[Bug ld/24226] Need advise on the binutils problem that generating wrong instruction like lw a3,-2048(a5) on RISC-V backend

[nelsonc1225 at sourceware dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=24226

Nelson Chu <nelsonc1225 at sourceware dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nelsonc1225 at sourceware dot 
org

--- Comment #6 from Nelson Chu <nelsonc1225 at sourceware dot org> ---
(In reply to Jim Wilson from comment #4)
> There is still a linker issue here, in that the linker should generate an
> error when the reloc overflows and computes the wrong address.

Seems hard to check this in linker since the lui and lw are not chained
together.

Consider the following reduced test case,

% cat tmp.s
lui a5, %hi(g_0x2c7fc)
lw  a4, %lo(g_0x2c7fc)(a5)
lui a5, %hi(g_0x2c7fc+4)
lw  a4, %lo(g_0x2c7fc+4)(a5)
lui a5, %hi(g_0x2c800)
lw  a4, %lo(g_0x2c800)(a5)
% riscv64-unknown-elf-as tmp.s -o tmp.o
% riscv64-unknown-elf-ld -defsym g_0x2c7fc=0x2c7fc -defsym g_0x2c800=0x2c800
tmp.o
/Users/nelsonc/work/build-upstream/build-elf64-upstream/build-install/bin/riscv64-unknown-elf-ld:
warning: cannot find entry symbol _start; defaulting to 00000000000100b0
% riscv64-unknown-elf-objdump -d a.out 

a.out:     file format elf64-littleriscv


Disassembly of section .text:

00000000000100b0 <__BSS_END__-0x1018>:
   100b0:       0002c7b7                lui     a5,0x2c
   100b4:       7fc7a703                lw      a4,2044(a5) # 2c7fc <g_0x2c7fc>
   100b8:       0002d7b7                lui     a5,0x2d
   100bc:       8007a703                lw      a4,-2048(a5) # 2c800
<g_0x2c800>
   100c0:       0002d7b7                lui     a5,0x2d
   100c4:       8007a703                lw      a4,-2048(a5) # 2c800
<g_0x2c800>

We don't chain the low instruction to it's lui, so we don't even know if the
lui is missing or cannot be shared (changed from 0x2c to 0x2d).  Compared to
auipc, since the low instructions are chained to quips, so we can check these
in details,
https://github.com/bminor/binutils-gdb/blob/master/bfd/elfnn-riscv.c#L1884

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