a - cur_orc_ip_table);
+ return orc_a->type == ORC_TYPE_UNDEFINED ? -1 : 1;
+}
+
+void unwind_module_init(struct module *mod, void *_orc_ip, size_t orc_ip_size,
+ void *_orc, size_t orc_size)
+{
+ int *orc_ip = _orc_ip;
+ struct orc_entry *orc = _orc;
+ unsigned int num_entries = orc_ip_size / sizeof(int);
+
+ WARN_ON_ONCE(orc_ip_size % sizeof(int) != 0 ||
+ orc_size % sizeof(*orc) != 0 ||
+ num_entries != orc_size / sizeof(*orc));
+
+ /*
+ * The 'cur_orc_*' globals allow the orc_sort_swap() callback to
+ * associate an .orc_unwind_ip table entry with its corresponding
+ * .orc_unwind entry so they can both be swapped.
+ */
+ mutex_lock(&sort_mutex);
+ cur_orc_ip_table = orc_ip;
+ cur_orc_table = orc;
+ sort(orc_ip, num_entries, sizeof(int), orc_sort_cmp, orc_sort_swap);
+ mutex_unlock(&sort_mutex);
+
+ mod->arch.orc_unwind_ip = orc_ip;
+ mod->arch.orc_unwind = orc;
+ mod->arch.num_orcs = num_entries;
+}
+#endif
+
+void __init unwind_init(void)
+{
+ size_t orc_ip_size = (void *)__stop_orc_unwind_ip - (void *)__start_orc_unwind_ip;
+ size_t orc_size = (void *)__stop_orc_unwind - (void *)__start_orc_unwind;
+ size_t num_entries = orc_ip_size / sizeof(int);
+ struct orc_entry *orc;
+ int i;
+
+ if (!num_entries || orc_ip_size % sizeof(int) != 0 ||
+ orc_size % sizeof(struct orc_entry) != 0 ||
+ num_entries != orc_size / sizeof(struct orc_entry)) {
+ orc_warn("WARNING: Bad or missing .orc_unwind table. Disabling unwinder.\n");
+ return;
+ }
+
+ /*
+ * Note, the orc_unwind and orc_unwind_ip tables were already
+ * sorted at build time via the 'sorttable' tool.
+ * It's ready for binary search straight away, no need to sort it.
+ */
+
+ /* Initialize the fast lookup table: */
+ lookup_num_blocks = orc_lookup_end - orc_lookup;
+ for (i = 0; i < lookup_num_blocks-1; i++) {
+ orc = __orc_find(__start_orc_unwind_ip, __start_orc_unwind,
+ num_entries,
+ LOOKUP_START_IP + (LOOKUP_BLOCK_SIZE * i));
+ if (!orc) {
+ orc_warn("WARNING: Corrupt .orc_unwind table. Disabling unwinder.\n");
+ return;
+ }
+
+ orc_lookup[i] = orc - __start_orc_unwind;
+ }
+
+ /* Initialize the ending block: */
+ orc = __orc_find(__start_orc_unwind_ip, __start_orc_unwind, num_entries,
+ LOOKUP_STOP_IP);
+ if (!orc) {
+ orc_warn("WARNING: Corrupt .orc_unwind table. Disabling unwinder.\n");
+ return;
+ }
+ orc_lookup[lookup_num_blocks-1] = orc - __start_orc_unwind;
+
+ orc_init = true;
+}
+
+static inline bool on_stack(struct stack_info *info, unsigned long addr, size_t len)
+{
+ unsigned long begin = info->begin;
+ unsigned long end = info->end;
+
+ return (info->type != STACK_TYPE_UNKNOWN &&
+ addr >= begin && addr < end &&
+ addr + len > begin && addr + len <= end);
+}
+
+static bool stack_access_ok(struct unwind_state *state, unsigned long addr,
