[Bug gdb/28371] gdb: watchpoint for local variable doesn't work on RISC-V

["shivam98.tkg at gmail dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=28371

--- Comment #11 from shivam98.tkg at gmail dot com <shivam98.tkg at gmail dot com> ---
OK, applying the diff that reverts that D69723 patch didn't work. clang does
emit function epilog CFI info. but debugging didn't work as expected.

clang new riscv assembly for function func2 :-

        .type   func2,@function
 59 func2:                                  # @func2
 60 .Lfunc_begin1:
 61         .loc    1 30 0                          # watch-cond.c:30:0
 62         .cfi_startproc
 63 # %bb.0:
 64         addi    sp, sp, -32
 65         .cfi_def_cfa_offset 32
 66         sd      ra, 24(sp)                      # 8-byte Folded Spill
 67         sd      s0, 16(sp)                      # 8-byte Folded Spill
 68         .cfi_offset ra, -8
 69         .cfi_offset s0, -16
 70         addi    s0, sp, 32
 71         .cfi_def_cfa s0, 0
 72         sd      a0, -24(s0)
 73 .Ltmp2:
 74         .loc    1 31 10 prologue_end            # watch-cond.c:31:10
 75         lui     a1, %hi(global2)
 76         lw      a0, %lo(global2)(a1)
 77         addiw   a0, a0, 1
 78         sw      a0, %lo(global2)(a1)
 79         .loc    1 32 1                          # watch-cond.c:32:1
 80         ld      s0, 16(sp)                      # 8-byte Folded Reload
 81         .cfi_def_cfa sp, -32
 82         ld      ra, 24(sp)                      # 8-byte Folded Reload
 83         .cfi_restore ra
 84         .cfi_restore s0
 85         addi    sp, sp, 32
 86         .cfi_def_cfa_offset 0
 87         ret
 88 .Ltmp3:
 89 .Lfunc_end1:
 90         .size   func2, .Lfunc_end1-func2
 91         .cfi_endproc
 92                                         # -- End function
 93         .globl  main                            # -- Begin function main
 94         .p2align        1
 95         .type   main,@function


Any Idea @Jim Wilson what else gdb expecting? Sorry for troubling you and the
forum.

I attached the clang-riscv, clang-x86, and gcc-riscv assembly and patch diff
file that I applied on llvm-project to generate the new assembly.

-- 
You are receiving this mail because:
You are on the CC list for the bug.