https://llvm.org/bugs/show_bug.cgi?id=23986
Bug ID: 23986
Summary: RIP-relative memory references
Product: libraries
Version: trunk
Hardware: PC
OS: All
Status: NEW
Severity: normal
Priority: P
Component: Backend: X86
Assignee: [email protected]
Reporter: [email protected]
CC: [email protected]
Classification: Unclassified
llvm-mc is encoding some RIP-relative instructions incorrectly by failing to
relax arith imm8 instruction forms. For example:
------------- t.s -----------------
bar:
.section imul
imul $foo, bar(%rip), %bx
imul $foo, bar(%rip), %ebx
imul $foo, bar(%rip), %rbx
-----------------------------------
This is the current behavior. Note that in each case, $foo is encoded as an
8-bit immediate even though it is not guaranteed to fit in 8 bits.
bash-4.3$ llvm-mc -filetype=obj -triple x86_64-pc-linux-gnu t.s -o t.o
bash-4.3$ objdump -d --section=imul t.o
t.o: file format elf64-x86-64
Disassembly of section imul:
0000000000000000 <imul>:
0: 66 6b 1d 00 00 00 00 imul $0x0,0x0(%rip),%bx # 8 <imul+0x8>
7: 00
8: 6b 1d 00 00 00 00 00 imul $0x0,0x0(%rip),%ebx # f
<imul+0xf>
f: 48 6b 1d 00 00 00 00 imul $0x0,0x0(%rip),%rbx # 17
<bar+0x17>
16: 00
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
LLVMbugs mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/llvmbugs