+ size_t len)
+{
+ struct stack_info *info = &state->stack_info;
+
+ if (on_stack(info, addr, len))
+ return true;
+
+ return !get_stack_info(addr, state->task, info) &&
+ on_stack(info, addr, len);
+}
+
+unsigned long unwind_get_return_address(struct unwind_state *state)
+{
+ return __unwind_get_return_address(state);
+}
+EXPORT_SYMBOL_GPL(unwind_get_return_address);
+
+void unwind_start(struct unwind_state *state, struct task_struct *task,
+ struct pt_regs *regs)
+{
+ __unwind_start(state, task, regs);
+ if (!unwind_done(state) && !__kernel_text_address(state->pc))
+ unwind_next_frame(state);
+}
+EXPORT_SYMBOL_GPL(unwind_start);
+
+
+static bool is_entry_func(unsigned long addr)
+{
+ extern u32 kernel_entry;
+ extern u32 kernel_entry_end;
+
+ return addr >= (unsigned long)&kernel_entry &&
+ addr < (unsigned long)&kernel_entry_end;
+}
+
+static inline unsigned long bt_address(unsigned long ra)
+{
+ extern unsigned long eentry;
+
+ if (__kernel_text_address(ra))
+ return ra;
+
+ /* We are in preempt_disable() here */
+ if (__module_text_address(ra))
+ return ra;
+
+ if (ra >= eentry && ra < eentry + EXCCODE_INT_END * VECSIZE) {
+ unsigned long type = (ra - eentry) / VECSIZE;
+ unsigned long offset = (ra - eentry) % VECSIZE;
+ unsigned long func;
+
+ switch (type) {
+ case EXCCODE_TLBL:
+ case EXCCODE_TLBI:
+ func = (unsigned long)handle_tlb_load;
+ break;
+ case EXCCODE_TLBS:
+ func = (unsigned long)handle_tlb_store;
+ break;
+ case EXCCODE_TLBM:
+ func = (unsigned long)handle_tlb_modify;
+ break;
+ case EXCCODE_TLBNR:
+ case EXCCODE_TLBNX:
+ case EXCCODE_TLBPE:
+ func = (unsigned long)handle_tlb_protect;
+ break;
+ case EXCCODE_ADE:
+ func = (unsigned long)handle_ade;
+ break;
+ case EXCCODE_ALE:
+ func = (unsigned long)handle_ale;
+ break;
+ case EXCCODE_SYS:
+ func = (unsigned long)handle_sys;
+ break;
+ case EXCCODE_BP:
+ func = (unsigned long)handle_bp;
+ break;
+ case EXCCODE_INE:
+ case EXCCODE_IPE:
+ func = (unsigned long)handle_ri;
+ break;
+ case EXCCODE_FPDIS:
+ func = (unsigned long)handle_fpu;
+ break;
+ case EXCCODE_LSXDIS:
+ func = (unsigned long)handle_lsx;
+ break;
+ case EXCCODE_LASXDIS:
+ func = (unsigned long)handle_lasx;
+ break;
+ case EXCCODE_FPE:
+ func = (unsigned long)handle_fpe;
+ break;
+ case EXCCODE_BTDIS:
+ func = (unsigned long)handle_lbt;
+ break;
+ case EXCCODE_WATCH:
+ func = (unsigned long)handle_watch;
+ break;
+ case EXCCODE_INT_START ... EXCCODE_INT_END - 1:
+ func = (unsigned long)handle_vint;
+ break;
+ default:
+
+ func = (unsigned long)handle_reserved;
+ break;
+ }
+
+ return func + offset;
+ }
+
+ return ra;
+}
+
+bool unwind_next_frame(struct unwind_state *state)
+{
+ struct stack_info *info = &state->stack_info;
+ struct orc_entry *orc;
+ struct pt_regs *regs;
