[Bug ld/27566] [RISC-V] relocation truncated to fit: R_RISCV_GPREL_I against aymbol

[lifang_xia at linux dot alibaba.com]

https://sourceware.org/bugzilla/show_bug.cgi?id=27566

--- Comment #12 from lifang_xia at linux dot alibaba.com ---
> 2024年10月25日 14:43，i at maskray dot me <sourceware-bugzilla@sourceware.org> 写道：
> 
> https://sourceware.org/bugzilla/show_bug.cgi?id=27566
> 
> Fangrui Song <i at maskray dot me> changed:
> 
>           What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                 CC|                            |i at maskray dot me
> 
> --- Comment #11 from Fangrui Song <i at maskray dot me> ---
> (In reply to lifang_xia from comment #9)
>> Hi we get this error a again.....
>> -------------
>> .text
>> 
>> hello:
>>        .rept 6000
>>        lla a0, ARCHOR0
>>        .endr
>> 
>> 
>> .section .rodata
>>        .set ARCHOR0, . + 4598
>>        .fill 100, 4, 0x12345678
>> 
>> .data
>> .align 3
>> world:
>>    .rept 860
>>    .long 0x1000
>>    .endr
>> -------------
>> build command:
>> 
>> binutils/build/gas/as-new -o 1.o 1.s -march=rv32gc -mabi=ilp32d
>> binutils/build/ld/ld-new -o 1 1.o -e hello -m elf32lriscv
>> 
>> we can get the error message:
>> 
>> 1.o: in function `hello':
>> (.text+0x0): relocation truncated to fit: R_RISCV_GPREL_I against `ARCHOR0'
>> ---------------------------------
>> It looks like the max alignment should include page size if the symbol is
>> not defined in the same section of GP.
>> 
>> Any idea about this?
> 
> This error still reproduces with GNU ld 2.43. lld can handle it correctly.
> 
> This is might be BFD riscv's linker relaxation with `.set ARCHOR0, . + 4598`
Yes. the relaxation like “.set ARCHOR0, . + 4598” being relaxed at 2nd pass or
3rd pass would be fine.
> 
> -- 
> You are receiving this mail because:
> You reported the bug.

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