[PATCH v3] gdb/record: Support fense.tso, zicond, bitmanip, sinval and zihintntl in RISC-V

[Nelson Chu]

Just keep your original version and go ahead ;)  If it's convenient for
you, maybe some comments to mark these macros are used only for gdb now is
pretty good, so that later if people want to change something, they will
know the difference (Same instruction but use different macros) between
opcodes and gdb.  Like,

/* These are only used for gdb for now.  */
#define MATCH_BCLRI_RV32 0x48001013
#define MASK_BCLRI_RV32 0xfe00707f
#define MATCH_BEXTI_RV32 0x48005013
#define MASK_BEXTI_RV32 0xfe00707f
...

Thanks
Nelson

On Tue, Nov 18, 2025 at 2:44 PM Timur Golubovich <timurgol007@gmail.com>
wrote:

> So, should I put these macros to header or you are satisfied with initial
> variant?
>
> On 18 Nov 2025, at 05:44, Nelson Chu <nelson@rivosinc.com> wrote:
>
> 
>
>
> On Fri, Nov 7, 2025 at 6:49 PM Timur Golubovich <timurgol007@gmail.com>
> wrote:
>
>> Hi!
>> I tried to avoid adding new MATCH and MASK macros and now patch looks
>> like this (in total 25 new lines):
>>
>> > +#define MATCH_BCLRI_RV32 0x48001013
>> > +#define MASK_BCLRI_RV32 0xfe00707f
>> > +#define MATCH_BEXTI_RV32 0x48005013
>> > +#define MASK_BEXTI_RV32 0xfe00707f
>> > +#define MATCH_BINVI_RV32 0x68001013
>> > +#define MASK_BINVI_RV32 0xfe00707f
>> > +#define MATCH_BSETI_RV32 0x28001013
>> > +#define MASK_BSETI_RV32 0xfe00707f
>> > +#define MATCH_RORI_RV32 0x60005013
>> > +#define MASK_RORI_RV32 0xfe00707f
>> > DECLARE_INSN(rorw, MATCH_RORW, MASK_RORW)
>> > DECLARE_INSN(roriw, MATCH_RORIW, MASK_RORIW)
>> > +DECLARE_INSN(fence_tso, MATCH_FENCE_TSO, MASK_FENCE_TSO)q
>> > +DECLARE_INSN(rev8, MATCH_GREVI|MATCH_SHAMT_REV8_64,
>> MASK_GREVI|MASK_SHAMT)
>> > +DECLARE_INSN(zext_h, MATCH_PACKW, MASK_PACKW|MASK_RS2)
>> > +DECLARE_INSN(zip, MATCH_SHFLI|MATCH_SHAMT_ZIP_32,
>> MASK_SHFLI|MASK_SHAMT)
>> > +DECLARE_INSN(unzip, MATCH_UNSHFLI|MATCH_SHAMT_ZIP_32,
>> MASK_UNSHFLI|MASK_SHAMT)
>> > +DECLARE_INSN(orc_b, MATCH_GORCI|MATCH_SHAMT_ORC_B, MASK_GORCI |
>> MASK_SHAMT)
>> > +DECLARE_INSN(brev8, MATCH_GREVI|MATCH_SHAMT_BREV8,
>> MASK_GREVI|MASK_SHAMT)
>> > +DECLARE_INSN(rev8_rv32, MATCH_GREVI|MATCH_SHAMT_REV8_32,
>> MASK_GREVI|MASK_SHAMT)
>> > +DECLARE_INSN(rori_rv32, MATCH_RORI_RV32, MASK_RORI_RV32)
>> > +DECLARE_INSN(zext_h_rv32, MATCH_PACK, MASK_PACK|MASK_RS2)
>> > +DECLARE_INSN(bclri_rv32, MATCH_BCLRI_RV32, MASK_BCLRI_RV32)
>> > +DECLARE_INSN(bexti_rv32, MATCH_BEXTI_RV32, MASK_BEXTI_RV32)
>> > +DECLARE_INSN(binvi_rv32, MATCH_BINVI_RV32, MASK_BINVI_RV32)
>> > +DECLARE_INSN(bseti_rv32, MATCH_BSETI_RV32, MASK_BSETI_RV32)
>> > +DECLARE_INSN(c_sext_w, MATCH_C_ADDIW, MASK_C_ADDIW|MASK_RVC_IMM)
>>
>> but unfortunately I came across a mistake that now it doesn't compile,
>> because such macroses:
>>
>
> Oops...
>
>
>> #define MASK_RS1 (OP_MASK_RS1 << OP_SH_RS1)
>> #define MASK_RS2 (OP_MASK_RS2 << OP_SH_RS2)
>> #define MASK_RD (OP_MASK_RD << OP_SH_RD)
>> #define MASK_CRS2 (OP_MASK_CRS2 << OP_SH_CRS2)
>> #define MASK_CRS2S (OP_MASK_CRS2S << OP_SH_CRS2S)
>> #define MASK_IMM ENCODE_ITYPE_IMM (-1U)
>> #define MASK_RVC_IMM ENCODE_CITYPE_IMM (-1U)
>> #define MASK_UIMM ENCODE_UTYPE_IMM (-1U)
>> #define MASK_RM (OP_MASK_RM << OP_SH_RM)
>> #define MASK_PRED (OP_MASK_PRED << OP_SH_PRED)
>> #define MASK_SUCC (OP_MASK_SUCC << OP_SH_SUCC)
>> #define MASK_AQ (OP_MASK_AQ << OP_SH_AQ)
>> #define MASK_RL (OP_MASK_RL << OP_SH_RL)
>> #define MASK_AQRL (MASK_AQ | MASK_RL)
>> #define MASK_SHAMT (OP_MASK_SHAMT << OP_SH_SHAMT)
>> #define MATCH_SHAMT_REV8_32 (0b11000 << OP_SH_SHAMT)
>> #define MATCH_SHAMT_REV8_64 (0b111000 << OP_SH_SHAMT)
>> #define MATCH_SHAMT_BREV8 (0b00111 << OP_SH_SHAMT)
>> #define MATCH_SHAMT_ZIP_32 (0b1111 << OP_SH_SHAMT)
>> #define MATCH_SHAMT_ORC_B (0b00111 << OP_SH_SHAMT)
>> #define MASK_VD (OP_MASK_VD << OP_SH_VD)
>> #define MASK_VS1 (OP_MASK_VS1 << OP_SH_VS1)
>> #define MASK_VS2 (OP_MASK_VS2 << OP_SH_VS2)
>> #define MASK_VMASK (OP_MASK_VMASK << OP_SH_VMASK)
>> /* Vendor-specific (CORE-V) masks. */
>> #define MASK_CV_IS3_UIMM5 ENCODE_CV_IS3_UIMM5 (-1U)
>>
>> are defined only in riscv-opc.c. So probably the simplest variant for
>> now is either just add these new instructions and macroses as I
>> proposed in the beginning either put these macroses in riscv-opc.h.
>>
>
> Okay, thanks.
>
> Nelson
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://sourceware.org/pipermail/binutils/attachments/20251119/75c9ade5/attachment.htm>