+ unsigned long *p, pc;
+
+ if (unwind_done(state))
+ return false;
+
+ /* Don't let modules unload while we're reading their ORC data. */
+ preempt_disable();
+
+ if (is_entry_func(state->pc))
+ goto end;
+
+ orc = orc_find(state->pc);
+ if (!orc) {
+ orc = &orc_fp_entry;
+ state->error = true;
+ }
+
+ switch (orc->sp_reg) {
+ case ORC_REG_SP:
+ state->sp = state->sp + orc->sp_offset;
+ break;
+ case ORC_REG_BP:
+ state->sp = state->fp;
+ break;
+ default:
+ orc_warn("unknown SP base reg %d at %pB\n",
+ orc->sp_reg, (void *)state->pc);
+ goto err;
+ }
+
+ switch (orc->bp_reg) {
+ case ORC_REG_PREV_SP:
+ p = (unsigned long *)(state->sp + orc->bp_offset);
+ if (!stack_access_ok(state, (unsigned long)p, sizeof(unsigned long)))
+ goto err;
+
+ state->fp = *p;
+ break;
+ case ORC_REG_UNDEFINED:
+ /* Nothing. */
+ break;
+ default:
+ orc_warn("unknown FP base reg %d at %pB\n",
+ orc->bp_reg, (void *)state->pc);
+ goto err;
+ }
+
+ switch (orc->type) {
+ case UNWIND_HINT_TYPE_CALL:
+ if (orc->ra_reg == ORC_REG_PREV_SP) {
+ p = (unsigned long *)(state->sp + orc->ra_offset);
+ if (!stack_access_ok(state, (unsigned long)p, sizeof(unsigned long)))
+ goto err;
+
+ pc = unwind_graph_addr(state, *p, state->sp);
+ pc -= LOONGARCH_INSN_SIZE;
+ } else if (orc->ra_reg == ORC_REG_UNDEFINED) {
+ if (!state->ra || state->ra == state->pc)
+ goto err;
+
+ pc = unwind_graph_addr(state, state->ra, state->sp);
+ pc -= LOONGARCH_INSN_SIZE;
+ state->ra = 0;
+ } else {
+ orc_warn("unknown ra base reg %d at %pB\n",
+ orc->ra_reg, (void *)state->pc);
+ goto err;
+ }
+ break;
+ case UNWIND_HINT_TYPE_REGS:
+ if (state->stack_info.type == STACK_TYPE_IRQ && state->sp == info->end)
+ regs = (struct pt_regs *)info->next_sp;
+ else
+ regs = (struct pt_regs *)state->sp;
+
+ if (!stack_access_ok(state, (unsigned long)regs, sizeof(*regs)))
+ goto err;
+
+ if ((info->end == (unsigned long)regs + sizeof(*regs)) &&
+ !regs->regs[3] && !regs->regs[1])
+ goto end;
+
+ if (user_mode(regs))
+ goto end;
+
+ pc = regs->csr_era;
+ if (!__kernel_text_address(pc))
+ goto err;
+
+ state->sp = regs->regs[3];
+ state->ra = regs->regs[1];
+ state->fp = regs->regs[22];
+ get_stack_info(state->sp, state->task, info);
+
+ break;
+ default:
+ orc_warn("unknown .orc_unwind entry type %d at %pB\n",
+ orc->type, (void *)state->pc);
+ goto err;
+ }
+
+ state->pc = bt_address(pc);
+ if (!state->pc) {
+ pr_err("cannot find unwind pc at %pK\n", (void *)pc);
+ goto err;
+ }
+
+ if (!__kernel_text_address(state->pc))
+ goto err;
+
+ preempt_enable();
+ return true;
+
+err:
+ state->error = true;
+
+end:
+ preempt_enable();
+ state->stack_info.type = STACK_TYPE_UNKNOWN;
+ return false;
+}
+EXPORT_SYMBOL_GPL(unwind_next_frame);
diff --git a/arch/loongarch/kernel/vmlinux.lds.S b/arch/loongarch/kernel/vmlinux.lds.S
