[Bug ld/30930] Broken BTI veneers: ld-2.41 links mame in a way which gets stuck on aarch64

[nsz at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=30930

--- Comment #27 from Szabolcs Nagy <nsz at gcc dot gnu.org> ---
for the record a minimal reproducer when a bti veneer branches to itself:

$ cat a.s
.global _start
.type   _start, %function
_start:
        b       foo
.zero   0x07000000
$ cat b.s
.zero   0x01000000
.global foo
.type   foo, %function
foo:
        nop
$ as -o a.o a.s
$ as -o b.o b.s
$ ld -z force-bti a.o b.o 2>/dev/null
$ objdump -d a.out

a.out:     file format elf64-littleaarch64


Disassembly of section .text:

0000000000400108 <_start>:
  400108:       15c00004        b       7400118 <__foo_bti_veneer>
        ...
 740010c:       d503201f        .word   0xd503201f
 7400110:       14000008        b       7400130 <__foo_bti_veneer+0x18>
 7400114:       d503201f        nop

0000000007400118 <__foo_bti_veneer>:
 7400118:       90000010        adrp    x16, 7400000 <_start+0x6fffef8>
 740011c:       91046210        add     x16, x16, #0x118
 7400120:       d61f0200        br      x16
        ...

0000000008400130 <foo>:
 8400130:       d503201f        nop


i posted a patchset (version 1) to fix this:
https://sourceware.org/pipermail/binutils/2023-November/130299.html

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