index b1686af..bb50d83 100644
--- a/arch/loongarch/kernel/vmlinux.lds.S
+++ b/arch/loongarch/kernel/vmlinux.lds.S
@@ -2,6 +2,7 @@
#include <linux/sizes.h>
#include <asm/asm-offsets.h>
#include <asm/thread_info.h>
+#include <asm/orc_lookup.h>
#define PAGE_SIZE _PAGE_SIZE
#define RO_EXCEPTION_TABLE_ALIGN 4
@@ -74,6 +75,8 @@ SECTIONS
}
#endif
+ ORC_UNWIND_TABLE
+
.got : ALIGN(16) { *(.got) }
.plt : ALIGN(16) { *(.plt) }
.got.plt : ALIGN(16) { *(.got.plt) }
diff --git a/arch/loongarch/power/Makefile b/arch/loongarch/power/Makefile
index 58151d0..bbd1d47 100644
--- a/arch/loongarch/power/Makefile
+++ b/arch/loongarch/power/Makefile
@@ -1,3 +1,5 @@
+OBJECT_FILES_NON_STANDARD_suspend_asm.o := y
+
obj-y += platform.o
obj-$(CONFIG_SUSPEND) += suspend.o suspend_asm.o
diff --git a/arch/loongarch/vdso/Makefile b/arch/loongarch/vdso/Makefile
index a50308b..e6a6b79 100644
--- a/arch/loongarch/vdso/Makefile
+++ b/arch/loongarch/vdso/Makefile
@@ -1,6 +1,8 @@
# SPDX-License-Identifier: GPL-2.0
# Objects to go into the VDSO.
+OBJECT_FILES_NON_STANDARD := y
+
# Include the generic Makefile to check the built vdso.
include $(srctree)/lib/vdso/Makefile
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index d7779a1..df29ddb 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -116,6 +116,14 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
*/
#define __stringify_label(n) #n
+#define __annotate_reachable(c) ({ \
+ asm volatile(__stringify_label(c) ":\n\t" \
+ ".pushsection .discard.reachable\n\t" \
+ ".long " __stringify_label(c) "b - .\n\t" \
+ ".popsection\n\t"); \
+})
+#define annotate_reachable() __annotate_reachable(__COUNTER__)
+
#define __annotate_unreachable(c) ({ \
asm volatile(__stringify_label(c) ":\n\t" \
".pushsection .discard.unreachable\n\t" \
@@ -128,6 +136,7 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
#define __annotate_jump_table __section(".rodata..c_jump_table")
#else /* !CONFIG_OBJTOOL */
+#define annotate_reachable()
#define annotate_unreachable()
#define __annotate_jump_table
#endif /* CONFIG_OBJTOOL */
diff --git a/scripts/Makefile b/scripts/Makefile
index 32b6ba7..d62d85f 100644
--- a/scripts/Makefile
+++ b/scripts/Makefile
@@ -29,7 +29,10 @@ ifdef CONFIG_UNWINDER_ORC
ifeq ($(ARCH),x86_64)
ARCH := x86
endif
-HOSTCFLAGS_sorttable.o += -I$(srctree)/tools/arch/x86/include
+ifeq ($(ARCH),loongarch)
+ARCH := loongarch
+endif
+HOSTCFLAGS_sorttable.o += -I$(srctree)/tools/arch/$(ARCH)/include
HOSTCFLAGS_sorttable.o += -DUNWINDER_ORC_ENABLED
endif
--
2.1.0
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH v1 0/6] Add objtool and orc support for LoongArch
2023-07-25 8:15 [PATCH v1 0/6] Add objtool and orc support for LoongArch Tiezhu Yang
` (5 preceding siblings ...)
2023-07-25 8:15 ` [PATCH v1 6/6] LoongArch: Add ORC unwinder support Tiezhu Yang
@ 2023-07-25 12:01 ` Peter Zijlstra
2023-08-09 9:55 ` Tiezhu Yang
6 siblings, 1 reply; 18+ messages in thread
From: Peter Zijlstra @ 2023-07-25 12:01 UTC (permalink / raw)
To: Tiezhu Yang
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On Tue, Jul 25, 2023 at 04:15:04PM +0800, Tiezhu Yang wrote:
> To be honest, there still exist some objtool warnings, I will try my best
> to fix them in the later versions or the other separate patches.
Please give us some examples of what's still going wrong. Are you
running into jump-tables? IIRC I've not seen code for those yet.
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH v1 0/6] Add objtool and orc support for LoongArch
2023-07-25 12:01 ` [PATCH v1 0/6] Add objtool and orc support for LoongArch Peter Zijlstra
@ 2023-08-09 9:55 ` Tiezhu Yang
0 siblings, 0 replies; 18+ messages in thread
From: Tiezhu Yang @ 2023-08-09 9:55 UTC (permalink / raw)
To: Peter Zijlstra
Cc: Josh Poimboeuf, Huacai Chen, loongarch, linux-kernel, loongson-kernel
On 07/25/2023 08:01 PM, Peter Zijlstra wrote:
> On Tue, Jul 25, 2023 at 04:15:04PM +0800, Tiezhu Yang wrote:
>
>> To be honest, there still exist some objtool warnings, I will try my best
>> to fix them in the later versions or the other separate patches.
>
> Please give us some examples of what's still going wrong. Are you
> running into jump-tables? IIRC I've not seen code for those yet.
>
Sorry for the late reply, I did more analysis and test recently.
I have already addressed most of the objtool warnings, discussed
offline with compiler engineers, sent separate patch to maillist
or modified the related code locally.
For example,
arch/loongarch/mm/fault.o: warning: objtool: no_context.part.0() falls
through to next function do_sigsegv()
[PATCH] LoongArch: Remove noreturn attribute for die()
https://lore.kernel.org/loongarch/1690871581-23944-1-git-send-email-yangtiezhu@loongson.cn/
Additionally, when I update the latest upstream gcc and binutils,
it generates more objtool warnings, like this:
init/version.o: warning: objtool: early_hostname+0x20: sibling call from
callable instruction with modified stack frame
[fedora@linux 6.5.test]$ gcc --version
gcc (GCC) 14.0.0 20230803 (experimental)
[fedora@linux 6.5.test]$ as --version
GNU assembler (GNU Binutils) 2.41.50.20230803
[fedora@linux 6.5.test]$ objdump -M no-aliases -D init/version.o
0000000000000000 <early_hostname>:
0: 00150085 or $a1, $a0, $zero
4: 1a000004 pcalau12i $a0, 0
8: 02ffc063 addi.d $sp, $sp, -16
c: 02810406 addi.w $a2, $zero, 65
10: 02c00084 addi.d $a0, $a0, 0
14: 29c02061 st.d $ra, $sp, 8
18: 54000000 bl 0 # 18 <early_hostname+0x18>
1c: 0281000c addi.w $t0, $zero, 64
20: 6c001584 bgeu $t0, $a0, 20 # 34 <.L2>
24: 1a000004 pcalau12i $a0, 0
28: 02810005 addi.w $a1, $zero, 64
2c: 02c00084 addi.d $a0, $a0, 0
30: 54000000 bl 0 # 30 <early_hostname+0x30>
0000000000000034 <.L2>:
34: 28c02061 ld.d $ra, $sp, 8
38: 00150004 or $a0, $zero, $zero
3c: 02c04063 addi.d $sp, $sp, 16
40: 4c000020 jirl $zero, $ra, 0
A preliminary analysis shows that the common functions is_sibling_call()
and add_jump_destinations() in tools/objtool/check.c need to be modified
to handle the case "reloc->sym->type == STT_NOTYPE", I will send a patch
in the next week.
Thanks,
Tiezhu
^ permalink raw reply [flat|nested] 18+ messages in thread